Academia.eduAcademia.edu
Download Free PDF
paper cover icon
Download Free PDF
paper cover thumbnail

Contributions to Building Physics

Contributions to Building Physics

Profile image of Bob MartensBob Martens

The present proceedings provides a collection of papers presented at the second Central European Building Physics Symposium (CESBP 2013), held in September 2013 at the Vienna University of Technology, Aus- tria. CESBP 2013 was organized by the Department of Building Physics and Building Ecology. Contributions were solicited in a broad range of topics in building physics, including heat and mass transfer in building ma- terials, building envelope, and whole buildings, buildings' energy performance, indoor climate and thermal comfort, hygrothermal building performance/moisture, air flow and ventilation, daylight and illumination en- gineering, building and room acoustics, urban physics, environmental impact, and life-cycle assessment. In response to the call for papers, a total of 206 abstracts were received. After the abstract and full paper re- view, 134 papers were accepted for inclusion in the proceedings. The papers’ authors come from 30 countries in European, as well as from Australia, Chile, China, Japan, and the USA. Thanks to these contributions, the present proceeding represents an excellent compendium of ongoing research and development work in build- ing physics. As the conference chairs, we thank all authors for their good work. We would also like to ex- presses our appreciation for the support we received from the organizers of the preceding CESBP2010 sym- posium. Furthermore, we recognize the support of the members of the international scientific committee and especially those colleagues who participated in the paper review process. Finally, the members of the local organizing committee delivered valuable input. We feel the present proceedings, with the depth and breadth of its contributions, represents a valuable asset to all those engaged in research and education in the field of building physics.

Related Papers

Research in Building Physics : Proceedings of the Second International Conference on Building Physics, Leuven, Belgium, 14-18 September 2003

2020 •

H. Hens

View PDF
Building Physics Conference (IBPC): The Role of Building Physics in Resolving Carbon Reduction Challenge and Promoting Human Health in Buildings. (pp. 377-384). Kyoto, Japan: Kyoto University

2012 •

Anker Nielsen

Use of sensitivity analysis to evaluate hygrothermal conditions in solid brick walls with interior insulation

View PDF
Nordic Symposium on Building Physics, NSB 2017, June 11, 2017

2017 •

Tore Kvande

In the next decades, buildings and infrastructure will be exposed to significantly different climatic strains than they are today. Still, building standards and design guidelines presuppose use of historic weather data. Thus, we need a point of departure for the support of decision-making aimed at reducing risk and climate vulnerability in the built environment. We therefore propose a new climate adaptation framework in compliance with the Norwegian Planning and Building Act; A general framework for climate adaptation and moisture-resilient buildings, tailor-made for implementation in both national and international standards, certification schemes and design guidelines.This paper has been written within the Klima 2050 Centre for Research-based Innovation. The authors gratefully acknowledge all our partners and the Research Council of Norway. A special thanks to CAD operator Remy EikpublishedVersio

View PDF
7th International Building Physics Conference IBPC 2018 Proceedings: Healthy, Intelligent and Resilient Buildings and Urban Environments

2019 •

Cedric Vuye

In cold and moderate climates, poorly designed construction joints can lead to local low surface temperatures, which entails local radiant asymmetry. An experiment was set up to test the hypothesis that overall and local thermal sensation is influenced by temperature difference, and position and distance of local radiant asymmetry. In the experiment, 18 subjects participated where they were introduced in a room at 21°C and 45% relative humidity. The subjects were exposed to local radiant asymmetry created by a cooling plate. This plate was positioned at three different heights, and controlled for temperatures at 3, 6 or 10°C below room air temperature. The data was analyzed using general linear modelling. The results show that thermal sensation is not influenced by local radiant asymmetry directly, but that the deviation from base comfort level is linked to height and temperature of the cold plate, as well as distance of the plate to the subject. This last effect proved to be the st...

View PDF
Energy and Buildings 51 (2012) 131–142 Contents lists available at SciVerse ScienceDirect Energy and Buildings

Principi Paolo

View PDF
IEA Annex 41 whole building heat, air, moisture response: Closing seminar, Nordic Building Physics Conference

2008 •

Monika Woloszyn

View PDF

Energies

Improving Building Energy Efficiency through Measurement of Building Physics Properties Using Dynamic Heating Tests

Ljubomir Jankovic

Buildings contribute to nearly 30% of global carbon dioxide emissions, making a significant impact on climate change. Despite advanced design methods, such as those based on dynamic simulation tools, a significant discrepancy exists between designed and actual performance. This so-called performance gap occurs as a result of many factors, including the discrepancies between theoretical properties of building materials and properties of the same materials in buildings in use, reflected in the physics properties of the entire building. There are several different ways in which building physics properties and the underlying properties of materials can be established: a co-heating test, which measures the overall heat loss coefficient of the building; a dynamic heating test, which, in addition to the overall heat loss coefficient, also measures the effective thermal capacitance and the time constant of the building; and a simulation of the dynamic heating test with a calibrated simulati...

View PDF

Renewable Energy

A field study on building inertia and its effects on indoor thermal environment

2012 •

Armando Oliveira

View PDF
A Simulation Method for Measuring Building Physics Properties

2019 •

Ljubomir Jankovic

Designing building energy performance with confidence requires accurate information on the properties of building materials and on assemblies of these materials. However, thermal properties of the building in use are rarely compared to manufacturers’ specifications. The approach reported in this paper determines building thermal properties using simulations of dynamic heating tests. It replaces the need for physical tests with the actual buildings, by conducting these tests with calibrated simulation models using data from energy performance monitoring of that building during its normal operation. The resultant method represents a tool for establishing physics properties of buildings in use before and after retrofit, and facilitates quality control of retrofit projects. The results pinpoint major discrepancies between theoretical and actual thermal properties before and after the retrofit, giving practical guidance for safety margins in relation to technical specifications of buildi...

View PDF

Energy and Buildings

Thermal behaviour of a building in its environment: Modelling, experimentation, and comparison

2018 •

sihem guernouti

View PDF
2013 A. Mahdavi B. Martens 2nd CENTRAL EUROPEAN SYMPOSIUM ON BUILDING PHYSICS ON BUILDING PHYSICS CRACOW - LODZ, SEPTEMBER 13-15, 2010 VIENNA, AUSTRIA, SEPTEMBER 9-11, 2013 2nd CENTRAL EUROPEAN SYMPOSIUM ON BUILDING PHYSICS • heat and mass transfer in building materials, building envelope, and whole buildings • buildings’ energy performance • indoor climate and thermal comfort • hygrothermal building performance/moisture • air flow and ventilation • daylight and illumination engineering • building and room acoustics • urban physics • environmental impact and life-cycle assessment Edited by: VIENNA, AUSTRIA,EUROPEAN SEPTEMBERSYMPOSIUM 9-11, 2013 1st CENTRAL The Symposium topics are research results on: CESBP Edited by: A. Mahdavi A. Mahdavi B. Martens B. Martens ISBN 978-3-85437-321-6 CESBP2013_Cover_final.indd 1 20.07.13 12:12 Proceedings of the 2nd Central European Symposium on Building Physics 9-11 September 2013, Vienna, Austria Contributions to Building Physics           Edited by: A. Mahdavi B. Martens Vienna University of Technology - Faculty of Architecture and Regional Planning, Vienna, Austria                                 Vienna, 2013 Copyright © 2013 Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria All rights reserved. No part of this publication or the information contained herein may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, by photocopying, recording or otherwise, without written prior permission from the publisher. Although all care is taken to ensure the integrity and quality of this publication and the information herein, no responsibility is assumed by the publishers or the author for any damage to property or persons as a result of operation or use of this publication and/or the information contained herein. Technical edition: U. Pont Book cover design: T. Waśniewski; H. Peter Printed by: RSA, Vienna, Austria ÖKK-Editions – Freundgasse 11 – A-1040 Vienna ISBN 978-3-85437-321-6 Table of Contents Preface Conference organisation xi xiii 1. Plenary lectures Building science with a human face: Reflections on the Austrian School of Building Physics Mahdavi, Ardeshir The energy turnaround in Europe and its consequences for renewable generation, energy infrastructure and end-use Brauner, Günther Intelligent traffic systems: Why should building physicists, architects, and urban designers care? Knoflacher, Hermann 3 9 13 2. Conference papers Recent advances in SEMERGY: A semantically enriched optimization environment for performance-guided building design and refurbishment Pont, Ulrich; Shayeganfar, Ferial; Ghiassi, Neda; Taheri, Mahnameh; Sustr, Christian; Mahdavi, Ardeshir; Heurix, Johannes; Fenz, Stefan; Anjomshoaa, Amin; Neubauer, Thomas; Tjoa, A Min Multi-objective optimization in the SEMERGY environment for sustainable building design and retrofit Heurix, Johannes; Taheri, Mahnameh; Shayeganfar, Ferial; Fenz, Stefan; Pont, Ulrich; Ghiassi, Neda; Anjomshoaa, Amin; Sustr, Christian; Neubauer, Thomas; Mahdavi, Ardeshir; Tjoa, A Min A comprehensive building model for performance-guided decision support Ghiassi, Neda; Pont, Ulrich; Shayeganfar, Ferial; Mahdavi, Ardeshir; Fenz, Stefan; Heurix, Johannes; Anjomshoaa, Amin; Neubauer, Thomas; Tjoa, A Min Functional and technological definition of BIM-aware services to assess, predict and optimize energy performance of buildings Rojicek, Jiri; Fisera, Radek; Kontes, Giorgos; Giannakis, Georgios; Lilis, Georgios; Rovas, Dimitrios Towards an ontology representing building physics parameters for increased energy efficiency in smart home operation Kofler, Mario Jerome; Kastner, Wolfgang 19 27 35 43 51 Recent advances in BIMSUSTAIN: The application of building information modeling in the context of building physics and building ecology Kiesel, Kristina; Skoruppa, Linda; Mahdavi, Ardeshir 59 Effect of drying methods, sample sizes and RH paths on sorption isotherms Feng, Chi; Janssen, Hans; Feng, Ya; Meng, Qinglin 65 Effect of temperature on the sorption isotherm and vapor permeability Feng, Chi; Janssen, Hans; Feng, Ya; Meng, Qinglin 71 Simulation of hygric performance of hysteretic building material exposed to cyclic changes of relative humidity Koronthalyova, Olga; Mihalka, Peter 77 Assessment of scale of the microstructure impact on capillary transport in cement-based composites with polypropylene fibers Wygocka, Agata; Garbalińska, Halina 83 Concrete with pozzolanic admixtures: An environmental-friendly solution Kulovaná, Tereza; Vejmelková, Eva; Keppert, Martin; Černý, Robert 87 Metamodelling in robust low-energy dwelling design Van Gelder, Liesje; Janssen, Hans; Roels, Staf 93 Inverse modelling to predict and characterize indoor climates Kramer, Rick; van Schijndel, Jos; Schellen, Henk 101 Individualised climate in future buildings. Fact or fiction? Dovjak, Mateja; Shukuya, Masanori; Krainer, Aleš 109 Evaluation of the applicability of the quasi-steady-state overheating indicator for wooden buildings Goethals, Kim; Smet, Lieselot; Janssens, Arnold; Laverge, Jelle 117 Overheating - an unexpected side-effect of decreased heating demand Kisilewicz, Tomasz 125 Assessing thermal comfort conditions in transitional states Wu, Yu-Chi; Orehounig, Kristina; Pont, Ulrich; Schuss, Matthias; Mahdavi, Ardeshir 131 Review of methods for evaluation of building energy enhancements Braun, Reiner; Dubisch, Florian; Judex, Florian; Kelly, Blaise; Vukovic, Vladimir 135 The mapping of simulated climate-dependent building innovations van Schijndel, Jos 141 Thermal performance effect of hollow ceramic microspheres coating assessed by dynamic outdoor testing in the summer season Čekon, Miroslav; Kalousek, Miloš Optimization of window installations in deep energy retrofits using vacuum insulation panels Misiopecki, Cezary; Kosny, Jan; Fallahi, Ali; DuPont, William 149 157 Effect of various Nordic climates on attaining the Passive House standard Gullbrekken, Lars; Nilsen, Espen Hobber; Kvande, Tore; Geving, Stig Study of hygrothermal behavior of advanced masonry components made with utilization of secondary row materials Zach, Jiri; Hroudova, Jitka; Korjenic, Azra 165 169 Applicability of capillary condensation sorption hysteresis model for burnt clay bricks Matiasovsky, Peter; Mihalka, Peter 175 Anti damp preservation of internally insulated brick walls Wójcik, Robert 179 Hygrothermal performance of internally insulated brick wall in a cold climate: field measurement and model calibration Klõšeiko, Paul; Arumägi, Endrik; Kalamees, Targo 185 Thermal performance of a test cell in a hot and humid climate: the impact of thermal insulation Tan, Pak Hooi; Pont, Ulrich; Müller, Veronika; Mahdavi, Ardeshir 193 Thermal Implications of radiant roof barriers: A field study in a hot and humid climate Müller, Veronika; Pont, Ulrich; Tan, Pak Hooi; Mahdavi, Ardeshir 199 An analysis of indoor environmental quality in an office: The case of a university campus in Istanbul Sunar, Pınar; Tanrıöver, Sezin 207 Thermal environment in detached houses with atrium: Towards proper utilization of atrium space in traditional dwellings “Kyo-machiya” Iba, Chiemi; Ito, Shihono; Hokoi, Shuichi; Ogura, Daisuke 215 Thermal performance analysis of traditional housing in Kosovo Deralla, Albana; Mahdavi, Ardeshir A comparison of projected and actual energy performance of buildings after thermal retrofit measures Housez, Pierre Pascal Housez; Pont, Ulrich; Mahdavi, Ardeshir 223 229 Design and construction of a plus energy house Schoch, Torsten 237 Simulation of thermal performance and retrofit of a historic greenhouse Ward, Rebecca Mary; Choudhary, Ruchi; Mortada, Adnan 245 The benefits of FEM-SS-BES (Finite Element Method, State-Space, Building Energy Simulation) modeling exchange for building physics van Schijndel, Jos; Kramer, Rick 253 Experimental and numerical energy performance analysis of PCM-enhanced building envelope products and systems Kosny, Jan; Fallahi, Ali; Shukla, Nitin 261 Influence of selected calculation tool on a design process: A case study Massetti, Marco; Morishita, Naomi; Bednar, Thomas 269 Generalizing roof geometry from minimal user input for building performance simulation Hammerberg, Kristopher; Ghiassi, Neda; Mahdavi, Ardeshir 277 Experimental hygrothermal study in wood and wood-based materials Shukla, Nitin; Elliott, Diana; Kumar, Devendra; Misiopecki, Cezary; Kosny, Jan 285 Laboratory investigation of drying of built-in moisture in wood frame walls at passive house level Dalehaug, Arvid; Geving, Stig; Gaare, Maret; Løtveit, Kirsti; Holme, Jonas 291 Wooden beam ends in masonry with interior insulation – A literature review and simulation on causes and assessment of decay Kehl, Daniel; Ruisinger, Ulrich; Plagge, Rudolf; Grunewald, John 299 A hygrothermal analysis of international timber frame wall assemblies tested under temperate maritime climatic conditions Corcoran, Lee Christopher; Duffy, Aidan; Rouholamin, Sima 305 Long-term measurements and simulations of five internal insulation systems and their impact on wooden beam heads Ruisinger, Uli 313 Derivation of an evaluation method for the hygrothermal and biohygrothermal behaviour of straw as insulation Klatecki, Marc; Maas, Anton 321 Smart vapour barriers in unventilated wooden roofs in a Nordic climate – laboratory study of drying effect under shaded conditions Geving, Stig; Stellander, Markus 327 Computer modelling to evaluate the risks of damage to objects exposed to varying indoor climate conditions in the past, present, and future Schellen, Henk; Huijbregts, Zara; Martens, Marco; van Schijndel, Jos 335 Thermal comfort of individual rooms in the design of commercial buildings Nowak-Dzieszko, Katarzyna Justyna; Nowak, Katarzyna; Rojewska-Warchal, Malgorzata 343 The aid of TRNSYS simulation for the conservation of an artwork. A case study De Backer, Lien; Janssens, Arnold; De Paepe, Michel; Van Belleghem, Marnix 351 Experimental measurements and analysis of the indoor conditions in Italian museum storerooms: A study case Cappelletti, Francesca; Birra, Anna; Romagnoni, Piercarlo 357 Variability assessment of summer comfort conditions in social housing using in situ measurements Curado, António José Candeias; de Freitas, Vasco P.; Ramos, Nuno M. M. 365 Thermal environment in a room with dynamic infrared fireplace heater Ponechal, Radoslav 373 Investigation of ceiling fans for improving summer thermal comfort Voss, Karsten; Voss, Tjado; Otto, Julius; Schweiker, Marcel; Rodriguez Ubinas, Edwin 381 Effect of temperature on water vapour transport properties Fořt, Jan; Pavlík, Zbyšek; Žumár, Jaromír; Černý, Robert 387 A new approach to measure liquid transport in capillary active interior insulation Binder, Andrea Birgit; Zirkelbach, Daniel Maria; Künzel, Hartwig 393 Detailed heat, air and moisture transport modelling in cavity walls Van Belleghem, Marnix; De Backer, Lien; Janssens, Arnold; De Paepe, Michel; Steeman, Marijke 401 On the use of the logarithmic of the capillary pressure for numerical simulation of moisture flow Rode, Carsten; Lasse, Juhl 409 The extent and implications of the urban heat island phenomenon in Central European region Kiesel, Kristina; Vuckovic, Milena; Mahdavi, Ardeshir 415 Urban energy and microclimate: Wind tunnel experiments and multiscale modeling Carmeliet, Jan; Allegrini, Jonas; Moonen, Peter; Saneinejad, Saba; Dorer, Viktor 421 Spatial distribution of wind-driven rain in the urban environment Kubilay, Aytac; Derome, Dominique; Blocken, Bert; Carmeliet, Jan 429 The use of vegetation for social housing renovations: A case study in the City of Palermo Pastore, Luisa; Corrao, Rossella; Heiselberg, Per 437 Impacts of building performance monitoring on integrated energy management Menzel, Karsten; Katzemich, Frank; Mahdavi, Ardeshir 443 Alternative ways for advanced energy management Browne, Donal; Foley, Raimond; Schuß, Matthias; Simonis, Helmut; de Torre, Cristina; Willis, David 451 A middleware platform for integrated building performance management Schülke, Anett; Floeck, Martin; Schmidt, Mischa 459 Capabilities of IFC 4 for advanced building performance management Menzel, Karsten; Weise, Matthias; Liebich, Thomas; Valmaseda, Cesar 467 Thermal energy flow balancing for optimizing energy performance in public swimming pools with solar thermal micro generation: A case study de Torre, Cristina; Simonis, Helmut; Macía, Andrés; García, Miguel A.; Valmaseda, César 475 Degradation of the great buddha monument in the Sukhothai ruins Hokoi, Shuichi; Yoshida, Yuri; Ogura, Daisuke 483 Hygrothermal modelling of flooding events within historic buildings Huijbregts, Zara; van Schijndel, Jos; Schellen, Henk 491 Degradation of mural paintings of Mogao Cave 285 in Dunhuang Ogura, Daisuke; Hokoi, Shuichi; Uno, Tomoko; Abuku, Masaru; Okada, Ken; Hase, Takahide 499 Optimization of a wall base ventilation system to control rising damp ventilation with outside or inside air? Guimarães, Ana Sofia; de Freitas, Vasco P.; Delgado, João 507 Prevention strategies for risk of moisture related damages due to renovation of historical buildings Guizzardi, Michela; Derome, Dominique; Vonbank, Roger; Carmeliet, Jan 513 A study of visual and non-visual effects of daylighting in an office Košir, Mitja; Krainer, Aleš; Kristl, Živa 521 Healthy transparency: A novel aspect of daylighting design Hraska, Jozef; Zeman, Michal; Hanuliak, Peter; Mankova, Lucia; Stebelova, Katarina 529 Bio-inspired responsive façades Park, Jong Jin; Dave, Bharat 537 A new tool for quick room acoustic assessment in architectural education Hennings, Detlef; Voss, Karsten 545 Renovating a lecture hall with a glass roof: A case study of performance based design Atça, Emre; İlal, Mustafa Emre; Başaran, Tahsin; Kazanasmaz, Tuğçe; Durmuş Arsan, Zeynep 551 Two case studies in optimization-based thermal building performance model calibration Taheri, Mahnameh; Tahmasebi, Farhang; Mahdavi, Ardeshir 559 Method to assess the load shifting potential by using buildings as a thermal storage Braun, Reiner; Judex, Florian; Brychta, Markus; Popovac, Mirza 565 Model-assisted control through co-simulation for intelligent Building Energy Management Systems design Giannakis, Georgios; Katsigarakis, Kyriakos; Kontes, Giorgos; Rovas, Dimitrios 571 Simulation of movable translucent aerogel shutters Evins, Ralph; Dowson, Mark 579 Calculation models for the diffuse fraction of global solar radiation Vazifeh, Ehsan; Dervishi, Sokol; Mahdavi, Ardeshir 587 Numerical quality of a model for coupled heat and moisture transport in COMSOL multiphysics Bianchi Janetti, Michele; Ochs, Fabian; Feist, Wolfgang 591 Implementation of an efficient numerical solution method to simulate freezing processes in porous media Sontag, Luisa; Nicolai, Andreas 599 Modelling salt effects on sorption isotherm of porous materials Abuku, Masaru; Ogura, Daisuke; Hokoi, Shuichi 605 Comparison of simplified moisture penetration profile models with experimental data Matiasovsky, Peter; Holubek, Matus 613 Model and program for the prediction of the indoor air temperature and indoor air relative humidity Häupl, Peter; Bishara, Ayman; Hansel, Frank Thermal mass effect of solid block Aerated Autoclaved Concrete Urban, Bryan; Elliott, Diana; Shukla, Nitin; Fallahi, Ali; Kosny, Jan 619 627 Effect of cavity filler on the effective thermal conductivity of hollow bricks: A computational analysis based on accurate input data Černy, Robert; Kočí, Jan; Maděra, Jiří; Jerman, Miloš 635 An experimental and numerical study on the efficiency of the polypropylene fibres admixture in reducing pore pressure in heated concrete Witek, Arkadusz; Gawin, Dariusz 641 Thermal calculations of the thermal performance of hollow ceramic blocks Moga, Ligia Mihaela; Moga, Ioan 647 Influence of the moisture in the thermal conductivity of Expanded Polystyrene insulators Lakatos, Ákos; Varga, Sándor; Kalmár, Ferenc 653 Simulation-powered virtual sensors in building monitoring systems Zach, Robert; Mahdavi, Ardeshir 657 Simulation studies method to identify occupancy schedules from indoor climate measurements Freudenberg, Peggy 665 Variability of energy and water consumption of school buildings Almeida, Ricardo M.S.F.; Ramos, Nuno M. M.; Simões, M. Lurdes; de Freitas, Vasco P. 671 Influence of the user behavior in the total energy consumption of social housing in Chile Rojo, Claudia; Fissore, Adelqui 679 Rebound effects in heating energy consumption after thermal refurbishment of appartment buildings Sölkner, Petra Johanna; Mahdavi, Ardeshir 685 Evaluation of two-eddy viscosity turbulence models to predict temperature and velocity indoor environments as exercise for new CFD users Cortés, Magdalena; Bustamante, Waldo; Rao, Jiwu; Fazio, Paul; Vera, Sergio 693 Numerical analysis of the indoor comfort and ventilation characteristics for an office room equipped with a façade-integrated ventilation system Popovac, Mirza 701 Stochastic character of local wind speed in application to reliability model of ACH Pietrzyk, Krystyna 709 Flow around and through a building storey with fully opened or tilted windows Teppner, Renate; Langensteiner, Bernd; Meile, Walter; Brenn, Günter; Kerschbaumer, Sybill 717 Numerical experiment investigation on turbulent natural convection in an enclosed tall cavity: Boussinesq-Approximation effect wang, changshan; Grunewald, John Thermal properties of contemporary lightweight cavity bricks: A semi-scale experimental study Pavlik, Zbysek; Fiala, Lukas; Jerman, Milos; Fort, Jan; Černý, Robert Measurement of hydration power and heat of hydration of Portland cement with fine ceramic dust using isothermal calorimetry Zákoutský, Jan; Tydlitát, Vratislav; Černý, Robert 725 733 739 Experimental evaluation of drying kinetics of building materials Barreira, Eva; Delgado, João M.P.Q.; Ramos, Nuno M.M.; de Freitas, Vasco P. 745 Experimental determination of the local interior heat transfer coefficient Vereecken, Evy; Janssen, Hans; Roels, Staf 751 Addressing surface temperature build-up due to the concentration of solar radiation created by curved glass facades Shaxted, Matthew; Raines, Ben; Stark, Don; Weismantle, Peter; Burton, Craig; Roudsari, Mostapha; Stochetti, Alejandro Effect of horizontal overhangs and glazing with spectral radiative properties on annual thermal balance of the building and thermal comfort Nowak, Henryk; Nowak, Lukasz; Sliwinska, Elzbieta; Staniec, Maja 759 767 Quantification of wind-driven rain and evaluation of façade humidification de Freitas, Sara Stingl; Barreira, Eva; de Freitas, Vasco P. 775 The impact of pressure equalization on watertightness of rainscreen systems Van Den Bossche, Nathan; Lacasse, Michael; Janssens, Arnold 783 Interior insulation of masonry walls – Assessment and design Steskens, Paul; Loncour, Xavier; Roels, Staf; Vereecken, Evy 791 Determination of the natural ventilation by measuring the CO2-content and estimating the number of occupants in a museum building Göttig, Roland; Schmidt, Simon; Nguyen, Mai-Khanh 799 Enhanced U-value of a wall structure using interior sensor driven forced or natural convection Hagentoft, Carl-Eric 805 Measuring air change rates in rooms using tracer gas and radon monitoring Becker, Rachel; Haquin, Gustavo; Kovler, Konstantin 811 Effects of combined shading and air exfiltration on lightweight building elements with high outer diffusion resistance Nusser, Bernd; Teibinger, Martin 819 Experimental investigations about the air flow in the ventilation layer of low pitched roofs Nusser, Bernd; Teibinger, Martin 827 A method for estimating amount of stored heat in PCM that exchange heat with flowing water Fujita, Koji; Matsushita, Takayuki 835 Structural building components as energy storage systems – Experimental investigations Cusnick, Lars; Nasrollahi, Kamyar; Pahn, Matthias 841 Influence of using treated wood on thermal properties and heat losses in basements of wooden buildings Vahalova, Eva; Peuhkuri, Ruut; Suhajda, Karel 847 Environmental and energy performance evaluation of wooden buildings and their comparison with constructions from different materials Badurova, Silvia; Ďurica, Pavol; Časnocha, Peter 855 Development and optimization of advanced silicate plasters materials for building rehabilitation Korjenic, Azra; Zach, Jiri; Hroudova, Jitka; Petranek, Vit; Korjenic, Sinan; Bednar, Thomas 863 Selection of critical weather year for hygrothermal analyses in the Czech Republic Koci, Jan; Madera, Jiri; Černý, Robert 869 Definition of generic materials by using a cluster analysis method Zhao, Jianhua; Grunewald, John; Plagge, Rudolf 875 Influence of materials’ moisture absorption properties on prediction of indoor relative humidity level Borodinecs, Anatolijs; Zemitis, Jurgis 883 Hygrothermal function of basement walls Blom, Peter; Bohlerengen, Trond 889 Damage of cement mortar microstructure induced by salt crystallization Koniorczyk, Marcin; Gawin, Dariusz; Konca, Piotr 895 Application of modelling tools to the design and performance validation of a life safety system Wilson, Laurence; Shaxted, Matthew 901 Performance of natural, exhaust, demand controlled exhaust and heat recovery residential ventilation systems as prescribed by the standards in 5 European countries. Laverge, Jelle; Pollet, Ivan; Janssens, Arnold; Vens, Anneleen Does a ventilation rate optimized for good health and low heat loss depend on built form? Das, Payel; Shrubsole, Clive; Davies, Michael; Jones, Benjamin; Hamilton, Ian; Chalabi, Zaid; Milner, James; Wilkinson, Paul; Ridley, Ian Parametric study of the summer behavior in timber houses – air flow in detached houses considering the influence of insect screens Ehrlich, Florian; Teibinger, Martin; Bednar, Thomas Numerical analysis of convective drying of gypsum boards Derome, Dominique; Defraeye, Thijs; Houvenaghel, Geert; Carmeliet, Jan Methodology to integrate energy efficiency, safe moisture performance and indoor environment quality in building renovation projects Ojanen, Tuomo; Holopainen, Riikka; Viitanen, Hannu; Lehtovaara, Jorma; Vinha, Juha; Kero, Paavo Multi-objective optimization for school buildings retrofit combining artificial neural networks and life cycle cost Almeida, Ricardo M.S.F.; de Freitas, Vasco P. Energy efficiency and comfort improvement in historic buildings: A methodology for diagnosis and interventions evaluation García-Fuentes, Miguel Ángel; Hernández-García, José Luis; Meiss, Alberto; de Torre, Cristina; García-Gil, Daniel Oekohaus: A case study of monitoring-based building performance assessment Skoruppa, Linda; Schuss, Matthias; Zach, Robert; Pont, Ulrich; Mahdavi, Ardeshir 909 917 923 931 937 943 951 959 Comparison of probabilistic approaches to the mould growth problem Sadovsky, Zoltan; Koronthalyova, Olga; Matiasovsky, Peter 965 Author index 969 Keywords index 975 Preface The present proceedings provides a collection of papers presented at the second Central European Building Physics Symposium (CESBP 2013), held in September 2013 at the Vienna University of Technology, Austria. CESBP 2013 was organized by the Department of Building Physics and Building Ecology. Contributions were solicited in a broad range of topics in building physics, including heat and mass transfer in building materials, building envelope, and whole buildings, buildings' energy performance, indoor climate and thermal comfort, hygrothermal building performance/moisture, air flow and ventilation, daylight and illumination engineering, building and room acoustics, urban physics, environmental impact, and life-cycle assessment. In response to the call for papers, a total of 206 abstracts were received. After the abstract and full paper review, 134 papers were accepted for inclusion in the proceedings. The papers’ authors come from 30 countries in European, as well as from Australia, Chile, China, Japan, and the USA. Thanks to these contributions, the present proceeding represents an excellent compendium of ongoing research and development work in building physics. As the conference chairs, we thank all authors for their good work. We would also like to expresses our appreciation for the support we received from the organizers of the preceding CESBP2010 symposium. Furthermore, we recognize the support of the members of the international scientific committee and especially those colleagues who participated in the paper review process. Finally, the members of the local organizing committee delivered valuable input. We feel the present proceedings, with the depth and breadth of its contributions, represents a valuable asset to all those engaged in research and education in the field of building physics. Ardeshir Mahdavi Bob Martens xi Conference organisation International Scientific Committee Ardeshir Mahdavi - Austria, chair Dariusz Gawin - Poland, vice-chair Robert Černý - Czech Republic, vice-chair John Grunewald - Germany, vice-chair Peter Matiasovsky - Slovakia, vice-chair Jesper Arfvidsson, Sweden Mark Bomberg, USA Vasco Peixoto de Freitas, Portugal Stig Geving, Norway Carl-Eric Hagentoft, Sweden Peter Häupl, Germany Shuichi Hokoi, Japan Arnold Janssens, Belgium Jan Kosny, USA Elżbieta Kossecka, Poland Jaroslav Kruis, Czech Republic Hartwig Künzel, Germany Bob Martens, Austria Leo Pel, Netherlands Carsten Rode, Danmark Staf Roels, Belgium Henk Schellen, Netherlands Matthias Schuß, Austria Libor Vozar, Slovakia Local Organizing Committee Ardeshir Mahdavi - chair Vienna University of Technology, Austria Bob Martens - vice-chair Vienna University of Technology, Austria Kristina Kiesel, VUT Ulrich Pont, VUT Matthias Schuß, VUT Robert Zach, VUT Elisabeth Finz, VUT Josef Lechleitner, VUT xiii Review Committee Jesper Arfvidsson, Sweden Martin Bauer, Germany Thomas Bednar, Austria Mark Bomberg, Canada Reiner Braun, Germany Markus Brychta, Austria Jan Carmeliet, Switzerland Robert Černy, Czech Republic Vasco de Freitas, Portugal Dominique Derome, Switzerland Sokol Dervishi, Austria Clemens Felsmann, Germany Dariusz Gawin, Poland Stig Geving, Norway Stefan Glawischnig, Austria John Grunewald, Germany Carl-Eric Hagentoft, Sweden Peter Häupl, Germany Shuichi Hokoi, Japan Andreas Holm, Germany Emre Ilal, Turkey Arnold Janssens, Belgium Kristina Kiesel, Austria Azra Korjenic, Austria Jan Kosny, United States Elisabeth Kossecka, Poland Jaroslav Kruis, Czech Republic Hartwig Künzel, Germany Sergio Leal, Austria Ardeshir Mahdavi, Austria Bob Martens, Austria Peter Matiasovsky, Slovakia Christoph Nytsch-Geusen, Germany Kristina Orehounig, Switzerland Leo Pel, Netherlands Olivier Pol, Austria Ulrich Pont, Austria Mirza Popovac, Austria Anita Preisler, Austria Carsten Rode, Denmark Staf Roels, Belgium Henk Schellen, Netherlands Matthias Schuß, Austria Klaus Sedlbauer, Germany Tim Selke, Austria Georg Siegel, Austria Wolfgang Streicher, Austria Georg Suter, Austria Renate Teppner, Austria Christoph Van Treeck, Germany Libor Vozar, Slovakia Vladimir Vukovic, Austria Robert Zach, Austria xiv 1. Plenary lectures Building science with a human face: Reflections on the Austrian School of Building Physics A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: This contribution deals with the essential characteristics of a particular tradition of teaching, practice, and research in the building physics area. We refer here to this tradition as the "Austrian School of Building Physics" (ASBP). The main features of this school are: tight integration with architectural design and construction, multi-scale and integrative approach, environmental orientation, and incorporation of human-ecological reasoning. The contribution is dedicated to Erich Panzhauser, as the distinguishing characteristics and accomplishments of the ASBP can be attributed to his prolific teaching and research tenure at the Vienna University of Technology. The present discussion focuses on the essential characteristics of a specific tradition of teaching, practice, and research in the building physics area, referred to here as the "Austrian School of Building Physics" (ASBP). It provides a number of reflections regarding the personal contributions and contextual circumstances that could help understand the main features of this school of building physics. These are, in a nutshell, ASBP's tight integration with architectural design and construction, its multi-scale and integrative approach, its persistent environmental orientation, and – perhaps most characteristically – its inclusion of human-ecological reasoning. These distinguishing characteristics as well as the evolution and accomplishments of the ASBP can be mainly attributed to the long and productive teaching and research tenure of Professor Erich Panzhauser at the Vienna University of Technology. Note that the purpose of this treatment is not to comparatively evaluate Erich Panzhauser's (and the ASBP's) particular style of building physics pursuit in reference to other traditions and surely not to demonstrate its superiority. As in many other scientific and technical domains, each tradition of inquiry has its points of strengths and insights. Thus, while here the focus is on the ASBP, there is no doubt that, given the increased interconnectedness of professionals and widespread exchange of ideas and scientific findings, the multitudinous traditions in building physics can advantageously inform – and learn from – each other. Accordingly, the present contribution focuses on a few hallmarks of the ASBP that could be of interest to the building physics community at large. SOURCES OF DIVERSITY IN APPROACHES TO BUILDING PHYSICS As a scientific and technical discipline, building physics is invested in a rational and technically sound approach to the conception, realization, and operation of buildings. The orientation, types, and even the character (style or "flavor") of building physicists' research and professional activities, however, vary considerably due to multiple parameters and circumstances. For one thing, there is the aspect of domain, i.e., if thermal, visual, acoustical or other aspects of building performance are the focus of inquiry or project. The scale of the inquiry also plays a role, as the focus may be on building materials and elements, whole buildings, or even urban-level questions. Educational background is yet another factor: Building physicists come from various engineering disciplines such as physics, mechanical engineering, civil engineering, and – less frequently – architecture. The sophistication of prevailing construction techniques and practices, the existence and rigor of applicable standards, and the economical and administrative boundary conditions of the building delivery process also influence the orientation of the professional community associated with building physics. Less tangible but no less essential parameters in the emergence of a specific tradition in building physics pertain to prevailing societal and intellectual currents. For example, the level of ecological awareness, intellectual discourse, or conceptions regarding professional ethics in a community imprint the outlook, thinking style, and problem solving approaches of building physics researchers, experts, and teachers. 3 infrared thermography. The results of these studies, together with the application of economic optimization methods led Panzhauser to recommend rather extensive thermal insulation requirements for buildings (Panzhauser et al. 1980). These recommendations were well (some four decades) ahead of their time, and are being only recently implemented on a broad basis. A further instance of Panzhauser's recognition of the importance (and early adoption) of rigorous building physics methods for building construction can be discerned in Panzhauser's collaboration with mathematician Walter Heindl, which, amongst other things, resulted in one of the early comprehensive computational (numeric) treatments of the heat transfer phenomena in thermal bridges (Heindl et al. 1987). Figure 1. Professor Erich Panzhauser. GROUNDING IN ARCHITECTURE AND BUILDING CONSTRUCTION The home of the ASBP and the first academic Department of Building Physics in Austria was established at the Institute of Building Construction, Faculty of Architecture and Planning, Vienna University of Technology. The institutional setting, together with the professional background of its first Head, Erich Panzhauser, can shed light on certain characteristics of the work conducted at this Department. Panzhauser studied both chemistry and architecture. He was thus in a position to introduce a scientific thinking style from natural sciences into the domain of architecture and building construction (Panzhauser et al. 1987, 1991, Haider et al. 1987, 1991, Ertl et al. 1987). The Department emerged from the Institute of Building Construction (Hochbau Institut), which, under the leadership of Karl Kupsky, had developed – in the sixties and the seventies of the last century – a solid reputation in teaching excellence as related to building detailing and construction. Panzhauser continued this tradition toward its logical development within the then emerging field of building physics. His approach to building physics was hence firmly grounded in an understanding of the rich building construction tradition in German speaking countries. He recognized the critical potential of building physics toward explicit formalization of physical (hygrothermal) processes as an opportunity to ground the design of building construction details and assemblies on a solid and accountable basis. Toward this end, Panzhauser pioneered in Austria the deployment of both diagnostics and computational methods. Immediately after the 1973 oil crisis, Panzhauser conducted large scale studies of the thermal quality of existing buildings in Austria via Figure 2. Numeric heat transfer analysis of thermal bridges (from Heindl et al. 1987). 4 principles. At the same time, the strategy embraced the significant importance of architecture's cultural and societal standing. Finding and maintaining this balance is as such not trivial and continues to represent a challenge even today. The ASBP advocated a view of architects and building physicists not as adversaries, but as collaborators pursuing common objectives. HUMAN ECOLOGY Perhaps the single most characteristic feature constitutive of the ASBP is its association with human ecology. Ecology denotes in general the scientific discipline that deals with the relationships between organisms and their surrounding world. Accordingly, human ecology may be simply defined as the ecology of the Homo sapiens. There are many traditions in (and approaches to) human ecology. Relevant for the ASBP is the "Vienna School of Human Ecology" (Knötig 1992a, 1992b, Mahdavi 1996a). Helmut Knötig, a biologist by background and a long-term collaborator of Erich Panzhauser, was the main intellectual driver behind the Vienna School of Human Ecology and responsible for the articulation and dissemination of its underlying conceptual instruments. Convinced of the broad explanatory utility of these concepts, Knötig continuously reiterated and refined them. The encounter with Panzhauser provided an opportunity to inform and enrich the treatment of environmental design and building physics at the Vienna University of Technology with human ecological insights. To introduce freshmen students to the subject, Knötig and Panzhauser initiated in the late seventies of the last century an influential interdisciplinary seminar on human ecology (Humanökologisches Proseminar). The course brought together experts from natural and social sciences as well as engineering disciplines not only to broaden the students' intellectual horizon, but also to demonstrate the critical necessity for an integrative approach to problem solving in the domain of the built environment. Knötig and Panzhauser realized that the emergent divergence of multiple historically evolved terminologies and conceptual instruments in various disciplines can represent an obstacle both to shared understanding and to practical collaboration in view of tasks and problems in the built environment. They saw in the rather high-level conceptual constructs of human ecology the potential to provide a common frame of reference. Thinking with foresight, they used this high-level conceptual frame of reference to facilitate a common understanding and a culture of collaboration amongst scientists, engineers, and other professionals of diverse backgrounds. Figure 3. Sample cover of research report ("planning of window ventilation") from a series of publications pertaining to research projects initiated by Erich Panzhauser. A further important component of the Department's work implied hands-on involvement in the building delivery process. There are building physics institutes that excel, for example, in the empirical – laboratory based – study of building materials and components properties. Others collaborate closely with industry with regard to product specification and development. Under Panzhauser, the Building Physics Department was engaged in numerous collaborative activities with architects pertaining to actual building projects (both new building construction and retrofit). Experiences gained from such "in the trenches" activities were constitutive of the ASBP knowledge base and provided an effective bridge between the engineering and design aspects of the building delivery process. Collaboration and interaction with architects as primary building designers influenced also the ASBP with regard to teaching, didactic strategies, and dissemination approaches. Specifically, the standing of the ASBP within the Faculty of Architecture and Planning posed both interesting possibilities and formidable challenges. A relevant case in point pertains to question of approaching – and reaching – "artistically minded" architecture students. Panzhauser followed a pedagogical strategy that did not favor an over-simplified presentation and treatment of building physics 5 In this context, it would be useful to briefly discuss here a few terms and concepts applied in the Vienna School of human ecology. Such a discussion is likely to clarify the relevance of human ecology as constitutive intellectual resource for the ASBP. In human ecology, architecture and the entire endeavor of building construction and operation may be viewed as an integral part of the totality of largely regulatory operations initiated by human beings as they interact with their surrounding world. Amongst other things, human ecology offers a useful way of thinking about these interactions via a central pair of concepts, namely the human beings’ ecological potency and the surrounding world’s ecological valency (Knötig 1992a, Mahdavi 1996b). Stated in simple terms, ecological potency refers to a dynamic human repertoire of capabilities and means of dealing (i.e., coping and interacting) with the surrounding world. On the other hand, ecological valency denotes the totality of that surrounding world’s characteristics (resources, possibilities, opportunities, challenges, risks, hazards, pitfalls) as it relates to, confronts, or accommodates the human ecological potency repertoire. Note that this latter concept (ecological valency) was essentially dealt with by Uexküll (Uexküll 1920) and is also akin to Gibson's concept of affordance (Gibson 1977, 1979). Using this conceptual framework, we can describe the main consideration in human ecology as the complex and dynamic relationships between the ecological potency of human beings and the ecological valency of their surrounding world. We can thus broadly characterize the entire building construction and operation endeavor in human ecological terms: Buildings are constructed and maintained with the implicit or explicit intention to favorably influence the relationship between people's ecological potency and the ecological valency of their surrounding world. Such an intention expresses itself, for example, in the "shelter function" of vernacular architecture (Mahdavi 1996c, 1989). Given this background, it should be clear why such human ecological concepts can aid structured thinking in building physics. We need to understand people's ecological potency – and its variations in the population – in order to define quality requirements pertaining to, for example, indoor climate in buildings. In today's building delivery process, this intention is often expressed explicitly and formally, for example when desirable indoor environmental conditions are specifically defined and expected to be maintained in the course of building operation. Provision of desirable occupancy conditions, or in other words, maintaining a sufficient degree of "habitability", may be thus seen as the main objective of building activity and an important concern of building physics. In human ecological parlance, habitability is afforded by proper fine-tuning of the ecological valency of the built environment. The environmental discourse naturally emerges in the ASBP from the realization that habitability can be secured sustainably only if it is achieved with a minimum on energy use, resource depletion, and negative environmental impact. In this context, Panzhauser, in tandem with many other building physics professionals around the world, worked on developing standards and associated calculation methods for description and declaration of the energy performance of buildings. A compact and "user-friendly" formulation of associated guidelines was one of his main contributions as a long-term leading participant in Austrian and European standardization bodies. However, he was also one of the first proponents of a more integrative (multidisciplinary and holistic) assessment of building performance. In this context, it is worth mentioning that Panzhauser was responsible for one of the early formulations of a total performance assessment scheme involving an eco-point assignment method with respect to multiple performance criteria. This benchmarking scheme encompassed three broad categories (energy, context, and health) with a corresponding set of sub-categories including very diverse items such as building enclosure, heating and cooling systems, infrastructure, water, biodiversity, ventilation, insolation, daylight, acoustics, air quality, and barrier-free design (Panzhauser 2000). Evaluating the habitability degree of the built environment can benefit from a second pair of concepts in human ecology. This pair of concepts concerns distinct aspects that can be attributed to the relationships between people and their surroundings. Thereby a high-level distinction is made between the material-energetic and information-related aspects of these relationships (Knötig 1992a, Mahdavi 1996a). These two aspects can be assigned to every entity, state, and process. The material-energetic aspect refers to the assumption that nothing exists unless some amount of matter or energy is involved. The information-related aspect refers to the assumption that matter and energy have a certain distribution in space and time, which can be represented in terms of a structure. An information content can be correlated with this structure. The relevance of this pair of concepts in the context of the ASBP and human ecology becomes clear if we ask ourselves how the habitability of the built environment is to be measured. An important component of this judgment lies in people's subjective experiences and attitudes. We can argue that such subjective evaluation processes of the built environment involve both the material-energetic and the information-related aspects of the relationships between inhabitants and the built environment. 6 A common approach to "operationalize" such evaluation processes in planning and operating involves the use of "psycho-physical" scales. The idea is that exposure to various levels of physical (material-energetic) stimuli translates – in a more or less predictable way – into corresponding subjective experiences. For example, exposure to increasing levels of sound intensity are said to translate into an experience of increased loudness and associated stress (annoyance). But psychophysical scales have been shown to be debatable. It would be highly problematic to postulate a deterministic relationship between measurable environmental factors and occupants' evaluation of environmental conditions (Mahdavi 2011a, 1996a, 1996b). In a nutshell, it appears that human evaluation processes are generally easier to describe and predict in exposure situations dominated by the materialenergetic aspect of the environmental relationships. In extreme cases of high-intensity exposure, the necessity for protective regulations is self-evident due to the obvious health hazards for the involved individuals (e.g., irreversible physical damage to the organism). It is thus not surprising that most efforts toward predicting the outcome of human evaluation processes have focused on the identification of a measurable material-energetic scale (such as sound pressure level) to which subjective judgments (such as the degree of annoyance) are expected to correlate. However, Knötig and Panzhauser consistently emphasized the impact of internal information processing on the degree of expressed dissatisfaction associated with various energetic levels of exposure. The information-related aspect of the environmental relationships plays indeed an important role in people's evaluation of exposure situation. This has been demonstrated in many experimental studies in psycho-acoustics (Mahdavi 2011a). For example, in an experiment conducted originally by Manfred Haider (Knötig 1983) and later repeated by the author (Mahdavi 2004), two groups of participants (university students) provided significantly different assessments of the same acoustical event (recorded white noise). Participants in the first group, who were told the recording was of a waterfall, judged it much more favorably than the second group, who was told the recording was of a factory. People's attitude toward the alleged source of an acoustical event clearly influenced their evaluation of the exposure, despite the absence of any objective difference in the nature of the event (Mahdavi 2011a). In another experiment (Schönpflug 1981), participants were exposed to white noise of different intensity while performing certain tasks (time estimations). The participants who received positive feedback about their performance ranked the same acoustical exposure as less annoying than those who received negative feedback. But the feedback messages were manipulated and did not reflect the true performance. Hence, their effect on participants' subjective evaluation of the noise exposure situation cannot be explained in terms of an acoustically induced impairment. From the human ecological point of view, the explanation lies rather in the nature of the information processing that was triggered by the combined effect of acoustical exposure and negative feedback. The degree of annoyance due to noise was apparently higher, once it was identified as the reason for one's (alleged) failure. Human ecological thinking, as adapted and promoted by Knötig and Panzhauser, not only provided original insights toward a deeper understanding of the "sick buildings" syndrome (Knötig et al. 1987), but can be perhaps viewed as intellectual predecessor of later inquiries in the domain of adaptive thermal comfort theory (Humphreys and Nicol 1998). While the efforts to ground the adaptive approach to thermal comfort on a solid scientific basis (Nicol et al. 2012) are not entirely convincing, insights from associated studies are highly valuable. They suggest that multi-level complexities involve in human processes of environmental sensation, perception, and evaluation cannot be captured with simple heat-balance based models of thermal comfort. Future advances may more successfully explain the combined physiological, psychological, and cognitive underpinnings of thermal comfort evaluation processes. Meanwhile, however, building design and operation professionals must be sensitive to the considerable variance in thermal responses of people to similar thermal conditions. The human ecological framework deployed by Knötig and Panzhauser provides a conceptual high-level (and qualitative) perspective toward the understanding of such evaluative variances in terms of the importance of not only the material-energetic aspect but also the information-related aspect of the environmental perception phenomena. CONCLUDING REMARKS Within the brief framework of the present treatment only a highly condensed impression of the work and impact of the ASBP and Erich Panzhauser could be presented. But it is not an exaggeration to identify Panzhauser as the central intellectual instance behind the ASBP. Nor is it an exaggeration to characterize his work toward establishing and promotion of ASBP as visionary. The ideas, concepts, and insights of the ASBP not only accompany progressive professionals in the building physics domain today, but point to the challenges and requirements of the future. These ideas and insights can be summarized as follows: 7 • the importance of the conversation between scientific-technical problem solving approaches on the one side and creative design thinking on the other side; • the critical necessity for multi-disciplinary, interdisciplinary, and integrative approaches, • the important practical role of adequate standards and benchmarking systems as well as accountability in building products and systems specification, • the life-cycle view of the building delivery and operation process, • the consistent development and deployment of evaluation criteria pertaining to the environmental sustainability of buildings and building products, and • the central human ecological importance of information processing phenomena in the evaluation of built environments. We, at the Department of Building Physics and Building Ecology at the Vienna University of Technology, would like to think that our ongoing research and teaching activities continue the intellectual tradition and professional spirit of the ASBP into the 21th century. But we are not alone. Through Panzhauser's prolific academic and professional work in the last half century, numerous architects, engineers, consultants, teachers, researchers, as well as professionals in standardization bodies and municipalities in Austria and beyond have been inspired by the ASBP's vision of a building science with a human face. Knötig H. 1992a. Human Ecology - The exact science of the interrelationships between Homo sapiens and the outside world surrounding this living and thinking being. The sixth meeting of the Society for Human Ecology "Human Ecology: Crossing Boundaries". Snowbird, Utah, USA. Knötig H. 1992b. Some essentials of the Vienna School of Human Ecology. Proceedings of the 1992 Birmingham Symposium; Austrian and Britisch Efforts in Human Ecology. Archivum Oecologiae Hominis. Vienna, Austria. Knötig H., Kurz I., and Panzhauser E. 1987. "Sick Buildings" – A phenomenon of internal information processing? Proceedings of the 4th International Conference on Indoor Air Quality and Climate (Indoor Air '87). Vol. 2, pp. 497 – 504. ISBN 3-89254-034-9. Mahdavi A. 2011a. The human dimension of building performance simulation. Keynote: Building Simulation 2011 - IBPSA 2011, Sydney, Australien; (Soebarto V, Bennetts H, Bannister P, Thomas PC, Leach D.: Editors). ISBN: 978-0-646-56510-1. pp. K16 - K33. Mahdavi A. 2011b. People in building performance simulation. In: Building performance simulation for design and operation. Spon Press. ISBN13: 978-0-415-47414-6. Mahdavi A. 2004. Reflections on computational building models. Building and Environment. Volume 39, Issue 8, August 2004. ISSN 0360-1323. pp. 913 – 925. Mahdavi A. 1996a. Approaches to Noise Control: A Human Ecological Perspective. Proceedings of the NOISE-CON 96 (The 1996 National Conference on Noise Control Engineering). Bellevue, WA, USA. pp. 649 – 654. Mahdavi A. 1996b. Human Ecological Reflections on the Architecture of the "Well-tempered Environment". In Proceedings of the 1996 International Symposium of CIB W67. Vienna, Austria. pp. 11 - 22. Mahdavi A. 1996c. A Human Ecological View of "Traditional" Architecture. Human Ecology Review (HER). Volume 3, Number 1. pp. 108 - 114. REFERENCES Ertl H., Fail A., Panzhauser E., Boisits R., Heiduk E., and Mahdavi, A. 1987. Natürliche Schachtslüftungsanlagen; Messung – Beurteilung – Planung. Archivum Oecologiae Hominis. Mahdavi A. 1989. Traditionelle Bauweisen in wissenschaftlicher Sicht. Bauforum. Vol. 132. pp. 34 - 40. Nicol F., Humphreys M., and Roaf S. 2012. Adaptive thermal comfort. Routledge. ISBN: 978-0-415-69159-8. Gibson J. 1979. The Ecological Approach to Visual Perception, ISBN 0-89859-959-8. Panzhauser E. 2000. Bauökologische Qualität von Gebäuden – Beispiele zur bauökologischen Deklaration – Strukturierung und Bewertung der humanökologischen Bauqualität. Report 2000-07. TU-Wien. Gibson J. 1977. The Theory of Affordances. In Perceiving, Acting, and Knowing, Eds. Robert Shaw and John Bransford, ISBN 0-470-99014-7. Panzhauser E., Fail A., Haider M., Heindl W., Knötig H., and Mahdavi A. 1991. Planung der (konventionellen) Fensterlüftung. Archivum Oecologiae Hominis. Haider M., Heindl W., Knötig H., Mahdavi A., Panzhauser E., et al. 1991. Bedeutung von Pufferräumen. Archivum Oecologiae Hominis. Panzhauser E., Fail A., Ertl H., Bednar et al. 1987. Formaldehydbelastung in österreichischen Wohnungen. Archivum Oecologiae Hominis. Haider M., Jansen P., Kafka-Lützow A., Knötig H., and Panzhauser, E. 1987. Vorzugstemperatur und IRReflexion von Bauteilinnenflächen. Archivum Oecologiae Hominis. Panzhauser E., Hariri M. R., and Gsandtner A. 1980. Betriebswirtschaftliche Optimierung des Wärmeschutzes raumumschließender Bauteile. Laboratorium für Hochbau, Technische Universität Wien. Heindl W., Krec K., Panzhauser E., and Sigmund A. 1987. Wärmebrücken: Grundlagen, einfache Formeln, Wärmeverluste, Kondsation, 100 durchgerechnete Baudetails. Springer. Schönpflug W. 1981. Acht Gründe für die Lästigkeit von Schallen und die Lautheitsregel. From Akustik zwischen Physik und Psychologie. In SCHICK, A. (Ed.), Akustik zwischen Physik und Psychologie. Stuttgart (Klett-Cotta). Humphreys, M.A. and Nicol, J.F. 1998. Understanding the adaptive approach to thermal comfort. ASHRAE Transactions 104:1. pp. 991 – 1004. Uexküll J. 1920. Kompositionslehre der Natur. (Edited by Thure von Uexküll). Frankfurt am Main. Knötig H. 2003. Personal communication. 8 The energy turnaround in Europe and its consequences for renewable generation, energy infrastructure and end-use G. Brauner Vienna University of Technology, Institute for Energy Systems and Electrical Drives, Vienna, Austria ABSTRACT: The energy turnaround in Europe represents a transition process, by which fossil resources are replaced by renewable energy. As the hydropower potentials are to a large extend exploited in Europe, wind and photovoltaic represent the technologies with the highest growing rates in the future. As the potential of these renewable resources is limited, efficiency in the end-use becomes in future a precondition for a successful turnaround. The sectors of buildings, households and mobility represent the highest efficiency potential. Energy active settlements with integration of renewable energy sources, local demand side management and storage capacities will form the backbone of transition. Replacement of combustion engines in cars by electric power train will bring improvement of efficiency by 70% and substitution of fossil fuel by renewable energy. The battery of cars can also be used as flexible energy storage in Smart Grid or Micro Grid management systems either grid to car or also to a smaller extend car to grid.  Providing balancing energy for the fluctuating renewable generation from more flexible thermal power stations or from pumped hydro power stations or local battery storages in adequate capacity  Reduction of end-use demand by efficiency measures, especially in the sectors of buildings and mobility  Development of technologies for energy efficient buildings with low energy demand  Integration of renewable energy sources especially photovoltaic in buildings  Development of Smart Grid and Micro Grid and “intelligent energy” technologies. This necessitates interdisciplinary research and development under ecological and economical aspects. 1 INTRODUCTION Europe’s energy strategy tends to low carbon and high renewable contend in energy. In history the Kyoto Protocol and low carbon dioxide dominated the energy strategy until about the year 2005. To reach the emission targets a broad assessment reaching from carbon capture and storage to new gas fired power stations with higher efficiency and lower emissions was initiated. Today renewable generation and efficiency in the end-use form the main targets. This is laid down in the package for climate protection SEC(2008) 85/3, giving for each member state of the EU individual target values for emission reduction and renewable energy content in the end-use of energy. The so called “20-20-20 until 2020” target aims to reduce in the mean of all EU member states the emissions by 20%, increase renewable energy content to 20% and improve end-use efficiency, related to the case without measures, by 20%. In Europe historically and according to diverse preconditions, the member states have different intentions in their energy strategy:  Keeping nuclear as main energy source  using coal from local resource for energy  Full turnaround to renewable energy In the long range until the year 2050 the overall target will be a portion of 80% of renewable energy in the end-use and will result in a synchronization of the strategies of EU member states. The transition process from predominant fossil to mainly renewable is linked to some preconditions:  Local potential of renewable energy such as wind, photovoltaic, hydropower  Extension of the energy infrastructures for integration of renewable energy sources 2 SHORT AND LONG TERM STRATEGIES The energy turnaround is characterized mainly by replacing fossil energy by renewable energy sources and thus avoiding CO2 emissions. In history hydro power formed the main source of renewable energy. As the hydro potential in Europe is limited, in future wind and photovoltaic will be the main sources. Wind has a big potential up to about 300 to 500 GW in the EU. PV will have the same potential. Wind generation is connected mainly to the transmission grid and necessitates extensions for higher grid capacities in the existing grid or of building a new super grid in Europe with extra high voltages of 800kV of 1.000 kV. Photovoltaic will in future be mainly situated on the roof or the facade of buildings, thus avoiding land use. PV will therefore mainly have influence on the low or medium voltage distribution grid. 9 Here mainly the integration of PV in buildings will be considered as this paper is focused on the decentralized aspects of the energy turnaround. 2.2 Long term strategies In the long term range until 2050 the renewable energy sources can significantly exceed the peak load of the grid. In this case the provision of storage capacities is necessary. But long term storage capcities are very expensive, so in future it will be better to cut-off surplus generation instead of providing these capacities. In Austria in a research project it was found, that for long term storing of a surlus of about 8% of the annual demand, the existing storage capacities would have to be extendet by the factor 100. This seems to be uneconomic and is also limited by the resources available for this purpose. In the long range it will thus be necessary to include also decentralized energy balancing concepts in the electrical energy system. At the one hand by increasing the efficiency in end-use, the deman can be reduced. At the other hand the by local demand side management (DSM), the energy demand can be better adopted tor the generating charachteristic of the renewable energy sources. In Figure 3 the change in energy demand in an energy active settlements from today to the year 2050 is shown: 2.1 Short term strategies In the short term range until 2020 PV and wind energy will grow in its generating capacities. A first mark is, if the installed power of both reaches the peak load in the grid. This can result in partly fully displacement the fossil operated power stations by fluctuating renewable energy. Figure 1. Solar and wind power in relation to peak load in European Member states (2020).  Thermal insulation of buildings will reduce the demand for heating.  New efficient appliances and illumination will reduche the electricity demand.  The fossil operated car will be replaced by an electrical vehicle, which can be charged by PV and has the ability to feed a limited amount of energy back to the grid (car to grid, C2G). As shown in Figure 1 according to the National Renewable Action Plan [NREAP] in two European countries, Germany and Spain by high installation rates this could happen until 2020. To guarantee security of supply the backup power stations have to be operated more flexible than in history. The number of start-ups and shut-downs will nearly be doubled and the power stations have to be able to bring high gradients of power for shut down e.g. in the morning during sunrise and start-up in the afternoon till sunset. As the energy demand in the private sector represents about 30% of the overall demand of a country and the demand for mobility a further 30%, by using electrical cars and heat pumps for heating about 60% of the fossil energy demand can be replaced by renewable electricity from wind, photovoltaic and hydropower [Power Vision]. Figure 2. Residual load balancing. As shown in Figure 2 the difference between load and renewble generation , which is called residual load, has to be provided from storage capacities or from flexible power stations. 10 Table 1. Efficiency potential in households [Ghaemi 2011]. Number of persons demand demand efficiency per household today efficient potential kWh/a kWh/a % 1 2,830 1,840 -23 2 3,580 2,612 -27 3 5,750 3,240 -44 4 5,820 3,700 -36 In general, most significant potentials in energy savings are in the sectors of building and mobility.  Thermal insulation and heat pump application with renewable electricity can bring down the demand to about 10% of today.  Replacing fossil operated cars by electrical vehicles can improve the power train efficiency by 70% and substitute fossil energy by renewable electricity mainly from wind and photovoltaic. Figure 3. Reduction of energy demand in households until 2050. Thermal insulation of buildings, heat pumps, efficient end-use, and electrical cars can reduce the energy demand in to 30% related to a conventional household of today. 3 CONCLUSION Buildings and mobility are the sectors with highest efficiency potential in the end-use of energy. In the long term run, here with renewable generation the energy turnaround can take place. Still some research and technological development is necessary to reach this target. As the rate of refurbishment is under realistic aspects only some percent per year, the energy turnaround will need some decades to be successful. Figure 4. Development of electricity demand until 2050. 4 REFERENCES The electricity demand will be increased in household to about 150% until 2050, as a consequence of the energy turnaround (Fig. 4). Ghaemi S. 2011. Efficiency potential in private sector in ADRES. Thesis, Vienna University of Technology 2011. Beurskens L.W.M., Hekkenberg M. 2011 Renewable Energy Projections as Published in the National Renewable Energy Action Plans of the European Member States. Document ECN-E-10-069 European Environment Agency, 1 February 2011. NREAP: National Renewable Action Plan towards 2020 targets. Power Vision 2012. Power Vision 2050, EUREL Study, 2012. SEC(2008) 85/3 Package of the Implementation means for the EU’s objective on climate change and renewable energy for 2020. In Figures 3 and 4 for the year 2050 an improvement of efficiency in end-use according to an investigation in Austria is assumed. By a questionnaire the potential was investigated by Ghaemi [Ghaemi 2011], as shown in table 1. About 70% of all households consist today on only one or two persons. Here smaller flats and newer appliances are used, so the efficiency potential is relatively small and between 23 to 27%. Households with children have a higher efficiency potential, because of lower income and thus using elder appliances. 11 Intelligent Transport Systems Why should building physicists, architects and urban designers care? H. Knoflacher Vienna University of Technology, Institute of Transportation, Research Center of Transport Planning and Traffic Engineering, Vienna, Austria ABSTRACT: Limited resources of energy for mobility have forced the society to develop compact multifunctional and liveable cities for hundreds of years. Cheap and easy available fossil fuel (coal and oil) have created first the industrial city around public transport modes and finally agglomerations, the city for the car. The former solar driven intelligent transport system was replaced by unintelligent individual optimized mechanical transport system consuming huge amount of energy and space and producing environmental, social and economic problems in the city. For the future it will be necessary that not only transport engineers and urban planners should develop the urban structures, but also building physicists and architects understanding the system behaviour. Life of Great American Cities: Random Houses, New York). Energy for the transport system was solar, human and animal body energy, the use of water and wind power. The city had for a long time a very strict noise control during the night, but had a vibrant daily life. (There were some exceptions, when Roman emperors organized good transport at night time like in Milet.) Artists and architects were employed to design famous buildings for the empress and rich people since about 2.500 years within the constraints of limited resources of energy and material. Under these circumstances intelligent urban planning was necessary to survive. Two questions suggest itself: − Is the intelligent transport system the precondition for an intelligent City? − Is it possible to organize an intelligent transport system in an unintelligent city? 1 INTRODUCTION Most of the time of 10.000 years of urban development the professions of building physicists, architects and urban designer didn’t exist. Nevertheless the societies in different parts of the globe were able to develop a kind of sustainable settlements, villages and cities which still exist till today. (Mumford, 1968, The City in History: Mariner Books). These building structures are not only functional well designed, they are also very often beautiful. They can maintain not only their basic functions till today when modern technical innovations like sanitary facilities, electricity and electronical equipment are incorporated, they have also proved their flexible adaptation to changing social and economic situations over hundreds of years. They are intelligent structures, since they can react on new challenges with problem-solving opportunities. They provide not only a high quality of life but also opportunities for all varieties of trades and workshops as well as culture. These are the „cities of short distances“, the socalled compact city, the dream of urban planners of the end of the 20th and beginning of the 21st century – but still an ideal for modern educated professionals. This compact city covers all necessary functions of a society on the smallest amount of space. This is only possible due to the multifunctionality of most of the buildings. The ground floor is used for trade, workshops, the second floor for living or offices and most of the cities have not more than four or five floors and a population density of up to 500 people per hectare like some inner districts of the city of Vienna. Multifunctionality is not only restricted to the building, the public space was always multifunctional, a place for trade, meeting, moving, transport of goods, cultural activities, social contacts, education by learning, playground, just to mention some of its main functions (Jacobs Jane, 1961, the Death and 2 INTELLIGENT TRANSPORT SYSTEMS (ITS) OF TODAY… …have nothing in mind with the structures of a city. It seems they are totally independent from any built structures if we follow the Wikipedia and EUdefinition of ITS: „Intelligent transport systems (ITS) are advanced applications which, without embodying intelligence as such, aim to provide innovative services relating to different modes of transport and traffic management and enable various users to be better informed and make safer, more coordinated, and 'smarter' use of transport networks. (Wikipedia, The free encyclopedia). Although ITS may refer to all modes of transport, EU Directive 2010/40/EU of 7 July 2010 on the framework for the deployment of intelligent transport systems in the field of road transport and for interfaces with other modes of transport defines 13 and tried to integrate the car into the new urban concept. Housing was separated from working, shopping and leisure, connected with wide roads and fast transport systems. His followers tried to optimize these principles without understanding the inbuilt wisdom of the historical city, the compactness and multifunctionality. The extrapolation of Le Corbusier’s idea was finally the system of detached one family houses – urban sprawl - social isolation and car dependency. Suburbia was born and the exodus from the city centres began. „The city centre is there, where no there is there“ is a common saying among urban planners in the United States. The European urban structures were a little bit more resistant against this damage of the urban functions in the second half of the last century. Rail based public transport, especially trams on the surface were an obstacle for the increasing demand for space for cars and many cities made the mistake to remove them. Today we have the opposite direction, since planners have realized that rail based public transport on the surface stabilizes the urban structure and the urban economy, while cars are undermining both. Car traffic, road noise, air pollution and accidents in the public space segregated the society and finally damaged local shops and workshops. Cars are the precondition for the priority of big international corporations above democratic municipalities and even big cities. ITS as systems in which information and communication technologies are applied in the field of road transport, including infrastructure, vehicles and users, and in traffic management and mobility management, as well as for interfaces with other modes of transport.“ (Directive 2010/40/EU oft the European Parliament and oft the Council, Official Journal oft he European Union 7. July 2010) Why should building physicists, architects and urban designer care, if this definition is valid and useful? 2.1 The industrial city Steam engine used for agriculture production and the railway removed the former constraints of urban development extremely fast as never before in human history. Urban growth and mass production on the one side, land exodus, urban proletariat and social problems were the flipside of this coin. Cities were not built anymore around the needs of the people, but around the railway or tramlines. Exergy driven transport systems shaped the cities more and more. Compactness was not necessary everywhere, nor multifunctionality of buildings. Public space became increasingly dominated by technical transport modes with high speeds. In the 19th century the „binding force“ of the compact city, the pedestrian, was partly replaced by public transport systems like trams and railways extending the daily travel distance over the former city borders. Former suburbs were annexed to the city and became districts, attached together with the new public transport system. New kinds of buildings were necessary to provide the society with this kind of technical modes. Elevated railways, underground tunnels, rails in the road, stations had to be built and maintained and became a part of the organism of the city. The intervention of engineers into the social, economic and cultural structure of cities and environment accelerated since then, supported by cheap and easy available energy from fossil sources, coal and later oil. The hierarchy of values changed. Technical transport systems became priority over human needs. 2.3 The professional problem: individual optimization Each profession tried to optimize the projects within their professional domain (border). Architects optimize the buildings, transport planners optimize the transport system, land use and urban planners provide the space for the needs of the society. In general the relationships and interrelationships are not taken into the consideration and the sum of the single optima doesn’t guarantee a system optimum. There is nowhere such a thing like an isolated building, isolated settlement, village, city, etc. Everything is connected to the environment, whether it is natural or artificial like everything which is manmade. The transformation of the transport system from human and animal movements and good transport based on pure solar energy into a fossil fuel driven technical system was so fast that none of the professions was able to understand the effects on the society, economy, culture and environment (Knoflacher, 1987, Verkehrsplanung für den Menschen, Orac Verlag). The individual benefits of these wonderful new tools for mobility were so predominant that the professions removed all obstacles for free movement of cars and provided a perfect environment for car users. Architects are not building anymore houses for people, urban planners not cities 2.2 Urban sprawl (Bruegmann Robert, 2005, Sprawl, A compact History; University of Chicago Press) and Agglomerations Transport infrastructure is not an issue of architects; it is much more the business of civil engineering. The invention of motorized individual vehicles and its use as a mass transport system overrun the society, the municipalities, the politicians and the professionals, first in the United States and then in the rest of the globe, especially in Europe. Architects like Le Corbusier were enthusiastic about the new opportunities of this wonderful individual transport mode 14 pression of human intelligence. Urban sprawl and the car-oriented city is the built expression of stupidity of all professions engaged in this business. Since the demand for stupidity is limitless, this behavior of professionals in the urban land use and engineering field must create conflicts with the limited resources of the globe. In contradiction to the individual experience there is no travel-time-saving in the transport system possible. for the urban society of people, but for a society of car users. The building code was changed following the principles of the §2 of the Reichsgaragenordnung 1939: „For each apartment, workshop, shop or whatever … parking space for the existing and future expected number of cars has to be provided on the ground or in the immediate neighbourhood.“ This paragraph was not seen as a problem for building physicists, architects and urban planners so far. Each discipline tried to do the best, building physicists optimized buildings, architects the shape and design of houses and urban planners provided the public space for cars. The building code and guidelines provide carriageways and not public space anymore. Traffic engineers optimize traffic flow and believe on ITS. The real behaviour of man and the system was not understood. 3.3 Modal choice is always restricted Finally the freedom of modal choice is restricted, alone by financial restrictions but much more by the inner structure of the people. People are selfish, intelligent and lazy. If we provide them with a car park close to their home, they cannot escape from the car, because the car compensates not only body energy but provides the car driver with opportunities he can never get as a normal pedestrian. With less than half of the body energy of a slow pedestrian, the car driver can reach a speed which is ten to twenty times faster than the walking speed. The car slips into the centre of his brain and controls his thinking and his value system. This was discovered in 1975 by the author. Under the prevailing circumstances people cannot escape from the car, because the binding force between the man and the car is energetic, physical and stronger than to the city or even to the children. Since we know now the behaviour of man in this artificial environment and the system behaviour we can now solve the problems. 3 LACK OF UNDERSTANDING SYSTEMS IS A LACK OF EXPERIENCE Urban and transport planners have lost their historical wisdom collected over 10.000 years and couldn’t replace it by sound scientific based facts. Instead they believe on three myths (Knoflacher, 2012, Grundlagen der Verkehrs- und Siedlungsplanung, Bd.2 Siedlungsplanung, Böhlau): 1 Growth of mobility, more cars more mobility. 2 Time saving with increasing speed. The faster the transport system, the more time saving for the society is possible. 3 Freedom of modal choice. More kind of vehicles, more modal choice. 4 THE SOLUTION 3.1 There is no growth of mobility The solution is a reorganization of parking. (Knoflacher, 2013, Zurück zur Mobilität, Überreuter) − Car parks have to be separated totally from all human activities. − The walking distance to and from the cars must be longer than the walking distance to the public transport stop. Otherwise people have no chance of choice and we cannot solve the environment, social and economic problems of the city. − When people are walking along the buildings, money is on the street and if money is on the street shop owners will open on the ground floor. − If the carriageway doesn’t exist anymore as a deadly environment for children and nonmotorized and barriers of parked cars along the sidewalks are removed, public space in settlements and cities can be used for economic, cultural, personal purposes for recreation and meeting place or can even be used as built-up areas in wide roads. − If cars are removed from human activities the travel speed is reduced substantially and if travel Terms are important: If we use wrong terms, we create wrong structures. Mobility is not the movement of vehicles on carriageways, people are mobile to compensate the deficits from the origin at the destinations. Each trip has a purpose and the number of purposes is not dependent on car ownership. People have to go to work, to school, shopping, recreations, social contacts and business, whether they have a car or not. People have to compensate the deficits of urban planners with an increasing demand for mobility. There is no such a thing like „Growth of mobility“, if we define mobility in the scientific sound way, related to a purpose. 3.2 Purposeless city, the base of purposeless transport and urban planning Travel time saving doesn’t happen in the transport or urban system. If travel speed increases, the distance increases and changes the urban structures. The low speed of pedestrians was and is the secret of the compactness of the historical city. It is the built ex15 − − − − − per square meters and number of purposes per joule. There is a very simple ranking between the transport modes. For speed up to 10 km/h pedestrians and cyclists have priority over all mechanical modes. For speed faster than 10 km/h public transport systems have to be preferred – without too many compromises. From the point of energy efficiency, public transport is about ten times less efficient than pedestrians and cyclists. Cars are at least 100 times less efficient than non-motorized traffic. Based on this fundamental indicators it is clear that an ITS system comprises pedestrians and cyclists as the dominant mode in all settlements and urban areas. These modes must have undisturbed, uninterrupted access to the stops of public transport. No private car traffic can penetrate into these efficient parts of urban structures. Cars have to be parked outside of the city, outside of the village. Car traffic is only allowed for transport of goods, for handicapped people, emergency trips and of course outside of built-up areas – if necessary – for private car use. If we continue to park the cars close to human activities of all kinds no intelligent transport system is possible any more. It is the physical structure, which prevents the intelligent system behaviour. The precondition to implement an ITS system is a car-free brain of experts, politicians and administrators. As long as the car is controlling the human thinking, the human value system, the thinking of architects and urban planners who provide life space for cars and destroy life space for people and the living organism of the city for the benefits of the car, it is useless to discuss the problem of ITS. Now we can answer the questions: • It is not possible to install an intelligent transport system in an unintelligent planned urban structure. • An intelligent Transport System is dependent on an intelligent urban structure and vice versa. Intelligent transport systems can only be developed in cooperation between architects, urban and transport planners, building physicists, if they understand the real human behaviour and the systembehaviour of Transport and City. speed is reduced substantially the destination must come close to the origins. The multifunctional intelligent city can be created only if these principles are consequently applied. If the structures are changed in such a way, the need for car use is reduced substantially. 80 - 95 % of car trips of today are not necessary anymore under these structural conditions. If public transport stops are accessible every 500 meters (Vienna has an average distance between public transport stops between 270 and 290 meters) and parking opportunities are one kilometre from each other (maximum walking distance to the parked and from the parked car is about 500 meters) more than 90 % of car trips will not happen. Pedestrians, cyclists and public transport will become the environmental and sustainable main transport modes for cities and settlements in which this kind of policy is implemented. This kind of cities is much less vulnerable, social stability will increase, local economy will flourish and environmental problems will disappear. 4.1 Intelligent traffic systems A traffic system cannot be more intelligent than the structure for which it serves. If a building has a wrong location, a wrong function, if architects provide car parks close to every human activity, an intelligent traffic system is not possible anymore. Intelligent systems are solving and not creating problems. All the existing attempts for which the society, the European Union, the ministries of the national states are spending billions of Euros for ITS (intelligent transport system) by enhancing the speed, reducing congestion, removing bottlenecks, building new transport infrastructure, improving the information system is nothing else than wasting money to continue the inbuilt stupidity of the whole system. This can only happen, when experts, politicians and the administrations have not understood the real human and system behaviour. Traffic flow whether it is on the road or on the rail is only a symptom and not the cause of the visible problem. This approach can therefore not solve the problems, it is an useless measure. The organization of the origins and destinations of trips and the location of parking places, public transport stops in such a way, that intelligent modes like pedestrians and cyclists get the best opportunities for a safe and free mobility is a precondition of an intelligent transport system. An intelligent transport system captures travel time within the body of a city and is not losing travel time over the urban or municipality borders by enhancing the speed. An ITS system is slow. It doesn’t need very much exergy from fossil or other sources, an intelligent transport system is efficient in number of purposes 5 REFERENCES Bruegmann R. 2005. Sprawl, A compact History. University of Chicago Press. Directive 2010/40/EU of the European Parliament and of the Council 7. July 2010. Official Journal of the European Union. Jane J. 1961. The Death and Life of Great American Cities: Random Houses, New York. Knoflacher H. 2012. Grundlagen der Verkehrs- und Siedlungsplanung, Bd. 2 Siedlungsplanung. Böhlau. Knoflacher H. 1987. Verkehrsplanung für den Menschen, Orac Verlag. Knoflacher H. 2013. Zurück zur Mobilität, Überreuter. Mumford L. 1968. The City in History: Mariner Books. Wikipedia, The free encyclopedia. 16 2. Conference papers Recent advances in SEMERGY: A semantically enriched optimization environment for performance-guided building design and refurbishment U. Pont, F. Shayeganfar, N. Ghiassi, M. Taheri, C. Sustr, A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria S. Fenz, J. Heurix, A. Anjomshoaa, T. Neubauer, A.M. Tjoa Vienna University of Technology, Institute of Software Technology and Interactive Systems, Vienna, Austria ABSTRACT: The SEMERGY project explores the utility of semantic web technologies toward populating the required input data for performance based building design optimization. A major portion of the required information (e.g. material properties, climatic data and standards, etc.) exists on the internet in extensive, yet ill-structured form. To facilitate a structured access to this data for different calculation modules the following steps were taken: i) Development of a comprehensive building data model for the internal flow of information. ii) Creation of an ontology of linked building product data. iii) Development of a rule-based system to automatically identify valid construction alternatives for building components. iv) Development of an optimization procedure with multiple criteria (energy use, environmental impact, and investment costs). v) Creation of a web-based Graphical User Interface to enable entry of data and user interaction. The present contribution provides an overview of the progress made in the above mentioned domains. Such information includes buildings' geometry, building components' technical properties, occupants' presence and actions, micro-climatic data, internal conditions, building systems, etc. Conventional methods toward collecting such information are time-consuming and error-prone (Ghiassi et al. 2012, Maile et al. 2007). Although much of this required data is freely available on the World Wide Web, the use of this data is rendered difficult by a lack of structure and proper formatting. Moreover, performance guided decision making in building design is hampered by the complexity of the problem, a consequence of financial, environmental, technical, and legal factors. Identification of different design alternatives that are code compliant and are within the limits of the available technological and financial resources is no trivial task. Evaluation of these alternatives against various criteria (investment and operation costs, environmental foot print and energy performance), requires an extensive amount of time and expertise. Nonetheless, such a manual optimization attempt may prove inconclusive and fail to capture the full scope of the available potentials. Semi-automated optimization is, however, not commonly integrated in performance assessment tools and used mainly in advanced research projects. In this context, SEMERGY is an ongoing research and development project, intended to explore developmental opportunities toward effective optimization environments for comparative assessment of alternative building design and retrofit options in view of construction materials and building products. Such options are to be benchmarked according to their functional, 1 INTRODUCTION Buildings are identified as major emitters of greenhouse gasses and are responsible for a large share of the global energy use. Efforts to reduce the energy demand of buildings include the tightening of building regulations toward higher thermal efficiency of building envelope, thermal retrofit of buildings, implementation of more effective building technologies for HVAC (Heating, Ventilation, AirConditioning), etc. Designers as well as clients find it increasingly difficult to cope with the complexity involved in generating high quality buildings. Paired with the common economic and time constraints of the AEC-field, planning and decision making becomes a major challenge for all stakeholders. The EU countries have implemented standard calculation methods for certification of buildings in view of their energy performance. A framework for different assessment methods was suggested in EN 13790 (ASI 2008), ranging from very simple steadystate calculation schemes to dynamic simulation. However, despite the availability of a wide range of performance assessment tools, performance computing has not been sufficiently integrated in the design process. The use of such tools is commonly limited to the final stages of the design process (Hensen et al. 2004), when the most crucial decisions concerning the future performance of the building are already made (Domeshek et al. 1994). One of the major barriers towards integration of such methods in the earlier design stages is the cumbersome and costly task of data accumulation and entry of the required information. These computational procedures require various detailed information on the building and its context. 19 evaluation of candidate design options supported by multi‐objective decision support methods (Mahdavi et al. 2012b). ecological, and economical performance (Mahdavi et al. 2012a). The present paper offers a general introduction to the project and reports on the recent progresses. Previous advances and in-depth description of certain aspects of the projects can be found in Mahdavi et al. 2012a, 2012b, Shayeganfar et al. 2013a, 2013b, Ghiassi et al. 2012, 2013, and Hammerberg 2013. 2.2.3 The reasoning interface The reasoning interface involves the assessment methods and calculation procedures that the model is to undergo. Given that different design scenarios are to be rated based on their functional, ecological, and economical performance, the reasoning interface needs to accommodate cost estimation, life cycle analysis, and other performance assessment procedures. SEMERGY involves both simple (e.g. normative) performance calculation procedures and advanced numeric analysis and simulation applications on the back ground, and is coupled with an elaborate optimization engine to identify, evaluate, and rank different design options (Mahdavi et al. 2012b). 2 OVERVIEW OF STRUCTURE AND WORKFLOW IN THE SEMERGY PROJECT 2.1 Target users SEMERGY targets three different potential user groups: i. Novice users with little or no knowledge of the building sector (building products, laws, applicable subsidies, etc.). For this user group a convenient guidance through the data entry process is necessary. ii. Architects and building designers who demand a highly flexible and rapidly adaptable environment capable of generating adequate responses to problems in the planning process. As such, data transfer via known formats (CAD, BIM) is of great importance to this group. iii. Municipalities, developers, and other authorities, interested in a toolbox for fast evaluation of buildings at a larger scale (neighborhood, town). 2.3 Workflow The initial design intention is conveyed by the user through the User Interface, either through direct geometry input (web based graphical user interface for novice users) or data import (via BIM or GIS). This data is structured in the form of a building data model. Based on the preferences of the user (e.g., available budget, desired performance level, and construction system), alternative design options regarding construction materials of various building components are identified by the semantic and the reasoning interfaces. The building model subsequently inherits the semantic information pertaining to various alternatives from the semantic interface and undergoes evaluations. These iterations continue until the optimal solution(s) have been detected in view of cost, performance, and environmental impact. Figure 1 provides a schematic depiction of this workflow. 2.2 Structure Different activities within the SEMERGY environment are organized under three different interfaces: the user interface, the reasoning interface and the semantic interface. The constant interaction and iterations between these interfaces result in the generation of the ultimate optimal solution set. 2.2.1 The user interface The User Interface supports the acquisition of data from the user and enables user interaction. The format of this interface and the interaction method is tailored to the specific needs and expertise of the aforementioned target groups. According to the use case, the data required for computation is to be provided by the user through a web‐based Graphical User Interface, extracted automatically from an advanced building design software (CAD, BIM), or derived from GIS data (Mahdavi et al. 2012b). 2.2.2 The semantic interface The key feature of the SEMERGY environment is the incorporation of semantic web technology toward efficient search for and compilation of input information required for comprehensive analysis and 20 Figure 1. SEMERGY workflow. potential for the SEMERGY project. Lack of a clear space-based structure and the work-load required for the extraction of the desired format hindered the adoption of the IFC representation for the SEMERGY project, despite its many advantages. The gbXML data model complies with various requirements of thermal performance modeling. However, the gbXML representation scheme is not extensible. Considering that SEMERGY is not merely a performance analysis environment and includes other calculation procedures (e.g., cost estimation and Life Cycle Analysis), the incapacity to integrate additional data, crucial to these procedures, renders gbXML inadequate for the internal storage of data in SEMERGY. Shared object model (SOM) is a hierarchically structured template (a class hierarchy in objectoriented programming terms) to capture the essential elements of a building and their properties, to the extent required by simulation applications for inquiries of a certain range of informational resolution (Mahdavi 2000, Mahdavi et al. 2002). Although the SOM does not include all the information necessary for the various calculation methods incorporated in the SEMERGY environment, its clear space-based and object oriented structure proved a reasonable template for the desired building representation scheme. SOM schema was therefore extended to meet the various requirements of SEMERGY in view of necessary input information, identified through the reverse engineering of several computational engines. 3 RECENT ACTIVITIES IN THE SEMERGY PROJECT Towards creation of the aforementioned optimization environment, the following activities have been (or are being) carried out. 3.1 Building information representation The transfer of data between the aforementioned interfaces of the SEMERGY environment is enabled by a building data model, which includes all necessary information for the integrated computational engines. Appropriate data is selected from this base model and subjected to various calculation methods for evaluation purposes. To arrive at a working model of the input data required for SEMERGY, we have conducted a reverse‐analysis of the informational requirements of both normative procedures (annual and monthly) and a detailed simulation application. Once the scope and format of the required information was determined, two standard building representation schemes, Industry Foundation Classes (Building Smart 2013) and Green Building XML (gbXML 2013), as well as the Shared Object Model developed for the SEMPER research and development project (Mahdavi 2000, Mahdavi et al. 2002) were extensively studied in view of their 21 The resulting building representation, SEMERGY Building Model (SBM) is a comprehensive spacebased building representation, tailored to meet the requirements of advanced energy simulation procedures, normative calculation routines, as well as life cycle analysis. Due to its object-oriented design, it can be extended in view of further functionalities with minimal effort. It enables efficient iterations within the optimization procedure and allows for simplification of data entry modalities to reduce user effort. The general structure of SBM is shown in figure 2. 3.2 Acquisition of user-dependent data For the first use case (novice user), a browser-based graphical user interface (GUI) for general data entry has been created. Mapping schemas for data import from gbXML format, as well as import modalities for two-dimensional CAD drawings have been developed to support the professional design community. For this use case, an extended version of the graphical user interface, with a higher degree of freedom is under development. Acquisition of appropriate information from GIS data and digital recreation of missing information (3D geometry) is to be addressed in the future phase of the project. The data entry modalities are described in more detail below. Figure 2. SEMERGY Building Model structure. 22 2) Automatic validity check of drawn floor plans (no loose ends) and identification of spaces (closed polygons). 3) Simple addition of windows and doors via drawing tools. 4) Assignment of space attributes by simple selection from a pre-defined list of space functions associated with detailed operational information on the background (infiltration and ventilation rates, temperature and humidity set points, etc.),(Figure 3). 5) Easy definition of building orientation and site obstructions by means of rotating the plan on a Google map underlay, generated according to building location (Figure 4). 6) Simple selection of integrated construction templates, which allow for identification of design alternatives. In the aforementioned step-by-step-routine, a set of simplifications was implemented to facilitate the entry of data by the user. For instance, the entry of the complex geometrical representation of roof was replaced by minimal user input such as roof type (gabled, hip, etc.), pitch angle and ridge direction. However, this minimal user input does not provide sufficient information depth for the building representation and further computational purposes. To close this gap a post-processing method was developed. This method statistically approximates missing information in user input (see Hammerberg et al. 2013). For operational data such as internal conditions and occupancy profiles, default templates based on standards were predefined for different room functions. This information may be edited or redefined by users in the professional use case. 3.2.1 SEMERGY Graphical User Interface The GUI for the first use case is intended for convenient and easy collection of user-dependent data. The development started after a number existing online as well as offline tools (e.g., Energy Globe 2012, Raiffeisen 2012, Archiphysik 2013, Energy Plus 2013) were examined in view of their data entry modalities and user friendliness. Some of these tools are intended for inexperienced users and incorporate very basic building data (gross area, volume, roof format, number of stories, construction year) with user friendly interaction methods. Others allow the entry of very specific information, via more complicated interfaces. As SEMERGY is a web-based decision-support environment, the GUI is browser-based. It is intended to be used as an easy-to-use, step-by-step interface, which collects all the data that can be provided by the laymen user. The main challenge was to reach a balance between ease of user interaction, accuracy and resolution of acquired data, and technical limits of browser based environments. The GUI is composed of three different parts: 1) General Information: location, date of construction, building usage, number of stories, roof form, heating system. 2) Geometry and operational information: floor plans, doors, and windows dimensions, wall types, space height, orientation, space functions. 3) Semantic information and material properties: selection of construction systems, determination of frameworks of optimization (user-defined constraints). Features of the GUI include: 1) Drawing of 2-dimensional (orthogonal) plans of building floors on a grid (with “snap”-functions), tool-tip keyboard entries, and possibility to copy perimeters or complete plans from previous floors. Figure 3. SEMERGY Graphical User Interface: Selection of room functions. 23 Figure 4. SEMERGY Graphical User Interface: Adjustment of the orientation of the building on Google maps. various sources. For instance price information gathered from a reseller website can be linked to information gathered from product manufacturer websites to create a complete profile of a certain product including its physical properties and actual market prices (which often differ from catalogue prices). Creation of parallel categorization systems and association of new properties to different groups of products is another capability of such data ontologies (Shayeganfar et al. 2013). 3.2.2 Data Import from BIM and CAD-Applications To facilitate the import of already existing geometry data from different design applications used by professional users (use cases 2 and 3), some common data exchange formats including gbXML and dwg were studied. For the use of gbXML data, a specialized parser was developed to extract relevant data from BIM software and to create a corresponding SBM-model. For 2D-CAD applications (still widely used in the AEC-field) a method has been developed to enable data transfer to SBM. This includes a set of guidelines to create or modify two dimensional CAD files (for instance assignment of specific layers to different building components), which allow the transfer with simple routines. 3.4 Optimization procedure In SEMERGY, the optimization procedure’s goal is to identify the optimal combinations of building material configurations (constructions) with respect to multiple criteria pertaining to heating demand, investment costs, and environmental footprint. Automatic generation of alternative constructions is facilitated by a rule-based logic. These rules for combination of materials according to their various properties are extracted based on a study of common construction systems for various building components. To enable the implementation of this rule-based logic, product descriptions extracted from the web and stored in the ontology had to be enriched with further attributes regarding their form, functions or possible position in a construction. This was rendered possible by attaching new property classes to the extracted product categories in the ontology. The logic implements the defined rules on the information stored in the product ontology to search for appropriate materials for different layers of a construction. Once alternative components are generated as potential candidates for the optimization problem, they are subjected to computation. Results are rated according to the above-mentioned criteria and the 3.3 Semantic data ontologies Supported by semantic web technologies (BernersLee et al. 2006, Shayeganfar et al. 2008), SEMERGY extracts building product information available in various web-based sources and restructures it to generate the SEMERGY building product Ontology. This enables the semi-automatic delivery of necessary material-related information to computational engines in the right format for assessment purposes. The product ontology is a hierarchical classification of building product information, which differs from static databases in that it preserves the links between the represented product information and manufacturer websites. As such, it is possible to automatically and effortlessly update the information or acquire new information from these websites on a regular basis as long as the source website retains its original structure. Moreover, such ontologies allow for establishment of links between data acquired from 24 Economy, Family and Youth (BMWFJ). The SEMERGY team includes, in addition to the authors, K. Hammerberg, V. Jain, D. Wolosiuk, and A. Wurm. best performing solution set is represented to the user (Heurix et al. 2013). 4 CONCLUSION 6 ABBREVIATIONS SEMERGY is a multi-objective optimization environment to support building designers and decision makers in selection of materials and building products towards design and construction of energy and cost efficient buildings. Simplification of data entry and automatic populating of required semantic information help reduce user effort and error, invested time, required expertise and resources. Semantic web technologies were utilized to create an ontology of linked building product information. This has facilitated the incorporation of real market products in the optimization procedure. Thus, the feasibility of the proposed solutions is guaranteed. Moreover, such an approach can promote the use of new and more efficient building products and technologies, which may be less familiar to the professional community, thereby shifting the focus of manufacturers from advertisement to quality improvement. In order to develop the full potential of such an environment, however, the collaboration of building product manufacturers is necessary. Coherence and consistency in information release formats and wellstructured and stable websites can enable effortless extraction of web-based data and regular updates. Development of a universal data format for the release of building product information, within the frameworks of semantic web movement, can help establish such collaboration, beneficial to all stakeholders. A format, which captures various aspects of building products (physical, economical and ecological), can allow for the retrieval of suitable information according to the desired field of inquiry. Such well-structured and web-enabled information may be linked and accumulated in dedicated ontologies with minimal effort. A similar strategy may be envisaged to structure data pertaining to building codes and regulations, tax exemptions, subsidies, and other funding opportunities. Provision of guidelines for restructuring of such data helps distribute the work load among numerous organizations and authorities, yet retain the integrity of the information, thereby facilitating its accumulation and utilization by various actors. 2D: 3D: AEC: BIM: CAD: gbXML: 2 Dimensional 3 Dimensional Architecture Engineering and Construction Building Information Modeling Computer Aided Design Green Building Extensible Markup Language GUI: Graphical User Interface HVAC: Heating, Ventilation, Air-Conditioning IFC: Industry Foundation Classes SBM: SEMERGY Building Model SOM: Shared Object Model 7 REFERENCES Archiphysik. 2013. [http://www.archiphysik.at/], Last visited February 2013. ASI (Austrian Standards Institute). 2008. ÖNORM EN ISO 13790: Energy performance of buildings - Calculation of energy use for space heating and coolin. Vienna:Austrian Standards plus GmbH Berners-Lee, T., Gödel, K., and Turing, A. 2006. Thinking on the Web. New Jersey: John Wiley & Sons, Inc. Building Smart. 2013. [http://www.buildingsmart-tech.org/], Last visited February 2013. Domeshek E.A., Kolodner J.L., and Zimring C.M. 1994. The design of tool kit for case-based design aids. In J.S. Gero, F. Sudweeks (ed.), Artificial Intelligence in Design’94: 109126. Alphen aan den Rijn: Kluwer Academic Publishers. Energy Plus. 2013. Energy Plus Energy Simulation Software. [http://apps1.eere.energy.gov/buildings/energyplus/], Last visited February 2013. Energy Globe. 2012. [http://www.energyglobe.com/], Last visited September 2012. gbXML. 2013. [http://www.gbxml.org/aboutgbxml.php], Last visited February 2013 Ghiassi N., Shayeganfar F., Pont U., Mahdavi A., Fenz A., Heurix A., Anjomshoaa A., Neubauer T., and Tjoa A.M. 2012. Improving the usability of energy simulation applications in processing common building performance inquiries. In O. Sikula, J. Hirs (ed), Simulace Budov a Techniky Prostredi. Brno: Ceska Technika - nakladatelstvi CVUT Ghiassi N., Shayeganfar F., Pont U., Mahdavi A., Fenz S., Heurix J. Anjomshoaa A. Neubauer T., and Tjoa A.M. 2013. A comprehensive building model for performanceguided decision support. CESBP 2013:Accepted Hammerberg K., Jain V., Ghiassi N., and Mahdavi A. 2013. Generalizing roof geometry from minimal user input for building performance simulation. CESBP 2013:Accepted Hensen J., Djunaedy E., Radošević M., and Yahiaoui A. 2004. Building performance simulation for better design: Some issues and solutions. Proceedings of the 21th conference on passive and low energy architecture. Eindhoven. Heurix J., Fenz S., Anjomshoaa A., Neubauer T., Tjoa A.M., Taheri M., Shayeganfar F., Pont U., Ghiassi N., Sustr C. , and Mahdavi A. 2013.Optimization in the SEMERGY 5 ACKNOWLEDGEMENTS The SEMERGY project is funded under the FFG Research Studio Austrian Program (grant No. 832012) by the Austrian Federal Ministry of 25 environment: Multi-objective optimization for sustanable building design and retrofit, CESBP2013: Accepted Mahdavi A. 2000. Supporting collaborative design via integrated building performance computing. Proceedings of the 12th International Conference on System Research, Informatics and Cybernetics: Advances in computer-based and WEB-based collaborative systems:91-102. BadenBaden Mahdavi A., Suter G., and Ries R. 2002. A presentation scheme for integrated building performance analysis. Proceedings of the 6th International Conference: Design and Decision Support Systems in Architecture: 301-316. Ellecom: DDSS. Mahdavi A., Pont U., Shayeganfar F., Ghiassi N., Anjomshoaa, A., Fenz S., Heurix J., Neubauer T., and Tjoa A.M. 2012a. SEMERGY: Semantic web technology support for comprehensive building design assessment. In G. Gudnason. R. Scherer. (ed), eWork and eBusiness in Architecture, Engineering and Construction:363-370. Reykjavík: Taylor&Francis. Mahdavi A., Pont U., Shayeganfar F., Ghiassi N., Anjomshoaa A., Fenz S., Heurix J., Neubauer T., and Tjoa A. M. 2012b. Exploring the utility of semantic web technology in building performance simulation. Proceedings of BauSIM 2012: Gebäudesimulation auf den Größenskalen Bauteil, Raum, Gebäude, Stadtquartier:58-64. Berlin: Universität der Künste Berlin. Maile T., Fischer M., and Bazjanac V. 2007. Building Energy Performance Simulation Tools, A Life-Cycle and Interoperable Perspective. Stanford: Stanford University, Center for Integrated Facility Engineering [http://cife.stanford.edu/sites/default/files/WP107.pdf] Last visited February 2013. Raiffeisen. 2012. Raiffeisen Bausparkasse. [http://wohnen.raiffeisen.at/Pages/Portal], Last visited September 2012. Shayeganfar F., Mahdavi A., Suter G., and Anjomshoaa A. 2008. Implementation of an IFD library using semantic web technologies: A case study, In A.S. Zarli, R. Scherer (ed), ECPPM 2008 eWork and eBusiness in Architecture, Engineering and Construction: 539-544. Sophia Antapolis: ECCPM. Shayeganfar F., Anjomshoaa A., Heurix J., Sustr C., Ghiassi N., Pont U., Fenz S., Neubauer T., Tjoa A.M., and Mahdavi A. 2013. An ontology-aided optimization approach to ecoefficient building design, ECCPM 2013: Accepted 26 Multi-objective optimization in the SEMERGY environment for sustainable building design and retrofit J. Heurix, S. Fenz, A. Anjomshoaa, T. Neubauer, A.M. Tjoa Vienna University of Technology, Institute of Software Technology and Interactive Systems, Vienna, Austria M. Taheri, F. Shayeganfar, U. Pont, N. Ghiassi, C. Sustr, A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: This paper reports on a specific effort (optimization procedure) within the on-going research and development project SEMERGY. This project is geared towards the development of a decision making environment for performance-guided building design and retrofit. The present contribution illustrates advances towards the creation of a multi-objective assessment-based optimization procedure for the selection of building products and materials in view of the ecological, economical, and energy performance of the building. SEMERGY associates the preferences and constraints of the user regarding construction systems, performance level, environmental foot-print and investment costs, with existing building products on the market. Supported by semantic web technologies, SEMERGY extracts building product information from various web-based sources and restructures them into an ontology of building products. The products are enriched by a set of additional properties which enable a rule-based automatic identification of valid construction alternatives for different building components. These alternatives are evaluated and benchmarked against various criteria, to determine the optimal solution(s). 1 INTRODUCTION Nowadays, the implementation of optimization in particular design or retrofit projects is becoming increasingly common. As such, developing an optimization platform to reach reliable and realistic results through an automated procedure, applicable to a vast group of problems, is highly demanded (Coffey 2008). In this regard, the SEMERGY project is an attempt towards the development of a performanceguided optimization environment, in which accumulation of input data required for performance assessment computations is supported by semantic web technologies (Mahdavi et al. 2012a, 2012b). In fact, the SEMERGY project aims at assisting decision-making by creating an optimization procedure which identifies a set of optimal solutions addressing the involved input parameters in view of various user defined objectives. Here the efficiency indicators are the heating demand, environmental impacts and investment costs which constitute the main objective functions in the multi-criteria optimization process. Considering a number of objective functions gives rise to consistent and realistic outcomes. The main goal is to identify optimal building product configurations (constructions) for various building elements pertaining to the heated building envelope. The optimal configuration is defined to be selected among a set of predefined alternatives. The pool of alternative constructions is automatically generated based on product attributes and a rule-based logic imbedded in SEMERGY. As such, the mentioned Due to severe limitations in the natural resources and time as well as financial restrictions, optimization problems are extensively involved in design processes. It is a very competitive and challenging topic in every engineering field to design a product that consumes the least while delivering the most. Basically the optimization problem emerges in terms of one or more mathematical objective functions whose variables are subjected to some restrictions. Treating an optimization problem deeply depends upon the complexity and shape of those functions within the domain of feasible solutions. In the building performance area, some expected demands in the design or retrofit phase are minimizing the building operation cost, energy consumption and environmental impact or maximizing occupants comfort. Therefore, the optimization problems are defined to fulfill these anticipations (Wetter 2008). In such optimization problems, usually a large number of variables are involved. This further complicates the assessment of the design for different sets of available options (Mahdavi & El-Bellahy 2005). Additionally, it has to be mentioned that traditional “manual” approaches of trying to find optimal solutions (mainly by trial-and-error) are timeconsuming, expensive and inconclusive. In other words, there is in general no guarantee to find optimal solutions in such attempts. 27 multi-objective optimization (MOO) problem is related to a large number of discrete variables. Therefore, choosing a stochastic optimization strategy seems reasonable. In the SEMERGY optimization toolbox, a Genetic algorithm among the family of evolutionary algorithms is constructed. Given the fact that the evaluation criteria are opposing (e.g., reducing the heating demand will most likely require an increase in the investment costs), optimal solutions are determined by Pareto dominance in SEMERGY. In this context, this contribution represents the effort toward creation of a performance-based optimization environment, considering three objectives, i.e., ecological, economical and energy performance of the buildings, for design and retrofit projects. whereas optimize can be either maximize or minimize, depending on whether the problem is a maximization or minimization problem. The solution vector x=[x1,x2,…,xI] is the vector of decision variables and X the set of feasible solutions, also called the solution space. For discrete variables, the MOO problem is called a multi-objective discrete optimization (MODO) problem. The so called point x x x x z = z , z , …, z 1 2 J =  f ( x) (2) represents the image of a solution x in the J objective space Z, such that () x = f x for j = 1, … , J j j z (3) Using the concept of Pareto-dominance, (in case of minimization) point z1 strictly dominates z2 if 2 OPTIMIZATIONS IN BUILDING PERFORMANCE ANALYSIS 1 2 < z for j = 1, … , J ; j j z Generally optimization is the process of finding optimal values for a set of user-defined parameters in order to minimize or maximize an objective function. In building performance analysis, usually the optimization is tasked with evaluation of a certain building in a single dimension like the total energy consumption indicator. The problem with this approach is that it captures a particular building’s performance from one view point only, though there are other factors that need to be considered at the same time. For instance, the knowledge of which insulation material is the best with respect to heating demand is practically useless if the price of the selected material exceeds the available budget. This example also demonstrates a particular problem with having multiple objectives for optimization; namely that these objectives are often opposing each other. Therefore the optimal solutions in these cases are trade-offs between different objectives. Identification of such holistic optimal solutions as a multiobjective optimization problem (MOO) is described below. (4) point z1 (loosely) dominates z2 if 1 2 ≤ z for j = 1, … , J j j z (5) and 1 2 < z for at least one j; j j z (6) if neither point dominates (strictly or loosely), they are called incomparable. A solution x∈X is called Pareto-optimal (efficient), if there is no x'∈X that dominates x. The image z∈Z of a Pareto-optimal solution is called non-dominated or non-inferior. The set of optimal solutions X*⊆X represents the Pareto (-optimal) set, whereas its image Z*⊆Z in the objective space represents the nondominated set or Pareto front. A Pareto front approximation is a set A of points, and their corresponding solutions are such that there are no z1, z2∈A that z1 dominates z2. That means the set A is totally composed of mutually non-dominated points. Pareto optimization deals with finding the Paretooptimal front or a set that represents a good approximation to that front. This technique is quite successful because there is no single-best solution (there are several solutions that represent different trade-offs between the objectives) and it is rather difficult to find a preferable ordering of the criteria. 3 MULTI-OBJECTIVE OPTIMIZATION The general multi-objective optimization (MOO) problem can be written as: (Jaszkiewicz 2004, Silva et al. 2004, Zitzler et al. 2004) Optimize { f1 {x} = z1 , f2 {x} = z2 , … , f j {x} = z j }(1) x∈X and z∈Z 28 4 EVOLUTIONARY ALGORITHMS 5 META- HEURISTICS-BASED MULTIOBJECTIVE OPTIMIZATION IN SEMERGY The straightforward approach to solve optimization problems is to iterate through all possible combinations of decision variables (solution vector). With rising numbers of decision variables and complexity of the valuation functions, this can quickly become computationally infeasible. Thus, instead of identifying the exact Pareto front, decision makers and analysts are usually satisfied with finding a reasonablyaccurate approximation of the Pareto front. This approximation can be determined by applying metaheuristics such as (population-based) evolutionary algorithms, or more specifically, Genetic Algorithms (Holland 1992). Evolutionary algorithms (Zitzler et al. 2004) mimic the generic natural evolutionary process of survival-of-the-fittest and are characterized by these three features: i.) A set of solutions is maintained; ii.) There is a mating selection process executed on the set of solutions; iii.) Selected solutions are combined to generate new solutions by genetic processes like crossover and mutations. In evolutionary algorithms, the solution candidates are called individuals, while the set of individuals is denoted as population. Each individual encodes its manifestations of the decision variables (solution vector) and is characterized by a certain fitness value, determined by the fitness function. The optimization process is executed iteratively and (in its simplest case) consists of the following steps: i.) Selection: Two individuals from the population are selected; ii.) Recombination: With a specified probability, the two individuals called parents are recombined and exchange sections of their solution vectors in a crossover operation. Furthermore, certain parts of the resulting new solution vector are changed randomly with a relatively low probability, denoted as mutation. Crossover is applied to preserve better performing code segments, while mutation ensures that the search is directed to unexplored areas of the search space. Crossover and mutation probabilities have to be carefully selected in order to prevent that the search is either too restricted or too open (resulting in simple random search); iii.) Evaluation: The new solutions called children are evaluated according to the fitness function and compared with their parents. If better results are produced, the parents are replaced with the children. In SEMERGY, the optimization procedure’s goal is to identify the optimal combinations of building material configurations (construction) with respect to certain objectives. Contrarily to simple functions like aggregations, the evaluation functions are more complex and involve different properties of the building products. These objectives include the following (all objectives are to be minimized): − Heating demand − Investment costs − Environmental footprint. The population consists of different combinations of building products for various elements (e.g., walls, roofs, floors, windows and doors) in a building in new design or retrofit cases. For example, in a refurbishment project, the individuals encode the different product possibilities for insulation of a particular external wall composition, as well as different window and door products. The optimization procedure’s main goal is to identify alternative building configurations, which are to be compared with the initial design encoded in the initial building model. Alternative Solutions are generated automatically by SEMERGY. Once alternative components are generated as potential candidates for the optimization problem, these candidates are rated according to the abovementioned criteria. For instance, available funds can be classified as limited resources, potentially with an absolute upper bound. Hence the optimization should be defined in a way to minimize the costs, while staying below the defined hard-limit. On the other hand, the efficient performance of building components in terms of thermal aspects is a benefit which has to be maximized. 6 DEFINING RULES FOR TREATING OPTIMIZATION PROBLEMS IN THE SEMERGY ONTOLOGY Supported by semantic web technologies, SEMERGY extracts building product information available in various web-based sources and restructures it to generate the SEMERGY building product ontology. This enables the semi-automatic delivery of necessary material-related information to computational engines, in the right format, for assessment purposes. 29 One of the web-based data repositories of building products, incorporated within the SEMERGY environment is Baubook (Baubook 2013). The Baubook database provides the ecological and physical properties of building products, required for computational purposes. It enables building product manufacturers to define their products by properties and physical attributes such as conductivity, specific heat, density, etc. and ecological indicators such as global warming potential (GWP), acidification potential (AP), embodied primary Energy (PEI) etc. In the SEMERGY project, a class hierarchy of building products is defined in the Resource Description Framework (RDF) format. A customized RDFizer component extracts Baubook product properties via a dedicated page parser and stores them in the ontology. In order to facilitate the automatic generation of alternative constructions, common construction systems for various building components are studied and certain rules are extracted for combination of materials, according to different predefined templates. These rules depend on various properties of materials which might not be present in available databases (e.g., form, function, etc.). To enable the implementation of this rule-based logic, product descriptions need to be enriched by further attributes regarding their form or functions. For instance, in a defined optimization problem with specified objectives, when searching for alternatives for the exterior layer of an external wall, a gypsum board, which is described as “sensitive to humidity” in the ontology, may not be selected. Thus, different product alternatives for different building components are identified within the ontology and subjected to the optimization process. For this purpose, a semantic data interface is incorporated within the SEMERGY environment, which functions as an interface between computational methods and the semantic repositories. This mediator, on the one hand, translates the constraints and requirements of the design into semantic queries in SPARQL format. On the other hand, it is responsible for mapping the extracted RDF data and query results to required fields in SEMERGY Building Model (SBM). In fact, by providing a set of SPARQL query components instead of static predefined queries, this semantic data interface enables the dynamic composition of the desired queries at execution time. For instance, the interface allows to query by type, class, property (or its value), as well as applying filters on SEMERGY-specific expressions. Therefore, the SPARQL interface of the prod- uct archive allows the users to apply wide ranges of constraints in the optimizations. In SEMERGY, the user specifies the basic information such as geometry and thermal properties of the building components as well as the available budget and/or desired building performance level. This platform has been established for both novice and expert users which will provide different levels of informational resolution for further calculations (Mahdavi et al. 2012b). In different SEMERGY usecases, the constraints can be divided into two main categories: i) the product type constraints, which are based on the SPARQL query including SEMERGY’s classes, and, ii) product property constraints, which consider threshold values for certain properties of products such as thermal conductivity, price, or thickness. The SPARQL queries in SEMERGY include a combination of the above mentioned constraints. For example, assume an optimization problem searching for wood-concrete wall elements with the following constraints: − Thickness under 30 cm − Thermal conductivity under 0.5 W.m-1.K-1. The semantic interface uses the query components pertaining to product type, thickness, and conductivity to create a suitable query. Listing 1 demonstrates the resulting SPARQL query to address the constraints. Figure 1 represents the results of this query and available product alternatives. SELECT ?name ?conductivity ?thickness WHERE { ?product a semergy:WoodConcreteWallElement. ?product rdfs:label ?name . ?product semergy:thickness ?thickness . FILTER (?thickness < 30) . ?product semergy:thermalConductivity ?conductivity . FILTER (?conductivity < 0.5) } Listing 1. Sample product query. Figure 1. SPARQL Query Results. For every building component with single or multiple layers, with different constraints regarding each 30 (e.g. external wall, roof, etc.), the user is asked to select a construction type from a given list of choices. These choices are different templates according to which actual constructions can be configured (Figure 3). layer, multiple alternative constructions can be generated. Combinations of these constructions for the entire building can be subjected to optimization against the defined objective functions. Evidently, this approach generates a large number of possible alternative candidates. Thus, by further filtering and applying other constraints such as decreasing the cost or environmental impacts, a manageable number of candidates can be achieved. In the SEMERGY optimization logic, a candidate is a particular combination of different layers constituting the construction of a particular building component, i.e., a particular possibility of how a building’s wall can be constructed. Each resulting solution of the optimization problem, includes specific candidates for each building component, i.e., external wall, internal wall, window, doors, etc. In order to make sure that the candidates are selected properly (e.g. a particular product combination is used as an external wall candidate instead of an internal wall or a window construction), the candidates are also organized into so-called candidate classes. These in turn influence the recombination logic of the optimization logic by defining the potential crossover positions to make sure that a wall candidate is actually interpreted correctly as wall candidate and not for instance as a window candidate. The logic’s goal is to evaluate these candidate combinations in terms of the pre-defined objective functions. Simply put, it checks how a particular combination of constructions (e.g. wall1, floor1, roof1, window1, door1, etc.), fares against another construction combination (e.g. wall2, floor1, roof3, etc.) with respect to the overall building heating demand, total investment costs, and the environmental footprint. Note that the constructions within two sets need not be mutually exclusive (e.g. a certain wall construction can be present in more than one alternative set). Finally, one or more combinations of constructions are proposed as optimal solutions. These offer suggestions for market-available materials and products for all different architectural elements within the building. These solutions lie within the limits of the user’s financial resources, are compliant with construction codes (with regard to thermal performance), and guarantee optimal energy and environmental performance. The example below (Figure 2) helps to illustrate the work-flow of the optimization procedure as implemented in the SEMERGY environment. The procedure begins with drawing the floor plan of the building in SEMERGY web-based GUI by means of pre-defined line types. Each line type represents a different construction. For each line type Figure 2. Example building. Figure 3. SEMERGY Graphical User Interface. Based on the user-selected template (e.g., massive construction), and according to different required characteristics pre-defined for each layer within this template, the semantic interface identifies all materials within the product ontology which meet these requirements. For instance the insulation layer of the external walls can assume all materials within the ontology which have the function "insulation", the format stiff, are appropriate for a "massive construction", and can be mounted on "the outer layer" of "external walls" (Table 1). As discussed before, this information is either included in the material profile of the original data base (e.g., conductivity), or has been attached to the product after its integration in the ontology (form, etc.). The layers, which have little influence on the thermal performance of the construction (e.g., finishing, vapor-barriers, etc.) are populated by default materials. The generated alternative constructions are then tested for their compliance with maximum U-value regulations. The valid alternatives constitute the gene pool in the optimization process. In this example only the external walls, windows and the roof are subjected to optimization. 31 Table 2. Resulting solution set. Solution U-Value U-Value No Wall Roof #1 0,195 0,19 #2 0,164 0,155 #3 0,213 0,20 #4 0,189 0,19 #5 0,142 0,19 #6 0,176 0,19 #7 0,131 0,16 #8 0,224 0,20 Table 1. Alternative insulation layers, as identified by SEMERGY, and their thermal (Conductivity) and ecological (PEI, GWP, AP) properties. Product Name Material Conductivity PEI GWP AP [MJequi. [CO2equi. [SO2equi . kg-1] kg-1] .kg-1] 0,041 102 3,45 0,0223 0,050 23,30 1,64 0,0105 0,045 15,70 0,84 0,0022 0,046 14,70 0,04 0,0000 0,040 102 3,44 0,0800 [W.m-1.K-1] InsulationProduct 1 EPS Insulation- Mineralw Product 2 ool InsulationProduct 3 Foamglass Insulation- Sheep- Product 4 wool InsulationProduct 5 … XPS … … … … U-Value – Windows 1,0 1,0 1,0 1,0 1,0 1,0 0,8 1,0 Table 3. Set of Solutions – grey dominated solutions (extracted from the results). SoHeating OI3 Cost dominated / lutio non domiDemand n nated No [kWh.m-2.a-1] [-] [€] #1 59,60 47,25 180.000 #2 51,51 53,22 200.000 #3 61,17 93,85 220.000 dominated by #1 and #2 #4 58,77 42,57 195.000 #5 46,95 47,90 250.000 #6 55,20 57,25 230.000 dominated by #2 #7 33,20 85,25 279.000 #8 67,33 60,52 195.000 dominated by #1 and #4 … These constructions are then combined to form complete design solution packages including potential constructions for all different building components. These solutions are subjected to optimization iterations until the optimal solution set is derived. Tables 2 shows the optimization individuals and Table 3 the resulting construction configurations. These configurations are represented here only by the Uvalues of the building components and the resulting performance indicators (Heating Demand, compare OIB 2011, OI3-Index, compare IBO 2006, and cost). Yet, each of these potential solutions are associated with detailed descriptions of the real world products incorporated in their design. Note that, for the purposes of the present illustrative example, the cost information (rough estimations) was explicitly entered in the system. Ideally, the cost information should also be extracted from the web and attached to different materials already present in the ontology. The resulting solutions all comply with thermal construction codes, fulfill the desired energy and environmental performance criteria, and lie within the limits of the user’s financial resources. None of these solutions is dominated by any other. In other words, if one solution performs better in one domain (cost, environmental impact or energy use) than another solution, it is dominated by the second solution in another domain. As seen in Table 3, solutions which failed to meet the criteria are excluded from the final suggested set. 7 CONCLUSION The research conducted in the present paper is aimed at helping the designers to find their desired products and optimal constructions which meet their predefined preferences in both building design and retrofit projects. The SEMERGY project offers an optimization platform to identify optimal alternatives for different building components through an automated procedure. The available design’s efficiency indicators in the SEMERGY performance-based optimization environment are heating demand, environmental impact, and investment cost. Therefore, the optimization problem in SEMERGY is defined as a Pareto-based multi-objective optimization problem in order to properly take these different criteria into consideration. In the optimization process, the optimal configuration is selected among a set of automatically-generated alternatives, which are established relying on semantic knowledge based on product attributes and rules pre-defined in the SEMERGY ontology. To manage the large number of discrete variables in the mentioned multiobjective optimization, genetic algorithms among the family of evolutionary algorithm are applied. 32 OIB 2011. OIB Richtlinie 6 – Energieeinsparung und Wärmeschutz, Ausgabe Oktober 2011 http://www.oib.or.at/RL6_061011.pdf last visited March 2013 Silva J.D.L., Burke E.K., and Petrovic S. 2004, An introduction of multiobjective metaheuristics for scheduling and timetabling. In Metaheuristics for Multiobjective Optimisation, SpringerHolland J., 1992, Genetic algorithms, Scientific American, p.66-72 Srinivas N., and Deb K. 1994, Multiobjective Optimization Using Nondominated Sorting in Genetic Algorithms, Evolutionary Computation 2 (3), p.221-248 Wetter M., 2001. GenOpt- A Generic Optimization Program. Berkeley, USA, U.S. Department of Energy. Zitzler E., Laumanns M., and Thiele L. 2001, SPEA2: Improving the strength pareto evolutionary algorithm for multiobjective optimization, Evolutionary Methods for Design Optimization and Control with Applications to Industrial Problems, p. 95-100 As a result, the optimization process implemented in the SEMERGY project provides a set of best solutions, each including proper combinations of constructions for various building components, which meet the user-specified settings and lead to optimum overall performance of the building, within the limits of the user-defined constraints. Since these constructions are composed using products and materials existing within legitimate web-based data-bases, they are commercially available to the user. Thus, the user is not required to establish an association between abstract values and real world products (as is the case with most performance-oriented optimization attempts). 8 ACKNOWLEDGEMENT The SEMERGY project is funded under the FFG Research Studio Austrian Program (grant No. 832012) by the Austrian Federal Ministry of Economy, Family and Youth (BMWFJ). In addition to the authors, the SEMERGY team includes: K. Hammerberg, V. Jain, D. Wolosiuk, and A. Wurm. 9 REFERENCES Baubook 2012. http://www.baubook.at/ , last visited March 2013 Cheret P. 2010, Baukonstruktion. Handbuch und Planungshilfe.,Dom Publishers, ISBN-13: 978-3869220338, p. 45. Coffey B., 2008. A Development and Testing Framework for Simulation-Based Supervisory Control with Application to Optimal Zone Temperature Ramping Demand Response Using a Modified Genetic Algorithm. Montreal, Canada: Concordia University. Holland J. 1992, Genetic algorithms, Scientific American, p.66-72 IBO 2006. Der OI3-Index: Die quantitative ökologische Gebäudeoptimierung mit dem vom IBOentwickelten OI3Index hält Einzug in die Wohnbauförderungen; http://www.ibo.at/documents/OI3index.pdf last visited March 2013 Mahdavi A., El-Bellahy S. 2005. Effort and effectiveness considerations in computational design evaluation: a case study. Building and Environment, 40 (2005), pp. 1651 - 1664. Mahdavi A., Pont U., Shayeganfar F., Ghiassi N., Anjomshoaa A., Fenz S., Heurix J., Neubauer T., Tjoa A.M. 2012a. SEMERGY: Semantic web technology support for comprehensive building design assessment. In G. Gudnason. R. Scherer. (ed), eWork and eBusiness in Architecture, Engineering and Construction:363-370. Reykjavík: Taylor&Francis. Mahdavi A., Pont U., Shayeganfar F., Ghiassi N., Anjomshoaa A., Fenz S., Heurix J., Neubauer T., Tjoa A.M. 2012b. Exploring the utility of semantic web technology in building performance simulation. Proceedings of BauSIM 2012: Gebäudesimulation auf den Größenskalen Bauteil, Raum, Gebäude, Stadtquartier:58-64. Berlin: Universität der Künste Berlin. 33 34 A comprehensive building model for performance-guided decision support N. Ghiassi, F. Shayeganfar, U. Pont, A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria J. Heurix, S. Fenz, A. Anjomshoaa, A.M. Tjoa Vienna University of Technology, Institute of Software Technology and Interactive Systems, Vienna, Austria ABSTRACT: This paper reports on a specific effort within the ongoing research and development project SEMERGY. This project is geared toward the development of a performance-guided decision making environment for building design and retrofit. A multi-objective optimization environment requires a dynamic representation of the building, which includes all necessary data for the underlying calculation engines (e.g., energy demand calculation, cost estimation, Life Cycle Analysis). This paper explores the capacities and deficiencies of some existing building data models (IFC and GBXML), within the context of SEMERGY project and reports on the development of "SEMERGY Building Model" (SBM). SBM is a comprehensive building representation, tailored to meet the requirements of advanced energy simulation procedures and normative calculation routines for common inquiries. It can be extended for further functionalities with minimal effort, enables efficient optimization iterations and allows for data entry simplification to reduce user effort and error. can interpret calculation results and draw comprehensible conclusions. SEMERGY (Mahdavi et al. 2012a, 2012b, Pont et al. 2013), represents an attempt towards development of a performance guided building optimization environment. Its aim is to overcome usability issues of available tools for designers and novice users by simplifying the accumulation and entry of data for assessment purposes, as well as to ameliorate system intelligence to analyze assessment results and propose clear design solutions. The SEMERGY environment is intended to support users in selection of appropriate building products and construction materials from the large variety of products available on the market, in order to achieve the best energy and environmental performance possible within the limits of the user’s financial means. It targets three different user groups: i. Novice users with little or no knowledge of the building sector. ii. Architects and building designers who demand a highly flexible and rapidly adaptable environment capable of generating adequate responses to problems in the planning process. iii. Municipalities, developers, and other authorities, interested in a toolbox for fast evaluation of buildings at a larger scale. According to the expertise of the user, the initial design intentions, as well as desired performance level and available funds, are communicated to SEMERGY through a web-based Graphical User Interface (novice users), extracted automatically from an advanced building information model (CAD, BIM), or derived from GIS data (Mahdavi et al. 2012a). 1 INTRODUCTION Despite the advances in the development of building performance assessment tools over the past decades, the uptake of such tools within the design community has been relatively slow. Their integration in the design process is limited to the late stages, mainly to analyze the final design solution for certification purposes (Hensen et al. 2004). Little attention has been paid to their potential to support the generation of design alternatives, to make informed choices between different design options, or to optimize building and/or systems (De Wilde 2004). There is evidence that decisions taken during conceptual design have a disproportionate impact on the final building energy performance (Domeshek et al. 1994) and that the cost, time, and effort needed for implementing changes early in the design process are substantially lower (McGraw Hill 2007). Nonetheless, assessment tools are rarely used to support early design phase tasks such as feasibility studies and conceptual design evaluations. The low application of performance assessment tools has been in part attributed to low market interests and high time-cost of applying them. On the other hand, studies show that the number of tools developed in the past decade to address specific needs and preferences of architects, who are the main decision makers in the early design stages, is very limited compared to those developed for experts in the field of building physics (Attia 2011). Architects' main criteria for selection of a performance assessment tool are usability and intelligence of the software (Attia 2011). In other words, they seek an easy-to-use application, which 35 2 DEVELOPMENT OF THE BUILDING MODEL Keeping the geometry unaltered, SEMERGY automatically identifies various legitimate design alternatives with regard to material combinations and construction configurations. Through optimization iterations, optimal solutions for all various building components are selected and suggested to the user. SEMERGY uses different computational engines to evaluate the proposed design in view of multiple criteria of investment costs, energy demand and environmental impact. It currently incorporates both simple normative procedures (steady-state monthly and annual calculation methods based on standards and guide-lines) (e.g. Austrian Standards Institute 2009, OIB 2007) and advanced simulation engines (e.g. Energy Plus 2013) to arrive at the value of the pertinent performance indicators. The functionalities of SEMERGY will be further extended by integration of other calculation procedures such as life cycle analysis and cost estimation. To reduce user effort, SEMERGY explores the potential of the semantic web technologies (BernersLee et al. 2001, Shayeganfar et al. 2008) towards populating the input data for calculation purposes via the navigation of the extensive but currently illstructured web-based information space pertaining to building materials, elements, components, and systems, as well as microclimatic, financial and legal information. Such information is extracted from the web and restructured into ontologies of semantic data, which facilitate queries and imposition of rule based logics to help automatically retrieve relevant information for computation purposes. In-depth description of the optimization procedure and the incorporated semantic technology can be found in Shayeganfar et al. 2013, and Heurix et al. 2013. To enable the transfer of data between different components of the SEMERGY environment (User Interface, computational engines, semantic interface), building related information is structured in the form of a building data model, which includes all required data for different calculation engines integrated in SEMERGY. Appropriate data is then selected from this base model and dispatched to various calculation engines for evaluation purposes. The present paper explores various features an appropriate data representation in the context of the SEMERGY environment, points out some of the short-comings of standard building representations (Industry Foundation Classes and Green Building XML) for the purpose of this project, and reports on the development of the SEMERGY Building Model (a.k.a. SBM). Prior to any attempt towards development of a building model for the SEMERGY environment, the scope and format of the required input information for the embedded calculation procedures had to be determined. For this purpose, the selected computational engines (Energy Plus and normative calculation routines) were studied to identify essential input information. In a second step, two standard building representations (Industry Foundation Classes and Green Building XML), as well as a third building data model (Shared Object Model) developed for the SEMPER project (Mahdavi 2000), were explored in detail in view of their potential for the SEMERGY project. Finally, adopting SOM as the underlying schema, the necessary input information was structured into the SEMERGY Building Model. 2.1 Structure and content of Input information for SEMERGY In order to better grasp the advantages and disadvantages of each of the available building representation schemes, the fundamental requirements of a suitable building model for the SEMERGY environment are briefly discussed, below. 2.1.1 Space-based, extensible Structure Performance assessment is a key functionality of the SEMERGY environment. Due to the nature of the calculations, performance assessment applications require a space-based view of the building in which different building components are represented not just as juxtaposed material entities, but by their relationship to various spaces. A study of the informational requirements of a large class of behavioral analysis applications (e.g. Energy, light and sound simulation) clearly shows their need for a rich description of space, containing geometric, topologic, and semantic information on bounding enclosure components, orientation in the overall building and site context, and adjacency relationships to other spaces (Mahdavi et al. 2002). In a space-based representation, building components are viewed as enclosure elements of a certain space and the borders between this space and others. Such a relational understanding of building components and spaces is crucial for the implementation of heat transfer and airflow algorithms embedded within the performance assessment tools. The concept of space-based thermal modeling applies to all various sorts of performance assessment applications. Building representations suited to more elaborate calculation engines such as dynamic simulation tools have a more complex 36 routines, in view their of required input information, is the resolution of the information. topological formation and higher resolution than those developed for steady-state normative methods which are typically less finely grained, yet the principles remain the same. In other calculation routines such as cost estimation and Life Cycle Analysis, the notion of space, is of lesser importance to the calculation procedure and the definition of spaces can be ignored as long as the required properties of the building elements are properly provided. In addition to having a space-based structure, a suitable building representation for SEMERGY has to include information required for the various incorporated calculation engines. Since SEMERGY is intended to incorporate more calculation procedures in the future, extensibility and flexibility of the building model are crucial to the project. 2.2 Study of some existing building representation schemes 2.2.1 Industry Foundation Classes The Industry Foundation Classes (IFC) data model for building and construction industry data is meant to universally support data exchange in the building industry. It is an object-based data model promoted by building SMART (International home of openBIM) to facilitate interoperability in the architecture, engineering and construction (AEC) industry, and is a commonly used format for Building Information Modeling (Building Smart 2013). However, the IFC building representation has not been fully endorsed by the developers of building performance assessments tools: certain requirements of energy models are not fully accommodated by IFC. In contrast to the described space based vision, IFC views a building as a composition of different building components such as walls, ceilings, roofs and floors. These are identified as "architectural" objects expressed as material entities, which together constitute the building. Although spaces are identified in the IFC model and space boundaries are geometrically expressed, a clear relationship between spaces and their composing building elements is not established. Such a representation may well accommodate structural and constructionrelated inquiries but does not fully meet the requirements for performance inquiries. Starting from the IFC-compliant geometric representation of spaces and their boundaries, it is possible to establish the topologic links required for a performance model. This requires, however, substantial systematic post-processing to identify bounding elements of spaces, allocate building components accordingly and assign appropriate adjacency status and semantic information to them (Bazjanac 2010). There have been various efforts towards the implementation of IFC models in performance assessment tasks (Bazjanac 2010, 2008, Granlund 2013). During the optimization procedure in the SEMERGY environment, numerous models need to be generated and evaluated through multiple procedures (performance, life cycle analysis, and cost evaluation). Unnecessary calculation load (e.g. intricate data reformatting) can severely jeopardise the performance of the system. Therefore, despite its advantages, we did not adopt the IFC model for performance assessment in the SEMERGY environment. This was mainly due to our requirements in view of appropriate space-based structure and the work-load required for the 2.1.2 Scope of required input information To determine the scope of the required data for the calculation procedures, for the first phase of the project, the performance assessment modules of the SEMERGY reasoning interface (Energy Plus and OIB calculation routines) were considered. Four different data categories were identified (Ghiassi et al. 2012): 1) Calculation parameters: This data category is fundamental to the running of the programs and generating results but not part of the representation of the building. Such data include weather data, simulation parameters, output settings, calculation methods and algorithms used by the applications. For the purpose of common inquiries, default data sets for such values can be stored in external repositories and linked to the building model. 2) Physical Data: This category holds the most essential data packages, which constitute the physical aspects of the building model and are therefore crucial to the calculations. Physical data includes two main types: Geometry (site description and location, building geometry) and Semantics (construction and material properties). 3) Operational Data: This category is dedicated to information associated with the building usage and systems. Internal conditions, occupancy, lighting, equipment, set points and schedules, air flows, and HVAC settings are examples of such data. 4) Optional data: The last category includes data which is required only for advanced calculations and research purposes (e.g. Phase change materials and intricate shading calculations). Since SEMERGY is tailored for common performance inquiries, this category was not considered relevant in the building representation. An appropriate building model for SEMERGY has to include physical and operational information and provide links to the case-appropriate calculation parameters. The main difference between the simple (normative) and elaborate(simulation-based) 37 extraction of the desired format. It remains however, a valid possibility for data import from various CAD applications. 2.3 Green Building XML The Green Building XML schema, referred to as "gbXML", was developed to facilitate the transfer of building information stored in CAD building information models, enabling integrated interoperability between building design models and a wide variety of engineering analysis tools and models currently available. Today, gbXML has the industry support and wide adoption by the leading CAD vendors. With the development of export and import capabilities in several major engineering modeling tools, gbXML has become a defacto industry standard schema (gbXML 2013). In contrast to the IFC representation, gbXML has experienced a large uptake by performance assessment software developers (e.g. DOE-2, EcoDesigner, EDSL-TAS, Ecotect, eQuest, etc.). This is mainly due to the appropriate structure of gbXML, which follows the afore-mentioned spacebased concept. The gbXML schema includes all necessary information for common performance inquiries and requires no reformatting, which makes it to a certain extent suitable for SEMERGY. However, the main disadvantage of gbXML is its rigid structure. Even though, XML (the underlying object based data structure of gbXML) is extensible, gbXML itself is not. If any additional information (e.g. space or material properties) are stored in a gbXML file, applications seize to recognize it as a valid file format. Consequently, some information potentially available within the original BIM environment may be lost due to an insufficient prevision of placeholders when transformed into the gbXML format. Considering that SEMERGY is not merely a performance analysis environment and includes other calculation procedures (e.g. cost estimation and LCA), the incapacity to integrate the additional data crucial to these procedures renders gbXML inadequate for the internal storage of data in SEMERGY. Nevertheless, gbXML is an ideal data import solution due to its large uptake by the CAD industry and its compatibility with performance applications. Figure 1. Structure of the Shared Object Model for SEMPER (Mahdavi et al. 2002). SOM is a hierarchically structured template (a class hierarchy in object-oriented programming terms) to capture the essential elements of a building and their properties, to the extent required by simulation applications for inquiries of a certain range of informational resolution. However, the shared model itself does not contain the entire building information; rather, it contains a tightly structured notation of constitutive building elements, with pointers to the detailed information on such elements in the data repository for the persistent storage of such information. While the Shared Object Model may allow retrieving all the necessary building geometry, material, and context information that such applications require, it is not sufficient on its own for a building performance simulation application to function. In the SEMPER environment, for each disciplinary domain (calculation method), a Domain Object Model had to be generated upon filtration and modification of information in the shared model according to the specific view of the building in that domain. Domain specific entities may have to be added to what is inherited from the shared model (Mahdavi 2002) (Fig. 1).Although the Shared Object Model does not include all the information necessary for the various calculation methods incorporated in the SEMERGY environment, its clear space-based and object oriented structure provides a reasonable template for the desired building representation scheme. 2.4 Shared Object Model The Shared Object Model was developed for the SEMPER project, a prototypical computational environment for integrated building performance modeling (Mahdavi 2000). 38 A short description of the main classes in SBM and their corresponding attributes and property sets is given in Table 1 and Table 2. 2.5 SEMERGY Building Model In order to arrive at an appropriate building data model for the SEMERGY Project, SOM schema was adopted as a basis and extended to meet the various requirements of the SEMERGY environment in view of necessary input information. Figure 2 illustrates the hierarchical structure of SBM. As seen in the diagram, all crucial data categories (physical and operational data and calculation parameters) are accounted for in the resulting building representation. The calculation parameter objects do not hold all calculation related data. Rather, they establish links to the external repositories of calculation and weather data templates based on clues provided in the physical data (e.g. building type, location, construction system, etc.). Geometrical data is either directly provided by the user in the right format (e.g. imported from BIM) or generated based on simple user input (two dimensional drawings and room heights in the case of the novice user and GIS data in the case of municipalities and governmental authorities) (Mahdavi et al. 2012a, 2012b). The operational data entry can be facilitated for novice users through provision of pre-defined data templates for all various objects. To further simplify the task, an additional space property “Space Function” has been foreseen, which operates as a short-cut to help automatically select the suitable predefined data-set for “Furniture”, “Required Internal Condition”, “Equipment”, “Occupant” and “Lighting”. A comprehensive ontology of building products, supported by semantic web technologies and structured in accordance with the requirements of SBM, helps populate the semantic portion of the physical data. SBM has been designed to store all relevant data for different integrated calculation modules in the most general format (e.g. Geometry of all enclosures and apertures is stored by coordinate points in an absolute coordinate system). However, not all stored building information is relevant in every calculation module. For instance, normative calculation methods require data on enclosures and apertures, which separate the heated volume from unheated space. In such cases, following a rule-based logic (e.g. based on adjacency relations), the appropriate portion of the data is automatically selected and dispatched to the calculation module. Sometimes, aggregate values are required to reduce the resolution of the data for simpler methods (e.g. U-value instead of detailed material properties).These aggregate values are automatically calculated and stored in the building model prior to the running of assessment and optimization procedures to reduce the impromptu (on-the-fly) calculation load. 39 Figure 2. SEMERGY Building Model schema. 40 Table 1. Definition of various classes in SBM: Geometry. Class Definition Site Site is the top-level physical entity including properties such as site topography and data-sets location and features (natural or man-made elements on the site). One or more Buildings are associated with a Site. Features Natural (e.g. Trees) or man-made (e.g. Walls) detached shading elements on the Site defined by coordinates Location Longitude, latitude and elevation of the Site. One weather data template for each calculation method is associated with location. Building Section Building is a logical unit, part of a Site and decomposed into Sections. HVAC and Zone objects may be associated with a Building. One calculation Setting template for each calculation method is associated with the Building. Building also includes properties such as year of construction and usage which help determine default values (e.g. construction sets) A Section is a grouping of spaces. Each Section is decomposed into one or more spaces. Zone An additional grouping scheme for spaces. Zone object accounts for user-defined thermal zoning. This class is currently not used as each space is considered an independent thermal zone. Space A Space is a volume enclosed by a collection of Enclosures forming a polyhedron. A space can have operational data-sets such as Furniture, Occupant, Equipment, Lighting, and Required Internal Condition. It has auto-calculable properties such as Volume and Floor area. Enclosure Table 2. Definition of various classes in SBM: Semantics. Class Definition Construction A Construction is (directly or indirectly) composed of one or more Layers and can be of Types Enclosure/Partition Construction, Door Construction and Window Construction. The Window Construction object is decomposed into Frame and Glazing. Physical boundaries of a space. Enclosures can be of three types: Wall, Roof/Ceiling or Floor. Each Enclosure is defined by its coordinates and adjacency status, has a Construction and additional autocalculable properties such as area. One or more Apertures may be associated with an Enclosure. Partition A Space can be further divided by Partitions (e.g. Dividing walls, raised floors, etc.) while remaining a single zone. A Partition is defined by its coordinates and has a Construction. Aperture An opening in an Enclosure. It can be of type Door or Window/Glazed Door. The latter can be associated with Fixed or Adjustable Shading elements. Simple Shading Attached shading elements (e.g. Fins or Overhangs) Adjustable Shading Shading elements such as Shades, Blinds, etc. Layer A Layer is composed of a Material with the added property of Thickness. Layers can be of four types: Homogeneous Opaque, Inhomogeneous Opaque, Unventilated Cavity and Glazing Layer. The first two types are associated with Enclosures and Partitions, the third only with Enclosures and the last, with the Glazing object under Aperture. Glazing/ Opaque Materials Material objects hold various properties for thermal, environmental or cost analysis. Glazing material can be of types Gals, Gas, Shade, Blind and Screen and is associated with the Glazing object associated with the Glazing object under an Aperture. Equipment/ Occupants/ Lighting Elements contributing to the internal gains within a Space. These objects are defined by single values or varying values based on a schedule. These objects can be populated by pre-defined templates based on the Function property of Space. Required Internal Conditions Holds property-sets Ventilation, Infiltration, Thermostat and Humidistat, all defined by single values or scheduled values. These objects can be populated by pre-defined templates based on the Function property of Space. HVAC Heating, ventilating, and air-conditioning systems’ components (the associate subclasses are currently under development) Calculation Settings Includes various default settings for different calculation methods. May be extended if new calculation methods are implemented. Weather Data Weather information templates linked to the building model according to Location 3 CONCLUSIONS SEMERGY Building Model (SBM) is a comprehensive space-based building representation, tailored to meet the requirements of advanced energy simulation procedures, normative calculation routines, as well as life cycle analysis. Due to its object-oriented design, it can be extended in view of further functionalities with minimal effort. It enables efficient iterations within the optimization procedure and allows for simplification of data entry modalities to reduce user effort. 41 applications in processing common building performance inquiries. In O. Sikula, J. Hirs (ed), Simulace Budov a Techniky Prostredi. Brno: Ceska Technika - nakladatelstvi CVUT Granlund. 2013. RIUSKA Software. [http://www.granlund.fi/en/software/riuska/], Last visited February 2013 Hensen J., Djunaedy E., Radošević M., and Yahiaoui A. 2004. Building performance simulation for better design: Some issues and solutions. Proceedings of the 21th conference on passive and low energy architecture. Eindhoven. Heurix J., Fenz S., Anjomshoaa A., Neubauer T., Tjoa A.M., Taheri M., Shayeganfar F., Pont U., Ghiassi N., Sustr C., and Mahdavi A. 2013.Optimization in the SEMERGY environment: Multi-objective optimization for sustanable building design and retrofit, CESBP2013: Accepted Mahdavi A. 2000. Supporting collaborative design via integrated building performance computing. Proceedings of the 12th International Conference on System Research, Informatics and Cybernetics: Advances in computer-based and WEB-based collaborative systems:91-102. BadenBaden Mahdavi A., Suter G., and Ries R. 2002. A presentation scheme for integrated building performance analysis. Proceedings of the 6th International Conference: Design and Decision Support Systems in Architecture: 301-316. Ellecom: DDSS. Mahdavi A., Pont U., Shayeganfar F., Ghiassi N., Anjomshoaa A., Fenz S., Heurix J., Neubauer T., and Tjoa A.M. 2012a. SEMERGY: Semantic web technology support for comprehensive building design assessment. In G. Gudnason. R. Scherer. (ed), eWork and eBusiness in Architecture, Engineering and Construction:363-370. Reykjavík: Taylor&Francis. Mahdavi A., Pont U., Shayeganfar F., Ghiassi N., Anjomshoaa A., Fenz S., Heurix J., Neubauer T., and Tjoa A. M. 2012b. Exploring the utility of semantic web technology in building performance simulation. Proceedings of BauSIM 2012: Gebäudesimulation auf den Größenskalen Bauteil, Raum, Gebäude, Stadtquartier:58-64. Berlin: Universität der Künste Berlin. McGraw-Hill Construction. 2007. Interoperability in the Construction Industry: Smart Market Report. New York: McGraw Hill. Österreichisches Institut für Bautechnik. 2007. ÖIB-Richtlinie 6: Energieeinsparung und Wärmeschutz. Vienna:ÖIB Shayeganfar F., Mahdavi A., Suter G., and Anjomshoaa A. 2008. Implementation of an IFD library using semantic web technologies: A case study, In A.S. Zarli, R. Scherer (ed), ECPPM 2008 eWork and eBusiness in Architecture, Engineering and Construction: 539-544. Sophia Antapolis: ECCPM. Shayeganfar F., Anjomshoaa A., Heurix J., Sustr C., Ghiassi N., Pont U., Fenz S., Neubauer T., Tjoa A.M., and Mahdavi A. 2013. An ontology-aided optimization approach to ecoefficient building design, ECCPM 2013: Accepted SEMERGY graphical user interface for novice designers is already equipped with a logical core to generate the detailed contents of the SBM based on minimal user input. A specialized gbXML to SBM parser is currently under-development to facilitate data import from BIM software. Generation of building models based on GIS data for large scale assessments is the following step. Integration of a cost estimation module and HVAC systems, with particular focus on solar technologies, are other up-coming challenge of the SEMERGY project. The SBM schema shall be extended to support these future functionalities. 4 ACKNOWLEDGEMENTS The SEMERGY project is funded under the FFG Research Studio Austrian Program (grant No. 832012) by the Austrian Federal Ministry of Economy, Family and Youth (BMWFJ). In addition to the authors, the SEMERGY team includes: K. Hammerberg, V. Jain, T. Neubauer, C. Sustr, M. Taheri, D. Wolosiuk and A. Wurm. 5 REFERENCES Attia S. 2011. State of the art of existing early design simulation tools for net zero energy buildings: A comparison of ten tools. Louvain La Neuve: Université catholique de Louvain. Austrian Standards Institute. 2009. ÖNORM B 8110-6: Wärmeschutz im Hochbau. Vienna: Austrian Standards plus GmbH. Bazjanac V. 2008. IFC BIM-based methodology for semiautomated building energy performance simulation. In L. Rischmoller (ed.), Proceedings of the 25th CIB W78 conference: Improving the management of construction projects through IT adoption. Santiago: Universidad de Talca. Bazjanac V., 2010. Space Boundary requirements for modeling of building geometry for energy and other performance simulation. Proceedings of the 27th CIB W78 International Conference. Cairo. Berners-Lee T., Gödel K., and Turing A. 2006. Thinking on the Web. New Jersey: John Wiley & Sons, Inc. Building Smart. 2013. [http://www.buildingsmart-tech.org/], Last visited February 2013. De Wilde P.J.C.J. 2004. Computational Support for the Selection of Energy Saving Building Components. Delft: Delft University Press. Domeshek E.A., Kolodner J.L., and Zimring C.M. 1994. The design of tool kit for case-based design aids. In J.S. Gero, F. Sudweeks (ed.), Artificial Intelligence in Design’94: 109126. Alphen aan den Rijn: Kluwer Academic Publishers. EnergyPlus. 2013. Energy Plus Energy Simulation Software. [http://apps1.eere.energy.gov/buildings/energyplus/], Last visited February 2013. gbXML. 2013. [http://www.gbxml.org/aboutgbxml.php], Last visited February 2013 Ghiassi N., Shayeganfar F., Pont U., Mahdavi A., Fenz A., Heurix A., Anjomshoaa A., Neubauer T., and Tjoa A.M. 2012. Improving the usability of energy simulation 42 Functional and technological definition of BIM-aware services to assess, predict and optimize energy performance of buildings J. Rojicek & R. Fisera Honeywell Labs, Prague, Czech Republic G.D. Kontes, G.I. Giannakis, G.N. Lilis & D.V. Rovas Department of Production Engineering and Management, Technical University of Crete, Chania, Greece ABSTRACT: There are a number of important elements in designing of building energy management systems – when data collection, aggregation and management is usually well addressed by existing building management systems, actual analytical components allowing to diagnose energy-prone and/or user comfort compromising behaviors are far less mature. It is not only about developing algorithms for such tools, but also proper design of a hosting platform and its viability – it should not only enable access to sensor readings, but also provide access to other building data like Building Information Models and allow collaboration and interconnection of such analytics. BaaS project calls such tools Assess, Predict and Optimize services. Developing a smart platform supporting these services naturally leads to a concept of the building as a service ecosystem (BaaS) where any new tool can be plugged in the system and can benefit from already existing components. In the present work, the high level architecture of the BaaS platform is presented and the ability of BaaS system to act as a platform enabling the building contextual data as well as dynamic data (sensor readings) to software modules is demonstrated through a use-case example on a simple one zone office building. (HESMOS Project) provides advanced simulation 1 INTRODUCTION capabilities to decision makers and attempts to close the gap between BEMS and BIM in an effort to outBuildings are important contributors to the total encome energy- and cost-minimizing decisions ergy consumption, thus Building Energy Managethroughout the whole building life-cycle. ment Systems (BEMS) are key ingredients towards Within BaaS Project, a smart platform supporting enabling parsimonious use of energy resources. Exthe whole ensemble of APO services (i.e. Control isting BEMS, although facilitating near-complete Design Optimization, Fault Identification and Enermechanisms for data collection, aggregation and gy Benchmarking) is developed, where each new management, lack the analytical components allowservice can be plugged into the platform and benefit ing diagnosing a behavior leading to excessive energy from already existing components, thus leading to a consumption and/or compromised occupants’ comfort. concept of the building as a service ecosystem (BaaS). BaaS project calls such tools Assess, Predict and OpThis also implies “Software as a Service” (SaaS) martimize (APO) services. keting model. Even though research effort has been focused on Such solution not only improves system perfordeveloping algorithms for such tools showing signifmance by detecting and correcting inefficiencies, but icant results (e.g. see PEBBLE Project, OptiControl it also increases user awareness: having all relevant Project), proper design of the hosting platform and Key Performance Indicators (KPIs) in one place it is its viability are essential. Such platform should supeasier to monitor discrepancies from their expected port access to Building Information Models (BIM), values or any other KPI deterioration. KPIs usually thus providing a common interface between the varcover energy consumption, occupants’ comfort and ious analytics and enable collaboration and interecological friendliness of the system. Even if KPIs communication between them in a transparent way. evaluation cannot be done automatically, or if the Several research projects aiming to provide such system control cannot be altered manually (e.g. for a holistic approach are on the way: the Monitoring safety reasons) the man in the loop has all necessary System Toolkit (MOST Project) is a set of tools eninformation at his hands, so it is easier to make qualabling effortless measurement, processing and visuified decisions. alization of in-building data streams; Control and In the present work, the high level architecture of Automation Management of Buildings and Public the BaaS kernel hosting the APO services is presentSpaces (CAMPUS21 Project) develops a Hardwareed and the ability of BaaS system to act as a platSoftware-Platform for the integration of existing form enabling the building contextual data as well as ICT-subsystems supporting energy, building, and dynamic data (sensor readings) to software modules security systems management for energy-efficient is demonstrated through an illustrative example on a operation of public buildings and spaces; and ICT simple one zone office building. Platform for Holistic Energy Efficiency Simulation and Lifecycle Management of Public Use Facilities 43 exist, facilitating a vast diversity in the shape of the controller, the necessity of a model of the building, the required inputs and so on. Despite the differences between these approaches, the general control problem can be described by defining some basic components (for a detailed description see Kontes et al. 2012b). To start, let’s consider that the physical system (building) can be described by a thermal simulation model, which is able to predict the thermal state of the actual building, taking into account the current building states (like wall and air temperatures, humidity, etc.), the predicted weather conditions and the control actions (like heating and cooling loads, shading angle, etc.) applied to the building. Having such a model at hand, allows designing series of control inputs that lead the physical building to a set of states that are (near-) optimal with respect to a performance measure, using the model for the design process. In buildings domain, the performance measure is modelled as a constraint optimization problem, facilitating the following components: − A performance-indicating KPI, usually correlated with operational cost or energy consumption (e.g. minimization of the total energy consumption, minimization of the grid-supplied energy, maximization of the net energy produced, etc.). − A set of KPIs acting as constraints, ensuring comfortable in-building conditions for the occupants (e.g. visual, acoustic, thermal comfort constraints, etc.). The availability of the building model, along with the stochastic nature of the occupancy patterns and weather conditions consist the use of model-assisted control (or model-predictive control, (MPC)) design optimization techniques (Goodwin et al. 2005, Bertsekas 1995) suitable for solving the above constraint optimization problem – see (Ma et l. 2010, Oldewurtel et al. 2010, 2012, Giannakis et al. 2011, Pichler et al. 2011, Kontes et al. 2012a, Cigler et al. 2012) for successful application in buildings. Here, based on the available building model and weather predictions, the optimization problem is solved for the period of time accurate weather predictions are available (prediction horizon), while the resulting controller is applied for a shorter period of time, called the control horizon. Besides the predictions, this approach necessitates the availability of sensor data from the real building for the previous days, since they will be used for the warming-up process, i.e. the model will be simulated using the actual building conditions for the previous days, in order to assimilate the actual thermal state of the real building at the beginning of the optimization process. 2 STATE OF THE ART OF APO SERVICES APO services address three main areas: Fault Detection and Diagnostics (FDD); Energy Management (EM) and Control Design and Optimization (CDO). Though they are treated separately, one can clearly see that they can be closely related. Recent commercially available FDD services are typically provided in form of SaaS. The algorithms running on the cloud are typically designed based on accumulated expert knowledge and are searching for unusual patterns in the data collected from building sensors. As the contextual information for the target commercial buildings is available usually only in very limited form and with quality varying from site to site, the FDD solution, aspiring to wide applicability without complex settings required, needs to be sufficiently robust. This implies that the algorithms should be capable of producing accurate results even with hardly any contextual information available. The lack of context can, in some cases, limit the fault detection and mainly diagnostics capabilities of the service. Basically, there are two main approaches in the FDD research field: rule-based (Kukal et al. 2009, Schein et al. 2006) and model-based (e.g. Du and Liang 2007) (and combinations). Typically rulebased methods are continuously evaluating simple rules, or sets of rules, using sensor data collected from the building, while a fault reasoning process subsequently decides on particular fault’s presence. Model-based methods typically compare the measured value of proper KPI against the modelled one. These referential values are constructed exploiting various approaches ranging from black-box, when hardly any context information is available, to white-box, when the monitored equipment is known in detail. Often only one piece of equipment is focused separately neglecting the fact, that it is commonly a part of a complex building system, where faults diagnostics is a difficult task (fault masking & propagation) hardly solvable without having a contextual information at disposal (e.g. building equipment connectivity model). The availability of building context (BIM) can thus significantly improve the performance of fault detection and mainly fault diagnostics algorithms. Moving forward, energy management services are often a part of BEMS. Typically the energy consumption on the whole building level and several major energy consumers are monitored. More advanced systems provide baselining or benchmarking functionality, usually on the building envelope level. In the former case, the referential energy consumption is constructed from past data (e.g. searching for similar driving conditions in the history) and in the latter case the reference is taken from a similar building (typical for the retail stores chains). Finally, towards designing intelligent BEMS, a variety of control design optimization approaches 44 (MVD) has been adopted within BaaS. An MVD defines the smallest possible subset of the full IFC schema required to satisfy one or many exchange requirements, thus, the exchange requirements for each APO service are defined and made publicly available. Within this context, if two software components are to interact they need to exchange sufficient information – all the exchange requirements so that this communication is complete are defined in the MVD. So the “sending” component (let's call it the writer), should create all the information to be sent (in conformance to the MVD), and the “receiving” component (let's call it the reader), should know how to use the information (which comes in conformance to the MVD), to perform some useful task. So both the “reader” and the “writer” should be designed to satisfy the requirements posed by the MVD (i.e. understand the MVD). Now, it is conceivable that there are many “writer” components, like CAD tools or GUI interfaces that populate aspects of the data model. BIM acts as the aggregator of such information, and the provider to clients (via available interfaces) of the requested information. Moreover, the availability of the BIM and the MVD description allows the generation of queries to the BIM based on the MVD, since the MVD actually determines which queries are supported, i.e. we can expect some meaningful data in the response. This way a “library” of queries for each exchange requirement (FDD, CDO, etc.) can be generated, that will be automatically supported by all ifc files compliant to the MVD. Finally, following this approach, the APO service modules are equipped with auto-configuration capabilities, while new modules can be imported to the system through a trivial process, as long as they ensure compatibility with the exchange requirements of the respective service, i.e. they are MVDcompliant. There are a number of software tools allowing querying and other manipulation of BIM data – the BaaS project focuses mainly the part of obtaining relevant data from BIM and using them efficiently in the building services aiming for control optimization, fault detection and diagnostics, etc. All such functionality should be enabled automatically, without any human interaction. BaaS can be seen as a platform enabling the building contextual data as well as dynamic data (sensor readings) to software modules providing the actual service or functionality needed. Such platform is supposed to enable required inputs to many other features and software modules and impact significantly the market, especially if it is open to public. 3 BAAS SYSTEM 3.1 Motivation The major obstacle for the deployment of new smart building control and monitoring technologies is the deployment cost. Typically, more advanced the technology is, more contextual information is required for the proper setup. In addition, since expert engineers are required for the task, the overall cost further increases. The reason is that due to lack of standardization (too many proprietary standards) in the building automation area, a lot of work has to be done manually requiring deep knowledge about the particular system. This problem is most evident in the commercial buildings domain where the need for advanced technologies grows quickly. Typically an engineer responsible for designing control algorithms decides rather to deploy safe but robust solutions – usually a simple rule based controller; and such solutions can hardly attain the performance of advanced control algorithms like MPC. In ideal case, the customer should be able to select a software package capable of providing desired functionality (building energy management, fault detection and diagnostics, control optimization, etc.), and the local installation or remote connection (in case of SaaS business model) is done automatically, so the tool is enabled to be used in a short time – and this is exactly the way the BaaS is aiming at. The target is to reduce required human interaction during deployment to some necessary minimum, ideally handled by a friendly GUI which would cut the deployment costs down. BaaS is thus pioneering the major enabler of advanced building technologies, like MPC that were mainly used only in the industrial domain and unlike commercial solutions uses open standards to achieve it. 3.2 Building Information Models Within BaaS project, the path selected for treating the aforementioned limitations is the use of properly populated Building Information Model (BIM) with all the contextual information about the building in a standard way, along with the development of tools enabling utilization of such information. Under this perspective, the use of Industry Foundation Classes (IFC) data model is used as a standard way of describing building contextual information. The use of BIM and IFC tools allows for semiautomatic deployment and operation of APO services in all buildings at hand, regardless of variations on the building types, construction, location and available systems. On the other hand, the use of BIM and IFC alone inserts more complexity to the problem, rather than simplifying the task, since requiring by all software components to provide support for the entire IFC schema is not a viable solution (Bazjanac 2007). Due to this fact, the concept of Model View Definition 3.3 Simulation models In addition to simulation as a design and decisionsupport tool, we take the stance within BaaS, that simulation is an essential ingredient to providing APO services. In this operational-phase utilization of 45 simulation models, the usage scenarios are different: simulation models are consumed by APO services that provide useful functionalities with respect to the building operation. The availability of factual (sensed) data, along with forecasts for pertinent parameters (e.g. weather, occupancy) can be exploited and actively used to bridge the “simulated” and “real” worlds, reducing or even mitigating design-phase uncertainties. With respect to the existing calculation methodologies for simulation models, quasi-static and CFD calculation methodologies are primarily useful in the design phases, either due to the resolution of their predictions (annual basis for quasi-static) or due to the inherent assumptions and modelling detail required – as such, they are of lesser importance within BaaS. The use of time-steps in the range of a minute to one hour allows to account for the dynamics of active climate control systems, but also to incorporate control strategies that use state measurements as inputs to compute actuation commands. The desire to use simulation as a forecasting tool, also suggests that a “small” time step might be warranted. In view of the comments above, in Figure 1, the type of calculation methodologies of interest to BaaS can be identified. Simulations and their respective calculation methodologies will be used with BaaS to accomplish a variety of different tasks: − Energy performance estimation: In this task the energy performance of the whole building is estimated. Energy performance includes total energy needs, including energy used for conditioning the spaces. In the transient calculation methodologies above thermal comfort parameters can also be computed. − Energy performance forecasting: The goal of energy performance forecasting is to estimate building energy needs in order to preserve comfort conditions in building spaces, during a finite future time horizon. The use of forecast data obtained from various sources is necessary in this case. As it can be expected the validity of the forecasting process depends strongly on the quality of the forecasts. Integration of past data and forecast data (obtained from different sources) is essential here and the abstractions of the middleware will facilitate access to these data, so that the problem can be correctly set up. − Model calibration: Although models are designed to predict the real behaviour of buildings and their systems as accurately as possible, their predictions may differ from real sensor measure- ments, due to a variety of reasons including: sensor measurement errors, modelling insufficiencies, or incorrect model parameter value’s estimations. Model calibration tasks rely on past sensor measurements in order to change the model parameter values and bridge the above gap. − Components validation: System performance can degrade over time, leading to out-ofspecification operation. This can have adverse effects with respect to energy performance and thermal comfort. Anomaly detection and identification using simulation-based methodologies can be one of the ways, to identify such events − Control design: The general purpose of (supervisory) control design is to design a controller that given state parameter values will return operation schedules and commands of controllable building elements. In model-based control design, the calculation methodology (here synonymous to “model”) is used in combination with modelpredictive control algorithms to generate such strategies. − Control design optimization: The generated control actions, using simplified state-space models, can have poor performance when applied to the real system. For this reason, the resulting controllers can be improved using more “accurate” building models, by performing a second optimization step. Uses of calculation methodologies can be an invaluable asset in finetuning/optimizing controller parameters. 3.4 System architecture Within the BaaS architectural design, a three layer architecture is envisaged: the data layer serving static and dynamic data needs through the implementation of an extended Building Information Model (eBIM) comprising a data warehouse and a BIM server; the communication layer acting as an abstraction layer to facilitate communication between the physical and ICT layers; and the APO Service Layer to provide the reasoning and analytics services. These functionally disjoint layers operate independently, communicating through the use of properly-defined (software) interfaces. The term APO collectively refers to continuously recurring tasks during building operation: assessment of the current building state; prediction of the effects that various decisions will have to KPIs; and, optimization of performance as measured through relevant KPIs. 46 Figure 1. Calculation methodologies of simulation models. the building envelope to the individual building equipment) and to identify critical levels for effective operation in order to take measures for respective maintenance − Control Design and Optimization, containing services which provide control-related analytics to design monitor and optimize applied control strategies by identifying control faults and inefficiencies. At the APO Layer, what is collectively denoted as services should be understood as functional components implemented as a collection of software modules. These modules are either developed during BaaS project or can be provided by interested stakeholders to implement analytics (fault detection and diagnostics, control design, etc.). From a business perspective these modules can be part of the business intelligence and solutions portfolio provided as a service to building owners and occupants. A schematic representation of the APO kernel, containing all the necessary functional components, is shown in Figure 2. The main modules of the APO kernel are: − The module registry, where various modules are available for use by any APO service, since all APO services that will be deployed later on to the system will have to select and use control design, control design optimization, energy monitoring, fault detection and simulation modules that are available through the module registry library. − The simulation manager is responsible for providing a fully functional simulation model of the building to any APO service that requests it. − The service handler, which is responsible for coordinating all APO services, by invoking the control design, control design optimization and fault detection modules, through the control and fault detection managers. Figure 2. APO Services Layer – Core Components. Functionally the APO services layer is intended to host the necessary algorithms to analyze the collected in-building data, interactions of processes, and generate control strategies for effective energy management. Specifically, the services will provide core intelligence, building/facility assessment and monitoring, prediction and optimization services, utilizing information made available by the data layer services. The aforementioned activities can be grouped together in three functionally-disjoint groups of services: − Fault Detection & Diagnostics, containing services which provide analytics to detect and possibly find a root cause (diagnosis) of various equipment malfunctions and faults − Energy Management, containing services which provide analytics to monitor equipment performance at various building hierarchy levels (from 47 − The control manager, which is responsible for any control-related action within BaaS. − The fault detection manger, which is responsible for detecting and identifying as many problems as possible of the building actuating components. Based on the user inputs, or predefined default set of faults to be monitored, the fault manager instantiates and manages a variety of fault detection and identification (FDI) abstract components. − The signal handler, which handles all data (past, present, future) coming through the middleware layer to the given APO services. A signal is a generic software abstraction used throughout the APO kernel able to accommodate any type of data. − The event handler, which is responsible for the event management: distribution to selected blocks, priority handling, repeating mechanism for unacknowledged events, logging, event timeof-life, etc. − The BaaS connector, which secures the data connection (data access layer) between the APO services kernel and modules. − The configuration manager, which collects the user requirements (cost function formulation – KPIs, constraints) entered through simple GUI and forms the setup for each task solved by the APO services. − The permissions and user manager, which takes care of the security aspects related to the APO services. Number of user profiles can be generated with different privilege levels. New users added to the system are then assigned by the selected user profile. − The time control, which is responsible for the proper timing of all APO services kernel actions. perature, thus saving energy by shutting-down the HVAC system. In an effort to capture behaviours that lead to energy leakage, the fault detection manager instantiates a fault detection/identification object pertinent to a fault description specified by user or experts during the BaaS configuration phase. In this setting, a fault is indicated if the related zone HVAC system is operating and the window is open at the same time, since this is an energy inefficiency that needs to be reported and treated. The fault detection object instantiates initially a relevant symptom object that will evaluate a specified set of rules to detect incorrect behaviour. If the symptoms supporting the fault hypothesis are observed, i.e. the HVAC is operating while the window is open, the fault likelihood is increasing and after a period of time (fault reasoning) it exceeds a predefined threshold and a fault event is generated by the fault detection object. In order to automatically adapt the specific fault detection logic to any building type at hand, a set of queries to the available BIM server are required, with the following order: 1. Get all window contact sensors of the building. 2. For each contact sensor, identify the specific window it is mounted to. 3. For each window, get the room it belongs to. 4. Get the air terminal and air terminal box serving each room. 5. Get the upstream HVAC structure (AHU, chiller) for each terminal. 4 EXAMPLE Consider the simple one-zone office building shown in Figure 3, located in Germany. The building, as shown in Figure 4, is equipped with a temperature, occupancy and a window contact sensor, while a sensor on the roof is used to measure the outside dry bulb temperature. Moreover, the building is served by a dedicated HVAC system. Finally, a BIM server containing the building description in IFC format is available, along with a Data Warehouse scheme that contains historical sensor measurements for the building. During the summer period, a static rule-based controller is applied, facilitating the following rule: when there is a demand for cooling (room temperature above setpoint) during working hours of the building, the window should open if the outside temperature is at least 3°C lower than the room tem- Figure 3. Architectural view of the building. Once the querying process has been completed, a collection of fault detection objects as the one described earlier is created and associated with each contact sensor – HVAC pair, while, at the same time, is able to request historical operational data for the sensor measurements and HVAC operation from the DW. Note here that this fault detection process assumes that all involved entities work properly (window opening sensor, zone temperature sensor, cooling valves, etc.). 48 room temperature and occupancy) into setpoints, while the KPIs of the optimization problem are defined as the total energy consumption and the zone temperature. In Figure 6, a draft sketch of the whole process is presented. First of all, the control manager requests historical data for the warming-up phase and reference occupancy data from the DW, as well as weather predictions from external services, through the middleware. The response signals are preprocessed and forwarded to the simulation manager. The simulation manager, using the simulation setup guidelines provided by the control manager, requests the available simulation model and creates the simulation and co-simulation objects, through which the historical, predicted and reference data are injected to the simulation, Finally, the control design optimization object created by the control manager is used to design the control strategy to be applied to the real building. Figure 4. Available building sensors. Since the entities dependencies are known through the querying process; the fault dependence object checks all relevant fault identification and detection objects outputs for possible faults. Should any involved entity be faulty, the FDI process must not be applied or its results must be discarded. An example of relevant objects tree is described in Figure 5, where the analyzed one zone example is put into wider context of the HVAC distribution system of a commercial building. The chiller (or chiller plant) is producing a cooling for the whole building, i.e. the chilled water is distributed to cooling coils in Air Handling Units (AHU) and optionally also in air terminal boxes serving the individual zones or rooms. Figure 5. The HVAC system. Figure 6. Functional flow chart for the one-zone simple example. Assuming the fault detection object identifies a malfunctioning contact sensor; an event containing the information on the discovered fault is created and announced on the event handler and to the building manager. Subsequently, using pre-defined fault recovery logic (like a decision tree), the control manager postpones the use of the static rule-based controller and forces application of a model-assisted control design optimization module for the control of the zone. The new control approach assumes that the window is uncontrollable and attempts to regulate the HVAC operation in order to save energy and maintain comfort in the room. For the new control design optimization problem, the warming-up period is set to two days, while the prediction horizon to three days. The new control function for the HVAC is a linear controller, transforming a set of building states (outside temperature, 5 CONCLUSIONS In the present work, the high-level architecture of the kernel hosting the BaaS APO services has been presented. BaaS system manages to consist a platform enabling the building contextual data as well as dynamic data (sensor readings) to software modules, allowing for semi-automatic deployment and operation of Asses, Predict and Optimize services in all buildings at hand, regardless of variations on the building types, construction, location and available systems, through the use of BIM and IFC tools and incorporation of MVDs. 49 OptiControl Project - Use of weather and occupancy forecasts for optimal building climate control, http://www.opticontrol.ethz.ch/ Pichler M.F., Dröscher A., Schranzhofer H., Kontes G.D., Giannakis G.I., Kosmatopoulos E.B. and Rovas D.V. 2011. Simulation-assisted building energy performance improvement using sensible control decisions. Proceedings of the Third ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Buildings, pp. 61-66, ACM Schein J., Bushby S.T., Castro N.S. and House J.M 2006. A rule-based fault detection method for air handling units. Energy and Buildings Volume 38, Issue 12, December 2006, Pages 1485-1492. 6 ACKNOWLEDGEMENTS The research leading to these results has been partially funded by the European Commission FP7ICT-2011-6, ICT Systems for Energy Efficiency under contract #288409 (BaaS). 7 REFERENCES Bazjanac V. 2007. Impact of the US national building information model standard (NBIMS) on building energy performance simulation. Presentation at the Building Simulation 2007 conference, Beijing. Bertsekas DP 1995. Dynamic Programming and Optimal Control. Athena Scientific CAMPUS21 Project- Control and Automation Management of Buildings and Public Spaces, http://www.campus21project.eu/index.php/en/ Cigler J., Prívara S., Váňa Z., Žáčeková E. and Ferkl L. 2012. Optimization of predicted mean vote index within model predictive control framework: Computationally tractable solution. Energy and Buildings, 52, pp. 39-49,2012 Du R., and Liang J. 2007. Model-based fault detection and diagnosis of hvac systems using support vector machine method. International Journal of Refrigeration, Volume 30, Issue 6, September 2007, Pages 1104-1114. Giannakis G. I., Kontes G.D. , Kosmatopoulos E.B. and Rovas D.V. 2011. A model-assisted adaptive controller finetunning methodology for e_cient energy use in buildings. In: 2011 Mediterranean Conference on Control & Automation, pp 49 – 54. Goodwin G.C., Seron M. and de Dona J. 2005. Constrained control and estimation: an optimisation approach. Springer Verlag, 2005. HESMOS Project - ICT Platform for Holistic Energy Efficiency Simulation and Lifecycle Management of Public Use Facilities, http://hesmos.eu/ Kontes G.D., Giannakis G.I., Kosmatopoulos E.B. and Rovas D.V. 2012a. Adaptive-fine tuning of building energy management systems using co-simulation. 2012 IEEE International Conference on Control Applications (CCA). pp. 1664-1669, IEEE. Kontes G.D., Giannakis G.I., Kosmatopoulos E.B. and Rovas D.V. 2012b. Demand-shifting using model-assisted control. ICHMT DIGITAL LIBRARY ONLINE, 2012. Kukal J., Macek K., Rojicek J. and Trojanova J. 2009: From Symptoms to Faults: Temporal Reasoning Methods. In the Proc. 2009 Int. Conference on Adaptive and Intelligent Systems, Klagenfurt, Austria, 2009. Ma Y., Borrelli F., Hencey B., Coffey B., Bengea S. and Haves, P. 2010. Model predictive control for the operation of building cooling systems. American Control Conference (ACC). 2010. pp. 5106-5011. MOST Project – Monitoring System Toolkit, http://most.bpi.tuwien.ac.at/ PEBBLE Project - Positive Energy Buildings thru Better Control Decisions, http://www.pebble-fp7.eu/ Oldewurtel F., Parisio A., Jones C.N., Morari M., Gyalistras D., Gwerder M., Stauch V., Lehmann B. and Wirth K. 2010. Energy efficient building climate control using Stochastic Model Predictive Control and weather predictions. Proceedings of the 2010 American Control Conference (ACC). 2010, pp. 5100-5105 Oldewurtel F., Parisio A., Jones C.N., Gyalistras D., Gwerder M., Stauch V., Lehmann B. and Morari M. 2012. Use of model predictive control and weather forecasts for energy efficient building climate control. Energy and Buildings, 45:15-27, 2012. 50 Towards an ontology representing building physics parameters for increased energy efficiency in smart home operation M. J. Kofler & W. Kastner Vienna University of Technology, Institute of Computer Aided Automation, Automation Systems Group, Vienna, Austria ABSTRACT: As building automation devices become more affordable, the holistic control of buildings is a major research area. Control strategies for automated homes need to consider the environment and act according to detected conditions to be able to optimize building energy behavior. While obviously the habits of residents as well as the state of appliances appear to be important, influences often overlooked are building physics and architectural characteristics. Parameters from these domains may be used to fine-tune control strategies and reach a faster introduction phase when installing a new building automation system. In modern building design and construction, a Building Information Model (BIM) is often used to exchange characteristics between different workflow tools. This publication proposes the transfer of parameters collected in a BIM to the operational phase for reuse in automated building control. Hereby, a semantic information layer is described that uses the Web Ontology Language (OWL) to create a sophisticated view of the building. physics may not be neglected. Building material, state of walls (chilled or tempered), as well as general layout to a certain extent influence energyintensive tasks like heating and cooling, and therefore may be integrated into smart home control to optimize building performance. To operate heating and cooling in accordance to the state of the building and its structure promises at the same time an improved thermal comfort as well as overall energy footprint. When the building is to be controlled by a smart home control system, this system needs a general view of the environment. Sophisticated modeling techniques from the semantic web domain can hereby be used on the one hand to take away complexity from the system relying on it and on the other hand to describe relations between parameters in a machine-interpretable form. Therefore, in this paper the modeling of the building and its characteristics with the help of the Web Ontology Language (OWL) is proposed. The usage of a complex, logic-based language like OWL can in this context aid the creation of a model that accurately reflects reality and represents knowledge about the building environment. After a brief introduction into the semantic web vision, a semantic representation of building parameters is presented. The main components of the representation and their interdependencies are explained. It is further shown how the created ontology can enhance information retrieved from a Building Information Model (BIM): Hereby, inherent logical reasoning of the applied semantic web language and the design of logical rules for sophisticated inferences is shortly explained. As a possible application, an outlook at the integration into a smart home multi-agent system is given. 1 INTRODUCTION Building automation is becoming more attractive for the home environment and the home of the future is filled with devices helping the resident to control the environment according to personal comfort and needs (Rosslin & Tai-hoon, 2010). To reach comfortable conditions in a building always goes hand in hand with the general building energy consumption. While in some cases already a more or less intelligent system helps the user to reach personal comfort and increase energy efficiency seldom all possibilities are exploited. Many times, energy-intensive tasks such as temperature control are realized by simple two-level controllers, solely attempting to reach a certain pre-defined setpoint, leaving the operating environment and building conditions unconsidered. In case the control system optimizes energy efficiency, the general focus often lies in improving the consumption of the associated devices (e.g., Jahn et al., 2010) more than the global building situation. While this is unquestionably an important step when coping with energy optimization, there are also other factors that need to be considered to get a global view of the environment. When it comes to heating setpoint control in order to adjust room temperature and bring it to a comfortable level according to the user schedule, there often just exist very simple and inflexible schemes that can be programmed by users in advance stating the planned occupancy times. How fast the desired temperature setpoint is reached, hence the preparation time, is however dependent on several prerequisites: While obviously the current indoor air temperature of a space directly influences this preparation time, also exterior conditions as well as the building architecture and associated building 51 Figure 1. Top-level concepts and hierarchical classification of building information and parameters. The Web Ontology Language is defined as a description language to represent meanings of entities on the World Wide Web as general as possible. An ontology hereby may be defined as a formal description of related concepts in a world of interest (Gruber, 1993). A representation in OWL facilitates information federation and allows sophisticated computational reasoning on stored knowledge when restricting the language to Description Logic (OWLDL). Based on RDF, OWL again represents knowledge through data triples and allows to model expressive semantic descriptions for entities. To extend expressivity beyond the capabilities of OWL-DL and still preserve computability the W3C proposes rule languages. One of the most important and widely used semantic web rule languages is SWRL. SWRL rules are a rule extension for OWL and follow the form of a Horn clause: such a logical clause describes implication rules consisting of a rule head and rule body while the rule body is the condition that needs to be met, so that the implication modeled by the rule can be drawn (Hebeler et al., 2009). The rule head further describes the conclusion that can be inferred. An SWRL rule may thus be seen as simple if-statement in a programming language. Finally, the defined query language for RDF is named SPARQL. SPARQL is the same for an RDF triple store as the well-known database language SQL for relational databases or XQuery to an XML database. SPARQL is a graph-based query language, what means it allows defining graph patterns that will be searched for in a given RDF graph. Modeling information with RDF/OWL instead of plain XML brings the advantage, that meaning of information can be expressed. Therefore, a query to an RDF triple store can retrieve stored information ac- 2 THE SEMANTIC WEB IN A SMART BUILDING CONTEXT The semantic web vision was first defined by Tim Berners Lee as a web in which computers become capable of analyzing content, links and transactions between people and computers (Berners-Lee & Fischetti, 1999). As such, it may be seen as an extension of the World Wide Web (WWW) in a way, so that information stored on the web can be annotated with metadata describing the meaning of the represented data in a machine-interpretable form. In this way, information on the Internet should be interpretable for humans and machines at the same time. To realize this vision, several proposals of the World Wide Web Consortium (W3C) have been developed which allow the annotation and formal description of information on the World Wide Web. With the grounding in computational logic and knowledge representation for intelligent systems, automated reasoning may be performed on the metadescription of data and thus an understanding of data context as well as relations between information sources can be reached. The Resource Description Framework (RDF) was developed to structure information in graphs for representation of knowledge and information exchange on the World Wide Web. As such, RDF builds its structure on subject-predicate-object triples. A subject in this case is the entity that is described with the help of a predicate that connects the subject to a specific characteristic of this subject represented as object. Connecting single triples to each other forms an acyclic RDF graph. 52 about the building already collected in the design and construction phase through architects, planners and building physicists. The inclusion of information collected in a Building Information Model therefore forms a fundamental basis for the envisioned semantic layer as often a BIM is used to collect information about a building throughout the building workflow. A BIM is more than a format for computer-aided design (CAD) tools: The objective of a BIM is to preserve information available about the building over the whole building lifecycle (Azhar et al., 2008). Using a BIM in architecture and construction brings interoperability among different domains and software tools and besides other things may already lead to improved, energy efficient design of a building by making transfer of building data into energy analysis tools available (Stumpf et al., 2011). Certainly, it can be beneficial for a sophisticated building automation system to include this knowledge about the building, its structure and environment in provided control strategies. At the moment no single BIM standard, but several interchange formats exist. One of the best known BIM is the IFC, the Industry Foundation Classes. IFC is an extensible framework model that was developed to create a large set of data representations for exchange between software applications used to document building design and construction (Eastman et al., 2008). Because of the general nature of IFC, the model schema itself is split into several layers representing diverse building information classes based on the ISO-STEP (Pratt, 2001) modeling standard for building information exchange. The general nature of IFC allows the unambiguous transfer of data back and forth between different software tools, which however also raises the complexity of this exchange format. For a smart home system, the detail that is offered by IFCs is however not needed. Therefore, because of the high overhead induced by the model the IFC cannot be seen as the perfect choice for direct usage in a smart home system. Another open BIM specialized in the domain of energy analysis is the Green Building XML schema (www.gbxml.org). The model focuses on exchange of energy-related data between construction and energy analysis tools. Because of the use of XML instead of ISO-STEP as data format, the structure of the model is less expressive but also inherently simpler than IFC. At the same time however, energycritical parameters that may be useful at building operation time are represented to a sufficient extent in this schema. Therefore, a transformation process from the building information model Green Building XML (gbXML) into OWL is a first step to create a semantic building information layer and was introduced in (Kofler & Kastner, 2009). The fundamental OWL representation of building data, retrieved from gbXML through the transformation, may be used as a basis for the creation of a sophisticated building in- cording to the meaning of the data while in XML a query will mostly be restricted to the content. While at the moment XML is still prevalent when modeling information for building automation systems (e.g., OSGi, OPC UA, oBIX), there also exist efforts to describe knowledge about building systems based on RDF. The DogOnt project (Bonino & Corno, 2008) focuses on the detailed modeling of the building device landscape including functionalities and commands for different devices. The OWL model rudimentarily allows the definition of the architectural entities which are mainly used to associate a device in the building with a specific room, while building physics and exact architectural dimensions are not mapped. Other projects like (Wolf et al., 2009) focus more on residents, their context, behaviors and interaction with the devices. Again, information about the building is reduced to a simple room entity that may include certain devices. The BOnSAI project (Stavropoulos et al., 2012) defines an ontology that models functionality, hardware users and context. The defined ontology also fails to describe the architectural and building characteristics beyond the concepts of locations, rooms and floors. Yet other articles focus on transformation of Building Information Models into semantic web ontologies for exchange of information between different stages of the building lifecycle. (Beetz et al., 2009) describe the process of transforming the EXPRESS schema that underlies Industry Foundation Classes (IFC) into an OWL representation. In their article they show how the underlying model of IFC may be semantically represented with OWL, as well as arising limitations with respect to automatic conversion. (Venugopal et al., 2012) proposes an ontology-based model exchange between stakeholders of the building construction process. The article focuses on the improvement of interoperability between BIM tools by developing consistent model views using the Web Ontology Language. Therefore, the authors describe a layered approach basing their specific model on already defined standard and domain ontologies. The present paper finally proposes the use of information collected in a BIM for representing building physics and architectural information in a smart home system ontology. 3 SEMANTIC BUILDING INFORMATION LAYER Particularly for autonomous building systems it is essential to completely represent information available about the building and its characteristics in a machine-interpretable form. As this information can seldom be retrieved to a full extent at the stage of building usage, it is desirable to reuse information 53 gy demand simulation may also influence the energy optimization task during building operation (e.g., people heat gain, blower door value). These parameters can be seen as static constants that among other factors may be used to calculate the actual heating or cooling requirement of a building and accordingly optimize the control of associated facilities like for example a geothermal heat pump. Finally, the identifier element is defined to group XML ID attributes that need to be imported from gbXML as well. The following section goes into detail for the parameter hierarchy and explains each class and its defined usage. formation ontology representing building physics and architectural parameters in a machineinterpretable way. 3.1 OWL conceptualization of building physics As the information retrieved from gbXML is extracted from an ordinary XML file following the gbXML schema, the initial OWL hierarchy is flat. This results from the fact that an inheritance relation (i.e. subsumption) as it is defined for hierarchically ordered concepts in OWL cannot generally be presumed in XML. Further, because gbXML describes a format for interoperability between development tools of the Architecture Engineering and Construction (AEC) domain, most of the elements are described in a very general way, what means, that many describing elements are optional. For example, the definition of the “Material” element in gbXML includes parameters like density, thickness, conductivity or R-value among others, however the only required paramter is an ID. This great amount of optional elements makes the definitions retrieved from gbXML rather unsuitable for the automatic deduction of a class hierarchy. Therefore, an asserted OWL subsumption hierarchy was manually created in order to classify the retrieved gbXML parameters. As shown in Figure 1, the top level concepts include the most general groups of data represented in gbXML while subhierarchies define partitions of subgroups. Firstly, the gbXML element describes the gbXML document itself, and includes one campus (i.e. site) for which the document was created. A campus further includes facts about the exterior environment of the building as well as other information specifically relevant for energy analysis. Further, one or more buildings are defined for a campus site. Each building has several associated building elements, a class defined to generally group relevant building parts (e.g., building storeys, zones, spaces, materials, surfaces). Hereby, a further distinction is made between physical and conceptual building elements. gbXML further allows to describe different types of equipment (e.g., air loops, lighting system) and therefore the top-level class equipment was created. The main top-level hierarchy is however the parameter classification: this hierarchy classifies all parameters that may be specified for buildings and equipment and groups them according to their usage type (e.g., geometric, temporal, building) as well as building lifecycle phase. The relevance of each parameter for a specific phase can hereby be classified through the ParameterSignificance group which currently identifies three different stages: analysis, construction and operation. The parameter significance concept was created as a BIM generally spans several stages of the building lifecycle and for example not all parameters available for energy analysis also have an influence on the operation of a building. Yet, other parameters that are used in building ener- 3.1.1 Building information parameter classification The parameter subhierarchy is mainly split into two main types: lifecycle-related and application-related. Due to XML limitations of the parameter description in gbXML the application-related hierarchy has to be manually asserted, while the lifecycle-related concepts allow logical reasoning to be applied. The following list provides the main parameter groups that are defined in the shared vocabulary and may be used to group gbXML elements. • Calculation parameters: general parameters that may be used for calculating performance and cost of equipment should be classified in this group. • Building parameters: parameters that are related to the building itself and its surroundings are represented in this concept. • Equipment parameters: this group is concerned with parameters regarding the design and operation of equipment in the represented building. As there exist several equipment parameters in the gbXML schema that are solely used for shade control simulation, an own subconcept has been created for this group of parameters. • Analysis method parameters: this group encloses items that are solely significant within a specific design method used for building analysis. • Temporal parameters: contains elements relevant to model time concepts in the building information model. The parameters are needed for example to describe schedules for energy simulations. • File parameters: this group represents auxiliary parameters that are used to describe the gbXML file itself and properties like its creator, modification date and name of the creation program. The building parameters concept group may further be divided into several distinct subgroups that again denote different fields of application. 54 parameters retrieved from gbXML are just relevant for the building performance analysis phase, e.g., parameters for specific load calculation methods, while others like the blower door value or various energyrelated parameters about equipment also prove to be useful for the operation phase of the building. Therefore a relation of parameters to the hierarchy representing parameter significance was created by means of the property hasSignificance. An example of a concrete building instantiation with selected related individuals is shown in Figure 2. Hereby, it can be noted that few relations need to be manually asserted: a sufficiently rich rule base and defined relationships enables the inference of several useful relationships (e.g., adjacent surfaces, defined construction type for surfaces). Further, it is not necessary to manually classify individual entries: the inherent structure of the OWL ontology facilitates the deduction of class membership from relations between individuals. The following section goes into detail for the defined upper ontology and elaborates possibilities that arise through the usage of OWL for building information representation. • Environmental parameter: this class contains all parameters that can be used to describe the interior and exterior environment of the building. • Structure parameter: covers information items that depict structures and more specifically materials used in the designed building. • Position parameter: parameters that are necessary to describe the global position of the building or campus, e.g., location (longitude/latitude) and elevation of the site is grouped in this class. • Geometric parameter: this set of parameters groups the items relevant for representing geometric forms in the building model and as such mainly includes concepts derived from CAD programs, e.g., height, length, polyloop or point data. 3.2 Reasoning and rule processing for queries to the semantic building information layer One of the main advantages that comes with an ontological knowledge representation in the OWL language is semantic reasoning. This can be exploited in several ways when describing building information for smart home systems. The general reasoning task in OWL languages is realized by a computational reasoner, e.g., Pellet (Sirin et al. 2007) and can be split into two parts: on the one hand the reasoning on the schema and the classes that have been defined and on the other hand the reasoning on a specific instantiation of the schema, hence in the present case the concrete building modeled in gbXML. Referring to Figure 1, there may be two different types of classes defined in an OWL ontology, primitive and defined classes (Horridge 2011): a primitive class is hereby described by subclass axioms that may be used to manually create class hierarchies or specify property relations of the concept. Defined classes on the other hand are described with equivalence axioms and allow the reasoner to classify other concepts as subclasses of the defined class as well as to reason about instance membership. An example equivalence axiom is shown in Listing 1. The defined classes IntSurface and ExtSurface of Figure 1 specify if the surface described in gbXML is internal or external. As surfaces represent abstracted walls in gbXML, this classification in other words allows the differentiation between internal and external walls. The definition of these classes is realized via a simple OWL axiom relating class membership to the value of the property Figure 2. Example of a specific building instantiation with selected relations and classifications. Typically a lot of information in OWL ontologies is represented through relations between elements, socalled properties. The transformation from gbXML already creates a wide variety of properties relating specific gbXML elements to each other. Again, the possibilities of automatically creating properties from an XML schema are limited to the expressivity of the schema. In order to create connections between the defined top-level elements, additional relations have been created, that have not been defined in gbXML and thus cannot be retrieved directly through transformation. The isBuildingElementOf relation hereby directly relates a building element to a specific building on the campus. Each equipment may be part of a campus or a building and therefore another property isEquipmentOf is defined. It is further necessary to relate each parameter represented in this ontology to a building or equipment which is realized with the isParameterOf relation. Several 55 hasExposedToSunValue retrieved from gbXML (cf. Listing 1). On the other hand also subclasses may be automatically deduced with the definition of equivalence axioms: As Figure 1 shows, the lifecycle-related parameter concepts are defined classes: The hasSignificance property is hereby responsible for creating a connection to the ParameterSignificance hierarchy and thus being able to classify parameter concepts that are significant for a specific phase. As a concrete example, the BlowerDoorValue is defined as having a significance for building operation additionally to its importance for the building analysis phase. With the axiom hasSignificance some OperationSignificance it can therefore be automatically classified by the reasoning mechanism to be a subclass of the operation phase parameter class. Another example for the need of rule extensions is the definition of inverse relations. Considering adjacent walls for a space, in gbXML surfaces are related to spaces via the AdjacentSpaceId element. The other direction, hence the relation from spaces to surfaces, may be modeled with the help of the SpaceBoundary element. While the schema provides possibilities to define both relation directions, none of them is mandatory. For example, only the first relation exists in the retrieved gbXML instance of the created test building. As it is however common to query for space objects to determine adjacent spaces and walls, a rule was created that extends the gbXML building information ontology with the property hasDefinedAdjacentSurface describing the adjacent surfaces (i.e. walls) for a specific space (cf. Listing 3). This way, the second relation direction can be inferred by the reasoning mechanism even without the existence of a SpaceBoundary element. Surface and (containsAdjacentSpaceId some IdentifierElement) and (hasExposedToSunValue value true) Listing 1. Axiom describing the membership in the defined class ExtSurface. Space(?sp1),Surface(?su1), containsAdjacentSpaceId(?su1,?aid1), hasIdValue(?sp1,?id2), hasSpaceIdRefValue(?aid1, ?id1), stringEqualIgnoreCase(?id1, ?id2) -> hasDefinedAdjacentSurface(?sp1, ?su1) Listing 3. SWRL-rule to determine the adjacent walls of a room. While at the moment the parameter significance hierarchy is only being used for subsumption reasoning, an individual member of the hierarchy could also be used for an extension of the building ontology in order to give more information about parameter usage throughout the building lifecycle (e.g., which processes it influences). Because of the fact that the data format used by gbXML is ordinary XML and as such is not as expressive as the OWL language, logical rules were used to extend the knowledge representation and simplify dependencies derived from gbXML. One of the main reasons to define rules in this particular case is the mechanism of XML to define references between elements: An element with a unique id is referenced via the XML schema simple type idref and thus creates a simple form to describe relations between objects in an XML document. In the created OWL ontology it is however desired to be able to directly relate concepts to each other without the need to match id/idref pairs in a user-defined query. In this case, SWRL rules may be used to connect concepts that are described with a reference relation in the original gbXML document (cf. Listing 2). These examples show how the internal reasoning mechanism of OWL and SWRL rule extensions may be used to enhance the information that is represented in a building information model and infer information in order to generate a more complete view of the building environment. Finally, the utilization of the building information ontology in a multi-agent controlled smart home system as one possible area of application is outlined. 4 EXAMPLE APPLICATION: A SMART HOME MULTI-AGENT SYSTEM A multi-agent system (MAS) is a software architecture for implementing a smart home system which is getting more attention recently. In the building domain, the implementation of a MAS may especially aid in overcoming conflicting situations arising when concurrently striving for optimized energy consumption and individual comfort (Mo & Mahdavi, 2003). In classic building automation building networks and operation are divided into three levels, the field, the automation and the management level (Kastner et al., 2005). A MAS hereby may be seen as an additional, abstracted level above the management level learning from the environment and acting according to global goals (Reinisch & Kastner, Opening(?o1), WindowType(?w1), hasIdValue(?w1, ?id1), hasWindowTypeIdRefValue(?o1, ?idref1), stringEqualIgnoreCase(?id1, ?idref1) -> hasDefinedWindowType(?o1, ?w1) Listing 2. SWRL-rule to connect entities referenced to each other via id and idref attributes in the original gbXML document. 56 as the building itself are some of the most important domains to be represented in the knowledge base. A semantic description of building information as it is proposed in this paper can thereby be beneficial for a variety of applications in a smart home reaching from optimized heating and cooling strategies to comfort parameters like lighting and air quality. As a concrete use case the energy efficient achievement of thermal comfort may be observed: the information represented in the building ontology can be directly used to include possible influence factors in a control strategy responsible for the thermic preparation of spaces in the building prior to expected occupancy. The knowledge about how many adjacent walls of a space are exterior walls (cf. Figure 2) combined with their structural properties and a simple representation of weather information is just one of the possibilities of how knowledge modeled in the building ontology can be useful for control strategies in multi-agent controlled smart homes. For example, in this case, a control strategy may reduce the preparation time of the space if it is known that three of four walls are adjacent to spaces that are already conditioned. Such optimizations in building control can only be reached with the representation of building information in a complete way as ontology. 2011). In this case, a knowledge base (KB) as it is presented in the present paper can be beneficial in different respects. First, it represents a shared source of universal knowledge for different agents, providing an unambiguous view of the operable world describing the entities that are of interest for a particular application. If the application is the optimization of energy efficiency, the inclusion of a building model is a valuable addition. With this information, the multi-agent system can take into account the material of the construction, orientation of a particular space and its location inside the building to optimize start/stop times of energy intensive tasks like heating and cooling. Another advantage of an ontology for a smart home system is that it facilitates the definition of relationships of sufficiently high complexity to generate a sophisticated view of the domain that the MAS operates on. Considering the example instantiation in Figure 2 it becomes obvious that a lot of inherent information can be directly modeled in the ontology, taking away complexity from the software system operating on it. This means, an ontology permits the separation of domain and operational knowledge (Noy & McGuinness, 2001) which in turn leads to a clearer system design and portable, reusable systems. The inherent form of reasoning can in this case be used to achieve a mostly complete view on the environment considerably supporting the global reasoning that needs to be achieved by the multi-agent system. 5 CONCLUSION In this article, a semantic layer representing building information as part of the environmental knowledge for an autonomously controlled smart building is proposed. Considering a BIM as source of information hereby makes it possible to rely on information already gathered by building physicists and architects and make it available to autonomous software controlling the smart building. In this respect, the present work describes the definition of a hierarchical classification scheme in the Web Ontology Language for the gbXML Building Information Model. The use of the OWL language hereby facilitates the sophisticated description of the building environment while it becomes possible to rely on inherent logical reasoning for the inference of additional facts and classification. Reasoning capabilities as well as the extension of the OWL model with logical rules are explained, while special focus is put on improving an OWL model retrieved from gbXML through a simple XML transformation. It is shown how OWL and semantic rules can further extend and simplify the relations between retrieved classes. The resulting OWL knowledge representation for gbXML is publicly available at https://www.auto.tuwien.ac.at/downloads/thinkhome /ontology/gbBuildingOntology.owl. While the general upper hierarchy of this OWL representation may be used to unify different Building Information Models by relying on a single, shared vocabulary, Figure 3. The proposed building information KB as part of the environment representation in a multi-agent system. Considering an agent-based system as one possible incarnation of a smart home control system, the building ontology shall be seen as one part of the global knowledge base representing the agents’ view of the world. To define a complete shared ontology in the building automation sector, several different domains need to be covered (cf. Figure 3): information about users, interior states of devices as well 57 Stavropoulos T. G. et al. BOnSAI: a smart building ontology for ambient intelligence. Proc of the 2nd International Conference on Web Intelligence, Mining and Semantics 2012. Article No. 30. Stumpf A. L. et al. 2011. Early Design Energy Analysis Using Building Information Modeling Technology. Champaign: US Army Corps of Engineers. Venugopal M. et al. 2012. An Ontological Approach to Building Information Model Exchanges in the Precast/PreStressed Concrete Industry. Proc. of the Construction Research Congress 2012. pp. 1114-1123. Wolf P. et al. 2009. Applying Semantic Technologies for Context-Aware AAL Services: What we can learn from SOPRANO. Workshop on Applications of Semantic Technologies 2009. (Vol. 9). the ontology is also ready to be used by system designers to integrate building physics parameters into a smart building system. Enhancing a system with detailed information about the building as such allows the definition of optimized control strategies taking into account the building structure and its physical properties. 6 REFERENCES Azhar S. et al. 2008. Building Information Modeling (BIM): A New Paradigm for Visual Interactive Modeling and Simulation for Construction Projects. Proceedings of the 1st International Conference on Construction in Developing Countries (ICCIDC-I) 2008. pp. 435-446. Beetz J. et al. 2009. IfcOWL: A case of transforming EXPRESS schemas into ontologies. Artificial Intelligence for Engineering Design, Analysis and Manufacturing 23:89101. Berners-Lee T., and Fischetti M. 1999. Weaving the Web: Origins and Future of the World Wide Web. San Francisco: Harper. Bonino D., and Corno F. 2008. DogOnt – Ontology Modeling for Intelligent Domotic Environments. Proc. of the 7th International Semantic Web Conference (ISWC) 2008. pp. 790-803. Eastman C. et al. 2008. BIM Handbook. Hoboken,NJ: John Wiley & Sons Inc. pp. 73. Gruber T. R. 1993. A translation approach to portable ontologies. Knowledge Acquisition 5(2):199-220. Hebeler J. et al. 2009. Semantic Web Programming. Hoboken, NJ: John Wiley & Sons Inc. pp. 234-257. Horridge M. et al. 2011. A Practical Guide To Building OWL Ontologies Using Protégé 4 and CO-ODE Tools. Edition 1.3. The University Of Manchester. Jahn M. et al. 2010. The Energy Aware Smart Home. Proc. of the 5th International Conference on Future Information Technology (FutureTech) 2010. pp. 1-8. Kastner W. et al. 2005. Communication Systems for Building Automation and Control. Proceedings of the IEEE 93(6):1178-1203. Kofler M. J., and Kastner W. 2009. A Knowledge Base for Energy-Efficient Smart Homes. Proc. of the 1st International Energy Conference (Energycon) 2010. pp. 85-90. Mo Z., and Mahdavi A. 2003. An agent-based simulationassisted approach to bilateral building systems control. Proc. of the 8th International Conference on Building Simulation (IBPSA) 2002. pp. 888-894. Noy N. F., and McGuinness D. L. 2001. Ontology Development 101: A Guide to Creating Your First Ontology. Stanford Knowledge Systems Laboratory Technical Report KSL-01-0. Pratt M. J. 2001. Introduction to ISO 10303 - the STEP Standard for Product Data Exchange. Journal of Computing and Information Science and Engineering 1(1):102-103. Reinisch C., and Kastner W. 2011. Agent based control in the Smart Home. Proc. of the 37th Conference on Industrial Electronics Society (IECON) 2011. pp. 334-339. Rosslin, J. R., and Tai-hoon, K. 2010. Applications, Systems and Methods in Smart Home Technology: A Review. International Journal of Advanced Science and Technology 15: 37-48. Sirin E. et al. 2007. Pellet: A practical OWL-DL reasoner. Web Semantics: Science, Services and Agents on the World Wide Web 2007. 5(2):51-53. 58 Recent advances in BIMSUSTAIN: The application of building information modeling in the context of building physics and building ecology K. Kiesel, L. Skoruppa, A. Mahdavi Vienna University of Technology, Institute for Architectural Design, Vienna, Austria ABSTRACT: Building information modeling (BIM) is a process involving the generation and management of digital representations regarding physical and functional characteristics of a construction. This paper reports on the progress of a funded research project called BIMsustain. One focus of this project is to explore the practicability of data interchange from/to building performance calculation and simulation tools. Towards this end a case study was conducted: The designing and planning process for a medium size office building was simulated including architectural design, structural engineering, energy evaluation, and HVAC aspects. Special attention was given to the processes, workflows, and data interoperability. For the designing and planning process different established software products were used and their potential for data exchange was compared. The intermediate results of this case study are presented with a special focus on the usability of building performance simulation and calculation tools in building information modeling. 1 INTRODUCTION 2 MOTIVATION AND BACKGROUND Building Information Modeling (BIM) is concerned with exchange of building data between project partners during the whole lifecycle of a building. BIM "represents the process of development and use of a computer generated model to simulate the planning, design, construction, and operation of a facility" (Azhar et al. 2008). BIM tools target structured and efficient working processes. To meet building performance and energy efficiency requirements, professionals must increasingly collaborate to design, construct and operate buildings. This highlights the importance of communication. BIM represents an interdisciplinary platform in order to optimize communication and collaboration during design, construction and the operation of buildings (Succar 2008). In this context, this present contribution reports on the preliminary results of a funded research project called BIMsustain. In order to explore the practicability of data exchange between different professions, an experimental case study was conducted that involved students of different background (architecture, civil engineering, and building physics). Various BIM tools and additional calculation and simulation software were deployed to explore BIM's potential advantages and shortcomings. 2.1 Motivation The complexity of building design and construction has been increasing, given the backdrop of changing requirements (e.g., energy efficiency, indoor climate) and advanced technical systems (Scheurer 2005). Hence, increasingly effective team decision making and multi-disciplinary collaboration is necessary in order to provide a functional, energy efficient and sustainable building (Succar 2008). Architects, civil engineers, building physicists, and building automation specialists are in need of intensive and recurrent information exchange. Building physicists can provide vital input at different levels and stages of the building delivery process toward energy efficient building design and operation. Thereby, simulation-based evaluation of design options provides a potentially powerful design support utility. However, in current practice, this possibility is used – if at all – after the building design process is more or less completed (Schlueter and Thesseling 2008). With the assistance of BIM tools, energy performance simulation could be implemented in an earlier phase and thus inform the decision making process on multiple levels. Accordingly, one of the aims of this research project is to analyze the workflow of the building design process, the use of BIM tools, and the cooperation of different professions at different states of the project development. 59 namely the Institute of Interdisciplinary Construction Process Management, the Department of Building Physics and Ecology, and the Institute of Management Science. The idea was to analyze the workflow using BIM tools in multi-disciplinary teams. Within a course offered by the three departments, students of different disciplines (architecture, civil engineering, building science) were required to use and test various BIM tools in a joint design course. The task was to design an energy efficient office building near Vienna (Austria). Table 1 gives an overview on the main requirements of this assignment. 2.2 Background Recent studies have addressed the use of BIM from various perspectives. Peterson et al. (2010) conducted a case study in order to analyze the working process with BIM tools. Students from two universities were asked to deploy BIM tools in their working process for specific courses. This helped the instructors to provide a realistic planning scenario as it occurs outside the university. As a consequence, the students learned how to organize the process efficiently toward handling real life scenarios. Schlueter and Thesseling (2008) developed a tool to integrate energy calculation in an early design stage including the use of BIM tools. This tool enables the designers to improve the building towards energy efficiency in a quick and efficient manner during the planning process. In the ‘Rosewood’-experiment (Sacks et al. 2009), the façade of a 16-story office building was drawn in CAD and additionally modeled with a BIM tool. The analysis of this study showed an efficiency gain of more than 50% compared to common 2D drawings. Recently, the academic community has realized that successful BIM deployment for integrated design is not only a technical matter. Rather, the design process itself needs to be reconsidered (Succar 2010, Penttilä 2008), including internal organization and standardization of the workflows, role descriptions and related responsibilities of the stakeholders, as well as a general commitment towards collaborative planning. Rekkola et al. (2010) argue that the knowledge regarding workflow and business practices can be crucial. In their case study, they identified problems and benefits of a BIM-supported integrated process and arranged them in terms of the following categories: (1) people (competence or knowledge problem), (2) process (workflows, timing, contracts, roles), and (3) technology (software). Furthermore, they argued that for an enhanced integrative practice, a participatory process is necessary. Moreover, the slow rate of BIM adoption in the practice is attributed to the difficulty of interrelation (triangulation) of the people-process-technology problems. In this context this paper reports on the initial results of a case study involving the deployment of BIM tools in a multidisciplinary design course. This paper focuses on the results from building science. Table 1. BIM sustain assignment requirements. type office gross area 7500 m² employees 300 location Vienna concept low energy construction concrete heating demand < 50 kWh.m-2.a-1 Several software companies supported this project with the aim to receive feedback on the usability of their BIM tools. In this regard, interfaces and data exchange possibilities were considered in addition to internal cooperation aspects. Prior to the presentation of the assignment information on the software skills of the students was collected. The idea was to let the students use software they already know. Using this information, 11 teams were put together. Each team included at least one architect (A), one civil engineer (C), and one building physicist (B). The architect’s assignment was to provide the preliminary design for a sustainable building. The civil engineer had to analyze the structure, and the building physicist was responsible for thermal performance, lighting, and ventilation concepts. The project was subdivided into different phases. In a first step, the architect generated a first concept for the design of the building. During this process, the other team members gave advice regarding structural problems and building physics. The second phase started with the handover of A's design to C and B via file export. Subsequently calculations regarding statics and energy performance followed. To do so, discipline specific tools had to be used by the teams. In phase three, the information was handed back to A. Finally the design was adjusted and specified by A. These steps had to be repeated until the aim of an energy efficient building was achieved. This process is illustrated in Figure 1. 3 METHODOLOGY The aforementioned research project BIMsustain was conducted at the Vienna University of Technology. Three departments took part in this research, 60 Archiphysik generates an energy certificate according to Austrian standards via steady state calculation. EDSL Tas is used for dynamic simulations. Dialux is a lighting simulation tool. Table 3 summarizes the software constellation in each team. Table 3. Overview of all the teams and the assigned software. Team For this process, different software applications were assigned to each team. For the architectural part Archicad, Revit, or Allpan were used. The Cs used Dlubal, Scia, and Sofistik. The Bs used Archiphysik and EDSL Tas to calculate thermal performance, Dialux for lighting, and Plancal, Allplan or Revit for ventilation. Table 2 summarizes the applied software applications. FEM SCIA SOF B MEP ALL REV 3 ARCHC TEK DLU PLA 4 ARCHC ALL DLU PLA 5 REV ALL SCIA PLA 6 ARCHC ALL DLU REV 7 ALL TEK SOF REV 8 REV TEK SCIA ALL 9 ARCHC DLU REV 10 ARCHC DLU REV 13 ARCHC REV ALL TEK TEK SOF REV TAS ARCHP DIA 4 RESULTS AND DISCUSSION The analysis of team and software constellations considered both technical workflow and interpersonal communication. Deficiencies in communication were reported by all teams. Fault tree analyses (FTA) illustrate the workflow and occurring discrepancies. Figure 2 exemplifies the FTA of one specific team. In this team, A started a draft with Archicad. To export the data from Archicad to Archiphysik, an aps-file was required. This format was generated with a plugin in Archicad (Archicad add-on). For EDSL Tas and Dialux gbxml-files were needed and again this format was generated with a plugin (Encina plugin). For Plancal and Dialux ifc-files were generated by Archicad. Table 2. Software used for the case study. Archicad 16 Allplan 2012 Revit 2013 Sofistik Tekla Dlubal Scia Plancal 8.1 EDSL Tas 9.2 Dialux 4.9 Archiphysik 10 C CAD ALL REV Due to this software variety, several interfaces and data exchange options had to be explored. In this context, exported information by each tool was analyzed towards completeness and usability. The most common data interfaces were Industry Foundation Classes (ifc 2x3) and Green Building XML (gbxml). The characteristics of these two formats differ significantly. Gbxml uses Extensible Markup Language (xml). Dong et al. (2007) describe it as a sturdy and non-proprietary format. It is a simple text-based format, which can be used for representing structured information (W3C 2013). Ifc is described as graphical modeling annotation to represent application interpreted models (Dong et al. 2007). Figure 1. Workflow of the case study. CAD CAD/MEP CAD/MEP FEM FEM FEM FEM MEP Thermal Simulation Lighting Energy Certificate 1 2 A CAD ALL REV ARCHC ALL REV SOF TEK DLU SCIA PLA TAS DIA ARCHP 61 Figure 2. Fault tree analysis of the workflow (Team 3). exported (or imported) correctly. The models showed for example missing or incomplete/incorrect building elements and room stamps. Wall intersections were occasionally not recognized. In EDSL Tas the exterior walls were not recognized and had to be redrawn. Dialux only imports the volume of rooms. Moreover, "irregular" vertical extrusions of floor plans are not recognized. In Plancal the room stamps and the intersections were not recognized. This information had to be added manually. As a consequence, the model had to be amended and reworked several times by the building physicist. The results from Archiphysik and EDSL Tas were transferred back to the architect via a performance report either in terms of a written document or orally. Dialux data cannot be directly exported back to Archicad. But there is the possibility to export the data to Plancal via stf-file. Plancal was then used to create an ifc-file and the information could be transferred back to Archicad. The Plancal data was exported to Archicad via ifc-file. Figure 3 gives an overview over the file format exchange possibilities. Linked programs use the same file format to transfer data. Software, which is not connected to the bottom applications have to use other add-ons or plugins in order to exchange information. 5 CONCLUSION These early findings of the BIMsustain experiment point to a number of challenges that interdisciplinary cooperation efforts face. In order to meet the assignment requirements, the participating students needed, aside from organizational and communication skills, the willingness and patience to deal with software constraints, especially with regard to file exchange possibilities. The import and export of files frequently resulted in incorrect models. Thus the writing or interpretation of the data was incomplete and had to Figure 3. File format exchange possibilities between assigned software. Even though the students found solutions for data transfer, the information was in most cases not 62 be corrected manually. Due to different file formats used by the applications, the data exchange processes had to be adjusted. The multitude makes the workflow more complex and time consuming. Improvement of the interoperability would make the workflow more efficient. These experiences shall be further analyzed. The lessons learned are to be integrated in designs of further future experiments involving multiple applications and students representing different disciplines. 7 REFERENCES Archicad 16, 2013. http://www.graphisoft.at (last access May 2013). Archicad add-on, 2013. http://www.archiphysik.at/?page_id=17 (last access May 2013). Archiphysik 10, 2013. http://www.archiphysik.at (last access May 2013). Allplan, 2013. http://www.nemetschek-allplan.at (last access May 2013). Autocad, 2013. http://www.autodesk.de (last access May 2013) Azhar S., Hein M., Sketo B., 2008: Building Information Modeling (BIM): Benefits, Risks and Challenges; McWhorter School of Building Science, Auburn University, Auburn, Alabama. Dialux 4.9, 2013. http://www.dial.de (last access May 2013). Dlubal REFM, 2013: http://www.dlubal.de (last access May 2013). Dong B., Lam K.P., Huang Y.C., and Dobbs G. M. 2007. A comparative study of the IFC and gbXML informational infrastructures for data exchange in computational design support environments, Porceedings: Building Simulation (2007) 1530-1537. Edsl Tas 9.2, 2013. http://www.edsl.net/main (last access May 2013). Encina plugin, 2013. http://www.encina.co.uk/swdownload.html (last access May 2013). Penttilä H., and Elger D. 2008. New Professional Profiles for International Collaboration in Design and Construction, 26th eCAADe Conference Proceedings, 17-20 September 2008, Antwerpen (Belgium), 333-340. Peterson F., Hartmann T., Fruchter R., and Fischer M. 2010. Teaching construction project management with BIM support: Experience and lessons learned, Automation in Construction 20 (2011) 115-125. Plancal 8.1, 2013. http://www.plancal.com (last access May 2013). Rekkola M., Kojima J., and Mäkeläinen T. 2010. Integrated Design and Delivery Solutions, Architectural Engineering and Design Management 6:264-278. Revit Architecture, Structure, and MEP, 2013. http://www.autodesk.de (last access May 2013). Sacks R., Kaner I., Eastman C., and Jeong Y. 2009. The Rosewood experiment – Building information modelling and interoberability for architectural precast facades, Automation in Construction 19 (2010) 419-432. Scheurer F. 2005. Getting complexity organised using selforganisation in architectural construction, Automation in Construction 16 (2007) 78-85. Schlueter A., and Thesseling F. 2008. Building information model based energy/exergy performance assessment in early design stages, Automation in Construction 18 (2009) 153-163. Scia, 2013. http://www.scia-software.de (last access May 2013). Sofistik. 2012. http://www.sofistik.de (last access May 2013). Succar B. 2008. Building information modeling framework: A research and delivery foundation for industry stakeholders, Automation in Construction 18 (2009) 357-375. Succar B. 2010. The five components of BIM performance management, in: Proceedings of CIB World Congress, Salford. Tekla Structures. 2013. http://www.tekla.com (last access May 2013). W3C 2013. The World Wide Web Consortium, http://www.w3.org/standards/xml/cor (last access May 2013). 6 ACKNOWLEDGMENTS The research presented in this paper is funded by FFG (Austrian Research Funding Agency). The following software companies support this research: A-Null, Artaker, b.i.m.m. Gasteiger, Dlubal REFM, Plancal, Sofistik, Construsoft Tekla, and Nemetschek Allplan. We gratefully acknowledge the contribution of our academic partners: Prof. C. Achammer, Dr. I. Kovacic, C. Müller, L. Oberwinter (Institute for interdisciplinary Construction Process Management), Prof. S. Kösegy, Dr. M. Filzmoser, (Institute for Management Sciences). We also acknowledge Mr. F. Brauner’s contribution towards preparing the graphical material for this paper. The research could not be conducted without the participation of a motivated group of master students of our university. 63 64 Effect of drying methods, sample sizes and RH paths on sorption isotherms C. Feng & Q. Meng Building Environment and Energy Laboratory (BEEL), State Key Laboratory of Subtropical Building Science, South China University of Technology, Guangzhou, Guangdong, P.R. China H. Janssen Building Physics Section, Department of Civil Engineering, KU Leuven, Leuven, Belgium Y. Feng China Southwest Architectural Design and Research Institute Corp. LTD, Chengdu, Sichuan, P.R. China ABSTRACT: The sorption isotherm characterizes the moisture storage of materials in the hygroscopic range, and is usually determined by the static gravimetric method. During the determination, many measures are adopted to improve test efficiency, but the influence of these measures is not well studied. In this paper, the influence of drying methods, sample sizes and intermediate RH levels in the static gravimetric tests for determining the sorption isotherms of autoclaved aerated concrete is analyzed experimentally. Oven drying at 70°C is found to give almost the same dry mass as desiccant drying at room temperature does, but is much faster than the desiccant drying method and avoids potential risks at 105°C. Moreover, large and small samples provide nearly identical equilibrium moisture content, while small samples shorten test time obviously. Last but not least, as long as one-way adsorption process is kept without reverse, the effect of intermediate relative humidity levels on the equilibrium moisture content is negligible. Sorption isotherms of porous building materials have been extensively determined all over the world. Many (inter)national projects – such as the EC HAMSTAD project (Roels, Carmeliet et al. 2003) and IEA Annex 24 project (Kumaran 1996) – have been executed, and relatively complete databases of hygrothermal properties are available for many common building materials. Usually the static gravimetric method is adopted for the determination of sorption isotherms. Many standards – such as the ISO 12571 (ISO 2000) and ASTM C1498 (ASTM 2010) – provide detailed descriptions for the test procedure. The static gravimetric method is simple and reliable. However, it usually takes several weeks or even months to complete the drying process and sorption tests (Janz and Johannesson 2001; Peuhkuri, Rode et al. 2005; Poyet 2009; Rode and Hansen 2011). To improve experimental efficiency, many measures have been taken. Unfortunately, these measures involve various factors – such as drying methods, sample sizes and intermediate RH levels – that may have significant influence on the test results but have received inadequate attention. Thus the reliability and applicability of currently available data are open to discussion. 1 INTRODUCTION 1.1 Background Moisture transfer in porous media is important in many areas, such as in food processing (Datta 2007), energy and power engineering (Missirlis, Donnerhack et al. 2010), soil science (Tuli, Hopmans et al. 2005), as well as in environmental protection (Stoltz, Gourc et al. 2010). When it comes to building physics, a good insight in and adequate control of moisture transfer processes allow us to prolong the service life of building components (Geving and Holme 2010), to reduce the energy consumption by the HVAC system (Kwiatkowski, Woloszyn et al. 2011), to mitigate indoor temperature and humidity fluctuations (Zhang, Yoshino et al. 2012), as well as to improve indoor air quality (Abuku, Janssen et al. 2009). In building physics, the sorption isotherm is often used to describe the moisture storage of building materials in the hygroscopic range. Moreover, many other hygrothermal properties – such as the thermal conductivity, water vapor permeability and moisture diffusivity – depend heavily on materials’ moisture content, and are therefore closely related to the sorption isotherm (Kumaran 1996; Roels, Carmeliet et al. 2003). Consequently, the sorption isotherm is indispensable for the modeling of materials’ hygrothermal properties and the analysis of buildings’ heat, air and moisture (HAM) performance (Steeman, Van Belleghem et al. 2009; Scheffler and Plagge 2010). 1.2 Existing uncertainties in sorption isotherm measurement The dry mass of test samples is of vital significance, because it is the basis for determining and express65 2 MATERIAL AND METHODS ing the material’s moisture content. When it comes to building materials, two drying methods are commonly used: desiccant drying at room temperature and oven drying at elevated temperature. It is widely accepted that drying methods have noticeable influence on the dry mass and hence on sorption isotherms. However, no consensus has been reached concerning the choice of the best drying method, even facing the same material (Roels, Carmeliet et al. 2003; Koronthalyova 2011; Rode and Hansen 2011). Although the desiccant drying method poses no threat to materials’ structures and chemical components, it takes a long time to completely remove physically bound water in the material. On the contrary, oven drying at elevated temperature is much more efficient, and is therefore more popular. The choice of oven temperature, however, remains controversial. One important reason for the long measurement interval in sorption tests is the adoption of large samples (over 10 grams, as required in the ISO 12571 standard (ISO 2000)). To improve test efficiency, large samples are sometimes crushed or even powdered. This strategy is very effective (Swami, Das et al. 2005; Jamali, Kouhila et al. 2006), but its effect on the equilibrium moisture content (EMC) is uncertain and not many researchers have looked into the issue (Peuhkuri, Rode et al. 2005). A first step can be made by comparing the EMC obtained with large samples (in the 10 grams order-of-magnitude) and small samples (in the 1 gram order-ofmagnitude), as is discussed in this paper. Another strategy for reducing test time is to expose different samples into different RH levels simultaneously (Roels, Carmeliet et al. 2003; Swami, Das et al. 2005; Jamali, Kouhila et al. 2006; Poyet 2009; Choudhury, Sahu et al. 2011), rather than putting the same group of samples in sequential RH levels (Peuhkuri, Rode et al. 2005; Arslan and Toğrul 2006; Rode and Hansen 2011). Obviously putting different samples into various RH levels simultaneously will skip many intermediate RH levels and thus requires large RH steps (from RH 0% to 95%, for example). Peuhkuri showed that the influence of the RH step size on the EMC and sorption curves is negligible for five measured materials (Peuhkuri, Rode et al. 2005). However, some further investigations are indispensable, because not much support is currently available. 2.1 Material description Autoclaved aerated concrete (AAC) is a popular building material in many countries, and is chosen as the target material in this paper. Test samples are cut from AAC blocks provided by Dongguan Chuangjian New Building Materials Co., Ltd. The blocks satisfy B07 density grade (the dry density is no more than 700 kg/m3) and A5.0 strength grade (the compressive strength of cubes is no less than 5.0 MPa) in China. 80 samples of two sizes, roughly 4×4×2 cm3 (18-20 g) and 3×3×1 cm3 (4-5 g), are employed. These samples are contained in small glass weighing cups (inner diameter: 6.5 cm, depth: 2.5 cm) during all drying and sorption processes. 2.2 Sample groups The bulk density of each sample is calculated with the dry mass obtained from oven drying at 70°C and the dimensions measured at room temperature. One large sample and two small ones with abnormal bulk densities are discarded, and so are redundant samples. Finally 32 large samples and 32 small ones are retained. Large samples are grouped into 8 groups of 4 duplicates. The grouping criterion is that the density distribution in each group should be similar. The small samples are similarly distributed into 8 groups. 2.3 Drying methods Three drying methods are adopted in our study: desiccant drying at room temperature, and oven drying at 70°C and 105°C. For the desiccant drying method, powdered CaCl2 has been heated to 200°C for over 24 hours and then cooled in sealed desiccators for use. Two groups of large and two groups of small samples are kept in the desiccator with desiccant at room temperature (around 20°C) for one month, and then weighed at intervals of 7-10 days. For the oven drying at 105°C, the same two groups of large samples and two groups of small samples are used. For the oven drying at 70°C, all samples are used. These samples are kept in an oven at fixed temperatures for five days and then weighed at intervals of 2-3 days. In the actual weighing process for the oven dried samples, all weighing cups (with samples inside) are taken out from the oven and tight-fitting lids are put on quickly. The weighing cups are then placed in desiccators with desiccant (CaCl2), until samples have naturally cooled to room temperature. Subsequently, weighing cups with samples are weighed by an analytical balance reading 0.0001g. After that samples are returned to oven for further drying and weighing. For the desiccant drying method, no cooling process is involved and all other processes are the same. Equilibrium is assumed to have been reached when three successive weighing results at 1.3 Objectives The aims of this paper are to investigate the influence of different drying methods on the dry mass, the influence of different sample sizes on the EMC, as well as the influence of intermediate RH levels on the EMC. 66 time needed for the cooling process, which leads to experimental errors. To sum, desiccant drying is the best in terms of the weighing precision. But basically the weighing precisions are satisfactory for all three drying methods. intervals of at least 48 hours do no longer show monotonous mass evolutions and when their relative deviations are below 0.1%. 2.4 Adsorption and desorption tests The static gravimetric method is used to determine the EMC of samples. Eight desiccators with different saturated salt solutions are placed in a climate chamber where temperature is 25±0.3°C. The RH in desiccators ranges from 11.3% to 97.3%. Oven dried samples (at 70°C) are put into desiccators for adsorption tests. Each desiccator holds one group of large samples and one group of small ones. Samples are weighed at intervals of at least two days. When three successive weighing results are within 0.1% change, equilibrium is assumed to have been reached. After adsorption samples at higher RH levels are subsequently placed into lower RH levels for desorption tests. The starting RH levels of desorption are different, so points for different scanning desorption curves are obtained. For each test condition in desorption tests one group of large samples and one group of small ones are used. 3.1.2 Dry mass comparison The direct comparison of all samples’ average dry mass obtained from three different drying methods is meaningless because samples of roughly the same size do not necessarily have the same dry mass. To solve this problem, we set the dry mass from one drying method as the reference and convert the dry mass from the other two methods accordingly. Since desiccant drying at room temperature is the most stable in terms of the weighing precision, we set the dry mass obtained from this method as 100 units. Table 1 summarizes the average values of converted dry mass and the standard deviations. According to Table 1, the dry mass obtained from desiccant drying at room temperature and that obtained from oven drying at 70°C are quite close. The mean difference is merely 0.14%, and can most probably be attributed to small differences in the final levels of RH: the RH in the desiccator for desiccant drying is less than 2%, while it is about 4% in the oven at 70°C (the oven takes in air from the lab environment). To sum, oven drying at 70°C and desiccant drying at room temperature can give almost the same dry mass, except for minor differences. Table 1 furthermore shows that the dry mass obtained from oven drying at 105°C is over 1% smaller than that obtained from desiccant drying at room temperature. This difference is quite significant, because it can result in more than 1% EMC difference (kg/kg). For a material with small EMC (such as AAC), such deviation is rather big. The main reason for this difference is that water in materials exists in various physical or chemical states. Besides physically bound water, oven drying at 105°C also evaporates chemically bound water (Al-Muhtaseb, McMinn et al. 2002). Last but not least, for large and small samples the converted dry mass from the same drying method is almost identical, and the differences between large and small samples are much smaller than the differences among different drying methods. This indicates that the differences in dry mass originate from drying methods, rather than from sample’s sizes. 3 RESULTS AND DISCUSSION 3.1 Drying methods 3.1.1 Weighing precision The dry mass of each sample is taken as the mean value of three successive weighing results at intervals of at least 48 hours showing non-monotonous relative deviations below 0.1%. The standard deviation of these three weighing results reflects the fluctuations of one sample’s dry mass readings. By averaging the relative standard deviation of all samples, we can obtain an indicator of both the weighing errors involved and the differences between samples. As the differences between samples within groups can be assumed reasonably similar from group to group, the calculated results should mainly reflect the stability of different drying methods. The detailed figures are 0.034%, 0.062% and 0.089% for desiccant drying at room temperature, oven drying at 70°C, as well as oven drying at 105°C respectively, showing that the desiccant drying method provides the most stable dry mass readings while oven drying at 105°C comes last. This can be explained by the Table 1. Comparison of dry mass from different drying methods. Desiccant drying at Drying method Oven drying at 70°C Oven drying at 105°C room temperature 100.00(0) 100.13(0.08) 98.98(0.04) Mean relative Large samples dry mass * Small samples 100.00(0) 100.16(0.06) 98.94(0.03) *Standard deviations are indicated in parentheses 67 EMC(% kg/kg) 8 7 6 5 4 3 2 1 0 0 the AAC blocks from which large and small samples are cut. Thus the EMC differences should be most probably attributed to these porosity differences, rather than reflecting the actually differences between large and small samples. Last but not least, from the results of the EC HAMSTAD project (Roels, Carmeliet et al. 2003), in which many round robin tests are performed to determine the hygrothermal properties of three building materials (the autoclaved aerated concrete, the calcium silicate board and the ceramic brick), we can easily tell that the EMCs of the same material at the same RH level obtained by different laboratories are noticeably different. The discrepancies among laboratories are much larger than the EMC differences between large and small samples in our case. To sum, large and small samples can provide almost the same EMCs throughout the whole RH range, except for some minor differences. large samples (4×4×2 cm3) small samples (3×3×1 cm3) 20 40 RH(%) 60 80 100 EMC(% kg/kg) a. Adsorption process 8 7 6 5 4 3 2 1 0 0 large samples (4×4×2 cm3) small samples (3×3×1 cm3) 20 40 RH(%) 60 80 3.2.2 Test efficiency It is well known that by using small samples the test efficiency of sorption tests can be improved dramatically. In our adsorption tests, small samples save up to 50 days to reach equilibrium (different improvements for different RH paths), while the overall test period for large samples is nearly 3 months. Obviously the improvement is impressive. Moreover, the advantage of small samples could be further enhanced. At the moment, the combination of the still air in the desiccator and of the use of weighing cups yields a slow vapour exchange with the samples, making the surface vapor resistance a dominant factor for the diffusion into or out of the samples. Installation of small fans in the desiccators can substantially reduce such transfer resistances, and in that case smaller, thinner samples will become even more advantageous. In our current follow-up tests with such set-up, it takes less than one month for samples to reach equilibrium, even the sorption path is from the oven dried state to RH 97.3%. 100 b. Desorption process Figure 1. The EMC of large and small AAC samples at 25°C. 3.1.3 Conclusion on drying methods The test efficiency, material protection and weighing precision can be used as the criteria for choosing the best drying method. Oven drying at 70°C can accelerate the drying process impressively and achieve almost the same dry mass as desiccant drying at room temperature does, without sacrificing experimental precision or introducing severe deformation to the material. Thus we recommend oven drying at 70°C as the best drying method for AAC, as well as for other materials which do not experience chemical reactions or irreversible structural changes at 70°C. All data reported in the following parts of this paper are based on the dry mass obtained from oven drying at 70°C. 3.2.3 Conclusion on sample sizes Firstly, small samples improve test efficiency a lot. Secondly, small samples provide almost the same results as large samples do without sacrificing experimental precision. Consequently, we advocate small samples. However, it should be kept in mind that the samples should not be too small in comparison to the accuracy of the balance – the accuracy of the balance should be at least 0.01-0.1 percent of a sample’s mass, as required by ISO 12571 (ISO 2000) and ASTM C1498 (ASTM 2010). It should be noted that, due to the original experimental arrangement, we carry out most tests with large samples. To keep things consistent, all data reported in the following part of this paper are based on the EMCs of large samples. 3.2 Sample sizes 3.2.1 EMC comparison Figure 1 illustrates the EMCs of large and small samples. It is worth mentioning that the EMCs of small samples at RH 84.3% in desorption are flawed by experimental mistakes, and thus not shown. Fig.1 shows that the EMCs of large samples slightly exceed the EMCs of small ones throughout the studied RH range. However, the differences are small when compared with the standard deviations of duplicate samples in the same group. In addition, as revealed by vacuum saturation tests, large and small samples have average open porosities of 75.5% and 74.7% respectively. This difference stems from the different original positions in 68 Table 2. EMC comparison for different RH paths (adsorption). EMC (% kg/kg) Temperature RH path (%) Standard (°C) Mean deviation Oven dried-11.3-33.3 0.93 0.08 15 Oven dried-33.3 0.82 0.06 Oven dried-33.3-43.2 0.98 0.06 15 Oven dried-43.2 1.06 0.16 Oven dried-43.2-75.3 1.75 0.12 25 Oven dried-75.3 1.88 0.13 Oven dried-75.3-93.6 3.12 0.12 25 Oven dried-93.6 3.28 0.12 c. The effect of intermediate RH levels on the EMC is negligible, as long as one-way adsorption process is kept without reverse. It is therefore acceptable to expose samples into different RH levels simultaneously to speed up the overall tests. 5 ACKNOWLEDGEMENTS This project is supported by National Key Technology R&D Program of China (Project No. 2011BAJ01B01). The authors express sincere thanks to Professor Yufeng Zhang, as well as Master students Chenchen Wu, Xi Yu and Chao Jiang in South China University of Technology for their help in carrying out the experiments. 3.3 Intermediate RH levels The influence of intermediate RH levels on the EMC is analyzed by comparing the EMCs of samples with different sorption histories throughout the whole RH range. Table 2 summarizes the results. It should be noted that due to experimental arrangement, not all tests in Table 2 are conducted at the same temperature. But this should have no influence on the conclusion. It is clear that the EMCs in the adsorption process remain almost the same if the final RH level is the same. In most cases the differences are less than 0.1% (kg/kg) while the maximum difference is merely 0.16% (kg/kg). These limited differences are in the same order of the EMC scatters within groups and are negligible in most situations. Thus we observe that the EMC of samples in the adsorption process is not affected by samples’ previous sorption histories, as long as a one-way adsorption process is imposed. It is therefore reasonable to expose many samples into different RH levels simultaneously to speed up the overall adsorption tests. 6 REFERENCES Abuku M., Janssen H. et al. 2009. Impact of wind-driven rain on historic brick wall buildings in a moderately cold and humid climate: Numerical analyses of mould growth risk, indoor climate and energy consumption. Energy and Buildings 41(1): 101-110. Al-Muhtaseb A.H., McMinn W.A.M. et al. 2002. Moisture Sorption Isotherm Characteristics of Food Products: A Review. Food and Bioproducts Processing 80(2): 118-128. Arslan N., and Toğrul H. 2006. The fitting of various models to water sorption isotherms of tea stored in a chamber under controlled temperature and humidity. Journal of Stored Products Research 42(2): 112-135. ASTM 2010. ASTM C1498 - 04a: Standard Test Method for Hygroscopic Sorption Isotherms of Building Materials. Choudhury D., Sahu J.K. et al. 2011. Moisture sorption isotherms, heat of sorption and properties of sorbed water of raw bamboo (Dendrocalamus longispathus) shoots. Industrial Crops and Products 33(1): 211-216. Datta A.K. 2007. Porous media approaches to studying simultaneous heat and mass transfer in food processes. II: Property data and representative results. Journal of Food Engineering 80(1): 96-110. Gallé C. 2001. Effect of drying on cement-based materials pore structure as identified by mercury intrusion porosimetry: A comparative study between oven-, vacuum-, and freezedrying. Cement and Concrete Research 31(10): 1467-1477. Geving S., and Holme J. 2010. The Drying Potential and Risk for Mold Growth in Compact Wood Frame Roofs with Built-in Moisture. Journal of Building Physics 33(3): 249269. Hagentoft C.E., Kalagasidis A.S. et al. 2004. Assessment method of numerical prediction models for combined heat, air and moisture transfer in building components: benchmarks for one-dimensional cases. Journal of Thermal Envelope and Building Science 27(4): 327-352. ISO 2000. ISO 12571: 2000(E) Hygrothermal performance of building materials and products - Determination of hygroscopic sorption properties. 4 CONCLUSIONS This paper focuses on the ad- and desorption isotherms of autoclaved aerated concrete (AAC) at 25°C, for which the influence of drying methods, sample sizes and intermediate RH levels is analyzed. Results are summarized as follows: a. The dry mass obtained from oven drying at 70°C is almost the same as that obtained with desiccant drying at room temperature, while oven drying at 105°C results in 1.04% lower dry mass than desiccant drying at room temperature does. Based on test efficiency, weighing precision and material protection, oven drying at 70°C is recommended. b. In both adsorption and desorption processes the EMCs of large and small samples are close. The minor difference can be explained by the slightly different open porosities of large and small samples. Small samples reduce test time obviously, and are therefore preferable. 69 Jamali A., Kouhila M. et al. 2006. Sorption isotherms of Chenopodium ambrosioides leaves at three temperatures. Journal of Food Engineering 72(1): 77-84. Janz M.Å. & Johannesson B.Ö.F. 2001. Measurement of the moisture storage capacity using sorption balance and pressure extractors. Journal of Building Physics 24(4): 316334. Koronthalyova O. 2011. Moisture storage capacity and microstructure of ceramic brick and autoclaved aerated concrete. Construction and Building Materials 25(2): 879885. Korpa A., and Trettin R. 2006. The influence of different drying methods on cement paste microstructures as reflected by gas adsorption: Comparison between freezedrying (F-drying), D-drying, P-drying and oven-drying methods. Cement and Concrete Research 36(4): 634-649. Kumaran M.K. 1996. IEA Annex 24: Heat, Air and Moisture Transfer in Insulated Envelope Parts. Final Report, Volume 3, Task 3: Material Properties. Kwiatkowski J., Woloszyn M. et al. 2011. Influence of sorption isotherm hysteresis effect on indoor climate and energy demand for heating. Applied Thermal Engineering 31(6-7): 1050-1057. Missirlis D., Donnerhack S. et al. 2010. Numerical development of a heat transfer and pressure drop porosity model for a heat exchanger for aero engine applications. Applied Thermal Engineering 30(11-12): 1341-1350. Peuhkuri R., Rode C. et al. 2005. Effect of method, step size and drying temperature on sorption isotherms. 7th Nordic Symposium on Building Physics. Poyet S. 2009. Experimental investigation of the effect of temperature on the first desorption isotherm of concrete. Cement and Concrete Research 39(11): 1052-1059. Rode C., and Hansen K.K. 2011. Hysteresis and temperature dependency of moisture sorption - new measurements. 9th Nordic Symposium on Building Physics. Roels S., Carmeliet J. et al. 2003. HAMSTAD Work Package 1: Final Report - Moisture Transfer Properties and Materials Characterisation. Scheffler G.A., and Plagge R. 2010. A whole range hygric material model: Modelling liquid and vapour transport properties in porous media. International Journal of Heat and Mass Transfer 53(1-3): 286-296. Steeman H.J., Van Belleghem M., et al. 2009. Coupled simulation of heat and moisture transport in air and porous materials for the assessment of moisture related damage. Building and Environment 44(10): 2176-2184. Stoltz G., Gourc J.-P. et al. 2010. Liquid and gas permeabilities of unsaturated municipal solid waste under compression. Journal of Contaminant Hydrology 118(1-2): 27-42. Swami S.B., Das S.K. et al. 2005. Moisture sorption isotherms of black gram nuggets (bori) at varied temperatures. Journal of Food Engineering 67(4): 477-482. Tuli A., Hopmans J.W. et al. 2005. Comparison of air and water permeability between disturbed and undisturbed soils. Soil Science Society of America Journal 69(5): 1361-1371. Zhang H., Yoshino H. et al. 2012. Assessing the moisture buffering performance of hygroscopic material by using experimental method. Building and Environment 48(0): 2734. 70 Effect of temperature on the sorption isotherm and vapor permeability C. Feng & Q. Meng Building Environment and Energy Laboratory (BEEL), State Key Laboratory of Subtropical Building Science, South China University of Technology, Guangzhou, Guangdong, P.R. China H. Janssen Building Physics Section, Department of Civil Engineering, KU Leuven, Leuven, Belgium Y. Feng China Southwest Architectural Design and Research Institute Corp. LTD, Chengdu, Sichuan, P.R. China ABSTRACT: The sorption isotherm and water vapor permeability are important hygrothermal properties of porous building materials. However, adequate data at different temperatures are lacking for analyzing potential temperature dependence. This paper studies the influence of temperature on the sorption isotherm and water vapor permeability of autoclaved aerated concrete by performing tests at 15°C, 25°C and 35°C. The sorption isotherm is measured by the static gravimetric method while water vapor permeability is measured via the cup tests. Results show that no consistent pattern of temperature dependence can be observed. Moreover, the fitting deviations of fitting all experimental data obtained at different temperatures into a single overall curve are only slightly larger than the fitting deviations of fitting data at respective temperatures with separate curves. To sum, the influence of temperature on the sorption isotherms and water vapor permeability of autoclaved aerated concrete is negligible in the range of 15-35°C. EC HAMSTAD (Roels, Carmeliet et al. 2003) and the IEA Annex 24 (Kumaran 1996) – have been executed, and a huge number of independent tests have been carried out by a lot of research institutes (Quenard, Xu et al. 1998; Beck, Al-Mukhtar et al. 2003; Mukhopadhyaya, Kumaran et al. 2005; Koronthalyova 2011). As a result, the data of the sorption isotherm and water vapor permeability are available for a huge number of building materials. Unfortunately, most reported tests are carried out at only one temperature (usually at room temperature, 20-25°C), thus the temperature dependence of these hygrothermal properties remains largely unknown. Notwithstanding that some researchers have looked into it, no consensus on this issue has been reached yet. It is therefore of great significance to perform more tests to study the influence of temperature on the hygroscopic storage and transport properties of porous building materials. This paper takes autoclaved aerated concrete (AAC) as the test material, and determines its sorption isotherms and water vapor permeability at 15°C, 25°C and 35°C respectively, in order to analyze the influence of temperature on these two properties. 1 INTRODUCTION Moisture transfer in porous media is important in many areas, such as in food processing (Datta 2007), soil science (Tuli, Hopmans et al. 2005), energy and power engineering (Missirlis, Donnerhack et al. 2010), and environmental protection (Stoltz, Gourc et al. 2010). When it comes to building physics, a good insight in and an adequate control of moisture transfer processes help prolong the service life of building components (Geving and Holme 2010), to reduce the energy consumption by the HVAC systems (Kwiatkowski, Woloszyn et al. 2011), to mitigate indoor temperature and humidity fluctuations (Zhang, Yoshino et al. 2012), as well as to improve indoor air quality (Abuku, Janssen et al. 2009). In building physics, the sorption isotherm and water vapor permeability are the most important hygric material properties of porous building materials in the hygroscopic range. The sorption isotherm describes a material’s capacity for moisture storage and is often determined by sorption tests at given relative humidity (RH) levels (as described in the ISO 12571 standard (ISO 2000)), while the water vapor permeability characterizes a material’s capability for vapor transmission and is usually determined via cup tests where fixed vapor pressure gradients are maintained (as described in the ISO 12572 standard (ISO 2001)). During the past two decades, many (inter)national projects concerning material properties – such as the 2 MATERIAL AND METHODS 2.1 Material description Autoclaved aerated concrete (AAC) is popular in many countries, and is chosen as the target material 71 settings. In most cases, the RH level for preconditioning is the lower one of the cup pair, and resultantly an adsorption process is imposed. For instance, if a sample is to be used in the cup setting with RH 43.2% and 84.3%, it is first oven dried, and then allowed to absorb moisture at RH 43.2% until a constant mass has been reached. However, at 25°C, we also have three groups of samples undergo desorption processes from the higher RH levels of cups’ RH pairs for pre-conditioning. These three groups of samples are used in cup sets with RH 043.2%, 43.2-84.3% and 84.3-97.3% respectively. After pre-conditioning, samples are sealed onto the cups for water vapor transmission tests. For each condition, one or two groups of samples are used. When the vapor transmission rates have reached constant levels, the total mass of samples and cups are weighed seven times and the water vapor permeability of each sample is calculated. The balance used for assembly weighing has an accuracy of 0.01g. After water vapor transmission tests, samples are broken and the moisture content of their central parts is determined. in this paper. Test samples are cut from AAC blocks provided by Dongguan Chuangjian New Building Materials Co., Ltd. The blocks are of B07 density grade (the dry density is no more than 700 kg/m3) and A5.0 strength grade (the compressive strength of cubes is no less than 5.0 MPa) in China. 2.2 Sorption tests For the determination of sorption isotherms the static gravimetric method is adopted, and the ISO 12571 standard (ISO 2000) is employed, albeit with some modifications. AAC blocks are cut into samples sized 4×4×2 cm3 for the sorption tests. After oven drying at 70°C, the bulk density of each sample is calculated from its dry mass and dimensions. Then samples are grouped into 8 groups of 4 duplicate samples according to their bulk densities, so that the density distributions in each group are similar. After that eight desiccators with different saturated salt solutions are placed in a climate chamber where temperature is controlled within ±0.3°C, with the RH in the desiccators ranging from 11.3% to 97.3% (at 25°C). Each desiccator holds one group of samples for adsorption. When constant mass of all samples in one group has been reached, groups at higher RH levels are subsequently placed into lower RH levels for desorption. Each sample’s equilibrium moisture content (EMC, kg/kg) is then calculated based on its dry mass and wet mass. It should be noted here that the starting RH levels of desorption tests are different, so the obtained points are for different main/primary desorption curves. Equilibrium is assumed to have been reached when three successive weighing results at intervals of at least 48 hours do no longer show monotonous mass evolutions and when their relative deviations are below 0.1%. 3 RESULTS AND DISCUSSION 3.1 Test results For the sorption isotherms, all samples’ EMCs are plotted against the related ambient RH levels, as illustrated in Figure 1 and Figure 2. When it comes to the water vapor permeability, it is common practice to express the permeability in function of the average RH of the cup pair. This expression method is depicted in Figure 3. However, a closer look into the test results at 25°C (where samples pre-conditioned in different ways are used with the same cup RH pairs) reveals that the influence of pre-conditioning is rather obvious: even when using the same cup RH pair (thus the same average RH), samples pre-conditioned to different EMCs via ador desorption processes show different permeabilities. This can be easily explained by the hysteresis phenomenon and the fact that it is the moisture content of a sample, rather than the setting of RH levels in cups, that directly influences the water vapor permeability. It is therefore more reasonable to plot the water vapor permeabilities against samples’ respective moisture content, as is shown in Figure 4. 2.3 Cup tests For the determination of water vapor permeability the cup method is adopted, and the ISO 12572 standard (2001) is employed, albeit with some modifications. AAC blocks are cut into samples sized 12 cm in diameter and 3 cm in thickness for cup tests. Similarly, samples are grouped into groups according to their bulk densities, and each group contains 3 duplicate samples. Three sets of cups are prepared and placed in a climate chamber where temperature is controlled within ±0.3°C. Desiccant and saturated salt solutions are used to control the RH levels in the cups. The RH pairs in different cup sets are 0-43.2%, 43.284.3% and 84.3-97.3% respectively (at 25°C). Samples are pre-conditioned at different RH levels before the water vapor transmission tests. The pre-conditioning temperature is the same as the test temperature, while the RH level depends on the cup 3.2 Qualitative analysis To judge the influence of temperature, we refer to some existing assumptions. Most observations of the effect of temperature on ad- and desorption isotherms indicate a negative correlation between EMC and temperature: a higher temperature leads to a lower EMC at any given RH level (Al-Muhtaseb, 72 1.5 1.0 0.5 Permeability(×10-4 g/m⋅h⋅Pa) 3.0 15 oC 25 oC 35 oC 2.5 2.0 1.5 1.0 0.5 Figure 4. Water vapor permeability at different temperatures (in function of moisture content). 4 15 oC 25 oC 35 oC 2 0 -2 20 40 60 RH(%) 80 -4 0 100 Figure 1. EMC at different temperatures (adsorption). 16 4 15 oC 25 oC 35 oC 12 20 40 60 RH(%) 80 100 Figure 5. Residuals of overall adsorption curve fitting. Residuals (% kg/kg) EMC(kg/kg,%) 2.0 Moisture content(kg/kg,%) 4 8 15 oC 25 oC 35 oC 2 0 -2 4 0 0 2.5 0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 8 0 0 15 oC 25 oC 35 oC RH(%) Figure 3. Water vapor permeability at different temperatures (in function of relative humidity). 15 oC 25 oC 35 oC 12 3.0 0.0 0 10 20 30 40 50 60 70 80 90 100 Residuals (% kg/kg) EMC(% kg/kg) 16 Permeability (×10-4 g/m⋅h⋅Pa) McMinn et al. 2002; Basu, Shivhare et al. 2006). Moreover, when the same saturated solutions are used at different temperatures, their respective ambient RHs go down when temperature goes up (ISO 2000). We thus expect to observe that EMCs obtained at a higher temperature do not exceed those obtained at a lower temperature. However, as is obviously reflected in Figures 1 and 2, this is not readily noted. Generally speaking, the influence of temperature on the sorption isotherms does not appear consistent or significant. For the water vapor permeability, it is commonly accepted that a material’s water vapor permeability changes with temperature in the same way as still air does (Kumaran 1994). A simple calculation reveals that from 15°C to 35°C, the water vapor permeability of still air increases by 5%. However, a higher temperature reduces the RHs in the cups (the same saturated salt solutions) and thus reduces the samples’ moisture content indirectly. Consequently temperature has contradictory influence on our experimental setting, and no expected outcome can be predicted here. Again however, Figure 4 does not reveal any consistent or significant influence of temperature on the water vapor permeability. In what follows, these initial qualitative analyses are complemented with a quantitative examination of the results. 20 40 60 RH(%) 80 -4 0 100 20 40 60 RH(%) 80 100 Figure 6. Residuals of overall desorption curve fitting. Figure 2. EMC at different temperatures (desorption). 73 Residuals (×10-5 g/m⋅h⋅Pa) 12 8 4 15°C, 25°C and 35°C respectively), the internal scatters within groups are quite dramatic. Moreover, in the adsorption tests, 8 saturated salt solutions are used for EMC determination while in the desorption tests only 7 (K2SO4 is not used). To avoid the impact of these very large scatters (which may confuse the analysis), and to make the comparison between adsorption and desorption more consistent, we do not use the data obtained for the K2SO4 solution in the adsorption process in the following analysis. 15 oC 25 oC 35 oC 0 -4 -8 -12 0.00 0.01 0.02 0.03 Moisture contetn(kg/kg) 0.04 Figure 7. Residuals of overall permeability fitting. 3.3.2 Mean deviations from the overall fitting curve For our quantitative analysis, we define the mean as: deviation from the overall fitting curve D overall fit 3.3 Quantitative analysis D overall = fit To quantitatively analyze the influence of temperature on the sorption isotherms and water vapor permeability, we first perform the curve fitting in different ways, then analyze the fitting results, and finally look into experimental errors. where xij is the measured value for duplicate i in group j while x overall is the predicted value at condipj tion j according to the overall fitting curve. For the sorption tests, i and j represent all samples (i) in the group exposed to the same RH level (j), while for the cup tests, i and j represent all samples (i) preconditioned in the same way and used in the same RH pair (j). If any consistent and significant influence of temperature exists, then the signs for D overall at 15°C fit and 35°C should always be opposite – because points obtained at these two temperatures should generally distribute on the opposite sides of the overall fitting curve – and the absolute values of at 25°C should be the smallest and close to D overall fit 0. However, the results in Table 1 do not comply with these trends. For the adsorption isotherm, the signs of D overall are negative at 15°C and 25°C, and fit positive at 35°C. At the same time, it is exactly opposite for the desorption isotherm. When it comes to the water vapor permeability, no consistent trend is values are negaobserved either, since the D overall fit tive at 15°C and 35°C, and positive at 25°C. These inconsistent deviations can also be identified visually in Figures 5, 6 and 7, where we find alternatively positive and negative residuals. Moreover, the absolute values of deviations at 25°C in Table 1 are not always the smallest. These findings hence indicate that the influence of temperature does not supersede experimental errors and thus cannot be clearly identified, at least not in our test temperature range. 3.3.1 Curve fitting To achieve the quantitative analysis, we fit the obtained data points with continuous sorption and water vapor permeability curves. As a very good fit is of the utmost importance, we prefer the elaborate expression below over other more standard formulations for sorption isotherms: u = k1 ln(ϕ + 1) + k2 ln(100 − ϕ ) + k3 exp(ϕ ) (1) − k2 ln 100 − k3 where u is the EMC (kg/kg), ϕ is the RH (%), and ki are fitting parameters. In the desorption process the obtained data points are for different main/primary curves and thus fitting them with one curve has no physical meaning. However, this is necessary for the statistical comparison and in this paper we still call it desorption curve/isotherm. For water vapor permeability, the following standard formula is adopted, since it provides quite good fitting results: (2) µ = k1 + k2u k 3 where µ is water vapor permeability (g/m·h·Pa), u is moisture content of samples (kg/kg) and ki are fitting parameters. This formula has the same form as suggested by Galbraith (Galbraith, Tao et al. 1993), but the independent variable used here is not the average RH of cup pairs but moisture content. Data obtained at different temperatures are fitted separately with Eq.(1) and Eq.(2) first (the separate fittings). Then all points obtained at three different temperatures are taken together and fitted with a single curve (the overall fitting, shown in Figures 1, 2 and 4). Figures 5-7 illustrate the residual plots of the overall curve fitting. As is clearly reflected in Figure 5, at the highest RH levels (where saturated K2SO4 solution is used, with resultant RHs 98.2%, 97.3%, and 96.4% at 1 1 ) ∑ ∑ (xij − x overall pj j j  i i  (3) Table 1. Mean deviations from the overall fitting curve. AdsorpVapor transDesorption Temp./°C tion* mission** 15 -0.075 0.051 -1.20 25 -0.072 0.067 4.18 35 0.150 -0.129 -2.98 *As in following tables, the unit of all deviations for sorption processes is kg/kg (%) ** As in following tables, the unit of all deviations for vapor transmission is ×10-6 g/(m·h·Pa) 74 Table 2. Enlarged deviations in the overall fitting. Process Adsorption Desorption Vapor transmission Temp./°C D fit ,abs D overall fit ,abs 15 25 35 Average 15 25 35 Average 15 25 35 Average 0.083 0.077 0.114 0.091 0.188 0.096 0.125 0.136 4.94 11.75 8.69 8.46 0.124 0.106 0.185 0.138 0.248 0.150 0.210 0.203 6.83 12.04 11.72 10.20 As is clearly reflected in Table 2, the enlarged deviations are very small when compared to the deviations for separate fitting curves at different temperatures. This, from another perspective, shows that no significant influence of temperature can be observed. D overall fit ,abs − D fit ,abs 0.041 0.029 0.071 0.047 0.060 0.054 0.085 0.067 1.89 0.20 3.01 1.70 3.3.4 Experimental errors To evaluate experimental errors, we refer to the mean absolute deviation from the group average ( Dgroup ,abs ), as defined by: 1 1 (6) ( ∑ xij − x j ) ∑ j j i i where x j is the average value of all duplicates of group j and other symbols have the same meanings as in previous equations. Table 3 summarizes the calculation results. Clearly, the scatter within each group ( Dgroup ,abs ) is larger than the enlarged deviation resulting from adopting the overall fit, showing that any impact of temperature in our test range is smaller than our experimental uncertainty, and thus again not significant. Dgroup ,abs = Table 3. Mean absolute deviations from group average. Vapor Desorption Temp./°C Adsorption transmission 15 0.075 0.136 2.49 25 0.073 0.084 4.36 35 0.082 0.077 9.27 Average 0.077 0.099 5.37 3.3.3 Enlarged deviations in the overall fitting In section 3.3.2, we calculate the mean deviations from the overall fitting curve. However, it is entirely possible that the deviations of data points obtained at one temperature from the separate fitting curves for different temperatures are smaller. Moreover, if the influence of temperature is significant, then the difference between deviations from the overall fitting curve and from the separate fitting curve should be obvious. We define the mean absolute deviation from the separate fit ( D fit ,abs ) and the mean absolute deviation 4 CONCLUSIONS The influence of temperature on the sorption isotherm and water vapor permeability of autoclaved aerated concrete in the range of 15-35°C is studied experimentally in this paper. Following conclusions are obtained: a. By neglecting the influence of temperature and fitting all data points obtained at different temperatures into an overall curve, deviations of measured results from fitting curves are enlarged. However, for both sorption isotherms and vapor permeability, these enlargements are much smaller than our experimental errors. b. The pattern of temperature’s influence can be analyzed by the mean deviation from the overall fitting curve. However, no systematical pattern can be observed on either sorption isotherms or on vapor permeability. To sum up, neither consistent nor significant influences of temperature on the sorption isotherms and water vapor permeability of autoclaved aerated concrete in the range of 15°C to 35°C are observed. from the overall fit ( D overall fit ,abs ) as: 1 1 ( ∑ xij − x pj ) ∑ j j i i 1 1 ) = ∑ ( ∑ xij − x overall pj j j i i D fit ,abs = (4) D overall fit ,abs (5) where all the symbols remain the same as in Eq.(3) except that in Eq.(4) the predicted value according to the overall fit ( x overall ) is replaced by the predicted pj value according to separate fits at different temperatures ( x pj ). The reason why we use absolute values here is to avoid the elimination of positive and negative deviations. Obviously, D overall fit ,abs − D fit ,abs stands 5 ACKNOWLEDGEMENTS This project is supported by National Natural Science Foundation of China (No. 51278478). The authors express sincere thanks to Professor Yufeng Zhang, as well as Master students Chenchen Wu, Xi Yu and Chao Jiang in South China University of Technology for their help in the experiments. for the enlarged deviations in the overall fitting by neglecting the influence of temperature and fitting all points into an overall curve. Table 2 summarizes the calculation results. 75 Tuli A., Hopmans J.W. et al. 2005. Comparison of air and water permeability between disturbed and undisturbed soils. Soil Science Society of America Journal 69(5): 1361-1371. Zhang H., Yoshino H. et al. 2012. Assessing the moisture buffering performance of hygroscopic material by using experimental method. Building and Environment 48(0): 2734. 6 REFERENCES Abuku M., Janssen H. et al. 2009. Impact of wind-driven rain on historic brick wall buildings in a moderately cold and humid climate: Numerical analyses of mould growth risk, indoor climate and energy consumption. Energy and Buildings 41(1): 101-110. Al-Muhtaseb A.H., McMinn W.A.M. et al. 2002. Moisture Sorption Isotherm Characteristics of Food Products: A Review. Food and Bioproducts Processing 80(2): 118-128. Basu S., Shivhare U.S. et al. 2006. Models for Sorption Isotherms for Foods: A Review. Drying technology 24(8): 917-930. Beck K., Al-Mukhtar M. et al. 2003. Characterization, water transfer properties and deterioration in tuffeau: building material in the Loire valley – France. Building and Environment 38(9-10): 1151-1162. Datta A.K. 2007. Porous media approaches to studying simultaneous heat and mass transfer in food processes. II: Property data and representative results. Journal of Food Engineering 80(1): 96-110. Galbraith G.H., Tao Z, et al. 1993. Separation of moisture flow through porous building materials into vapour and liquid components. Building Services Engineering Research and Technology 14(3): 107-113. Geving S., and Holme J. 2010. The Drying Potential and Risk for Mold Growth in Compact Wood Frame Roofs with Built-in Moisture. Journal of Building Physics 33(3): 249269. ISO 2000. ISO 12571: 2000(E) Hygrothermal performance of building materials and products - Determination of hygroscopic sorption properties. ISO 2001. ISO 12572: 2001(E) Hygrothermal performance of building materials and products - Determination of water vapour transmission properties. Koronthalyova O. 2011. Moisture storage capacity and microstructure of ceramic brick and autoclaved aerated concrete. Construction and Building Materials 25(2): 879885. Kumaran M.K. 1994. Temperature dependence of water vapour permeability of glass fiber insulation from heat flow meter measurements. First ISHMT-ASME Heat and Mass Transfer Conference. Kumaran M.K. 1996. IEA Annex 24: Heat, Air and Moisture Transfer in Insulated Envelope Parts. Final Report, Volume 3, Task 3: Material Properties. Kwiatkowski J., Woloszyn M. et al. 2011. Influence of sorption isotherm hysteresis effect on indoor climate and energy demand for heating. Applied Thermal Engineering 31(6-7): 1050-1057. Missirlis D., Donnerhack S. et al. 2010. Numerical development of a heat transfer and pressure drop porosity model for a heat exchanger for aero engine applications. Applied Thermal Engineering 30(11-12): 1341-1350. Mukhopadhyaya P., Kumaran M.K. et al. 2005. Use of the modified cup method to determine temperature dependency of water vapor transmission properties of building materials. Journal of Testing and Evaluation 33(5): 316322. Quenard D.A., Xu K., et al. 1998. Microstructure and transport properties of porous building materials. Materials and structures 31(5): 317-324. Roels S., Carmeliet J. et al. 2003. HAMSTAD Work Package 1: Final Report - Moisture Transfer Properties and Materials Characterisation. Stoltz G., Gourc J.-P. et al. 2010. Liquid and gas permeabilities of unsaturated municipal solid waste under compression. Journal of Contaminant Hydrology 118(1-2): 27-42. 76 Simulation of hygric performance of hysteretic building material exposed to cyclic changes of relative humidity O. Koronthalyova & P. Mihalka Institute of Construction and Architecture, Slovak Academy of Science, Bratislava, Slovakia ABSTRACT: The hygric performance of a hysteretic building material – burnt clay brick - exposed to cyclic changes of relative humidity (RH) was experimentally determined. The measurements consisted in monitoring mass changes of the brick specimens exposed to changing RH in climatic chamber. The test started with 24/24 hours cyclic changes of RH between 29.5% and 87%. After reaching the quasi steady state, the tests proceeded with the changes of surrounding air RH between 29.5% and 74%, in order to involve a transition from higher to lower moisture level. Simulation of the hygric performance of the brick was done using a simplified approach, based on a concept of ‘mean scanning curve’. The advantage of this approach is that it doesn’t need the experimental data about scanning adsorption and desorption curves. A comparison of the results obtained using the suggested simplified approach with the experimental results confirmed the usability of the simplified simulation. simulated. In the simulation, the hysteric behaviour is involved by a simplified way using the Mualem algorithm. A capability of this simplified approach is evaluated by a comparison with the experimental results. 1 INTRODUCTION In simulations of hygro-thermal performance of building structure the actual moisture content of building materials is most frequently calculated using the main adsorption curve or the mean between the adsorption and desorption curve. However, in real circumstances, the materials in building structures are exposed to changing boundary conditions. Therefore – due to hysteretic effects - their actual moisture content corresponds to the scanning curves between the adsorption and desorption curve. The reason why the hysteretic effects are not usually involved in simulations is that the existing complex models of hysteretic behaviour (e. g. Pedersen 1998) are relative time consuming and need data about scanning curves that are not commonly available. In the previous work (Koronthalyova 2011), a possibility of a prediction of the scanning isotherms in the hygroscopic range by two algorithms – Slope (Jaynes 1984) and Mualem (Mualem 1974) was analysed for two building materials: ceramic brick and autoclaved aerated concrete. It was concluded that both algorithms gave acceptable coincidence between the predicted and experimentally determined scanning curves. The idea of using Mualem algorithm for the prediction of scanning curves has been applied previously for other types of materials: wood (Carmeliet et al. 2005) and paper (Derluyn et al. 2007). In both cases, the obtained coincidence between the predicted and measured scanning curves was not ideal but was considered acceptable. In this work the hygric performance of a hysteretic building material – ceramic brick - exposed to cyclic changes of RH is experimentally determined and 2 EXPERIMENTAL PART The measurement was done for a ceramic brick. The tested brick is commonly used burnt clay brick produced by a Slovak manufacturer. Its basic material parameters - bulk density, total open porosity and capillary moisture content as well as water vapour sorption isotherms were determined in previous work (Koronthalyova 2011). Water vapour main adsorption, desorption isotherm from 98% RH and scanning curves were determined by standard gravimetric desiccator method, which consists in conditioning the samples in desiccators under constant RH and temperature (23°C) until the static equilibrium is achieved (EN ISO 12571). Water vapour permeability of the brick was measured by the standard cup method (EN ISO 12572) using the RH differences: 0 – 53%, 53 – 85% and 100 – 53%. At the end of the cup measurements the tested samples were oven dried at temperature of 105°C in order to determine their actual moisture content. The hygric behaviour of the brick under changing RH was determined in a climatic chamber with precise control of temperature and RH. The experiment consisted in monitoring mass changes of the samples. Three brick specimens with the dimensions of 100.8 x 101.9 x 26.38 mm, 101.8 x 101.6 x 24.23 mm and 101.2 x 101.6 x 26.04 mm were tested. The specimens were sealed on all but two surfaces by the 77 whole porous body but only to the part of the body where the assumption of the domains independency can be better fulfilled. The moisture storage of a hygroscopic material is described by a ‘mean scanning curve’ for the considered RH interval. The ‘mean scanning curve’ is determined as a mean of the 1st scanning desorption and the 1st scanning adsorption curve, predicted by the Mualem algorithm. According to this algorithm the moisture content during desorption, after the RH path φmin → φ1 (i. e. after one processes of adsorption) is expressed by the relation (1): epoxy resin in order to guarantee 1D water vapour flow. At the beginning of the measurement, the samples were oven dried at temperature of 105°C and then conditioned in climatic chamber at constant temperature and RH of 23°C and 27% respectively. The unsteady test started with 24/24 hours cyclic changes of surrounding air RH between 29.5% and 87%. After reaching the quasi steady state, the tests proceeded with the changes of surrounding air RH between 29.5% and 74% (Fig. 1). Starting from a dry state was chosen in order to allow evaluation of accuracy of the used adsorption isotherm fit. The applied schedule of RH changes enabled also monitoring a process of transition from higher to lower moisture contents. u (ϕ ) = u ads (ϕ ) + (u ads (ϕ1 ) − u ads (ϕ )) ⋅ u des (ϕ ) − u ads (ϕ ) (1) u max − u ads (ϕ ) During adsorption, after the RH path φmin → φ1 → φ2 (i. e. after one processes of adsorption and one processes of desorption) the moisture content is calculated from the relation (2): u (ϕ ) = u ads (ϕ ) + (u ads (ϕ1 ) − u ads (ϕ )) ⋅ u des (ϕ 2 ) − u ads (ϕ 2 ) (2) u max − u ads (ϕ 2 ) where u = moisture content (m3/m3), φ = relative humidity (-), φmin is starting RH (-); (φmin ≈ 0), uads = the moisture content corresponding to main adsorption curve, udes = the moisture content corresponding to main desorption curve, umax = maximum moisture content; (in the used procedure with partial hysteretic loop umax corresponds to moisture content at RH = 0.98), φ1 and φ2 correspond to the maximum and minimum of RH during the cyclic changes of RH. The main adsorption and the desorption isotherm from 98% RH as well as the ‘mean scanning curve’ are approximated by the relation of van Genuchten type (3) or (4): Figure 1. RH and temperature courses during unsteady test. The temperature was kept at constant value of 23°C. The relative humidity and temperature in the climatic chamber were registered in one minute intervals. The mass of the samples was weighed in the chosen time intervals by the electronic balance with the accuracy of 0.01g. The air flow velocity near the samples varied between 0.20 and 0.30 m/s. Based on similarity relations and the Lewis relation the value of the surface film coefficient for diffusion β = 5.6.10-8 s/m was determined, which was used in all simulations.   ln ϕ  n1   u (ϕ ) = A ⋅ 1 +  −   B     −n2   ln(ϕ + 0.02)  n1  u (ϕ ) = A ⋅ 1 +  −     B    (3) −n2 (4) where A, B, n1, n2 are parameters 3 SIMULATION OF HYSTERETIC BEHAVIOUR The process of the ‘mean scanning curve’ estimation for the tested ceramic brick and RH interval between 30 and 85% is illustrated in Fig. 2. The obtained scanning curves predicted by the Mualem algorithm are compared with the experimental ones, measured by standard desiccator method (Fig. 2). The numerical simulation of the hygric performance of the tested bricks was done by onedimensional simulation tool NEV3M. It is a modification of the former simulation tool NEV3 (Koronthalyova & Matiasovsky 1998), based on solution of the coupled partial differential equations for heat and moisture transfer. In the simplified evaluation of moisture storage, a conceptual model of hysteresis, suggested by Mualem (Mualem, 1974) was applied. It is based on Neel’s similarity hypothesis and Everett’s independent domain model. Based on the previous work (Koronthalyova 2011), the partial hysteretic loop formed by the main adsorption curve and the desorption isotherm from 98% RH is used for a prediction of the scanning curves inside this loop. From the point of view of the Mualem model it means that the independent domain theory is not applied to the 78 µ (u ) = µ 0 ⋅ (a + b ⋅ (1 − u / ucr ) 3 + c ⋅ (1 − u / ucr ) 4 ) Table 1. Parameters of van Genuchten equations for tested brick. Curve A B n1 n2 Eq. [m3/m3] Main 0.07056 0.037 2.0 0.35088 (4) adsorption Desorption 0.07056 0.06 1.22 0.33898 (4) from 98% Mean scan0.37 0.00088 1.0 0.513 (3) ning curve RH 29-82% Mean scan0.37 0.0013 1.0 0.56 (3) ning curve RH 29-71% Mean be0.37 0.0015 1.0 0.5 (3) tween main adsorption and desorption from 98% (5) where μ0, a, b, c, ucr are parameters: μ0 = 17, a = 0.25, b = 0. 31, c = 0.44 and ucr = 0.055; ucr corresponds to critical moisture content. A comparison of the measured μ values and their approximation by the relation (5) is shown in Figure 3. Table 2. Basic material properties of tested ceramic brick. Bulk density Open Capillary moisture content [kg/m3] porosity [-] [m3/m3] 1370 0.42 0.37 Table 3. Measured water vapour resistance factor values of tested ceramic brick. RH applied Moisture content [m3/m3] μ [-] 0 / 53 % 0.0096 12 53 / 85 % 0.0186 7.2 53 / 100 % 0.0274 5.9 Figure 2 shows comparison between the measured and approximated curves of the partial hysteretic loop. The used parameters of the van Genuchten equation (4) are presented in Tab. 1. Note that all presented measured values correspond to mean values obtained from three specimens. Figure 3. Comparison between measured and approximated vapour resistance factor moisture dependence. The simulations of hygric performance of the brick samples started with using the main adsorption isotherm as the moisture storage function. Then calculations were repeated with moisture storage function approximated by the “mean scanning curve” and the mean between the adsorption and desorption from 98%. These calculations started at the end of the first adsorption (at the 48th hour; cf. Fig. 1) where the moisture profiles, calculated with the main adsorption curve were used as initial values. At the end of the first adsorption, the calculated RH values varied within the sample thickness from 79.5 to 85 %, with the average of 82%. This relatively small dispersion of RH enabled using single “mean scanning curve” for the whole specimen. It was determined by the procedure explained in part 3 Figure 2. Illustration of generation of ‘mean scanning curve’ (ceramic brick, RH interval 30-85%). 4 RESULTS AND DISCUSSION The basic material properties of the ceramic brick are in Table 2. The measured values of water vapour resistance factor values (μ) and corresponding moisture contents at the end of the cup measurements are presented in Table 3. In all numerical simulations the water vapour resistance factor moisture dependence was approximated by the relation (5): 79 Figure 4. Comparison between measured and simulated specimen mass changes (specimen 1). Figure 5. Comparison between measured and simulated specimen mass changes (specimen 3). Figure 6. Comparison between measured and simulated specimen mass changes (specimen 2). 80 Table 4. Comparison between the measured and calculated average moisture content of the samples (m3/m3). Specimen 1 2 3 At end of the 1st adsorption (after 48 hours) Measured 0.022 0.018 0.022 Main adsorption 0.020 0.020 0.021 Maximum at end of 29.5 /87% cycle (after 288 hours) Measured 0.023 0.020 0.023 Main adsorption 0.020 0.020 0.021 Mean scanning curve (Mualem) 0.022 0.022 0.023 Mean adsorption / desorption 0.030 0.031 Minimum at end of 29.5 /87% cycle (after 312 hours) Measured 0.010 0.009 0.009 Main adsorption 0.008 0.008 0.008 Mean scanning curve (Mualem) 0.010 0.010 0.010 Mean adsorption / desorption 0.015 0.014 Maximum at end of 29.5 /72% cycle (after 432 hours) Measured 0.016 0.014 0.015 Main adsorption 0.015 0.015 0.015 Mean scanning curve (Mualem) 0.016 0.016 0.016 Mean adsorption / desorption 0.023 0.023 Minimum at end of 29.5 /72% cycle (after 453 hours) Measured 0.009 0.008 0.009 Main adsorption 0.007 0.007 0.007 Mean scanning curve (Mualem) 0.008 0.008 0.008 Mean adsorption / desorption 0.014 0.014 The agreement between the values calculated using “mean scanning curve” and measurement is very good in case of specimen 3, in case of specimen 1 is quite acceptable. In case of specimen 2, there is noticeable shift between the measured and calculated mass values. Mass values calculated with “mean scanning curve” are for the most part higher than the measured ones (Fig. 6). This difference is probably caused by the above-mentioned effect of material non-homogeneity. Using the mean between the adsorption and desorption from 98% as the moisture storage function results in significant overestimation of calculated moisture content/mass values for all specimens (Figs. 4-6, Tab. 4). Taking into account the sensitivity of the calculated results to the applied moisture storage function and the present effect of material non-homogeneity it can be concluded that the used simplified approach evaluates the hysteretic behavior of the tested brick with acceptable accuracy. for RH interval from 29 to 82%. Similarly in the simulation of the next period of the test with the RH changes between 27 and 72% the “mean scanning curve” determined for RH interval from 29 to 71% was applied. Parameters of the used curves are given in Table 1. Generally, in case of larger spread of RH different “mean scanning curves”, corresponding to the actual RH values should be used. A comparison between the measured and calculated values is shown in Figures 4-6 and Table 4. In evaluating the simulation results it is important to take into account that calculated values of specimen moisture content/mass are very sensitive to the applied moisture storage function. In case of specimen 3, the used approximation of the main adsorption curve gives quite close coincidence with the measurement. In case of specimen 1 the moisture content calculated with the main adsorption is slightly underestimated while for specimen 2 it is overestimated. The obtained differences are not very high (see Tab. 4) and can be explained by the fact that the main adsorption curve as well as water vapour resistance factor moisture dependence were approximated using the mean of the measured values whereas the tested brick is not ideally homogenous material. The results obtained for the period of cyclic changes of RH confirm a presence of hysteretic behavior that is more significant after the transition from 29.5/87% cycle to 29.5/72% cycle (Fig. 4-6). However, more significant hysteretic effect could be expected if higher RH values are applied during the first cyclic phase. Hygric performance of a hysteretic building material – ceramic brick – exposed to cyclic changes of RH was experimentally determined and simulated. Simulation of the hygric performance of the brick was done using a simplified approach, based on a concept of ‘mean scanning curve’, that was verified by comparison with the experimental results. Taking into account the sensitivity of the calculated moisture content/mass to the used moisture storage function, the agreement between the results of simplified model and experiment is satisfactory. The advantage of the suggested approach is that it doesn’t need the experimental data about scanning adsorption and desorption curves. 5 CONCLUSIONS 6 ACKNOWLEDGMENTS This research was supported by the Scientific Grant Agency VEGA (Grant No. 2/0145/13) and the Slovak Research and Development Agency (Grant No. APVV-0031-10). 7 REFERENCES Carmeliet J., de Wit M., and Janssen H. 2005. Hysteresis and Moisture Buffering of Wood. In Proceedings of the 7th Symposium of Building Physics in the Nordic Countries, 13–15 June, 55–62, Reykjavik, Iceland. Derluyn H. Janssen H. Diepens J., Derome D., and Carmeliet J. 2007. Hygroscopic behavior of paper and books. Journal of Building Physics 31 (9): 9-34. 81 EN ISO 12571:2000 Hygrothermal performance of building materials and products - Determination of hygroscopic sorption properties. EN ISO 12572:2001 Hygrothermal performance of building materials and products - Determination of water vapour transmission properties. Jaynes D. B. 1984/1985. Comparison of soil-water hysteresis models, In Journal of Hydrology, Vol. 75: 287-299. Koronthalyova O. 2011. Water vapour sorption of building materials – modelling of scanning curves. In Proceedings of the 9th Nordic Symposium on Building Physics, NSB 2011, Editors: J. Vinha, J. Piironen, K. Salminen, Department of Civil Engineering, Tampere University of Technology, Tampere, Finland, p. 655-662. Koronthalyova O., and Matiasovsky P. 1998. Factors Influencing Correctness of the Simulation Model of the Building Structures Hygrothermal Behaviour. In New Requirements for Materials and Structures, Editors: V. Broza, R. Cerny, B. Kubat and J. Witzany, Czech Technical University, Prague, p.160-165. Mualem Y. 1974. A Conceptual Model of Hysteresis. Water Resources Research. Vol. 10 ( No 3): 514-520. Pedersen C. R. 1990. Combined Heat and Moisture Transfer in Building Constructions. PhD Thesis, Thermal Insulation Laboratory, Technical University of Denmark. 82 Assessment of scale of the microstructure impact on capillary transport in cement-based composites with polypropylene fibers A. Wygocka & H. Garbalińska West Pomeranian University of Technology, Faculty of Civil Engineering and Architecture, Szczecin, Poland ABSTRACT: The paper presents results of the microstructure and capillarity tests of four cement mortars reinforced with polypropylene fibers with length of 19 mm. The recipes of the particular composites were varied by applying superplasticizer, air-entraining agent and silica fume. In order to identify differences in the internal structures of the mortars as a result of application of selected modifiers, microstructure tests using mercury intrusion porosimetry (MIP) were performed. The capillarity experiments were conducted for two months with reference to each composite. Obtained data of samples mass changes were used to determine values of water sorption coefficient that characterizes the rate of water absorption. The influence of structure modification on the course of capillary flow was assessed. The conducted studies provide some practical guidelines for a controlled regulation of the capillary water transport processes in porous materials with a microstructure modified through appropriate use of selected admixtures and additives. continuity of the capillaries in the cement paste. It is to be expected that in microstructures formed so differently, the process of capillary transport of water will represent a distinctly different course. This paper was devoted to laboratory studies, testing the scale of impact of all aforementioned structure-creating factors on the process of capillary water transport in differently modified cement composites. 1 INTRODUCTION Capillary transport of water in porous building materials is of great practical importance. Due to the danger of introducing aggressive substances into the material by water, as well as the emergence of tensile stress during formation of ice crystals, the processes of capillarity largely determine durability of the material (Basheer et al. 2001, Garbalińska 2002, Panesar & Chidiac 2010). Since the water absorption rate and the scale of the phenomenon are strongly dependent on the internal structure of the material, it was assumed that by a proper control of its microstructural parameters the course of these processes can be regulated on purpose. One of the main factors affecting the structure of porosity in the cement matrix is the value of the water-cement ratio, a significant reduction of which, and thereby some improvement of the microstructural properties of the composite, can be achieved by application of certain plasticizing and liquefying admixtures. The literature also indicates that application of silica fume into the mix results in limiting the amount of capillary pores to the benefit of the gel ones in thickening the transition layer at the aggregate-paste interface (Alexander & Magee 1999, De Gutiérrez et al. 2005). On the other hand, application of polypropylene fibers leads to formation of a more homogenous structure of cement composites by reducing the cracks associated with the plastic shrinkage of the cement paste (Banthia & Gupta 2006, Sanjuán & Moragues 1997). Another method of modification, aimed at durability improvement of such composites, is application of an air-entraining agent (Chatterji 2003, Hale et al. 2009). Formed as a result of its performance, spherical air pores disrupt 2 MATERIALS The study of the structure and capillarity was performed on four mortars based on cement CEM I 42.5 R and natural quartz sand with grain 0÷2 mm. In all of the composites micro-reinforcement was used in the form of PP fibers, with a length of 19 mm and a diameter of 100 μm. Fibers were added directly to the mix in a standard amount of 0.9 g/dm3. In formula I, the w/c ratio was adopted as 0.55. Formulas II and III were enriched with a superplasticizer, permitting reduction of the water admix by 20%. Additionally, in composite III 5% of the cement mass was replaced by silica fume. Comparing to I, formula IV was modified by application of some air-entraining agent in amount of 0.3% of the cement mass. The mix proportions of the four tested composites are given in Table 1. From each mix, 16 cylindrical samples were produced, and were kept for 1 year in a water bath right after demoulding. Such a long water conditioning period was supposed to ensure hydration processes be so advanced that they do not bring about any measurable changes in the microstructure during the main experiments. 83 Table 1. Mixture proportions of the tested mortars. Mortar I II III IV Ratio w/c or w/b [-] 0.55 0.41 0.41 0.55 CEM I 42.5 R [g/dm ] Water + admixture [g/dm3] Water [g/dm3] 490 525 499 490 270 216 216 270 270 210.2 210.2 268.5 Sand [g/dm ] 1519 1629 1629 1519 Superplasticizer [g/dm3] - 5.78 5.78 - Silica fume [g/dm3] Air-entraining agent [g/dm3] PP fibers [g/dm3] - - 26.25 - - - - 1.47 0.9 0.9 0.9 0.9 3 log differential intrusion [mL/g] 3 Mortar I 0,20 0,18 0,16 0,14 0,12 0,10 0,08 0,06 0,04 0,02 0,00 1000,000 10,000 0,100 diameter [micrometers] 0,001 Mortar II log differential intrusion [mL/g] 0,14 3 EXPERIMENTS 3.1 Capillarity test In order to ensure uniform initial conditions all the tested cylinders were dried up. The drying process was performed cyclically until a constant mass of the samples was reached. To dry cylinder cores up and to prevent cracking due to thermal stress, the temperature in a single cycle was raised gradually up to the maximum of 105oC, and then the samples were gradually cooled down to the laboratory temperature. Before the capillarity tests all the cylinders were insulated on the sides with polythene foil to ensure one-dimensional water flow during the experiment (Garbalińska & Wygocka 2007). The experiment on capillarity began upon placement of the dried cylinders in dishes filled with distilled water, on plastic grills providing them some points of support. Measurements of the changing sample masses were conducted at pre-defined intervals, adapted to process dynamics, throughout 2 months. In the initial phase of experiment (intense water absorption) weighing was performed relatively frequently – every 1, 2, 4 hours. With the capillary suction slowdown measurements were carried out at longer intervals (2÷3 days). 0,12 0,10 0,08 0,06 0,04 0,02 0,00 1000,000 10,000 0,100 diameter [micrometers] 0,001 Mortar III log differential intrusion [mL/g] 0,14 0,12 0,10 0,08 0,06 0,04 0,02 0,00 1000,000 10,000 0,100 diameter [micrometers] 0,001 Mortar IV log differential intrusion [mL/g] 0,25 3.2 Microstructural test After capillarity experiment all composites in question underwent additional structure tests. Tests were conducted using Mercury Intrusion Porosimetry by means of porosimeter AutoPore II 9220. Measuring range of apparatus enabled to measure the pores within the range from 3 nm to 360 μm. From each mortar one cylinder was selected for tests. In order to measure the most representative and homogenous structure testing samples of 3÷4 g were obtained from the cores of the cylinders. 0,20 0,15 0,10 0,05 0,00 1000,000 10,000 0,100 diameter [micrometers] 0,001 Figure 1. Pore size distribution of tested mortars I, II, III and IV. 84 4 TESTS RESULTS Mortar I 35 4.1 Microstructural test results 30 Δmt/F [kg/m2] The results of the structural tests allowed to prepare the curves of pore size distribution, showed in Figure 1. On the basis of the results of the conducted measurements it was stated that application of tested modifiers (admixtures and silica fume) considerably influences the porosity structure of cement composites. Pore distribution in the mortar I (prepared on the basic components and reinforced with PP fibers) was unimodal, with predominant pore diameter of 0.075 μm. Application of superplasticizer in mortar II and additionally silica fume in mortar III yield more diverse porosity structures. In the both composites increase of predominant pore size was found – 0.186 μm and 0.336 μm respectively. The use of an air-entraining agent in mortar IV resulted in a bimodal pore size distribution, with predominant pore diameter similar to mortar I. Another effect of application of air-entraining agent was 50% increase in effective porosity of composite IV in relation to reference mortar I. 25 20 15 I-1 I-4 I-7 I-10 10 5 I-2 I-5 I-8 I-11 I-3 I-6 I-9 I-12 0 0 10 20 30 t 1/2 [h1/2] 40 Mortar II 35 Δmt/F [kg/m2] 30 25 20 15 II-1 II-4 II-7 II-10 10 5 II-2 II-5 II-8 II-11 II-3 II-6 II-9 II-12 0 4.2 Capillarity test results 0 ∆mt F ⋅∆ t 20 30 t 1/2 [h1/2] 40 Mortar III Data obtained at capillarity tests were used for preparation of graphs of mass changes referred to water contact area Δmt/F in relation to square root of time √t. Graphs for each of 12 samples produced from given mortar are presented in Figure 2. In Figure 3 resultant graphs for the four tested composites are shown. From the range in which linear dependence of mass gain on the square root of time occurred, value of water sorption coefficient A for each sample was determined, using formula: A= 10 35 III-1 III-4 III-7 III-10 Δmt/F [kg/m2] 30 25 III-2 III-5 III-8 III-11 III-3 III-6 III-9 III-12 20 15 10 5 0 0 10 20 30 (1) t 1/2 [h1/2] 40 Mortar IV where A = water sorption coefficient; Δmt = change in the mass of the material sample during the time t [kg]; F = sample surface area exposed to water [m2]; t = time [h]. Values of water sorption coefficient for particular samples and mean values for mortars I, II, III and IV are presented in Table 2. Microstructure modifications of tested mortars resulted in significant changes in the values of the parameters describing capillary water flow. In the case of all the composites modified with admixtures and silica fume, slowdown of the rate of water capillary suction with reference to mortar I was observed. 35 Δmt/F [kg/m2] 30 25 20 15 10 5 IV-1 IV-4 IV-7 IV-10 IV-2 IV-5 IV-8 IV-11 20 30 IV-3 IV-6 IV-9 IV-12 0 0 10 t 1/2 [h1/2] 40 Figure 2. Changes of samples mass with respect to the square root of time of the four mortars under examination. 85 tained for samples of mortar produced with airentraining agent (IV) diverge from the model description of the phenomenon, with two clearly separated phases of different rate of water uptake. After intense water absorption phase, significant mass gain of samples was observed. As a result, considering lower bulk density of composite IV, the final moisture content increased about 11% in comparison to mortar I. 30 Δmt/F [kg/m2] 25 20 15 10 5 I II III IV 0 0 10 20 5 SUMMARY 30 t 1/2 [h1/2] 40 The analysis of the obtained results of porosimetry tests indicated plainly that thanks to the application of chemical admixtures and silica fume it is possible to produce materials of strongly diversified internal structures. Microstructure modifications of tested composites resulted in changes in their capillary properties. The influence of the microstructure manifested itself not only in the different values of the sorption coefficients describing the water absorption rate in the first hours of contact of the material with water. In the subsequent part of the experiment, a different course of the capillarity process was observed, which resulted in a different degree of the final water content in the mortars in question. Figure 3. The resultants of the mass change with respect to the square root of time of the four mortars under examination. Table 2. Water sorption coefficients A [kg/(m2h1/2]. Mortar Sample No I II III IV 1 3.204 2.643 1.445 1.953 2 2.419 2.597 2.200 2.239 3 3.747 2.547 1.590 1.726 4 2.467 2.381 1.787 2.656 5 2.598 2.796 1.493 2.278 6 3.120 2.811 2.155 2.506 7 3.372 2.265 2.243 1.984 8 2.339 2.655 1.966 2.176 9 3.208 2.541 1.897 2.243 10 3.763 2.912 2.147 2.214 11 2.832 2.716 1.531 2.126 12 2.584 2.385 1.639 2.106 Mean 2.971 2.604 1.841 2.184 Standard 0.504 0.193 0.299 0.244 deviation 6 REFERENCES Alexander M.G., and Magee B.J. 1999. Durability performance of concrete containing condensed silica fume. Cement and Concrete Research 29: 917-922. Banthia N., and Gupta R. 2006. Influence of polypropylene fiber geometry on plastic shrinkage cracking in concrete. Cement and Concrete Research 36: 1263-1267. Basheer L., Kropp J., and Cleland D.J. 2001. Assessment of the durability of concrete from its permeation properties: a review. Construction and Building Materials 15: 93-103. Chatterji S. 2003. Freezing of air-entrained cement-based materials and specific actions of air-entraining agents. Cement and Concrete Composites 25: 759-765. De Gutiérrez R.M., Díaz L.N., and Delvasto S. 2005. Effect of pozzolans on the performance of fiber-reinforced mortars. Cement and Concrete Composites 27: 593-598. Garbalińska H. 2002. Kapillarer Wassertransport in Zementmörtel. Experimentelle Bestimmung der Koeffizienten des kapillaren Saugens. Bauphysik 24: 87-92. Garbalińska H., and Wygocka A. 2007 Prüfkörperabdichtung und der Wasserabsorptionskoeffizient von mit Polypropylenfasern modifizierten Zementmörteln. Bauphysik 29: 436-441. Hale M.W., Freyne S.F., and Russell B.W. 2009. Examining the frost resistance of high performance concrete. Construction and Building Materials 23: 878-888. Panesar D.K., and Chidiac S.E. 2010. Capillary suction model for characterizing salt scaling resistance of concrete containing GGFS. Cement and Concrete Composites 31: 570-576. Sanjuán M.A., and Moragues A. 1997. Polypropylene-fibrereinforced mortar mixes: optimization to control plastic shrinkage. Composites Science and Technology 57: 655660. Reduction of w/c ratio and application of superplasticizer (composite II) resulted in decrease of water sorption coefficient by 12%. The most significant decrease of this parameter – up to 38% with reference to mortar I, was noted in the case of composite III, where 5% of cement was replaced by silica fume. Effect of using superplasticizer and silica fume also revealed in lower amount of water absorbed by the samples of mortars II and III after two moths of experiment duration. The final moisture content decreased 18% and 31% respectively. Favourable influence of application of airentraining agent in composite IV was revealed in the first phase of process, resulting in slowdown of capillary water absorption, which led to reduction of water sorption coefficient by 26%. Having analysed graphs shown in Figure 3, distinct course of mass changes in relation to square root of time for mortar IV was stated. In mortars I, II and III the intense water absorption phase was followed by stabilizing of the samples mass. In some cases, at the final period of experiment, decrease of cylinders mass were recorded. This was probably related with water front reaching the un-insulated upper surface of the cylinder and evaporating to the environment. Graphs ob86 Concrete with pozzolanic admixtures: An environmental-friendly solution T. Kulovaná, E. Vejmelková, D. Koňáková, M. Keppert, & R. Černý Czech Technical University in Prague, Faculty of Civil Engineering, Department of Materials Engineering and Chemistry, Prague, Czech Republic M. Ondráček Brno University of Technology, Faculty of Civil Engineering, Institute of Technology of Building Materials and Components, Brno, Czech Republic ABSTRACT: High performance concrete (HPC) with natural zeolite and ceramic powder as Portland cement replacement in the amount of up to 40% is analyzed in the paper. Natural zeolite and ceramic powder belong to the pozzolanic materials. Utilization of pozzolanic materials is accompanied with lower demand on energy needed for Portland clinker calcination which usually results in lower production costs and lower CO2 emissions. Besides, the application of pozzolanic materials can improve properties of hardened concrete, namely strength and durability-related parameters. Basic physical characteristics, hygric properties and thermal properties are studied and compared with reference concrete. The effective amount of pozzolanic materials used as a partial replacement of cement for concrete production is found. Measurement results show that natural zeolite and ceramic powder have a potential to replace a part of Portland cement in concrete in building industry. It is shown that although from both environmental and economical points of view it would be desirable to use its highest possible amounts in concrete production, the extent of Portland cement replacement which could be chosen in preparation of concrete mixes has relatively strict limitations. 1 INTRODUCTION in high performance concrete (HPC), replacing Portland cement in the amount of up to 40 % of mass. Properties of several concrete mixes are studied and compared with reference concrete. In recent years, there has been a renewed interest in the use of these materials such as pozzolana in concrete production. Natural zeolite is a type of aluminosilicate mineral containing large quantities of reactive SiO2 and Al2O3 (Naiqian et al. 1992). Zeolites are taking part in hydration reactions and contribute to strength of cementitious materials (Poon et al. 1999). The use of ceramic powder as a partial replacement of cement replacement in concrete was reported only lately (Pacheto-Torgal et al. 2010). Most solutions, which were developed until now, rather deal with the usage of crushed bricks, which replace parts of the aggregate in concrete (Cabral et al. 2010). Cracked or broken ceramic fragments, roof tiles ground, damaged bricks are not accepted as commercial products and, therefore, the unsold waste of the ceramic industry becomes an environmental problem (Layat et al. 2009). Natural zeolite and ceramic powder are used as alternative binders Table 1. Composition of studied concretes [kg m-3]. Component CR CEM I 42.5 R 484 Natural zeolite Ceramic powder Aggregates 0-4 mm 812 Aggregates 8-16 mm 910 Plasticizer Mapei Dynamon SX 5.3 Water 172 CZ 10 436 48.4 812 910 5.3 194 2 MATERIALS The composition of the studied concrete mixes (denoted CZ 10, CC 10, CZ 20, CC 20, CZ 40, CC 40) used in experimental work in this paper is shown in Table 1. A part of cement (10-40% by mass) was replaced by natural zeolite and ceramic powder. For the sake of comparison, also a reference mix CR with only Portland cement CEM I 42.5 R as the binder was studied. The total mass of binder in the reference mix was the same in all mixtures. The specific surface area of Portland cement was 341 m2kg-1 of ceramic powder 582 m2kg-1, and for natural zeolite it was 589 m2kg-1. The chemical composition is shown in Table 2. CC 10 436 48.4 812 910 5.3 160 87 CZ 20 387 96.8 812 910 5.3 221 CC 20 387 96.8 812 910 5.3 160 CZ 40 305 179.2 812 910 5.3 244 CC 40 305 179.2 812 910 5.3 160 Amount of used water was determined based on the same consistence of all studied mixtures. The measurement of material parameters of hardened concrete mixes was done after 28 days of standard curing. It took place in a conditioned laboratory at the temperature of 22±1°C and 25-30% relative humidity. 3.2 Pore structure Characterization of pore system was performed by mercury intrusion porosimetry. The experiments were carried out using the instruments PASCAL 140 and 440 (Thermo Scientific). Since the size of the specimens is restricted to the volume of approximately 1 cm3 and the studied materials contained also aggregates about the same size, the porosimetry measurements were performed on samples without coarse aggregates. Table 2. Chemical composition of applied materials [%]. Cement Natural zeolite Ceramic powder SiO2 21.89 66.84 63.45 Al2O3 5.60 12.62 13.98 Fe2O3 3.75 3.33 5.39 TiO2 0.30 0.37 0.77 CaO 62.33 6.71 8.18 MgO 1.04 0.59 4.50 K2O 0.92 8.13 2.43 Na2O 0.11 0.85 0.9 SO3 2.88 0.1 3.3 Mechanical properties The measurement of compressive strength was done by the hydraulic testing device VEB WPM Leipzig having a stiff loading frame with the capacity of 3000 kN. The tests were performed according to ČSN EN 12390-3 (ČSN 12390 2002) after 28 and 360 days of standard curing. The measurement was done on the samples with the dimensions of 150 x 150 x 150 mm. 3 EXPERIMENAL METHODS 3.1 Basic physical properties 3.4 Water vapor transport properties Among basic properties, the bulk density, the matrix density and the open porosity were measured using the water vacuum saturation method (Roels et al. 2004). The measurement was done on the samples with the dimensions of 50 x 50 x 50 mm. Each sample was dried in a drier to remove majority of the physically bound water. After that the samples were placed into the desiccator with deaired water. The specimen was then kept under water not less than 24 hours. From the mass of the dry sample md, the mass of water saturated sample mw, and the mass of the immersed water saturated sample ma, the volume V of the sample was determined from the equation The dry-cup and wet-cup methods were employed in the measurements of water vapor transport parameters (Roels et al. 2004). The water vapor diffusion coefficient D [m2 s-1], water vapor diffusion permeability δ [s] and water vapor diffusion resistance factor μ [-] were determined. The measurement was done on the samples with the dimensions of 50 x 50 x 20 mm. V= mw − ma ρl 3.5 Water transport properties The water absorption coefficient A [kg m-2 s-1/2] and apparent moisture diffusivity κ [m2 s-1] were measured using a water sorptivity experiment (Kumaran 1999), (Vejmelková et.al 2009). The measurement was done on the samples with the dimensions of 50 x 50 x 20 mm. (1) where ρl = density of water. The open porosity ψ0, the bulk density ρ and the matrix density ρmat were calculated according to the equations ψ0 = mw − md Vρ l ρ= ρ mat = 3.6 Thermal properties Thermal conductivity λ [W m-1 K-1] and specific heat capacity c [J kg-1 K-1] were measured using the commercial device ISOMET 2104 (Applied Precision, Ltd., Figure 1). The measurement is based on analysis of the temperature response of the analyzed material to heat flow impulses. The heat flow is induced by electrical heating using a resistor heater having a direct thermal contact with the surface of the sample. The measurement was done on the samples with the dimensions of 70 x 70 x 70 mm. (2) md V (3) md (1 − ψ 0 )V (4) 88 other concretes with supplementary cementing materials). 4.2 Pore structure The pore size distribution curves (Figure 2) revealed diverse influence of two studied admixtures to material’s pore system. The fine ground ceramics did not change nature of the pore system compared to reference mixture CR but caused moderate increase of the total pore volume. That was due to more pores of diameter between 0.1 and 1 μm. The admixing of zeolite increased total pore volume in higher extent than ceramics; it also changed the nature of pore system. One observed (similarly to ceramics) higher volume of pores of 0.1 and 1 μm in diameter but presence of very small pores (of diameter in range of just a few nm) was indicated as well; these were a residue of primary zeolite microstructure which is extremely (micro)porous. Figure 1. ISOMET 2104. 4 EXPERIMENTAL RESULTS AND DISCUSSION 4.1 Basic physical properties Basic physical properties of studied concrete measured by the water vacuum saturation method are shown in Table 3. Table 3. Basic material properties. Material Bulk density Matrix density [kg m-3] [kg m-3] CR 2244 2590 CZ 10 2194 2601 CZ 20 2132 2601 CZ 40 2036 2623 CC 10 2420 2730 CC 20 2370 2720 CC 40 2330 2700 Open porosity [%] 13.4 15.7 18.0 22.4 11.4 12.8 13.9 In comparison of reference material CR with material containing pozzolanic admixture CC achieved higher value of the bulk density than materials with content of natural zeolite and lower value than material with ceramic powder. The bulk density decreased with increasing amount of pozzolanic admixture in cases of zeolite as well as ceramic powder. Values of the bulk density of materials with content of natural zeolite were lower than the bulk density of material with ceramic powder in all mixtures. Value of the matrix density of reference material CR was always lower in comparison with other studied material with pozzolanic admixture. Materials with content of ceramic powder achieved slightly higher values of the matrix density than materials with natural zeolite in all cases. The open porosity increased with increasing amount of pozzolanic admixtures in both cases. This agreed with values of the bulk density. The lowest porosity achieved CC 10 with 10% ceramic powder. The highest porosity had CZ 40 with natural zeolite (about 20-50% higher than the Figure 2. Pore size distribution. 4.3 Mechanical properties Table 4 shows the compressive strength of the studied concretes after 28 and 360 days. Table 4. Mechanical properties. Material Compressive strength [MPa] 28 days 360 days CR 71.9 77.7 CZ 10 63.7 64.1 CZ 20 54.2 58.4 CZ 40 36.4 42.8 CC 10 65.7 77.0 CC 20 60.2 68.9 CC 40 42.6 61.8 Reference material CR achieved the highest compressive strengths. Materials containing ceramic 89 powder had higher values of compressive strength than materials with natural zeolite in all cases. Values of compressive strength after 28 days showed that up to the CC 10, CC 20 and CZ 10 the concretes maintained their high performance character. These materials met the basic criterion of the compressive strength of 60 MPa to be considered as highperformance concrete. For higher amounts of natural zeolites and ceramic powder in mixtures (CZ 20, CZ 40 and CC 40) the compressive strength was significantly lower which was an expected outcome, taking into account the properties of pozzolanic materials. Differences in the compressive strength of individual mixtures after 360 days were in similar relationship as in 28 days but obviously somewhat higher. The mixtures with ceramic powder featured higher increase of strength in time which implies that ceramic powder was more effective as pozzolana than the used zeolite. The values of compressive strength of concrete with ceramic powder were after 360 days up to 45% higher at mixture CZ 40. In a combination with the open porosity data (Table 3), mechanical properties indicated that the materials CZ 20, CZ 40 and CC 40 might have certain durability problems due to their lower strength and higher porosity as compared with the other three mixtures. increasing amount of pozzolanic admixture in concretes in wet-cup and dry-cup arrangement as well which is consistent with open porosity (Table 3) and pore size distribution (Figure 2). One can see presence of big amount of little pores, especially at material with natural zeolite. Values of the water vapor diffusion coefficient of studied materials with ceramic powder in dry-cup were lower than for concrete with content of natural zeolite. In wet cup arrangement were values of materials with ceramic powder almost the same. The data that were measured show that, the values of water vapor diffusion coefficient corresponding to the lower values of relative humidity (5/50 %) were always lower than those for higher relative humidity values (97/50 %). This is related to the partial transport of capillary condensed water in the wet-cup arrangement. 4.5 Water transport properties The liquid water transport parameters (Table 7) systematically increased with the increasing amount of natural zeolite and ceramic powder in the mixture. Values of the water absorption coefficient of mixture with natural zeolite were higher at samples with ceramic powder in all mixtures. The highest ability of liquid water transport had the material CZ 40 with 40% of natural zeolite which is consistent with its higher open porosity (Table 3) and presence of higher amount of pores with diameter around 1 µm. Material CC 10 with 10% of ceramic powder had the water absorption coefficient five times lower than CZ 40. This agreed with the porosity data in Table 3. CZ 20 with 20% natural zeolite could be considered a reasonable limit, as for the liquid water transport parameters. 4.4 Water vapor transport properties Table 5 and Table 6 show water vapor transport properties of studied materials with natural zeolite and ceramic powder. Table 5. Water vapor transport properties in dry-cup arrangement. Material δ D µ [s] [m2s-1] [-] CR 1.58E-12 2.18E-07 106.69 CZ 10 2.05E-12 2.82E-07 81.87 CZ 20 2.85E-12 3.92E-07 58.78 CZ 40 4.72E-12 6.49E-07 35.42 CC 10 1.93E-12 2.65E-07 86.91 CC 20 2.09E-12 2.87E-07 80.02 CC 40 2.23E-12 3.06E-07 75.08 Table 7. Water transport properties. Material A [kg m-2 s-1/2] CR 0.0086 CZ 10 0.0096 CZ 20 0.0153 CZ 40 0.0317 CC 10 0.0067 CC 20 0.0077 CC 40 0.0101 Table 6. Water vapor transport properties in wet-cup arrangement. Material δ D µ [s] [m2s-1] [-] CR 1.92E-12 2.64E-07 89.75 CZ 10 2.43E-12 3.34E-07 68.87 CZ 20 3.57E-12 4.91E-07 49.45 CZ 40 6.51E-12 8.95E-07 29.80 CC 10 2.33E-12 3.20E-07 71.97 CC 20 3.16E-12 4.34E-07 53.03 CC 40 4.67E-12 6.41E-07 35.87 κ [m2 s-1] 4.28E-09 3.92E-09 7.42E-09 20.8E-09 3.79E-09 3.83E-09 6.23E-09 4.6 Thermal properties Thermal properties of studied concretes in dry state are shown in Table 8. Value of the water vapor diffusion coefficient of studied materials with natural zeolite increased with 90 Table 8. Thermal properties of studied concretes in dry state. Material λ c [W m-1 K-1] [J kg K-1] CR 1.623 738 CZ 10 1.513 732 CZ 20 1.397 729 CZ 40 1.167 706 CC 10 1.530 678 CC 20 1.397 729 CC 40 1.533 783 Figure 3 shows that the dependence of thermal conductivity of studied materials on moisture content was significant; up to 50% increase of thermal conductivity was observed for water saturated specimens as compared with the dry state. The specific heat capacity increased considerably with increasing moisture content (Figure 4) which was related to the high specific heat capacity of water. 5 CONCLUSIONS Results were in a basic qualitative agreement with the porosity results (Table 3). The value of the thermal conductivity decreased with increasing amount of natural zeolite in the mixture and with increasing amount of ceramic powder was relatively similar. The lowest value of thermal conductivity achieved CZ 40 with 40% of natural zeolite. Values of the specific heat capacity slightly decreased with the increasing amount of zeolites in mixture and were lower than CR. The values of specific heat capacity for mixture with ceramic powder have shown an opposite trend. However, the differences in specific heat capacity were within the error range of the measuring method. The experimental results presented in this paper showed that the replacement of Portland cement by supplementary cementing materials may be a viable solution for high performance concrete. Natural zeolite and ceramic powder can be considered an environmental friendly binder with a potential to replace a part of Portland cement in concrete in building industry, but their dosage is to be considered carefully due to the worsening of mechanical properties for higher zeolite and ceramics amounts. The mechanical and liquid water transport parameters were satisfactory up to the 20% replacement level for both pozzolanic materials. The results of measuring water vapor transport parameters were within an expected range. Thermal parameters were acceptable for all studied mixes. Summarizing the above results, it can be concluded that the replacement of Portland cement by natural zeolite and ceramic powder in the amount of 20% by mass can be considered the most suitable option among the studied mixes. Thermal conductivity [Wm-1 K-1 ] 2.5 2 1.5 CR CZ 10 1 CC 10 CZ 20 CC 20 0.5 6 ACKNOWLEDGEMENTS CZ 40 CC 40 0 0 5 10 15 20 Moisture [%m3 m-3 ] 25 This research has been supported by the Czech Science Foundation, under project No P104/12/0308 and under project SGS13/165/OHK1/3T/11. 30 Figure 3. Thermal conductivity of studied concretes. 1200 7 REFERENCES Specicic heat capacity [Jkg-1 K-1 ] 1100 Cabral A., Schalch V., Molin D., and Ribeiro J. 2010. Mechanical properties modeling of recycled aggregate concrete, Construction and Building Materials: 24:421– 430 ČSN EN 12390-3 2002. Testing of hardened concrete – Part 3: Compressive strength. Czech Standardization Institute, Prague. Kumaran M.K. 1999. Moisture Diffusivity of Building Materials from Water Absorption Measurements. Journal of Thermal Envelope and Building Science, 22:349-355. Layat A., Trezza M., and Poggi M. 2009. Characterization of ceramic roof tile wastes as pozzolanic admixture. Waste management 29(5):1666-74 1000 CR CZ 10 CC 10 CZ 20 CC 20 CZ 40 CC 40 900 800 700 600 500 0 5 10 15 20 Moisture [%m3 m-3 ] 25 30 Figure 4. Specific heat capacity of studied concretes. 91 Naiqian F., Changchen M., and Xihuang J. 1992. Natural zeolite for preventing expansion due to alkali–aggregate reaction, Cement, Concrete and Aggregates 14:93-96 Pacheco-Torgal F., and Jalali S. 2010. Reusing ceramic wastes in concrete. Construction and Building Materials;24:832– 838. Poon C.S., Lam L., Ko S.C., and Lin Z.S. 1999. A study on the hydration rate of natural zeolite blended cement pastes. Construction and Building Materials 8: 427–432 Roels S., Carmeliet J., Hens H., Adan O., Brocken H., Černý R., Pavlík Z., Hall C., Kumaran K., Pel L., and Plagge R. 2004. Interlaboratory Comparison of Hygric Properties of Porous Building Materials. Journal of Thermal Envelope and Building Science,27:307-325 Vejmelková E., Pavlíková M., Jerman M., and Černý R. 2009. Free Water Intake as Means of Material Characterization. Journal of Building Physics 33:29-44. 92 Metamodelling in robust low-energy dwelling design L. Van Gelder, H. Janssen & S. Roels KU Leuven, Department of Civil Engineering, Building Physics Section, Leuven, Belgium ABSTRACT: Deterministic simulations are commonly used in building design to calculate for instance the energy use. Many influential parameters are however inherently uncertain. As a result, deterministic optimisation unreliably predicts the impact of design measures. In order to improve the optimisation process, uncertainties need to be taken into account through a robust design method. Such a simulation based optimisation is often extremely time-consuming and hence unrealistic. To overcome this barrier, metamodelling is of high interest as a metamodel aims to imitate the original numerical model with a simplified fast model. In this paper, metamodels are constructed for the energy demand and indoor temperature of a semi-detached dwelling. Polynomial regression and multivariate adaptive regression splines (MARS) are compared as well as the number of training samples. For the current case, MARS models show to be more accurate. Furthermore, the models appeared more reliable for some outputs than for others. To ensure the reliability of the metamodels, a cross-validation strategy is proposed to construct metamodels with as less training sets as possible. rameters are involved. Because only 100 samples remain for each potential U-value, it is nonguaranteed that the other influential parameters are equally distributed, resulting in a potentially unreliable comparison. This example indicates the need for more samples in a robust design method. Coupling such a robust design method to dynamic energy simulation models is generally very timeconsuming, which makes metamodelling extremely interesting since a metamodel aims at mimicking the original model, but with a strongly reduced calculation time. In building physics, metamodels were recently used as predictive models based on simulations (Caldera et al. 2008) and measurements (Catalina et al. 2013) and in building optimisation (Eisenhower et al. 2012). In this paper, metamodels are constructed for the energy demand and indoor temperature of a semidetached dwelling to exemplify the benefits and drawbacks. Section 2 clarifies the robust design method; section 3 explains the metamodelling techniques used. The case study is described in section 4. In section 5, used techniques are compared as well as the number of training samples. The reliability of the best model is investigated and a strategy to construct reliable metamodels is proposed. 1 INTRODUCTION Building energy efficiency is raising concern regarding climate change and imminent fossil fuel depletion. In research and design of low-energy buildings deterministic simulations are commonly used to calculate the energy use and indoor climate and to optimise buildings by adjusting design parameters. However, many influential parameters such as user behaviour are inherently uncertain. As a result, deterministic simulations are unable to determine the optimal design measures for energy use and indoor climate, potentially resulting in solutions which are not favourable in all conditions. Accurate optimisation is however essential to investigate the most effective and robust measures as both governments and dwelling owners need confidence in optimal design solutions. In order to improve the optimisation process, uncertainties need to be taken into account through a robust design method in which for all performance criteria the target value is optimised and the impact of uncertain conditions is minimised. In building physics, uncertainty analysis was recently introduced to reliably calculate building performances under uncertainties (MacDonald & Strachan 2001, de Wit & Augenbroe 2002, Haarhoff & Mathews 2006). Advanced sampling schemes allow reducing the amount of simulations and thus calculation time, making such an uncertainty analysis realistic (Janssen 2013). Unfortunately, due to fewer samples, several design options are unreliably compared. As an example, one can calculate the output distributions with 200 optimal-schemed simulations instead of 2000 randomly selected, but comparing the two possible U-value solutions included in the simulations, becomes though if numerous pa- 2 ROBUST DESIGN METHOD Both occupants in particular and society in general need confidence in selected measures in optimisation as dwelling owners require guaranteed returns on their investments in energy efficiency and indoor climate, while governments must ensure that their subsidy programs have the desired impact. The de93 scheme used in a robustness study is a crossed array scheme with the same sampling scheme for the uncertain parameters of all design options, allowing the user to calculate the effectiveness and robustness of every design option (Sanchez et al. 1996). A full factorial scheme, in which all chosen values of design options are combined with all considered values of uncertain parameter, can be used, but more optimal schemes are available as well (Dehlendorff et al. 2011). Weight factors might be introduced to give priorities to several performance criteria and e.g. to effectiveness over robustness. Optimisation by a crossed array scheme can become time consuming if several uncertain and design parameters are taken into account, which is generally true for robust design problems in building physics. Even if the two layers in the design are constructed with an optimal scheme, still many combinations need to be taken into account. As metamodelling can enormously reduce calculation time, the next section will introduce some simple metamodels applicable for the current problem. y y50+P/2 (xi) y50+P/2 y50 (xi) y50-P/2(xi ) y50+P/2(xj ) y50 y50(xj) y50-P/2 (xj) y50-P/2 ymin xi xj x Figure 1. Effectiveness and robustness definition. velopment and promotion of effective and robust building envelope and service solutions is thus an important step to optimise the performances while limiting the spread, and thus to minimise the impact of uncertain conditions. In robust design, effectiveness and robustness of design options are evaluated at the same time for all performance criteria and all uncertain conditions and design options. The effectiveness ε and robustness RP of a design option xn for performance y are illustrated in Figure 1 and defined as (Van Gelder et al. 2013): ε (xn ) = 1 − y50 ( xn ) − y min y50 − y min RP ( x n ) = 1 − y50+ P / 2 ( xn ) − y50− P / 2 ( xn ) y50+ P / 2 − y50− P / 2 3 METAMODELLING Metamodels, also known as surrogate models, have the intention to stand in for the original model. The major advantage is theirs highly reduced calculation time. While for extreme cases, the original model might take days for one simulation, the metamodel only takes about a second. A metamodel is constructed for every output parameter with training data and validated on validation data coming from the original model as we want to design a metamodel which obtains good performance on unseen data. In general, all input and output data is standardised (zero mean, unit variance) to overcome influences from parameter units. One of the most important steps in metamodelling is to select validation criteria (Kleijnen & Sargent 2000) as the accuracy required for a predictive metamodel is usually very high. Polynomial regression and multivariate adaptive regression splines (MARS) are selected for this case study due to their simplicity. Both techniques quickly build models with short execution time. (1) (2) with P the user specified percentage of included sample points, yk the kth percentile under full uncertainty and yk(xn) the kth percentile after selecting design option xn. ymin corresponds to the simulated minimal value which is not an outlier, whereby an outlier is defined as a sample point smaller than y25-1.5(y75-y25). In this definition the performance indicator y is defined in such a way that it is greater or equal to zero and to be minimised. Effectiveness is thus defined as the improvement the median performance of a design option makes in proportion to the best possible reduction. The robustness is analogously determined as the improvement the performance spread of a design option makes in proportion to the spread under full uncertainty. According to this definition a measure with an effectiveness and robustness of one is the best possible, while negative values are to be avoided. Figure 1 shows as an example that design option xi is more robust, while xj is more effective. In robust design, effectiveness and robustness are included in the optimisation of design variables according to the performance criteria. The sampling 3.1 Polynomial regression Polynomial regression is one of the most known metamodelling techniques and fits a relation between the sampled input and output data using the method of least squares. In general, the model is a function of the form m k m m k k y = b0 + ∑ ∑ bni xin + ∑ ∑ ∑∑ bnpij xin x jp n =1 i =1 94 n =1 p =1 i =1 j =1 (3) with y the estimated output parameter, x the input parameters, k the amount of input parameters, m the order of the polynomial and b the regression coefficients (Jin et al. 2001, Wikipedia 2013a). First, second and third order polynomials are studied in this paper using Matlab tool polyfitn (D’Errico 2012). An additional advantage of first order polynomial regression is that the significance of the parameters can be identified from the coefficients. Table 1. Design parameters. Parameter Options* Type of ventilation . natural ventilation system . exhaust ventilation . balanced ventilation Infiltration rate U(0.45,12.5) m³/m²h Construction type . massive . timber frame U-value roof U(0.1,0.3) W/m²K U-value floor U(0.1,0.3) W/m²K U-value wall U(0.1,0.3) W/m²K U-value door U(0.8,2) W/m²K Type of window . U 2.07 W/m²K - g 0.631 . U 2.07 W/m²K – g 0.521 . U 1.29 W/m²K – g 0.631 . U 1.31 W/m²K – g 0.551 . U 0.7 W/m²K – g 0.407 Type of sunscreen . none . transmission 0.3 southern windows . transmission 0.3 all windows . transmission 0.1 southern windows . transmission 0.1 all windows Type of sunscreen . automatic on solar irradiance control . automatic on solar irradiance and indoor temperature . automatic on indoor temperature in summer . manual on in summer * Explanation of the symbols used: U(a,b): uniform distribution between a and b 3.2 MARS Multivariate adaptive regression splines (MARS) is a regression method that takes stepwise nonlinearities into account. The models are constructed through a forward/backward iterative approach and are of the form k y = ∑ ci Bi ( x ) (4) i =1 with y the estimated output parameter, x the input parameters, k the amount of basis functions Bi and ci the weight factors (Friedman 1991, Jin et al. 2001). A basis function is a constant, a hinge function or a product of hinge functions to take interactions into account (Wikipedia 2013b). A hinge function has the form max(0,xp-constant) or max(0,constant-xp). In the backward stage the least effective model terms are deleted to improve its generalization ability. The MARS models in this paper are piecewisecubic regression models created with the Matlab tool ARESLab (Jekabsons 2011). Table 2. Uncertain parameters. Parameter Distribution* Occupancy profile day zone 4 discrete profiles Occupancy profile night zone 3 discrete profiles Set temperature occupancy day zone N(21,1.35) °C Set temperature absence day zone D(15,no reduction) °C Set temperature occupancy night N(19,2) °C zone Recovery efficiency ventilation sysU(0.7,0.95) tem (only for balanced ventilation) Air change rate day zone Natural ventilation L(-0.8,0.528) 1/h Other L(-0.8,0.235) 1/h Air change rate night zone Natural ventilation L(0,0.39) 1/h Other L(0,0.115) 1/h Internal heat gains persons U(35,175) W Basis internal gains appliances U(20,180) W Summer internal gains appliances U(130,1000) W Winter internal gains appliances U(180,1300) W Spring and autumn internal gains apU(140,1150) W pliances * Explanation of the symbols used: N(µ,σ): normal distribution with mean value µ and standard deviation σ D(a,b): discrete uniform distribution between a and b U(a,b): uniform distribution between a and b L(µ,σ): lognormal distribution with mean value µ and standard deviation σ 4 CASE STUDY 4.1 Simulation model A semi-detached dwelling is modelled with two thermal zones and simulated in a transient simulation tool developed in Modelica (Baetens et al. 2012) for the reference climate year of Uccle, Belgium. This dwelling has a volume of 450 m³ and sun shading sheds, while the indoor temperature of the adjacent dwelling is assumed constant at 19 °C (Van Gelder et al. 2013). The hourly heat demand and indoor temperatures are calculated in these simulations under stochastic boundary conditions taking several design options into account. The resulting total annual heat demand, maximal temperature and amount of hours with temperatures exceeding 25 °C for the day zone are successively computed based on the simulation output. 4.2 Probabilistic approach Many parameters influencing heat demand and indoor temperature are either inherently uncertain, such as user behaviour and workmanship, or design variables to be optimised. Both design and uncertain 95 heat demand day zone 2 maximal relative error r² heat demand day zone 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 MARS first order second order third order 0 1 2 3 4 5 6 7 8 9 1.5 MARS first order 1 second order third order 0.5 0 10 0 1 2 3 4 5 6 7 8 9 10 training sets maximal temperature day zone 1.2 maximal relative error r² maximal temperature day zone 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 MARS first order second order third order 0 1 2 3 4 5 6 7 8 9 1 MARS first order 0.8 second order 0.6 third order 0.4 0.2 0 10 0 1 2 3 4 5 6 7 8 9 10 training sets TE 25°C day zone maximal relative error r² TE 25°C day zone 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 MARS first order second order third order 0 1 2 3 4 5 6 7 8 9 10 first order 14 second order 12 third order 10 8 6 4 2 0 0 training sets MARS 16 1 2 3 4 5 6 7 8 9 10 training sets Figure 4. Comparison of maximal relative errors of MARS and first, second and third polynomial regression for heat demand, maximal temperature and temperature exceedings of 25 °C of day zone with different number of training sets. Figure 2. Comparison of r²-values of MARS and first, second and third order polynomial regression for heat demand, maximal temperature and temperature exceedings of 25 °C of day zone with different number of training sets. heat demand day zone [kWh] simulation 20000 parameters are considered stochastic. The former are uniformly distributed (Table 1), while the latter are assigned a probability distribution inspired by a measurement campaign (Table 2) (Staepels et al. 2013, Van Gelder et al. 2013). A Monte Carlo simulation in Matlab is coupled to the dynamic simulation tool to process all uncertainties and design variables. The maximin Latin Hypercube advanced sampling technique was applied to reduce calculation time (Janssen 2013), when ten sets of 60 samples are created to enable bootstrapping. 15000 10000 y = 1.0479x - 389.4 R² = 0.9446 5000 0 0 5000 10000 15000 20000 metamodel Figure 3. Comparison of simulation and MARS model with six training samples for heat demand day zone. 96 5 RESULTS 1 5.1 Polynomial regression versus MARS 0.8 CDF Because ten simulation sets are available, we can compare both the metamodelling techniques and the number of training sets. One to nine sets are used to train the metamodels, while the tenth set is used for validation. As mentioned before, the simulation outputs heat demand, maximal temperature and amount of hours with temperatures above 25 °C (TE25°C) are modelled for the day zone based on the input parameters. Figure 2 shows the coefficients of determination r² for the outputs of both techniques in function of the number of training sets. The coefficient of determination r² indicates how accurate the original model is approximated by the metamodel as shown in Figure 3. A perfect correlation is given by an r²-value of 1 and trend line y=x. Figure 4 also shows the maximal relative error (MRE), which is the deviation between simulation and metamodel divided by the average simulation value, as both the overall fit and the maximal error are important in robust design. Comparing the validation indicators for the different outputs learns that MARS models have a better approximating ability than the polynomial regression models as higher-order polynomials tend to overfit the training data. As can be seen in Figures 2 and 4, the heat demand and maximal temperature are much easier to mimic than the temperature exceedings of 25 °C, even for few training sets. Furthermore, more training sets do not guarantee a better fit. For each output parameter, the MARS model based on six training sets shows good generalisation ability and is chosen as investigation object for the next subsection. 0.6 r² = 0.941 MRE = 0.33 0.4 simulation metamodel 0.2 0 0 10000 20000 30000 heat demand day zone [kWh] 1 CDF 0.8 r² = 0.9083 MRE = 0.06 0.6 simulation 0.4 metamodel 0.2 0 15 20 25 30 35 maximal temperature day zone [°C] 1 CDF 0.8 0.6 r² = 0.7827 MRE = 2.98 0.4 simulation metamodel 0.2 0 0 500 1000 1500 TE 25°C day zone [h] Figure 5. Cumulative distribution function of heat demand, maximal temperature and temperature exceedings of 25 °C of day zone based on four validation sets for original simulation and metamodel. 5.2 Reliability of metamodels The selected MARS models are now tested on the four remaining sets. The cumulative distribution function of the considered outputs of both original simulation and metamodel are compared in Figure 5. The coefficient of determination r² and the maximal relative error of the metamodel are given as well. One can conclude that the overall agreement is sufficient for the heat demand model. This model can be perfectly used in robust design. If one wants to include indoor comfort in the optimisation as well, one preferably opts for the maximal temperature model instead of temperature exceeding model because of its better similarity. In the previous subsection, several training sets were available and different models were built to select the best performing model. When making a metamodel in robust design optimisation, one wants to judge the constructed model independently of other potential models. Due to time constraints, one wants to create as less simulation sets as possible. Therefore the next paragraph investigates the reliability of a constructed model and thus the influence of training and validation sets on the goodness of the fit. Seven training set combinations with six out of nine available sets are used to construct metamodels as shown in Table 3. One can see that the metamodels based on some training set combinations perform better than on other sets. On the other hand, the validation set used also has its influence on the validation indicators, as presented in Table 4. The metamodels selected in the previous subsection, are validated on each of the four remaining sets. As it is important that the validation indicators do not differ in function of the used training and validation sets, it is recommended to test the metamodel for as many samples as possible. Because of calculation time, this is unfortunately impossible. 97 Table 3. Reliability of metamodel – different training sets Output parameTraining set r² MRE ter combination 1 0.9446 0.26 2 0.9402 0.30 3 0.9308 0.38 Heat demand 4 0.9515 0.26 day zone 5 0.9516 0.26 6 0.9595 0.27 7 0.9654 0.23 1 0.88 0.06 2 0.8274 0.07 3 0.8672 0.07 Maximal 4 0.8649 0.07 temperature 5 day zone 0.8017 0.13 6 0.8572 0.06 7 0.8429 0.09 1 0.7745 2.98 2 0.7346 3.40 3 0.6265 3.57 Temperature exceedings of 4 0.8228 2.46 25 °C day zone 5 0.8775 1.60 6 0.8858 1.81 7 0.865 2.43 Table 5. Building metamodel – cross-validation indicators The number of training sets, minimal, maximal and average r²-value and maximal relative error of all models after crossvalidation are indicated. Output r² MRE parame- Sets min max avg min max avg ter 2 0.64 0.80 0.72 0.48 0.85 0.67 3 0.90 0.95 0.93 0.28 0.33 0.31 Heat demand 4 0.85 0.95 0.91 0.28 0.34 0.30 day zone 5 0.93 0.95 0.94 0.23 0.34 0.27 6 0.93 0.95 0.94 0.24 0.39 0.29 2 0.85 0.89 0.87 0.05 0.07 0.06 3 0.90 0.92 0.91 0.06 0.06 0.06 Max. temp. day 4 0.85 0.91 0.88 0.04 0.08 0.06 zone 5 0.86 0.90 0.89 0.05 0.07 0.06 6 0.89 0.94 0.91 0.04 0.07 0.06 2 0.69 0.74 0.72 3.62 4.57 4.04 3 0.49 0.76 0.63 2.74 6.13 4.47 TE25°C 4 0.59 0.77 0.68 2.04 5.09 3.25 day zone 5 0.65 0.84 0.74 1.92 5.16 3.31 6 0.70 0.83 0.76 1.94 3.41 2.83 egy. The first step is to simulate two sampling sets. One of these sets is selected as training set, the other as validation set. A metamodel is constructed and r²value and maximal relative error are calculated for the validation set. To control the reliability of this model, an f-fold cross-validation is performed (Wikipedia 2013c). This implies that each set is once used as validation set, while the other set is a training set. In this way, as many metamodels are constructed and validated as simulation sets are available. Both r²-values and maximal relative errors of these metamodels are averaged. The minimal, maximal and average validation indicators are then compared with the validation criteria. If these criteria are met, the constructed metamodel can be used for robust design optimisation; otherwise, an additional training set is generated. Again, metamodels are constructed and a cross-validation is performed. The validation indicators are recalculated and again compared with the criteria. These steps are repeated until a metamodel is constructed which meets the validation criteria as exemplified in Table 5. The validation indicators of the metamodels after crossvalidation are presented. A comparison with validation criteria will be possible in further research, allowing the judgement of the constructed metamodels. As can be seen in Table 5, the metamodels for heat demand and maximal temperature models will meet these criteria with less simulations than the metamodel for temperature exceedings. Table 4. Reliability of metamodel – different validation sets Output Validation set r² MRE parameter 1 0.95 0.33 2 0.935 0.27 Heat demand day zone 3 0.9347 0.29 4 0.9446 0.26 1 0.939 0.05 Maximal 2 0.8955 0.06 temperature 3 0.9173 0.05 day zone 4 0.88 0.06 1 0.7817 1.83 Temperature 2 0.8101 1.58 exceedings of 3 0.7769 2.66 25 °C day zone 4 0.7745 2.98 5.3 Reliable metamodel construction The previous section shows that metamodels should be constructed and validated with care. Nevertheless, one wants to limit calculation time. To judge the metamodel accuracy, validation criteria, which are dependent on the goal of the model, are therefore needed. As stated before, the accuracy of predictive models is preferably very high. The required metamodel reliability has to be studied in further research to know which accuracy is sufficient to obtain the same robust design as with the original simulation tool. Based on that research, validation criteria will be determined. Both the overall fit and maximal errors are considered, of which the overall fit is probably most important. Once validation criteria are determined, metamodels can be constructed based on the proposed strat- 6 CONCLUSION This paper focussed on metamodelling in the context of robust design for low-energy dwellings. Although 98 Royal Statistical Society, Applied Statistics, 60(1), pp.31– 49. D’Errico J. 2012. Matlab extension polyfitn, available at http://www.mathworks.com/matlabcentral/fileexchange/347 65-polyfitn. De Wit M.S., and Augenbroe G. 2002. Analysis of uncertainty in building design evaluations and its implications. Energy and Buildings, 34(9):951–958. Eisenhower B., O’Neill Z., Narayanan S., Fonoberov V.A., and Mezic, I. 2012. A methodology for meta-model based optimization in building energy models. Energy and Buildings, 47:292–301. Friedman J.H. 1991. Multivariate adaptive regression splines. The Annals of Statistics, 19(1):1–141. Haarhoff J., and Mathews E. 2006. A Monte Carlo method for thermal building simulation. Energy and Buildings, 38(12):1395–1399. Husslage B., Rennen G., van Dam E.R., and den Hertog D. 2008. Space-filling Latin hypercube designs for computer experiments. Janssen H. 2013. Monte-Carlo based uncertainty analysis: Sampling efficiency and sampling convergence. Reliability Engineering & System Safety, 109:123–132. Jekabsons G. 2011. ARESLab Adaptive Regression Splines toolbox for Matlab/Octave. Riga. Jin R., Chen W., and Simpson T.W. 2001. Comparative studies of metamodelling techniques under multiple modelling criteria. Structural and Multidisciplinary Optimisation, 23:1– 13. Kleijnen J.P.C., and Sargent R.G. 2000. A methodology for fitting and validating metamodels in simulation. European Journal Of Operational Research, 120:14–29. Macdonald I.A., and Strachan P. 2001. Practical application of uncertainty analysis. Energy and Buildings, 33(3):219–227. Sanchez S.M., Sanchez P.J., Ramberg J.S., and Moeeni F. 1996. Effective engineering design through simulation. International Transactions in Operational Research, 3(2):169-185. Staepels L., Verbeeck G., Roels S., Van Gelder L., and Bauwens G. 2013. Do ventilation systems accomplish the necessary indoor air quality in low energy houses? In Clima 2013: 11th REHVA world congress & 8th international conference on IAQVEC, Prague, 16-19 June (to be published). Van Gelder L., Janssen H., Roels S., Verbeeck G., and Staepels L. 2013. Effective and robust measures for energy efficient dwellings: probabilistic determination. In 13th International Conference of the International Building Performance Simulation Association, 25th-30th August 2013, France (to be published). Wikipedia 2013a. Polynomial regression. http://en.wikipedia.org/wiki/Polynomial_regression. Last modified on 10 March 2013. Wikipedia 2013b. Multivariate adaptive regression splines. http://en.wikipedia.org/wiki/Multivariate_adaptive_regressi on_splines. Last modified on 31 March 2013. Wikipedia 2013c. Cross-validation (statistics). http://en.wikipedia.org/wiki/Cross-validation_(statistics). Last modified on 5 April 2013. deterministic simulations are commonly used in building optimisation, many influential parameters are inherently uncertain, with the result that deterministic simulations are unable to determine the optimal design measures. Taking the uncertainties into account through a robust design, which was explained in this paper, results in an improvement, but with the disadvantage of longer calculation time. To strongly improve the time efficiency, metamodelling was proposed. Metamodels were constructed for the energy demand and indoor temperature of a semi-detached dwelling. Comparison of polynomial regression and multivariate adaptive regression splines (MARS) learned that the fitting agreement of the latter was significantly better. The metamodels of heat demand and maximal temperature of the day zone seemed to better match the original simulation than the temperature exceedings of 25 °C. This showed the importance of selecting the most suitable output parameters for metamodelling. To construct a reliable metamodel with as less simulation sets as needed, this paper proposed a metamodelling strategy. Because training and validation sets may be of influence for the validation of the model, cross-validation is preferred. Sample sets are added to construct a metamodel meeting the validation criteria. However, more research is needed to define what accuracy is needed for a metamodel in robust design and to determine the validation criteria. These criteria handle the overall fit and maximal error of validation and cross-validation. 7 ACKNOWLEDGEMENTS The authors are very thankful for the fundings of the Flemish government for the IWT TETRA BEP2020project. They would like to thank Bart Husslage and Gijs Rennen from the Tilburg University as well for sharing their Matlab code for calculation of ‘maximin’ designs (Husslage et al. 2008). 8 REFERENCES Baetens R., De Coninck R., Van Roy J., Verbruggen B., Driesen J., Helsen L., and Saelens D. 2012. Assessing electrical bottlenecks at feeder level for residential net zeroenergy buildings by integrated system simulation. Applied Energy, 96:74–83. Caldera M., Corgnati S.P., and Filippi M. 2008. Energy demand for space heating through a statistical approach: application to residential buildings. Energy and Buildings, 40(10):1972–1983. Catalina T., Iordache V., and Caracaleanu B. 2013. Multiple regression model for fast prediction of the heating energy demand. Energy and Buildings, 57:302-312. Dehlendorff C., Kulahci M., and Andersen K.K. 2011. Designing simulation experiments with controllable and uncontrollable factors for applications in healthcare. Journal of the 99 100 Inverse modelling to predict and characterize indoor climates R.P. Kramer, A.W.M. van Schijndel, H.L. Schellen Eindhoven University of Technology, Department of the Built Environment, Eindhoven, The Netherlands ABSTRACT: Computational research on monumental buildings yields three problems regarding current building models: tedious modelling, relatively long simulation times, difficult to characterize the building by models parameters. A simplified building model with physical meaning is developed which is capable of simulating indoor temperature and humidity. The parameters of the model are identified by an optimization algorithm which fits the output of the model to measurements. The method consisted of: developing simplified building models based on insights from a literature review; fitting the models to a 17th century castle’s indoor climate; based on three criteria the best performing model is chosen and validated. The validation consisted of: residual analysis; applying the chosen model to a 16th century Cathedral, a parameter analysis. Concluding: inverse modelling is applied successfully to reproduce the free floating indoor climate of monumental buildings; characterization of the building remains a challenge. 1 INTRODUCTION timum parameters. Therefore, we focus on Linear Time Invariant models, specifically state space models, to ensure simulation speed. The article is structured as follows. Section 2 elaborates on the methodology, section 3 presents the results: 3.1 deals with state space performance, 3.2 with simplified model development and assessment and section 3.3 deals with model validation. Section 4 provides a discussion and section 5 presents the conclusions. Monumental buildings and their collections are important assets that form our cultural heritage. Unfortunately, they are exposed to all kinds of agents of deterioration including incorrect temperature and incorrect relative humidity (Michalski 1994). Computational research is conducted to relate deterioration to the indoor climate (Martens 2012). Moreover, it’s important to consider the monumental building itself (Michalski 2004). Especially moisture needs constant attention (Conrad 1996). Finally, effects of climate change on the cultural heritage for the years 2000 until 2100 are researched (Huijbregts et al. 2012). Computational modeling and simulation plays a vital role in performing research on the built cultural heritage. Given the aforementioned context, three problems are identified with respect to current building modeling: (i) detailed modeling of the buildings requires much effort: monumental buildings are old and protected. Therefore, blueprints are hard to find and destructive methods to obtain building material properties are often not allowed; (ii) simulation run times are long due to long simulation periods (up to 100 years with time step 1 hour) and detailed physical models; (iii) detailed models obstruct an easy characterization of the building. A simplified building model with physical meaning is desired which is capable of simulating temperature and humidity in which the parameters are identified from measurements by inverse modeling. The simplified model is needed for the prediction of indoor temperature and indoor humidity and for building characterization. Inverse modeling implies many repetitive simulations in order to find the op- 2 METHODOLOGY 2.1 Data acquisition Measured outdoor climate data are provided by the Royal Netherlands Meteorological Institute: temperature, relative humidity and global irradiance on the horizontal plane. Indoor measurements of Amerongen Castle’s king’s chamber are used for model development and selection (Fig. 1, left). Amerongen Castle is situated in the center of the Netherlands. It is a 17th century building, surrounded by a moat, with massive walls varying from 0.7 to 1.5 m thick. The building covers five floors. The main building materials are brick, wood and slate roof covering. The king’s chamber is located on the second floor and has windows oriented to the south and to the east. The indoor climate is free floating. One adjacent room is not free floating but has limited heating (set point 10°C). Indoor climate measurements of St. Bavo’s Cathedral’s south transept are used for model validation (Fig. 1, right). St. Bavo’s Cathedral is situated in Belgium. The measurements are performed in the south transept. The influence of solar irradi101 ance and the massive construction make the measurement data particularly valuable for model validation. The input Pe of the hygric model is calculated from measurements of outdoor relative humidity RHe and outdoor temperature Te according to (Künzel 1995). The solar inputs Irradi on vertical planes are calculated from global irradiance on the horizontal plane (Perez et al. 1987). Other climate data as precipitation and wind, are excluded to keep the amount of parameters at a minimum. This is an important aspect since more parameters lead to a higher uncertainty in parameter estimation (Ljung 1999). 2.3 Inverse modelling Inverse modeling is the inverse of traditional modeling. In traditional modeling, the system is known and the output is unknown. By modeling the system, the output can be simulated. In inverse modeling the output is known, e.g. measured, but little is known about the system’s parameters. The goal is to find the parameter set minimizing the error between the simulation result and measurements, formulated as an objective function. If the solution space includes multiple minima the goal is to find the global minimum, called global optimization. The optimization process is repeated with different solvers. Although Genetic Algorithms (GA) gain popularity by researchers (Huang & Lam 1997, Wang & Xu 2006a, Xu & Wang 2007), it is known that such Evolutionary Algorithms only find a nearoptimum solution. However, GA can be used to find an approximate solution and set this as the new initial value (Maeder et al. 2004, Babu & Murty 1993).Then the process is repeated with the PatternSearch algorithm to search the global solution space thoroughly and eventually the process is repeated with fmincon, a local solver, to check the solution or improve it. These algorithms are included in MATLAB’s Global Optimization Toolbox and Optimization Toolbox. Figure 1. Used case studies: Amerongen Castle [left] and St. Bavo’s Cathedral [right]. The indoor temperature and relative humidity are measured with combined humidity and temperature sensors. The relative humidity sensors have an absolute accuracy of ±2% RH from 10 to 90% RH and the temperature sensors have an accuracy of ±0.5 °C from 0 to 40 °C (Sensirion 2012). The sensors inside the buildings are placed at a height of 1-2 m above floor level. To obtain uniform room conditions, sensors are placed at the best possible locations avoiding the influence of air flow through the opening of doors and windows and heat loss through external walls (Huijbregts et al. 2012). 2.2 Hygrothermal model development The developed simplified model is a lumped building model. Based on the literature research (Kramer et al. 2012), different lumped model structures are developed which represent a building zone. The ordinary differential equations are derived from these RC-networks and transformed into state space matrices. State space models are Linear Time Invariant (LTI) models. A consequence is that the ODEs’ parameters are fixed coefficients. A linear model is often sufficiently accurate to describe the system dynamics (Ljung 1999). 2.4 Performance validation To quantify a model’s accuracy, three performance criteria are used: the Mean Squared Error (mse), Mean Absolute Error (mae) and Goodness of Fit (fit). The mse is calculated according to, (1) ∑ The first equation is known as state equation where x(t) is the state vector and u(t) is the input vector. The second equation is referred to as the output equation. A is the state matrix, B is the input matrix, C the output matrix and D the direct transition matrix. Thermal model inputs used: outdoor temperature (Te [°C]); solar irradiation on vertical planes oriented on north, east, south and west (IrradN, IrradE, IrradS, IrradW [W/m2]); fixed temperature node (Tfixed [°C]). Hygric model inputs used: outdoor vapor pressure (Pe [Pa]) and fixed vapor pressure node (Pfixed [Pa]). (2) and the mae is calculated according to, ∑ | | 3 and the fit is calculated according to, 100 ∙ 1 4 where = measured signal; = simulated signal; = average measured signal; N = number of samples. 102 is the Euclidean length of the vector , also known as the magnitude. Equation (4) therefore calculates in the numerator the magnitude of the simulation error and in the denominator, it calculates how much the measured signal fluctuates around its mean. Consequently, the goodness of fit criterion is robust with respect to the fluctuation level of the signal. 3.2 Hygrothermal model development The developed thermal models are presented in Figure 3. The ability of these models to reproduce the indoor climate of a whole building zone is assessed. Two variants are tested of all these thermal models: variant (a) and variant (b). The variation is the capacitance on which the input solar irradiance is placed: Cw (walls), Ci (indoor air) or Cint (interior). The variable Gfast represents fast heat losses, e.g. ventilation and transmission through glazing. The variable Tfixed represents ground contact or a connection to another steady temperature zone. 3 RESULTS 3.1 Simulation speed and accuracy The simplified hygrothermal building model consists of a set of Ordinary Differential Equations. The ODEs of the model are usually solved by an algorithm, e.g. the ode23 algorithm in the MATLAB environment which is a variable step-size solver. However, the step size is in the order of seconds resulting in long simulation times for longer simulation periods. On the contrary, if the ODEs are transformed into a state space model structure according to Equation 1, the simulation step size can be increased for building simulations to 1 hour. A comparison of the simulation run time between ode23 and state space for model 1 in Figure 3 is given in Table 1. Table 1. Calculation time [s] of model 1 (top in Fig. 3) solved as state space or ode23 for different simulation periods. 1month 1 year 10 years 100 years [s] [s] [s] [s] ode23 5 89 ̴ 900* ̴ 9000* state space 0.016 0.016 0.050 0.45 * Obtained by extrapolation. The results demonstrate the simulation time savings of the state space approach. The state space simulation is thousands of times faster. But does it influence the simulation accuracy? Figure 2 shows the simulation result for the model solved in both ways. The state space output coincides with the ode23 output accurately. Therefore, the state space approach with time steps of 1 hour may be used confidently for building simulations of monumental buildings which are free floating. Figure 3. Developed simplified thermal models, model 1 at the top and model 5 at the bottom. Hygric models are identical, but with vapor pressure as input and excluding solar irradiation. The hygric models’ structure is identical, with the vapor pressure being the driving force and excluding solar irradiance as input. Consequently, the hygric model has only two inputs being the outdoor vapor pressure Pe and the fixed vapor pressure Pfixed. The models’ outputs are fitted by the optimization procedure to indoor measurements of Amerongen Castle’s king’s chamber. Model performance is assessed by the three criteria mse, mae and fit (Equations 2-4). See Table 2 for thermal performance and Table 3 for hygric performance. Thermal model 2 without Tfixed is also fitted to the king’s chamber showing a significantly worse fit than thermal model 2 including Tfixed. Thermal model 3a is considered to be the simplest model structure Figure 2. State space model versus ode23 simulation accuracy. 103 reproducing the thermal indoor climate accurately. It is able to reproduce the indoor climate sufficiently accurate with a mse of only 0.63°C for a simulation period of 8 months. Therefore model 3a is chosen to be the best thermal model structure and will be validated. (5) Table 2. Thermal models’ performance (fitted to indoor measurements of king’s chamber). Model 3a is chosen. model mse [°C2] 1a - - - 1b - - - 2a 0.92 / 4.05* 0.75 / 1.72* 78.37 / 54.57* 2b 0.86 / 3.97* 0.72 / 1.70* 79.00 / 55.00* 3a 0.63 0.61 82.11 3b 0.86 0.72 79.01 4a 0.95 0.76 84.35 4b 0.95 0.76 84.32 5a 0.96 0.76 84.29 5b 0.69 0.65 86.68 mae [°C] • fit [%] These ODEs are transformed into state space matrices A, B, C, D to form the state space structure. 3.3 Validation 3.3.1 Residual analysis The part of the measured signal which is unexplained by the model forms the residuals. Hence, (6) * Without fixed temperature node. where = residuals; = measured signal; = simulated signal. Possible reasons for the remaining residuals are measurement errors, erroneous modeling of inputs, missing inputs, over simplified model. The residual analysis consists of two tests: The whiteness test and the independence test. The whiteness test is used to analyze the autocorrelation between the residuals. Ideally, the residuals only consist of measurement errors as white noise and the autocorrelation is within acceptable limits. If the model fails on the whiteness test, there is a strong indication that inputs are missing. The independence test is used to analyze the cross correlation between residuals and inputs. A significant cross correlation indicates that the influence of input on output is not correctly described by the model. This denotes an incorrect model structure. Figure 4 shows the autocorrelation and cross correlation for the models. Both models’ autocorrelation exceed the tolerated bandwidth. This is an indication of missing inputs. However, Ljung states that less attention should be paid to the autocorrelation function if no error model is included (Ljung 1999). To verify this, a state space model including error model is fitted using MATLAB’s System Identification Toolbox. Although the goodness of fit is only 0.33% higher, the auto correlation fits within the tolerated band width. This is an indication that the autocorrelation results from a missing error model and not primarily from missing inputs. The cross correlation of both models is within the tolerated bandwidth for all used inputs: this shows that the models’ structures are correct. Table 3 shows the results of the hygric models. Hygric model 2 is considered to be the simplest model structure reproducing the hygric indoor climate accurately. It is able to reproduce the king’s chamber humidity with a fit of 86%. Therefore model 2 is chosen for further validation. Table 3. Hygric models’ performance (fitted to indoor measurements of king’s chamber). Model 2 is chosen. model mse [Pa2] mae [Pa] fit [%] 1 4440 51 82.76 2 3023 45 85.77 3 3016 45 85.79 4 3023 45 85.77 5 3019 45 85.78 The ODEs of the selected models are given by Equation (5). From top to bottom: the first equation describes the temperature change of the building envelope, i.e. the walls; the second equation describes the temperature change of the indoor air; the third equation describes the temperature change of interior parts; the fourth equation describes the partial vapor pressure change of the building envelope; the fifth equation describes the partial vapor pressure change of the indoor air. These equations result in a 3rd-order thermal model and 2nd-order hygric model. 104 3.3.2 Case study The simplified hygrothermal model is fitted to indoor climate measurements of St. Bavo’s Cathedral. Recall that the hygric model’s input Pe is calculated from measured Te and RHe. Consequently, the hygric model’s output is simulated indoor vapor pressure Pisim. Because RH is more relevant for object and building conservation, simulated indoor relative humidity RHisim is calculated from Tisim and Pisim. The figures below show the results of the inverse modeling procedure in time plots: the entire simulation period in Figure 5 and a detail view in Figure 6. Autocorrelation of residuals from Tisim 1 0.5 0 -0.5 0 5 10 15 20 25 Cross correlation between input Te and residuals from Tisim 0.05 0 -0.05 -20 -15 -10 -5 0 5 10 15 20 Cross correlation between input IrradN and residuals from Tisim 0.2 0 meas sim -0.2 -20 -15 -10 -5 0 5 10 15 Ti [°C] 20 20 Cross correlation between input IrradE and residuals from Tisim 0.2 15 10 0 1000 -0.2 -20 -15 -10 -5 0 5 10 15 20 Cross correlation between input IrradS and residuals from Tisim RH [%] 0 -20 -15 -10 -5 0 5 10 15 4000 90 0.2 -0.2 2000 3000 time [hours] 70 meas sim 50 20 Cross correlation between input IrradW and residuals from Tisim 0.2 1000 2000 3000 time [hours] 4000 Figure 5. Thermal performance [top] and hygric performance [bottom] for St. Bavo’s Cathedral (period = 6 months). 0 -0.2 -20 -15 -10 -5 0 5 10 15 20 Cross correlation between input Tfixed and residuals from Tisim 0.5 Ti [°C] 22 0 -0.5 -20 -15 -10 -5 0 5 10 15 21 20 20 Autocorrelation of residuals from Pisim 1 1050 1150 time [hours] 1250 1050 1150 time [hours] 1250 0.5 0 75 0 5 10 15 20 Cross correlation between input Pe and residuals from Pisim RH [%] -0.5 25 0.5 70 65 60 0 -0.5 -20 -15 -10 -5 0 5 10 15 20 Figure 6. Thermal performance [top] and hygric performance [bottom] for St. Bavo’s Cathedral (period = 2 weeks). Cross correlation between input Pfixed and residuals from Pisim 0.5 0 -0.5 -20 -15 -10 -5 0 5 10 15 Figure 5 shows that seasonal dynamics are reproduced very well by the simplified model. Figure 6 shows that the model also reproduces the diurnal dynamics accurately. 20 Figure 4. autocorrelation and cross correlation functions of models fitted to king’s chamber measurements. The gray area represents the tolerated bandwidth. 105 Figure 7 shows the cumulative distribution function of the thermal and hygric errors. Recall that the measurement accuracy of the T-sensor is ±0.5°C and the measurement accuracy of the RH-sensor is ±2%. It shows that 80% of the thermal errors are within ±0.5°C and 70% of the hygric errors are within ±2% RH. Moreover, both distribution functions are symmetrically positioned around zero indicating that the residuals are unbiased. 1 RH [%] T [°C] probability 0.8 0.6 1,E+02 1,E+01 1,E+00 1,E-01 1,E-02 1,E-03 1,E-04 1,E-05 1,E-06 1,E-07 1,E-08 1,E-09 1,E-10 1,E-11 1,E-12 1,E-13 0.4 Figure 8. Thermal parameters for St. Bavo’s Cathedral. 0.2 0 −6 −4 −2 −1 0 1 2 error T [°C] and RH [%] 4 6 1,E+04 Figure 7. Cumulative Distribution Function of simulation errors for case study St. Bavo Cathedral. 1,E+03 1,E+02 1,E+01 Table 4 shows the thermal and hygric performance expressed in mse, mae and fit (Equations 24). To show the influence of sun irradiation, the model has also been fitted to St. Bavo´s Cathedral indoor measurements excluding solar input. The results are indicated in Table 4 by *. This demonstrates unambiguously that sun irradiation influences the thermal indoor climate significantly and should also be included in simplified building models. 1,E+00 1,E-01 1,E-02 1,E-03 1,E-04 1,E-05 1,E-06 1,E-07 1,E-08 Table 4. Models´ performance expressed in three criteria (fitted to measured indoor climate of St. Bavo’s Cathedral). mse mae fit thermal model 0.74* /0.17°C2 2 hygric model 1870 Pa * without sun irradiation. 0.69*/0.32°C 32 Pa Figure 9. Hygric parameters for St. Bavo’s Cathedral. 81*/91% 4 DISCUSSION 86% The thermal and hygric models are presented in state space form resulting in fast simulations (100y, time step 1h → 0.5s on i5 processor 2.4 GHz). Hudson & Underwood (1999) used state space for their building model, but implemented the model in Simulink which uses a 4th order Runge-Kutta solver yielding simulation times equal to the ode23 algorithm. It is shown that the state space model achieves the same accuracy with a time step of 1 hour as the ode23 algorithm with a small variable time step. Fast simulations are crucial for the feasibility of the optimization process: the optimization algorithm repeats the simulation thousands of times to identify the optimum parameters. With the state space model, this optimization only takes ̴ minutes. The best results have been obtained by: using Genetic Algorithm to determine feasible initial values, using di- The identified parameters for St. Bavo’s Cathedral are shown in Figure 8 and Figure 9 for the thermal and hygric model respectively. The ODEs of the models (Equation 5) show a capacitance Cx on the left of the equal sign and the remaining variables Gx on the right of the equal sign. Consequently, the identified parameters are a ratio of the variable and its connected capacitance, i.e. Gx/Cx. Due to the large differences between indoor air capacitance Ci and envelope capacitance Cw, the resulting parameters’ magnitude vary accordingly in the range of 1e+1 to 1e-12 for the thermal model and the hygric model’s parameters vary in the range of 1e-1 to 1e-10. This makes physical interpretation and applicability for characterization difficult. 106 that the identified parameters are a ratio Gx/Cx. Other researches concerning inverse building modeling assumed foreknowledge of certain parameters (Mathews et al. 1991), additionally measured specific signals, e.g. heat loads (Jiménez et al. 2008, Lundin et al. 2004, Penman 1990, Wang & Xu 2006b) or steady state situations eliminating the capacitance Cx (Norlén 1990). This allows for the identification of individual parameters rather than a ratio. The widely used approach of measuring heat loads is no solution since many monumental buildings are free floating. Secondly, identified parameter values are difficult to verify since they represent the effective part: the theoretical value, i.e. apparent value, for thermal capacitance of the envelope, which results from calculations including all envelope walls, deviates significantly from the effective heat capacitance (K. A. Antonopoulos & Koronaki 1998, 1999, Mathews et al. 1991). Be aware that the identification results are effective parameters rather than apparent parameters making verification difficult. rect search algorithm PatternSearch to search with generous solution space between lower and upper bounds of parameters, using fmincon towards the end of the optimization process to check or fine-tune the optimum. Because PatternSearch is a gradient free algorithm, it is able to handle unsmooth and discontinuous solution spaces. Since fmincon is gradient based, it finds the optimum efficiently and is preferred in the final stage. Wang and Xu (2006a, 2006b) and Xu & Wang (2007) have used Genetic Algorithm to find their parameters and proclaim GA to be the best optimization algorithm for the task. Literature agrees that GA will only find a near optimal solution. It is found in this research that GA can search a vast solution space very efficiently, however, the so called ‘near optimal’ solution is often unsatisfying. Furthermore, since it is a stochastic algorithm, the solution will vary every time the algorithm is executed. Also Ljung (1999) recommends PatternSearch above GA. A state space model is a Linear Time Invariant model. Therefore the parameters in the model are fixed coefficients. This is a challenge if used for inverse modeling: the indoor climate is influenced by time varying processes, e.g. use of sun blinds, internal heat and moisture sources by people. Furthermore, nonlinear processes are unlikely to be reproduced correctly, e.g. opening and closing of doors. This research shows that monumental buildings’ indoor climates can be reproduced properly with the state space model: many monumental buildings are not used intensively and therefore little time varying processes occur. Secondly, many monumental buildings are characterized by high capacities and are therefore fairly insensitive to time varying disturbances, e.g. as is the case with St. Bavo’s Cathedral. This research shows that including a fixed temperature node Tfixed in the model is a simple, yet effective method for modeling ground contact of monumental buildings since these buildings often lack floor insulation. Since solar irradiation is a dynamical process, the following approach is applied for modeling solar irradiance: four signals are used as input, each representing solar irradiance on a vertical wall with orientations north, east, south and west. St. Bavo’s Cathedral has been simulated with and without sun irradiation. The result shows unambiguously that solar irradiation should be taken into account and that the used modeling approach gives accurate results. The simplified hygrothermal model is used for the simulation of indoor temperature and humidity. Additionally, it is desired for an easy characterization of the building by the identified parameters. The latter turns out to be very challenging for two main reasons. Firstly, this research assumes the absence of any foreknowledge of certain variables. Referring to the capacitances on the left side and remaining parameters on the right side in the ODEs, this implies 5 CONCLUSION A new simplified hygrothermal building model in state space form is presented with an inverse modeling technique to identify its parameters. Results show that the simplified 3rd-order thermal and 2ndorder hygric models are capable of reproducing the case studies’ indoor climates accurately. Moreover, the state space model results in fast simulations. This is essential for the optimization process: the optimization time is in the order of minutes. GA is recommended to find initial values, PatternSearch to search globally and fmincon to check or fine tune the optimum. Characterization by and validation of the parameter values is challenging: additional measurements are required to identify individual parameters and parameter values are effective values, not apparent values. The fixed temperature and fixed vapor pressure node are important aspects in the model’s structure and improve model performance significantly. Also the modeling of sun irradiation performs correctly. 107 6 REFERENCES Phanendra Babu G., and Narasimha Murty M. 1993. A nearoptimal initial seed value selection in K-means means algorithm using a genetic algorithm. Pattern Recognition Letters, 14(10): 763–769. Sensirion 2012. Retrieved January 13, 2012, from http://www.sensirion.com Wang S., and Xu X. 2006a. Simplified building model for transient thermal performance estimation using GA-based parameter identification. International Journal of Thermal Sciences 45(4): 419–432. Wang S., and Xu X. 2006b. Parameter estimation of internal thermal mass of building dynamic models using genetic algorithm. Energy Conversion and Management 47: 1927– 1941. Xu X., and Wang S. 2007. Optimal simplified thermal models of building envelope based on frequency domain regression using genetic algorithm. Energy and Buildings 39(5): 525– 536. Antonopoulos K. A., and Koronaki E. 1998. Apparent and effective thermal capacitance of buildings. Energy 23(3): 183–192. Antonopoulos K. A., and Koronaki E. 1999. Envelope and indoor thermal capacitance of buildings. Applied Thermal Engineering 19(7): 743–756. Conrad E. A. 1996. The Dews and Donts Of Insulating. OldHouse Journal 24(3): 36–41. Huang W., and Lam H. N. 1997. Using genetic algorithms to optimize controller parameters for HVAC systems. Energy and Buildings 26: 277–282. Hudson G., and Underwood C. P. 1999. A simple building modelling procedure for MATLAB / Simulink. Proceedings of the 6th International Conference on Building Performance Simulation: 777–783. Kyoto. Huijbregts Z., Kramer R. P., Martens M. H. J., Van Schijndel A. W. M., and Schellen H. L. 2012. A proposed method to assess the damage risk of future climate change to museum objects in historic buildings. Building and Environment 55(sept): 43–56. Jiménez M. J., Madsen H., and Andersen K. K. 2008. Identification of the main thermal characteristics of building components using MATLAB. Building and Environment 43(2): 170–180. Kramer R., Van Schijndel J., and Schellen H. 2012. Simplified thermal and hygric building models: A literature review. Frontiers of Architectural Research 1(4): 318–325. Künzel H. M. 1995. Simultaneous Heat and Moisture Transport in Building Components: One- and twodimensional calculation using simple parameters . Fraunhofer Institute of Building Physics. Ljung L. 1999. System Identification: Theory for the User (2nd ed.). Upper Saddle River (New York): Prentice-Hall. Lundin M., Andersson S., and Östin R. 2004. Development and validation of a method aimed at estimating building performance parameters. Energy and Buildings 36(9): 905– 914. Maeder M., Neuhold Y.M., and Puxty G. 2004. Application of a genetic algorithm: near optimal estimation of the rate and equilibrium constants of complex reaction mechanisms. Chemometrics and Intelligent Laboratory Systems 70(2): 193–203. Martens M. H. J. 2012. Climate Risk Assessment In Museums: degradation risks determined from temperature and relative humidity data. Eindhoven University of Technology. Mathews E. H., Rousseau P. G., Richards P. G., and Lombard C. 1991. A Procedure to Estimate the Effective Heat Storage Capability of a Building. Building and Environment 26(2): 179–188. Michalski S. 1994. A systematic approach to preservation: description and integration with other museum activities. In A. Roy & P. Smith (Eds.), Preventive Conservation: practice, theory and research: 8–11. London: International Institute for Conservation. Michalski S. 2004. Care and preservation of collections. Running a museum: a practical handbook: 51–90. Paris: International Council of Museums. Norlén U. 1990. Estimating Thermal Parameters of Outdoor Test Cells. Building and Environment 25(1): 17–24. Penman J. M. 1990. Second order system identification in the thermal response of a working school. Building and Environment 25(2): 105–110 Perez R., Seals R., Ineichen P., Stewart R., and Menicucci D. 1987. A new simplified version of the perez diffuse irradiance model for tilted surfaces. Solar Energy 39(3): 221–231. 108 Individualised climate in future buildings. Fact or fiction? M. Dovjak & A. Krainer University of Ljubljana, Faculty of Civil and Geodetic Engineering, Chair for Buildings and Constructional Complexes, Ljubljana, Slovenia M. Shukuya Tokyo City University, Laboratory of Building Environment, Yokohama, Japan ABSTRACT: The purpose of the paper is to design and test a user–centred heating and cooling (H/C) system that enables to create optimal conditions for individual user and at the same time minimal possible energy use for H/C of residential buildings. The user–centred system was designed with upgraded methodology of engineering design. The system was installed in a test active space and compared with conventional system. It includes six ceiling radiative panels connected with an integrated control system of internal environment on the basis of fuzzy logic. During the experiment, environmental parameters and energy use for H/C were measured. Thermal comfort conditions for individual users were analysed with the calculation of human body exergy balance for three virtual residential users. The system enables to create comfort conditions for individual user and it provides zoning of the space. Regulation of environmental parameters and creation of optimal conditions for individual users are in favour to lower energy use for H/C. 1 INTRODUCTION The first user–centred system that creates comfort and healthy conditions was designed by Dovjak (2012a, 2012b). The system was tested in highly demanding hospital environment, in a model room for burn patient. It created optional conditions for healthcare and treatment of patients. The purpose of this paper is to develop and test user–centred H/C system for application in residential buildings. The goal is to create optimal conditions for individual user and at the same time minimal possible energy use for H/C of active spaces. Flexibility of the system was proven on specific users of the space as well as for various activities. The system was compared with a reference conventional system. “Individuality is the state or quality of being an individual; a person separate from others and possessing his or her own needs, goals, and desires; single person with individual characteristics, who is considered separately (Crowie et al. 1994).” Build environment should be treated as interactive active space of users, environmental factors and specific activities. In active space various users are present with different demands and needs for thermal comfort conditions. Current heating, ventilation and air–conditioning systems are not designed on the requirements of individual users and often result in dissatisfaction, decreased productivity and high energy use for heating and cooling (H/C) purposes (Pheasant 1991). Selkowitz, head of the Building technologies department of the Environmental Energy Technologies Division at the Lawrence Berkeley Lab emphasised the importance of understanding a cause and an effect on the level of setting priorities and decision-making. He calculated that energy costs presents about $2 per square foot (approx. $21.53 per m2) per year, and people cost about $2152.78 per square foot (approx. $18.6 per m2) in an office building (Peyton 1999). Even a small improvement in productivity and reduction in absenteeism are be more worthy than any energy saving. Thus, any effort towards designing user friendly environment has great benefits. 2 METHODS 2.1 Design of a system The methodology of the design was based on the Morris Asimov’s morphology of engineering design that comprises seven phases (Figure 1) (Asimow 1962). The methodology was upgraded and applied considering the specifics of built environment. The main emphasis was on the first phase, i.e. Step 1– Definition of needs, demands and conditions, Step 2–Design problem, and Step 3–Synthesis of the possible solutions. Outputs of feasibility study present the basis for Phase II–Preliminary Design. 109 Needs, demands and conditions (Step 1) required for user comfort can be obtained optimally by analysing real state conditions. State–of–the art analysis was focused on searching for the data from the most relevant literature, regarding time, location (Slovenia, EU, world), type of literature (articles, studies, statistic data, regulations, guidelines, standards, manuals, scientific reports, monographs, etc.) and research filed. Actual conditions can be confronted to regulations and guidelines. At that level, irregularities and health impacts can be detected and also possible solutions can be defined. Design of the problem (Step 2) arose from the relation user–system–built environment. It includes detection and classification of the problems on the basis of real–time conditions. Design and application of matrix enable holistic consideration of impact factors and their interactions in all phases of engineering design. For the synthesis of the possible solutions, human body exergy balance model was used. Results of the executed phases of design are presented in more detail in the study by Dovjak (2012a). maintain human body in zero energy and mass balance (Bresjanac & Rupnik 1999). There are three essential components of all homeostatic control mechanisms: detector, integrator and effector (Fig. 2). The detector monitors and responds to stimulus in the environment. It sends information to an integrator that sets the range at which a variable is maintained. The integrator determines an appropriate response to the stimulus and sends signals to an effector. After receiving the signal, a change occurs to correct the deviation by enhancing it with feedback mechanisms (Bresjanac & Rupnik 1999). Figure 2. Basic elements of homeostatic control mechanism (Bresjanac & Rupnik 1999). The system works in such a way that the deviations between set point and the measured values is as small as possible. The result is stable cell environment (Bresjanac & Rupnik 1999). The tested user–centred system includes six low– temperature–heating and high–temperature–cooling ceiling radiative panels connected with an integrated control system of internal environment on the basis of fuzzy logic (ICSIE system). The system is installed into a test active space (7.5 m x 5.0 m x 4.0 m) that has one exterior wall with 15 m2 glazed window; other walls are interior. It is located at the Chair for Buildings and Constructional Complexes, Faculty of Civil and Geodetic Engineering, University of Ljubljana. Panels cover 9 m2 of the ceiling. The final layer of four panels is a contact stuck 1.25 cm gypsum board, and two panels have stone plates (one is a compact 2 cm marble plate and another a composite 12 mm Al-honeycomb with 3 mm tick stone plate). All panels are fixed with 4 steel screws into ceiling construction. Panels are connected into H/C system with valves for switching off every panel separately, and with a pump on thermostatic mixing valve. Switching between hot and cool water entering into the panels is manual. Panels are connected with ICSIE system that was developed by TrobecLah (2003) and upgraded by Košir (2008). It enables the control of indoor air temperature, CO2 and illuminance under the influence of outdoor environment and users` requests. The basic elements of the ICSIE Figure 1. System design. 2.2 User-centred design of a system „If a product (environment or system) is intended for human use, then its design should be based on the characteristics of its human users“ (Pheasant 1991). The basis for the design of the system presents vital physiological processes in human body that require a dynamic constancy or balance. The process in which the body maintains internal equilibrium and allows normal functionality is called homeostasis, and it is maintained by regulatory mechanisms that operate by negative feedback mechanisms (Bresjanac & Rupnik 1999). Homeostatic mechanisms 110 for every individual (Dovjak 2012a, 2012b; Regulation EU 305/2011; Preamble WHO 1946). system are the same as the elements of homeostatic control mechanism. They consist of sensor network system (detector), regulation system (integrator) and actuator system (effector). The basic architecture of the system is presented in Figure 3. Table 1. Individual demands and needs for thermal comfort conditions of various users with references. To = operative temperature; Tai = room air temperature; RHin = relative humidity of indoor air. User/activity To, Tai [°C] RHin [%] Reference Newborns: 1 kg body To =35°C 30-60% Hey and Katz weight 1970 >2.5 kg To =32°C 30-60% Hey and Katz body weight 1970 Burn patient To=32°C Up to 95% ASHRAE 2007 Elderly Tai =24°C 30-60% Rohles and Johnson 1972 WeightTai=18-20°C 40-60% IFA 2013 training Yoga Tai=27°C 40-60% IFA 2013 A human being’s thermal sensation is influenced by metabolic rate and clothing, as well as the environmental parameters (Fanger 1970; ISO 7730: 2005), individual characteristics (gender differences, anthropometric characteristics, cultural differences), and health status (Dovjak 2012a, 2012b; Hwang et al. 2007; Parsons 2002). Individual demands and needs for thermal comfort conditions of various users are presented in Table 1. For the purpose of our study three virtual residential users were simulated for the analysis of individual thermal comfort conditions. Users’ characteristics and specific activities are presented in Table 2. The user– centred system was tested regarding simulation of individual thermal comfort conditions and measured energy use. The efficiency of the user–centred system was compared with conventional system (oil-filled electric heaters and split system with indoor A/C unit). Figure 3. Basic architecture of the ICSIE system (Tai–room air temperature, Tao–outdoor air temperature, RHin–relative humidity of indoor air, RHout–relative humidity of outdoor air, ILin1 & ILin2–internal work plane illumination, ILout–external illumination, CCO2–concentration of CO2, Irgo–direct solar radiation, Irdo–reflected solar radiation, Wp–wind speed, Wd–wind direction, Pe–precipitation detection, Cheat–energy use for heating, Ccool– energy use for cooling) (Košir 2008). 2.3 Individual demands and needs The main goal in the built environment is to achieve health and comfort conditions (Preamble WHO 1946; Regulative 306/2011) and to satisfy basic physiological needs (Maslow 1943). Man as a physiological, sociological and psychological being is daily subjected to a large number of needs. The needs arise as a result of some imbalances inside of the human body or due to other outside factors (Musek & Pečjak 1997). As soon as the need arises, the aspiration for its fulfilment appears (Musek & Pečjak 1997). Needs can be fulfilled instinctively or reasonably. The physiological needs can be fulfilled with the mechanism of homeostasis or progressively (Musek & Pečjak 1997). American psychologist Abraham Maslow proposed a theory called Maslow's hierarchy of needs, where needs are listed in the shape of a pyramid. The largest and most fundamental physiological needs are positioned at the bottom level and the psychological needs are positioned at higher levels. Maslow's theory suggests that the most basic level of needs must be met before an individual will strongly desire or focus motivation upon the secondary or higher level needs. Therefore, environmental parameters of the thermal comfort present one of the basic physiological needs in Maslow's hierarchy of needs (Maslow 1943). Their satisfaction is necessary for the state of homeostasis of the human body. Consequently, the main guidance for the design of built environment is the creation of healthy and comfortable conditions Table 2. Users’ characteristics and specific activities. User/activity Metabolic Effective clothing rate [met] insulation [clo] Grandfather, 1.0 0.7 watching TV Teenager, 6.0 0.2 weight-training Mother, 1.2 0.7 Yoga In the simulation users were exposed to experimental conditions based on in–situ real–time measurements. Set up parameters for both systems are defined according to required conditions and studies (ANSI/ASHRAE Standard 55: 2004; IFA 2013; Rohles & Johnson 1972; Rules 2002). In the case of conventional system Tai was equal to Tmr (mean radiant temperature) and To. In the case of user– centred system Tai, Tmr and To differed. User–centred system enabled to set up different combinations of Tai and Tmr and To that resulted in optimal human body exergy balance for every individual separately. 111 Experimental conditions in Table 3 present one of the real–time conditions that were selected for further simulation of individual thermal comfort conditions. The designed system enables to provide the comfort parameters for three virtual persons at the same time. To maintain comfort conditions, it is important that the exergy consumption and stored exergy are at optimal values with a rational combination of exergy input and output. Individual thermal comfort conditions were analysed by human body exergy balance, calculated human body exergy consumption rates and predicted mean votes (PMV) index with spread sheet software developed by Hideo Asada (Shukuya et al. 2010). For exergy calculations, the reference environmental temperature (the outdoor environmental temperature, Tao) and RHout are set to be equal to Tai and RHin. Table 3. Real-time experimental conditions for the simulation of individual thermal comfort conditions. System User Tai Tmr v RHin [°C] [°C] [m/s] [%] Conventional All 21 21 0.1 40 User–centred Grandfather Teenager Mother 22 17 22 24 19 24 0.1 0.1 0.1 40 40 40 3 RESULTS OF TESTING PHASE AND DISCUSSION 2.4 Application of the human body exergy balance model 3.1 Human body exergy balance and conventional system For the analysis of individual thermal comfort conditions, exergy concept was introduced. Exergy analysis jointly treats processes inside the human body and processes in built environment. It enables to design health and comfort conditions for everyone. Human body is treated as a thermodynamic system based on exergy–entropy processes. The system consists of a core and a shell and is simulated in a test room with the environmental temperature. Thermal exergy balance of human body (Shukuya et al. 2010) was derived by combining the water balance equation, the energy balance equation and the entropy balance equation under steady–state condition. All of them are the resultant equations of the mathematical operations described in Shukuya et al. (2010), together with the environmental temperature for exergy calculation. The general form of the exergy balance equation for a human body as a system is represented in Eq. (1) (Shukuya et al. 2010): Figure 4 shows the example of the whole human body exergy balances for thee virtual users of the active space equipped with conventional system. Exergy input − Exergy consumption = Exergy stored + Exergy output Figure 4. Human body exergy balances for three virtual users of active space equipped with conventional system. All three human body exergy balances have in common that the input exergy presents thermal radiative exergy exchange between the human body and the surrounding surfaces of active space, which has influence on thermal comfort. Warm/cool radiant exergy rate absorbed by the whole skin and clothing surfaces is zero for all users, because Tai is equal to Tmr. The sum of exergy rate contained by the inhaled humid air is also zero (breath air in Figure 4), because room Tai and RHin are equal to outside conditions Tao and RHout. Warm/cool convective exergy rates absorbed by the whole skin and clothing surfaces is also zero, because Tai is equal to Tao, and even temperature of clothing (hereinafter Tcl) is higher than Tai. The main input exergy is presented by metabolic thermal exergy rate (inner part in Figure 4). This means that the thermal exergy rate of 4.35 W/m2 for mother, 28.86 W/m2 for teenager and 3.75 W/m2 for grandfather is generated by bio– chemical reactions inside the human body. It is (1) Table 4. Components of exergy input and output. Exergy input warm exergy generated by metabolism; warm/cool and wet/dry exergies of the inhaled humid air; warm and wet exergies of the liquid water generated in the core by metabolism; warm/cool and wet/dry exergies of the sum of liquid water generated in the shell by metabolism and dry air to let the liquid water disperse; warm/cool radiant exergy absorbed by the whole skin and clothing surfaces. Exergy outputs warm and wet exergy contained in the exhaled humid air; warm/cool and wet/dry exergy contained in resultant humid air containing the evaporated sweat; warm/cool radiant exergy discharged from the whole skin and clothing surfaces; warm/cool exergy transferred by convection from the whole skin and clothing surfaces into surrounding air. 112 mainly influenced by the metabolic rate and the differences between Tai, body core temperature (hereinafter Tcr) and skin temperature (hereinafter Tsk). The metabolic thermal exergy rate presents output exergy that has to be released into ambient environmental space to keep functioning of the body structure. The rate of warm exergy stored in the core and in the shell is 0 W/m2 for mother and grandfather and 0.213 W/m2 for teenager and presents a part of metabolic thermal exergy rate, which is influenced by the difference between Tai, Tcr and Tsk. The exergy rate of exhalation and evaporation of sweat is 0.37 W/m2 for mother, 1.89 W/m2 for teenager and 0.31 W/m2 for grandfather. It is influenced by the difference between Tcr and Tai, Tcl and Tai. Warm radiant exergy rate discharged from the whole skin and clothing surfaces emerges because of higher Tcl than Tai and presents 0.29 W/m2 for mother, 0.68 W/m2 for teenager and 0.28 W/m2 for grandfather. Exergy rates of 0.52 W/m2, 3.42 W/m2, 0.40 W/m2 are transferred by convection from the whole skin and clothing surfaces into surrounding air, mainly due to the difference between Tcl and Tai. The rate of exergy consumption valid for thermoregulation that presents the difference between the rate of input exergy, the rate of stored exergy and the rate of output exergy is 3.17 W/m2 for mother, 22.66 W/m2 for teenager and 2.76 W/m2 for grandfather. The largest human body exergy consumption rate among observed users has the teenager, mainly due to the highest activity level and metabolic thermal exergy rate. Results of previous studies (Isawa et al. 2003, Prek 2004; Shukuya 2009) proved that in thermally neutral conditions, lower human body exergy consumption rate appeared. Results of our study show that created conditions with conventional system are more comfortable for the mother and the grandfather due to lower human body exergy consumption rate valid for thermoregulation and PMV closer to 0 than for the teenager. Figure 5. Human body exergy balances for three virtual users of active space equipped with user–centred system. If the user–centred system was set up to Tai 22°C, Tmr 24°C and RHin 40% in the zone for mother, this could result in optimal human body exergy balance with 3.17 W/m2 consumption rate valid for thermoregulation, higher input rate by metabolic thermal exergy rate (4.38 W/m2) and warm radiation (0.03 W/m2), higher output rate by warm convection (0.57 W/m2) and warm radiation (0.32 W/m2) as well as lower evaporation and exhalation of sweat (0.35 W/m2) compared to the conditions created with conventional system. If the system was set up to Tai 22°C, Tmr 24°C and RHin 40% in the zone for grandfather, this could result in lower human body exergy consumption rate valid for thermoregulation (2.36 W/m2), as well as lower input exergy by metabolic thermal exergy rate (3.44 W/m2) and warm radiation (0.03 W/m2), higher output rate by warm convection (0.48 W/m2) and warm radiation (0.34 W/m2) as well as lower evaporation and exhalation of sweat (0.29 W/m2). Conditions with Tai 17°C, Tmr 19°C and RHin 40% in the zone for teenager could result in lower human body exergy consumption rate valid for thermoregulation (21.93 W/m2), higher input rate by metabolic thermal exergy rate (31.28 W/m2) and warm radiation (0.03 W/m2), higher output rate by warm convection (5.73 W/m2) and warm radiation (1.11 W/m2), lower storage exergy rate (0.16 W/m2) as well as higher evaporation and exhalation of sweat (2.38 W/m2). The conditions created with user–centred system result in optimal human body exergy balance with PMV 0 compared to the conditions created with conventional system. Since energy use was measured for the same space equipped with the designed and the conventional system in different periods, approximately the same conditions were selected for the systems’ comparison (equal set-point T, time period, Tao and Tai variate among systems ±0.5 K; 0.8% assumed error). The measured energy use for space heating was by 11% lower when using user–centred system compared to the conventional system. The energy use for 3.2 Human body exergy balance and individualized climate „Variatio delectat“ (Cicero). The more ergonomic it is, if it possible to change (Pheasant, 1992). Figure 5 shows the example of the whole human body exergy balances for thee virtual users of the active space equipped with user–centred system. The system can provide special zones for individual user of active space (Fig. 5). For example, for mother, teenager and grandfather individual comfort zones equal to thermal neutrality can be created (PMV = 0). Zones can be actively regulated by setting the user–centred system to such a combination of Tai, Tmr and RHin that would result in optimal human body exergy balance. Set up parameters can be defined according to user needs, demands and his/her activity level. 113 6 REFERENCES space cooling was by 73% lower for user–centred system (Table 5). Table 5. Results of energy use (Ccool, Cheat) for selected conditions Trange = set up temperature range. System Trange User– Conventional Reduction [°C] centred [%] HeatTai=23.3 °C Tai = 23.7°C 11 23–24 Tao= -2.60°C Tao = -2.60°C ing Cheat= Cheat = winter 2.630 MJ 2.950 MJ Cooling summer 24–25 Tai=24.83°C Tao=19.45°C Ccool = 1.116 MJ Tai = 24.30°C Tao = 19.88 °C Ccool = 4.068 MJ 73 4 CONCLUSIONS The main role of user–centred system is to create optimal conditions for various users and activities. These would result in an optimal human body exergy balance. If the system was set up to Tai 22°C, Tmr 24°C and RHin 40% in the zone for mother, this would result in neutral thermal comfort conditions (PMV = 0) and 3.17 W/m2 of human body exergy consumption rate. In the case of grandfather we could create neutral thermal comfort conditions (PMV = 0) in such way to set up the system to Tai 22°C, Tmr 24°C and RHin 40% and human body exergy consumption rate is 2.36 W/m2. In the case of teenager we could create neutral thermal comfort conditions (PMV = 0) to set up the system to Tai 17°C, Tmr 19°C and RHin 40% and human body exergy consumption rate is 21.93 W/m2. To maintain thermal comfort for all users and activities, it is important that exergy consumption and stored exergy are at optimal values with rational combination of exergy inputs and outputs. The presented analysis was carried out for three selected subjects with different demands and needs for thermal comfort. However, user–centred system is a flexible system. It is possible to create optimal microclimatic conditions for every individual user and activities. The system could be applied in residential or public buildings. 5 ACKNOWLEDGMENTS Research program Building Construction and Building Physics, UL FGG founded by the Ministry of Higher Education, Science and Technology, Republic of Slovenia, COST action C24 Analysis and design of innovative systems with LowEx for application in build environment, CosteXergy, TIGR Sustainable And Innovative Construction P13.1.1.2.03.0003, 3211-10-000465. 114 ANSI/ASHRAE 2004. Thermal environmental conditions for human occupancy, Standard 55 1-30. Atlanta. ASHRAE Handbook 2007. HVAC Applications, Health Care Facilities. Atlanta: ASHRAE. Asimow M. 1962. Fundamentals of engineering design: Introduction to design. 1st edition. New York: Prentice Hall Inc. Bresjanac M., and Rupnik M. 1999. Patofiziologija s temelji fiziologije. Druga izdaja. Ljubljana: Inštitut za patološko fiziologijo. Crowie A.P., Mackin R., and McCaig I.R. 1994. Oxford dictionary of English idioms. Oxford: Oxford University Press. Dovjak M. 2012a. Individualisation of personal space in hospital environment. Doctoral thesis, Nova Gorica: University of Nova Gorica. Dovjak M., Kukec A., Kristl Ž., Košir M., Bilban M., Shukuya M., and Krainer A. 2012b. Integral Control of Health Hazards in Hospital Environment, Indoor and Built Environment, doi:10.1177/1420326X12459867. Fanger P.O. 1970. Thermal Comfort. Copenhagen: Danish Technical Press. Hwang R.L., Lin T.P., Cheng M.J., and Chien J.H. 2007. Patient thermal comfort requirement for hospital environments in Taiwan, Building and Environment 42, No. 8, 2980– 2987. Hey E.N., and Katz G. 1970. The optimum thermal environment for naked babies. Arch Dis Child 45, No. 241, 328– 334. International Fitness Association, 2013. Gym Temperature and Noise Limits. Orlando: IFA. Isawa K., Komizo T., and Shukuya M. 2003. The relationship between human–body exergy consumption rate and a combination of indoor air temperature and mean radiant temperature, Transactions of Architectural Institute of Japan 570, 29-35. ISO 2005. Ergonomics of the thermal environment–Analytical determination and interpretation of thermal comfort using calculation of the PMV and PPD indices and local thermal comfort criteria, Standard 7730 1- 52. Geneva. Košir M. 2008. Integrated regulating system of internal environment of the basis of fuzzy logic use. Doctoral thesis. Ljubljana: University of Ljubljana. Maslow A.H. 1943. A Theory of Human Motivation, Psychological Review 50, No. 4, 370-96. Musek J., and Pečjak V. 1997. Psychology. Ljubljana: Educy. Parsons K.C. 2002. The effects of gender, acclimation state, the opportunity to adjust clothing and physical disability on requirements for thermal comfort. Energy and Buildings 34, No. 6, 593-599. Peyton C. 1999. Sunlight could perk up kids’ grades, store proits, Sacramento Bee: Bee. Pheasant S. 1991. Ergonomics, work and health. London: MacMillan Press. Preamble to the Constitution of the World Health Organization as adopted by the International Health Conference, New York, 19-22 June, 1946; signed on 22 July 1946 by and entered into force on 7 April 1948. New York: World Health Organisation. Prek M. 2004. Exergy analysis of thermal comfort. International Journal of Exergy 1, No.2, 303-315. Regulation EU 305/2011. Regulation EU 305/2011 of the European Parliament and of the Council of 9 March 2011 Laying Down Harmonized Conditions for the Marketing of Construction Products and Repealing Council Directive 89/106/EEC. Brussels: European Commission. Rules on the ventilation and air-conditioning of buildings. O.J. RS, No. 42/2002. Rohles F. H., and Johnson M. A. 1972. Thermal comfort in the elderly, ASHRAE Transactions 78, 131-137. Shukuya M. 2009. Exergy concept and its application to the built environment, Building and Environment 44, No.7, 1545-1550. Shukuya M., Saito M., Isawa K., Iwamatsu T., and Asada H. 2010. Human body exergy balance and thermal comfort. Working Report of IEA ECBS Annex 49. Low exergy systems for high–performance buildings and communities. Stuttgart: Fraunhofer Verlag. Trobec–Lah M. 2003. Harmonization of thermal and daylight fluxes with fuzzy logic. Doctoral thesis. Ljubljana: University of Ljubljana. 115 116 Evaluation of the applicability of the quasi-steady-state overheating indicator for wooden buildings K. Goethals, L. Smet & A. Janssens Ghent University, Department of Architecture and Urban Planning, Ghent, Belgium ABSTRACT: EN 13790 suggests some quasi-steady-state overheating indicator to evaluate the summer comfort in buildings. However, only few countries included this method in the rules primarily because the method still needs major improvement. A group of Belgian architects and builders, for instance, indicated that the overheating indicator overestimates the overheating risk in lightweight structures like wooden buildings and asked Ghent University to back up this assertion and, if necessary, to suggest improvements. This paper discusses the first step in the study: an evaluation of the applicability of the overheating indicator for lightweight structures and an analysis of the effect of several building characteristics and the climate on the calculation parameters of the quasi-steady-state method. To this end the authors performed quasi-steady-state calculations and multi-zone energy simulations with TRNSYS. The results revealed that the overheating indicator overestimates the overheating risk in, especially well-insulated and airtight, lightweight structures and that insulation level, thermal capacitance and climate influence the calculation parameters of the method the most. 1. INTRODUCTION rameters. Unfortunately, as only recently summer comfort is a growing point of attention, the method is not as advanced as the one for heating. A group of Belgian architects and builders, for instance, indicated that the quasi-steady-state overheating indicator overestimates the overheating risk in lightweight structures like wooden buildings. They asked Ghent University to back up this assertion and, if necessary, to suggest improvements. This paper discusses the first step in the study: an evaluation of the applicability of the overheating indicator for lightweight structures and an analysis of the effect of several building/system characteristics on the calculation parameters of the quasi-steady-state method. To this end the authors performed quasi-steady-state calculations and multi-zone energy simulations (TRNSYS (Solar Energy Laboratory 2006)) on a simple reference building. When building new structures or renovating old ones, designers try first and foremost to reduce the heating demand. For, currently space heating and other heat uses dominate the final energy demand in most building types (Directorate-General for Energy and Transport 2007). Designers usually provide a well-insulated and very airtight building skin in combination with minimum ventilation. Yet, at the same time these measures lead to a higher overheating risk. Certainly in wooden structures, which generally have less thermal capacitance, the indoor temperature can reach rapidly high levels. And this while wooden buildings grow more popular. It is only too obvious that both designers and lawmakers have a role to play. Designers need to tell which (combinations of) measures lead to acceptable summer comfort and lawmakers need to lay down clear comfort requirements. Coming up with general quantitative rules is difficult because summer comfort depends on a large number of parameters. However, a simple simulation tool to evaluate the summer comfort would be of great help to both parties. In European countries like Belgium, Germany and the Netherlands designers as well as ‘energy reporters’ already make use of the so-called overheating indicator. The way to determine this indicator is described in EN 13790 (CEN 2007) and is based on the quasi-steady-state methodology to calculate the heating/cooling demand in buildings. Such a quasisteady-state approach finds favour with many parties, including the co-workers of the ASIEPI-project (Van Orshoven and Alvarez 2009), because of its transparency, robustness, reproducibility, adequate (balanced) accuracy and limited number of input pa- 2. QUASI-STEADY-STATE CALCULATION METHOD OF THE OVERHEATING RISK 2.1 Origin The quasi-steady-state calculation method is essentially a correlation method. It calculates the heat balance of a building over a sufficiently long time (e.g. a month, a season) and takes the dynamic effects into account with gain and/or loss utilization factors that are distilled from many complex energy simulation runs. This principle goes back to the 80’s of the previous century, in which various types of correlation were developed (e.g. the LCR method developed by the Los Alamos Scientific Laboratory (Los Alamos Scientific Laboratory 1983), the method 117 In 2000 Rouvel and Kolmetz (Deutscher, Elsberger et al. 2000) suggested to include an overheating analysis in the then German regulation (DIN 1995). For, the code only enforced minimum requirements on maximum glazing percentages to limit the overheating risk. The method they proposed was strongly inspired by the space heating calculation in EN 832 (DIN 1995), which was an essential part of the German regulation. The input and calculation parameters were also related to space heating for practical reasons. The methods defined the overheating indicator as the sum of the monthly normalized excess heat gains, which were function of the unutilized heat gains and the overall heat transfer coefficients for transmission and ventilation (Equation (3)). The linear relationship between this overheating indicator and the temperature excess hours using a fixed threshold temperature showed its suitability. As consequence, to set a maximum allowable value of the overheating indicator, Rouvel and Kolmetz performed detailed energy simulations of an – at least to the authors – unknown set of buildings (zones). An overheating indicator of 11000Kh/year corresponded to temperatures higher than 26°C for 10% of the occupation time. proposed in the framework of the International Standard Organisation (ISO 1986)). Co-workers of the PASSYS I project (CSTB 1989) who looked for a European correlation-based design tool, found that the then available methods were, generally speaking, rather similar. Most methods predicted the monthly performance and made use of two correlated parameters. The first parameter led to the useful gains or directly to the heating needs (utilization factor, solar saving fraction…). The second one was a ratio of gains (solar or total) to losses (SLR, GLR, X…). The correlation between these two parameters generally depended on a third parameter which expressed the possibility of storage of the solar energy (mass of the building). The co-workers of PASSYS II (CSTB 1993) decided to work out the Th-BV method of the Centre Scientifique et Technique du Bâtiment (CSTB 1988). To determine the monthly heating demand QH (MJ) they put forward Equation (1), which was actually a reduced equation. Originally a term -(1-η)QH,gn (to account for the effect of unutilized heat gains) was part of the equation and QL was multiplied by a factor dθr (to account for increased heat losses due temperature rise). However, these two effects were assumed to compensate each other and, thus, they were left out of Equation (1). QH = QH ,ls − η ⋅ QH , gn 12 QNN = ∑ (1) i =1 where QH = heating demand (MJ); QH,ls = transmission and ventilation heat losses (MJ); QH,gn = internal and solar heat gains (MJ); ηH,gn = dimensionless gain utilization factor (-). The two main parameters influencing the utilization factor η were the heat balance ratio γ (-) and the time constant τ (h). The heat balance ratio is the ratio of the heat gains to the losses; the time constant the ratio of the energy stored in the zone when the interior undergoes a 1°C temperature rise, to the specific transmission and ventilation heat loss. The form of the function f(γ,τ) which gave η was derived from the Th-BV formula (Equation (2)). The numerical parameters of the equation were derived by means of a minimization routine performed on detailed energy simulations (ESP (Maver and Clarke 1984)). The coworkers of PASSYS II found that a = 1 + τ/16 led to an uncertainty less than 4% on the heating demand for a wide range of buildings (10 ≤ τ < 100h). ( )( η H , γn = 1 − γ a 1 − γ a +1 if ) wiτη a = c + d ⋅τ g = 1 then h H , gn = a (a + 1) (1 − η (H H , gn tr , adj )⋅ Q H , gn + H ve,adj ) (3) where QNN = overheating indicator (Kh); ηH,gn = dimensionless gain utilization factor (-); QH,gn = internal and solar heat gains (Wh); Htr,adj = overall heat transfer coefficient for transmission, adjusted for the indoor-outdoor temperature difference (W/K); Hve,adj = overall heat transfer coefficient for ventilation, adjusted for the indooroutdoor temperature difference (W/K). (2) Figure 1. Temperature excess hours TE(>26°C) as a function of the overheating indicator QNN (Deutscher, Elsberger et al. 2000). where η = dimensionless gain utilization factor (-); γ = heat balance ratio (-); τ = time constant (h); a, c, d = numerical parameters (-). The method was later extended to cooling and, sometimes after a small redraft, included in standards like EN 13790, the standard which describes how the European countries should calculate the energy use for space heating and cooling in buildings. Also the overheating indicator was later included in EN 13790. However, unlike the quasi-steady-state heating/cooling demand calculation it was only adopted in a few countries (like Belgium, Germany and the Netherlands). 118 Corrado and Fabrizio (2007) wanted to evaluate the quasi-steady-state cooling demand calculation for Italian buildings. To this end they derived for five buildings (two test cells, a single-family house, a multi-family house and an office building) the utilization factor and the loss/gain ratio from dynamic simulations (EnergyPlus (University of Illinois and Lawrence Berkeley National Laboratory 2009)) and set the derived relations between these two parameters against the results of the quasi-steady-state calculation. This comparison revealed that the function of the utilization factor fixed in EN 13790 (i.e. Equation (2)) only matched the dynamic simulation results when numerical parameter a in the function would also account somehow for the window area. Corrado and Fabrizio performed multiple linear regression to relate the numerical parameter a to the time constant, the internal heat capacity per unit volume, the ratio between the window area and floor area and/or the ratio between the window area and the envelope area. They found that when a is function of the time constant and the ratio of the window area to the floor area R squared is the highest (Equation (5)). Obviously this holds only for Italian buildings. 2.2 Suggested adaptations EN 13790 fixed for the greater part the equations and the numerical parameters of the quasi-steadystate calculation method, but left room for modifications at national/regional level. Some adjusted the numerical parameters to account for the local climate, the typical building practice and/or the peculiar inhabitants. Others went one step further: they included more parameters in the function of the utilization factor. This section discusses some of the most recent suggestions. Kalema, Pylsy et al. (2006) investigated, among other things, the applicability of the quasi-steadystate heating demand calculations for modern Finnish buildings. They derived for a single-family house and an apartment the relation between the utilization factor and the gain/loss ratio from dynamic simulations (several programs including TASE (Aittomäki and Kalema 1996) and VIP+ (StruSoft 2002)) and compared it with the one obtained with the quasisteady-state calculation. The buildings had four thermal capacitance levels. The researchers in question found that EN 13790 predicted with reasonable accuracy the utilization factor for most of the studied buildings. Only for the lightweight structures (without any massive surface) EN 13790 underestimated the utilization factor (which would lead to an overestimation of the heating demand). If such lightweight structures would be built more often, a revision of EN 13790 would be imperative. Jokisalo and Kurnitski (2007) redid the study of Kalema, Pylsy et al., using other simulation software and looking at other buildings. They determined with the dynamic simulation software IDA-ICE (EQUA Simulation AB 2002) and the EN 13790 quasi-steady-state method the heating demand of an apartment, a detached house and a five-storey office building. The buildings had two thermal insulation levels, two internal heat gain levels, three thermal capacitance levels and solar shading or not. The results showed that the quasi-steady-state calculation overestimated the heating demand especially for the lightweight structures. Jokisalo and Kurnitski took the view that the numerical parameters in the function of the utilization factor could be tweaked for residential buildings. They arrived at Equation (4) for numerical parameter a. The quasi-steady-state method would never be able to account for the difference in heat gains between day and night in nonresidential buildings. The authors of underlying paper, however, believe this is not true. For, the heat load profile in dwellings is generally the negative of the one in non-residential buildings. a = 6 +τ 7 a = 8.1 − 13 ⋅ ξ + τ 17 (5) where a = numerical parameter (-); ξ = ratio of window area to floor area (-); τ = time constant (h). Orosa and Oliveira (2010) tried to include the effect of the permeability of surface coverings in the utilization factor function, just like Corrado and Fabrizio did for the window area to the floor area. For, studies showed that permeable coverings improve the thermal comfort and lower the energy use. The fact is, they dampen the indoor relative humidity variations. To this purpose Orosa and Oliveira first measured the heat losses, the heat gains and the heating/cooling demand of 25 Spanish office buildings and then derived for these buildings new utilization factor functions. They included in the equation of the numerical parameter a a new constant which accounts for the permeability of coverings (Equation (6)) and suggested a new equation for the utilization factor (Equation (7) for heating). However, the authors of underlying paper found it strange that the utilization factor did not include the heat balance ratio and the time constant, only the new constant which accounts for the permeability of coverings. Next to it, the effort may not be worth it: the effect of internal coverings on the thermal comfort and the heating demand is without doubt smaller than the impact of the building envelope. (4) where a = numerical parameter (-); τ = time constant (h). a = a0 + τ 70 (6) η H = 0.2267 ⋅ ln(a0 ) + 0.6373 (7) where a, a0 = numerical parameters (-); τ = time constant (h); ηH = utilization factor for heating (-). 119 3. MATERIALS AND METHODS internal heat gains (W) 700 3.1 Simulation model The model resembled the monozone reference building of PASSYS I. Figure 2 shows that it concerned a floating beam-shaped volume. This implied that all the surfaces were adjacent to the outside climate, which simplified the calculations. The principal and secondary façades had multiple glazed openings while the side walls had none. The walls as well as the floor and the ceiling consisted of a sequence of concrete, mineral wool and again concrete. Further, a constant leakage flow rate was assumed, even though the rate of in/exfiltration generally depends on the porosity of the building envelope and the magnitude of the natural driving forces of wind and temperature difference. The model comprised one or more systems. A convective heating and/or cooling system, with unlimited capacity, controlled the indoor operative temperature. No additional mechanical ventilation system was foreseen. 600 500 400 300 200 100 0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324 time (h) Figure 3. Hourly profile of sensible heat. 3.2 Results processing The procedure to derive the correlation parameters from detailed energy simulations was based on the one developed by the co-workers of the PASSYS project. For, TRNSYS does not output all terms directly, just like ESP that was used in the PASSYS projects. Three simulation runs were required to derive the relevant terms. Table 1 explains the simulations and gives the respective basic equations. Table 2 shows how the relevant terms were derived from the simulation results. (a) Table 1. Required simulation runs and their basic equations. 1 Normal simulation: heating keeps the inside temperature equal to at least the set point temperature for heating QH 1 = QH = QH ,ls − η ⋅ QH , gn 2 Simulation without overheating: heating and cooling keep the inside temperature equal to the set point temperature for heating (b) QH 2 − QC 2 = QH ,ls − QH , gn 3 Simulation without losses: heating and cooling keep the inside temperature equal to the external temperature QH 3 − QC 3 = −QH , gn where QH,i = heating demand (MJ); QC,i = cooling demand (MJ); QH,ls = transmission and ventilation heat losses (MJ); QH,gn = internal and solar heat gains (MJ). (c) Figure 2. (a) Floor plan, (b) principal façade and (c) secondary façade of the first PASSYS reference building (CSTB 1989). Table 2. Equations to derive the relevant terms. QH = QH 1 The building was supposedly inhabited. An hourly profile of sensible heat (of which 70% was convective) imitated the occupancy, lighting and appliances. The total internal heat gains matched those used in the Flemish quasi-steady-state calculation (Figure 3). The profile was based on earlier work done as part of the EPIcool-project (Saelens, Leenknegt et al. 2011). QH , gn = QC 3 QH ,ls = QH 2 − QC 2 + QH , gn η = (QH ,ls − QH ) QH , gn γ = QH , gn QH ,ls where QH,i = heating demand (MJ); QC,i = cooling demand (MJ); QH,ls = transmission and ventilation heat losses (MJ); QH,gn = internal and solar heat gains (MJ); η = dimensionless gain utilization factor (-); γ = heat balance ratio (-) 120 a clear linear correlation, but this was not the case: the graph had some noise. However, once the data was subdivided according to thermal capacitance, four distinct linear regressions lines could be determined, as shown in Figure 4. This meant that the equations for the determination of the utilization factor did not correctly account for the thermal capacitance. More specifically, a lightweight structure which had in reality the same thermal comfort level as a heavy structure, would have a higher overheating indicator value. 3.3 Evaluation methods The authors intended this study as a preliminary exploration of the applicability of the overheating indicator for lightweight structures. First they set the overheating indicator against TRNSYS to see if the overheating indicator correctly accounted for thermal capacitance. Secondly, they identified the factors that influence the dimensionless gain utilization factor the most so that later the function of the utilization factor could be optimized. They regarded a simple screening design sufficient. They performed a one-at-a-time analysis, which evaluated the impact of each factor in turn. Table 3 lists the parameters and their respective values. The parameters values that acted as a reference, are underlined. To avoid incorrect user input and to speed up the analyses, the authors developed a fully-automated VBA code which built, executed and processed the TRNSYS simulations as well as the steady-state calculations. Table 3. Parameter values. Parameter 1 Insulation level U (W.m-2.K-1) 0.15 Thermal capacitance C (MJ.K-1) Glazing area Agl (m2) Main orientation ori (-) Climate clm (-) 2500 TE(>26°C) (h) 2000 1500 1000 500 2 3 4 51 (0.217V) 0 0 0.31 7 (0.027V) 16 (0.067V) 28 (0.117V) (12+15) (6+10) (3+5) south west Nancy Uccle east Copenhagen 10000 20000 30000 40000 Ioverh (Kh) light moderately heavy medium heavy heavy Figure 4. Temperature excess hours as a function of the overheating indicator (for interpretation of the references to colour, the reader is referred to the digital version of the article). To further exemplify this shortcoming, the authors first derived (with the procedure described in section 3.2) the correlations parameters, more specifically the utilization factor and the heat balance ratio, from TRNSYS simulations for the same set of parameter combinations as above. And, then, they compared the obtained relations between the two correlation parameters with the results of the quasisteady-state calculation, as shown in Figure 5 (min/max QSS corresponds to minimum and maximum values obtained with the quasi-steady-state calculation: different heat loss levels in the parameter combinations). The TRNSYS data shows a higher utilization factor, especially when the heat balance ratio is low, and this is more pronounced in case of lightweight structures. This means that the overheating indicator overestimates the overheating risk in especially well-insulated and airtight lightweight structures, such as wooden buildings – as shown above. 4. RESULTS 4.1 Evaluation of applicability of overheating indicator for lightweight structures The authors evaluated the applicability of the Flemish overheating indicator, designed for dwellings, for lightweight structures (the Flemish region is a part of Belgium). This version is to a large extent similar to the one in EN 13790. The equations for the determination of the heat gains, the heat losses and the utilization factor are very much alike and almost only numerical parameters which determine the gains and losses (not the utilization factor) differ. Only the default values of the effective thermal capacitance, which determines the time constant, are different: four, different classes including light, moderately heavy, medium heavy and heavy (categories 1 to 4 in Table 3). The authors determined the overheating indicator and the temperature excess hours for all the combinations of the first four parameters in Table 3 (the climate corresponded to the one used in the quasi-steady-state) and set, just like Rouvel and Kolmetz once did to show the applicability of the overheating indicator, the results against each other in a graph. It was assumed that the data would show 121 001 001 η (-) η (-) 001 000 000 000 0 1 2 3 4 5 01 01 01 01 01 01 00 00 00 00 00 000 001 QH,gn/QH,ls (-) light moderately heavy medium heavy heavy min/max QSS TRNSYS 002 003 004 005 QH,gn/QH,ls (-) U=0.15W.m-2.K-1 U = 0.31 W.m-2.K-1 Figure 6. Utilization factor as a function of the heat balance ratio for two insulation levels of the reference case. Figure 5. Utilization factor as a function of the heat balance ratio (for interpretation of the references to colour, the reader is referred to the digital version of the article). η (-) 4.2 Analysis of effect of building characteristics and climate on calculation parameters The authors continued to work on the Flemish overheating indicator, designed for dwellings, to identify the factors that influence the utilization factor the most. They plotted the relations between the utilization factor and the heat balance ratio for variations on the reference case wherein only one parameter was altered (Figure 6 to Figure 10). Variations of other reference cases revealed similar trends and, thus, are not included in the paper. The first thing that struck the authors was that both the insulation level and the thermal capacitance had a significant impact on the relation. This was expected since the quasi-steady-state calculation method used both parameters to define the time constant of a building/zone. Also the climate was an influential parameter. Each climate led to specific gain and loss profiles, which affected the relation utilization factor-heat balance ratio. The authors believed that for that reason it might be worth the effort to adapt the numerical parameters to the local climate. The glazing percentage and the orientation did not influence the relation. Yet, the authors believed that the setup of the PASSYS reference building caused this trend: the principal and secondary façades had both reasonable glazing areas. A case with only one façade with glass would probably have resulted in a noticeable sensitivity to the glazing percentage and the orientation. 01 01 01 01 01 01 00 00 00 00 00 000 001 002 003 004 005 QH,gn/QH,ls (-) 7 MJ.K-1 16 MJ.K-1 28 MJ.K-1 51 MJ.K-1 Figure 7. Utilization factor as a function of the heat balance ratio for four thermal capacitance levels of the reference case. 01 η (-) 01 01 00 00 00 000 001 002 003 004 005 QH,gn/QH,ls (-) Agl = (3+5) m2 Agl = (6+10) m2 Agl = (12+15) m2 Figure 8. Utilization factor as a function of the heat balance ratio for three glass percentages of the reference case. 122 η (-) 6. ACKNOWLEDGEMENTS 01 01 01 01 01 01 00 00 00 00 00 This work was established as part of the VIS project “DO-IT houtbouw” supported by the Institute for the Promotion of Innovations through Science and Technology in Flanders. 7. 000 001 002 003 004 Aittomäki A., and Kalema T. 1996. TASE - A computer program for energy analysis of buildings. Espoo. CEN 2007. EN 13790: energy performance of buildings - Calculation of energy use for space heating and cooling. Brussels. Corrado V., and Fabrizio E. 2007. "Assessment of building cooling energy need through a quasi-steady state model: Simplified correlation for gain-loss mismatch." Energy and Buildings 39(5): 569-579. CSTB 1988. Règles Th-BV 1988, Calcul du coefficient de besoins de chauffage des logements. Cahiers due CSTB. CSTB 1989. The PASSYS project phase 1 - Subgroup simplified design tools. CSTB 1993. The PASSYS project - Development of simplified design tools. Deutscher P., Elsberger M., et al. 2000. "Sommerlicher Wärmeschutz - Eine einheitlicht Methodik für die Anforderungen and den winterlichten und sommerlichen Wärmeschutz, Teil 2." Bauphysik 22(3): 178-184. DIN 1995. DIN EN 832: Wärmetechnisches Verhalten von Gebäuden - Berechnung des Heizenergiebedarfs Wohngebäude. Directorate-General for Energy and Transport 2007. European energy and transport trends to 2030 - Update 2007, Office for Official Publications of the European Communities: 156. EQUA Simulation AB 2002. IDA-ICE 3.0. Solna. ISO (1986). ISO TC 163: Thermal insulation - Calculation of space heating requirements - Residential buildings. Jokisalo J. and Kurnitski J. 2007. "Performance of EN ISO 13790 utilisation factor heat demand calculation method in a cold climate." Energy and Buildings 39(2): 236-247. Kalema T., Pylsy P., et al. 2006. Nordic thermal mass - Effect on energy and indoor climate (report 184). Tampere. Los Alamos Scientific Laboratory 1983. Performance analysis of passive solar heated buildings by the solar load ratio method. Design handbook, American Solar Energy Society. Maver T. W., and Clarke J. A. 1984. Major extensions to the ESP system. Science and Engineering Research Council. Glasgow. Orosa J. A., and Oliveira A. C. 2010. "Implementation of a method in EN ISO 13790 for calculating the utilisation factor taking into account different permeability levels of internal coverings." Energy and Buildings 42(5): 598-604. Saelens D., Leenknegt S., et al. 2011. EPICOOL - report 15: implementation adaptations energy performance regulation for cooling. Solar Energy Laboratory 2006. TRNSYS 16.01 - A transient system simulation program. Madison. StruSoft 2002. VIP+. Malmö. University of Illinois and Lawrence Berkeley National Laboratory 2009. EnergyPlus. Van Orshoven D., and Alvarez S. 2009. Summer comfort and cooling: calculation methods and requirements. Stimulating increased energy efficiency and better building ventilation. M. Papaglastra and P. Wouters. Brussels, INIVE EEIG: 573. 005 QH,gn/QH,ls (-) ori = east ori = south ori = west η (-) Figure 9. Utilization factor as a function of the heat balance ratio for three orientations of the reference case. 01 01 01 01 01 01 00 00 00 00 00 000 001 002 003 004 005 QH,gn/QH,ls (-) clm = copenhagen REFERENCES clm = uccle clm = nancy Figure 10. Utilization factor as a function of the heat balance ratio for three climates of the reference case. 5. CONCLUSIONS The first substudy revealed that the quasi-steadystate overheating indicator overestimates the overheating risk in, especially well-insulated and airtight, lightweight structures. The second substudy showed that in particular the time constant (based on the insulation level, the thermal capacitance) and the climate influence the calculation parameters of the quasi-steady-state method. This means that future studies should first evaluate the impact of a larger parameter set and then modify the numerical parameters and/or include more parameters in the utilization factor function. For example, new values for the numerical parameters c en d can account for the local climate. 123 124 Overheating – an unexpected side-effect of decreased heating demand T. Kisilewicz Cracow University of Technology, Division of Building and Building Physics, Kraków, Poland ABSTRACT: South facing windows with high thermal resistance are expected to be a source of net energy gains. A simple conclusion of this kind may encourage a designer to maximize window area in order to maximize free energy gains. But the results of such approach are very often frustrating, excessively glazed building is neither energy-saving nor comfortable. Difficult rational decision regarding south window area should be based on a building thermal characteristics to ensure the low demand on heating and avoid cooling load simultaneously. Maximization of the solar gains may contradict thermal comfort in south collecting spaces. Simplified but comprehensive enough tools are needed for solar building designer. The special measures to protect building against excessive gains should be taken in any case, because overheating became quite a common side-effect of low energy buildings. 1 INTRODUCTION 2 WINDOW AREA VS. HEATING DEMAND Design process for low energy buildings if often limited to the aspects of heating energy demand. According to the simplistic understanding of low energy architecture, based on photos rather than on detailed monitoring of existing buildings, large south-oriented windows supply a large amount of solar energy and thus minimize heating demand. But in fact a structure of the well insulated building’s energy balance is very sensitive to any design decision and may be reasonably predicted only by means of detailed simulation programe. Optimum glazing area considering only heating may easily contradict optimum cooling demand. Finally, in extreme cases the cooling load may even exceed the demand on heating. Computer simulation results, presented in this paper, prove that rational window sizing decision must be based on the technical characteristics of the building shell and climate conditions. Oversized windows and the large resulting energy gains, not accumulated in building thermal capacity and used immediately for heating, do not reduce energy demand but create unbearable conditions or big cooling load in living spaces (Cooper 1997, Pfafferott et al. 2007). What is of extreme importance, cooling demand is much more sensitive to window size and thermal capacity than heating needs (Kisilewicz 2007). Because of the complexity of dynamic heat flow in building shell every building is practically a unique system and therefore does not conform to simple formulas or simple stationary calculations. Awareness of this fact should result in the development of special tools and procedures for those who design low energy buildings. South facing windows with high thermal resistance are expected to be a source of net energy gains. A simple conclusion of this kind may encourage a designer to maximize window area in order to maximize free energy gains. But the results of such a strategy may be contrary to the optimistic expectations, Figure 1. E, kW·h/(m2·year) 27 25 I 23 II 21 III IV 19 V VI 17 VII 15 13 11 0 0,1 0,2 0,3 0,4 0,5 0,6 South glazing index Rw /f Figure 1. South zone. Heating demand versus thermal capacity and south window area – passive house insulation standard with 80% ventilation heat recovery, LE double glazing. Presented in Figure 1 data were calculated for the south-west zone of a building and they do not represent the whole building. It may be a storey of a single-family building or, after slight modifications; a repeatable unit of a multi-storey residential or office building. Although the dimensions of the entire floor 125 The curves displayed in Figure 1, joining separate data points obtained from the simulations have been approximated using 5-th order polynomial regression. It may be observed that well insulated south glazing may actually reduce space heating demand. But what is of the extreme importance, oversized windows increase again energy demand, even beyond the initial blind space level. It means that a simple and commonly used by designers rule of the thumb: “big window area of the south windows assures big savings of the conventional energy” is not true. Available big solar gains do not necessarily become conventional energy savings. There is always an optimum glazing index value for which space energy demand on heating would be minimized (Schnieders 2006). In case of the lightweight buildings optimum ratio should be equal to 0.12 and for the massive buildings 0.27. Further window area increase results only in rising demand on energy and higher investment costs. The heating demand curves and hence resulting conclusions would be completely different in case of the improved glazing, e.g. triple glazing with U value decreased to 0.627 W/(m2K) and total solar transmittance equal to 0.472, Figure 2. The given data are typical for a “passive window” glazing standard. Due to decreased transmittance of solar radiation and significantly reduced thermal losses the optimum window to floor ratio is within the range of 0.22 for extremely light structures and 0.56 in case of the very massive structures. Evidently lower values of the heating demand index E and corresponding much higher solar heating fractions (SHF) are achieved in this case. area of the simulated unit were 10 x 10 x 3 m, the main object of the simulation reported in this paper is the unit’s south-west part only (modeled as a separate thermal zone), with floor dimensions 5 x 5 m and a height of 3 m. It was assumed that:  variable high density internal layers are in good thermal contact with the space,  continuous heating and cooling were considered,  the minimum internal air temperature is set at 200C and the maximum at 250C,  the measure of the passive system’s thermal efficiency is the amount of purchased heating and cooling energy,  heating simulation period: 15.IX – 15.VI, four computational time-steps per hour,  ventilation rate: 1/2 air change per hour with highly efficient heat recovery (80%), no extra infiltration was assumed,  standard apartment occupancy heat gains are constant in time,  air or material humidity was not considered. Program EnergyPlus and meteorological data for Kraków were used for computer simulations. Sample results of the computer simulations presented above, were obtained assuming passive house insulation standard. Exterior wall insulation: 40 cm of Styrofoam, roof insulation: 45 cm of Styrofoam. Heating demand index E of the selected south-west blind zone (without windows) was 20 kWh/(m2year). South wall glazing consists of double pane with LE (low emission) coatings, xenon filled, glazing heat transfer coefficient Ug equal to 0.92 W/(m2K), solar transmittance 0.418 and total solar transmittance g = 0.516. Rw/f - called glazing index is the ratio of south window area to floor area. Thermal capacity is one of the most important factors, that influence actual heat demand of the space (Balaras 1996). To investigate this relation the seven thermal capacity variants have been investigated. Variant I corresponds with a space with no massive layers in the exterior walls and 2 cm thick plaster on the internal walls, variant II: 2 cm internal plaster all around, variant III: 10 cm concrete internal structure except floor. In variant IV there is 10 cm concrete all around the space, in variant V external wall’s concrete layer thickness is increased to 20 cm. Variant VI responds to 20 cm concrete layer all around the simulated space. The last variant VII is for a space with 20 cm thick concrete walls and an extra accumulating mass inside of the simulated zone. The high density internal materials are in direct thermal contact with the space. E, kW·h/(m2·year) 21 19 I 17 II 15 III IV 13 V 11 VI 9 VII 7 5 0 0,1 0,2 0,3 0,4 0,5 0,6 Rw /f Figure 2. South zone. Heating demand versus thermal capacity and south window area – passive house insulation standard with 80% ventilation heat recovery, LE triple glazing. Hence rational window sizing procedure should account for at least several independent variables, that characterize thermal performance of a building 126 shell and glazing. The main factors influencing passive solar system performance in specified climatic conditions are:  total heat transfer (transmission and ventilation losses),  solar gains,  internal heat gains,  thermal capacity of the building internal structure,  thermal resistance and total solar transmittance of glazing,  external climate conditions. Multi parameter procedure should be used to create a design tool that allows to combine optimum window area with thermal characteristics of building. Common design process is nowadays completely different. A decision regarding window size is usually based on aesthetic or lighting aspects only. According to the above comments it should follow the basic technical decisions regarding building structure, insulation level, ventilation intensity and glazing features (Keller et. al. 2006). Window sizing procedure should be obviously adjusted to the local climate conditions. Commonly used single criterion of minimum heating demand does not account for space protection against overheating. Therefore, it is necessary to introduce cooling demand index EC as a measure of energy that should be removed from the simulated space to keep internal air temperature not higher than +25oC, Figure 3. The sample results of the simulations, presented in Figure 3, were obtained for passive house insulation standard, efficient heat recovery and xenon filled triple glazing with LE coating. Total solar transmittance of this glazing g is equal to 0.472. EC values were calculated for period September to May only, summer month data were not included. It was assumed that ventilation intensity was kept at a constant level for the whole period of simulation. Thus ventilation was not used for a free cooling of the space, as it should happen in reality and the calculated values of EC should be considered as overestimated. Nevertheless they may explain basic relations between window area and cooling demand. High values of glazing index are connected with very intense space overheating. The absolute values of cooling index EC may be even a few times higher than heating demand. Even intensive ventilation would not remove completely excess energy and the cooling load would be still significant. The black arrows in Figure 3 indicate optimum Rw/f value for heating criterion. Optimized in this way south window area would minimize heating load of the zone but in the same time cooling load would be at least comparable, what is completely inconsistent with the idea of low energy housing. The passive standard building considered only in terms of heating load would not be passive anymore if the total load was taken into account. 3 WINDOW AREA VS. COOLING DEMAND All the discussed above issues concerned only demand on heating energy. Solar radiation, transmitted through oversized windows during a day and not stored efficiently for later use, can not disappear from the building space - as it could be concluded from the stationary heat balance equations with a limited efficiency of gain use coefficient. heating I 40 cooling 35 III 30 IV 25 E, EC, kWh/(m2 year) EC, kW·h/(m2·year) II V 25 VI 20 VII 15 10 5 20 15 10 5 0 0 0,09 0 0,1 0,2 0,3 0,4 0,5 0,17 0,26 0,35 0,41 0,46 0,56 0,6 Rw /f Rw /f Figure 3. South zone. Cooling demand versus thermal capacity and south window area – passive house insulation standard with 80% ventilation heat recovery, LE triple glazing. Figure 4. South zone. Massive building. Heating and cooling demand versus south glazing index – passive house insulation standard with 80% ventilation heat recovery, LE triple glazing. Especially in case of a very light and well insulated structure this problem becomes a major issue with serious consequences for internal thermal conditions. Overheating is nowadays a typical but still underestimated side-effect of a modern glass architecture. In Figure 4 a combined relation between heating and cooling demand versus glazing index was shown. It is the most efficient case of a passive solar system (VII): a massive building with very high thermal capacity. Growing step by step solar heat127 coatings, xenon filled, glazing heat transfer coefficient Ug equal to 0.795 W/(m2K), and total solar transmittance g = 0.33. E - south zone kW·h/(m2·year) ing fraction and decreased demand on conventional energy (black bars) is combined with quickly increasing cooling load (white bars). Suggested earlier optimum Rw/f value equal to 0.56 would result in a cooling load that is a few times higher than heating load (Figs 3 and 4). A conscious and smart compromise is necessary to achieve high solar heating fraction and in the same time to avoid severe space overheating. It was confirmed that the cooling load up to ca. 2.5 kWh/(m2·year) may be easily mitigated by intensive ventilation and effective shading (Kisilewicz 2009). Instead of Rw/f value equal to 0.56 even two times lower ( <0.3) value should be chosen to ensure the low demand on heating energy and the acceptable thermal conditions. Due to the relatively small inclination of the E-curve (Fig. 2) heating energy increase is not significant. In lightweight buildings such a compromise is even much more important and difficult to achieve. Window shading strategy would be of a special importance in this case (Kisilewicz 2009). 30 I 29 II 28 III 27 IV V 26 VI 25 24 23 22 21 20 0 0,1 0,2 0,3 Whole building Rw /f Figure 5. Internal air mixing model. South zone heating demand versus ratio of total south window area to the whole building floor area; passive house insulation standard with 80% ventilation heat recovery, spectral selective double glazing. 3.1 Internal heat exchange Excess energy gains in one zone of a building can be used to warm up the other zones of the building, especially those without south windows. Internal heat exchange by transmission through the internal walls and doors is already included in data given above. Non-insulated internal brickwork (12 cm) allows to transfer a fraction of collected energy to the colder part of the building. In this case total energy demand of the building is 10.8% lower than in case of the insulated internal walls. In case of the lightweight buildings, a lack of thermal capacity and decreased use of the solar gains would be combined with a thermal (acoustic) barrier between the zones, that would block internal heat exchange. In those circumstances, only a convective exchange of air would allow to transfer energy between overheated south zone and adjacent north zone. To check this possibility it was assumed that 2K air temperature difference between the south and north zones triggers internal air exchange. Natural or forced air exchange intensity depends on current conditions in the zones but may not exceed maximum 5 ach (air changes per hour) to sustain thermal and acoustic comfort conditions within the living or working spaces. It was checked also that a further substantial increase of air exchange intensity has practically no influence on the achieved results. Simulation results for internal air mixing model were shown in Figure 5 and 6. Horizontal axis is in both figures the same and displays whole building glazing index, i.e. ratio of the total south glazing area to the whole building floor area. In this batch of simulations spectral selective glazing was applied. Glazing consists of two panes with spectral selective Heating energy demand of the single south-west zone was in Figure 4 related to the whole building glazing index. Due to decreased solar transmittance of spectral selective glazing heating energy demand is significantly higher than in case of LE glazing. Optimum total south glazing index is within the range of 0.10 to 0.17. E - whole building, kW·h/(m2·year) 30 29 28 27 26 25 24 23 22 0 0,1 0,2 Whole building Rw /f 0,3 Figure 6. Internal air mixing model. Total building heating demand versus ratio of total south window area to the whole building floor area; passive house insulation standard with 80% ventilation heat recovery, spectral selective double glazing. The heating energy demand curves for the whole building with spectral selective glazing were shown in Figure 6. Whole building optimization process would lead to the significantly increased south window area. Optimum total south glazing index is within the 128 EC, kW·h/(m2·year) range of 0.18 to 0.28 of the whole building floor area (or 0.36 to 0.56 south-west zone glazing index). Increased (not more than 5 to 10 %) heating demand of a separate south zone (Fig. 5) would be counterbalanced by the reduced demand of the whole building (Fig. 6). However, optimum heating demand of the building is not very sensitive to south window area changes. The reverse statement is needed when space overheating is considered, Figure 7. Whole building optimization of the south glazing area would result in 6 to 8 kWh/(m2·year) cooling load in a south zone - black arrows in Figure 7, the left one for a lightweight and the right one for a massive structure. In this case sunny part of a building becomes a kind of a solar collector for the whole building but with unbearable thermal conditions inside or high demand on mechanical cooling. Local optimization of south oriented window area - white arrows in Figure 7 – leads to the limited overheating, that could be easily moderated by ventilation or efficient shading. 20 I 18 II 16 III 14 IV 12 V approach thermal conditions in a glazed zone are of little importance and in practice it would be used as a living space only occasionally. But if living or working aspects take priority, window sizing decision should be made as a wise compromise on heating and cooling demand of the considered zone only. Whatever is the adopted designing criterion it is clear by now that a window sizing procedure should be completely different than the present designing practice. Numerous technical features of a building and equipment influence its final demand on conventional energy. Optimum window area is closely related to energy demand, so it is too early to determine it at the initial stage of designing process. Fully rational decision would be possible after all the technical factors of energy balance are known (Kisilewicz 2008). It is easy to guess that this suggestion will be not quickly accepted by the architects. That is why it is so important to develop the efficient measures against overheating. But even the right decision regarding south window area should be followed by the standard protection measures against excess solar gains. 3.3 Overheating control VI 10 VII 8 In case of the south oriented windows efficient protection could be provided by e.g. balconies, roof overhang or openwork sun screens. Much more expensive and troublesome in use would be the external blinds. Internal blinds are not very efficient but they are widely used because of their simplicity and much lower price. Nowadays, overheating may be efficiently reduced by application of a smart glazing. A properly designed control schedule of solar transmission would enable the achievement of maximum solar gains when needed, and reduction of the gains to avoid overheating (Arasteh et al. 2006). Efficient smart glazing could be to a certain extent a chance to make up for damages done by the wrong designing decisions. All the considered above problems were connected with the excess solar gains transmitted by transparent part of building shell. Finally, it should be mentioned that overheating may also occur due to the heat waves conducted by the opaque part of a lightweight building shell and triggered by absorbed solar radiation (Henze 2007). In case of the large commercial or industrial buildings with hardly any windows and lightweight flat roof, solar waves and internal gains may be the basic sources of overheating. Solar radiation, intensively absorbed by dark coloured bituminous roof coating, combined with 6 4 2 0 0 0,1 0,2 0,3 0,4 0,5 0,6 Rw /f Figure 7. Internal air mixing model. South zone cooling load versus thermal capacity and zone glazing index; passive house insulation standard with 80% ventilation heat recovery, spectral selective double glazing. 3.2 Design scenario Due to its low transmittance of solar radiation spectral selective glazing may be considered as a kind of protection measure against overheating and is rather advised in a mild and very sunny climate (Oelhafen 2007). Maximum values of cooling index (Fig. 5) are much lower than shown before for LE triple glazing (Fig. 3). In case of a typical LE glazing with higher transmittance than of spectral selective glazing, overheating risk is much higher and reasonable reduction of glazing area even more important. Building designer has to consider various scenarios and make a deliberate decision. Glazed space may be treated as a solar collector only, i.e. source of renewable energy for the rest of the building. In such 129 diversified in respect of solar and thermal transmittance, so there could be no one standard solution. It is also not recommended to use any of the glazing indexes given above in this paper apart from the other thermal features of designed object and local climate conditions. high air temperature results in a high heat flow rate entering the building space. A dynamic approach to this problem proves that increased thermal resistance is not the only one and the most reasonable measure to prevent overheating underneath. Green roof with an additional massive substrate layer is efficiently reducing the risk of overheating due to its thermal capacity and time shift (delay) of the penetrating heat wave (Kisilewicz T. 2012). Space overheating in lightweight structures may be reduced by decreased amplitude of the internal heat flow rate (due to improved thermal resistance) and/or increased time lag, i.e. phase difference between heat flow rate on one side and surface temperature on the other side (due to increased thermal capacity). 4.5 New window sizing tools for a common practice are needed to enable designer on one hand to take into consideration comprehensive building characteristic and on the other to avoid time consuming numerical simulations and random search for optimal solutions. 4.4 Even a justified decision regarding south window area should be followed by design of protection measures against overheating (overhangs and blinds, smart glazing, increased thermal capacity). In case of a lightweight building shell not only its thermal resistance but also time lag of the transferred heat wave would vitally influence internal temperature amplitude. 4 SUMMARY 4.1. Every building shell is practically a unique dynamic system with a specific design and combination of thermal characteristics and climate data. Rational south window sizing procedure should account for several independent variables, that characterize thermal performance of building shell and the boundary conditions:  total heating demand  space thermal capacity  glazing thermal resistance and transmittance  and local climate. 5 REFERENCES Arasteh D., Goudey H., Huang J., Kohler Ch., and Mitchell R. 2006, Performance Criteria for Residential Zero Energy Windows, Lawrence Berkeley National Laboratory LBNL59190, California. Balaras C.A. 1996. The role of thermal mass on the cooling load of buildings. An overview of computational methods, Energy and Buildings 24: 1-10. Cooper K. 1997. Overheating as a Factor in House Design, CANMET Energy Technology Centre, Ontario. Henze G.P., Pfafferott J., Herkel S., and Felsmann C. 2007. Impact of adaptive comfort criteria and heat waves on optimal building mass control, Energy and Buildings 39: 221235. Keller B., and Magyari E. 2006. Ein Design-Tool für den Entwurf von Niedrigenergiebauten, Bauphysik 28, Heft 5: 321329. Kisilewicz T. 2007. Computer Simulation in Solar Architecture Design. Architectural Engineering and Design Management 2007 Volume 3: 106-123. Kisilewicz T. 2008. The Influence of Resisitive, Dynamic and Spectral Features of the Building Walls on the Thermal Balance of the Low Energy Buildings (in Polish). Kraków, Wydawnictwo PK, No. 364. Kisilewicz T. 2012. On attenuation of temperature oscillation by the enclosure of the building, Czasopismo Techniczne – Budownictwo; Kraków, Wyd. PK; 2-B/2012: 87-94. Oelhafen P. 2007. Optimized spectral transmittance of sun protection glasses, Solar Energy, 81: 1191–1195. Pfafferott J., Herkel S., Kalz D. E., and Zeuschner A. 2007. Comparison of low-energy office buildings in summer using different thermal comfort criteria, Energy and Buildings 39: 750-757. Schnieders J. 2006. Passive On - Work Package 1 - Review of Existing Low Energy and Passive Best Practice, Darmstadt, Passive House Institute. 4.2 Unfortunately, common design process is nowadays completely different. Decision regarding window size is often based on intuition or fashion only. Current fashion is encouraging building designers to maximize glazed area. Oversized windows results in increased heating demand and massive overheating, that could be much higher than the heating needs. 4.3 A conscious and rational compromise is necessary to achieve as high as possible solar heating fraction and in the same time to avoid or at least radically minimize overheating. Those two aspects should be considered simultaneously. 4.4 When a whole house optimization strategy is raised it is very difficult to avoid entirely south zone overheating. Even very intensive air mixing between the zones would not remove efficiently excess heat gains from collecting space. Discrepancy between thermal comfort in the south zone and whole building optimization should be considered on the base of intended space use. Open spaces and not insulated internal walls increase efficiency of solar energy use in the whole building scale. 4.5 It is not possible to use any longer historic glazing rates, e.g. lighting index, as a sole designing principle. Contemporary glazing units are extremely 130 Assessing thermal comfort conditions in transitional states Yu-chi Wu, Ulrich Pont, Matthias Schuß, Ardeshir Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: This paper presents the initial analyses of a thermal comfort assessment study regarding transitional thermal states. Thereby, multiple groups of participants moved through a number of spaces with different thermal conditions. The thermal sensation and comfort evaluations of the participants were assessed before transition, immediately after the spatial transition, and after a short period of adaptation. The main objective of the study was to explore the relationship between before-after temperature change and the resulting shift in participants' expressed thermal sensation and comfort. sensation as a consequence of moving through spaces with distinct thermal conditions. Toward this end, a set of experiments were conducted, which included a number of such transitional states (temperature down-steps and up-steps). The experiments were conducted in a laboratory setting (Department of Building Physics and Building Ecology, Vienna University of Technology, Austria). In this laboratory, two office-like cells (one heated and the other cooled) are located within a larger – unconditioned – space. During the experiments, participants were exposed to different transitional states including transition from outside to the unconditioned spaces and vice versa and from the unconditioned space to the heated (or cooled) cell and vice versa. Subsequent to each transition, participants' thermal sensation and thermal comfort perception were assessed. Note that our study did not address the experience of thermal pleasure in the course of transitional processes. The subjective expressions of thermal conditions were analyzed in the context of collected indoor environmental data during the experiments. The concurrent assessment of subjective and objective factors in the course of transitional states provides important insights with regard to comfort implications of temperature differences between indoor and outdoor spaces as well as between differentially conditioned indoor spaces. 1 INTRODUCTION Heating, ventilation, and air-conditioning technologies and systems are typically applied as means of adjusting indoor thermal environments toward desirable conditions for human occupancy. However, if a sudden change in temperature occurs, typical thermal comfort evaluation schemes (geared toward thermally adapted people) may not apply. People are frequently exposed to such transitional states, for example when they enter or exit buildings under circumstances involving significant indooroutdoor temperature differences. Subjective evaluations of the thermal conditions have been addressed in past research. For example, the thermal sensation responses immediately after a transition involving temperature increase were found to be close to the responses after adaptation, whereas the thermal sensation responses immediately after a temperature decrease dropped initially to return to a stable level after adaptation (de Dear et al. 1993, ISO 2005, ASHRAE 2004). Chun & Tamura (1998) emphasize the importance of temperature change for the perception of thermal comfort. They also suggest (Chun & Tamura 2005) that thermal comfort perception at a certain point in time is influenced by antecedent thermal conditions. Arens et al. (2006) and Zhang (2003) investigated thermal sensation and comfort in time series after rapid temperature changes using both thermal sensation and thermal comfort scales. Likewise, Chen et al. (2011) studied thermal sensation and comfort patterns after transitional states. In a recent paper, Parkinson et al. (2012) indicated that sudden changes in ambient temperature can induce thermal pleasure, given a positive alliesthesial effect. However, the same environmental step change invoked a displeasure response when the core temperature was stable. In this context, the specific objective of the present study is to investigate people's thermal 2 APPROACH Altogether, 313 students participated in the experiment in groups consisting of 12 individuals. At the outset of each session, the group was divided into two sub-groups who took two different paths throughout the experiment. The thermal resistance of the participants clothing, which remained unchanged throughout the experiment, was recorded at the 131 beginning of the experiment (around 0.6 ± 0.15 clo). Figure 1 schematically illustrates the experiment's spatial arrangement. Here, E denotes the external environment (open courtyard) and M is the general (unconditioned) laboratory space. A and B are two differentially conditioned cells. Thermal conditions in these spaces were monitored via sensors for air temperature, relative humidity, illumination, and CO2 concentration. Participants went through a number of transitions as summarized in Table 1. In our study, the adaptation phase lasted about 15 minutes. In literature, post-transition adaption phases of 10 to 20 minutes have been found appropriate (Arens et al. 2006, Zhang 2003, Nakano 2003). Before each transition, participants were adapted to thermal conditions. Immediately after transition, their thermal sensation and comfort vote was assessed via a questionnaire. After approximately 15 minutes, votes were collected again. The experiment was conducted in the beginning of May 2012 over a period of 5 days from 9 am to 6 pm. The outdoor temperature (E) range in this period was between 9 and 25ºC. The temperatures of the unconditioned space (M) fluctuated slightly around 24ºC. The temperatures of heated cell (A) and the cooled cell (B) were kept at 27ºC and 17ºC respectively. Participants expressed their thermal sensation (TSV) using a 7 points scale (-3: cold, -2: cool, -1: slightly cool, 0: neutral, 1: slightly warm, 2: warm, 3: hot) and their thermal comfort vote (TCV) using a 6 points scale (-3: very uncomfortable, -2: uncomfortable, -1: just uncomfortable, 1: just comfortable, 2: comfortable, 3: very comfortable). To compare participants' expressed thermal sensation votes (TSV) with steady-state model predictions, we calculated for all instances the PMV (Predicted Mean Vote) values using measured indoor environmental variables (e.g., air temperature, relative humidity) and known personal factors (clothing, activity). Figure 1. Schematic illustration of the test spaces. Figure 2 shows – for all indoor to indoor transitions cases (transitions 3 to 6 in Table 1) – the relationship between participants expressed TSV and calculated PMV after a thermal transition. Thereby, a distinction is made between TSV votes before and after adaptation (note that each dot represents the mean value of a sub-group of participants, typically consisting of six people). As it could be expected, calculated PMV values correlate better with participants' expressed TSV as compared to TCV. 4 3 R² = 0.81 Mean TSV 2 R² = 0.59 1 0 -1 -2 -3 -4 -4 3 RESULTS AND DISCUSSION -3 -2 -1 0 1 2 3 4 PMV Figure 2. Participants' thermal sensation vote (TSV) after transition versus calculated values of PMV. White dots, dashed line: before adaptation; dark dots, continuous line: and after adaption. To provide an initial impression of the results, we include in this paper a number of Figures, which mainly depict the implications of temperature increase (or decrease) in the course of a spatial transition for the participants' thermal sensation and comfort evaluations. Figure 3 shows changes in thermal sensation vote after indoor to indoor spatial transitions as a function of the temperature difference between the start room (θ1) and the end room (θ2). Thereby, ∆TSV denotes the difference between TSV values immediately after transition (before adaptation) and before transition. Hence, increase in TSV is positive, whereas decrease in TSV is negative. Note that in this Figure each dot represents the mean value of the votes of a group of participants who experienced the same transition. Moreover, this figure includes a Table 1. Overview of the spatial transitions (see Figure 1 for the room symbols) and the associated number of participants. Transition Number of participants 1 E_M 313 2 M_E 313 3 M_A 155 4 M_B 158 5 A_M 155 6 B_M 158 132 transition temperature was close to the neutrality temperature of 24˚C. To further explore this circumstance, we categorized the 626 individually experienced (indoor to indoor) transitions in terms of 6 categories (see Table 2) according to the position of start and end temperatures (θ1, θ2) with regard to an assumed neutrality temperature of 24 ˚C. linear correlation function for all data as well as separate linear regression lines for temperature increase and decrease cases. Figure 4 is similar to Figure 3, but it shows the relationship between temperature difference and the corresponding change in thermal comfort vote. These results suggest that, as compared to changes in TCV, changes in TSV display a stronger correlation with transition-related temperature differences. 3 2 6 R² = 0.86 4 0 TCV ΔTSV R² = 0.72 1 2 -2 M-R M-B R-M B-M R² = 0.35 -4 R² = 0.85 -6 -15 -10 -5 0 5 0 -1 10 θ2 -θ1 [K] -2 Figure 3. Change in thermal sensation (∆TSV) vote following a transition as a function of the temperature difference between the start room (θ1) and the end room (θ2) (120 groups). -3 -10 -5 0 5 10 θ2 -θn [K] 6 Figure 5. Mean thermal comfort vote (50 groups of participants) immediately after transition (M_A, M_B, A_M, B_M) as a function of temperature difference between posttransition temperature and an assumed neutrality temperature (θn) of 24˚C. 4 ΔTCV 2 0 R² = 0.56 R² = 0.39 -2 R² = 0.52 -4 -6 -15 -10 Table 2. Categorization of (indoor to indoor) transition categories experienced by participants according to the position of start and end temperatures (θ1, θ2) with regard to an assumed neutrality temperature of 24 ˚C. θ2 θ2-θ1 Number of Category θ1 [K] transition [˚C] [˚C] A < 24 < 24 >0 22 B > 24 > 24 <0 135 C < 24 < 24 <0 44 D > 24 > 24 >0 129 E < 24 > 24 >0 151 F > 24 < 24 <0 145 M-B M-R R-M B-M -5 0 5 10 θ2 -θ1 [K] Figure 4. Change in thermal sensation (∆TCV) vote following a transition as a function of the temperature difference between the start room (θ1) and the end room (θ2) (120 groups). Figure 5 shows the thermal comfort vote (50 groups of participants) immediately after indoor to indoor transition as a function of the difference between post-transition temperature (θ2) and the applicable value of the neutrality temperature (θn= 24˚C), calculated based on the mean monthly outdoor temperature (May 2012) at the laboratory location in Vienna using the Nicol and Roaf (1996) equation. Note that to generate this Figure the total number of 626 relevant transitions was subdivided into 50 sets comprising of participants who experienced similar temperature changes in the course of the transition. Thus, the results are aggregated in terms of fifty mean values (dots in the Figure). Each value represents the mean vote of a group. This figure suggests that the groups' mean thermal comfort votes were more positive, if the post- Figure 6 shows, for the six transition categories of Table 2, the changes in expressed TSV (black vector) and TCV (grey vector) together with the corresponding changes in temperature (dashed vector). As it could be expected, the information in this Figure implies that the direction of changes in TSV after a transition are consistent with the direction of changes in temperature. The direction of TCV change, however, may point in the opposite direction of temperature change (B, D): If, as a result of the transition, the distance to the neutrality temperature is increased, TCV change is negative. Otherwise, the TCV change is positive. 133 To further investigate this point, we can apply the concept of an effective temperature difference ∆θeff (Mahdavi 2012), which denotes the transition-related net thermal distance (expressed in terms of temperature difference) to neutrality temperature. It can be calculated as per the following equation: [K ] ∆θ eff = θ1 − θ n − θ 2 − θ n consistent with transition-related temperature differences. Moreover, TSV and corresponding PMV calculations display a certain level of congruence. As to TCV, our results imply an interesting relationship with the distance to thermal neutrality before and after transition: A high correlation could be found between the transitionrelated effective temperature difference (Mahdavi 2012) and the change in thermal comfort vote immediately after a transition. Hence, the net reduction (or increase) of distance to applicable neutrality temperature appears to provide an effective basis for the prediction of the change in people's expressed TCV immediately after a thermal transition circumstance. (1) Figure 7 shows the relationship between the change in thermal comfort vote due to a thermal transition as a function of the respective effective temperature difference. This rather strong relationship suggest that the short-term thermal comfort vote after a thermal transition could be reliably predicted based on start and end temperature values. 5 REFERENCES 3 32 Arens E., Zhang H., and Huizenga C. 2006. Partial- and Whole-body Thermal Sensation and Comfort, Part I: Uniform Environmental Conditions. Journal of Thermal Biology 31: 53-59. ASHRAE 2004. Thermal Environmental Conditions for Human Occupancy, Standard 55. Atlanta. Chen C., Hwang R., Chang S., and Lu Y. 2011. Effects of Temperature Steps on Human Skin Physiology and Thermal Sensation Response. Building and Environment 46: 23872397. Chun C., and Tamura A. 2005. Thermal comfort in urban transitional spaces. Building and Environment 40: 633-639. Chun C., and Tamura A. 1998. Thermal Environment and Human responses in Underground Shopping Malls vs Department Stores in Japan. Building and Environment 33: 151-158. de Dear R., Ring J. W., and Fanger P. O. 1993. Thermal Sensation Resulting from Sudden Ambient Temperature Changes. Indoor Air 3: 181-192. ISO 2005. Ergonomics of the Thermal Environment – Analytical Determination and Interpretation of Thermal Comfort Using Calculation of PMV and PPD Indicates and Local Comfort Criteria. ISO EN 7730-2005. Mahdavi A. 2012. Effective temperature difference as a variable of thermal transition states. Internal report. Nakano J. 2003. Evaluation of Thermal Comfort in Semioutdoor Environment. Ph. D. Thesis, Waseda University. Nicol F., and Roaf S. 1996. Pioneering New Indoor Temperature Standards: The Pakistan Project. Energy and Buildings 23: 169-174. Parkinson T., de Dear R., and Cândido C. M. 2012. Perception of Transient Thermal Environments: Pleasure and Alliesthesia. In: 7th Windsor Conference. Windsor, UK. Zhang Y., and Zhao R. 2009. Relationship between thermal sensation and comfort in non-uniform and dynamic environments. Building and Environment 44: 1386-1391. Zhang H. 2003. Human Thermal Sensation and Comfort in Transient and Non-uniform Thermal Environments. Ph. D. Thesis, CEDR, University of California, Berkeley. 1 0 24 TSV/TCV Air temperature 2 -1 -2 16 B A C Type -3 D E F Figure 6. Representation of temperature (dashed vector), TSV (black vector), and TCV (grey vector) change due to a transition aggregated for the six transition types of Table 2. 4 3 R² = 0.87 2 ΔTCV 1 0 -1 -2 -3 -4 -10 -8 -6 -4 -2 0 2 4 6 8 ∆θeff [K] Figure 7. Change in thermal comfort vote as a function of effective temperature difference ∆θeff [K]. 4 CONCLUSION We presented the results of a study pertaining to the subjective evaluations of thermal transitions between rooms of different temperatures. Similar to past research results (see de Dear et al. 1993), our study suggests that the general tendency of participants' thermal sensation (expressed in terms of TSV) was 134 Review of methods for evaluation of building energy enhancements B. Kelly, F. Judex, V. Vukovic Austrian Institute of Technology (AIT), Energy Department, Giefinggasse 2, 1210 Vienna, Austria R. Braun University of Applied Science, Schellingstraße 24, 70174 Stuttgart, Germany F. Dubisch Institut für angewandte Energiesimulation & Facility Management, 50829 Cologne, Germany ABSTRACT: Buildings are one of the biggest energy consumers in the European Union (EU). To improve energy efficiency requires changes in technology, policy and human behaviour. The energy management systems currently used involve the use of various tools and methodologies. The evaluation of the impact of applied measures and procedures on building energy efficiency is becoming more relevant. However the implementation of such measures can be time consuming and costly. The aim of this paper is to present a new model for the definition of the baseline and reporting period for the calculation of energy savings after the application of improved Building management control strategies. The final goal is to reduce the time needed to implement energy and cost saving methods in real world scenarios. 1 INTRODUCTION Another standard available to consider the long term implications of particular building enhancement measures on life-cycle energy cost assessment is EN15459 (CEN, 2007b). Such information can be used in conjunction with climate data corrections to account for variability in weather conditions between the building energy assessment periods stated in EN15603 (CEN, 2008a). Additionally, the user may be interested in investigating separate results for heating and cooling seasons, which are defined in EN13790 (CEN, 2008b). A major problem in the application of these methodologies is the creation of the baseline which should be used to quantify the improvements. Data and monitoring equipment are not always available and the building scenario may not easily fit into the methods documented. This may result in prohibitive costs or long delays for the application, hindering the progress towards improved building or process quality. This paper will describe such an example and demonstrate how to set up an energy assessment plan. It will look at the current standards and highlight and attempt to fill gaps in knowledge and recommend areas for future work. Through summarizing such a diversity of information available from various sources, the present review is expected to ease the frequent challenge faced by building practitioners in evaluating the effects of building energy performance enhancement measures. A variety of state-of-the-art standards in building construction (ASHRAE, 2011), design concepts (iPHA, 2013) and “green building” rating systems (US GBC, 2013) promote the use of renewable materials and innovative technologies to enhance building environmental performance. Typical sustainability indicators include water use, energy efficiency, indoor environmental quality, energy consumption, materials and resources, as well as post-occupancy restoration of construction sites and the preservation of natural habitats. However, no unique methodology is standardized for evaluating the impact of applied measures and procedures on the building energy efficiency. Rather, a range of procedures exist to partly address the complexity of the comprehensive evaluation of building energy efficiency enhancements. The often cited International Performance Measurement and Verification Protocol (IPMVP) and the ISO50001 standard prescribe a methodology for continuous improvement of energy management. The “Plan-do-check-act” scheme, promoted by the ISO 50001 standard (ISO, 2011), provides quality assurance processes but does not specify parameter definitions or equations for calculation of key energy performance indicators. Such equations and common variables are defined in previous European projects which also focused on the need for objective and consistent methodology to evaluate building energy efficiency enhancements (e.g. CONCERTO protocol). Key energy performance indicators considered in the equations, such as total and primary energy consumption, CO2 emissions and costs are defined by the governing standard EN15217 (CEN, 2007a). The main outcomes are: − To explore the diversity of common standards related to energy performance of buildings. 135 − To highlight the difficulties in calculating energy savings resulting from an implementation of different measures. − Review of monitoring methodologies and current and or finished projects. − Definition of project related standards and methodologies. − Development of an appropriate methodology for the definition of the baseline and the reporting period for the “Sounds for Energy Control of Buildings” (S4ECoB) project. 3.1 International Perfomance Measurement & Verification Protocol (IPMVP) The IPMVP prepared by the Efficiency Valuation Organization (EVO) (EVO, 2012) provides an overview of current best practice techniques available for verifying results of energy efficiency, water efficiency and renewable energy projects in commercial and industrial buildings. The IPMVP is published in two current and one archived Volume: − Volume I: Concepts and Options for Determining Energy and Water Savings. − Volume II: Indoor and Environmental Quality (IEQ) Issues (archived) − Volume III: Application This protocol is not a standard; therefore a formal compliance requirement does not exist. The main idea behind this protocol is to provide a common method for the documentation of achieved energy savings and to add transparency to energy saving reports. Volume I defines terminology and gives suggestions for good practice in documenting the effectiveness of efficiency projects that are implemented in buildings. Furthermore the preparation of “Measurement and Verification” (M&V) Plans. Figure 1 shows the M&V process defined in this protocol; the M&V Plan is the process of using measurement to reliably determining actual savings. 2 METHODOLOGY This work is based on the research conducted within the FP7 “Sounds for Energy Control of Buildings” (S4ECoB) project. Within this work more than fifty international standards related to energy efficiency in buildings, indoor comfort and energy management are analysed. Out of this analysis five standards were identified of interest for the S4ECoB project and are reviewed in more detail. In addition to this work three international projects and one international protocol (IPMVP) related to energy measurement are analysed. The measurement and verification method proposed in the IPMVP can be applied to any system. The presented equations for calculating the energy savings are general and should be individually adapted to the scope of the project. This is because the protocol is only a description of a methodology and not a detailed calculation method. The original basis for energy reduction calculation presented in the Information and Communication Technologies Policy Support Programme 2009 (ICT PSP, 2011) is a modified version of the IPMVP developed within the ICT PSP Work Programme 2009. In these projects it is shown how the IPMVP can be applied to a specific project. Overall it can be said that there is a high variety of monitoring approaches to assess the effectiveness of measures. Some methods are more general and are not dependent on the field of application e.g. ISO 50001. Figure 1. Measurement and verification process (IPMVP, 2012). The definition of the baseline energy consumption is one of the key aspects in the planning phase of the M&V process. The energy savings can be calculated with following equation: 3 REVIEWED DOCUMENTS Energy Savings = Base Year Energy Use – Post Retrofit Energy Use ± Adjustments To date various methods have been developed and introduced to measure the energy consumption of a building before and after Energy Conservation Measures (ECM) have been implemented. The following section presents a summary of the EVO, IPMVP, the ICT PSP Work Programme 2009, “ICT for Energy Efficiency in Social Housing” and the ISO 50001. In order to identify the energy savings the IPMVP provides four methods for determining the savings, which can either be applied for a whole building or only for a sub system such as a ventilation system. The baseline should include a full operation cycle and represent all operating modes of the system. The IMPVP provides following options for the calculation of the savings. 136 A. B. C. D. Retrofit Isolation: Key Parameter Measurement Retrofit Isolation: All Parameter Measurement Whole Facility Calibrated Simulation PSP option C, “Whole Facility” as defined in the IPMVP was certainly applicable. Option A, B and D was, in the opinion of the ICT PSP, not applicable and not used for evaluation of the achieved savings. The chosen option does not assume constant energy demand or that energy demand variation can be accurately modeled. The energy savings measurements are based on estimation of non-intervention consumption (baseline +/- adjustments). For the ICT PSP approach there are two different possible methods to calculate the energy savings: − Pre - post comparison − Applying a control group For option A, B and C, data for the baseline or reporting period must be available, option D allows the calculation of energy saving without measured data for the baseline. The "Adjustments" term in this general equation is used to re-state the use or demand of the baseline and reporting periods under a common set of conditions. This term distinguishes proper savings reports from a simple comparison of cost or usage before and after implementation of an Energy Conservation Measure (ECM). Simple comparisons of utility costs without such adjustments will report only cost changes and fail to report the true performance of a project. To properly convey “savings”, adjustments must account for the differences in conditions between the baseline and reporting periods. The baseline in an existing facility project is usually the performance of the facility or system prior to modification. This baseline physically exists and can be measured before changes are implemented. In new constructions, the baseline is usually hypothetical and defined based on code, regulation, common practice or documented performance of similar facilities. In either case, the baseline model must be capable of accommodating changes in operating parameters and conditions so “adjustments” can be made. Figure 2 shows the method to calculate the energy savings after the implementation has taken place. For the pre – post comparison the same buildings are used to calculated the savings. The main challenge by using the same building is to estimate the consumption savings which would have taken place had the intervention not been carried out. The identification of independent variables such as outside temperature and occupancy is an important step in this method. The identification of independent variables is a critical point for the selection of a baseline period. In some cases the best solution is for the base line period to cover a full year, or the baseline and reporting period cover the same months in different years. Figure 3 shows a comparison of the before and after analysis of one building. The advantage of this method is that the analysis can be carried out at the same object. It is important either to ensure the baseline and reporting period covers a full year, or to ensure baseline and reporting period are covering the same months in different years. Figure 3. Before and after analysis of one building as used in the ICT PSP approach (ICT PSP, 2011). For the pre – post comparison, prior consumption is used for the estimation of energy savings. Figure 4 shows the method by applying a control group. This is useful where no baseline energy consumption data is available. The challenge in this method is to find a second building which matches the characteristics of the experimental building. Figure 2. Measurement & Verification Process (IPMVP, 2012). 3.2 ICT PSP Work Programme 2009 “ICT for Energy Efficiency in Social Housing” In the ICT PSP Work Programme 2009, Objective 4.1 (ICT PSP, 2009) “ICT for energy efficiency in social housing”, three projects have been selected to develop a methodology for energy saving measurements in buildings. The original basis for the energy reduction calculation presented in this methodology is a modified version of the IPMVP protocol. In the context of ICT 137 energy use from heating, cooling and dehumidification, ventilation and humidification, hot water, lighting (optional for residential buildings) and other electrical appliances or industrial processes are calculated or measured. The standard explains what systems are included and excluded from these assessments. 3.6 EN217:2007 Figure 4. Control building design as used in the ICT PSP approach (ICT PSP, 2011 p. 24). This standard (CEN, 2007a) builds on the general rating calculated in 15603, expressing in a more specific manner to the actual building. It highlights how climate, building function, national energy policy, building size and shape as well as ventilation rates can affect the basic energy rating and gives methods for calculation of an energy performance indicator and validating this against reference buildings. The standard builds towards how this data should be documented and certified. 3.3 ISO 50001 The “ISO 50001:2011 energy management systems – requirements with guidance for use”, is a voluntary international standard giving organizations guidance on implementing energy management systems (EnMs). It is the first edition of the standard, published in June 2011 and applies to all sizes of organisations irrespective of geographical, cultural or social conditions. The standard is based on a Plan-DoCheck-Act (PDCA) management tool which can be outlined as follows: 4 USE CASE − Plan: conduct the energy review and establish the baseline, energy performance indicators (EnPIs), objectives, targets and action plans necessary to deliver results that will improve energy performance in accordance with the organization’s energy policy. − Do: implement the energy management action plans. − Check: monitor and measure processes and the key characteristics of operations that determine energy performance against the energy policy and objectives, and report the results. − Act: take the actions to continually improve energy performance and the EnMS Within the project S4ECoB a prototype for improving HVAC controls by integrating a low-cost novel network of audio sensors within the BEMS and the improvement of the strategies and algorithms of automation and conditioning is deployed. It will be calibrated and validated in real operational situation in order to ensure that energy savings and benefits justify the investment, providing new market solutions and supporting reduction of climate change. As the target group of the system to be developed is mostly large scale commercial buildings, an airport and two shopping malls have been selected as test sites. The sites are as follows: 4.1 Linate Airport, Milan 3.4 EN 15232:2012 Linate Airport is close to the city of Milan and is mainly used for domestic and short-haul international flights, its terminal is 75,000m2 and in 2009 it processed 8 million passengers. It is a business airport with connections to all main European cities. The overall airport surface is about 4 million m2, of which at least 100,000 m2 are built environment. The HVAC system in passenger boarding areas at the airport is controlled by a PID PLC controller that enables parametric control. It is supplied by hot water from a Combined Heat and Power (CHP) plant with an electrical output of 24MW and 84 MW thermal in summer and 72 MW in winter, making the airport completely independent from external power providers as well as supplying hot water to a local town nearby (Wärtsilä, 2007) (Sea Energia, 2011). This standard (CEN, 2012) allows a rating to be assigned for the level of control of individual elements of Building Automated Control Systems (BACS) , from the blinds on a window to the heating system pumps. In this way the level of Technical Building Management (TBM) can be assessed. Using the BACS factor method this information can be combined with data from the buildings HVAC systems to determine BACS efficiency factors. 3.5 EN 15603:2008 EN 15603:2008 (CEN, 2008a) describes two principle types of energy ratings for buildings applicable to new and existing buildings and details how they can be calculated or measured. A clear “system” boundary for measurement is defined and 138 chosen and the constraints from the project had to be found. 4.2 Principe Pio, Spain Principe Pio is a shopping centre located in the centre of Madrid, opened in 2004, connected to the historical building of the Principe Pio train station and to a transport interchanger. The building receives 12,000,000 visitors per annum and has a total area of 77,828 m2. The building is equipped with two independent HVAC systems for common areas one for the historical building and another for the new building. The BEMS controls the lighting, air conditioning production, heating and cooling units; AHU and fan coils. 5 S4ECOB APPROACH FOR THE ENERGY ASSESSMENT As previously mentioned the availability of precise data and monitoring equipment is not always available at the beginning of a project. Add to this that the baseline should also include a full operation cycle and represent all operating modes of the system and that research projects are nearly always limited in budget and time with the building not easily fitting into the methods described in past projects and standards and it is clear a new method needs to be developed. The main problems identified in the S4ECoB project are: 4.3 Maremagnum, Spain Maremagnum is a waterfront shopping centre located in the Port area of Barcelona. The building receives 13,000,000 visitors per annum. The mall was opened in 1994 and refurbished in 2005. The total area of the shopping centre is 22,890 m2. No air conditioning is provided to the common areas. The cooling and heating system for the commercial units rely on the three heat exchangers which exchange the heat in summer or the cold in winter with the water from the sea. - The developed system is influencing only a sub system and not the whole HVAC system of the building. - The implementation of new technologies needs to be integrated with respect of the daily work in an airport. - The reliability of the system needs to be proven before the project ends. - The assessment of the baseline energy consumption and the reporting energy consumption needs to be carried out at the same time. 4.4 Conclusion and impact on the Use Case The measurement and verification method proposed in the IPMVP protocol can be applied for any type of systems. The presented equations for calculating the energy savings are general and should be individually adapted to the scope of the project. This is because the protocol is only a description of a methodology and not a detailed calculation method. The original basis for energy reduction calculation presented in the PSP Work Programme 2009 and the Save energy project is a modified version of the IPMVP. In these projects it is shown how the IPMVP can be applied to a project. In general it can be said that there is a high variety of monitoring approaches to assess the effectiveness of measures. Some methods are more general and are not dependent on the field of application e.g. ISO 50001. To ensure the efficient and continual use of the developed system within the S4ECoB project, the energy management method described in ISO 50001 will be implemented. The key aspect of this method is the PDCA circle, as mentioned before it is not always possible to do all four steps within a certain time. Therefore trainings for persons working with the system will be organized during the project. Currently in the use case buildings the energy data recorded represents mostly the whole buildings. In the project only parts of the buildings are going to be equipped with the new developed network audio sensors. Due to lack of missing data it is nearly impossible to identify the energy consumed in different parts of the buildings. Therefore the measured energy consumption during the past years cannot be used for the definition of the baseline period. The first period of the project was used for the identification of the relevant systems in the demo sites. Figure 5 shows the approach suggested in this work. The time 𝛥t1 is the time of the measurement period and 𝛥t2 the time which is needed to engage the system after the change of the control strategy. If the HVAC system of the buildings reacts quickly to the change e.g. direct electrical heating systems, than the time 𝛥t1 could be one day. If the controls are changing set points in a hot water storage tank the Energy strategy Education and training of technical personal Setting an energy saving target The methodology developed for the S4ECoB project is essential for the achievement of the project. The main objective of this method is to reduce the energy consumption of the demo sites and to give organizations involved guidance for continuous improvements in their energy strategy. This of course leads to the problem of baseline definition, together with a tight time frame, as the evaluation is to be done within the projects duration. Therefore a solution satisfying both the methodology 139 time 𝛥t1 should be adapted to the dynamics of the storage. Performance Indicator t1 t2 t1 t2 t1 t2 Energy and Water Savings. EVO 1000 – 1:2012. Efficiency Valuation Organization (EVO), Washington DC, USA. Hicks T.W., and von Neida B. 2000. An Evaluation of America’s First ENERGY STAR® Buildings: The Class of 1999. US Environmental Protection Agency, Washington DC, USA. ICT PSP, 2009. Information Communication Technologies Policy Support Programme 2009. European Commission (EC). ICT PSP, 2011. The ICT methodology for energy saving measurement - A common deliverable from projects of ICT for sustainable growth in the residential sector. Version 2. European Commision, Bruxelles, Belgium. iPHA, 2013. International Passive House Association [WWW Document]. URL http://www.passivehouseinternational.org/ (accessed 3.1.13). ISO, 2011. ISO 50001:2011 Energy management systems - Requirements with guidance for use. International Organisation for Standardisation (ISO), Switzerland. Sea Energia, 2011. Linate Plant | Sea Energia [WWW Document]. URL http://www.sea-energia.eu/en/plants/linateplant (accessed 5.5.13). US GBC, 2013. U.S. Green Building Council | U.S. Green Building Council [WWW Document]. URL http://new.usgbc.org/ (accessed 3.1.13). Wärtsilä, 2007. Linate, Italy - Wärtsilä power plant reference [WWW Document]. URL http://www.wartsila.com/en/references/linate (accessed 5.5.13). t1 t2 t2 Change of Control Strategies Baseline Period Reporting Period Baseline Period Reporting Period Time Figure 5. S4ECoB approach for the definition of the baseline and reporting period. 6 CONCLUSION AND OUTLOOK The approach discussed in this paper can be considered as a good compromise between the requirements of the methodologies surveyed and the practical constraints encountered in actual projects. The limiting factor is of course that the approach is constrained to easily reversible measures. On the other hand, the improvement of building controls is a big issue in large scale commercial building, and these measures can lead to high savings (Hicks and von Neida, 2000). Furthermore, they can be used to quantify the change in energy consumption within a shorter timeframe, making the approach useful for projects with time and budget constraints that prohibit the establishment of the baselines suggested by the methodologies reviewed. 7 REFERENCES ASHRAE, 2011. 2011 ASHRAE Handbook—HVAC Applications. American Society of Heating, Refrigerating & AirConditioning Engineers (ASHRAE), Atlanta, Georgia. CEN, 2007a. BS EN 15217:2007 Energy performance of buildings - Methods for expressing energy performance and for energy certification of buildings. European Commitee for Standardisation (CEN), Brussels. CEN, 2007b. EN 15459:2007 Energy performance of buildings - Economic evaluation procedure for energy systems in buildings. European Commitee for Standardisation (CEN), Brussels. CEN, 2008a. EN 15603:2008 Energy performance of buildings - Overall energy use and definition of energy ratings. European Commitee for Standardisation (CEN), Brussels. CEN, 2008b. EN ISO 13790:2008 Energy performance of buildings - Calculation of energy use for space heating and cooling. European Commitee for Standardisation (CEN), Brussels. EVO, 2012. International Performance Measurement and Verification Protocol Concepts and Options for Determining 140 The mapping of simulated climate-dependent building innovations A.W.M. (Jos) van Schijndel Department of the Built Environment, Eindhoven University of Technology, Netherlands ABSTRACT: Performances of building energy innovations are most of the time dependent on the external climate conditions. This means a high performance of a specific innovation in a certain part of Europe, does not imply the same performances in other regions. The mapping of simulated building performances at the EU scale could prevent the waste of potential good ideas by identifying the best region for a specific innovation. This paper presents a methodology for obtaining maps of performances of building innovations that are virtually spread over whole Europe. It is concluded that these maps are useful for finding regions at the EU where innovations have the highest expected performances. rope. Figure 1 shows an exemplarily result of the application of a freezing event map. 1 INTRODUCTION Due to energy efficiency, there exist a lot of studies on innovative buildings systems. The performances of these innovations are mostly very dependent on the external climate conditions. This also means that a high performance of a specific innovation in a certain part of Europe, does not imply the same performances in other regions. Similar, innovations that did not perform very well due to local climate conditions, and therefore not commercialised, could still perform quite well in other climates. The latter can be seen as ‘wasted’ innovations. The mapping of simulated building systems performances at the EU scale could prevent this wasting of potential good ideas by identifying the best region for a specific innovation. This paper presents a methodology for obtaining maps of performances of building systems innovations that are virtually spread over whole Europe. Due to the novelty of the methodology it was quite difficult to find relevant references including both building simulation as well as EU mapping. Therefore the literature on both topics is presented separately. See the next Section and Section 2.2. Figure 1. Pan-European maps of average yearly freezing events in 30 years period 1961–1990 (top) and far future 2070–2099 (bottom) by Grossi et al. (2007). Similar maps as presented in Figure 1 are used to show the expected reduction of freezing and lowering the potential for frost shattering of porous building stone. The underlying data for these maps are based on regional climate models. This is the second research area where EU maps are commonly used. There is an enormous amount of literature on climate change and mapping. Therefore we illustrate the use of these maps by one state of the art regional climate model: REMO (Jacob 1997, Larsen 2010). Figure 2 shows the twenty-five-year mean modeled wind at 10 m height over Central Europe using REMO with a 10 km resolution. 1.1 Related work on maps In this Section we will focus on two important building related research areas where EU mappings are already common techniques. First, we start with cultural heritage and climate change. Grossi et al. (2007) are using maps to visualize the prediction of the evolution in frost patterns due to climate change during the 21st century and the potential damage to historic structures and archeological remains in Eu141 Figure 2. Twenty-five-year mean modeled wind at 10 m height over the entire domain REMO 10 km resolution (Larsen 2010). Figure 3. The distributions of the locations of the external climates in Europe. Maps like figure 2 are suited for wind energy assessment application in Northern Europe. Moreover, literature of the related work shows that a lot of EU maps of external climate parameters are available. Each climate file includes hourly based values for the common used external climate parameters: Horizontal global solar radiation [W/m2] (ISGH), Diffuse solar radiation [W/m2] (ISD), Cloud cover [0-1] (CI), Air temperature [oC] (TA), Relative humidity [%] (HREL), Wind speed [m/s] (WS), Wind direction [0-360o] (WD), Rain intensity [mm/h] (RN), Long wave radiation [W/m2] (ILAH). 1.2 Goal and Outline The maps presented in the previous Section are all based on external climate parameters. However, the goal of this work is to produce maps of indoor climate related building performances. The outline of the paper is as follows: Section 2 presents the methodology for obtaining maps of performances of similar buildings that are virtual spread over whole Europe. It provides a benchmark of the EU mapping of the Bestest building. The produced maps are useful for analyzing regional climate influence on building performance indicators such as energy use and indoor climate. Section 3 presents a methodology to produce maps of systems innovations using statespace models based on a commercial case study. In Section 4, the conclusions and future research are provided. 2.2 Whole building simulation model The whole building model originates from the thermal indoor climate model ELAN which was already published in 1987 (de Wit et al. 1988). Separately a model for simulating the indoor air humidity (AHUM) was developed. In 1992 the two models were combined and programmed in the MatLab environment. Since that time, the model has constantly been improved using newest techniques provided by recent MatLab versions. The current hourly-based model HAMBase, is part of the Heat, Air and Moisture Laboratory (HAMLab 2013), and is capable of simulating the indoor temperature, the indoor air humidity and energy use for heating and cooling of a multi-zone building. The physics of this model is extensively described by de Wit (2006). The main modeling considerations are summarized below. The HAMBase model uses an integrated sphere approach. It reduces the radiant temperatures to only one node. This has the advantage that complicated geometries can easily be modeled. Figure 4 shows the thermal network. Where Ta is the air temperature and Tx is a combination of air and radiant temperature. Tx is needed to calculate transmission heat losses with a combined surface coefficient. hr and hcv are the surface weighted mean surface heat transfer coefficients for convection and radiation. Φr and Φcv are respectively the radiant and convective part of the total heat input consisting of heating or cooling, casual gains and solar gains. 2 CREATING MAPS OF BUILDING INNOVATIONS USING HAMBASE The methodology used for obtaining the required simulation results and maps can be divided into three steps. These are presented in the following Sections. 2.1 External climate files Over 130 external hourly-based climate files were produced using commercially available software (Meteonorm 2011) using the so-called wac format. Figure 3 presents the distribution of the locations over Europe. 142 ΣΦxy ΣΦab Tx ing climate file of the Bestest is based on weather station near Denver (USA). For further details, see Table 1. Ta Lxa -h Φ /h Φcv cv r r Φr+hcvΦr/hr Table 1. Comparison of the HAMBase model with some cases of the standard test. Ca Case 600ff 600ff 600ff 900ff 900ff 900ff 600 600 600 600 900 900 900 Figure 4. The room radiative model as a thermal network. The HAMbase model is adapted in such a way that all climate (.wac) files in a directory are automatically processed. For each climate file a corresponding output file is produced containing hourly based values for the indoor climate and heating and cooling power. A separate Matlab mfile is developed for calculating annual means and peak values for each location (i.e. wac file) and together with the longitude and latitude stored in a single file suitable for mapping purposes. 2.3 Mapping of the results Nr. Simulation of model test min .. mean indoor temperature [oC] 25.1 24.2 .. min. indoor temperature [oC] -17.9 -18.8 .. o max. indoor temperature [ C] 64.0 64.9 .. mean indoor temperature [oC] 25.1 24.5.. -5.1 -6.4.. min. indoor temperature [oC] max. indoor temperature [oC] 43.5 41.8.. annual sensible heating [MWh] 4.9 4.3.. annual sensible cooling [MWh] 6.5 6.1.. peak heating [kW] 4.0 3.4.. peak sensible cooling [kW] 5.9 6.0.. annual sensible heating [MWh] 1.7 1.2.. annual sensible cooling [MWh] 2.6 2.1 .. peak heating [kW] 3.5 2.9 .. max 25.9 -15.6 69.5 25.9 -1.6 44.8 5.7 8.0 4.4 6.6 2.0 3.4 3.9 After this verification with the Denver climate, the case 600 building model was exposed to EU climate data, located at the weather stations of Figure 3. The energy use for heating and cooling the Bestest case 600 building is shown in Figures 6 and 7. The mean annual heating and cooling power maps are in general quite as expected, i.e. high heating amounts in the north of Europe and higher cooling amounts in the South. A MatLab mfile was developed for the visualization of the just mentioned mapping file. For the exact details of this mfile, we refer to the HAMLab website (HAMLab 2013). 2.4 Mapping benchmark: Bestest using HAMBase The Bestest (ASHRAE, (2001)) is a structured approach to evaluate the performance of building performance simulation tools. The evaluation is performed by comparing results of the tested tool relative to results by reference tools. The procedure requires simulating a number of predefined and hierarchal ordered cases. Firstly, a set of qualification cases have to be modeled and simulated. If the tool passes all qualification cases the tool is considered to perform Bestest compliant. In case of compliance failure the procedure suggests considering diagnostic cases to isolate its cause. Diagnostic cases are directly associated with the qualification cases (Judkoff and Neymark 1995). The first qualification case, case 600 (see Figure 5) was used for the performance comparison. The use of these types of maps is published in van Schijndel and Schellen (2012). Furthermore the presented maps below can be used as a future benchmark for other modeling and mapping tools. Figure 6. Mean annual heating power [W] of the Bestest case 600 building (This figure relies on color, see digital version of the paper). Figure 5. Bestest case 600 geometry The thermal part of HAMBase has been subjected to a standard method of test (Bestest ASHRAE, (2001)), with satisfactory results. The accompany143 The solar collector will be used for the heating of water that directly can be used or stored for later use. 3.1 Modeling A 3-State (3S) model was developed: Figure 7. Mean annual cooling power [W] of the Bestest case 600 building (see also digital version of the paper for color). So far the approach to produce maps was based on a building energy simulation (BES) tool HAMBase. In the next Section we continue with a new method (without BES) to produce maps using state space (SS) models. Where Inputs Tamb(t) ambient (external) air temperature [oC] Tsup(t) water supply temperature [oC] I(t) external solar irradiance [W/m2] States T1 external surface temperature [oC] T2 water return temperature [oC] = T3 internal wall temperature [oC] 3 CREATING MAPS OF SYSTEMS INNOVATIONS USING STATE-SPACE (SS) A commercial case study is presented in this Section. Due to the patent protection of the industrial partner, some specific information is omitted without loss of generality. The innovation consists of a novel heat exchanger built inside a construction acting as a solar collector. Figure 8 shows the principle construction of the solar collector (in reality this is much more complicated). Parameters: water mass flow [kg/s] c heat capacity of water [J/kgK] aI solar absorption factor [-] ( h heat transfer surface coefficient [W/m2oC] A surface [m2] d1 distance pipe to surface [m] d2 distance pipe to insulation [m] k heat conductivity of concrete [W/mK] R1 heat resistance [ K/W] = d1/(kA) R2 heat resistance [ K/W] = d2/(kA) Ci heat capacity [J/ K] The model was implemented using standard statespace modeling facilities of MatLab. The next Section shows the simulation and validations results. 3.2 Validation Laboratory experiments were used to validate the models. All experiments were simulated using the proper parameters and boundary conditions. The results were compared in order to evaluate the predictability of the model. In Figure 8 the results for a typical experiment A is shown. Figure 8. Construction of the solar collector. 144 The water supply temperature was constant held at 10 oC. The other two input signals: Ambient air temperature and solar irradiation were taken from the climate file. The main output signal is the return temperature (out). With this signal the output power can be calculated. This is shown in the next Section. 3.4 Performance evaluation Figure 10 shows details of the model A configuration performance results. Figure 8. The simulation of experiment A: Temperatures vs time of the measured supply water(sup), the measured ambient air (amb), the simulated return water (Ret sim 1 & 2) and the measured return water (Ret). From Figure 8 we observed that the predictability of model was satisfactory. All other tested configurations provided similar good results. Therefore we conclude that the model is quite usable for further use. Figure 10. Performance evaluation. Top: The simulated supply and return water temperatures versus time. Bottom: The heat flux [W/m2] of the incoming solar irradiation (Solar) and simulated output flux of the wall. 31.5 % of the year the wall system can be operated (PFt) The yearly mean efficiency is 41.5 % (PFp). 3.3 Simulation using a typical Dutch climate The model configuration A was simulated using a reference standard Dutch climate of deBilt. Figure 9 presents the result. The output flux Pout is calculated by: Pout(t) = *c*( Tret(t)-Tsup(t))/A [W/m2] The overall performance is evaluated as follows: Firstly, P50(t) is defined as Pout(t) with a threshold of 50 W/m2. Below 50W/m2, the water return temperature drops below 10.7 oC and the wall system is too inefficient. For these values P50(t) = 0. Secondly, two performance (PF) indicators are defined as follows: PFt = Percentage of time of Pout(t) above threshold of 50 W i.e. percentage of time of possible operation [%]. PFp = 100* sum(P50(t)) / sum(I(t)) , Figure 9. Simulation of model configuration A using a reference standard Dutch climate of deBilt. Temperatures versus time of the external wall surface (opp), the water return (out), the mid wall (con), the water supply (sup) and ambient air (amb). i.e. the yearly mean efficiency [%] 145 From Figure 10 it follows for configuration A, PFt=31.5% and PFp=41.5%. 3.5 The main parameter that affects the simulated performances is the mass flow of the water. Figure 11 provides the simulated performances PFt and PFp as functions of the mass flow. The following parameters were varied for the parameter study: * The distance from the pipe to the surface (default 35 mm) was varied: 20, 35 and 50 mm. * The mass flow (default 1 kg/min) was varied: 0.5, 1 and 2 kg/min. Parameter study The results are shown in Table 2 and 3. Table 2. Efficiency Performance. Simulated yearly mean efficiency PFp [%] d=20 d=35 d=50 mm mm mm MF=0.5 30.6 24.7 20.2 kg/min MF= 1 39.0 30.9 25.2 kg/min MF= 2 44.3 34.8 28.0 kg/min Table 3. Operation Time Performance. Figure 11. The simulated performances versus the mass flow. Simulated Operation time PFt [%] d=20 d=35 d=50 mm mm mm MF=0.5 29.8 26.5 23.7 kg/min MF= 1 33.1 29.5 26.5 kg/min MF= 2 34.5 30.9 27.7 kg/min Figure 12 presents the influence of the pump energy and surface heat transfer coefficient. The optimal efficiency performance for a Dutch climate is 44.3% with the accompanying mass flow of 2 kg/min and pipe depth of 20 mm. 3.6 EU Mapping of the standard configuration By replacing the Dutch climate with the climates of weather stations presented in Figure 3, it is quite easy to simulate the response of the system to each external climate. From the responses the performance indicators can be calculated (See previous Section). The results of the standard wall performances are shown in Figures 13 and 14. These results are still based on the standard wall configuration A. Figure 12. Influence of the pump energy and surface heat transfer coefficient. Top: The influence of a change in heat transfer surface coefficient. Bottom: Correction of the performances using pump energy. For further simulations a more realistic surface heat transfer coefficient of 25 W/m2K is used instead of 7 W/m2K from the indoor experiment. The latter (i.e. h=7 W/m2K) was used for the validation of the experiments. Furthermore, for the water mass flow, values between 0.2 and 2 l/min are used. 146 Figure 15. Optimized wall configuration Efficiency (PFp). Figure 13. Efficiency (PFp) of the standard wall configuration. From figure 10 it can be seen that large parts of Europe have efficiencies of at least 45%. Figure 14. Percentage of time operation (PFt) of the standard wall configuration. Figure 16. Optimized wall configuration Percentage of time operation (PFt). From figure 16 it can be seen that the areas near the Mediterranean have percentages of time of operation above 60%. The latter means that the wall collector is also operational during parts of the night. 3.7 Simulation of optimized wall configurations All nine configurations of the parameter study (see Table II and III) were also simulated on the EU scale. For each weather station the best configuration out of nine was selected. These optimized wall configuration performances are presented in Figures 15 and 16 4 CONCLUSIONS 147 Mapping using HAMBase The produced maps are useful for analyzing regional climate influence on building performance indicators such as energy use and indoor climate. This is shown using the Bestest building as a reference benchmark. An important application of the mapping tool is the visualization of potential building measures over the EU. Also the performances of single building components can be simulated and mapped. It is concluded that the presented method efficient as it takes less 6 REFERENCES than 15 minutes to simulate and produce the maps on a 2.6GHz/4GB computer. Moreover, the approach is applicable for any type of building ASHRAE 2001. Standard method of test for the evaluation of building energy analysis computer programs, standard 1402001. Grossi C.M., Brimblecombe P., and Harris I. 2007. Predicting long term freeze-thaw risks on Europe built heritage and archaeological sites in a changing climate. Science of the Total Environment 377, pp 273-281 HAMLab (2013), http://archbps1.campus.tue.nl/bpswiki/index.php/Hamlab IPCC, Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Core Writing Team, e.d. Pachauri, R.K and Reisinger, A., IPCC, Geneva, Switzerland (2007). Jacob D., and Podzun R. 1997. Sensitivity studies with the Regional Climate Model REMO. Meteorology and Atmospheric Physics, 63, pp 119-129. Judkoff R., and Neymark J. 1995. “International energy agency building energy simulation test (Bestest) and diagnostic method”, National Renewable Energy Laboratory, Golden, CO. Kramer R. 2012. From castle to binary code: the application of inverse modeling for the prediction and characterization of indoor climates and energy performances, Eindhoven. Larsen X.G., Mann J., Berg J., Gottel H., and Jacob D. 2010. Wind climate from the regional climate model REMO. Wind Energy 13, pp 279-206 Max Planck Institute for Meteorology, www.mpimet.mpg.de (accessed Feb 2013) Majewski D. 1991. The Europa-Modell of the Deutscher Wetterdienst. ECMWF Seminar on Numerical Methods in Atmospheric Models, Vol 2 (1991), pp. 147-191. Meteonorm (2013).http:/meteonorm.com/ Perez R. et al., 1987. A new simplified version of the perez diffuse irradiance model for tilted surfaces. Solar Energy, 39(3), pp. p.221-231. Schijndel A.W.M. van 2007. Integrated Heat Air and Moisture Modeling and Simulation, PhD Dissertation, Eindhoven University of Technology Schijndel A.W.M. van, and Schellen H.L. 2012. The simulation and mapping of building performance Indicators based on European Weather Stations. Proc. of the 5TH IBPC Conference Kyoto pp1-8 Schijndel A.W.M. van and Wit M.H. de 1999. A building physics toolbox in MatLab, 7TH Symposium on Building Physics in the Nordic Countries Goteborg, pp 81-88 Wit M.H. de, and Driessen H.H. 1988. ELAN A Computer Model for Building Energy Design. Building and Environment 23, pp 285-289 Wit M.H. de 2006. HAMBase, Heat, Air and Moisture Model for Building and Systems Evaluation, Bouwstenen 100, Eindhoven University of Technology Mapping using state space The main objective was to simulate and optimize the thermal performance of innovative solar collector under different EU climate conditions using state space modeling: (1) The solar collector was successfully modeled; (2) The validation of this model using existing measurements was satisfactory; (3) The solar collector model was successfully simulation using 130 EU weather stations; (4) Estimation of minimal and maximal performance was done by a parameter study; (5) EU Maps of the performance were created. Regarding the EU performance of the solar collector Large parts of Europe have solar collector efficiencies of at least 45%, the exact details are provided in Figure 15. Furthermore, areas near the Mediterranean have percentages of time of operation above 60% (exact details are shown in Figure 16). The latter means that the solar collector is even operational during parts of the night. It is concluded that this study shows that the solar collector could be applicable in large parts of Europe. However, the reader should notice that the solar collector simulation results in this study are based on two assumptions: The supply water temperature is constant at 10 oC and all heat produced by the wall collector is usable at any time. Under most circumstances this is not very realistic. Therefore it is recommended to include buildings, systems and controllers details into the modeling for more realistic performance simulations and design of promising integrated configurations. 5 FUTURE RESEARCH Including future climates Within the mentioned EU FP7 project ‘Climate for Culture’, detailed EU external climate files are currently under development for the period 1960 – 2100 using the REMO model (Jacob et al. 1997) and a moderate climate scenario. With these future external climate files we will be able to predict future building performance indicators. Together with the EU mapping tool this could be helpful to locate EU regions with the highest impact on the specific building performances. Towards a state space based mapping tool Currently we are working on a more general state space mapping tool in MatLab. This tool will become public available. 148 Thermal performance effect of hollow ceramic microspheres coating assessed by dynamic outdoor testing in the summer season A.M. Čekon & M. Kalousek Institute of Building Structures, Brno University of Technology, Faculty of Civil Engineering, Brno, Czech R. B.R. Ingeli Department of Building Structures, Slovak University of Technology, Faculty of Civil Engineering, Bratislava, Slovak Republic ABSTRACT: The paper presents the results of a comparative study investigating the thermal performance of coatings containing hollow ceramic microspheres, as regards their impact upon the energy balance of buildings during the summer period. This research has been inspired by the intention to reduce the costs of heating and cooling of buildings by using appropriate combinations of various surface coating modifications and verifying their features guaranteed by manufacturers. There are both expectations and doubts that have not yet been properly investigated in many cases. At the present day the building industries are already aware of more precise evaluation and measurement methods that are not yet commonly used in the area of building physics. We decided to investigate the thermal performance of the external building coatings by way of the Twin-Box method that is suitable for outdoor dynamic testing. Samples were prepared, and namely the coating of microspheres base to be compared and a standard facing for reference. The measured data testify to the difference of spectral reflectance properties as well as thermal performance effects made up app. 5% in comparison of the standard façade coatings of an identical color shade. The samples were monitored by dynamic outdoor testing especially under clear sky daytime conditions. The spectral reflectance properties in the region of visible and shortwave radiation were assessed by spectroscopy methods. Both used methods were finally confronted. microspheres do not possess the declared features, and concerning the reflective properties in the longwave radiation region they show no difference at all from current façade coatings (Čekon 2012). The examined coatings do not affect the energy balance as currently declared, and there is not the least reason to maintain that they could be alternatively used compared with standard thermal insulation materials (Čekon 2013). Another field that has not yet been given due attention relates to saving energy for cooling down the buildings. This is linked with the issue of optical properties of the building surface in the range of near infrared and visible radiation areas. The focus of the present article lies especially upon the possible savings of energy for cooling down the buildings and in examining the thermodynamic properties of coatings for the summer season. In general one of the most important roles in the mechanism of heat transfer by radiation is played by the optical properties of material depending, in an ideal case, upon the impact angle and the wavelength of radiation (Table 1). 1 INTRODUCTION Within the stipulated requirements as to the thermal protection of buildings the radiating heat flow represents, as a rule, a negligible proportion of the resulting energy flow. Under dynamic conditions of a building environment, however, the reflective properties of the surfaces of building structures can make up for a considerable part of the transition phenomenon in the longwave, the visible and the shortwave radiation regions. The starting point for investigating the efficiency of various coatings in correlation with the building physics was based upon the fact that currently building market offers promising coatings with all sorts of positive features predominantly declared by their manufacturers. One of the methods for the determination of the efficiency of such materials examines the thermodynamic properties of coatings by way of the spectral reflectance properties in the short-wave infrared and visible radiation regions (Sonel 2006). A number of studies have dealt with these problems as concerns excessive warming up and reducing the consumption of energy both for cooling and heating purposes (Griggs & Shipp 1988, Bretz & Akbari 1997, Akridge 1998, Hildebrandt 1998, Simpson & McPherson 1997, Konopacki 1998, Akbari 2003). As regards local sites, Čekon (2011) has investigated the effect of thermodynamic properties of reflective coatings upon the heat balance during the winter season, i.e. during the heating period. The results have shown that coatings based upon ceramic Table 1. Optical properties of building surfaces. Directional spectral reflectance Directional spectral absorbance Directional spectral transmittance Directional spectral emissivity ρg,l αg,l τg,l εg,l [-] [-] [-] [-] The materials of the building surfaces with their internal and external layers can be considered as not transmissive in the whole range of thermal radiation 149 2 METHODICAL APPROACH AND FOCUS (τg,l= 0). The wavelength is influenced by the thermodynamic magnitude of the source of thermal radiation T [K]. The effect of the radiating properties of materials under current conditions in the building field can be considered and determined independently from the source of thermal radiation, as long as the difference is small enough to avoid any change of the microstructure of material that might show such effect. The decisive range of the electromagnetic waves affecting the process of thermal radiation in the building field is roughly within 0.40 to 40.00 µm, comprising the visible electromagnetic radiation (0.38-0.78 µm), shortwave thermal radiation (0.782.50 µm) and longwave thermal radiation (2.5025.00 µm). The characteristic regions defined by these boundaries in the building applications are as follows: The main objective of this contribution resides in examining the thermodynamic behavior of coatings that are based upon hollow ceramic microspheres (HCM), as compared with the standard façade coatings for summer period. The first step used dynamic evaluation methods. The experiment under real climatic conditions allowed comparing the effects of outdoor dynamic conditions during the summer periods, examining the energy balance of samples coated on their exterior sides. In order to make a comprehensive analysis of the thermodynamic behavior of coatings, we decided to complement the dynamic outdoor testing under real summer conditions by the verification of our findings by laboratory methods. For this purpose spectral methods were applied; by way of reflective measurement techniques, determining the parameters of reflectance in the visible as well as the near infrared radiation region, with spectral correlation. − visible electromagnetic radiation - with application especially concerning the reflective features of the surface finish color schemes, − shortwave thermal radiation – predominantly on outdoor surfaces, the source of radiation being the Sun (T = 6000 K), or heat sources with thermodynamic temperature exceeding 800 K, − longwave thermal radiation - predominantly on indoor surfaces, the source of this type of radiation being, e.g., current heating systems achieving thermodynamic temperatures around 350 K. Each range of thermal radiation is of specific importance in the building physics. During the summer season the absorption capacity of the surface of building structures, affecting the transition phenomena of construction surfaces, is essential from the viewpoint of thermal radiation. The decisive range is the spectral length approximately 0.40-2.50 µm, the most important effect lying at the border between the visible and the shortwave infrared radiation. Thermal performance effect of this area is significantly dependent on irradiance level of the sun in particular wavelength. Solar radiation spectrum plays an important role. On the contrary, in winter or during transitory periods the building envelope tends to be markedly chilled due to radiating heat exchange, especially under clear and cloudless skies. The marginal condition of thermal radiation shifts towards longer waves in the longwave region. An optimal management of thermal radiation in the summer months with regard to thermal efficiency favors the elimination of thermal radiation through the building structures in the shortwave range (0.40-2.50 µm), and at the same time its increase in the longwave range of radiation (2.50-40.00 µm).. The maximum impact within these regions is seen approximately at 1 µm in the shortwave radiation and approximately around 10 µm in the longwave radiation region. 3 DYNAMIC OUTDOOR TESTING 3.1 Methodical approach The measurement apparatus Twin-boxes at the Central laboratories (Faculty of Civil Engineering in Bratislava) was used. This measurement apparatus was successfully integrated in the research facility, specifically for the application in dynamic outdoor testing of advanced materials. The basic concept is based upon the protected hot-box method with principles of dynamic outdoor testing (Strachan & Baker 2008, Wouters et. al 1993, Baker & v. Dijk 2008). It consists of two identical metering boxes for comparing the non-steady state heat transfer through the building components. It is intended for measuring heat flows and passive solar gains of examined samples. The samples are directly exposed to the outdoor conditions. More detailed information has been given in (Čekon 2011). The energy balance effects of the studied coatings were analyzed on several samples of 1.2 x 1.2 m2 dimensions. After the initial shaded phase they were finally exposed to the dynamic outdoor conditions (Fig. 1.). Each sample was coated from its exterior side. The difference between these samples consisted only in the external coating modification (Table 2). Each sample was accommodated upon a lightweight concrete base of 50 mm thickness with 5 mm fine plaster on both sides. The sample with a standard facing coating was used as reference. A sample of the most wide-spread HCM coating on the market and a sample coated with standard acrylic-based facing coating were used for comparative testing. 150 3.2 Results of outdoor dynamic testing and analysis of measured data The results of each measurement period are shown in four diagrams. The first diagram registers the temperature flow curves depending upon the outdoor air temperature, the balance of longwave radiation at the level of the samples, and a differential heat flow curve in absolute values. The next diagram shows the courses of solar radiation intensity and the courses of average surface temperatures within the area of the measured sample. The last diagram shows the difference of heat flow in an augmented scale. During the measurements also the parameters of relative humidity of ambient air and the velocity and direction of wind were registered, without becoming integrated in the final diagrams. The first measurement period (Figs. 2-5) allows the statement that the registered curves do not show differences in the course of the measured flow and the surface temperatures depending upon the intensity of solar radiation. This is due to the ambient climatic conditions, the intensity of solar radiation upon the measured samples having been rather low. (Fig. 3). Actually, it was overcast most of the time, with occasional rain. Accordingly, the surface and the optical properties of the investigated samples did not manifest themselves at all. The same applies to the efficiency values regarding the heat losses of the examined coatings under night conditions. A similar effect was documented in the second period of measurement (Figs. 6-9). Solar radiation did not achieve intensities sufficient for allowing the impact upon the monitored surface finish to get manifested as part of the energetic balance. One diagram (Fig. 6) documents the effect of HCM upon the course of the thermal flow when the intensity of the overall solar radiation achieved the average of 400 W/m2 from the morning hours to midday (Fig. 8). The result of the tested sample is about 5 W lower (Fig.9), corresponding with approx. 15% difference against the standard reference coating. During the remaining time no effects were documented, since the impact of solar radiation upon the monitored surfaces did not reach sufficient levels. During the last registered period, however, the effects of the investigated coatings were pronounced. This was due to the climatic conditions with clear sunny days. The results represented in diagrams (Figs. 10-14) show differences of thermal flow up to 10W (Fig. 14) during the day, a difference amounting to some 15 %. The maximum solar radiation intensities were monitored in area of measured samples at the level of 700 W/m2. Since the supplied samples that were subject to a detailed visual analysis did not look as being fully identical, specimens of them were taken for spectral laboratory analysis with the purpose of verifying their optical properties. Figure 1. External view of the samples, with coating consisting of HCM vs. standard facing coating. The measurements were carried out in three periods of about one week each. In order to eliminate both the radiation influence and the climatic conditions, the initial part of the first period was measured under shaded conditions. After about two days the samples were finally exposed to outdoor climatic conditions. The resulting heat flow through the measured samples was then analyzed together with monitoring a number of other physical parameters making up the climatic conditions, such as the surface temperatures, solar radiation and the longwave radiation balance between samples (imagined with ambient temperature) and their environment (Table 3). Table 2. Table of measured samples. Sample: Thickness Composition: [mm] Lightweight concrete F_ref 50 mm + standard facing with acrylic based paint Lightweight concrete F 50 mm + coating consisting of HCM Color: orange colored orange colored Table 3. Table of symbols, indices and markings of the measured and computed values in the diagrams. Symbol: Expression and description: Unit 1 index box 1 [-] 2 index box 2 [-] F, F_ref. Resulting heat flow [W] Llw,net Net longwave radiation balance [W/m2] diff Absolute heat flow difference [W] Tesa Average external surface temp. [°C] Tisa Average internal surface temp. [°C] Te Outdoor air temperature [°C] Internal temperature in the metering Ti [°C] box Igh Global horizontal solar radiation [W/m2] Igv Total vertical solar radiation [W/m2] Idh Diffuse horizontal solar radiation [W/m2] 151 Figure 2. Energy balance between measured samples, 18 July. 2011 (12:00 a.m.) – 22 July. 2011 (18:00 p.m.): Llw,net – net longwave radiation balance in front of both samples, F_ref – heat flow of reference sample, F - heat flow of comparative sample, diff. – difference between heat flows. Figure 3. Solar radiation intensity, 18 July. 2011 (12:00 a.m.) – 22 July. 2011 (18:00 p.m.): Igh – global horizontal solar radiation, Igv – global vertical solar radiation, Idh – diffuse horizontal solar radiation. Figure 4. Temperatures progresses, 18 July. 2011 (12:00 a.m.) – 22 July. 2011 (18:00 p.m.). Figure 5. Energy balance between measured samples, 18 July. 2011 (12:00 a.m.) – 22 July. 2011 (18:00 p.m.): diff. – difference between heat flows. 152 Figure 6. Energy balance between measured samples, 22 July. 2011 (18:00 p.m.) – 27 July. 2011 (08:00 a.m.): Llw,net – net longwave radiation balance in front of both samples, F_ref – heat flow of reference sample, F - heat flow of comparative sample, diff. – difference between heat flows. Figure 7. Solar radiation intensity, 22 July. 2011 (18:00 p.m.) – 27 July. 2011 (08:00 a.m.): Igh – global horizontal solar radiation, Igv – global vertical solar radiation, Idh – diffuse horizontal solar radiation. Figure 8. Temperatures progresses, 22 July. 2011 (18:00 p.m.) – 27 July. 2011 (08:00 a.m.). Figure 9. Energy balance between measured samples, 22 July. 2011 (18:00 p.m.) – 27 July. 2011 (08:00 a.m.): diff. – difference between heat flows. 153 Figure 10. Energy balance between measured samples, 16 August. 2011 (19:00 p.m.) – 19 August. 2011 (12:00 a.m.): Llw,net – net longwave radiation balance in front of both samples, F_ref – heat flow of reference sample, F - heat flow of comparative sample, diff. – difference between heat flows. Figure 11. Solar radiation intensity, 16 August. 2011 (19:00 p.m.) – 19 August. 2011 (12:00 a.m.): Igh – global horizontal solar radiation, Igv – global vertical solar radiation, Idh – diffuse horizontal solar radiation. Figure 12. Temperatures progresses, 16 August. 2011 (19:00 p.m.) – 19 August. 2011 (12:00 a.m.). Figure 13. Energy balance between measured samples, 16 August. 2011 (19:00 p.m.) – 19 August. 2011 (12:00 a.m.): diff. – difference between heat flows. 154 4 EXPERIMENTAL VERIFICATION OF VIS AND NIR SPECTRAL REFLECTANCE OF THE STUDIED COATINGS 4.1 Method and measuring instrument The key role was focused on the verification of the optical properties of the tested samples by way of laboratory methods in the visible (VIS) and shortwave (NIR) radiation regions. Whereas color and glossiness of the surface are the decisive factors for the visible region, the material composition and the texture of the surface are the leading ones for the shortwave range. The actual purpose of the analysis resided in the verification of the reflectance parameters, especially in the visible radiation range. There was an adverse factor for measuring under dynamic conditions in that the coatings contained certain pigments due to which the color was not identical in some moments. In order to compare identical samples without any colored effect by means of laboratory methods, the white specimen of each coating were prepared for comparison as well. An important aspect of thermal performance effect for this area is based on knowledge of spectral irradiance of the sun with particular dependence on wavelength (Fig 14). Figure 16. 150 mm Spectralon integrating spheres. 4.2 Results of Spectral Verification and Analysis of Measured Data The measured results were authenticated by repeated measurements of each sample. This allowed the precision of various methods and measuring devices to be evaluated. The comparison of multiple measurements of each specimen enabled us to achieve results with negligible deviation and, among others, to simplify the diagrams in showing only one single spectral curve representing one sample. The diagram (Fig. 17) shows the spectral curves of reflectance in the visible range of radiation (360 nm to 780 nm). The registered spectral course within the range from yellow to red light differs up to some 10% of reflectance for colored samples. There are also registered spectral curves of white samples each coating in order to compare identical color, where the difference was recorded up to 5%. The most important area goes to 1000 nm for the highest solar radiation impact. The next diagram (Fig. 18) shows the spectral curves of reflectance in the near infrared region between 1000 nm to 2500 nm. The registered spectral courses are in some regions not identical. Figure 14. Solar radiation spectrum (http://commons.wikimedia.org/wiki/File:Solar_Spectrum.png). An UV/VIS/NIR spectrophotometer Lambda 1050 (Fig. 15) with 150 mm Spectralon integrating sphere (Fig. 16) was used in the visible and shortwave radiation region. This apparatus can register the course of spectral reflectance properties in the region from 200 nm to 3300 nm. Figure 17. Spectral reflectance in VIS region. Figure 18. Spectral reflectance in NIR region. Figure 15. UV/VIS/NIR Spectrophotometer Lambda 1050 with 150 mm Spectralon integrating spheres. 155 impacts of climate, in comparison with acrylic based. In the final consequence it can protect the building against rain and influence its humidity regime. Also these aspects are substantial for the heating energy savings, while affecting especially the heat losses through the building construction. 5 CONCLUSION The paper presents an analysis of the thermal performance of coatings containing hollow ceramic microspheres, as compared with current facade coatings, carried out during the summer months. The results of measurements carried out under real climatic conditions indicate that the effect of coatings containing HCM can be proven in the light of the different course of heat flow against the standard façade coatings. However, the impact was proven only during clear and sunny days, with impacting intensities of solar radiation on the monitored surface amounting to at least 400 W/m2. The difference against the standard coating can be quantified on a level not exceeding about 15%, depending upon the measured composition and the climatic conditions. In addition to that, laboratory methods were used to analyze the reflective properties of samples in the visible and the near infrared region, in order to verify the results of dynamic testing and to do away with any possible doubt due to the not fully identical appearance of the tested samples. The results show that the value of spectral reflectance of the HCM coating is about 10% higher than that of the reference coating. This fact exerts a substantial effect upon the results under dynamic conditions. Accordingly, a statement can be made that the primary effect must have been due to different properties in the visible radiation region. In other words, the coatings did not have an identical color scheme, yet on the other hand, without allowing their visual identification. One of the important features of the compared samples, however, is the visually recognizable glossiness of the surfaces. Whereas the standard acrylic based façade coating is characterized by its matt surface, the coating of microspheres base, in general, is visibly glossier. The glossiness of coatings of this type is exactly due to their ceramic component. Within the infrared region, however, both in the near and the longwave one (Čekon 2012), the compared coatings can be assessed as having almost identical reflective properties, thus affecting radiant heat flow and the thermal parameters of building structures from the viewpoint of building physics in practically the same level. The laboratory tests and the results of dynamic testing allow formulating a final conclusion that the coatings based upon ceramic microspheres can contribute to positive energetic effect during the summer season, saving app. 5% as in comparison of the standard façade coatings of an identical color shade. On the other hand, a coating containing ceramic microspheres with its specific features, such as steadfast color, lifetime, mechanical strength and further factors, can contribute at least by small additions to the energetic effect of a building. From this viewpoint such coating can be characterized as an efficient protection of building structures against the This research was supported by the project CZ.1.07/2.3.00/30.0039 of Brno University of Technology, CZ.1.05/2.1.00/03.0097 of AdMaS Center and VEGA No. 1/0281/12 research project of Slovak University of Technology. 6 REFERENCES Akbari H. 2003. Measured energy savings from the application of reflective roofs in 2 small non-residential buildings, Energy 28 (2003) 953–967. Akridge J. 1998. High-albedo roof coatings - impact on energy consumption, ASHRAE Techical Data Bulletin 14 (2) (1998). Baker P.H., and van Dijk H.A.L. 2008, PASLINK and dynamic outdoor testing of building components, Building and Environment 43(2008), pp. 143–151. Berdhal P., and Bretz S. 1997. Preliminary survey of the solar reflectance of roofing materials, Energy and Buildings 25 (2) (1997) 149–158. Bretz S., and Akbari H. 1997. Long-term perf. of high-albedo roof coatings, Energy and Buildings 25 (2) (1997) 159-167. Čekon M. 2011. Thermodynamic Properties of Low-Emissivity Coatings of Building Surfaces (Termodynamické vlastnosti nízkoemisných náterov stavebných povrchov): dissertation thesis. Bratislava, 2011. 127 pp. (in Slovak). Čekon M. 2012. Spectral Emissivity Properties of Reflective Coatings. Slovak Journal of Civil Engineering Vol. 20, 2012, No. 2. p. 1--7. ISSN 1210-3896. Čekon M. 2013. Thermodynamic properties of reflective coatings. Advanced Materials Research Vol. 649, Trans Tech Publications, Switzerland, 2013. p. 179-183. ISSN 1022-6680. Griggs E.I., and Shipp P.H. 1988. The impact of surface reflectance on the thermal performance of roofs: an experimental study, ASHRAE Trans., ARHRAE Handbook of Fundamentals, 94 (1988), p. 1626 Hildebrandt E. et al. 1998. Assessing the impacts of white roofs on building energy loads, ASHRAE Technical Data Bulletin 14 (2) (1998). Konopacki S. et al. 1998. Demonstration of energy savings of cool roofs, Lawrence Berkeley National Laboratory Report LBNL-40673, Berkeley, CA, 1998. Simpson J.R., and McPherson E.G. 1997. The effect of roof albedo modification on cooling loads of scale residences in Tucson, Arizona, Energy Buildings 25 (1997) 127–137. Sloan J. et al. 2006. Laboratory Evaluation of Commercial Coatings for Use by Soldiers in the Field to Lower Operating Temperatures of Collapsible Fuel Tanks. Army Research Laboratory, ARL-TR-3728, Aberdeen Proving Ground, MD 21005-5069, ADA444745, p. 28 Strachan P.A., and Baker P.H. 2008. Special issue on outdoor testing, analysis and modelling of building components (Editorial), Building and Environment 43(2008), pp. 127–128. Synnefa A., Santamouris M., Livada I. 2005. A Study of the Thermal Performance of Reflective Coatings for the Urban Environment. In Solar Energy, 2005, vol. 80(8), pp. 968–981. Wouters P. et al. 1993. The use of outdoor test cells for thermal and solar building research within the PASSYS Project. Building and Environment, 28 (2) (1993), pp. 107–113 156 Optimization of window installations in deep energy retrofits using vacuum insulation panels Cezary Misiopecki Fellow at FraunhoferCSE, Building Energy Efficiency Group, Boston, MA, United States Ph.D. Candidate, Norwegian University of Science and Technology, Trondheim, Norway William C. duPont, Jan Kosny, Ali Fallahi & Nitin Shukla FraunhoferCSE, Building Energy Efficiency Group, Boston, MA, United States Arild Gustavsen Norwegian University of Science and Technology, Trondheim, Norway ABSTRACT: In order to reduce the heating and cooling demand in structures, it is often more cost effective to retrofit existing building envelopes with additional insulation or improved fenestration. This paper explores the optimum location of replacement windows in retrofit applications with thicker building wall systems due to the addition of insulation to existing walls. Replacement windows may be installed at different depths in the wall aperture, e.g. close to the interior, in the middle, or close to the exterior of the window opening. In addition, this project identifies the optimum location to install the replacement windows in two types of retrofitted wall systems; those using Vacuum Panel Insulation, and those using the thicker conventional insulation systems (e.g., Extruded Polystyrene). Two-dimensional computer modelling is used to determine the most energy-efficient position to install the window. In order to differentiate between the different window installation systems, the linear thermal transmittance of the window-wall connection, as determined by ISO 10211 and THERM is also calculated and presented. construction skills. The intent of this project is to improve the design and construction of energyefficient building envelopes when retrofitting existing structures. The Vacuum Insulation Panels (VIP) used in this project are 25mm thick flat panels, which have a thermal conductance that is an order of magnitude lower than the same thickness of conventional insulation. Typically, VIPs consist of an open porous core material (usually made of pressed fumed silica) sealed in a gas-tight envelope (i.e., metal foil) which maintains the vacuum that was created inside the panel during the manufacturing process. This kind of insulation was initially used to insulate refrigerators and freezers, and have been recently used for insulating building walls, floors and roofs (Erb et al., 2005). VIPs gives the building designer greater flexibility as they can create thin, yet highly-insulating wall systems. This insulation technology can also be successfully used in retrofit applications. As VIPs are much thinner than conventional insulation for a given thermal resistance, the existing window and door openings may not have to be significantly modified. Given these advantages, there are a number of features of VIPs that have presented challenges to their wide-spread acceptance in the marketplace. Their high cost is likely the greatest obstacle, but they are also relatively fragile compared to existing insulation materials, VIPs require careful installation by builders trained in this new building practice. If a VIP is punctured, it loses its vacuum, and the thermal conductivity of the VIP will significantly in- 1 INTRODUCTION Nowadays, saving energy and reducing carbon emissions is one of the top priorities for many countries and communities. Buildings are an important part of human existence, as people spend most of their lives inside enclosed structures. Energy use in buildings worldwide accounts for over 40% of the primary energy use and for around 24% of greenhouse gas emissions. This energy use and these emissions include the direct usage of fossil fuels (on-site), and the indirect use of energy in the form of electricity, district heating, district cooling and the embodied energy in construction materials (Voss and Musall, 2011). A study carried out by International Energy Agency (IEA) shows that older buildings currently in existence will use most of the energy consumed by all buildings in the future. Predictions show that new buildings will only contribute 10-20% of the total energy consumed by buildings by 2050 in most industrialized countries. This indicates that more than 80% of the energy used in structures will be consumed by buildings that already exist (IEA, 2011). Existing buildings may still be functional, occupied, and be located in a prime location, and therefore, cannot be easily or inexpensively replaced with new buildings. As an alternative to building a new structure, existing building envelopes can be retrofitted with additional insulation and/or new fenestration, and HVAC systems can be replaced to make existing structures more energy efficient. Usually, the process of retrofitting a building requires specific design and 157 crease. Some manufactures have introduced new designs that encapsulate VIPs in Expanded Polystyrene foam (EPS) to protect the encapsulated panels against accidental damage during installation 1. This product and other similar wall systems employing these new VIP technologies are commercially available. Fenestration is an integrated component of the building envelope. There has been a recent worldwide trend to increase the area of fenestration in residential and commercial construction. Windows and skylights can contribute to a better standard of living and useful solar heat gains, but on the other hand, can cause higher heat losses or non-desirable heat gains and glare. Currently, the thermal transmittance of most fenestration products is still much higher than for the equivalent area of modern wall systems. Studies show that up to 60% of the total energy loss through the building envelope can be attributed to windows (Gustavsen et al., 2008a). With the current increase in the use of fenestration in modern building practices, the selection and proper installation of windows is critical to improving the overall performance of building envelopes. Recent improvements in the designs of walls and windows contribute to better thermal performance by addressing thermal bridging in the building envelope. The effects of thermal bridging can be significant, and are often difficult to detect in the design stage. Significant thermal bridging can occur at the areas near the edge of the window apertures in wall systems; the window-wall interface. A study by Gustavsen et al. (2008b) shows that for a typical 160 m2 dwelling, the window-wall interface can be responsible for 40% of the total heat loss in the structure. By focusing on reducing the thermal bridging in the same building, the window-wall interfaces were responsible for only 17% of the total heat loss through all of the areas of thermal bridging. Similar results were reported in ISO 14683:2007 (ISO, 2007c), which specifies methods to calculate the thermal performance of thermal bridges. Using this methodology, relatively low performing generic buildings are calculated to lose 36% of their total energy through thermal bridges in the building envelope, whereas the window-wall connection is responsible for 38% of that value. These studies demonstrate the significance of heat transfer through the window-wall interface area. Although the heat transfer through the windowwall interface can be significant, it is often not accounted for when determining the thermal performance of the building envelope. Given that the heat transfer through the wall system is typically represented by an overall R-Value for the opaque area (which may be area-weighted to include framing components), and the thermal transmittance of the window system is represented by a U-Factor for the entire window area, the “flanking” heat flow around the base of the thinner window frame through the thicker window aperture in the wall is not typically included in the heat transfer calculations. As many windows are designed and sold as individual products, their energy performance is typically reported by the manufacturer or the certification agency for the window alone as they do not know where or how the window will be installed. Window U-Factors are typically calculated assuming that there is no heat transfer between the window frame and the adjacent wall (adiabatic boundary condition). The apertures created in the walls for windows must support the weight and forces created by the installed window, and protect the building envelope from water intrusion and air leakage. Maref et al. (2011) reported that the lowest surface temperatures on the building envelope are located in the areas of the window-wall interface, which makes these locations more susceptible to condensation. For retrofit applications, the installed location of replacement windows in the window aperture will depend on the chosen strategy for retrofitting existing building walls. If the existing building envelope is retrofitted with convectional insulation, the walls will be relatively thick. If the edge of the aperture in the wall is much thicker than the window frame, there is the opportunity to install the window at a number of different locations in the window opening (i.e., the interior, exterior, or between the faces of the wall). If VIPs are used to retrofit the existing wall, the overall wall thickness will be much thinner. As VIPs are a new technology, the optimum location of installation, sealing, and means to insulate any fasteners need to be studied and documented. Window placement in wall systems retrofitted with VIP systems can be challenging as it is not easy to balance thermal performance, solar heat gain, daylight distribution, moisture resistance, durability, structural integrity, and architectural esthetics. This study focuses on investigating different solutions for improving the thermal performance of window-wall interface area in retrofit applications by reducing thermal bridging in the wall assembly near the window aperture. A number of computer simulations were performed using two-Dimensional heat flux computer software called, THERM 6.3. The effects of thermal bridging through the window-wall interface were investigated for two wall types, and a variety of window installation locations and methodologies (e.g., use of setting blocks and flashing). To compare these different scenarios, the Linear Thermal Transmittance, ψ, as determined by ISO 10211 is reported. 1 Dow Corning, a U.S. manufacturer of VIPs, offers panels at a variety of thicknesses (6.3–38 mm) and sizes (up to 600 x 900 mm). The thermal conductivity of 25 mm thick panel is reported to be 0.00213 W/(m2 ·K) (Dow Corning Corporation, 2012) 158 area for clay block walls. This study simulated the heat flow through wooden-framed windows installed in two different wall constructions (a brick wall insulated from the outside, and a brick wall containing an insulated cavity) at three different positions (outside, intermediate, and inside) for each wall design. For each case the Linear Thermal Transmittance based on external dimensions was calculated (in accordance with EN/ISO 10211:2007 (ISO, 2007b)). It was found that the window position, the installation details, and the framing of the window aperture in the wall have a significant impact on the Linear Thermal Transmittance, which differs by 70-75% between cases. In addition, the paper proposes a methodology to combine the heat transfer of the thermal bridging in the wall area adjacent to the window with the rated window U-Factor. 2 LITERATURE REVIEW The topic of the window-wall interface has been studied before in the literature, however the issue of thermal bridging on the connection between the window and the wall has yet to be widely explored. Experimental studies by Maref et al. (2012) and (2011) investigated the influence of air leakage through the window-wall interface with respect to the formation of condensation. Another study by Lacasse et al., (2009) concluded that the windowwall interface area is important from a water intrusion point of view. The study assessed the potential of water intrusion through the window-wall interface by laboratory testing. SINTEF, which is the largest non-commercial and independent research organization in Scandinavia, has been a major contributor to the research on thermal bridging effects in building envelops. In various reports and guidelines, the location of windows in window apertures is referred to as an important parameter for minimizing thermal bridges in buildings. SINTEF building research design guideline 471.015 (Gustavsen and Roald, 2008) gives an example of the relationships between the window position and the linear thermal transmittance, ψ, for a wood-framed wall with 250 mm of insulation. This study shows that installing the window 35 mm from the outside face of the wall is the most favorable in terms of reducing thermal bridging. The optimum window location according to other SINTEF reports differs from this recommendation. SINTEF building research design guideline 523.701 (SINTEF, 2003a) and 523.702 (SINTEF, 2003b) recommend two different locations to install windows, which depend on the climate. For buildings located in coastal Norwegian regions, the window should be installed flush or close to the outside face of the wall. Windows should be installed close to the interior face of the wall at those building locations further from the coast where the occurrence of wind driven-rain is infrequent. By recessing the window deep inside the aperture in the wall, the indoor surface temperatures on the window will be warmer, which lowers the risk of condensation. Decheva (2012) also concluded that the best location to install widows in the apertures of thick building envelopes (i.e., retrofit) to minimize thermal bridging is around 35mm from the exterior sheathing. Moreover, alternative solutions for window framing where proposed in this study, which substantially reduce the heat loss due to thermal bridging. However these proposed solutions do not directly apply to construction practices in the United States. Another study by Cappelletti et al. (2011) investigated the influence of window installation locations and construction details on the heat flow by thermal bridging through the window-wall interface 3 SCOPE AND LIMITATION OF WORK The objective of this study is to determine the optimum position to install windows in thick wall apertures to minimize the Linear Thermal Transmittance, ψ, around the windows for walls retrofitted with convectional insulation and VIPs. It should be noted that this study does not investigate the air leakage or water drainage of the modeled solutions. In addition, air flow velocities next to individual surfaces were not simulated in detail, and instead simplified heat transfer coefficients were used. Finally, this study only modeled the window sill and the bottom edge of the window opening, and those characteristics were used to represent the performance of the window jambs and head, and the wall areas around them. 4 NUMERICAL SIMULATIONS Numerous heat flow calculations were performed to evaluate the effects of thermal bridging at the window-wall interface. Heat flow calculations were carried out using the computer program THERM (version 6.3), developed by Lawrence Berkeley National Laboratory (LBNL) in California, USA. THERM is a two-dimensional finite element program that calculates conduction, convection and radiation heat exchange, and also estimates the heat exchange due to convection and radiation in air cavities (LBNL, 2011). This software is used to create a model representing the assembly of materials, and to conduct two-Dimensional heat transfer simulations of the window sill area, and the wall area below the window. Window frame geometry was prepared in accordance with NFRC 100 (NFRC, 2012a), and the overall geometry of the window-wall interface area was created in accordance with ISO 10211 (ISO, 2007b). 159 Two different retrofit wall systems are evaluated in this study. Both retrofit systems are added to the outside of a conventional 2x6 wood-framed wall system (actual width of lumber is 140mm) insulated between the studs with fiberglass batts. The “conventional” retrofit strategy consists of the addition of 220mm thick extruded polystyrene (XPS) to the outside face of the existing wall. The thickness of extruded polystyrene insulation for this retrofitted wall was determined by matching the overall thermal transmittance of the VIP retrofitted wall. The external sheathing was not modeled in this scenario, and therefore, the outside boundary conditions have been adjusted as described below. The “VIP” retrofit strategy uses Dow Corning Vacuum Insulated Panels that are encapsulated with a layer of expanded polystyrene foam. The wall employing VIPs is about 70% thinner than the wall with XPS insulation mainly due to the very low conductivity of VIPs. As with most computer modeling, the thermal effects of defects in the actual materials or imperfections in the installation are not taken into account. In addition, the vapor barrier was omitted from the computer simulations due to its low thermal resistance. The configurations of the retrofit walls analyzed in this report are presented and described in Table 1. All simulations were performed using the sill and lower sash of the Clara Starrett EnergyCore window manufactured by Mathews Brothers Company. This window is triple-glazed with two low emissivity coatings (low-e), and contains Duralite non-metallic insulating spacers (Mathews-BrothersCompany, 2011). Initially, the window was modeled using actual glazing and a simplified sill geometry according to NFRC 100: Procedure for Determining Fenestration Product U-factors (NFRC, 2012a). This simulation generated a good match between the rated U-Factor from the NFRC Certified Product Directory, and the computer modeling results. In addition, the thermal transmittance of the window is calculated using ISO 10077-2:2012 (ISO, 2012) so the results can be compared with other similar studies performed in Europe. Keep in mind that ISO 100772 simulates the thermal transmittance of the window frame/sash with the glazing replaced by the same thickness of foam insulation. The difference between the results generated by ISO 10077-2 and the UFactors calculated using NFRC 100 are within 2% (with the actual difference of 0.025 W/(m2K)). These differences do not have a significant effect on this study as all the simulations are performed using the same window. This project is a comparison of different configurations using similar building envelope components, rather than an effort to report the actual heat flux expected. The U-Factor of the window is 1.19 W/(m2K) as calculated using NFRC 100 and THERM 6.3. A diagram of the window and a copy of the THERM cross-section are presented in Figure 1. In additional simulations, the glazing unit was re- placed with an foam insulating panel with a thermal conductivity of 0.035 W/(mK) as specified in ISO 10077-2.. In order to justify the replacement of glazing with a foam panel, the influence of the spacer on Linear Thermal Transmittance, ψ, was investigated. Simulations show that including the spacer in the model can change the Linear Thermal Transmittance due to changes in the thermal conductance of the window. Once again, the differences between comparisons of ψ of different window locations with and without the spacers are negligible. Finally, the vertical nailing fin was removed so that the window could be installed at any location within the window aperture. The U-Factor of the sill and lower sash of the windows with an insulating panel replacing the glazing is 1.11 W/(m2K). The wall and window dimensions and material properties are shown in Table 2. All the simulations were performed using the wall area that is 1.2 m below the edge of the window aperture, and the window was represented by the window frame sill and lower sash with a foam panel installed in the sash instead of the Insulated Glazing unit. A diagram showing the configuration of the wall and window model is presented in Figure 2. The boundary conditions representing the surface heat transfer coefficients on the inside and outside surfaces of the window and the wall differ depending on the retrofit wall system. Except for the case described below, the simplified ISO 10077-2:2012 (ISO, 2012) boundary conditions were applied to all the window and wall surfaces using THERM. For the simulations of the wall retrofitted with conventional insulation, the external sheathing and ventilated air cavities were not included in the model of the wall. For those cases, the outside surface of the drywall was treated as the surface facing the ventilated air cavity. The surface heat transfer coefficients were determined for this cavity by ISO 6946:2007 (ISO, 2007a). ISO boundary conditions were used in this study to compare the results from this project with other studies by Cappelletti et al. (2011) and Decheva (2012), which also evaluated the windowwall interface. The boundary conditions for all of the simulations are described in Table 3. The Linear Thermal Transmittance, ψ, was calculated according to the formula (1). where, L2D - is the thermal coupling coefficient obtained from a two-dimensional calculation of the wall and window components separating the two areas, Ui - is the thermal transmittance of the onedimensional component separating the window-wall interface (see Figure 2), 160 li - is the length within the two-dimensional geometrical model over which the value of Ui applies (see Figure 2). lines of setting the slope of the flashing at five (horizontal) to one (vertical). − An additional piece of 19 mm thick wood is added on top of any exposed insulation. In most cases this layer of wood covers the entire width of the wall, but in one of the conventional insulation retrofits, this piece of wood only covers the additional insulation at a slope. This piece of wood is added to only one of the VIP retrofit wall configurations, and to all but one of the conventional retrofit wall systems. − In a few cases, an additional piece of wood, which is as wide as of the base of the window frame, is added below the window sill. This additional piece of wood is called the “setting block” in this paper. The thickness of the setting block depends on the application. This setting block is recommended by literature focusing on retrofitting strategies for conventional insulation (Ueno 2010), and commercial solutions proposed by STO Corporation (STO 2005). Based on the guidelines in those references, setting blocks are used in 3 VIP retrofit wall systems, and 3 conventional retrofit wall systems. − Although the sill may be cut at a slope to accommodate the flashing, the thin metal or vinyl flashing was not included in the simulation model. The thermal effects of metal flashing were initially modeled on many of the wall configurations in this study. Although metal flashing can affect the local surface temperatures, it has a negligible effect on the Linear Thermal Transmittances for the configurations considered in this project. Vinyl flashing could also be used in these wall systems, and has even less thermal impact. 5 RESULTS The first set of simulations calculated the thermal performance of 14 different window installation locations in the window apertures of both of the retrofit wall systems. The intent of this investigation was theoretical, and therefore, setting blocks and flashing were not modeled. The window aperture in both of the retrofit wall systems was framed using identical 19 mm thick lumber. The location of the window is measured from the outside face of the external wall/drywall to the outside edge of the window sill, b, which is measured in mille-meters (See Figure 3). The results show that the location of the window in the aperture has significant influence on the Linear Thermal Transmittance, ψ (See Figure 3). There were significant variations in the ranges of ψ for both wall systems; however, the wall retrofitted with conventional insulation was not as sensitive to the installed location of the window. The optimal window location in terms of thermal performance was different for both wall systems. For both cases, the highest ψ (poorest thermal performance) are reported when the window is installed flush with indoor wall. It appears that for the wall retrofitted with conventional insulation, the optimal position to install the window is in the broad range between 35 and 340 mm from the outside face of the external gypsum board. For wall system retrofitted with VIPs, the most thermally efficient location to install the window is flush with the outside face of the wall, where b = 0 mm. At that position, the window is almost centered above the encapsulated VIP. Moving the window from this position toward the interior of the wall significantly increases the heat flow through the window-wall interface. The ψ increases around 50% when b = 80 mm, and 104% when b = 120 mm. This initial set of simulations identified the optimum location to install a window from a theoretical point of view. Most of the window locations considered would never be actually installed as modeled as they could not be sealed against water intrusion, and have inadequate water drainage. The next step of the study is to explore window installation locations taking into account published guidance regarding window installation and window flashing requirements. Based on a literature review, the following rules and assumptions were made: − The slope for all of the external window flashing is at least 5:1. Different sources, (ASTM, 2001), (SINTEF, 2003b) and (SINTEF, 2003a), did not agree on the optimum slope of exterior flashing. It was decided to follow the current SINTEF guide- Table 4 presents the window-wall interface configurations and the ψ results from the second set of simulations. Some of these models include a sloped sill to accommodate the flashing and setting block installations discussed in the previous assumptions. For those cases representing walls retrofitted with VIP wall systems, the most energy efficient location to install the window is flush with the outer face of the wall, where b = 0 mm. Results shows that installing the window flush with external face of the wall reduces the ψ by 80% compared to all of the other locations investigated. The window-wall configuration in this scenario is the same for the first set of theoretical simulations as for the second set of more realistic simulations. Installing the window at other locations either required a setting block, or the top of the expanded polystyrene insulation was cut at a slope to accommodate the 5:1 slope of the flashing. For cases marked with the letter “A,” a setting block was required for proper installation. The setting block was 12 mm thick for the location where b = 85 mm, and 25 mm thick for the location where b = 160 mm. In those cases marked with letter “B” the slope 161 less than 1% based on THERM documentation (LBNL, 2011). of the flashing is created by cutting the insulation at an angle. For a given configuration, cutting the exterior wall insulation at a slope is more energy efficient than adding a setting block. Those simulations where the wall is retrofitted with conventional insulation are also shown in Table 4. In those cases identified with the letter “C,” setting blocks were placed under the windows. At these locations, where b = 35 mm, 55 mm and 85 mm, setting blocks with a thickness of 6 mm, 12 mm and 19 mm, respectively, were added. Cases marked with letter, “D,” employ additional sloped insulation in front of the setting block under the location of the flashing. Windows installed at locations closer to the inside face of the wall system require thicker setting blocks, which contribute to additional heat loses. For this reason, windows installed within 55 mm from the outside face of the wall achieve the lowest Linear Thermal Transmittance, ψ. Similar results were also reported by Decheva (2012). Adding additional insulation under the location of the flashing slightly lowers the ψ for each of these cases. The final comparisons are simulated with the window on top of the conventional insulation (b = 440 mm). This configuration required that the exterior retrofit extruded polystyrene insulation (XPS) was cut at the 5:1 slope for the flashing. The configuration labeled, “E,” does not have a piece of wood on top of the sloped XPS, whereas the case labeled, “F,” represents the same configuration with a piece of 19 mm thick wood on top of the XPS, The latter configuration is based on solution presented by Ueno (2010). These cases where the window is installed 440 mm from the outside face of the wall have a 40% higher ψ than when the window is installed 35 mm from the exterior. The optimal position to install a window in the window aperture of a VIP retrofitted wall system generates lower Linear Thermal Transmittances (around 30%) than for the cases where the wall is retrofitted with conventional insulation. This is due to the thinner cross-section of VIP systems, which was also a conclusion reported by Decheva (2012). 7 CONCLUSIONS A well-insulated building envelope is essential for thermal comfort and energy savings. To meet these objectives, the individual building components and the connections between different materials need to be carefully designed. The window-wall interface is a significant component in the overall building envelope energy balance. This study shows that the location of the window in the window aperture is a complex subject. There are many factors and conditions related to the installation of windows, and the framing of the window aperture in the wall that need to be taken into account when selecting the best solution. The current study is focused mainly on thermal aspects of the window-wall interface. Results show that the window location in the aperture of a wall has a significant influence on the Linear Thermal Transmittance, and therefore, the overall heat flow through the entire system. If water drainage and installation details are not considered the most energy efficient window position is roughly in the middle of retrofit insulation (b = 0 mm for the wall retrofitted with VIPs, and b = 160 mm for the wall retrofitted with conventional insulation). When reviewing the results for those configurations that have a 5:1 sloped sill to accommodate flashing, the optimal location to install windows in walls retrofitted with VIPs were found to be flush with the outside face of the wall (b = 0 mm). The optimum location to install a window in the window aperture of a wall retrofitted with conventional insulation is within 55 mm from the exterior face of the wall. Although the heat transfer through the window-wall area in a conventionally retrofitted wall system is not as sensitive to the location of the window in the wall aperture, walls retrofitted with VIP systems have a 30% lower Linear Thermal Transmittance when comparing the optimal window installation locations. 6 UNCERTAINTIES A number of simplifications were made in the development of the computer model and assumed material properties that may have influenced the accuracy of the results. For starters, simplified boundary conditions were used. In addition, for the purpose of modeling, several intermittent layers were assumed continuous and simplified. Finally, other thermally insignificant components were not included such as the window flashing and the vapor barrier. The Energy Error Norm for all simulations was kept around 6% which yields U-Factor uncertainty of 8 FUTURE WORK The work presented in this paper is a good starting point for further investigation of window–wall interface properties. Further work might focus on: − Model all sides of the window-wall interface including the window jambs and the head, − Model the window-wall interface in threeDimensions to capture influence of the corners of the window and framing of the window aperture in the wall. 162 Condensation on Windows. Journal of Testing and Evaluation, 39 (4), 562-575. Maref W., Van Den Bossche N., Armstrong M., Lacasse M. A., Elmahdy H., and Glazer R. 2012. Condensation risk assessment on box windows: the effect of the window-wall interface. Journal of Building Physics, 36 (1), 35-56. Mathews-Brothers-Company 2011. Clara Starrett - New construction & replacement windows. Mathews Brothers Company. NFRC 2012a. Procedure for Determining Fenestration Product U-factors - 100. National Fenestration Rating Council NFRC 2012b. Procedure for Determining Thermophysical Properties of Materials For Use in NFRC-Approved Software. National Fenestration Rating Council. SINTEF 2003a. 523.701 - Innsetting av vindu i vegger av bindingsverk. (Installation of windows in wooden frame walls.), Olso. SINTEF 2003b. 523.702 - Innsetting av vindu i mur- og betongvegger. (Installation of windows in brick and concrete walls.), Olso. Ueno K. 2010. Residential Exterior Wall Superinsulation Retrofit Details and Analysis. Building Science Press. Voss K., and Musall E. 2011. Towards Net Zero Energy Solar Buildings. International Energy Agency. 9 REFERENCES ASHRAE, 2009. ASHRAE Handbook - Fundamentals (I-P Edition). American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. ASTM, 2001. Standard Practice for Installation of Exterior Windows, Doors and Skylights. Cappelletti F., Gasparella A., Romagnoni P., and Baggio P. 2011. Analysis of the influence of installation thermal bridges on windows performance: The case of clay block walls. Energy and Buildings, 43 (6), 1435-1442. Decheva L. M. 2012. Vinduer for energieffektive bygninger Kuldebroer ved vindusinnsetting, "Windows for energy efficient buidings - thermal briding effect on window-wall interface" Norway, NTNU. Dow-Corning-Corporation. 2012. Dow-Corning® Vacuum Insulation Panel. Erb M., Heinemann U., Schwab H., Simmler H., Brunner S., Ghazi K., Bundi R., Kumaran K., Mukhopadhyaya P., Quénard D., Sallée H., Noller K., Kücükpinar-Niarchos E., Stramm C., Tenpierik M., Cauberg H., Binz A., Steinke G., and Moosmann A. 2005. Vacuum Insulation Panel Properties and Building Applications. Gustavsen A., Arasteh D., Jelle B. P., Curcija C., and Kohler C. 2008a. Developing Low-Conductance Window Frames: Capabilities and Limitations of Current Window Heat Transfer Design Tools - State-of-the-Art Review. Journal of Building Physics, 32 (2), 131-153. Gustavsen A., and Roald B. K. 2008. Byggforskserien, Byggdetaljer 471.015 / Kuldebroer. Konsekvenser og dokumentasjon av energibruk. (Thermal bridges. Impact and documentation on energy usage.). Oslo, SINTEF Byggforsk. Gustavsen A., Thue J. V., Blom P., Dalehaug A., Aurlien T., Grynning S., and Uvsløkk S. 2008b. Kuldebroer : beregning, kuldebroverdier og innvirkning på energibruk. (Thermal bridges: Calculation, thermal bridge values and impact on energy use.). Oslo, SINTEF Byggforsk. IEA, 2011. Retrofit Strategies Design Guide Advanced Retrofit Strategies &10 Steps to a Prefab Module. International Energy Agency Energy Conservation in Buildings and Community Systems Programme. ISO 2006. ISO 10077-1:2006 - Thermal Performance of Windows, Doors and Shutters - Calculation of Thermal Transmittance - Part 1: General. International Organization for Standardization. ISO 2007b. ISO 10211:2007 - Thermal bridges in building construction - Heat flows and surface temperatures Detailed calculations. International Organization for Standardization ISO 2007c. ISO 14683:2007 - Thermal bridges in building construction - Linear thermal transmittance - Simplified methods and default values. International Organization for Standardization ISO, 2012. ISO 10077-2:2012 - Thermal performance of windows, doors and shutters - Calculation of thermal transmittance - Part 2: Numerical method for frames. Lacasse M. A., Armstrong M., Ganapathy G., Rousseau M., Cornick S. M., Bibee D., Shuler D., and Hoffee A. 2009. Results on Assessing the Effectiveness of Wall-Window Interface Details to Manage Rainwater. JOURNAL OF ASTM INTERNATIONAL, 6 (9). LBNL, L. B. N. L. 2011. THERM 6.3 / WINDOW 6.3 NFRC Simulation Manual. Windows and Daylighting Group, USA. Maref W., Van de Bossche N., Armstrong M., Lacasse M. A., Elmahdy H., and Glazer R. 2011. Laboratory Tests of Window-Wall Interface Details to Evaluate the Risk of 10 FIGURES AND TABLES Table 1. Materials, Dimensions and Properties of Wall Systems. Table 2. Properties of Wall System Materials. 163 Table 3. Boundary Conditions Used for the Simulations. Table 4. Simulation Results. Figure 2. THERM Model of the Window and VIP Wall System (Not to Scale). Figure 3. Linear Thermal Transmittance vs. Window Location in the Window Aperture for Both Retrofitting Wall Systems. Figure 1. Left: Cut-Away View of Mathews Brothers Window. (Mathews-Brothers-Company, 2011). Right: THERM Model of the Mathews Brothers Window. 164 Effect of various Nordic climates on attaining the Passive House standard L. Gullbrekken SINTEF Building and Infrastructure, Trondheim, Norway E.H. Nilsen, T. Kvande, S. Geving Norwegian University of Science and Technology (NTNU), Trondheim, Norway ABSTRACT: The main aim for this work is to show how three different prefabricated single family houses from a Norwegian house manufacturer can be adapted to the Norwegian Passive house standard (NS 3700:2010) for various Norwegian climates. The effect of orientation of the main façade (largest window area) in regard to solar gain is given extra focus. Energy calculations are conducted. Five different geographical locations of houses in Norway are studied to evaluate how the outdoor climate affects the energy performance of the building. The effects on the energy performance caused by different sun-shading devices and orientations of the houses are also studied. The results show that the heating demand is considerably affected by the particular climate. The two factors sun-radiance and outdoor temperature relate directly to energy demand. Orientation of the houses in regard to solar gain has a small affect on the heating demand. Table 1. Annual temperature and insolation from the sun for the five geographical locations. City Annual Total average Clitemperature insolation from mate ϴm [°C] the sun [kWh/m²year] Towards south, 30° slope Oslo 6,1 1149 D Bergen 7,8 958 C Trondheim 5,8 1080 D Tromsø 2,7 859 D Karasjok -2,1 859 D 1 INTRODUCTION The main aim for this work is to show how three different prefabricated single family houses from a Norwegian house manufacturer can be adapted to the Norwegian Passive house standard (NS 3700 2010) for various Norwegian climates. To be able to evaluate the robustness of the manufacturers building system regarding energy use, heating demand for buildings in five different geographical locations are calculated. The geographical locations are given in Figure 1 and their local climate is briefly described in Table 1. The effect of orientation of the main façade in regard to solar gain is given extra focus (largest window area facing the sun). According to Table 1 Bergen has the highest annual temperature of the five geographical locations while Oslo has the highest insolation value. Karasjok has the lowest annual temperature and the lowest insolation value. The climate given in Table 1 is based on Köppen Climate Classification System (developed by Wladimir Peter Köppen around 1900, with several later modifications). Climate "C" represents Moist Mid-latitude Climate with mild winters while "D" represents Moist Mid-latitude clilmatees with cold winters (Lisø 2006). Tromsø Karasjok Trondheim 2 METHOD Bergen Oslo Energy calculations are performed according to NS 3031 2007 and the energy performance is calculated stationary giving monthly values according to NSEN ISO 13790 2008. Software SIMIEN 4.505 is used for the calculations. The software is validated according to NS-EN 15265 2007 (class B). Figure 1. Geographical location of the five Norwegian cities included in the study. 165 Table 2. House information – Sans. The following parameters are varied in the study: Ground floor area, LxB Heated floor area Heated volume Area of windows and exterior doors related to heated floor area (Windows area facing south (Windows area facing north (Windows area facing west (Windows area facing east − Geographical location: The climate conditions affect the energy use for heating. Construction of passive houses in the north of Norway is more demanding due to colder climate and less insolation from the sun in the winter time. The five geographical locations of the houses were selected in order to span most of the relevant building sites in Norway Unit m m² m³ % % % % % Value 10,6 x 8,5 143,8 340,8 21 11) 4) 3) 3) − Orientation of the building: The possibility for insolation from the sun depend of the area of windows facing the sun. However, the façade with the largest window area may not be facing the sun due to the local orientation of the building. Energy use are calculated for orientation of the largest window area of the houses facing to the south, west, north and east respectively. Figure 2. Sans (Nordbohus 2013) 3 CASES 3.3 Orion "Orion" is a two story large house with room height 2.4m in both floors, see Figure 3 and Table 3. The ceiling is horizontal and the pitched wooden roof with cold attic is made of W-roof truss. Orion is the biggest house in our study. Area of exterior doors and windows for the different façades are given for south orientation of the house, see Table 3. 3.1 Nordbohus The Norwegian second most successful home builder manufacturer, Nordbohus, markets more than 50 different high developed standard houses. Three of the standard houses were selected for this study in order to give a picture of witch type of houses being suitable for passive house construction. The houses have different size, shape and structure, representing challenges for the manufacture designing passive houses. In the following are the houses identified by there marketing name used in advertising of the houses. Table 3. House information – Orion. Ground floor area, LxB Heated floor area Heated volume Area of windows and exterior doors related to heated floor area (Windows area facing south (Windows area facing north (Windows area facing west (Windows area facing east 3.2 Sans The standard house named "Sans" is a one and a half story large house, see Figure 2 and Table 2. The room height of the ground floor is 2.4m while the attic has a sloping ceiling. Area of exterior doors and windows for the different façades are given for south orientation of the house. Figure 3. Orion (Nordbohus 2013). 166 Unit m m² m³ % % % % % Value 12,4 x 8,8 200,6 482 17 8) 6) 0) 3) 3.4 Arcturus 4 RESULTS The standard house named "Arcturus" is a one floor house favourable for people with reduced functional ability. Arcturus is the smallest of the houses in the study. See Figure 4 and Table 4. Area of exterior doors and windows for the different façades are given for south orientation of the house, see Table 4. 4.1 Heating demand for Sans Figure 5 gives the heating demand calculated for Sans in the five different geographical locations and compared to the Norwegian Passive House standard (NS 3700 2010). The heating demand vary very little regarding to orientation. Sans fulfils the requirements of NS 3700 2010 for all of the five cities and for all orientations of the house. Table 4. House information – Arcturus. Heated floor area (BRA) Heated volume Area of windows and exterior doors related to heated floor area (Windows area facing south (Windows area facing north (Windows area facing west (Windows area facing east Enhet m² m³ % Verdi 129 296 22 % % % % 8) 1) 4) 9) Figure 5. Heating demand for Sans in different cities and orientations compared with the requirements of NS 3700 2010. 4.2 Heating demand for Orion Figure 6 gives the heating demand calculated for Orion in the five different geographical locations and compared to the Norwegian Passive House standard (NS 3700 2010). The heating demand vary very little regarding to orientation. Orion fulfils the requirements of NS 3700 2010 for all of the five cities except Karasjok. Figure 4. Arcturus (Nordbohus 2013). 3.5 Building system The component performance level of the building system is given in Table 5. Most of the design values used in the calculations are lower than the minimum requirements in NS 3700 2010. This is necessary in order to fulfill the requirements for energy use for heating for different types of houses located all over in Norway. Table 5. Values used in the calculations compared minimum demands in NS 3700 2010. Unit Used values U-value exterior wall W/m²K 0,10 U-value roof W/m²K 0,09 U-value floor W/m²K 0,09 U-value window W/m²K 0,70 Normalized thermal W/m²K 0,02 bridge value Efficiency of the heat % 81 exchanger SFP-factor kW/(m³/s) 1,5 Air change rate at h-1 0,6 50Pa, n50 with the NS3700 ≤0,15 ≤0,13 ≤0,15 ≤0,80 ≤0,03 Figure 6. Heating demand for Orion in different cities and orientations compared with the requirements of NS 3700:2010. 4.3 Heating demand for Arcturus ≥80 Figure 7 gives the heating demand calculated for Arcturus in the five different geographical locations and compared to the Norwegian Passive House standard (NS 3700 2010). ≤1,5 ≤0,6 167 7 REFERENCES The heating demand vary very little regarding to orientation. Arcturus fulfils the requirements in NS 3700 2010 for all of the five cities except Karasjok and for North and East orientation of the largest window area façade in Tromsø and Oslo. Byggforskserien 451.021 2009. Klimadata for termisk dimensjonering og frostsikring. Oslo: SINTEF Byggforsk. Byggforskserien 472.411 1991. Solstrålingsdata for energi og effektberegninger. Oslo: SINTEF Byggforsk. Lisø K.R. 2006. Building envelope performance assessments in harsh climates: Methods for geographically dependent desig. Doctoral Theses at NTNU, 2006:185. Trondheim: NTNU. Nordbohus 2013. http://www.nordbohus.no/orion, www.nordbohus.no/sans, www.nordbohus.no/arcturus, downloaded 08.02.2013. NS 3031 2007. Calculation of energy performance of buildings. Method and data. Oslo: Standard Norge. NS 3700 2010. Criteria for passive houses and low energy houses. Residential buildings. Oslo: Standard Norge. NS-EN ISO 13790 2008. Energy performance of buildings Calculation of energy use for space heating and cooling. Oslo: Standard Norge. NS-EN 15265 2007. Energy performance of buildings - Calculation of energy needs for space heating and cooling using dynamic methods - General criteria and validation procedures. Oslo: Standard Norge. Figure 7. Heating demand for Arcturus in different cities and orientations compared with the requirements of NS 3700 2010. 5 DISCUSSION Geographical location and thereby the climate directly influence the heating demand. As Table 1 shows, Bergen has the highest annual temperature. The consequence is given in Figure 5, 6 and 7 showing Bergen with the lowest calculated heating demand. The annual temperature and insolation in Oslo and Trondheim are about equal. The heating demands in these cities are therefore about equal. Karasjok has the lowest annual temperature at -2,1 °C and the lowest insolation value giving the highest calculated heating demand for all the houses. Arcturus is the smallest house. This gives the highest calculated heating demand normalized by the floor area. This is expected because a smaller house has a larger envelope area compared to the heated available area. Sans gives the smallest calculated heating demand due to a more favourable quadratic base area. 6 CONCLUSIONS The three different houses from Nordbohus fulfil the Norwegian Passive House Standard (NS 3700 2010) for moderate mild climates. Orientation of the houses in regard to solar gain has a small affect on the heating demand. Geographical location and thereby the climate directly influence the heating demand. 168 Study of hygrothermal behavior of advanced masonry components made with utilization of secondary row materials J. Zach, J. Hroudová & M. Sedlmajer Brno University of Technology, Faculty of Civil Engineering, Institute of Technology of Building Materials and Components, Brno, Czech Republic A. Korjenic Vienna University of Technology, Institute of Building Construction and Technology, Vienna, Austria ABSTRACT: Performance requirements for packaging structures in terms of thermal protection of buildings are constantly increasing which is related to the increasing demands on the properties of masonry materials. Development of new, thermally-insulating ceramic fittings for circuit design is a rather complicated process. One of the options of improvement of thermal insulating properties of ceramic fittings is to reduce the thickness of the inner ribs and change their geometric arrangement of shapes or sizes increase and thus the width of masonry construction. Alternative options are insulating fittings with very good insulation properties which will be integrated in the cavities shaped padding. Ceramic skeleton provides mechanical stability and fittings and integrated insulation layer of smaller or larger part in its share and possibly thermal acoustic insulation properties. This paper describes the study of hygrothermal behavior of ceramic masonry blocks developed using an integrated insulating panels on the basis of waste textile fibers. 1 INTRODUCTION of the air hollows and the volume of the ceramic body, as well as on the form and the size of the hollows, their arrangement and the thickness of the internal and peripheral walls of the masonry blocks. The trend in recent years has been to integrate insulating materials with a low thermal conductivity into the hollows of the ceramic shaped unit. This material must be easily applied into the shaped piece; it should have the stated minimum strength due to ensuring the spatial rigidity of the resulting masonry blocks and should not significantly increase the final price of the shaped unit. (Zach & Brožovský & Hroudová 2009) The paper addresses the development of ceramic walling shaped pieces for envelope walling with integrated insulating materials on the basis of textile fibres. At present, the higher thermal technical requirements are permanently stated for building constructions which are closely related to the thermal technical requirements of building materials and the elements from which they consist. In recent years, the aim of construction material manufacturers designated for peripheral constructions has been to develop elements with strong thermal insulation properties which would fully replace the use of additional thermal insulation. In general, for ceramic shaped units, there are the following possible variations for improving thermal insulating properties: a) decrease of the thermal conductivity for ceramic chips; b) proposal for the optimal geometry of the internal arrangement of the masonry blocks; c) application of thermal insulation materials with a low thermal conductivity value into the hollows of the masonry block. The decrease of the thermal conductivity of the ceramic chip is related to the quality of input raw material and the possibility of lightening. In the case of a decrease in density which leads to the decrease in the thermal conductivity, it is necessary to take into consideration the worsening of the mechanical properties and the increase of sensitivity during drying. Therefore, at present most manufacturers of ceramic products are seeking other alternative ways to increase thermal resistance for ceramic shaped pieces, see below. Up to now, the most effective manner of increasing thermal technical properties for ceramic masonry blocks was considered to be the lightening of masonry blocks using air hollows. The efficiency of this lightening depends on the ratio of the volume 2 INTEGRATION OF THERMAL INSULATION MATERIALS IN MASONRY BRICKS As previously mentioned, the most progressive way is to integrate suitable heat insulating material into the hollows of ceramic shaped pieces. This manner of improving thermal insulating properties for walling shaped pieces for the envelope walls may significantly decrease thermal conductivity in the hollows of ceramic shaped pieces and fully eliminate convection inside the shaped pieces. Insulating materials for filling the hollows of ceramic shaped pieces must: - have the lowest thermal conductivity (must always be lower than the thermal conductance of air in the masonry gaps, which must be filled by the insulator), 169 - must be easily applied into the hollows of shaped units. In general, two types of insulators are used: a) piece insulations which, after pressing, are inserted into the hollows of shaped pieces and by consequent relaxation are fixed inside the hollows (these insulators can only be inserted into large sized hollows), b) particular insulations which can be filled into hollows and consequently secured by sealing, thermal fixation, etc. - determination of the thermal conductivity in a steady status according to ČSN 707012, EN 12667, ISO 8301 (the measurement was taken at medium temperature of +10 °C and at temperature difference of 10 K on dried testing samples). - determination of sound absorption according to ČSN ISO 10534-1, and the measured values were evaluated according to EN ISO 11654. The results of the basic physical technical measurements are contained in Table No. 1. Table 1. Thickness and density of sets of samples. Set of testing Thickness [mm] Density [kg.m-3] samples A 63.3 31.9 B 77.1 41.4 C 47.4 64.5 The results of the measurement of thermal and acoustic insulating properties are contained in Table No. 2. Figure 1. Textile waste. 3 THERMAL INSULATION MATERIALS BASED ON TEXTILE WASTE FIBRES Table 2. Overview of average values of thermal and acoustic properties. Set of testing Thermal conductivity Weight value of samples in dry state sound absorption [W.m-1.K-1] coefficient [-] A 0.90 0.0346 B 0.90 0.0355 C 0.85 0.0357 During the research and the development of insulating materials on the basis of alternative raw material resources, insulating materials on the basis of textile fibres were selected as suitable materials for filling the hollows of shaped pieces. These insulating materials were produced from fiberized textile consisting of a minimum 95% ratio of cotton. Textile fibres were thermally bound by bicomponent fibres consisting of a core with hightemperature polystyrene and coated with lowtemperature polystyrene. The ratio of input raw materials mixture from textile fibres: bi-component fibres were 85:15. It is evident from the measured values that the test samples report very favourable thermalinsulation properties. The very high value of sound absorption which may positively improve the value of the air sound-transmission loss of the resulting partitioning construction should also be mentioned. Moreover, the test samples were the subject of the determination of thermal insulating properties. In particular, this concerned the determination of: - dependence of the thermal conductivity on temperature, - dependence of the thermal conductivity on moisture. During the development of insulating materials on the basis of alternative raw material sources, three types of samples were created on the basis of different density differing by the density. These were the testing sets: - A ... 30 kg.m-3, - B ... 40 kg.m-3, - C ... 60 kg.m-3. Thermal insulating properties are defined by the thermal conductivity λ [W.m-1.K-1]. The measurement was taken using the plate stationary method according to ČSN 72 7012-3 (ISO 8301) on the Lambda 2300 device at the thermal gradient 10 K. The measurement of the dependence of the thermal conductivity on temperature was conducted at laboratory humidity at medium temperatures -10 °C, 0 °C, +10 °C, +20 °C, +30 °C and +40 °C. The results are contained in Table 3. The key properties were determined on the prepared testing samples of insulating materials on the basis of textile fibres: - determination of the thickness according to EN 823 (measurement was taken at the nominal down pressure of 50 Pa), - determination of linear dimensions according to EN 12085, - determination of density EN 1602, 170 Set of testing samples Table 3. Dependence of thermal conductivity on temperature. Thermal conductivity λ [W.m-1.K-1] -10 °C 0 °C 10 °C 20 °C 30 °C Table 5. Dependence of water content in mass on relative humidity. 40 °C Water content [%] Sample 0% 33 % 55 % 80 % A 0.0319 0.0331 0.0345 0.0358 0.0371 0.0382 A 0 2.68 3.72 8.44 B 0.0330 0.0345 0.0365 0.0388 0.0396 0.0404 B 0 3.11 4.06 8.13 C 0.0333 0.0346 0.0365 0.0393 0.0407 0.0414 C 0 2.94 3.91 8.78 Measurement of the dependences of the thermal conductivity on humidity was conducted at a medium temperature of 23 °C on fully dry samples and on samples conditioned in the environment with relative humidity of 33%, 55% and 80%. The results of this measurement are contained in Table 4: Table 4. Dependence of thermal conductivity on relative humidity. Thermal conductivity λ [W.m-1.K-1] Sample 0% 33 % 55 % 80 % A 0.0346 0.0349 0.0357 0.0462 B 0.0355 0.0357 0.0367 0.0469 C 0.0357 0.0358 0.0368 0.0505 Figure 3. Dependence of the mass humidity of the test samples on the relative humidity of environments at the temperature 23 °C. The results of the experimental measurements served as the source materials for the simulation of design simulations during the development of new ceramic blocks with integrated insulating materials on the basis of waste textile fibres. 4 DEVELOPMENT OF NEW NONTRADITIONAL BRICK BLOCKS For the design simulations in the WUFI design program, a shaped walling piece was selected with a thickness of 500 mm with 13 hollows which were filled with the developed insulating materials on the basis of waste textile fibres. For this paper, the simulation of the behaviour of the shaped piece with textile insulation from test set C was conducted. The measured data was entered into the material database in the design program. Figure 2. Graph - Dependence of thermal conductivity on relative humidity. In the next step, the sorption properties for the test samples were determined. The sorption properties for the material were stated together with the determination of the dependence of the thermal insulating properties on relative humidity on the basis of the change of weight in the investigated samples in individual environments (0%, 33%, 55% and 80% relative humidity). The results are contained in Table 5 and Figure 3. Figure 4. Display of the detail of the shaped piece of the design calculation during the application of the developed heat insulations on the basis of textile fibres into the hollows of the shaped piece. 171 First of all, the geometry of the shaped piece in the simplified form was modelled. Then, the density of the points network was set in which individual properties will be monitored. In the following step, suitable materials were taken from the material database. The ceramic chip was identified as Wienerberger Solid Brick and the hollows were filled as mentioned above, with the developed textile insulation. In Figure 5 there is a simplified model for the design simulation of hygrothermal behaviour of the investigated shaped piece. construction of the shaped piece; the average value of the moisture content in the monitored area was approximately 0.8 hm. %. Figure 6. Profile of temperature in the shaped piece on 12.12.2010. In Figure 6 it is possible to monitor the course through the shaped unit, from the side of the exterior value which varied within the interval from -11 °C to -5.1 °C, from the side of the interior temperatures which were in the interval 18.5 °C to 24.4 °C. Figure 5. Proposed geometry of the shaped piece by individual layers (1-textile insulation). For the further procedure, input and climatic conditions were stated. Hydrothermal behaviour was monitored for the area of Vienna. The area of the shaped piece located in the exterior was proposed for the most loaded area – West. On the part of the interior, the procedure according to EN 13788 was used where Humidity Class 2 was proposed. Assumed characteristics for the exterior: - Heat transfer Coefficient: 25 W.m-2.K-1 - Short-Wave Radiation Absorptivity: 0.68 - Long-Wave Radiation Emissivity: 0.9 Assumed characteristics for the interior: - Heat transfer Coefficient: 8 W.m-2.K-1 Figure 7. Relative humidity in the shaped piece on 12.12.2010 at 12:00. In Figure 7 it is possible to monitor the relevant humidity from the side of the exterior values varied in the interval from 70% to 80%; from the interior side, the relative humidity values were in the interval 40% to 50%. Design simulation was conducted for the monitored period from 1. 10. 2010 to 13. 1. 2013. In Figure 6 is the thermal course on 12. 12. 2010 at 12:00. In Figure 8 there are isothermal curves on 12. 1. 2013. A further objective was to monitor equilibrium moisture content of textile insulation inside the ceramics block. In Figure 9 is the monitored area of insulation; the content of humidity in mass percentages in insulation on the basis of textile fibres is displayed in Figure 10. It is possible to observe the cyclic course of the moisture in the 172 Figure 8. Isometric curves in the developed shaped piece on 12. 1. 2013. Figure 10. Profile of the moisture content [weight %] in the monitored area of the hollow with textile insulation in the period from 1. 10. 2010 to 13. 1. 2013. 5 CONCLUSION The development of thermal insulating materials from alternative raw material sources contributed via several materials for their application as insulation recommended for application into the hollows of ceramic shaped pieces. During the determination of heat insulating properties and acoustic properties these materials reported excellent properties and represent suitable alternative materials for currently commonly used insulating materials which in most frequent cases are expanded polystyrene, mineral wool, etc. For selection of a suitable insulator into the hollows of a selected shaped piece, the set of test samples C was selected with the density of 60 kg.m-3; the thermal conductivity on the dried status was 0.0357 W.m-1.K-1. For the simulation of heated moisture behaviour in the developed shaped piece, the WUFI 2D design program was selected. On the basis of the results of these simulations, it can be stated that this group of materials represents a good alternative to the currently used fillings based on mineral wool. However, the textile insulations represent a more ecological variation with the use of waste fibres. Figure 9. Monitored area of the hollow with textile insulation in terms of moisture content. 173 6 ACKNOWLEDGEMENTS This paper was elaborated with the financial support of the project MPO FR-TI3/231 and GA 13-21791S. 7 REFERENCES Abdul Kadir A., Mohajerani A., Roddick F., and Buckeridge J. 2009. Density, strength, thermal conductivity and leachate characteristics of light-weight fired clay bricks incorporating cigarette butts, Proceedings of the World Academy of Science, Engineering and Technology. EN 12667 Thermal performance of building materials and products (2001). Determination of thermal resistance by means of guarded hot plate and heat flow meter methods. Products of high and medium thermal resistance, Brussels. EN 1602 Thermal insulating products for building applications (1997). Determination of the apparent density, Brussels. EN 823 Thermal insulating products for building applications (1995). Determination of thickness, Brussels. EN 1745 Masonry and masonry products (2002). Methods for determining design thermal values, European committee for standardization, Brussels. Kazragis A., and Gailius A. 2006. Composite materials and product containing natural organic aggregates. Vilnius: Technika. Vėjelienė J., Gailius A., and Vėjelis S. 2010. Analysis of Thermal Insulation from Renewable Resources, Engineering structures and technologies. Zach J., Brožovský J., and Hroudová J. 2009. Investigation of anti-noise absorptions walls development with utilization of waste from agriculture, in IABSE SYMPOSIUM BANGKOK, Switzerland. Zach J., and Hroudova J. 2009. Thermal technical properties of insulation materials made from easy renewable row sources. Journal Intersections. 174 Applicability of capillary condensation sorption hysteresis model for burnt clay bricks P. Matiasovsky & P. Mihalka Institute of Construction and Architecture, Slovak Academy of Sciences, Bratislava, Slovakia ABSTRACT: The analysis of results of the water vapour and nitrogen sorption for the lightweight plasters and calcium silicate boards proved the compatibility of the nitrogen sorption in water vapour sorption modelling and the possibility to identify the pore structure and to model the sorption hysteresis with the use of Barret, Joyner and Halenda (1951) model of capillary condensation. In this paper the validity of capillary condensation sorption model and the equivalence of its application for nitrogen and water vapour sorption of burnt clay bricks are proved. For four different burnt clay bricks the sorption model based on the application of Halsey equation was applied. The following criteria of the Halsey equation validity were used: the similarity of the specific surface areas and the similarity of the adsorbate volumes at 95 % relative pressure for the analysed sorbates. The relationship between sorption and desorption relative humidities was the criterion of the capillary condensation sorption hysteresis model. The results confirm an applicability of the model for all analysed burnt clay brick types. 1 INTRODUCTION 2 MEASUREMENTS In Mihalka et al. (2011) the analysis of results of the water vapour and nitrogen sorption in combination with the results of MIP for the lightweight carbonate plaster proved the compatibility of the used methods in water vapour sorption modelling and the possibility to identify the pore structure and to model the sorption hysteresis with the use of BJH model of capillary condensation. The approach is applicable for the type IV isotherms and type H1 hysteresis, typical for porous materials containing well defined cylindrical-like pores. In this paper the validity of capillary condensation sorption model and the equivalence of its application for nitrogen and water vapour sorption for various burnt clay bricks are proved. The validity criteria used were: the similarity of the specific surface areas and the similarity of the adsorbate volumes at 95 % relative pressure for the analysed sorbates. For four brick materials from various locations the measurements of the nitrogen and water vapour sorption were carried out. From the results of the measurements the sorption isotherms and the specific surface areas were determined. Nitrogen adsorption-desorption isotherms of nitrogen gas at liquid nitrogen temperature were obtained with Quantachrome Autosorb iQ Station 1 instrument. The isotherms were obtained for the 0.01 – 0.99 relative pressure range. Water sorption isotherms were determined with Aquadyne DVS Water Sorption Analyser, an integrated microbalance system for the automatic study of dynamic water sorption at ambient pressure. The isotherms were determined in 0.1 – 0.98 relative pressure range. The basic parameters of the analysed materials are presented in Table 1. The values of specific surface area for water vapour and for nitrogen are not identical in a case of burnt clay bricks. 175 Table 1. Basic parameters of analysed materials. Nitrogen specific surface area (m2/m3) Water vapour specific surface area (m2/m3) 0.45 7.04x106 4.36x106 1840 0.32 8.5x106 15.94x106 Pezinok 1500 0.50 21.31x106 48.74x106 Devinska Nova Ves 1724 0.30 1.95x106 4.36x106 Location of origin Bulk density Porosity (kg/m3) (-) Zilina 1510 Lucenec 3 ANALYSIS The analysis of the results of measurements was focused on two parameters: the specific surface area and the adsorbate volume at 95 % relative pressure. These parameters for each material were compared from the aspect of their similarity for analysed sorbates. The similarity was evaluated by the specific surface areas ratio determined for water and nitrogen Awater/AN, as well as by the ratio of the adsorbates volumes at 95 % relative pressure Vwater/VN. The obtained ratios were then compared. From (Fig. 1) it is evident that for all analysed materials the ratios of specific surface areas and adsorbed volumes are close. In the figure burnt clay bricks, volcanic ash particles (Delmelle, Villieras and Pelletier, 2005), interior plasters (Matiašovský and Bágeľ, 2009) and lightweight carbonate plasters (Mihálka, Matiašovský and Vasilkovová, 2011) are compared. For the analysed materials the values of Vwater/VN ratios represent ca 0.78 portion of the Awater/AN ratios. This result can be explained if we suppose that the thickness of the adsorbate layer t(p/po) on the adsorbate relative pressure p/po is expressed in (nm) by the empirical Halsey equation:   5 t =τ    ln( p0 / p )  Figure 1. Similarity between ratio of specific surface areas (Awater/AN) and ratio of sorbate volumes at 95 % rel. pressure (Vwater/VN), ---- equality reduced by ratio of effective diameters of water and nitrogen molecules equal to 0.78, ● burnt clay bricks, ○ volcanic ash particles, interior plasters, lightweight carbonate plasters. are proportional suggests that the multilayer adsorption on the pore surfaces is the dominant process and the nitrogen and water sorption can be modelled with use of Halsey equation as the product of adsorbate layer thickness and specific surface area: 1/ n (1) where τ = the effective diameter of sorbate molecule equal to 0.277 nm for water and 0.354 nm for nitrogen, n equals usually 3. The value 0.78 is identical with the ratio of effective diameters of water and nitrogen molecules and the following relationship between the measured water and nitrogen sorption parameters is valid: Vwater A = 0.78 ⋅ water VN AN V ( p / p0 ) = t ( p / p0 ) ⋅ A (3) In case of the sorption hysteresis the desorption is realised at lower relative pressures than the sorption, considering the cylindrical capillary model at the capillary condensation (Gregg and Sing, 1982). This results from a theoretical assumption that at the sorption the relative humidity corresponds to the creation of the cylindrical multilayer water film of the limited stability, spontaneously filling the capillary and converting into the spherical capillary meniscus and so the desorption relative humidity corresponds to this water meniscus existence. Under (2) The fact that the ratios of specific surface areas and volumes at 95 % relative pressure for both sor bates 176 this assumption there are the corresponding sorption and desorption relative humidities at a given equilibrium moisture content. In case of the sorption hysteresis the desorption is realised at lower relative pressures than the sorption, considering the cylindrical capillary model at the capillary condensation (Gregg and Sing, 1982). This results from a theoretical assumption that at the sorption the relative humidity corresponds to the creation of the cylindrical multilayer water film of the limited stability, spontaneously filling the capillary and converting into the spherical capillary meniscus and so the desorption relative humidity corresponds to this water meniscus existence. Under this assumption there are the corresponding sorption and desorption relative humidities at a given equilibrium moisture content. A modelling of the sorption hysteresis by capillary condensation model then is based on the use the classical Kelvin equation for water in the cylindrical pore: ln ( p / p 0 ) = − k ⋅ γ ⋅V rp ⋅ R ⋅ T Figure 2. Sorption isotherms of burnt clay brick from Zilina. (4) where γ = the surface tension, V = the molar volume, R = the universal gas constant, T = the temperature, rp = the pore radius in (nm) and k = 1 at adsorption and 2 at desorption and the following relationship results for relative humidities (Gregg and Sing, 1982): ϕ desorption = ϕ 2 sorption (5) The desorption isotherm then can modelled from the adsorption one replacing the relative pressures by their squares. The results for all analysed bricks have a similar character. Therefore the analysis will be presented for the brick from Zilina. In Figure 2 there are the measured sorption isotherms, including the adsorption curves and the -desorption scanning curves. In Figure 3 there is a comparison of measured and modelled data. The modelling of adsorption by Equation (1) is almost identical with measurement, considering the use of coefficients 1.2 and exponent 1.65 instead of coefficient 5 and exponent 3. The application of capillary condensation desorption model gives results comparable with measurement. Figure 3. Water vapour adsorption and desorption isotherms of burnt clay brick from Zilina, comparison of measured and approximated data. Equation (6) distinguishes the validity of BET equation and Halsey equation in ranges defined by 40 % RH. The BJH model was implemented to the burnt clay brick sorption hysteresis modelling by the following way (Matiasovsky et. al., 2011): The adsorption isotherm was calculated from the statistical film thickness obtained by Equation (1) and multiplied by the calculated specific surface area A. Its values are assigned to the relative pressures calculated by Equation (4) for k = 1. The desorption isotherm was calculated by a summation of liquid adsorbate changes calculated from mean pore and Kelvin radii rk and rp respectively changes of pore volume Vp , changes of film thickness t and adsorbed film surface area A: For all plasters the modelling by Equation (1) does not satisfy in relative pressures region corresponding to the validity of BET equation. Therefore the following combination of BET and Halsey equation was applied at approximation of adsorption isotherm: (6) w = (1 − ϕ ) ⋅ wBET + ϕ ⋅ wHalsey 2  rk  ∆Vl = ∆V p    + ∆t ∑ A r   p 177 (7) and desorption relative humidities at a given equilibrium moisture content and the relative humidity at desorption equals to second power of the relative humidity at sorption. The modelling of scanning curves was made based on an assumption that the transition between adsorption and desorption isotherms is caused only by the change of film thickness on the pore surface area reduced by the capillary surface area of the capillaries filled in with the condensate. The calculated values are assigned to the relative pressures calculated by Equation (4) for k = 2. The modelling of scanning curves issued from an assumption that the adsorption isotherm is represented by adsorbate film and the transition between adsorption and desorption isotherms is caused only by the change of film thickness on the pore surface area reduced by the capillary surface area of the capillaries filled in with the condensate. And the values of scanning curves are assigned to the RH calculated by Equation (4) for k = 1. 5 ACKNOWLEDGMENT This research was supported by the project APVV0032-10. 6 REFERENCES Barret E.P., Joyner L.G., and Halenda P.H. 1951. Determination of pore volumes and area distributions in porous substances. I. Computation from nitrogen isotherms, Journal of American Chemical Sociery, 73: 373 – 380. Brunauer S., Emmett P.H., and Teller E. 1938. Adsorption of Gases in Multimolecular Layers, In Journal of the American Chemical Society, vol. 60: 309 - 319. Delmelle P., Villieras F., and Pelletier M. 2005. Surface area, porosity and water adsorption properties of fine volcanic ash particles. Bulletin of Volcanology vol. 67: 160-169. Gregg S. J., and Sing K. S. W. 1982. Adsorption, Surface Area and Porosity, Academic Press, London. Ishida T., Maekawa K., and Kishi T. 2007. Enhanced modeling of moisture equilibrium and transport in cementitious materials under arbitrary temperature and relative humidity history, Cement and Concrete Research, vol. 37: 565-578. Lowell S., Shields J.E., Thomas M.A., and Thommes M. 2004. Characterization of porous solids and powders: surface area, pore size and density, Kluwer, Dordrecht, 4: 101-126. Matiašovský P., and Bágeľ Ľ. 2009. Compatibility of BET nitrogen and water adsorption/desorption isotherms of interior plasters. In Energy Efficiency and New Approaches: proceedings of the fourth international building physics conference: 11-16. Matiašovský P., Vasilkovová B., and Mihálka P. 2012. Capillary condensation sorption model and its validity for higly porous building materials. Thermophysics 2012 - Conference Proceedings 17th International Meeting of Thermophysical Society. Editor Oldřich Zmeškal. - Bratislava Institute of Physics, Slovak Academy of Science in Bratislava, 2: 116-122. Mihálka P., Matiašovský P., and Vasilkovová B. 2011. Sorption hysteresis of lightweight carbonate plasters. Thermophysics 2011 - Conference Proceedings. 16 international meeting of thermophysical society. Editor Oldřich Zmeškal. - Brno Brno University of Technology, Faculty of Chemistry: 157-165. Neimark A.V., and Ravikovitch P. I. 2001. Capillary condensation im MMS and pore structure characterization, Microporous and mesoporous materials vol. 45-46: 696-707. Sing K.S.W., Everett D.H., Haul P.A.W., Moscou L., Pierotti R.A., Rouquerol J., and Siemieniewska T. IUPAC Recommendations, Pure Applied Chemistry 57,603. Washburn E.W. 1921.The dynamics of capillary flow. In. Physical Review, 17: 273 - 283. Figure 4. Comparison of measured and modelled water vapour sorption isotherms and scanning curves for burnt clay brick from Zilina. The comparison of measured and modelled water vapour adsorption isotherms and scanning curves is in Figure 4. 4 CONCLUSIONS For four burnt clay brick materials the capillary condensation model of sorption based on the application of Halsey equation and BJH model was verified. The modelled data were determined from the results of nitrogen and water vapour sorption experiments. The results were compared with the measured sorption isotherms and confirmed an appropriate agreement. The approach is applicable for the type IV isotherms and type H1 hysteresis, typical for porous materials containing well defined cylindrical-like pores. It was confirmed that in case of the sorption hysteresis the desorption is realised at lower relative pressures than the sorption, considering the cylindrical capillary model at the capillary condensation (Gregg and Sing, 1982).This results from a theoretical assumption that at the sorption the relative humidity corresponds to the creation of the cylindrical multilayer water film of the limited stability, spontaneously filling the capillary and converting into the spherical capillary meniscus and so the desorption relative humidity corresponds to this water meniscus existence. Under this assumption there are the corresponding sorption 178 Anti damp preservation of internally insulated brick walls R. Wójcik University of Warmia and Mazury in Olsztyn, The Chair of Building Engineering and Building Physics, Olsztyn, Poland ABSTRACT: First and foremost, buildings submitted to internal insulation must be preserved against dampness. Currently, preservation methods are not guaranteed – where walls are prepared with injections (such as silicates, organics) they present a problem in becoming agrivated with harmful side effects. In The Building Physics Laboratory at the University of Warmia and Mazury, Olsztyn, a new method has been developed where previously injected walls remain damp-proof course (DPC). This method consists of thermo hermetic inactive injections towards the kerosene waxes in a wall. The article presents the research results that prove high effectiveness of such impregnations. This method is especially useful when used on walls submitted to internal insulation. 1 INTRODUCTION the development of the DPC method has patents proving its original and innovative characteristics. In cases where a building is insulated from the inside there is a necessity for complex anti damp preservations. Additionally, internal layers of the walls increase the risk of biological corrosion and frost damages (Grunewald & Ruisinger 2011, Hartwig & Künzel 2011). Moreover, internal insulation of inappropriately preserved walls accelerates concentration of harmful salts in the area of damp desiccation. An especially high risk of rising dampness is noticeable when insulation is made with thermo insulation layers characterized by high heat resistance and high diffusional resistance of water vapour. The advantages of tightly sealed insulation are: low costs and the possibility of free modelling heat resistance of walls, but in such cases the walls must be perfectly preserved against dampness. External humidity sources, such as rain and capillary raising groundwater, can cause damages. Dampness caused by groundwater containing hygroscopic salts is particularly dangerous. Commonly used structural insulation (DPC) with the use of silicone and silicone - organic preparations, such as thixotrophic creams, do not provide absolute preservation against rising dampness; therefore, a method providing much higher levels of certitude has been searched for. Thermal saturation of kerosene waxes was a method that was considered to provide high level of certitude. Preservations made according to this technology gave relatively good results--both combined with capillary materials conducting steam, as well as steam tight insulations. From my experiences with the preparations of dwindling and sealing capillary and hydrophobisation, a set of faults from these procedures has arisen, thus 2 RESTORING OF DPC BY THE THERMAL HERMETIC INJECTION METHOD The method of thermal hermetic injection is being used in The Laboratory of Building Physics at UWM in Olsztyn, and has nearly proved a way of restoring vertical insulations in existing buildings. It is especially useful in buildings previously submitted to ineffective insulation methods. It is based on the structural blockade made through thermal saturation of capillary – porous wall's structure with thermoplastic composite of kerosene waxes. The waxes can be applied in the walls that have already been saturated with silicones and hydrophobic preparations. In the thermal injection process, the heat treatment is connected with impregnation and these two processes are not separated. The warmed inject is applied to drilled holes into a damp wall without initial drying of the wall. Originally thermoplastic composite worked as a booster for the contact centre heating up. The thermal injection into the damp wall is possible due to the specially designed thermo packers. These devices ensure the encapsulation of the heated area. They work as heaters, passing accumulator, dispensers and valves allowing for pressure adjustment. When manually controlled and created in the blockade area, the gases’ decompression starts the phase of saturation. In the same time the water vapour flows out of the wall. The process in which the vertical blockade is made by the latest thermo hermetic injection is pre179 sented in Figure 1; the work realisation is presented in Figure 2. in the zones directly adjoined to it. The mechanism of the inject’s work that is based on sealing and hydrophobisation is long lasting and stable. At the same time the processes of drying, saturation with the inject and water transport result with the minimizing of the wall's deformation due to the mutual dealing with the results of drying, heating and swelling. The inject is also limited in moving outside the blockade area. Uncontrolled leaks are also limited. In a wall saturated with thermoplastic inject and in a non - isotherm temperature area the self-sealing of the blockade takes place. A high level of certitude of the blockade tightness - independent on the wall's dampness and surrounding thermal conditions is possible due to the visual control of the saturation process and physical neutralization of harmful salts in the blockade zone. The saturated area of the wall can be characterised by the minimal moisture-absorbing capacity of the material. Lasting many hours, heating causes the vulcanisation of the blockade area with existing insulation made on the basis of adhesive asphalt cements and other thermoplastic materials. Old asphalt roofing is regenerated at this time as well. In laboratory conditions research has been conducted over the possible removal of wax from the capillary – porous wall's structure. Such actions are possible therefore the whole process is reversible (which agrees with the Venice Charter's postulates). The benefit of this process is the lack of negative, harmful effects on people's and animals' health. The composite is not harmful if contact with food or drinkable water is made. Practical uses of thermo hermetic insulation are presented in Figure 2. Figure 1. An outline of a technology of restoring structural insulations (DPC) by thermal injection: 1 – wall, 2 – injection hole filled with kerosene wax composite, 3 – thermo packer, 4 – heating element, 5 – the tank thermal insulation, 6 – thermo packer tank, 7 – counter current steam drain (safety valve), 8 – feeding compressed air or negative pressure, 9 – steam transport, 10 – injection transport, 11 – injection circulation, 12 – the zone of injection convection, 13 – the area of injection dispersion, 14 – steam condensation zone (“water trap”). 15 – the zone of non-saturation with water, 16 – hermetic cover. In the wall (1) being at any dampness state (including a state of high saturation with water) on the level of expected blockade, the holes with a distance of 14-17cm apart are drilled (2). In the holes thermo packers are inserted. In the first phase, a composite of kerosene wax is inserted into to the accumulator (6). After the heating elements are started, the wax becomes liquid and is transported into the holes as well as to the cavern and slits connected to the holes where the wax undergoes transformation into elastic sealant. That is the manner in which the centre is initially sealed and what accelerates the process of heating up through stopping water vaporising. The range of melting the composite away is limited by the moving temperature front of 56o C, therefore thermoplastic injections do not leak outside the blockade area. After reaching the required temperature, in the initially sealed blockade the water becomes overheated. When the valves (7) are open the air-steam mixture is decompressed and saturation of the capillary – porous structure starts. Created around the injected areas “water plug” gives the steam an opportunity to travel outside through the injection towards the drilled hole. During this saturation, the wall is both dried in the blockade zone and Figure 2. Realisation of DPC according to thermo hermetic injection. 180 sealed thermo packer. This also means there are no significant compositional mobile changes or adsorbed moisture. At this point mobile moisture can slightly move in the opposite direction than the inject hole (into the wall), because in a thermo packer there is pressurized inject. More importantly thermal diffusion transportation is also noticeable. Due to this observation, during the first stage, inside moisture does not evaporate and can stay in the capillaries in overheated conditions until the temperature reaches (100+T*) ºC, where T* - the surplus of the temperature over the boiling water temperature. This condition causes excess energy in the material's skeleton (as in the moisture itself). Let's denote the stage time as t*. In practice it lasts about 1,5 hour and is characterised by the lack of anti-current out-flow of vapour water through a sealed thermo packer. This also means there are no significant compositional mobile changes or adsorbed moisture. At this point mobile moisture can slightly move in the opposite direction than the inject hole (into the wall), because in a thermo packer there is pressurized inject. More importantly thermal diffusion transportation is also noticeable. Due to this observation, during the first stage, inside moisture does not evaporate and can stay in the capillaries in overheated conditions until the temperature reaches (100 + T*) ºC, where T* – the surplus of the temperature over the boiling water temperature. This condition causes excess energy in the material's skeleton (as in the moisture itself). The second stage, which length will be marked as t**, starts from the moment the accumulator cover opens and this leads into releasing pressure and a fierce change of thermodynamic conditions on the whole system. Released heat energy causes inner sources of water vapour as well as surface moisture. Under the influences of its own pressure, the mixtures of steam and water start moving outside through the shortest way. This way usually leads to a hole in which there is a thermo packer attached. The temperature is the highest here and this is the place where the whole process of making and transporting steam moves to the outside. Closely to the hole more and more of capillaries become empty, and this creates a way for next moisture layers, located deeper into the injected area. This moisture also evaporates and leaves the area. Simultaneously the source of free moisture starts to become noticeable because mechanical and thermodynamic changes lead to releasing surface moisture as well as the adsorbed one. This way, in the second stage the injection area becomes more and more dehydrated. The process of dehydration itself cramps the way of the inject’s absorption through the capillary- porous area. The inject’s and the moisture’s directions are opposite and the homogeneity of the surroundings allows one of these phenomenon to happen. 3 MODELING OF THE THERMO HERMETIC INJECTION'S PROCESS The article prepares a mathematic description of the thermal injection, reflecting main properties of the examined phenomenon in transportation of mass and energy. Within the description information is included concerning the processes concurrent to the injection. For example, the equation of energy transportation is presented. In the process of saturation, the main role belongs to the heating and drying of the centre. It forces the need for the process's division into stages as well as the use of particular mathematic descriptions for each stage. In the centres of varied capillary diameters, moisture and inject transportation can occur simultaneously in spite of the fact that sometimes different moves in direction take place. This is explained by the fact that liquid inject flows into the centre through small diameter capillaries, they have the abilities of the strongest capillary “suction”, which the driving force is causing the movement of the liquid into the heated area. Mixtures of vapour and water can leave the centre through bigger diameter capillaries. In that case capillary influence on such components is smaller and under the influence of pressure, the mixture moves anti-currently towards the inject hole. The case of double porous structures is most commonly seen in practice. In the injection process two stages are distinguished. The first one is the loading of the centre with heat energy. It takes some time since starting a technological heat source lasts as long as mono-porous centres (consisting of capillaries with the same diameter). The second stage consists of simultaneously heating up the moisture and the inject during transportation. Both stages can be described with the same mathematic model. Additionally, a description of the injection process with the assumption of a mono-porous centre, of which the skeleton consists of all the same properties and where are capillaries only of the same diameter needs to be addressed. In practice this should be treated as a theoretical issue. Therefore the model describing this kind of centre can be generalised. In most cases the real injection process takes place somewhere between these two extreme cases. Injection in the mono-porous centre consists of three stages. The first one begins with starting a technological electrical heat source that warms up the injection area. This process takes place under the conditions of some excess pressure caused by hydraulic insulation of the composition consisting of the inject hole and thermo packer accumulator filled with liquid inject. Let's denote the stage time as t*. In practice it lasts about 1,5 hour and is characterised by the lack of anti-current out-flow of vapour water through a 181 – whole volumetric moisture inside the material in the shape of liquid – only as a gas. In the range (0 < < 1) a visible mixture for steam and water. For whole volumetric moisture in the material is in the shape of liquid, while for – only as a steam. In the range there will be visible a mixture of steam and water. Part of the volume filled with the inject will be marked with symbol – there is no inject, – there is neither volumetric moisture nor steam. In broad range, the coefficient of thermal conductivity – also can be defined with the formula including additivity: Active absorption of the inject is possible only when the dehydration of capillaries is finished. Its movement is partially caused by the pressure in a thermo packer, but mostly by capillary forces. Filling of the capillary-porous area with the inject starts, what constitutes third and the last stage of saturation t***. The whole length of the injection tI amounts then: tI = t* + t** +t***. Notably, during each injection stage qualitatively different phenomenon’s occur. It provides an opportunity to gather the whole process into one set of equations. The first stage is characterised only by energetic changes. The equation of energy transportation is also included. In the second stage additional factors should be added such as the equation on water transportation including the phases alternation and desorption. The third stage’s model should include the equation of the energy (present in each of the stages) and the units describing inject’s absorption. Primary causes of all the physical processes accompanying the thermal injection are the energetic changes occurring in capillary – porous centre, causing deviation of some area from thermodynamic balance. (Koniorczyk & Gawin 2008) These changes refer to physical – chemical conditions as well as to the mechanical of all components that are located in the area of influence of technological heat source and the inject. Modelling of the process can be described by an example of the energy equation: T t 2 c T r2 1 T r r RV c (t 0, 0 < r rz ) 1 1 5 c k k 3 1 1 4 2 (3) 5 Symbols 0, 1, 2, 3, 4, 5 characterize the thermal conductivity coefficient of the material’s skeleton, volumetric moisture, surface moisture, adsorbed moisture, steam and the inject’s. Finally we come with the equation of: 1 T t o 2 1 3 5 1 4 5 2 r c 2 1 T r r k k k 0 (4) RV 5 c k k k 0 The last component of the right side of the equation (4) can be precisely defined by the introduction of Heaviside’s formula: (1) where: T – temperature, t – time, r – radius, RV – heat source, c – heat capacity, – density, rz – depth of infiltration of heat interferences counted from the centre to the cylinder edge (the radius of distorted internal layer). Furthermore, the definition of c and coefficients, is characterized by thermal properties of capillary – porous material. Following denotations will be introduced: c0 – specific heat of the skeleton , c1 – specific heat of the volumetric moisture, c2 – specific heat of the surface moisture, c3 – specific heat of the adsorbed moisture, c4 – specific heat of the steam, c5 – specific heat of the inject. Therefore can be assumed as: c 1 o H x 0 dla x, 0, 1 dla x 0 (5) On this basis one can create a new formula: H T 0 dla T 373 T * 0, 1 dla T 373 T * 0, (6) where: T* – the surplus of the temperature over the boiling water temperature T = 373 K in the atmospheric conditions, caused by excess pressure of the inject in a thermo packer and by blockade of capillaries by a water plug created beyond the injection area, allowing outflow of the water mixture and the steam from the direct injection area; T – absolute temperature. Assuming in (6) with a temperature value on the external boarder r = R of a cylinder, the future area that will be filled with the inject T = T(R, t). If one continues to create the formula: (2) k 0 Symbol will be used for describing open porosity of the centre. Varied phase condition of the moisture will be easier described using the internal evaporation coef- RV H T R, t 182 373 T * (7) where in a more elaborate version can be presented as RV H T r , t 373 T * 0 dla T 373 T * 0, RV dla T 373 T * 0. 4 SUMMARY Developed technology of walls’ preservation with the use of thermal – hermetic injection of kerosene waxes was successfully applied in historic buildings that had already been submitted to previous preservation methods. In many cases harmful salts condensation was extremely high. Long lasting research of these objects proves the method as an effective way of preserving the buildings against dampness. Designed mathematical formulas accurately reflect occurring phenomenon. Research of these building conducted by The Building Research Institute in Warsaw, as approval actions, has shown full effectiveness of applied preservation. The kerosene waxes composite and the technology itself gained the technical approval of and recommendation of Building Research Institute in Warsaw. The method was applied for restoring DPC in St. Mary Sanctuary in (8) The last item indicates that in the moment when the whole injection area is heated , the inner heat sources RV start abruptly to the temperature of T 373 + T*, activation of which will be caused by the pressure release with simultaneous lowering of the pressure in the inject area. We will continue to assume that all kinds of water located in capillaries are characterized by the same coefficient , so 1 = 2 3 Including this and substituting (8) to the equation (4) we can transform it to the form of T t 1 o 2 1 5 4 c k k k 0 RV H T r , t 373 T 5 5 2 T 1 T r2 r r (9) historical and non-historical buildings across Poland, Latvia and Russia. * c k k k 0 5 REFERENCES The last equation with the boarder conditions is a model of energy transportation during the injection process. Until the excess pressure release, the water does not evaporate (no phase changes) and the heat source RV does not work. In the moment t* Heaviside’s formula turns off and this starts the energy outflow. Numerical software allows us to determine the temperature distribution in the wall at any given time, with different spans, hole diameters and the power of electrical heaters (Fig. 3). Grunewald J., and Ruisinger U. 2011. Feuchteatlas: Neue Bewertungskriterien für die Bemessung und Simulation von Innendämmlösungen. 1. Internationaler Innendämmkongress Tagungsunterlage. Technische Universität Dresden. Dresden. Hartwig M., and Künzel H. 2011. Bauphysik der Innendämmung und Bewertungsverfahren. 1. Internationaler Innendämmkongress Tagungsunterlage. Technische Universität Dresden. Dresden. Koniorczyk M., and Gawin D. 2008. Heat and moisture transport in porous building materials containing salt, Journal of Building Physics 31, No. 4, 279-300. Figure 3. The temperature distribution in the inject’s area at any given time. Similar solution was achieved for the mass transportation (water, steam and inject). 183 184 Hygrothermal performance of internally insulated brick wall in a cold climate: field measurement and model calibration P. Klõšeiko, E. Arumägi, T. Kalamees Tallinn University of Technology, Chair of Building Physics and Architecture, Tallinn, Estonia ABSTRACT: The interior thermal insulation is frequently one of the only possible solutions for thermal upgrade of the building envelope where the external appearance cannot be changed. In this study four insulation materials were used to build up in-situ test walls. The indoor climate in the test room was controlled to simulate the typical living space with high moisture load. The temperatures, relative humidity and heat flows were monitored. The measurement results over nine months were used to analyse the hygrothermal performance of four different insulation materials. The hygrothermal behaviour of insulation materials during drying and wetting period are presented. The results show that timing of the renovation works is a matter of consideration to avoid the hygrothermal risks inside the retrofitted wall assemblies. The results show that in all the cases, thermal comfort can be improved by increasing the inner surface temperature and decreasing thermal conductivity. However, in some cases the risk of mould growth and interstitial condensation were present inside the retrofitted wall assemblies. 1 INTRODUCTION Maurenbrecher (1998) monitored hygrothermal performance of an internally insulated 765 mm thick masonry wall. As a result of the renovation, the thermal resistance of the wall increased by 47 to 63%. The temperature between the insulation layer and masonry drops below freezing temperature for several months in the winter. Nevertheless the moisture levels did not seem to be a problem in the monitored wall sections. Hens (1998) studied the hygrothermal performance of masonry walls insulated on the inside face and indicated that it increased the thermal bridging effect of partition walls, which resulted in much lower thermal resistance than that calculated. Today’s living habits cause higher indoor air humidity and therefore structures face larger humidity loads. Using correctly dimensioned vapor barriers can prevent the water vapor penetration to the construction. However, the quality of the installation is not always guaranteed and therefore the risk potential remains. In addition of limiting the water vapour diffusion into structures, another possibility exists to use capillary active insulation materials. With these materials is possible to decrease initiation of moisture problems and can upgrade the thermal resistance of the existing walls (Scheffler 2003). Stopp (2001) found that with new materials and new technologies inside insulation can be a viable method. The authors preferred calcium silicate because this material can distribute the liquid moisture content in the structures, and, therefore, it accelerates the drying out process. During the last years, the campaigns have promoted and explained the need to save energy in buildings. The improvement of existing buildings has raised people’s awareness of the energy consumption. The surviving historic buildings are forming wellpreserved areas in all the larger cities in Estonia. These areas are nominated as milieu valuable areas and are protected by planning laws. The main focus is on the built-up environment and therefore the preserving the exterior of the buildings is essential. The simplest way to save energy in apartment buildings is to improve the building envelope by lowering the heat losses through the existing walls, ceiling and windows. By improving the building envelope with additional thermal insulation, it is possible to save energy and increase thermal comfort. Adding insulation to the external walls is used as a common solution. That can be easily used to lower the heat losses through the existing exterior walls if the buildings have no restrictions because of the architectural and historical values and the exterior can be changed. In historic buildings, often the architectural and historic values set limitations to the use of external thermal insulation. Due to the regulations no changes outside the external wall on the facade are allowed to be done in the buildings with architectural and historic values. There exists a strong pressure to use internal insulation as a energy conservation measure in historic buildings. 185 thermal performance of a brick wall. The paper describes the test wall setup, presents measurement results over one year and the validation of numerical simulation models. Häupl (2003, 2004) demonstrated the function of the insulation systems by presenting the measurement results of several internal insulation systems at a number of different outside wall construction. The application of capillary active inside insulation materials proved advantageous for the drying process of potential built-in moisture as well as for the limitation of the condensation amount during winter. The thermal transmittance of the building walls could approximately be halved in the presented cases without the necessity of vapour barriers. Toman (2009) presented long-term on-site assessment of hydrophilic mineral wool insulation system without water vapour barrier. The reconstructed building envelope exhibited very good hygrothermal performance. Nevertheless Morelli (2010) made computational analysis of the internal insulation solution of masonry walls with wooden floor beams in northern humid climate and showed that the solution would almost halve the heat loss through the wall section compared to the original one but the internal insulation reduces the drying potential of the wall which can lead to moisture problems. Scheffler (2011) introduced an innovative and sustainable internal insulation system based on a light-weight autoclaved aerated concrete. The results showed that though the moisture content inside the masonry structure increased, the overall moisture level was kept below critical value. Vereecken (2011) made hygrothermal comparison between capillary active and traditional vapor tight interior insulation system. The performance of the capillary active system is shown to be more sensitive to the different parameters (wind-driven rain load, orientation, catch ratio, finishing coat, thickness of the wall, etc). Additionally, there exists a large number of uncertainties concerning internal insulation (Nielsen 2011, Zhao, J. 2011). Most of these computational and experimental studies have focused on wall structures in Central European climate conditions. In cold climate, external thermal insulation is hygrothermally a much safer solution than internal thermal insulation. Frequently, in hygrothermal performance risks arise. To avoid the possible mold growth and condensation problems special attention should be paid during the design and installation phase of the renovation solution. A better understanding of the hygrothermal performance of the interior insulation retrofit approach in cold climate is needed. In this study, four solutions of internal thermal insulation for the brick wall are tested. An analysis was carried out to assess the impact of an interior insulation retrofit on the hygro- 2 METHODS Field measurements on a school building (Figure 1) and computer simulations were used in this study. Figure 1. View of the studied school building. 2.1 Field measurements To compare the hygrothermal performance of different internal insulation materials the same wall was insulated with four materials. Materials were selected so that diffusion open, capillary active, and vapour tight materials were used: − calcium silicate (CaSi): capillary active material with very high open porosity and low vapour diffusion resistance; − aerated concrete (AAC) with high open porosity, lower capillary activity and thermal conductivity; − polyurethane board with capillary active channels (IQ-T) - combines low thermal conductivity and a certain capillary activity; − polyisocyanurate board (PIR) with closed pores: low thermal conductivity and relatively high vapour diffusion resistance forming a vapour barrier in itself. The original brick wall was 73…75 cm thick, composed of three layers of brick with two air and insulation (peat) layers between them. The thickness of the insulation layer was selected to represent typical products and to avoid large thermal transmittance differences between the walls. 186 Figure 2. Plan, view, and section of the studied test walls. Each test wall was equipped with two temperature and relative humidity (RH) sensors between insulation and the original wall (∅ 5mm, Rotronic HC 2 SC05) and heat flow plates (Hukseflux HFP01) on the internal surface of insulation. Internal and external surface temperatures were also measured. Indoor climate was heated with the set point of thermostat +21 ºC and humidified to keep the moisture excess of +2.3…+4.4 g/m3. Indoor climate conditions were selected to represent the average conditions in dwellings with high humidity loads (Kalamees et al. 2012), see Figure 4. The climatization of the test room started by an attempt to imitate a typical renovation process: − Period 1 (P1): internal insulation works and starting to heat the room: 27.03.2012; ∆ν +0.1 g/m3 (not humidified); − Period 2 (P2): starting to humidify ~3 months after installation (4.07.2012) to represent time period between the renovation and moving back in, high humidity load; ∆ν +2.3 g/m3 (humidified). − Period 3 (P3): humidifying period to see the influence of different humidity loads on hygrothermal performance of test walls, medium high humidity load: 4.07.2012…8.10.2012; ∆ν +2.9 g/m3 (humidified) − Period (P4): medium humidity load, 8.10.201214.12.2012; ∆ν +4.2 g/m3 (humidified). − Period 5 (P5): with low humidification (from 14.12.2012) to see drying out potential of different walls; ∆ν +2.5 g/m3 (humidified according to indoor RH) To assess the conditions favourable for mould growth the Finnish mould growth model (Viitanen 2007, 2010) was used. Critical RH levels RHcrit>80% and RHcrit>95% depending on substrate category was used. RHcrit>80% in case of PIR was used, because the side which is in contact with the original wall is covered with paper. RHcrit>95% for stone materials (CaSi and AAC) and IQ-T was used. 2.2 Computer simulations The results of temperature and humidity measurements from test walls were compared in an 1D model with a commercial hygrothermal simulation program DELPHIN 5.7.4 (Grunewald 1997, 2000, Nicolai 2008). There were two main reasons for that comparison: − to acquire a better understanding of the hygrothermal performance of the internally insulated brick walls; − to validate the simulation model for future simulations with different initial and climatic conditions as well as with different dimensions of the building envelope. The material producer’s data and simulation program’s (DELPHIN) material libraries were used to provide material properties. The density and porosity of original brick were measured. Brick with similar material properties was selected from DELPHIN’s material library. The dependency of hygrothermal properties on the environmental conditions was taken into account: water vapour permeability, liquid water conductivity, thermal conductivity dependent on water content of a material. The main properties of materials, used in study are shown in Table 1. Table 1. Properties of insulation materials. Property Material PIR IQ-T AAC Thickness of used layer, mm 30 50 60 Density, kg/m3 32-38 45 90 Thermal conductivity λ, W/(mK) 0.027 0.031 0.047 Vapor diffusion resistance coefficient μ, 100 27 2 Water absorption coefficient Aw, kg/(m2 1.0⋅1 h0.5) 0-7 0.01 0.1 187 CaSi 50 360 0.063 4.6 1.17 3 RESULTS Moisture excess ∆νi, g/m 3 7 3.1 Climate conditions Indoor and outdoor temperatures as well as temperatures between the insulation and the original wall are shown in Figure 3. Outdoor air temperatures fluctuated from +17 to +30°C in July with an average of +19.4°C. Temperature until December was rather mild, with an average temperature of +5.4°C in November and minimum temperatures reaching +1°C on a number of occasions. According to Künzel (2011) freezing should be avoided to prevent damage to the materials. Although the outdoor air temperature stayed under 10°C for 8 days in December, none of the sensors detected temperatures below 0°C between the insulation systems and original wall. Conditioned indoor temperature stayed the same throughout the heating period (+21+/-0.5°C). A drop (minimum of +17.5°C) of the indoor temperature during the cold period in December was caused by low power setting of additional heating in the room. 4 3 2 1 0 -1 Period 1 -25 -20 -15 -10 -5 0 5 10 15 20 25 Exterior temperature t e,oC Figure 4. Indoor moisture excess dependent on exterior temperature. 3.2 Comparison of reactions to changes in climate conditions Comparisons of relative humidities during the period without humidification (Period 1) of the test walls are given in Figure 5. Drying out period finished fastest in the case of CaSi (24 days), AAC section dried to the same level in 38 days, whereas iQ-T reached the lowest RH of 86% at the start of humidification (3 months after the installation). tin 100 10 5 0 -5 -10 -15 tex (72h avg) -20 05.2012 07.2012 PIR 09.2012 IQ-T 11.2012 AAC 01.2013 CaSi installation on 27.03.2012 90 15 Relative humidity RH, % o Period 2 -3 20 Temperature t, C Period 3 5 -2 30 25 Period 5 Period 4 6 80 70 60 50 40 38 days 24 days 30 P1 20 17.03.2012 1.04.2012 16.04.2012 1.05.2012 17.05.2012 1.06.2012 Time (mm.yyyy) PIR Figure 3. Temperatures between insulation systems and original wall and indoor/outdoor air temperatures. IQ-T AAC CaSi RH_in Time (d.mm.yyyy) Figure 5. Drying period without humidification – relative humidities between insulation systems and original wall; time taken by wet systems to reach stable level. Figure 4 shows the values of moisture excess plotted against exterior temperature. On the graph, different periods of humidification can be identified. Average values of moisture excess were 0.1 g/m3, 2.3 g/m3, 2.9 g/m3, 4.2 g/m3 and 2.5 g/m3 for periods 1-5 respectively. Figure 8 to Figure 11 show the duration of the periods. Figure 6 shows the period of the start of humidification (transition from period 1 to 2). Two distinct behaviours are apparent – in the case of low vapour diffusion resistance (AAC, CaSi), the RH level behind the insulation follows the indoor conditions closely. PIR and iQ-T, on the other hand, reacted more slowly. 188 100 90 90 Relative humidity RH, % Relative humidity RH, % 100 80 70 60 50 40 30 Humidification ON P2 P1 20 28.06.2012 PIR 5.07.2012 IQ-T CaSi RH_in PIRmeasured PIRsimulated 80 70 60 50 Humidification ON 40 30 P1 20 05.2012 12.07.2012 AAC RHcrit P2 07.2012 P3 09.2012 P4 11.2012 P5 01.2013 Time (mm.yyyy) Figure 6. Start of humidification - relative humidities between insulation systems and original wall. Figure 8. Relative humidity between PIR board and original wall; periods of humidification. Second period of drying (period 5) is visible in Figure 7, permeable materials stabilize in about 5 days. Figure 9 shows the relative humidity between the IQ-T insulation system until the error of the measurement system. The value of RH stayed just below the limit of 95% which is also the region where the capillary transport of humidity increases significantly. As a reference, the RH of the sensor close to the ceiling (IQ-Thigh) is given. Relative humidity RH, % 90 80 70 60 50 40 30 P4 100 90 5 days P5 20 10.12.2012 PIR Relative humidity RH, % 100 Humidification to LOW on 14.12.2012 Time (d.mm.yyyy) 17.12.2012 IQ-T (up) AAC 24.12.2012 CaSi RH_in Time (d.mm.yyyy) Figure 7. Drying period with no humidification – relative humidities between insulation systems and original wall. Time for permeable materials to reach stable level. 80 iQ-Thigh iQ-Tcenter, meas. iQ-Tcenter, simul. RHcrit 95% RHcrit "80%" 70 60 50 Humidification ON 40 30 20 05.2012 P1 P2 07.2012 P3 09.2012 P4 11.2012 P5 01.2013 Time (mm.yyyy) 3.3 Moisture-performance of the test walls Figure 9. Relative humidity between iQ-T insulation system and original wall; periods of humidification. 3.3.1 Relative humidity between the insulation and the original wall Relative humidity behind the insulating layer and the surface of the original wall fro different walls are given in Figure 8 to Figure 11. These charts also show the corresponding value of the relative humidity as simulated in Delphin and the assessment criteria for the proper hygrothermal performance. Relative humidity behind the layer of AAC (Figure 10), due to its low water vapour diffusion resistance, reacted more quickly and drastically to the changes of indoor humidity. The period with higher moisture excess resulted in 59 hours of possible condensation. Figure 8 exhibits the relative humidity of the PIR test wall. . Measured values of the relative humidity exceeded the critical limit (RH~80%) for 78 days of the 230 of the test duration. Both the measured and simulated humidities dropped to safer levels during period 4, while the lack of humidification (second half of December 2012) lowered it even further. 189 The average values of surface temperature in November stayed within 19.2-19.8°C and the average for the uninsulated reference wall was 17.3°C (average indoor temperature 20.9°C). During the cold period in December, the surface temperature of the reference wall dropped to an average of 13.1°C, while the sections with insulation stayed 3-4°C higher (average indoor temperature 19.0°C) depending on the material. 100 80 AACmeasured AACsimulated 70 60 50 Humidification ON 30 P1 20 05.2012 P2 07.2012 P3 09.2012 P4 11.2012 P5 01.2013 1.20 2 40 Thermal transmittance U, W/(m *K) Relative humidity RH, % 90 RHcrit 95% RHcrit "80%" Time (mm.yyyy) Figure 10. Relative humidity between AAC insulation system and original wall; periods of humidification. Likewise to the AAC wall segment, the RH graph of CaSi (Figure 11) exhibits high fluctuations which follow the changes in the indoor relative humidity. However, the overall level stays lower with the maximum value of 95% during period 3. This is due to lower insulation thickness and higher thermal conductivity which leads to higher temperatures behind the insulation, while the absolute humidities stay practically on the same level. 1.00 Ureference wall = 0.94 W/(m2*K) 0.80 0.54 0.60 0.40 0.38 0.41 0.33 0.20 0.00 PIR iQ-T AAC CaSi Material Figure 12. Average thermal transmittances calculated from measured thermal resistances of reference wall and insulation layers during the period of 1.09.2012-28.12.2012. 100 Relative humidity RH, % 90 80 70 RHcrit 95% 4 DISCUSSION RHcrit "80%" CaSimeas. CaSisimul. The hygrothermal performance of four internally insulated test walls was compared in field conditions. 1000mm wide sections of thermal insulation systems were installed on the same wall in the test room. Because of the possible moisture and heat flux between the test wall sections the different parts were separated by 50 mm wide joints. In the existing wall the ~50 mm deep grooves were cut and cleaned. Afterwards the joints in the existing wall and between the different insulation sections were completed with aluminium foil and polyurethane foam. Although there may have been some minor leaks between the test walls, the potential for the vapour diffusion and heat conduction was low and possible heat and moisture movement between the test walls was probably minor. Another factor influencing the accuracy of direct comparison between the wall sections could be non-uniform distribution of insulation in the existing wall and air flows in cavities. This may be considered as general limitation of the field study. The temperatures measured in all the different insulation material sections and in reference wall section, of the brick façade were uniform. Façade temperature is mainly influenced by the outdoor temperature. The test wall is facing north, therefore the solar heat 60 50 Humidification ON 40 30 20 05.2012 P1 P2 07.2012 P3 09.2012 P4 11.2012 P5 01.2013 Time (mm.yyyy) Figure 11. Relative humidity between CaSi insulation system and original wall; periods of humidification. 3.3.2 Thermal conductivity and surface temperature Thermal transmittance was calculated based on heat flow and surface temperature measurements for the period of 1.09.2012-28.12.2012. The period was chosen with low temperature fluctuations and solar radiation. The average values for thermal transmittance are given in Figure 12. Higher water vapour content was found to have an inconsiderable effect on the thermal transmittance, as the moist values of all materials dropped ~3% during the period without humidification. This is quite unexpected as the reduction of the moisture content of the vapour permeable materials was more significant than on the PIR. 190 material layers. After switching on and off the humidifier, a much quicker response of the vapour open materials to the change than that of vapour tight materials was observed. Also, there is a difference in drying rate of the materials if two time periods are compared (period 1 and 5). The drying is accelerated because of the lowered moisture load and mechanical ventilation during period 5. The measurement results show high humidity levels in all the cases. In the case of AAC, possible condensation occurred. The conditions suitable to initiate the mould growth are described in (Viitanen 2007, 2010). The measured temperature and RH results show that the conditions suitable for the mould growth were fulfilled during most of the time period. The most critical are the conditions in case of the PIR (RHcrit>80%) and AAC (RHcrit>95%). The important aspect of any retrofit is the increase in thermal resistance of the building envelope. This results in reduced energy usage and improved thermal comfort for inhabitants. The thermal resistance was calculated using measured temperature difference across the brick wall, the insulation layers and the heat flow through the wall. Regarding the wall construction, the additional insulation resulted in a decrease of the thermal transmittance of the wall by 40 to 65%. The highest thermal resistance was achieved with IQ-T. There was evident temperature difference on the surface of the wall on the room side. During the winter period the temperature of the inner surface of the IQ-T wall was almost 3ºC higher than that of the reference wall. In all the cases, the temperature was higher than that of the reference wall and the thermal comfort was improved. The quality of the simulation results depends on the input variables, on the assumptions and simplifications made and on simulation settings. Simulations of hygrothermal performance using the HAMsimulation programs may contain different kinds of errors, for instance, the description of the existing wall assemblies in terms of the chosen material properties with retrofitted building constructions. No thorough laboratory tests were conducted to obtain data about the properties of the different building materials used in the existing wall construction. The material database of the simulation tool and material data from the literature was used. Therefore, the field measurements enable us to check the accuracy of the simulation model. The simulation models of four internally insulated brick walls were validated using field measurements. A satisfactory correlation between the calculated results and measured values was achieved. absorption was minimal. The uniform temperature can be explained by the relatively thick wall construction (typical thickness of external brick walls is 43…51 cm) and the different layers inside the wall that equalize the temperature distribution. The bigger temperature differences were on the surface of the inner face of the existing masonry wall. Largest temperature drop in the added insulation material was in case of the IQ-T and lowest in case of the CaSi, because of the difference in thermal transmittance. This result coincides with expected results. Freeze-thaw damage mechanism can occur at temperatures well below freezing when the wall construction is essentially saturated. With brick walls, the limit for freezing can be lower than 0ºC because of the dissolved salts in the brick pores. The temperature limit can be from 0 to -5 ºC (Vesikari 1998). The temperatures inside the wall assemblies were measured between the added insulation material and existing wall. In all the cases the temperature did not drop below the freezing temperature between the insulation and existing wall. During the first year no freeze-thaw cycles were observed directly behind the insulation even if the limit was set to 0 ºC. The lowest RH is with the PIR (dry installation) and the highest with the AAC (wet installation and vapour permeable material). During the installation of PIR no extra moisture is added to the wall and the RH between the insulation material and existing wall reflects the built in moisture of the existing wall. All the other materials were installed using glue-mortar and additional moisture was added to the construction that contributes to higher RH level. As all wet materials had ~5 mm layer of glue mortar according to installation instructions, the amount of water added to the wall was largely the same. With IQ-T, the topmost plaster level was thicker than on the others (10-15 mm vs 2 mm with the CaSi and 5 mm with AAC). The moisture added during the retrofit works can cause long periods of high RH, if the renovation works are done during the period when the outdoor and indoor conditions are not suitable for the drying out or the moisture is not dried out before the rooms are taken into use again. The results show that timing of the renovation works or the start point of the reuse need to be taken into account. The time after the renovation works is most critical for the IQ-T, for CaSi and AAC it is considerably shorter. It is also significant, that although more resistant to vapour diffusion, humidification also rises the RH behind PIR layer. To simulate the use of the room as a living space the humidifier and ventilation system were switched on two months after the installation of the insulation 191 national research project supported by Sustainable Renovation Information Centre SRIK. References The simulation models will be used for further studies to analyse the hygrothermal performance of different brick wall constructions with internal thermal insulation in the measured climate conditions and different outdoor and indoor climate conditions. Häupl P., Fechner H., and Petzold H. 2004. ‘Interior retrofit of masonry wall to reduce energy and eliminate moisture damage: Comparison of modelling and field performance’, Thermal Performance of the Exterior Envelopes of Buildings IX, Florida. Häupl P., Jurk K., and Petzold H. 2003. Inside thermal insulation for historical facades, In Proc. Of the 2nd International Conference on Building Physics, 2003, ISBN90 5809 5657, 463-469. Hens H. 1998. Performance predictions for masonry walls with inside insulation using calculation procedures and laboratory testing. In Journal of Thermal Envelopes and Building Science, Vol. 22, July, pp. 32-48. Künzel H. M. 2011. Bauphysik der Innendämmung und Bewertungsverfahren, In Proc. of the 1. Internationaler Innendämmkongress. 2011, ISBN 3-940117-07-6, 9-16 Nielsen A., Möller E. B., Rasmussen T. V., and Hansen E. J, 2012. Use of sensitivity analysis to evaluate hygrothermal conditions in solid brick walls with interior insulation. In 5th International Building Physics Conference (IBPC) – NSB 2011, ISBN 978-952-15-2574-2 (Vol. 1), 465-472 Scheffler G., and Grunewald J. 2003. Material development and optimisation supported by numerical simulation for capillary-active inside insulation material, In Proc. Of the 2nd International Conference on Building Physics, 2003, ISBN90 5809 5657, 77-85. Stopp H., Strangeld P., Fechner H., and Häupl P. 2001. The hygrothermal performance of external walls with inside insulation, In Thermal Performance of the Exterior Envelopes of Buildings VIII, Clearwater Beach, Florida. Toman J., Vimmrová, A., and Černý R. 2009. Long-term onsite assessment of hygrothermal performance of interior thermal insulation system without water vapour barrier, In Energy and Buildings, Volume 41, Issue 1: pp 51-55 Vesikari E. 1998. Prediction of service life of concrete structures by computer simulation. Helsinki University of Technology. Licentiate's thesis. 131p. Viitanen H., and Ojanen T. 2007. Improved model to predict mold growth in building materials, In Thermal Performance of the Exterior Envelopes of Buildings X, Florida. Viitanen H., Vinha J., Salminen K., Ojanen T., Peuhkuri R., Paajanen L., and Lahdesmaki K. 2010 Moisture and biodeterioration risk of building materials and structures, In Journal of Building Physics, 33 (3) , pp. 201-224. Zhao J., Plagge R., and Grunewald J. 2011. Performance assessment of interior insulations by stochastic method. In 9th Nordic Symposium on Building Physics – NSB 2011, ISBN 978-952-15-2574-2 (Vol. 1), 465-472 5 CONCLUSIONS Upgrading of energy performance for existing historic buildings is often possible only by internal insulation. The study concentrated on four different materials. In-situ measurements in the test room with regulated indoor climate were conducted. The results show that the added moisture during installation can cause high RH levels in a wall for a long time period that can lead to interstitial condensation. The temperature and RH condition inside the wall between the added insulation and brick wall favouring mould growth were fulfilled during most of the time period. As PIR board exceeded the RHcrit”80” convincingly, it cannot be recommended to be used in its current form in high humidity loads – e.g. using a product without the paper layer on the “cold” side to rise the critical RH criterion or applying additional moisture retarder could be considered. Furthermore, possible problems caused by moisture diffusion flux toward the interior will have to be evaluated. IQ-T behaved similarly to PIR, however the thicknesses differed. Thus the materials could not be compared directly and the effect of its capillary active channels will need further assessment. AAC and CaSi exhibited almost equal absolute humidity behind the insulation, with the difference in RH due to lower thermal transmittance of the AAC layer. To reduce the RH levels, lower thicknesses of insulation (in both AAC and IQ-T cases) should also be evaluated. The temperature on the inner surface was increased compared to the existing wall and thermal transmittance was halved almost in all the cases. Based on the measurement results the simulation models for the further computational analysis of different brick wall constructions and climate conditions were calibrated. 6 ACKNOWLEDGEMENT The research has been conducted as part of the IUT1−15 project “Nearly-zero energy solutions and their implementation on deep renovation of buildings“. The study utilizes the measuring data of the 192 Thermal implications of radiant roof barriers: A field study in a hot and humid climate V. Müller, U. Pont, P.H. Tan & A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: This paper studies radiant barriers (aluminium foil systems) in attics of buildings in the hotand-humid climate of Malaysia. Three similar 3x3m test cells were equipped with different configurations of roof tiles (clay, cement) and radiant barrier foils. The test cells were equipped with data loggers for indoor temperature and relative humidity in the attic and the room below. The local outdoor climate (temperature, relative humidity, solar radiation, wind speed) was measured with a roof mounted weather station. Different natural ventilation options for the attic were realized. The room below the attic was ventilated via passive cooling during night in all cases. The function and effectiveness of configurations with radiant barriers were compared with the baseline case. In general, the results showed no significant difference between the use of clay and cement roof tiles, while the use of radiant barriers did have some influence on the attic and room temperatures. Table 1. Temperature and relative humidity conditions in Ipoh, Malaysia. Max Min Monthly Means Temperature Non-Monsoon 35 22 25-27 [°C] Monsoon 30 22 25-27 1 INTRODUCTION In this contribution, the effect of radiant roof barriers and different roof tile materials on indoor thermal comfort in sample buildings in a hot and humid climate (Ipoh, Malaysia) is discussed. Three singleroom test cells were constructed and equipped with different roof constructions concerning roof tiles, additional counter battens and aluminium foil underlay. Different ventilation options were deployed and the resulting indoor conditions monitored. Objectives of the study were to find out if: i) minor changes in the roof design (roof tiles, and layer construction) influences heat flow from the attic to the room below, ii) the application of radiant roof barriers and additional implementation of counter-batten air spaces change the thermal behavior of the building and improve indoor climate. Rel. Humidity [%] Non-Monsoon 96 65 75-85 Monsoon 96 74 75-85 2.2 Heat migration via roofs Buildings in tropical countries like Malaysia normally feature a steep roof, given its good water drainage ability during Monsoon rain season. The attic spaces below the roofs tend to get very warm during daytime. This can influence the thermal conditions in the rooms below (Figure 1). To find potential technical solutions to reduce this effect, many studies and field experiments have been carried out. Thereby, the solar reflectance of roof surface, the thermal properties of roof tiles, radiant barriers, and increased air flow underneath roof tiles have been considered. Concerning the solar reflectance of roof surfaces, Miller et al. (2010) performed a detailed survey in this climatic context. They calculated the Solar Reflectance index (SRI) for different material and color combinations of roof tiles. As expected, in general, the darker the color, the higher was the solar absorption of the tiles. A study by Suehrcke et al. (2008) in the hot and humid climate of Townsville 2 BACKGROUND 2.1 Key outdoor climate data in Ipoh, Malaysia Malaysia is a tropical country. Table 1 shows an overview of the typical temperature and relative humidity ranges at the location of the experiments. Given typical absorption rates of roof surfaces (between 20 and 95%), non-shaded roof areas can be expected to critically influence indoor conditions (Suehrcke et al. 2008). 193 (Australia) compared clay and cements roof tiles and found no significant difference in terms of heat gain. Radiant barriers are thin and highly reflective sheets that are intended to reduce the heat transfer by radiation between the roof cladding and the interior of the attic. Michels et al. (2008) suggested that these systems can help reducing the effect of solar radiation. Concerning the counter batten construction involving air flow, Miller et al. (2010) suggested that a minimum of 3/4 inch (19 mm) is necessary for a ventilation air layer to dissipate heat. Vandewater (2007) applied CFD to explore the effect of counter battens ventilation for different roof angles. The orientation allowed direct solar radiation on both sides of the roof from 10 am to 4 pm, while during sunrise and sunset one side received direct radiation (see Figure 3). Different roof constructions were applied to the different test cells. Test cell A (Figure 4) was simply cladded with cement tiles, without an additional counter batten construction, and with no aluminium radiant barrier. Test cell B (Figure 5) was cladded with cement tiles above a counter batten construction, and additionally was equipped with an aluminium radiant barrier. Test cell C was constructed in a similar manner as test cell B, but instead of cement tiles clay tiles were used. The application of both counter battens and radiant barrier foil is known as “cool roof system” (Miller et al. 2010). Figure 3. Sun path and insolation with regard to the test cells. Figure 1. Heat flow from attic space to the room below in the climatic context of Ipoh, Malaysia. 3 APPROACH Three free standing test cells were constructed (in 2010) with an internal floor area of 3x3 m and 3.6 m room height (Figure 2). The walls were constructed out of 110 mm bricks and mortar. Each of the test cells feature a gable roof with a 45° pitch and a ceiling-to-ridge-height of 1.5 m. The orientation and layout of the test cells is illustrated in Figure 2. Figure 2. Plan and elevations of the test cells. Figure 4. Construction details of test cell A. 194 Figure 5. Construction details of test cell B. In all test cells temperature and relative humidity was monitored in the attic as well as in the room below. Additionally, outdoor temperature, relative humidity, solar radiation, and wind speed were monitored via a weather station at the location of the experiments. The monitoring took place between October and December 2010, measuring interval was 10 minutes. During the monitoring period, four different configurations ("rounds") were tested. The rooms below the attic were ventilated via the door in all ventilation strategies: During day time (7 am – 7 pm) the entrance doors of the cells were kept closed, while during night (7 pm – 7 am), the doors were opened to facilitate night-time ventilation. The windows stayed closed throughout the experiment period. Note that the actual rate of air change in the two cells could not be measured. However, given identical location, construction, and orientation of the two cells – as well as synchronized timing of the opening and closing of the cell doors – it is likely that the air change rates in the three cells were similar. The ventilation of the attic space, however, was different in each cell (see Figure 6 and Table 2). Thereby, the ventilation parameters in the attic were as follows: i) attic window opened or closed (Figure 7); ii) soffit board mounted or dismounted (Figure 8); iii) aluminium foil at the ridge opened or sealed (Figure 9). Each of the ventilation options were deployed for at least seven days. Figure 6. Ventilation options in each test cell. Figure 7. Attic window open (left) and closed (right). Figure 8. Soffit board dismounted (left) and mounted (right). 195 Table 2. Ventilation regimes in test cell A, B, and C. options Test cell A Round 1 Round 2 Round 3 Round 4 Test cell B Test cell C Attic window closed closed closed Soffit board mounted mounted mounted Radiant barrier NA sealed sealed Attic window closed closed closed Soffit board mounted mounted mounted Radiant barrier NA opened opened Attic window closed closed closed Soffit board dismounted dismounted dismounted Radiant barrier NA open Attic window open open open open Soffit board dismounted dismounted dismounted Radiant barrier NA open open cell C. This may be attributed to the only difference between these cells, namely cement roof tiles (B) versus clay tiles (C). 4.1.2 Round 2 In round 2, the radiant barriers in B and C were opened implying a slightly higher air change in the attic space. Room and attic temperatures are shown in Figures 12 and 13 respectively. Table 4 shows temperatures at 02:00 pm and 00:00 am. The room temperatures show a very similar course for all three test cells, however, C’s peak is about 1 K lower than those of A and B. Similar to round 1, Cell A does show the highest peak temperatures in the attic, but lower temperatures during night. Again, this can be attributed to the radiant barrier effect, which appears to compensate for the somewhat higher air change rate in cells B and C. Similar to round 1, peak temperatures in cell B are slightly higher than cell C. Figure 9. Radiant barrier opened (left) and sealed (right) at ridge. 4 RESULTS AND DISCUSSION 4.1.3 Round 3 For round 3, the soffit boards of all three test cells were dismounted. Table 5 shows the attic and room temperatures in all cells at 02:00 pm and 00:00 am, while Figures 14 and 15 illustrate the temperature courses. Given the higher ventilation rates in the attic spaces in this round, the temperature difference between A's attic on the one side (no radiant barrier), and the attic space of B and C (with radiant barrier) on the other side are smaller, as compared to rounds 1 and 2. Likewise, the roof tile material (cement versus clay) also does not appear to play an important role, once the attic ventilation is increased. As to the room temperature in this round, we cannot provide an explanation for the slightly lower temperature of the cell A. 4.1 Temperatures Each of the monitored rounds spans at least 5 days. For analysis, the monitoring period of each round was represented in terms of a 24 hour cycle. For each round the temperatures of the attic space and the room below were measured and analyzed. 4.1.1 Round 1 In round 1, the attics had no intended ventilation (see Table 2). Figures 10 and 11 show the room and attic temperatures in the three cells. Table 3 shows the temperatures at 02:00 pm and 00:00 am. Both the lower peak temperatures and higher night temperatures in cells B and C (as compared to cell A) may be attributed to the radiant barrier. The temperature drop in all rooms after 7 pm suggests that night-time ventilation of the test cells has a stronger effect than the heat exchange with the attic. Peak temperatures in cell B are slightly higher than 4.1.4 Round 4 Round 4 featured the maximum ventilation for attic spaces, as the attic window was opened. Table 6 shows the temperatures in roofs and rooms at 02:00 pm and 00:00 am. Figures 16 and 17 illustrate the 196 temperatures. As with the previous round, increased ventilation appears to lessen the difference between the cell temperatures. Table 3. Temperatures in Round 1. Cell A B C Time Outside Attic Room 02:00 pm 31.0 35.9 32.7 00:00 am 23.4 24.4 26.3 02:00 pm 31.0 33.0 31.6 00:00 am 23.4 26.0 26.3 02:00 pm 31.0 32.5 30.1 00:00 am 23.4 26.0 26.3 Figure 12. Room temperatures in test cell A, B, C, outdoor temperature and Solar Radiation in Round 2. Figure 13. Attic temperatures in test cell A, B, C, outdoor temperature and Solar Radiation in Round 2. Figure 10. Room temperatures in test cell A, B, C, outdoor temperature and Solar Radiation in Round 1. Table 5. Temperatures in Round 3. Cell Time Outside Attic Room A 02:00 pm 32.1 34.0 30.3 00:00 am 27.0 28.9 30.4 B C 02:00 pm 32.1 31.9 31.0 00:00 am 27.0 30.3 31.5 02:00 pm 32.1 31.6 30.3 00:00 am 27.0 30.2 31.3 Figure 11. Attic temperatures in test cell A, B, C, outdoor temperature and Solar Radiation in Round 1. Table 4. Temperatures in Round 2. Cell Time A 02:00 pm 32.5 38.0 32.3 00:00 am 24.5 25.7 26.4 02:00 pm 32.5 34.8 31.9 00:00 am 24.5 26.8 27.2 02:00 pm 32.5 34.2 31.4 00:00 am 24.5 26.8 27.2 B C Outside Attic Room Figure 14. Room temperatures in test cell A, B, C, outdoor temperature, and Solar Radiation in Round 3. 197 significantly higher percentage of time in comfort range. The cell with the clay roof tiles (C) performs slightly better than the cell with the cement tiles (B). Figures 19 and 20 compare the percentage of time in the aforementioned temperature range separately for daytime (07:00 am – 07:00 pm, no ventilation) and nighttime (07:00 pm – 07:00 am, ventilation via entrance door). The previously highlighted difference between room temperatures of the different cells at different rounds apply also to these results. As it could be expected, nighttime conditions are more favorable thermally. Figure 15. Attic temperatures in test cell A, B, C, outdoor temperature and Solar Radiation in Round 3. 100% Table 6. Temperatures in Round 4. Time Outside Attic Room A 02:00 pm 33.7 35.4 33.5 00:00 am 25.3 26.7 29.2 02:00 pm 33.7 33.7 33.5 00:00 am 25.3 28.1 30.0 02:00 pm 33.7 33.4 32.5 00:00 am 25.3 27.9 29.9 B C 90% Percentage of time in the comfort zone (22-27 °C) Cell 80% 70% 60% 50% 40% 30% 20% 10% Figure 16. Room temperatures in test cell A, B, C, outdoor temperature and Solar Radiation in Round 4. 0% Round1 Round2 Round3 Round4 Figure 18. Percentage of time in the 22 - 27 °C temperature range in each round (day and nighttime). 4.3 Predicted Mean Vote As a further means of comparison, we also considered the Predicted Mean Vote (PMV) in different cells and rounds (ASRHAE 2004). For PMV calculation, we used measured variables together with assumptions regarding metabolic rate (1.2 met), insulation of clothing (0.63 clo), and air velocity (0.10 m.s-1). Figure 21 summarizes the PMV calculations for all rooms in each round. Thereby, the percentage of PMV values within the range of -1 (slightly cool) to +1 (slightly warm) are shown. These results are in agreement with the previous inferences based on Figures 18 to 20. Given a basic level of ventilation, the conditions in the rooms of the cells without the radiant barrier appear to be better than those with the radiant barrier. Likewise, clay roof tiles lead to a slightly better performance as compared to cement tiles. Figure 17. Attic temperatures in test cell A, B, C, outdoor temperature and Solar Radiation in Round 4. 4.2 Distribution of data To compare the different rounds with each other, we considered a "comfort temperature range" of 22 to 27 °C. Figure 18 shows the percentage of time each room was in this range during each round. Cell A showed in the first round (no intended ventilation in attic space) a slightly lower percentage of time in the comfort zone, while in all other rounds, A shows a 198 70% 90% 60% 50% 80% Percentage Percentage of time in the comfort zone (22-27 °C) 100% 70% 60% 40% 30% 20% 50% 10% 40% 0% 30% Round1 A 20% Round2 B Round3 C Round4 Figure 21. Percentage of PMV-Values in the range of -1 to +1. 10% 5 CONCLUSION 0% Round1 A Round2 B Round3 C Round4 The application of the cool roof system with the radiant barrier does have a slight positive effect in the case of attic spaces with very low ventilation rates (round 1). However, this effect disappears with increased ventilation (rounds 2 to 4). Indeed, in the latter rounds, cells without radiant barriers performed slightly better in terms of room temperatures, percentage of time in the 22 – 27°C temperature range, and PMV values between -1 and +1. Clay roof tiles displayed a slightly better performance in comparison to cement tiles. In general, it should be noted that the application of radiant roof barriers in combination with counter batten construction has to be evaluated on a case by case basis. In case of highly air-tight roofs and attic spaces of buildings with air-conditioning systems, radiant barriers may have a net positive thermal effect. However, the application of radiant roof barriers in case of ventilated attics may be counterproductive, if the night-time radiant cooling of the building roof is diminished. Outdoor Figure 19. Percentage of time in the 22 - 27 °C temperature range in each round (daytime from 07:00 am – 07:00 pm, no ventilation though door). 100% Percentage of time in the comfort zone (22-27 °C) 90% 80% 70% 60% 50% 40% 30% 20% 6 ACKNOWLEDGMENT 10% The authors would like to thank Andreas Wurm for contributing to the preparation of the graphs and diagrams used in this contribution. 0% Round1 A Round2 B Round3 C Round4 Outdoor Figure 20. Percentage of time in the 22 - 27 °C temperature range in each round (nighttime from 07:00 pm – 07:00 am, ventilation via entrance door). 199 7 REFERENCES ASHRAE 2004. ASHRAE 55-2004: American Society of Heating, Refrigeration and Air-Conditioning Engineers. Michels C., Lamberts R., and Güths S. 2008. Theoretical/experimental comparison of heat flux reduction in roofs achieved through the use of reflective thermal insulators. Energy and Buildings 40, 2008, pp. 438-444 Miller W., Cherry N., Allen R., Childs P., Atchley J., Levinson R., Akbari H., and Berdahl P. 2010. Field Experiments to evaluate coolcolored roofing; Monier California Research Center Müller V. 2012. Thermal Implications of Radiant Roof Barriers: A Field Study in hot-humid climatic conditions. Master-Thesis, Vienna University of Technology. Suehrcke H., Peterson E.L., and Selby N. 2008. Effect of roof solar reflectance on the building heat gain in a hot climate. Energy and Buildings 40, Elsevier, pp. 1487-1497. Vandewater J. 2007. Above Sheating Ventilation in Tile Roof Installations; The Cool Colors project http://coolcolors.lbl.gov/ 200 An analysis of indoor environmental quality in an office: The case of an university campus in Istanbul P. Sunar & S. Tanrıöver Bahçeşehir University, Faculty of Architecture and Design, Istanbul, Turkey ABSTRACT: This study was focused on indoor environmental quality in an office. An existing office space in an university campus was addressed regarding its indoor environmental quality. The key features such as lighting, thermal comfort, and environmental noise in the space have been studied. Technical data including the architectural drawings, engineering data regarding HVAC, lighting and acoustics was obtained and analyzed; intervallic measurements of indoor lighting, noise levels, humidity, air velocity and temperature was completed, and finally occupants were asked to fill in a questionnaire concerning the key factors mentioned above and their personal experiences related to indoor environmental conditions in the space. By conducting this study, existing conditions were assessed, concretized and solid feedbacks were collected for the renovation to enhance the work environment. Furthermore, this study constituted an example for the analysis and assessment of an active office space, and an example to display the relationship between the measurements taken, the standards and the users' comments. 1 INTRODUCTION Ambient temperature which is one of the compound of the thermal comfort has significant effect on office occupant work performance. In Wang, Federspiel and Arens (2005) study, there is a direct correlation between temperature and occupant satisfaction degree. Several studies that were carried out in commercial buildings indicated that too cold and too warm workspaces may reduce the occupants work performance. Witterseh, Wyon and Clausen (2002) reported a study in an office setting to compare the relation between performance and temperature rise. The results revealed that occupants had difficulty in thinking and concentrating and it is also affected their performace. Another important parameter for thermal comfort is local air movement. In Wargocki, Wyan, Sundell, etg.(2000) study, occupants working performance increased with rising up the air movement rates and also significant decrease in sick building syndrome were observed. A model was developed by Fountain, Avens, de Bear, etg., 1994, to measure the percentage of satisfied occupant when the air movement is locally controlled. The results revealed that occupants preferred locally controlled air than centrally controlled HVAC system. The majority of occupant satisfaction studies related to the lighting conditions involve surveys and experiments in office buildings (Veitch, Newsham, 2000; Boyce, Veitch, Newcham, 2006). These studies point out the direct effects of the luminous conditions on the occupants in terms of visual comfort and perception of the occupant, and the visibility of the task. In metropolitans, people do spend nearly 90% of their working times in office buildings which, in the last 40 years turned into massive living organisms depending on the active systems providing occupants settings to work and live. When the aforementioned amount of the time spent indoors is considered, the affects of the indoor environmental quality and occupants’ physical and psychological needs and their satisfaction regarding physical setting turn out to be significant for increasing satisfaction and productivity. Recent studies have shown that indoor environment quality (IEQ) which depend on several parameters such as thermal comfort, lighting and environmental noise and acoustics have significant effects on occupants satisfaction and work performance in office buildings. Thermal comfort satisfaction is effected by the results of correlation between the human and the environmental factors in buildings. There is also subjective response or state of mind factor which is influenced by contextual, cultural factors and person’s sense of thermal comfort. In the assessment of thermal comfort, two models are described in ISO 7730 and ASHRAE 55 standards. These are PMV (Predicted Mean Vote) method and PDD (Predicted Percentage of Dissatisfied) index, the use of the PMV includes the information about environment factors; ambient air temperature, relative humidity, mean radiant temperature, air speed and human factors; metabolic heat production and clothing insulation. PDD index that is used for evaluating percentage of thermally dissatisfied occupants in spaces (Olesen, Berger). 201 pants dissatisfied with the tonal components in low and high frequencies in offices. Another research pointed out the occupants satisfied when they can control the noise level in their working place (Cohen, Weistein, 1981). An important parameter for acoustic satisfaction and job performance is related to the office setting. The survey conducted by Jens, Arens and Zageus (2005) studying noise levels and speech privacy in different office settings (private office, shared office, cubicle with high partitions, cubicle with low partitions or open office environment without partitions), revealed that occupants were dissatisfied with the lack of speech privacy and high noise levels; besides, acoustic satisfaction varied between different office types. The results of surveys and experiments revealed that that luminous environment limits visual performance due to the time-dependent conditions. Office occupants’ work performance and motivation to the task depend on the visual performances (Boyce,Veitch, Newcham, 2006). As referred to by Rashid and Zimrig (2008), some types of work depends very highly on excellent vision, therefore work performance varies with lighting levels and quality as reported in the study by Romm and Browning (1994) in which 6% increase in the performance of postal workers have been reported in increased lighting conditions. In another case, more rapid production of drawings by a drafting group has been reported after the bright reflections were reduced. Concerning user control of lighting conditions, an investigation done by Moore, Carter and Slater (2002) on how the occupants of open-plan offices perceived locally operable lighting systems. Results provided evidence to suggest that some users are pro-actively using lighting controls to set their preferred conditions, not in response to their discomfort. Besides the visual effects of lighting, biological effects have been investigated by Mills, Tomkins and Schlangen (2007). Employees' wellbeing, functioning and productivity were tested with a newly developed fluorescent light source with a high correlated color temperature. As a result, significant improvement in concentration, alertness, and work performance; and decrease in fatigue and daytime sleepiness were recorded. Many studies referred to by Rashid and Zimring (2008), physiological and psychological benefits of daylight in office environments are worth mentioning. Markus, cited by the scholars, reported that 96% of workers preferred to work under natural light as opposed to electric light and 86% of them preferred having sunshine in their offices. A study carried out by Baron, Rea and Daniels (1992) and referred to by Rashid and Zimring (2008), reported that lighting conditions in an office can also have a significant effect on the employees' social relations in an office. Occupants under warmwhite light had stronger preferences for resolving interpersonal conflicts rather than the occupants under cool-white light. Significant number of studies have investigated noise sources and their effects on occupant’s acoustic comfort, satisfaction and health in office buildings (Hay, Kemp, 1972; Lee, 2010; Rashid, Zimring, 2008). Researches revealed that the most annoying sources of noise in offices are; people, telephones and mechanical equipment and they affect occupant’s concentration and performance depending on the nature and complexity of the work (Rashid, Zimring, 2008). Landström (1995) conducted two different studies about the noise level and tonal components which are exposed in working environments during a working day. Results indicate that occu- In the literature review, it was observed that besides environmental factors, office occupants’ indoor environmental satisfaction is as closely related with the office setting. In a study by Ahlin, Westlender (1991) and Duffy (1999) office types were classified in seven categories; cell, shared room, small open plan, large open plan, flex and combined offices. The IEQ Survey generated by the Center for the Built Environment (CBE) at the University of California, Berkeley has been conducted to a number of occupants of the office buildings, in which the occupants have defined and designated the office spaces due to the degree of openness and enclosure, as enclosed private, enclosed shared, cubicles with high or low partitions and open offices (Lee, Guerin, 2010) The main purpose of this study was to assess the environmental quality of the office and the occupant’s satisfaction in a university campus and to compare the results of the measurements and occupants' comments with standards set for thermal, visual and acoustic environments. Furthermore, occupants' evaluations regarding their own working spaces in terms of these three parameters in their office settings enabled researchers to gain solid feedbacks for the renovation to enhance the conditions of the work environment. 2. MATERIALS AND METHODS 2.1 Materials Physical Setting The office space studied is the faculty of architecture and design of a university in Istanbul. The office is situated in one of the three renewed warehouses positioned side by side on the shore. These three buildings are in the shape of long, narrow rectangular prisms sited very close to each other and are interconnected. 202 These buildings face the garden on the shore and sea with one of their narrow sides, and a high traffic artery of the city with the other (Fig.1.) Figure 3. Pictures of the ceiling. All rooms and open offices utilize 60x 60cm ceiling mounted fluorescent lamp fixtures, with 4 fluorescent tubes in each with 18 W. These fixtures were placed regularly on the ceiling for general illumination. In total, central space is illuminated directly with 10 fixtures located right above the desktops, and 6 others indirectly from the transitional areas on two sides. Lighting conditions in the central space can be controlled from four different points. In the enclosed offices, numbers of lighting fixtures vary from two to three according to the size of the room. A-type closed offices have 3 and B type rooms have two ceiling mounted fixtures and are controlled from one point in each room. Figure 1. Left;The site photo of Bahçeşehir University.Right; The ste drawing of Bahçeşehir University. The office occupies 1/2 of the 1st floor of Block B (Fig.2.a) and faces sea on the East, and other two neighboring buildings (buildings A and C) on the north and south. On the north, rooms look out to a narrow atrium space with a glass roof in between A and B Blocks; the others on the south, to a clearance of approximately 2m wide in between the buildings B and C. (where the exhausts of air conditioning systems of two buildings are given out) There are two different types of office spaces in the faculty. Open-offices with semi-enclosed partitions in the central area and are used by the teaching/research assistants; the closed offices surrounding the central area from three sides, shared by two or three academic staff (Fig.2.).Closed offices, with regard to their orientation as A and B as shown in Fig.2, create 2 sub groups within closed office type. Occupants The users of the office differ according to age, gender and the academic title. The ages vary from 23 to 67; gender from female to male; academic titles from teaching assistants to professors and administrative titles, from chairperson to dean of the faculty. The determinant factor for the locations of the offices are occupants' academic and administrative title, as mentioned in the previous sections. Work Duration According to the academic calendar except August, office is highly occupied and active whole year. Office work hours are 08.30 - 17.30, with an hour lunch break at noontime. Since all users are teaching staff and attend to the classes in different times of the day, time spent in open or enclosed offices by each changes according to the syllabuses. Figure 2. Left; Location of the office setting, Right; The plan layout of Architecture and Design Faculty. 2.2 Methods B type closed offices on the East side are occupied by the dean, vice deans and chairpersons of the departments, and have direct access to natural light and ventilation. Others on the North and South sides are used by the teaching staff and have limited access to natural light and ventilation. The openoffices are occupied in the central area on the other hand, by the research assistants and have no access to natural light and fresh air (Fig.3). Therefore, the indoor comfort conditions here have been enhanced and balanced by the use of artificial HVAC system and lighting fixtures. The study was conducted in three phases; review of the literature regarding the three key factors of indoor comfort conditions in office spaces, analysis of technical data of the office analyzed and finally, evaluations and comments of the occupants collected via questionnaires and short interviews. 203 2.2.1.Technical Data Compilation Method 2.2.2. Statistical Analysis Method Thermal Comfort Thermal measurements (temperature and humidity) were taken from the point where seat is located for each work unit, at the desk height. The measurements were repeated for each work unit in all three offices (A, B and C) in 3 different times of the day (morning, noon, afternoon). In thermal comfort measurements occupants’ mean metabolic rate (met) and clothing insulation (clo) values were not taken into consideration. Questionnaire The questionnaire was prepared in reference to the recent literature on the subject and the IEQ Survey made by the Center for the Built Environment (CBE) at the University of California, Berkeley. Questions were arranged in four parts such as, personal information, physical characteristics of the workspace, assessment of the existing conditions in the workspace and the user satisfaction levels regarding key components of indoor environmental comfort conditions. Fifty-five occupants of the faculty from different age groups, gender and academic titles have been given the questionnaire of 34 questions in total. Questions in the first part were prepared with the intention of gathering information about users’ age, gender and the academic title; and time spent in the office per week and the total months/years spent in the office. Ones in the second part focused on the physical conditions of the office such as type, location, and orientation of the office space. Third part comprised of questions that define the existing conditions regarding thermal comfort, lighting and noise control conditions. Finally, last part targeted at recognizing whether or not the occupants are satisfied with their office environment. The results were recorded and analyzed with the statistical analysis program SPSS, and were studied with t-tests and f-tests. Data was analyzed with respect to the office type and the time spent in the office. Figure 6 reveals the general levels of satisfaction/dissatisfaction in terms of thermal, visual and acoustic conditions in the space. Visual Comfort/Lighting Illumination in the office has been also been evaluated in 3 different times of the day (morning; 10.0010.11, noon; 13.00-13.12, afternoon;16. 26- 16.36). Significant changes in the values have been observed especially in the rooms having outside connections. The amount of illumination on each user’s desktop, which is approximately 75 cm high from the ground, has been measured with Sekonic L-358 light meter (Fig.4). Figure 4. Left: Sekonic light meter L-358, Right: Plan of the sensor locations (light meter, sound level meter). Audial Comfort/Noise Control Same as the other environmental factors, measurements regarding the environmental noise in the space were taken in the morning (10.00-10.11), noon (13. 00- 13. 12) and afternoon (16. 26- 16. 36) times of a workday, with closed window conditions. These periods were decided in reference to the standard occupation times of the users. Sound level meter and sound pressure equipments (Fig.5) were set in the middle of the each office and central area with the height of 150cm (ear level) and then used for comparison with the standards IEC 60651 and IEC 60804. 3. RESULTS AND DISCUSSION Questionnaire When data is analyzed in detail general satisfaction regarding thermal comfort appeared as 22%. Results regarding temperature and humidity varied due to air movement, in terms of the variable, office type. Satisfaction regarding temperature and humidity in offices A and B were about 80%, in contrasts with the offices C where dissatisfaction regarding temperature and humidity was more than 60%. Concerning air movement, occupants of office C are much more (80%) dissatisfied than the occupants in offices A and B (50%) when compared (Fig.7). The results of temperature and humidity perception of the users did not indicate a significant affect however, the impact of air movement changed due to the time spent in the offices. The occupants which spend 10 or less and 11-30 hours in their workspaces complained about the air movement (Fig.8.). Figure 5. Acoustic sound level meter and sound pressure equipments. 204 In terms of lighting, the evaluation of the data reveled that approximately 22 % of the occupants were satisfied and 45 % were dissatisfied with the existing lighting conditions in the office. When data analyzed according to the office types, occupants of A and B types appeared to have lower satisfaction levels with about 45% when compared to the occupants of office C with satisfaction levels above 50 % (Fig.2). Regarding the time spent in the offices. occupants which spent 10 or less and 11-30 hours in office are the most dissatisfied groups with 85% and 76%. The less dissatisfied group is spent more than 30 hours (40%) in office (Fig.8.). Technical Data Compilation Measured indoor temperatures ranged from 26, 07°C to 27, 35°C (Fig.9) and the relative humidity ranged between 39, 63% to 45, 88% in all offices (Fig.10). According to the ASHRAE 55 standards, acceptable levels for indoor temperatures and humidity levels for occupants’ comfort are between 20- 25, 5°C and 30- 60%. In Turkish Regulation (TS 825) ideal temperature and humidity levels are stated to be differentiating between 20- 26°C and 30- 70%. When average temperatures and humidity levels are evaluated due to the Turkish Regulation (TS 825) and ASHRAE 55 standard, it was observed that measured average indoor temperatures are over the acceptable limits but the humidity levels are in the limits of comfort zone. Comparing the data of satisfaction and measurements, the most dissatisfied occupants about temperature and humidity are the users of office C, on the contrary to the measured values. Another significant parameter for thermal comfort is air movement. Occupants appeared to be highly dissatisfied with the air movement in offices. Figure 6. Percentage of satisfaction degree for IEQ conditions. in their offices. Figure 7. Office occupants’ dissatisfaction about IEQ parameters in their workspace. Figure 9. Measured temperature levels (C°). Considering the measurements about the illumination levels in offices, it is observed that the highest illumination levels are obtained in office B (Fig.11.). Office A and C average lux values (265, 45; 211, 99) are below the standard levels. It is stated in TS EN 12464 regulation that the illumination level in offices should be 500 lux. Correlation between the statistical data for satisfaction and measured illumination levels are compatible with each other according to the office types. The most dissatisfied occupants are the users of office C (above 50%) and measured average lux value 211, 99 is far below the standards. Figure 8. Dissatisfaction percentage of occupants due to the time (hour) spent in offices. When occupants evaluate the noise level and sound privacy in offices, it can be seen that open office (C) occupants were the most dissatisfied group. Satisfaction level about the noise is 10 % in office C and is above the 50% in office A and B (Fig.7.). Assessment the impact of the noise alteration due to the time spent in offices, the data indicate that the 205 Figure 12. Measured noise levels (dB) in all office types throughout the day. Figure 10. Measured humidity levels (%). 4. CONCLUSION The survey focused on indoor environmental quality of an office space located within the body of a university, and aimed at assessing and concretizing existing conditions of the workspace by determining feedbacks from its users, collecting data by taking measurements regarding the key features of indoor environmental quality; lighting, thermal comfort, and environmental noise and comparing the data with related standards and codes. Among the features mentioned above, thermal comfort and environmental noise appeared to influence the perception of environmental quality the most. Although the measured values of temperature was slightly higher than the acceptability range indicated by TSE and ASHRAE, and humidity stayed with the acceptability range, still the occupants stated their discomfort especially in Office C. Air movement was the content criticized by the occupants, strictly. Therefore the further study will include the air movement measurements and ‘met’ and ‘clo’ values as for the renovation and enhancement practices of the office. Environmental noise, which usually is the major problem in open-plan offices, appeared to be as significant as the thermal comfort conditions. In the office where the study takes place, sound privacy and noise were the features that the occupants are most dissatisfied with. Although the occupants of Office C seemed to be more dissatisfied than the other users, enclosed-shared offices appeared to have the highest dB values, due to the outdoor environmental noise, coming from highly active urban setting. Lighting, although criticized less than the other features, seemed to lack the most in the work environment when the measurements taken to reveal the existing conditions are compared with the codes and standards of TS EN 12464. As in the other features, occupants of Office C were more dissatisfied with the lighting conditions than the occupants of Offices A and B. Moreover, a significant relationship between the age and illumination levels has appeared in the statistical analysis. Figure 11. Average illumination levels (lux) in all office types throughout the day. According to the Turkish regulation (TS EN ISO 3745); Environmental noise assessment and management, indoor environmental noise level is limited between 45 (close window)- 55 (open window) dBA. Indoors office noise level found to be exceeding the limit in office A (53,5 dBA). In office B’ occupants are exposed to the motor noise and warning beeps of the sea crafts, since the rooms face the seaside. According to the results, value measured in office B is appeared to be higher (60,1 dBA) than the upper limit indicated by the standards for indoor noise levels. The results of three observation times indicated that indoor environment noise in open office (Office C) is 54, 3 dBA and closed offices (A, B) are above the acceptable level for closed window conditions (Fig.12). In the statistical analysis, dissatisfaction regarding environmental noise in the office is about 85%, and the most satisfied group appeared to be the occupants of office C. Despite being enclosed offices, the high values recorded during the measurements were predicted to occur due to the sudden singular events such as horns of the sea craft, speech noise, falling object, phone ring etc. 206 Landström U., Akerlund E., Kjellberg A., and Tesarz M. 1995. Exposure Levels, Tonal Components and Noise Annoyance in Working Environments, Environment International, pp.265-275. Lee Y.S. 2010. Office Layout Affecting Privacy, Interaction and Acoustic Quality in LEED- Certified Buildings, Journal of Building and Environment, pp. 1594- 1600. Lee Y.S., and Guerin D.A. 2010. Indoor Environmental Quality Differences Between Office Types in LEED- Certified Buildings in the U.S., Journal of Building and Environment, pp. 1104- 1112. Mills P. R., Tomkins S.C., and Schlangen L.J.M. (2007). The effect of high correlated color temperature office lighting on employee well being and work performance. Journal of Circadian Rhythms, 5, 2. doi: 10.1186/1740-3391-5-2. Moore T.A., Carter D.J., and Slater A.I. 2002. A field study of occupant controlled lighting in offices. Lighting Research and Technology vol 34 issue 3 pp 191-205 Olesen B.W., and Brager G.S. 2004. A Better Way to Predict Comfort, ASHRAE Journal, pp. 20-26. Rashid M., and Zimring C. 2008. A Review of the Empirical Literature on the Relationships between Indoor Environment and Stress in HealthCare and Office Settings, Journal of Environment and Behavior, pp. 151-190. Romm J.J., and Browning W.D. 1994. “Greening the Building and the Bottom Line: Increasing Productivity Through Energy-Efficient Design,” Snowmass, CO: Rocky Mountain Institute. TS, 2008. Thermal Insulation in Building, Standart, 825. Ankara TS, 2012. Acoustics - Determination of sound power levels and sound energy levels of noise sources using sound pressure Precision methods for anechoic rooms and hemi-anechoic rooms, EN ISO 3745. Ankara. TS, 2013. Light and lighting - Lighting of work places - Part 1: Indoor work places, EN 12464. Ankara. Veitch J.A., and Newsham G.R. 2000. Preffered Luminous Conditions in Open Plan Offices: Research and Recommend tions, Journal of Lighting Research and Technology, pp. 199-212. Wang D., Federspiel C.C., and Arens E., 2005. Correlation between Temperature Satisfaction and Unsolicited Complaint Rates in Commercial Buildings, Indoor Air, pp.1318. Wargocki P., Wyon D., Sundell J., Clausen G., and Fanger P.O. 2000. The effects of outdoor air supply rate in an office on perceived air quality, sick building syndrome (SBS) symptoms and productivity, International Journal of Indoor Air Quality and Climate, pp.222- 236. Witterseh T., Wyon D., and Clausen D. 2002. The Effects of Moderate Heat Stress and Open-Plan Office Noise Distraction an Office Work. The results put forth that the thermal conditions and noise are more influential in the satisfaction of the users of than lighting conditions in a workspace. In addition, the study as a whole exemplifies a methodology for the investigation, analysis and assessment of an active office space, which reveals the relationship between the users' comments, measurements taken, and the codes and standards. Hopefully, the results revealed and discussed above will be used in the renovation to enhance the work environment. 5 ACKNOWLEDGMENTS The support from BUTECH (Technology Development Center of Bahçeşehir University) to this research study is gratefully acknowledged. We would like to thank Nil Girgin Kalıp, Kaan Alper and Çınar D. Kurra for their comments and technical assistance. Thanks are due to all occupants who responded to the questionnaire. 6 REFERENCES Ahlin J., and Westlander G. 1991. Kontorslokaler ochkontorsarbete—två perspektiv på kontoretsom a rbetsplats[Office spaces and office work—two perspectives on the office asa workplace]. Solna, Sweden: Arbetsmiljöinstitutet (The Swedish National Institute forWorking Life). ASHRAE 2001. Thermal Comfort Conditions for Human Occpancy, Standart 55-81. Atlanta Baron R. A., Rea M. S., and Daniels S. G. 1992. Effects of Indoor Lighting (illuminance and spectral distribution) on The Performance of Cognitive Tasks and Interpersonal Behaviors: The Potential Mediating Role of Positive Affect. Motivation and Emotion, 1, 1-33. Boyce P.R., Veitch J.A., Newsham G.R et al. 2006. Lighting Quality and Office Work: Two Field Simulation Experiments, Journal of Lighting Research and Technology, pp. 191-223. Boyce P.R., Veitch J.A., Newsham G.R., Jones C.C., Heerwagen J., and Myer M. 2006. Occupant use of switching and dimming controls in offices. Lighting Research & Technology;38:358-378. Chung T.M., and Burnett J. 2000., Lighting Quality Surveys in Office Premises Journal of Indoor Built Environment, pp. 335- 341. Cohen S., and Weinstein N. 1981. Nonauditory Effects of Noise on Behavior and Health, Journal of Social Issues, pp. 36-70. Duffy F. (1999). The new office (2nd ed.). London: Conran Octopus Limited Hay B., and Kemp M. F. 1972. Measurements Of Noise in AirConditioned Landscaped Offices, Journal of Sound and Vibration, pp. 363-373. ISO, Ergonomics of the thermal environment – Analytical determination and interpretation of thermal comfort using calculation of the PMV and PPD indices and local thermal comfort criteria, 7730, Switzerland. Jensen K.L., Arens E., and Zagreus L., 2005. Acosutical Quality in Office Workstations, as Assessed by Occupant Surveys, Proceedings, Indoor Air 2005, Sept. 4-9, Beijing, China. 207 208 Thermal environment in detached houses with atrium: Towards proper utilization of atrium space in traditional dwellings “Kyo-machiya” C. Iba, S. Hokoi & D. Ogura Kyoto University, Department of Architecture and Architectural Engineering, Kyoto, Japan S. Ito Takenaka Corporation, Tokyo, Japan ABSTRACT: With the aim of realization of comfortable and energy-saving indoor environment in traditional dwellings, which have generally atrium space, the influence of atrium on the indoor thermal environment was evaluated. First, the measurements of temperature and airflow distributions were conducted in a real-scale detached house in summer and in a different house in winter, separately. And then the measurements were simulated by using computational fluid dynamics analysis. From the results of the summer measurements, when the windows were closed and air conditioner was operated in the house, the cooler air blown from the air conditioner tended to move downward through the atrium. Thus the upper floor was seemed to not be effectively cooled. Furthermore, the calculation results for summer condition clarified that the cross ventilation might be better in the case without an atrium under certain summer condition. The measurement and calculation for the winter condition showed that the combination of the atrium and staircase could cause an undesirable cold draft. Thus, it can be summarized that an atrium necessarily does not have good influence on the indoor thermal environment. Undesirable airflow can be avoided by properly opening or closing the partitions between the atrium space and the living space. The information obtained in these investigations will be made use of the reasonable utilization technique of the atrium space in traditional dwellings. indoor thermal environment was emphasized, particularly focusing air flows in a house. Indoor environmental factors such as temperature and airflow rate were measured in two real-scale detached houses with an atrium, and then, the airflow and the temperature distributions in the house were simulated by using computational fluid dynamics (CFD) analysis. The analysis model was verified by comparing the calculation results with the measurement results. By using this analysis model, the cases with and without an atrium were calculated. 1 INTRODUCTION Traditional dwellings “Kyo-machiya” in Kyoto, Japan, generally have atrium space, which is expected to be a path of breeze and daylight. However, the effect of atrium spaces on the indoor thermal environment has not been clarified, particularly concerning with the effect in winter. Also in modern detached houses in Japan, the number of people who prefer an atrium space have increased, because atrium can provide an open, spacious and bright space to their houses. Furthermore, an atrium has an advantage that the residents who stay upstairs and downstairs can easily communicate with each other. On the other hand, it is quite difficult to design air conditioning system effectively for such modern houses with an atrium space. For example, the air in upper floors tends to become too warm in summer, while the air temperature in lower floors is hard to rise in winter. This is because the warmer air moves upward in the atrium space (Nishimori & Nakamura 1998, Matsuoka & Matsumoto 2007, Noda et al. 2007). A major goal of our research is to propose proper ways of living, which enable comfortable and energy-saving indoor environment in traditional dwellings. In this paper, the effect of atrium space on the 2 INDOOR ENVIRONMENT IN SUMMER In this section, an indoor environment measurement in the summer season and the corresponding simulation are described. 2.1 Measurement of indoor environment The measurement of summer condition was carried out in August 18th, 2009 in a two-story gabled house (call this house ‘House A’) in Kyoto, Japan. The weather was fine on that day, and the outdoor air temperature was kept at approximately 32 °C during the measurement. 209 Table 1. Properties of windows. Window Opening size No. Width Height W-1 Opening type 820 × 2030 Sliding window W-2 1640 × 570 Sideslip window W-3 640 × 570 Sideslip window W-4 640 × 570 Sideslip window W-5 405 × 2230 Projected window W-6 1640 × 570 Sideslip window W-7 820 × 1370 Sliding window W-8 820 × 1370 Sliding window W-9 820 × 2030 Sliding window Figure 1. Plans and measuring points of House A (Mesurement in summer). Figure 1 shows the plans of House A. The first floor of this house has a large space consisted of kitchen, dining/living room and Japanese style room. Two rooms on the second floor are connected by a corridor which faces to an atrium. Each room can be open to the atrium by opening the partition. An air conditioner (AC) is installed in each floor near the ceiling. The property of each window is shown in Table 1. Previous research suggested that the sizes and locations of the windows have a significant influence on airflow and thermal comfort (Ravikumar & Prakash 2011). In order to examine the influence of window opening pattern, nine windows of this house and the air conditioners were operated as follows; − Pattern 1: All windows were closed, and each air conditioner was ON (cooling, set at 27°C). − Pattern 2: All windows were open, and each air conditioner was OFF. − Pattern 3: Windows on the 1st floor were closed, and windows on the 2nd floor were open. Each air conditioner was OFF. 35 Pattern 1 34 Pattern 2 Pattern 3 Temperature [°C] 33 Measuring point 2 (2cd floor: north) 32 31 Measuring point 3 (2cd floor: center) 30 29 28 Measuring point 1 (1st floor) 27 26 14:45 14:30 14:15 14:00 13:45 13:30 13:15 13:00 12:45 12:30 12:15 12:00 11:45 11:30 25 Time Figure 2. Temperature profile (Measurement in summer). Figure 3. Simple streamer setup. 2.1.1 Measurement of indoor temperature The air temperature and relative humidity were measured at the three points shown in Figure 1. The air temperature and the relative humidity were measured using a portable thermo-hygrometer which has a thermistor temperature sensor and a polymer humidity sensor. The measurement interval was set 5 min. In this paper, the results of humidity are omitted. Figure 2 shows the temperature profiles at three measurement point shown in Figure 1. The temperatures on the 2nd floor were nearly constant in Pattern 1, while the temperature on the 1st floor fluctuated according to ON/OFF operation of the AC. The temperature at the measuring point 3, which is at the centre of 2nd floor, the temperature is about 1 °C higher than that at the other points. As the operation of AC on the 1st floor became stable, the temperature on the 1st floor became the lowest, probably because cool air was blown from the AC on the 2nd floor and moved to the 1st floor through the atrium. After the ACs were turned off and the all windows were open (in Pattern 2), the indoor temperatures increased by about 4 °C within 40 min. 2.1.2 Measurement of air flow distribution Indoor airflow speed was measured at several points in the house using a hot-wire anemometer. At the same point, airflow direction was checked using a simple streamer (Figure 3). The air flow distribution in each pattern is shown in Figures 4, 5 and 6 (in the next page). The arrow directions were corresponding with the air flow directions, and the lengths of arrows indicate the airflow speed. In pattern 1, where all the windows were closed and ACs were on, the cool air from the AC on the 1st floor flowed straight ahead and reached the op210 posite wall (Figure 4). On the other hand, as conjectured from the measured temperature, the cool air from the AC on the 2nd floor flowed downward to the 1st floor through the atrium. In pattern 2 (Figure 5), where all the windows were open and ACs were off, the main flow on the 1st floor was seen from W-1 (south side) to W-5 (east side). On the 2nd floor, the air flowed mainly from W-8 (north side) to W-9 (south side). Airflow was observed in the whole house, along with the flow through the atrium. In pattern 3 (Figure 6), where only the windows on the 2nd floor were open and the ACs were off, strong air flow was measured only on the 2nd floor. Airflow through the atrium was observed slightly. Height above the floor: 300mm 1,000mm 1,800mm Figure 4. Airflow distribution (Pattern 1; 11:35-12:00). Height above the floor: 2.1.3 Summary of measurement When the ACs were operated, the cool air from the AC flowed downward through the atrium. As a result, the 2nd floor was not sufficiently cooled, while the 1st floor was overcooled. Such a situation is considered generally undesirable in terms of indoor thermal comfort and saving energy. This house was adjacent to the traffic road on the east and south sides. Therefore, when the windows were open and a large car passed near the house, particularly strong airflow was observed. When the windows were open on just one floor with the ACs off, airflow appears only on that floor. On the other hand, air flows in the whole house when the windows on both floors open. However, in this case, the influence of the atrium on the indoor airflow was unidentified. 300mm 1,000mm 1,800mm Figure 5. Airflow distribution (Pattern 2; 13:40-14:15). Height above the floor: 300mm 1,000mm 1,800mm 2.2 CFD analysis of indoor airflow Indoor airflow obtained in the measurements was simulated using Computational Fluid Dynamics (CFD) analysis. The software STAR-CD 4.10 was used for the simulation. Figure 6. Airflow distribution (Pattern 3; 14:30-14:45). 2.2.1 Calculated model and condition In this analysis, the standard k-ε turbulence model was used with the assumption of an incompressible flow. The choice of the k-ε turbulence model seemed a good compromise between the realistic description of the turbulence and computational efficiency (Jones & Whittle 1992). House A was modeled as shown in figure 7. The case of window opening pattern 3 is indicated in this figure. 40 × 40× 40 grid cells were used for the simulation. The measured results were used as the boundary conditions; the inlet air temperature and air speed were set at 32 °C and 1.0 m/s, respectively. No-slip condition was applied to the external walls, while their surface temperature was set at 32 °C. At the partition walls, no-slip and no-heat-flux conditions were used. Bird’s-eye view Figure 7. Simulated model for House A (Pattern 3) 2.2.2 Comparison between the calculated and measured results Figure 8 shows the calculated airflow on the 2nd floor for Pattern 3. In comparison with Figure 6, the airflow direction and speed agree mostly with the measured results. In both results, main airflow occurred along the corri211 dor, and strong airflow observed near W-8. On the other hand, there are some differences in the air flow direction particularly in low-air-speed areas; this might be partly due to the difficulty in identifying the wind direction in the measurements. The calculated temperatures are shown in Figure 9. Measured temperatures in Pattern 3 are added to these figures. It is seen that there was 0.6°C difference between the measured and calculated results at the centre on the 2nd floor. This is presumably because solar radiation was neglected in the calculation, or because the temperatures on 2nd floor had not reached the steady state in the measurement. However, it can be said that the calculated results are generally in harmony with the measured results. Figure 8. Calcurated airflow (Pattern 3; 2nd floor) (unit: m/s). 32.0°C 31.4°C Without atrium N N 31.4°C 30.2°C 32.0°C Vertical distribution Horizontal distribution on 2nd floor * Measured data are shown in the box Figure 9. Calcurated temperature (Pattern 3). Inlet Inlet: W-8 Window 8 Inlet Inlet: W-8 Window 8 N S Outlet Outlet: W-1 Window 1 With atrium Atrium Outlet Outlet: W-1 Window 1 No atrium Without atrium Figure 10. Calculation model with and without atrium. flowed into the house more than in the case without atrium. In the case with atrium, the greater part of the inlet air goes straight to the south wall, while the rest flows downward through the atrium. Small part of 2.3.2 Calculated results The calculated airflows in both cases are shown in Figure 11. In the case with atrium, the outdoor air With Atrium 31.4°C 30.5°C Here, we simulate the cases with an atrium and without an atrium for House A. By comparing these results, we can examine the influence of an atrium and a staircase on the airflow crossing the floors. Airflow distribution on 2nd floor 31.3°C 31.4°C 2.3 Influence of atrium space on indoor airflow 2.3.1 Calculated model and condition The calculation model with and without atrium space are shown in Figure 10. For investigating the cross ventilation between floors, air inlet was set on the 2nd floor (W-8), and air outlet was set on the 1st floor (W-1). In this calculation, a steady, isothermal flow was assumed. The air flowed into the house through W-8 (at a pressure of 1 Pa), and blown out through W-1 (at a pressure of 0 Pa). At the wall surfaces, no-slip and adiabatic (no heat flux) boundary conditions were assumed. 31.3°C Airflow distribution on 1st floor Vertical distribution of airflow N N Figure 11. Comparison of airflow distributions between in case with atrium and without atrium (unit: m/s). 212 3.1.1 Measurement of indoor temperature The temperatures were measured at several points in this house in the same way described in Section 2.1.1. Figures 13 and 14 show the temperature profiles from 10:30 to 15:30 on the 1st and 2nd floors, respectively. In the living space (living room (P1), dining room (P2), kitchen (P3), and Japanese style room (P4)), the temperatures rose up slowly after the heating system was heated up, and reached to 20.1°C in average (at 15:00). The temperatures in the entrance (P5) and the utility (P6) sometimes dropped because some visitors opened the entrance door. At the point P7, the temperature dropped for a short while around 12:30 by opening and closing the door leading to the outdoor air. The temperatures on the 2nd floor increased in the same way as those on the 1st floor, and reached to 18.4 °C in average (at 15:00). At the point P7 and P11 which are facing south, the temperatures were about 1 °C higher than those at other points. From the measured temperatures, it can be said that such a radiant heating system realized a com- the southward airflow collides against the partition wall and flows into the northwest closet. The airflow going down through the atrium flows out of W-1. On the 1st floor, strong wind occurs only near the outlet window. A slight airflow goes upward through the staircase. In the case without atrium, large amount of inlet air goes straight to the south wall, and the rest flows down through the staircase. Strong downdraft occurs in the staircase, thus the airflow spreads over the 1st floor. 3 INDOOR ENVIRONMENT IN WINTER The indoor environment in winter was also examined by a measurement in a different detached house and corresponding CFD simulation. 3.1 Measurement of indoor environment The measurement of winter condition was conducted in January 16th, 2011 in a two-story gabled house (call this house ‘House B’) in Kyoto, Japan. Figure 12 shows the plans of House B. The living room on the 1st floor had a large atrium. A staircase, which was installed in the living room, was spatially connected to the atrium. In this house, a floor radiant heating system was equipped. Hot water piping was installed in the space between the ceiling of the 1st floor and the flooring boards of 2nd floor. The heating system was put into operation 3 days before the measurement, and was heated up at 10:00 on January 16th. 25 P4(Japanese style room) Temperature[°C] 20 P1 (Living room) P3(Kitchen) P2(Dining room) P6(Utility) 15 P5(Entrance) 10 5 Outside 15:30 15:00 14:30 14:00 13:30 13:00 12:30 12:00 11:30 11:00 10:30 0 Time Figure 13. Temperature profile on 1st floor. 25 P11(near window) P7(Room 1) Temperature[°C] 20 15 P9(Room 3) P8(Room 2) P10(Closet) 10 5 Outside Time Figure 14. Temperature profile on 2nd floor. 0.14m/s (820mm high) 0.10m/s (410mm high) 0.25m/s (1,230mm high) 0.22m/s (1,640mm high) 0.07m/s (on the beam) 0.05m/s (on the beam) Figure 12. Plans and measuring points of House B (Measument in winter). Figure 15. Airflow distribution at 14:30 (with doors on the 2nd floor closed). 213 15:30 15:00 14:30 14:00 13:30 13:00 12:30 12:00 11:30 11:00 10:30 0 fortable indoor thermal environment on both the 1st floor and the 2nd floor. staircase. However, at the upper position in the corridor, the airflow in the opposite direction occurred. In the headroom of the atrium, the air flowed from the living room to the 2nd floor. Near the beams across the atrium (at the same height of 2nd floor level), irregular airflows were observed. Such airflow patterns are caused by the space structure, the combination of the atrium and the staircase, and the thermal stratification. In this case, the warmer air in the 1st floor tends to move upward, while the cooler air in the 2nd floor flows down. 3.1.2 Measurement of air flow distribution Figure 15 shows the indoor airflows distribution observed when the doors on the 2nd floor were closed. In this case, only the airflows on the 1st floor were measured. The airflow was mainly observed around the atrium and the staircase. In the staircase, the air flowed downward from the 2nd floor to the 1st floor. Figure 16 shows the airflow distribution in the case when all the doors on the 2nd floor were open. It is noted that the air speed became higher than in the case when the doors were closed, especially in the staircase. Figure 17 is a sketch of the vertical airflow distribution at the staircase. At the lower position in the 2nd floor corridor, the air flowed down along the 0.19m/s (1,800mm high) 0.41m/s (820mm high) 0.25m/s (410mm high) 0.14m/s (on the beam) 0.35m/s (1,640mm high) 3.2 CFD analysis of indoor airflow The indoor airflows and temperatures in House B in winter condition were simulated by using CFD analysis. The calculation methods are same as shown in Section 2.2. 0.22m/s (2,650mm high) 0.05m/s (1,000mm high) 0.05m/s (on the beam) corridor 0.40m/s (100mm high) inner Inner garden garden 0.05m/s (550mm high) 0.17m/s (on the beam) Figure 16. Airflow distribution at 14:30 (with doors on the 2nd floor open). Table 2 Wall boundary conditions (Wall temperature) (unit: °C). 1F Floor 1F Wall 1F Ceiling 2F Floor 22.0 21.5 25.5 22.0 A) Airflow on 1st floor Figure 18. Calculated airflow distributions. B) 2F Exterior wall 18.5 Airflow on 2nd floor A) Temperature on the 1st floor B) Temperature on the 2nd floor Figure 19. Calculated horizontal and vertical temperature distributions. 214 Figure 17. Airflow distribution (vertical cross-section at the staircase). C) 2F Partition 20.0 2F Ceiling 19.0 Vertical distribution of airflow C) Vertical distribution also the negative aspect that the hot external air can warm the inside of the house. The measurements conducted in winter showed that a cold draft occurred through the atrium and the staircase even though there was little vertical temperature difference. If the residents use an air conditioner for heating, a stronger cold draft would occur. From the results of the CFD calculations in winter, it was found that the cold air going down through the staircase caused a complicated flow around the atrium. The atrium and the stairs should therefore be separated to prevent this cold draft. The wall boundary conditions were determined based on the measured results, as shown in Table 2. In this calculation, the case with open doors was treated. The calculated airflow distributions were shown in Figure 18 (in previous page). Strong airflows are seen in the living room. The air in the 2nd floor flows toward the staircase, and airflows like a vortex are indicated in the headroom of the atrium. As can be seen in vertical distribution in Figure 18, the air flows down along the staircase, and in the upper area of the corridor, the reverse airflows occurs. The calculated results agree well with the measured results. Figure 19 shows the calculated temperature distributions. The temperature on the 1st floor is within the range from 20.5 to 21.0 °C, and the temperature in the 2nd floor, from 19.5 to 20.5 °C. The calculated results were in good agreement with the measurement. These results suggest that even in a house which has a small temperature difference in the vertical direction owing to the radiant heating system, a strong draft can be generated at the staircase and around the atrium. 5 CONCLUSIONS In order to evaluate the influence of an atrium on cross ventilation, the measurements of indoor thermal environment were carried out in a two-story detached house with an atrium, in summer and in winter. In both measurements, the atrium combined with the staircase had a significant influence on the indoor airflow. CFD calculations were also carried out in order to compare the cases with and without an atrium. The results showed that under certain conditions in summer, the cross ventilation might be better without atrium than with atrium. In winter, the combination of the atrium and staircase causes an undesirable cold draft to flow down through the staircase. If some partitions are installed in proper position and are operated reasonably, undesirable airflow can be avoided. The traditional dwellings “Kyo-machiya”, our main research object has generally an atrium and partitions. In the near future, we are planning to measure indoor environment factors in some Kyomachiyas, and conduct corresponding CFD analysis. Then we will propose the reasonable utilization technique of the atrium space toward more comfortable and energy-saving indoor environment in traditional dwellings. 4 DISCUSSIONS From the measurements in a house with an atrium in summer, it was shown that when the air conditioner was working, the cool air from the 2nd floor went down to the 1st floor. Such airflows can cause the thermal conditions uncomfortable. When the windows were opened for natural ventilation, the wind from the outside mainly flowed only on each floor, and affected slightly the other floor. This was partly because the atrium had a balustrade on the 2nd floor, which prevented airflow. Thus, it cannot be said that an atrium is always suitable for cross ventilation. When only one window was open on each floor, a large amount of air came in through the window. Along this strong airflow, several circulating secondary flows, which were rather weak, were generated. However, the cross ventilation seems to have a cooling effect only in a limited area between the open windows. This means that most of the airflow cannot be used effectively for cooling. In the case without the atrium, although the less air came into the house, sufficient airflow occurred on both floors because of the downdraft through the staircase. Therefore, under certain conditions, the more agreeable indoor ventilation might be obtained in the house without the atrium. When the outdoor hot air comes into the house, the indoor temperature can increase. Therefore, we should take into account not only the positive effects of the ventilation, such as heat removal from the hot interior and wind cooling for thermal comfort, but 6 REFERENCES Jones P. J., and Whittle G. E. 1992. Computational Fluid Dynamics for Building Air Flow Prediction – Current Status and Capabilities. Building and Environment, Vol. 27, No. 3: 321-338 Matsuoka D., and Matsumoto T. 2007. Study on Thermal Environment of Void Space in Detached House. Part.1 Measurement of Vertical Temperature Distribution and Room Air Velocity. Summaries of technical papers of Annual Meeting Architectural Institute of Japan. D-2: 61-62. Nishimori A., and Nakamura Y. 1998. Investigation of Summer Indoor Thermal Environment and Design Method of Detached House with Open Ceiling, Summaries of technical papers of Annual Meeting Architectural Institute of Japan. D-2: 1019-1020. 215 Noda M., Matsumoto T., and Matsui I. 2007. Thermal comfort of the room with void space Part.2 Field measurement of thermal environment in winter. Summaries of technical papers of Annual Meeting Architectural Institute of Japan. D2: 433-434. Ravikumar P., and Prakash D. 2011. Analysis of thermal comfort in a residential room with insect proof screen: A case study by numerical simulation methods. Building Simulation, Vol. 4, Issue 3: 217-225 216 Thermal performance analysis of traditional housing in Kosovo A. Deralla & A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: This paper reports on thermal performance analyses carried out in five traditional residential buildings in Kosovo known as "Kulla". The survey was performed over a one year period. Measurements of indoor environmental parameters and parametric simulation runs were deployed to capture existing conditions and explore possible improvements. Thermal comfort assessments suggest that the indoor environments in Kullas during summer could be perceived as slightly cool (mean PMV = -0.1; mean PPD = 16 %). Simulations of existing Kulla conditions (using standard user assumptions) point to annual heating energy demands of up to 196 kWh.m-2, which could be reduced by 75% via proper thermal improvement measures. 1 INTRODUCTION buildings. Subsequently, the impact of potential improvement scenarios was explored using parametric simulations including a base case and seven improvement scenarios. Traditional architecture around the world provides numerous examples of low-energy yet comfortable buildings. Kullas are considered as environmentally sustainable in terms of suitability to climatic conditions, energy use, and use of local building materials (Rassam 2001). A Kulla is a fortified residential building with massive load bearing stone walls (Drançolli 2001). Past studies on Kullas were mostly focused on historical and typological observations. Aside from a paper by the authors (Rexhepi and Mahdavi 2010), we know of no previous study that specifically addressed the physical and environmental performance of Kullas. The objective of the present study is to further investigate the thermal performance of this type of traditional housing via long-term monitoring of indoor environment thermal parameters in selected buildings as well as assessment of thermal improvement scenarios using calibrated simulation models. 2.1 Overview of case studies Five buildings situated in western Kosovo were selected for this study (Table 1). One (K3) was fully occupied (Full), two (K1, K5) were partially occupied (Part), and two (K2, K4) were unoccupied (None). Kullas are heated in winter. During summer, no cooling mechanism is utilized. Figures 1 to 3 show exterior views of K1, K3, and K5. 2.2 Indoor Environmental Measurements Indoor air temperature and relative humidity levels were monitored over a period of one year. Data loggers were located in various spaces inside four Kullas (living rooms, fire room, and ground floor hall). Measurements took place from August 2008 until July 2009 in 15 minutes time intervals. Figure 4 includes sample floor plans (K1, K2). Outdoor weather data was provided from a local meteorological station in Prishtina (IHMK 20082009), complemented with synthetic weather information generated via Meteotest 2008. 2 METHODOLOGY The main focus of the present study was the thermal performance of Kullas. Toward this end, we collected indoor environmental data (via installation of data loggers in selected objects) as well as weather data obtained from a local meteorological station. The collected data was analyzed and interpreted in view of the thermal performance of and comfort conditions in the buildings. Moreover, the collected data was used to calibrate digital performance simulation models of the selected Table 1. Overview of selected buildings with approximate date of construction, floor area, and occupancy state. Code K1 K2 K3 K4 K5 Construction date 1800 1810 1800 1825 1850 Total floor area [m²] 228 263 303 183 220 Occupancy state Part Non Full Non Part 217 2.3 Measurements data analysis The collected long term data was deployed to assess Kullas’ thermal performance including comfort conditions. Measurement results were plotted in psychrometric charts together with applicable thermal comfort zones according to the adaptive thermal comfort theory (Szokolay 2007). Furthermore, PMV (Predicted Mean Vote) and PPD (Predicted Percentage of dissatisfied) values were derived (Fanger 1970). The related assumptions for the latter calculations are summarized in Table 2. Figure 1. Exterior view of building K1. Table 2. Overview of assumptions for thermal comfort calculations (for day time from 8:00 to 22:00). clo 1 Fall/Spring Winter Summer met 1.2 air speed [m/s] 0.1 clo 1.5 met 1.2 air speed [m/s] 0.1 clo 0.7 met 1.2 air speed [m/s] 0.1 2.4 Calibration of Simulation Models Figure 2. Exterior view of building K3. Thermal simulation models of the selected buildings were generated using TAS software (EDSL 2010). Using indoor measurements, the simulation models were calibrated (Mahdavi et al. 2007). Since the available local weather files did not contain sufficient information for simulation, they were compared with a synthetic file for the region (Meteotest 2008). Calibration was conducted for intervals with a sufficiently good match between the above mentioned two sources of weather information. Figure 5 shows, as an example, temperature data (both local weather information and synthetic weather file) pertaining to a sample day used for calibration. Figure 3. Exterior view of building K5. 2.5 Parametric Simulation For parametric simulations, a base case scenario (BC) and various combined improved scenarios (S1 to S7) were considered. Improvement scenarios, involved the thermal enhancement of the building envelope elements. Thereby, double-pane window glazing was substituted with low-e triple glazing and the thermal insulation of the roof and external wall were improved. Additional thermal insulation for external walls was limited to internal insulation option. The related condensation risk was further investigated and was not found to be critical. Table 3 shows construction information for existing conditions and the thermal improvement scenarios. Table 4 specifies the scenarios in view of Figure 4. Monitored room examples (fire room, 3rd floor); left (K1), right (K2). 218 the improvement measures considered. Note that all scenarios were run with an overall air change assumption of 0.4 h-1. Table 5 summarizes user profile settings adopted for Kullas under the assumption of full occupancy. 3 RESULTS 3.1 Existing indoor conditions Figure 6 shows the fraction of time when measurements were within thermal comfort zones. Figures 7 to 9 illustrate measured indoor conditions in buildings K1 (partially occupied), K3 (fully occupied) and K4 (unoccupied) for the month of July. Table 6 shows derived mean monthly PMV values in buildings K1, K3 and K4. Temperature [°C] 30 25 20 15 10 Fraction [%] 0 ExT WS ExT MT 10 20 30 40 50 60 70 80 90 100 LR-K4 FR-K3 LR-K3 FR-K1 june Figure 5. Sample day with a good match between air temperature data from the obtained local weather station (WS) and from weather file generated using Meteotest (MS). july Table 3. Overview of construction information. K1, K3, K4, K5 K2 Base Case UWall = 2 W.m-2.K-1 UWindow = 2.9 W.m-2.K-1 URoof = 0.5 W.m-2.K-1 gWindow = 0.75 Improved Scenarios UWall = 0.3 W.m-2.K-1 UWindow = 0.96 W.m-2.K-1 URoof = 0.18 W.m-2.K-1 gWindow = 0.55 URoof = 2 W.m-2.K-1 UWindow = 5.8 W.m-2.K-1 gWindow = 0.85 UWall = 0.3 W.m-2.K-1 UWindow = 0.96 W.m-2.K-1 URoof = 0.18 W.m-2.K-1 gWindow = 0.55 august september Figure 6. Fraction of time when measurements were within comfort zones (LR: living room: FR: fire room). Table 4. Overview of base case and improvement scenarios. Scenarios Windows Walls Humidity Ratio [kg H2O /kg dry air] Improved component Roof BC S1 X S2 X S3 S4 X X S5 X X S6 X S7 X X X X X 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0 Table 5. User profile settings according to ÖNORM 2002 for single family houses. Set point temperature for heating 20°C Air change rate Internal gains (Equipment, Light, Occupants) 0.4 h-1 5 10 15 20 25 Dry Air Temperature [oC] 30 Figure 7. Psychrometric chart, K1-living room, July. 3.75 W.m-2 219 35 3.2 Simulation results Humidity Ratio [kg H2O /kg dry air] 0.000 Figures 10 to 14 show simulated annual heating loads for all buildings and scenarios (BC, S1 to S7). 0.000 0.000 0.000 0.000 180 0.000 160 0.000 140 0.000 0.000 120 0.000 100 0.000 0 5 10 15 20 25 Dry Air Temperature [oC] 30 80 35 60 40 Figure 8. Psychrometric chart, K3-living room, July. 20 0 BC S2 S3 S4 S5 S6 S7 -2 -1 -2 -1 -2 -1 Figure 10. Simulated heating loads of K1 [kWh.m .a ]. 0.000 Humidity Ratio [kg H2O /kg dry air] S1 0.000 0.000 0.000 180 0.000 160 0.000 140 0.000 0.000 120 0.000 100 0.000 80 0.000 0 5 10 15 20 25 o Dry Air Temperature [ C] 30 35 60 40 Figure 9. Psychrometric chart, K4-living room, July. 20 0 BC Table 6. Derived mean monthly PMV values. K1 S1 S2 S3 S4 S5 S6 S7 Figure 11. Simulated heating loads of K2 [kWh.m .a ]. K3 K4 Month LR FR LR FR LR August 0.4 0.7 0.5 0.3 0.2 180 September -1.0 -1.0 -0.4 -1.5 -1.5 160 October -1.3 -1.2 -0.7 -1.7 -1.5 November -1.2 -1.2 -0.1 -1.4 -1.4 December -2.0 -2.0 -0.2 <-2 <-2 January <-2 <-2 -0.1 <-2 <-2 February <-2 <-2 0.2 <-2 <-2 March -2.0 <-2 -0.2 -1.8 <-2 60 April -1.5 -1.4 -0.6 -1.7 -1.7 40 May -0.7 -0.5 0.0 -0.5 -0.8 20 June -0.8 -0.6 -0.4 -0.8 -1.0 0 July - 0.1 0.1 0.0 -0.3 140 120 100 80 BC S1 S2 S3 S4 S5 S6 S7 Figure 12. Simulated heating loads of K3 [kWh.m .a ]. 220 Table 7. Simulated Overheating hours [Kh], buildings K1, K3, and K4. 180 160 June BC 140 LR 120 246 97 144 457 391 165 0 28 34 10 97 135 244 89 131 363 357 131 0 33 31 10 93 132 238 73 S7 -A 267 744 723 263 S7-B 47 133 180 36 BC 142 230 313 96 S7 -A 300 744 720 256 S7-B 74 176 197 38 BC 177 310 365 144 S7 -A 343 744 739 273 S7-B 116 282 272 88 BC 140 175 266 65 S7 -A 269 740 655 239 S7-B 100 203 227 61 BC FR S7 -A 60 S7-B 40 BC LR 20 0 BC S1 S2 S3 S4 S5 S6 K3 S7 Figure 13. Simulated heating loads of K4 [kWh.m-2.a-1]. FR 180 LR 160 140 K4 120 FR 100 Sep. 129 S7-B 80 Aug. 83 S7 -A K1 100 July 80 60 Table 8. Simulated PMV values for scenarios BC, S7-A, and S7-B. 40 June July Aug. Sep. BC 0.6 1.0 1.0 0.4 S7 -A 0.9 1.3 1.2 0.7 S7-B -0.1 0.3 0.3 -0.3 BC 0.6 1.0 1.0 0.3 S7 -A 0.8 1.2 1.2 0.3 S7-B -0.2 0.3 0.2 -0.4 BC 0.5 1.0 1.0 0.3 S7 -A 1.1 1.7 1.6 1 S7-B 0.5 1.0 0.9 0.3 BC 0.7 1.1 1.1 0.4 S7 -A 1.2 1.7 1.6 1.0 S7-B 0.6 1.0 0.9 0.3 BC 0.3 0.8 0.8 -0.1 S7 -A 1.0 1.6 1.5 0.7 S7-B 0.2 0.8 0.7 -0.1 BC 0.1 0.7 0.6 -0.3 S7 -A 0.8 1.4 1.3 0.4 S7-B 0.1 0.6 0.6 -0.2 20 0 BC S1 S2 S3 S4 S5 S6 S7 LR Figure 14. Simulated heating loads of K5 [kWh.m-2.a-1]. K1 FR 3.3 Overheating and Thermal Comfort Table 7 shows calculated cumulated monthly overheating hours in buildings K1, K3, and K4 for base case (BC) scenario and for scenarios S7 (best performing improvement scenario) with the default air change rate of 0.4 h-1 (S7-A) as well as with an increased night ventilation of 1 h-1 (S7-B). Note that overheating denotes here a room temperature exceeding 27 oC. Table 8 includes calculated PMV values (buildings K1, K3, and K4) for BC, S7-A, and S7-B scenarios. LR K3 FR LR K4 FR 221 4 DISCUSSION 6 REFERENCES The measurement results suggest that the problem of overheating does not occur in traditional Kullas. However, the percentage of time in comfort zone is rather low (see Figures 6 to 9). This is mainly due to partially low indoor air temperatures (attributable in part the buildings' very high thermal mass) and the associated high relative humidity values. The derived PMV values also suggest that indoor spaces in Kullas are likely to be perceived as rather cool (see Table 6). With regard to heating energy demand, simulation results suggest that thermally improving windows (S1) or roof area (S3) alone is not particularly effective (see Figures 10 to 14). This is mainly due to the relative small areas involved. However, thermal improvement of the wall insulation does significantly reduced heating demand (S2). Combining wall insulation with other measures brings about only a marginal additional improvement of the energy performance (Figures 10 to 14). Results pertaining to scenario S7 suggests that the combination of all improvement options could reduce heating energy demand of the existing Kullas about 75%. Thermal improvement reduces evidently the heating energy demand. However, calculated overheating hours (as well as PMV values) for scenario S7-A (see Tables 7 and 8) might suggest that better thermal insulation has a negative effect on overheating tendency. Yet these results must be interpreted carefully. Increase in overheating hours (and PMV values) in the S7-A scenario simulations can be attributed to the "heat retaining" effect of the better insulated building envelope. If thermal insulation is coupled with proper ventilations rates (see scenario S7-B), not only the heating energy demand is reduced, but also the overheating hours. The Simulation results thus suggest that thermal improvement measures could not only reduce heating energy demand, but also overheating tendency, if they are accompanied by an appropriate natural ventilation regime. Drançolli F. 2001. Kulla shqiptare. Kosovo. EDSL 2010. Tas Version 9.1 Environmental Design Solustions Limited. http://edsl.net. Fanger PO. 1970. Thermal Comfort. Danish Technical Press. Denmark. IHMK 2009. Instituti i Hidro Meterologjisë së Kosovës. Kosovo. Meteotest 2008. Meteonorm version 6.0. www.meteonorm.ch. Mahdavi A., Orehounig K., Mikats N. Lambeva L., and ElHabashi A. 2007. Analyzing Traditional Buildings via Empirically Calibrated Building Performance Models. Proceeding of IBPSA 2007: 71-78. Bejing, China. ÖNORM 2002. ÖNORM B8110-5. Wärmeschutz im Hochbau - Niedrig- und Niedrigstenergie-Gebäude - Teil 5: Anforderungen und Nachweisverfahren. Österreich. Rassam S. 2000. Kulla: A Traditional Albanian house type in Kosovo. http://archnet.org. Rexhepi A., and Mahdavi A. 2010. Empirical and Computational Study of the Thermal Performance of a Traditional Housing type in Kosovo. Proceeding of BAUSIM 2010: 86-91. Vienna, Austria. Szokolay S. 2004. Introduction to Architectural Science: 1622. Great Britain. Elsevier. 5 ACKNOWLEDGMENTS The Ph.D. study of Albana Deralla was supported in part by the Ministry of Education, Science and Technology of Republic of Kosovo. 222 Thermal performance of a test cell in a hot and humid climate: the impact of thermal insulation P. H. Tan, U. Pont, V. Müller & A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: This paper deals with the impact of thermal insulation on indoor thermal conditions in a hotand-humid climate (Malaysia). To explore this question empirically, two small masonry test cells were constructed, only one of which was thermally insulated. Both test cells were not occupied during the 10 day long test phase and only naturally ventilated. During daytime these openings remained closed, while they were opened in the nighttime for passive cooling. The test cells were equipped with data loggers for temperature and relative humidity. Parameters of the local outside climate were measured via a roof-mounted weather station. According to the experimental results, the insulated cell displayed a more stable temperature regime and somewhat preferable thermal comfort circumstances. The disadvantage of night-time radiative cooling effect reduction was compensated by lower day-time temperatures due to insulation-based reduction of conductive heat transfer. 1 INTRODUCTION Keeping buildings and their rooms thermally comfortable is one of the key tasks of architectural design. In most regions the traditional vernacular building design has adapted to the climate. Instances of such adaptation have been discussed for example in Mahdavi 1996 (for the arid climate) and in Tan 2012 (for the Malaysian climate). In the latter, the traditional Malay house is described as a wooden structure of low thermal mass that during daytime has all openings closed and/or shaded. Modern Malaysian buildings tend to be constructed with bricks. Thus, a key difference to the traditional architecture is the high building mass. In this contribution the effect of thermal insulation on such massive structures is discussed for the climatic context of Malaysia. Malaysia features a hotand-humid (tropical) climate. Figure 1 shows the typical minimum, average and maximum temperatures as well as the relative humidity on a monthly basis. Two small test cells, built of massive masonry were constructed (in Malaysia) and monitored to examine the effect of thermal insulation and passive cooling. Three main research questions were addressed: i) Is indoor thermal comfort affected in the thermally insulated cell due to the reduction of night-time radiant cooling? ii) How do the two cells compare regarding thermal comfort during a 24 hour cycle? iii) How do the humidity levels behave in both test cells? Figure 1. Minimum, average, and maximum temperatures and relative Humidity on monthly basis for Kuala Lumpur, Malaysia (CLIMA 2013) 2 BACKGROUND A study similar in approach to the present one – however for a very different (hot and arid) climate – was conducted in Israel involving test cells (Givoni 2009). This study implied that passive cooling methods such as night ventilation work well in regions with a diurnal temperature swing of more than 8 Kelvin. Kubota et al. (2009) studied the effect of nighttime ventilation on the indoor climate in residential buildings in Malaysia. They compared the behavior of two wooden built light-weight row houses (one with night ventilation, the other one without any ventilation at all). Night ventilation was shown to bring about lower daytime temperatures (between 223 0.5 and 2 K in comparison to the non-ventilated building). However, in this study the buildings featured the possibility of cross ventilation and they were not insulated at all. Various other studies (e.g., Al-Hamoud 2003) highlight the importance of thermal insulation on buildings in hot climates, as well as the necessary thoughtful planning of insulation of building elements. . Figure 2. Plan and elevations of the test cells. 3 APPROACH The two free standing test cells were constructed (in 2010) with an internal floor area of 3x3 m and 3.6 m room height (Figure 2). The walls were constructed out of 110 mm bricks and mortar. One test cell (referred to as A) had no additional insulation on the exterior walls. The other test cell (B) was treated 50 mm of EPS insulation with an external layer of thin brick cladding (Figure 3). This system is called EIFS (Exterior Insulation Finishing System). Both test cells were insulated with 50 mm EPS on the top of the attic floor to reduce the heat flow between attic and the room below. Figure 2 shows a plan and two elevations of the basic setup of both test cells, while Figure 3 illustrates the EPS/EIFS wrapping of test cell B. After finishing the primary construction of the test cells (i.e., before the application of EPS/EIFS to B), the temperature in both test cells was monitored over multiple days to ensure that their thermal behavior was congruent (see Figure 4). Subsequently, the thermal behavior of the insulated and non-insulated cells were extensively monitored for 10 days in December 2010. Both test cells were equipped with a set of data loggers for measurement of indoor temperature and relative humidity. The six sensor positions in each cell included: (i) above ceiling, (ii) 3.3 m above ground (centered), (iii) 1.5 m above ground (centered), (iv) 0.3 above ground (centered), (v) on the north wall and (vi) on the west wall (Figure 5). Additionally, a weather station was mounted next to the test cells to monitor outdoor ambient temperature, outdoor relative humidity, solar radiation and wind speed and direction. During day time (9am – 7pm) the entrance doors of the cells were kept closed, while during night (7pm – 9am), the doors were opened to facilitate night-time ventilation. The windows stayed closed throughout the experiment period. Note that the actual rate of air change in the two cells could not be measured. However, given identical location, construction, and orientation of the two cells – as well as synchronized timing of the opening and closing of the cell doors – a similar rate of air change can be considered as likely. Figure 3. EPS/EIFS Application on Test Cell no2. Figure 4. Correlation between measured temperatures in the two cells prior to treatment of cell B with insulation. Figure 5. Sensor positions within each cell (left), roof-mounted weather station (right). 4 RESULTS AND DISCUSSION 4.1 Temperature and relative humidity To illustrate the indoor temperature and relative humidity trends, a typical day within the observation period is selected (16th of December 2010). For this day, Figure 6 depicts the course of temperatures and 224 solar radiation, Table 1 includes the minimum and maximum outdoor and indoor temperatures, while Table 2 shows the values of temperature and relative humidity at specific times. Obviously, the temperature amplitude in the non-insulated cell A is significantly larger (over 10 K, as compared to 3 K in cell B). Thermal insulation does appear to weaken the night-time radiative cooling of the cell, but the day time temperature peak is significantly dampened, given cell B a more comfortable overall thermal regime: Indoor temperature in cell B remains considerably longer below an assumed upper comfort temperature level of of 27 °C. This assertion is further supported by considering the indoor temperature distribution for the entire observations period (14th – 22nd December 2010). The corresponding results are shown in Figure 7, which depicts indoor temperature distribution in the two test cells. Figure 8 shows the same data in terms of a cumulative distribution graph. Figure 6. Temperatures indoor and outdoor and solar radiation on 16th of December 2010. Table 1. Maximum and minimum indoor and outdoor temperatures for 16th of December 2010. Maximum Minimum Difference [°C] [°C] ("amplitude") [K] Outdoor 33 23 10 Test cell A 31 28 Test cell B 24 25 7 3 Table 2. Temperature and relative humidity values at specific times on 16th of December 2010. Time 00:00 04:00 08:00 12:00 16:00 20:00 Figure 7. Distribution of indoor temperature measurements over the whole observation period. θe [°C] 23.7 22.9 23.5 31.5 26.8 23.3 RHe [%] 93 94 92 65 77 91 θi,A [°C] 26.0 25.4 25.4 27.2 31.0 27.0 82 84 86 83 70 75 26.0 24.6 24.2 26.5 27.6 26.4 84 84 84 85 80 80 RHi,A [%] θi,B [°C] RHi,B [%] Figure 8. Cumulative distribution of indoor temperature measurements over the whole observation period. Figure 9 shows the relative humidity levels during the aforementioned reference day. The apparent divergence between the relative humidity in the two cells can be explained in part by the difference in the air temperature, as evidenced by the comparison of absolute humidity levels over a period of three days (see Figure 10). Figures 9 shows the indoor temperature distribution in the test cells for. Figure 9. Indoor and outdoor relative humidity levels for 16th of December 2010 (included are also solar radiation data). 225 5 CONCLUSION Two test cells (insulated and non-insulated) in a hotand-humid climate (Malaysia) were constructed and monitored with regard to indoor environment. The insulated cell displayed a more stable temperature regime. The disadvantage of night-time radiative cooling effect reduction was compensated by lower day-time temperatures due to insulation-based reduction of conductive heat transfer. Generally speaking, the insulated test cell B provided somewhat preferable thermal comfort circumstances: Its temperature remained below 27 °C for nearly 80% of the monitoring time (65% for test cell A). As such, the nighttime temperatures in the insulated cell could have been lower, if higher air change rate could have been realized via cross ventilation (not possible in the present set up). Future works could further explore the potential effect of cross ventilation toward increased nighttime cooling. Moreover, construction alternatives with lower or higher thermal mass and more or less thermal insulation should be considered. Meanwhile, the monitoring results are been deployed to calibrate thermal simulation models of the cells (see Appendix for a brief description). Parametric simulations based on such models can be applied to virtually realize and evaluate variations of cells' thermal mass, insulation thickness, and ventilation rates toward thermal performance optimization. Figure 10. Absolute humidity levels outdoor and in the test cells and solar radiation for the run of three days during the monitoring period. 4.2 PMV and PPD To obtain a deeper sense of thermal comfort conditions in the test cells, measured temperature and relative humidity values were used to calculate PMV (Predicted Mean Vote). Thereby, the assumed values for metabolic rate, clothing, and air speed was 1.2 met, 0.63 clo, and 0.1 m.s-1 respectively. Mean radiant temperature values were calculated using an existing computational application (Academy 2006). The results (PMV distribution in both cells) are shown in Figure 11. It could be argued that, in the light of adaptive thermal comfort theory findings, PMV would not be an ideal indicator of thermal comfort for the present case (naturally ventilated cells in a hot-humid context). Nonetheless, the results do provide an impression of the indoor climate differences in the two cells and appear to confirm the conclusion made earlier based on indoor temperature alone. 6 REFERENCES Academy 2006. Academy of HVAC Engineering Ltd. 2006. Al-Homoud M. 2003. The Effectiveness of Thermal Insulation in Different Types of Buildings in Hot Climates. Journal of Thermal Envelope and Building Science 2004; 27; 235, DOI: 10.1177/1097196304038368 CLIMA 2013. http://www.malaysia.climatemps.com/#table (last visited March 2013). EDSL 2011. Software A-Tas, www.edsl.com (last visit March 2013) Givoni B. 2009. Indoor temperature reduction by passive cooling systems. Solar Energy, Elsevier, doi:10.1016/ j.solener.2009.10.003 Kubota T., Hooi Chyee, D.T., Supian, A. 2009 The effects of night ventilation technique on indoor thermal environment for residential buildings in hot-humid climate of Malaysia. Energy and Buildings 41, 2009, Elsevier, pp829–839. Mahdavi A. 1996.. A Human Ecological View of "Traditional" Architecture, Human Ecology Review (HER), 3 (1996), 1; P. 108 - 114. METEOTEST 2009. Software Meteonorm, version 6; www.http://meteonorm.com/de/ (last visited March 2013). Tan P.H. 2012.. Thermal Performance of a test cell in a hothumid climate: the impact of thermal insulation. MasterThesis, Vienna University of Technology. Figure 11. PMV in Test Cell A and B (distribution over the observation period). 226 APPENDIX Dynamic thermal simulation The simulation application TAS (EDSL 2011) was used to model the test cells in terms of two thermal zones (attic space and main space). To represent the climatic boundary condition, weather data from the aforementioned weather station on site was used, together with additional generated via METEONORM application (METEOTEST 2009). Thermal properties of the materials were obtained from corresponding specifications of the manufacturers. Input information regarding ventilation schedule was set according to the actual ventilation regime. However, the ventilation rates had to be estimated. Figure A-3. Correlation between measured and simulated temperatures for cell A. Comparison with measurements Figures A-1 and A-2 display measured and simulated indoor temperatures in cells A and B respectively. Associated regression analyses are shown in Figures A-3 and A-4. In case of cell A, simulation results are rather close to the measured temperatures. The simulation does overestimate the indoor temperature by about 0 to 0.5 K (up to 1.5 K on few occasions during daytime). A lesser degree of agreement between measurements and simulations applies to cell B. Temperature overestimations by the simulation of about 0.5 - 2 K during daytime can be observed. The night-time temperatures are only slightly overestimated (0 - 0.5 K). Figure A-4. Correlation between measured and simulated temperatures for test cell B. Further simulation runs are necessary to better calibrate the numeric thermal simulation model. Subsequently, parametric simulations are to be conducted in order to virtually realize and evaluate variations of cells' thermal mass, insulation thickness, and ventilation rates toward thermal performance optimization. Figure A-1. Measured and simulated indoor temperatures in cell A (also shown is the outdoor temperature). Figure A-2. Measured and simulated indoor temperatures in cell B (also shown is the outdoor temperature). 227 228 A comparison of projected and actual energy performance of buildings after thermal retrofit measures P. P. Housez, U. Pont & A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: This paper addresses the discrepancies between projected and actual energy performance of thermally retrofitted buildings. It is based on the detailed observation of seven recently thermally retrofitted residential buildings in Austria. These include five multifamily residences in Vorarlberg, the upper level of a duplex in Lower-Austria, and a residential complex for the elderly in Styria. During the heating season 2009/2010 (1st October – 30th April), indoor temperature and relative humidity were measured and logged in these buildings. For each building, the actual energy use during this period was derived based on bills for gas and electrical power. Additionally, existing energy calculations (energy certificates) were examined. In six of seven cases, a large discrepancy between projected and actual space heating demand was noted. To explore the reason for this discrepancy, for each of the buildings an energy certificate was generated and a detailed thermal simulation was conducted. Thereby, the main input parameters for energy calculations (air change rate, indoor air temperature, outside air temperature, internal gains) were taken into consideration. If all of these parameters are set to the standard (default) values as suggested by Austrian standards, the above mentioned discrepancy cannot be explained. Measurements in the buildings, as well as interviews with the building’s inhabitants implied that standard-based input data assumptions were not reliable. A subsequent multifactor study suggested that specifically the assumptions regarding air change rates might be responsible for the large deviations of the calculated values from the actual heating demand. For instance, Guerra Santin et al. (2009) state that user behavior can influence the energy demand of buildings as much as the thermal properties of the building envelope. Castagna et al. (2008) present a case study of an Italian low energy building (CasaClima Gold) with a calculated heating demand of 9 kWh.m-2.a-1. Due to user-driven ventilation, climatic conditions, and technical issues, the actual heating demand was found to be 42 kWh.m-2.a-1. Guerra Santin et al. (2009), Iwashita and Akasaka (1997), and Richter et al. (2003) all identify users' ventilation behavior as a crucial influencing factor on heating demand. This is especially true for lowenergy buildings. Higher indoor temperatures are mentioned as a far smaller influencing factor. In this context, the so-called Rebound-Effect has been mentioned. It involves the reduction of the effect of energy efficiency measures due to postretrofit changes in user behavior. Van Raaj and Verhallen (1985) describe five different groups of behavioral patterns of building inhabitants ("Conservers", "Average", "Spenders", "Cool" and "Warm"). Some of these groups tend to raise the room temperature and/or increase the ventilation rate, resulting in the reduction of the projected savings (see also Hens et al. 2010 and Hoes et al. 2008). In this contribution, seven recently thermally retrofitted residential buildings are examined in view of discrepancies between their projected and actual space heating demand. The former is documented via existing energy certificates and energy calculations done by the authors, while the latter is derived from buildings' energy bills. 1 INTRODUCTION One of the main goals of thermal retrofit of existing buildings is to lower their heating demand. This can not only reduce buildings' operational cost, but also their environmental impact (e.g., greenhouse gas emissions). Most refurbishment cases involve one or more of the following steps to reduce the building’s energy demand are applied: i) Application of thermal insulation to the building envelope to reduce transmission losses, ii) Renewal of the building’s HVACsystems toward higher energy efficiency, iii) Implementation of technologies such as controlled mechanical ventilation systems for reduction of ventilation losses. Refurbishment planning therefore needs an in-depth knowledge of building construction technology, building physics, and advanced HVACsystems. Usually, in the course of thermal retrofit projects, certain energy performance indicators (e.g., heating demand) are calculated for before and after conditions. To facilitate benchmarking, calculation methods are standardized in view of underlying calculation methods and input parameters pertaining, for example, to occupancy, ventilation, and indoor temperature. Real user-related data can be very different from assumed values, contributing thus to the deviations between reality and calculated indicators. Consideration of user behavior in building's performance calculation is not a trivial matter (Zimmermann 2007). This represents a problem, as several studies point to user behavior as a strong influencing factor on energy demand of buildings. 229 2 METHODOLOGY 2.1 Selected buildings We selected seven buildings (A to G), which were recently retrofitted to (nearly) meet "Passivhaus" standard. Furthermore, Objects A to E as well as G were equipped with solar-thermal collectors. Table 1 provides an overview of the key data of these objects. Figures 1 – 5 show some of the objects before, during, and/or after refurbishment. Figure 3. Object E after refurbishment. Figure 4. Object F before (l.), during (m.), and after (r.) refurbishment. Figure 1. Object A before (l.) and after refurbishment (r.). Figure 5. Object F during (l.) and after (r.) refurbishment. 2.2 Inhabitants The inhabitants of the objects A-E include singles, couples, and families. All inhabitants lived in their flats already before the thermal retrofit. In case of building F, the refurbishment was planned by a technically knowledgeable senior member of the family owning the building. Object G is a home for elderly inhabited by 115 retirees. According to the building's administrative staff, high thermal comfort for the inhabitants of this building has priority over energy saving. Furthermore, most of the inhabitants were not aware of the fact that their home was refurbished to low-energy standards. Figure 2. Object D after refurbishment. Table 1. Key data of the selected buildings. Object Function Construction year Refurbishment year Net Volume [m³] Gross Area [m²] Um Opaque [W.m-2.K-1] Um Transparent [W.m-2.K-1] Window/floor ratio [%] lc [m] LEK [-] A apartment building 1968 2007 1026 475 0.173 0.842 10.0 1.67 19.64 B apartment building 1968 2007 1111 512 0.178 0.835 9.6 1.69 19.6 C apartment building 1978 2007 4005 1813 0.117 0.876 11.7 2.22 14.6 D apartment building 1978 2008 2569 1142 0.158 1.113 8.7 2.17 17.32 E apartment building 1975 2009 2084 928 0.142 0.792 10.8 2.08 15.6 F Extension 1985 2005 437 168 0.110 0.790 10.9 1.43 16.12 G Home for elderly 1975 2005 14171 5490 0.303 1.230 17.8 3.79 24.21 230 able tools (OIB 2003, A-Null 2009) were deployed. The hourly-based thermal simulations were performed using TAS software (EDSL 2010). Necessary settings for both the simple and the complex calculations were derived from the refurbishment planning, inspections, and interviews with the inhabitants. 2.3 Measurement of indoor climate data During the monitoring period (from 1st of October 2009 to 30th April 2010) indoor conditions (temperature, relative humidity) were monitored in selected rooms of each of the objects in 15 minute intervals (four data loggers in objects A-E, three in F, and five in G). 2.8 Overview of the derived indicators To illustrate and structure the results properly, the buildings' heating demand were derived in terms of five indicators. These indicators and their respective underlying derivation methods are as follows: The heating demand based on the existing energy certificate data are denoted as ECex. Our own heating demand calculations based on the energy certificate method are denoted as ECa (annual method) and ECm (monthly method). Actual heating demand (estimated based on energy bills) is denoted as HDact, and simulated heating demand is denoted as HDsim. 2.4 Weather data Data from local weather stations was obtained for the monitoring period, including outdoor air temperature, relative humidity, hours of daily sunshine, solar radiation, wind speed, and wind direction. 2.5 Electrical power and gas consumption For all objects the energy bills were collected. Objects A-E use gas burners for heating and warm water, together with solar collector units. Gains by the solar collector units are not measured. Hence, their contribution was estimated based on solar radiation data and panel sizes. Object F is heated by an oil burner. The monthly oil demand was documented by the inhabitants on the monthly basis. Object G is connected to a local district heating provider. The solar collector units on the roof of G are solely used for warm water. As energy bills only provide a lump sum for both heating and warm water energy demand, the share of warm water was estimated based on OENORM 2010 (35 Wh.m-2.d-1 for A-F and 70 Wh.m-2.d-1 for G). The final estimation of actual space heating demand took the estimated contribution of solar collector units as well as the assumed system efficiencies into account. 2.9 Parametric studies For the base case, for all of the above indicators, internal gains assumptions were based on standard values. Standard data was used also for weather information, with the exception of HDsim, which was based on local weather station data. Starting from the base values, a number of parametric calculations were performed in order to explain the gap between calculated and actual heating demand values: − Weather Data. To explore the effect of actual local weather conditions during the observation period on calculation results the real measured weather data was adapted and used. − Air change rate: Assumed air change rates in the base case calculations use standard values. A variation of this input variable can show the influence of deviating user behavior. − Indoor temperature: To explore the effect of actually maintained indoor air temperatures, monitored data were used instead of fix standard-based data. 2.6 Existing energy certificates The existing energy certificates of the buildings were collected. For objects A, B, D, and G an (older) annual method had been used (consistent with the applicable standard at the compilation time). The energy certificates for the other objects were calculated with a (newer) monthly method. The Heating Demand stated in these certificates was used as a reference value. 3 RESULTS & DISCUSSION 2.7 Calculations and simulation models 3.1 Estimation of the actual heating demand Given the mentioned differences in calculation methods and abundance of errors in preparation of energy certificates (Kaiser 2009), we conducted our own calculations, using both normative calculation procedures and detailed numeric simulation. The normative procedure included both the old (annual), and the current (monthly) calculation method for Austrian Energy certificates. Toward this end, avail- Table 2 summarizes the steps toward the derivation of the estimated values of the buildings' actual heating demand (see section 2.5). 231 3.2 Results comparison for base case Figure 6 entails a comparison of calculated (ECex, ECa, ECm, and HDsim) and actual heating demand (HDact) for all buildings. As compared to energy certificate results, the detailed thermal simulations provide higher estimates of heating demand. Nonetheless, they are still significantly lower than the actual demand data, except for building F. Figure 7. Comparison of ECex, HDAct and weather-adapted calculations and simulation (ECa, ECm, HDsim). Figure 6. Base case Heating demand results (ECex, ECm, EXa, HDsim, HDAct). 3.3 Adaptation of weather data For this variation (see Figure 7) the measured weather data for the heating period 2009/2010 was used in the energy certification calculations and the thermal simulation instead of standard weather data. These results suggest that – despite some slight improvement – the weather-related model input assumptions cannot explain the differences between calculated and estimated heating demand. Figure 8. Monitored monthly mean indoor temperatures in representative rooms in the selected buildings. 3.4 Adaptation of indoor temperature As mentioned before, the actual indoor air temperature was monitored in selected rooms of the buildings. Figure 8 shows monitored monthly mean indoor temperatures in representative rooms in the selected buildings. Figure 9 illustrates the cumulative frequencies of indoor temperatures in buildings A-F. Figure 9. Cumulative frequencies of the indoor temperature in the objects A-F. Table 2. Estimation of heating demand (derived from bills). Object A B C D E F G Total heating energy use [kWh] Energy use according to gas/oil bills [kWh] Solar gains via collectors [kWh] Estimated energy use for warm water [kWh] Estimated heating energy use [kWh] Assumed heating system Efficiency [-] Estimated actual total heating demand [kWh] Estimated actual area-related heating demand [kWh.m-2] 34682 36195 26072 27585 8610 8610 6.064 6543 28618 29652 0.925 0.925 26471 27428 55.76 53.55 121196 64051 53528 5688 430500 93767 51751 40528 5688 390500 27429 12300 13000 40000 23161 14586 11852 2146 140270 98035 49465 41676 3542 290231 0.925 0.925 0.925 0.770 0.880 90682 45755 38550 2727 255403 50.02 40.07 41.55 16.23 46.52 232 Figure 10 compares the modified ECm calculations (with monitored indoor air temperatures) with the ECex and HDact. The results show that the use of the actually measured indoor temperatures instead of the standard value of 20°C (24°C in the home for the elderly) does change the calculated heating demand, however only slightly. Table 3. HDact and ACH assumptions in ECex together with ACH assumptions for annual (ECa) and monthly (ECm) calculation methods that would yield close estimation of actual heating demand. Build- HDact ACH asModified Modified ACH ing [kWh. m- sumption in ACH asassumption in 2 -1 ECex [h-1] sumption ECm [h-1] .a ] in ECa [h 1 ] A 55.76 0.14 0.92 0.91 B 53.55 0.14 0.87 0.87 C 50.02 0.10 0.66 0.64 D 40.07 0.25 0.63 0.65 E 41.55 0.14 0.63 0.64 F 16.23 0.10 0.07 0.07 G 46.52 0.13 0.78 1.95 Figure 10. Comparison of ECex, HDAct and monthly method (ECm) with adapted indoor temperatures. 3.5 Adaption of air change rate In energy certificate calculations according to OENORM 2010 a default air change rate of 0.4 h-1 is used. For calculations pertaining to buildings with highly insulated envelope and controlled ventilation systems (with heat recovery) lower ventilation rate assumptions are suggested. This circumstance applies to the buildings selected for the present study. However, the real ventilation rate might be different due to the users' window operation behavior. To address this issue, we explored the following question. What air change rates could have resulted in the observed heating demands? Table 3 illustrates the results of this exploration. Figure 11 compares modified ECa und ECm calculations (with adjusted air change rates) with HDact. These results clearly demonstrate the major impact of air change rate assumptions on the magnitude of the calculated heating demand. In order to reproduce actual heating demand values, model assumptions for air change rate need to be higher than the energy certificate assumptions by a factor of four to six (with the exception of building F). Note that – for buildings, which facilitate manual window operation – the modified air change rate assumptions are not unrealistic. Figure 11. Figure 11. Comparison of ECex, HDAct and monthly method results with adapted ACH (ECm, adapted). 3.6 Overview and multi-variable analysis Table 4 facilitates the comparison of actual heating demand with multiple calculation results involving different methods (ECa, ECm, HDsim) and multiple input assumptions (weather data, indoor temperature, air change rates). This data confirms the aforementioned conjecture (see section 3.5): Air change rate assumptions appear to be – in the case of present 233 buildings – the key factor for the explanation of the divergence between projected and actual heating demand values. Note that building F results deviate from the other buildings. In this case, the original calculations with the standard assumption appear to closely match the actual heating demand. The reason for this may lie in this building's specific ownership and occupancy circumstances. This building is in complete ownership of one family. The thermal retrofit was planned by a senior member of this family (an engineer by profession). It can be thus speculated that, in this case, occupants were more cognizant of the impact of their control-oriented actions (e.g., window operation) on the building's energy performance. Table 4. Overview of the variance in heating demand estimation results due to differences in deployed methods as well as assumptions pertaining to boundary conditions (weather), indoor temperatures, and ventilation rates. Heating demand [kWh.m-2.a-1] (associated air change rate written in italics in h-1) Method Weather Indoor temperature HDact ECex ECa Standard Monitored A B C D E F G 55.76 53.55 50.02 40.07 41.55 16.23 46.52 (0.14) (0.14) (0.10) (0.25) (0.14) (0.10) (0.13) 8.80 9.22 11.10 15.39 11.42 16.83 11.62 (0.14) (0.14) (0.10) (0.25) (0.14) (0.10) (0.13) 8.65 9.10 7.90 15.54 6.64 14.72 11.49 (0.14) (0.14) (0.10) (0.25) (0.14) (0.10) (0.13) 10.93 11.20 6.59 15.21 8.16 13.73 17.17 (0.14) (0.14) (0.10) (0.25) (0.14) (0.10) (0.13) 11.56 12.08 5.18 11.82 11.78 6.24 11.46 (0.14) (0.14) (0.10) (0.25) (0.14) (0.10) (0.13) Standard Standard ECm Monitored Standard HDsim Monitored Standard ECm Standard Monitored ECa Standard Standard ECm Standard Standard ECa Monitored Monitored ECm Monitored Monitored HDsim Monitored Monitored 13.68 22.96 12.48 21.37 11.76 17.49 13.09 (0.14) (0.14) (0.10) (0.25) (0.14) (0.10) (0.13) 55.74 53.33 50.50 40.23 41.54 16.29 46.79 (0.92) (0.87) (0.66) (0.63) (0.63) (0.07) (0.78) 55.54 53.7 50.35 39.98 41.34 16.05 46.59 (0.91) (0.87) (1.95) (1.95) (1.95) (1.95) (1.95) 55.82 53.83 50.02 40.10 41.49 16.03 46.42 (0.76) (0.51) (0.62) (0.53) (0.61) (0.09) (0.33) 55.74 53.91 50.18 40.41 41.74 16.26 46.24 (0.97) (0.59) (0.80) (0.58) (0.74) (0.16) (1.63) 56.23 53.71 49.82 40.24 41.58 15.98 47.01 (0.67) (0.49) (0.67) (0.54) (0.57) (0.26) (0.62) 4 CONCLUSION calculation method assumptions. But we cannot necessarily speak of an instance of the rebound effect, as indoor thermal conditions prior to the retrofit measures are not known. Air change rate assumptions, however, could provide an explanation for the gap. Manual window operation is possible in all buildings studied. Hence, occupancy-driven window operation might have resulted in air change rates higher than those suggested in energy certificate calculations. Another piece of information, which gives credence to this speculation pertains to the energy accounting practice (distribution of energy costs amongst tenants) in these buildings. As buildings do not possess heat flow me- This paper compared real and projected heating demand for seven buildings, which were recently retrofitted to meet the Passivhaus standard. Results showed a large difference between projected and estimated heating demand. The estimated actual heating demand for six of the seven buildings were found to be significantly larger (by a factor of 4 to 6) than the calculated values. Input assumptions regarding boundary conditions (weather data) and indoor temperatures could not sufficiently explain the gap between calculations and observations. The monitored indoor temperatures were generally higher than 234 ters for individual units in buildings A to E, the distribution of thermal energy costs is based on the unit size. Hence, the accounting method does not necessarily provide an incentive for the individual occupants toward energy-efficient behavior (for example with regard to indoor air temperature settings or the frequency of natural ventilation via manual window operation). 5 REFERENCES A-Null 2009. Archiphysik - Software zur Erstellung von bauphysikalischen Berechnungen; A-Null Bauphysik GmbH, www.archiphysik.com Castagna M., Schmitt Y., and Troi A. 2008. Energy performance of buildings in the autonomous province of south Tyrol, Italy World sustainable Building Conference, Melbourne, Australia 21st - 25th September 2008 – 7 p. Guerra Santin O., Itard L., and Visscher H. 2009. The effect of occupancy and building characteristics on energy use for space and water heating in a Dutch residential stock, Buildings & Energy 41 (2009) pp. 1223-1232 EDSL 2010. A-TAS. www.edsl.net Hens H., Parijs W., and Deurink M. 2010. Energy consumption for heating and the rebound effects Energy and Buildings, Vol. 42, Issue 1, January 2010, pp 105-110 Hoes P., Hansen J.L.M., and Loomans M.G.L.C. 2008. User behavior in whole building simulation Energy and Buildings 41; 2008 – p 2 Iwashita G., and Akasaka H. 1997. The effects of human behavior on natural ventilation rate and indoor air environment in summer — a field study in southern Japan Energy & Buildings 25 (1997) pp 195-205 Kaiser J. 2009. Tickende Zeitbomben; http://www.report.at/bau-immo/wirtschaftpolitik/item/33151-tickende-zeitbomben (last visit March 2013) OENORM 2010. OENORM B 8110-5. Wärmeschutz im Hochbau - Niedrig- und Niedrigstenergie-Gebäude - Teil 5: Anforderungen und Nachweisverfahren; Österreichisches Institut für Normung 2010 OIB 2003. Excel-Tool zur Erstellung von Energieausweisen, Oesterreichisches Institut für Bautechnik, 2003 Richter W. Ender T., and Hartmann T. 2003. Einfluss des Nutzerverhaltens auf den Energieverbrauch in Niedrigenergie- und Passivhäusern– in Bauforschung für die Praxis, Stuttgart :Fraunhofer IRB Verlag; 2003 Band 63 – 127 p. Van Raaij F., and Verhallen T. 1983. Patterns of residential energy behavior, Journal of economic Psychology 4 (1983) pp 85-106 Zimmermann G. 2007. Modeling and simulation of individual user behavior for building performance predicitons. in Proceedings of the Summer Computer Simulation Conference, San Diego 235 236 Design and construction of a plus energy house T. Schoch Xella Technologie- und Forschungsgesellschaft mbH, Kloster Lehnin The latest version of EPBD (European Energy Performance of Buildings Directive) calls for all new buildings to be nearly-zero energy buildings by the end of 2020. The term “zero” has yet to be clearly defined, but it is expected that “zero” will focus on the primary energy demand of the building. The construction industry faces a huge amount of changes in terms of future construction and several questions have arisen regarding the continuity of using well known proven design and constructions. With regards to this challenges the ministry of construction in Germany set up a funding program for design and construction of so called efficiency-houseplus, which is defined as a house with lower energy consumption than it can deliver for the energy grid. A team of designer, energy system planners and contractors have pick up this challenge and took part on this funding program recently. The house was built close to Berlin and has been handed over to the user in December 2012. This is the first massive plus one family energy plus house within the funding program. The paper reports about all difficulties in the design and construction phase and how they could be overcome. The paper shows also the selected monitoring system and the expected results. 1 INTRODUCTION following sections provide an overview of our findings so far and possible future implications. The residential sector is the largest consumer of energy in the German economy, accounting for just over a third of energy consumed, which makes it one of the sectors responsible for the highest CO2 emissions. Heating and hot water account for around 46% of CO2 emissions generated by private households. 2 TRANSFERRING THE “ENERGY-PLUS HOUSE” STANDARD TO SOLID MASONRY CONSTRUCTIONS The challenge in constructing energy-efficient residential and non-residential buildings is not dependent on the type of construction. Masonry construction, which is the dominant construction style for new buildings in Germany, will have an important role to play in that it combines the benefits of this construction method (which relate not merely to energy efficiency) with this very characteristic without forfeiting substantial advantages such as loadbearing capacity, fire protection and sound insulation. Advocates of masonry construction have always played a part in the development of new concepts, such as the passive house, the low-energy house or even the ultra-low-energy house; not just superficially, but fundamentally by bringing new observations and accents to the discussion and – more importantly – ensuring their practical implementation. These initial considerations were reason enough to become involved in the BMVBS programme to develop an energy-plus house and to build the first single-family masonry house as part of this programme. But how do you go about satisfying such requirements? Ideas, a plot of land and a group of hard-working designers full of initiative and enthusi- It is therefore no surprise that European Union policy is particularly concerned with responding to climate change and making corresponding improvements in energy efficiency. The targets set by the European Council are ambitious: A 20% reduction in CO2 emissions from 1990 levels by 2020 and a 20% increase in the share of renewables for overall primary energy consumption by 2020. Projected energy consumption for 2020 is also to be cut by 20%. The German Federal Government's “energy concept” launched on 28 September 2010 provides guidelines for an environmentally friendly, reliable and affordable energy supply, creating the first roadmap for a renewable age. In June 2011 the energy concept was specified in greater detail. By 2020 primary energy consumption is to be 20% lower than in 2008, and 50% lower by 2050. The share of renewable energies is to increase to 60% by 2050.[1] Xella and the housebuilder Elbehaus have joined forces to work on a pilot project for developing energy efficient houses funded by the Federal Ministry of Transport, Building and Urban Development. The 237 asm who are willing to undertake the preliminary work. A team of designers from Xella Technologieund Forschungsgesellschaft (TF) and the housebuilding company Elbehaus Schwerin took up this challenge and developed the “M1” energy-plus house and then built it on a plot in Brieselang near Berlin. ceilings of the house are of solid masonry construction. The house is equipped with the necessary plant and equipment to act as a “mini power station” – see the energy balance data below. The car and carport are a component of the project which extends its energy efficiency to include mobility. The same applies to the household appliances and the heating and hot water systems. With a floor space of around 130 m², the building has been designed for a family of four. The floor plans are shown in Figure 3. Figure 1. The M1 energy-plus house concept Admittedly, this is a noble aspiration and it's not the first time that it has been expressed. But this aspiration presupposes that the concept and its outcome are transparent at all times, not just in those areas which find widespread favour – and here we are referring to the anticipated energy consumption – but also in terms of costs. The project team went through different stages of discussion until ultimately the house was designed on paper in its present form and subsequently constructed on the plot. Not everything in your imagination can always be implemented in its purest form, and this was no less true here. Apart from our own ideas, the ideas expressed in the funding guidelines and by the local planning authorities played a not insignificant role. The result is a house which, externally at least, resembles many other contemporary houses. Figure 2 illustrates what the M1 looks like: Figure 3. Floor plans of the M1 energy-plus house According to the BMVBS funding guidelines, the house must achieve a value below zero for both the final and primary energy demand. This challenging requirement inevitably led us to conclude that minimising the building's heat demand would fundamentally affect the choice of possible building materials and construction methods. Every kilowatt hour of surplus energy for heating the building or providing hot water has a direct impact on the choice of technical building equipment. Mindful of the saying that Figure 2. M1 energy-plus house The M1 has been designed and constructed as a single-story family home with a developed roof space. The internal and external walls, floor slab and 238 or vapour barriers are required neither to maintain the integrity of the structure nor to make the building airtight. Figure 4 shows the structure of the YtongEnergy-Plus wall system. the first decision sets us free whilst the second one enslaves us, we compared different types of insulation in terms of both performance and cost. In the end we realised that the heat transfer coefficients of individual components of the opaque building envelope should be comparable with the passive house standard (approx. 0.15 W/(m²K)). The window requirements are also similar – a heat transfer coefficient of 0.80 W/(m²K) for energy transmittance through the glazing of around 0.55 W/(m²K). The individual assemblies of the opaque building envelope and their corresponding U-values are shown in Table 1. Table 1. Structure of the opaque building envelope Structural Structure element (starting from the inside) Floor slab 60 mm floor screed 100 mm expanded polystyrene 200 reinforced concrete 120 mm extruded polystyrene External wall 10 mm gypsum plaster (Ytong 400 mm Ytong Energy Plus Energy Plus) 15 mm exterior plaster Roof 12.5 mm gypsum plasterboard 260 mm mineral fibre/timber roof covering U value in W/(m²K) 0.16 Figure 4. Ytong Energy Plus – the monolithic wall system 1 = load-bearing AAC, raw density: 400 kg/m³ 2 = non-load bearing AAC, raw density: 100 kg/m³ 3 = non-load bearing AAC, raw density: 400 kg/m³ 0.15 The requirement to construct a masonry building using approved standards also prompted the need for straightforward design detailing for craftsmen to work with. As an essential criterion, the building design demonstrates how simple detailing, such as can be achieved with a masonry house, makes a substantial contribution to the quality of an energy-efficient house. The following images of the construction demonstrate the straightforward detailing, although it has to be said that in this project too, the craftsmen were often called upon to make adjustments. 0.16 All internal walls in the M1 house are made from calcium-silicate blocks with a minimum bulk density of 1800 kg/m³. The “Ytong Energy-Plus” used to construct the external walls is an Ytong design which can also be described as a compound or sandwich solution. These blocks create a monolithic structure which has the advantage of combining different bulk densities in the same raw material – in this case autoclaved aerated concrete. The internal and external layers of the “Ytong Energy Plus” consist of autoclaved aerated concrete with a bulk density of approximately 350 kg/m³. Sandwiched in between these is a factoryfitted layer of Multipor AAC. This combination of different bulk densities has already been used successfully for the past two years in Denmark, where new buildings are famously required to have U values of 0.15 W/(m²K); not to comply with funding schemes, but as an everyday requirement. On account of the thermal properties of the autoclaved aerated concrete conferred by the different bulk densities, no further insulation measures are required for the external wall. The core of the structure remains monolithic, water vapour permeable and not thermally decoupled from the internal space. In spite of (or indeed because of) the low U- values, users of the building will not notice any differences from other houses of solid masonry construction built according to the latest standards set out by the German Energy Saving Ordinance (EnEV). Vapour retarders Figure 5. Preparing the foundations 239 Figure 6. Applying the base layer of insulation Figure 10. Erecting the roof trusses Figure 7. Preparing the top layer of the floor slab Figure 11. Fitting the windows and the roller shutter boxes These images show that the design and construction of the M1 house is very similar to that of conventional houses – and so it should be. But what about the energy balance? Can these construction methods be used in the future to build houses with a final energy demand calculated as an annual balance that is below zero? Basically we are talking about an energy demand figure that users can measure in terms of cents and euros. The monitoring phase which will be taking place over the next two years will ultimately provide the answer. In the meantime we have to rely on mathematical calculations to determine the energy demand when presenting our arguments. In this case an energy balance for the M1 house was prepared using the DIN V 18599 tool. The first parameter – you could call it the mother of all energy-saving strategies – is the heating demand (also referred to as the useful (i.e. effective) energy demand). This is controlled mainly via losses and gains across the building envelope, internal and solar heat gains and the underlying ambient climate. On the basis of the aforementioned standard (2011 version) we calculated an annual heating demand of 3905.50 kWh for the building, which corresponds to 20.90 kWh/m² based on the usable floor space. Figure 12 illustrates the monthly calculated heating demand. We did not include a useful energy demand for cooling the building as there is no provision for active air conditioning. Even with high external temperatures, sun-screening devices for the windows combined with the high thermal capacity of the cal- Figure 8. Laying the external block work Figure 9. Concreting the reinforced concrete ceiling 240 cium-silicate block internal walls is enough to prevent rooms from overheating, which is commonly associated with lightweight constructions. Figure 13. Monthly yields of photovoltaic In compliance with the funding guidelines, a fixed annual value of 2500 kWh was stipulated as the final energy demand for lighting and household appliances. The solar thermal inputs are derived from a model calculation based on DIN EN 12976-2 and DIN V EN 12977-2 for reference conditions in Würzburg, Northern Bavaria. Heat generated by the air-to-water heat pump was calculated on the basis of the algorithms listed in DIN V 18599. The energy balance ratios from both of these sources are shown in Figure 14. The solar thermal installation and heat pump combined can generate a total of 6874 kWh for the building during the course of the year. The light gray bar in figure 14 is related to the yield of solar thermal, the dark gray is for the yields of the heat pump. Figure 12. The building’s heating demand To determine the building's final energy demand from the heating demand, you have to include the technical building equipment used. Considerations which ultimately determined the choice of technical building equipment can be found in Section 10. At this stage we will simply point out that inputs from the photovoltaic panels, the solar thermal installation and the air-to-water heat pump were included in the energy balance. The renewable energy input was calculated on the basis of DIN V 18599-9 Version 2011. The solar input was calculated on the basis of the “reference climate” described in DIN V 18599-10: 2007. The system performance factor was adjusted to ensure that the efficiency of the system calculated according to the standard was consistent with the system actually used. The results of the energy input calculations based on DIN V 18599-9:2011 are shown in Figure 13. The results from the energy input calculations for the photovoltaic installation are compared with the results from a simulation. This suggests that a higher yield would be possible by including the results from the simulation (light gray) and calculations (gray). Figure 14. Yields from heat pump and solar thermal The results of the final and primary energy demand calculations are shown in Tables 2 and 3 respectively. 241 Table 2. Final energy demand Annual final energy demand specific absolute (based on calorific value) [kWh/(m²a)] [kWh/a] Heating (yields from heat 10.14 1,894.66 pump and solar thermal are already considered) Hot water (yields from heat 1.45 270.89 pump and solar thermal are already considered) Total 11.59 2,165.55 energy balance Lighting and household appli- 13.38 2,500.0 ances (default) Renewable energy generation (only power generation) - for own use -17.24 -3,222.4 - fed into the grid -29.24 -5,464.8 Total -21.52 -4,021.7 Table 3. Primary energy demand Annual primary energy demand (based on calorific value) specific [kWh/ (m²a)] 24.33 constructing a monolithic masonry structure, this baseline condition does not exist per se because the external wall assumes both the load-bearing and insulating function. The development of energyefficient masonry houses sets very high standards for minimising additional losses through thermal bridges. Blanket approaches such as those typically found in EnEV certification procedures result in high energy demand values in the building's energy balance. The objective must be to prevent these losses occurring in the first place through careful attention to detail. Right from the design stage, all design details for the M1 energy-plus house could be catalogued and calculated on the basis of DIN EN ISO 10211 in connection with DIN 4108 Supplement 2. The necessary adjustments which inevitably occur in any design stage were incorporated into the technical build specifications. Thermal bridges were detected on the basis of both the draft design and subsequent detailed design. In this regard, connections to the floor slab and windows, including the reveals, are of critical importance for the Ytong Energy-Plus external blockwork. Every detail of the final design was submitted to the design engineer’s verification management system in a scale of 1:50, analysed using the thermal bridge program “Psi-Therm” and then discussed in the group project meetings to determine the potential for optimisation. Figure 15 illustrates one such design detail. absolute [kWh/a] Heating (included yields from 4,547.18 heat pump and solar thermal) (final energy multiplied with primary energy factor 2.4) Hot water (included yields form 3.48 650.14 heat pump and solar thermal) (final energy multiplied with primary energy factor 2.4) Total 27.81 5,197.32 Primary en- Primary Ratio su- specific absolute ergy deenergy peri[kWh/(m²a)] [kWh/a] mand acfactor or/inferior cording to calorific energy value fHS/HI source Increased scope of energy balance with energy-plus house Lighting and 2.40 1.0 32.11 6,000.0 household appliances (default) Renewable energy generation: - for own 2.40 1.0 -41.38 -7,733.8 use - fed into 2.80 1.0 -81.88 -15,301 the grid Total -63.34 -11,838 Figure 15. Design detail 3 MINIMISING THE EFFECTS OF THERMAL BRIDGES The results of individual thermal bridge calculations are shown in Table 4. It also shows the temperature factor fRSI of the thermal bridges. DIN 4108 requires a minimum value of 0.70 to prevent mould growth on the surface. Long before the development of the passive house we have known that the significance of additional losses through thermal bridges rises in line with increasing levels of insulation in the building envelope. It is often said that “shell insulation”, in other words a separation of the load-bearing and insulating function of the heat-transferring building envelope, represents the ultimate insulation measure. If you are 242 Table 4. Thermal bridging detail (selection) Figure 16. Water heating system Key components are: − − − − heat pump, external section heat pump, internal section 1000 litre hot water tank, tank within a tank solar thermal installation with tube collectors The power required to drive the system components shown in Figure 16 is supplied by the on-roof photovoltaic panels, which have a net surface area of 65.51 m² and an output of 9.75 kWp. Thin-film photovoltaic modules on the building facade (0.75 kWp) and additional modules on the solar carport (2.07 kWp) also generate electricity. A residential ventilation system with heat recovery provides the building's ventilation. An infinitely variable control system enables the user to adjust the level of ventilation required. Based on ventilation systems typically used in contemporary newbuilds, these components are in fact far from typical. Today window ventilation is still typically controlled by the user alone. Ventilating energy-plus houses solely via window ventilation is critical in terms of the energy balance, since apart from transmission heat losses, ventilation heat losses must also be kept to a minimum. But this design allows users themselves to decide whether to adjust this ventilation regime during transitional or summer months by shutting down the system. Since the ventilation system is not the sole heat source, as is the case in a passive house for example, this option is always available. The monitoring phase which will take place over the next two years will show the extent to which the building's users (the test family) will have recourse to this option. The system itself is a state-of-the-art ventilation system with a high heat recovery rate (≥ 0.85) and low fan power consumption (≤ 0.40 W/m³h). The Multiwatt energy system can store electricity generated from renewable energy sources. The one used in the M1 house has a capacity of 24 kWh. The Multiwatt ® energy manager monitors the energy coming from the PV installation. This electricity is delivered to the house as a priority, and any surplus When all losses through thermal bridges are added up and divided by the heat-transmitting building envelope, the M1 house achieves a value of 0.01 W/(m²K). 4 WHAT TECHNICAL BUILDING EQUIPMENT SHOULD BE INCLUDED IN THE HOUSE? This question cropped up in many discussions right from the initial stages of the M1 project. After all, the aim was to create a positive energy balance which took account of all energy-consuming components. How many photovoltaic panels does the house need, is solar thermal necessary, what affordable alternatives are available? And: Which of them will be acceptable to the user, which systems will he or she gets along with? In the end we were left with a combination of systems which are already used in newbuilds today: Photovoltaics, solar thermal, 1000 litre water tank, heat pump, ventilation system with heat recovery. The M1 energy-plus house is ultimately a house powered by electricity. Figure 16 gives an overview of the system components for heating water (for heating and hot water) and how they connect with one another. For further and more detailed information please see www.m1-ernergieplus.de. 243 6 REFERENCES is used to charge the battery. Any surplus power remaining after that is supplied to the power supply company (sold). When the PV installation does not generate enough power (at night or during bad weather), the energy manager switches to grid consumption, i.e. connected consumers are supplied, but no power from the battery is fed into the grid. When the PV installation generates enough power again, the energy manager switches back to self-consumption, charges the battery with PV energy and supplies power to the consumers in the house. In the event of a power failure, the energy manager switches the connected consumers completely (all poles) from the grid and continues to supply them entirely independently. This is referred to as “blackout switching”, and it remains active until the power supply company restores the power. The controller continuously displays charging and discharging cycles to enable times of consumption and capacities to be analysed. Hegner H.-D.: „Energieausweise für die Praxis Handbuch für Energieberater, Planer und Immobilienwirtschaft“, 2. Auflage 2010, Bundesanzeiger Verlag Köln/Fraunhofer IRB Verlag Stuttgart Hegner H-D., Schoch T.: Zukunftssicher Bauen, wie die Energiewende das Bauen verändert, Mauerwerk-Kalender 2013, Verlag Ernst und Sohn, Berlin 2013 Information der Bundesregierung unter: http://www.bundesregierung.de/Content/DE/_Anlagen/2011 /06/2011-06-06-energiekonzept-eckpunkte.pdf „Der Weg zur Energie der Zukunft“ Schoch T.: Energieeinsparverordnung 2009, Wohnbau, 2.Auflage 2010, Bauwerk-Verlag Berlin Schoch T.: Auswirkung der Energieeinsparverordnung auf den Mauerwerksbau, Das Mauerwerk, Heft 3, Verlag Ernst und Sohn Berlin 5 CONCLUSION Using highly efficient technical building equipment combined with the benefits of solid masonry construction, we have developed a house which generates more energy than it consumes as part of a research project. This development has demonstrated that an energy-plus house can be offered on the market. In the heating technology sector, heat pumps will gain a greater market share, whilst various cogeneration systems will also capture a share of the market. Although conventional heat generation will remain in use in the existing housing stock, it must be capable of working in combination with renewable energies. In the coming years further improvements are expected to be made to various efficiency parameters. The rapid fall in the cost of photovoltaics alone makes it possible to generate affordable electricity which can compete with grid costs. The installation of energy-generating systems will be high on the agenda. Various forms of solar thermal heating, photovoltaics and biomass based on a variety of different technologies must be incorporated into the structure and architecture of buildings. One of the greatest challenges we will have to overcome in the next few years is how to embed this technology into architectural design. Another new consideration is that buildings which generate energy must also be able to manage this energy. The aim must be to achieve a particularly high level of self-consumption. Furthermore, connection to other homes in the neighbourhood and to electric vehicle charging points must also be feasible. 244 Simulation of thermal performance and retrofit of a historic greenhouse R. Ward, A. Mortada & R. Choudhary Energy Efficient Cities initiative, Cambridge University Engineering Department, Cambridge, UK. ABSTRACT: Typical building energy simulation programs do not adequately describe the physical processes of heat and mass transfer which occur in a greenhouse, primarily because they do not include the interaction of the plants with their environment. This is of concern when simulation studies are required in order to assess different options for improving the greenhouse thermal performance. A previous paper (Brown et al 2012) has described the development of a model for simulation of ornamental glasshouses. Further development, in order to represent more accurately the physical processes and assess their significance, is described in this paper. The development has been undertaken in support of retrofit analysis for the glasshouses at the Royal Botanic Gardens, Kew, in London, which present unique challenges in terms of their historic nature, construction details and design and have provided an opportunity to investigate the relative importance of different physical effects on the energy consumption. Within this context, the simulation has been used to explore options for improving greenhouse thermal performance. analysis and concludes with directions for future work. 1 INTRODUCTION Ornamental glasshouses are built with the primary purpose of allowing the plants inside to thrive, despite external weather conditions being substantially different from the plants’ native climate. This requires the glasshouse to be maintained at beneficial temperature and relative humidity conditions. However, the plants interact with the surrounding air in a way which acts to reduce the temperature, so the heat required to maintain set-point temperatures may be higher for a glasshouse than for a building housing people as the primary occupants. The appearance of ornamental glasshouses can also be important for heritage and aesthetic reasons, which can mean that technologies which could facilitate reduced energy consumption may not be applicable. A model has been adapted from a previously developed commercial greenhouse model to be applicable to an ornamental glasshouse (Brown et al 2012). The first section describes the model and its adaptation to incorporate additional physical processes. The glasshouses at the Royal Botanic Gardens (RBG), Kew in London, UK, are described next as they form the portfolio of buildings considered in this study. Application of the model to the RBG Kew glasshouses is described and validation of the simulation against metered gas consumption data is presented. Retrofit opportunities are explored using the simulation as a basis for predicting energy saving potential. The paper finishes with a discussion of the usefulness of the simulation for retrofit 2 MODEL DESCRIPTION A one-dimensional model has been developed to simulate the flow of heat through an ornamental glasshouse incorporating the interaction of the vegetation with its surroundings, specifically the heat and mass flow due to transpiration (Brown et al 2012). The model is based on the Gembloux Dynamic Greenhouse Climate Model (GDGCM) (Pieters and Deltour 2000) originally created for application to commercial greenhouses, Differences between commercial and ornamental glasshouses are primarily in geometry and operation; commercial greenhouses tend to be of low, uniform construction with few architectural development restrictions, operated under tight controls to maximise crop production, whereas ornamental glasshouses may be tall, multizone and designed equally for aesthetic impact as well as operational efficiency, operated to maximise display impact and plant preservation. Simulating an ornamental glasshouse has therefore required extension of the model to consider heat transfer between different climatic zones within the glasshouse and vertical temperature stratification. Using the model for simulation of retrofit opportunities has required additional consideration of ventilation and infiltration of external air. 2.1 Basic model The model is composed of a system of onedimensional heat and mass balance differential equa245 is the temperature at ground level, the temperatures in the layers are then calculated using the defined temperature gradient. Heat demands are linearly approximated in a similar way. In order to use this approach, it is necessary to define the magnitude and variation of the temperature gradient which will be dependent on a number of factors including the building geometry, time of day/year and weather conditions. The temperature inside the Palm House was measured half-hourly at 4 different heights for a period of 2 months from March to May. The average daily profile of air temperatures is shown in Figure 1. It was noticed that the diurnal temperature followed an approximately sinusoidal pattern, reaching maximum in midafternoon. Not evident from the figure, but observable from the raw data, is that the period of the sinusoidal variation increases with increasing day length. Based on these observations, and extrapolating for the entire year, the monitored data has been used as a basis for approximating how the temperature gradient up the height of the building, dT/dz, changes throughout the year. i.e. tions which calculate the temperature variation in defined layers of the greenhouse with time, subject to the external weather conditions and deep soil temperature. The layers consist of the soil, vegetation, inside air and cover, sandwiched between the sub-soil and outside air and sky. Fluxes between layers are vertical, with the exception of solar radiation for which the angle of incidence varies with time. Each layer in the model has a finite heat capacity with temperature varying according to the heat exchanged between the layers. The heat transfer processes considered include convection, farinfrared radiation, solar radiation and latent heat, together with convective heat transfer from the heating system. Heat transfer can also occur between zones, calculated using a simple U-value calculation. Details of the methodology for simulation of the dominant heat transfer processes are given in Brown et al (2012) and in Pieters and Deltour (2000). A significant contributor to the overall heat flux is the latent heat flux caused by plant transpiration, determined by the plant’s stomatal resistance, rst. This value is simulated here using the approach developed by Jolliet and Bailey for a tomato crop (Jolliet and Bailey 1994), which uses an empirically derived equation dependent on the solar radiation flux density, qSint and the difference between the vapour pressure in the air and the saturated vapour pressure at the leaf surface, vpd, i.e. 𝑟𝑠𝑡 = 244 200 −0.22𝑣𝑝𝑑 1−0.66 𝑞𝑆𝑖𝑛𝑡+200 𝑑𝑇 𝑑𝑧 600 = 𝐴 + 𝐵. sin(𝐶. 2𝜋. �𝑚𝑖𝑛 − 1440�) (2) A and B are empirically derived parameters based on the maximum and minimum temperature gradients throughout the day. For the case of the Palm House, A=0.6 and B=0.3; A+B is the maximum observed temperature gradient and A-B is the temperature gradient observed at dawn and dusk (oC/m). C is a parameter dependent on the length of the day, and min is the time of day in minutes. (1) As the solar radiation levels rise, the transpiration rates rise, counteracting potentially damaging increases in leaf temperature. 2.2 Model adaptation 2.2.1 Vertical temperature stratification Inherent in the GDGCM model is the assumption that the air temperature inside the greenhouse is uniform. The validity of this approximation decreases as the height of the greenhouse increases and the temperature varies significantly from floor to roof level. For a tall greenhouse, the model will therefore overestimate the heat demand. The incorporation of the heat exchanges occurring within the movement of air would require extensive CFD modeling which would be cumbersome and would penalize the model in terms of calculation time. A simple method to incorporate the effects of stratification is to divide the glasshouse into layers based on elevation heights, where each layer is assumed to have a uniform temperature. The temperature gradient up the height of the building is defined, and assuming that the internal temperature calculated by the simulation Figure 1. Average daily profile of air temperature in the Palm House, March – May. 2.2.2 Ventilation Heat transfer to the external air across the building envelope via controlled and uncontrolled air flow is a primary cause of heat loss, and it is essential to include adequate representations of ventilation and in246 𝑆𝑒′ = filtration in the model. The original model assumed a constant fresh air rate, but this approach is not adequate to allow investigation of the effect of improving glazing seals or modifying ventilation controls. The ventilation model has been adapted to simulate the operation of air vents as the internal air temperature changes. The ventilation rate is at a minimum value when the internal air temperature is lower than the ventilation set-point temperature, Tv_min. As the temperature rises above the set-point, the ventilation rate is assumed to be proportional to the difference between the actual inside air temperature and the ventilation set-point temperature, representing the opening of vents and subsequent increase in ventilation rate. At a defined temperature, Tv_max, it is assumed that all vents are open and that a maximum possible ventilation rate, Ra_max has been reached. 𝐹𝑜𝑟 𝑇𝑖 < 𝑇𝑣_𝑚𝑖𝑛, 𝑅𝑎 = 𝑅𝑎_𝑚𝑖𝑛 𝐹𝑜𝑟 𝑇𝑖 > 𝑇𝑣_𝑚𝑎𝑥 , 𝑝𝑤 = 0.5 𝐶𝑝 𝜌𝑜 𝑣 2 𝑝𝑠 = −𝜌𝑜 𝑔ℎ(1 − 𝑇𝑜 ⁄𝑇𝑖 ) 1 2𝑆𝑒′ (3) and 12 𝜇 𝐿 𝑏2 𝐴 (8) The Royal Botanic Gardens at Kew, London is a UNESCO World Heritage Site, comprising 121 hectares of gardens and over 50 buildings, many of which are of heritage status. There are 3 large glasshouses, illustrated in Figures 2 – 4, namely the Palm House, an iconic 19th century structure, the Temperate House, the largest surviving Victorian glasshouse in the world and currently undergoing extensive restoration work, and the Princess of Wales Conservatory (POWC), a multi-zone energyefficient glasshouse (Bunn 1986) opened in 1987. The three glasshouses present different challenges for simulation. The POWC is the most complex of the glasshouses, with 10 inter-connected climatic zones each maintained at different temperature and humidity levels to replicate various climatic conditions from around the world. The total floor area covered is 4490m2, with a volume of 19630m3. The largest zone, the ‘Wet Tropics’, which occupies more than 50% of the volume is maintained at a daytime temperature of 20oC, dropping to 18oC at night. The low-profile design of the POWC, oriented approximately on a N-S axis, is intended to maximize the amount of winter sun, while facilitating solar reflection in summer to prevent overheating. The plants housed are primarily herbaceous, and hence the maximum height of the building is just 7m. By comparison, the Palm House houses tall tropical plants and has a maximum height of 20m in the high central section. It has a relatively small floor area, just 2273m2, but a volume of 17678m3 and is oriented approximately on a NW-SE axis. It is maintained at the same temperature as the POWC Zone 1, but its height means that there is significant temperature stratification from ground to roof level. 𝑅𝑎 = 𝑅𝑎_𝑚𝑎𝑥 (4) where for a crack of length L (m), depth b (m), cross-sectional area A (m2), internal air density ρa (kg/m3) and dynamic viscosity µ (Ns/m2); 𝑆𝑔 = (7) 3 ROYAL BOTANIC GARDENS, KEW 𝑅𝑎_𝑚𝑎𝑥 − 𝑅𝑎_𝑚𝑖𝑛 . (𝑇𝑖 − 𝑇𝑣_𝑚𝑖𝑛 ) 𝑇𝑣_𝑚𝑎𝑥 − 𝑇𝑣_𝑚𝑖𝑛 (�𝑆𝑔2 + 4 ∆𝑃 𝑆𝑒′ − 𝑆𝑔 (6) In which Cp is the static pressure coefficient assumed to be 0.62 (Eurocode 1 2005), ρo is the external air pressure (Pa), v is the windspeed (m/s) at the reference height, h (m) and To and Ti are the external and internal temperatures respectively (K). The flow rate through large gaps, such as open vents, has been calculated using the standard orifice flow equation, assuming turbulent flow. 2.2.3 Infiltration Infiltration has been simulated by calculating the contributions from wind pressure and stack effect according to external and internal wind speed and temperature. The flow rate through small gaps in the cover i.e. between glazing panes and frames has been calculated using the approach presented in (Hagentoft 2001) for air flow through a gap in an air-tight building envelope. This approach calculates the flow rate through a gap for laminar flow, Ra, according to the difference between the internal and external pressure, ΔP (Pa), the pressure loss inside the gap, Sg, (Pa.s/m3) and the air flow resistance, Se (Pa/(m3/s)2) , according to the equation; 𝑅𝑎 = 2 𝐴2 The pressure difference has been calculated from a quadratic sum of the wind pressure, pw and the stack pressure, ps (Pa), where; 𝐹𝑜𝑟 𝑇𝑣_𝑚𝑖𝑛 < 𝑇𝑖 < 𝑇𝑣_𝑚𝑎𝑥 , 𝑅𝑎 = 𝑅𝑎_𝑚𝑖𝑛 + 1.8 𝜌𝑎 (5) 247 data for the POWC for 2009-2010 and 2012-2013 and the Palm House cluster for 2012-2013. The cluster comprises the Palm House, the Waterlily House, which is a small glasshouse in comparison and a nearby café/shop complex, which may consume up to 10% of the total metered gas consumption. Even taking this into account it is clear that, despite its smaller floor area, the gas consumption of the Palm House is of a similar order of magnitude to that of the POWC. Figure 2. Palm House. Figure 3. Temperate House. Figure 5. Comparison of metered gas consumption. 4 SIMULATION AND VALIDATION 4.1 Princess of Wales Conservatory An initial simulation of the POWC using this approach has been described in detail elsewhere (Brown et al 2012). The six largest zones have been simulated, amounting to 97% of the total floor area. Zone geometries, including volumes, floor and glazing areas have been taken from construction drawings, and set-point temperatures and humidity levels have been provided by RBG Kew. Vegetation characteristics have been derived from a global database (Scurlock et al 2001). For the purposes of retrofit analysis, however, it is important to include an adequate representation of ventilation and infiltration so that changes to these parameters can be investigated. The original model assumed a constant ventilation rate and did not consider infiltration explicitly. For this analysis, ventilation and infiltration have been simulated in more detail. As detailed previously, the ventilation model simulates the opening of vents once a ventilation setpoint temperature, Tv_min of 2oC above the day-time set-point temperature is reached. The minimum ventilation rate is assumed to be currently zero, while the maximum ventilation rates, Ra_max are estimated based on the vent opening area of each zone, and are assumed to be reached once the internal air temperature reaches a value of 10oC above the ventilation set-point temperature. Set-point temperatures Figure 4. Princess of Wales Conservatory. The Temperate House is the largest of the glasshouses and also the coolest. It has a floor area of 4571m2 and a volume of 42012m3, with a height of 19m in the central largest zone. As for the POWC, it is oriented approximately on a N-S axis and comprises 5 zones connected linearly, with a main building, flanked on the North and South sides by smaller buildings. It is maintained at a minimum temperature of 7.25oC, rising to 13oC in the daytime in the warmest zone. These three glasshouses present a significant heat demand, with the POWC alone consuming 32% of the total gas consumption for the site. In order to validate the simulations, it is necessary to extract gas consumption data where available. However, due to the nature of the site, gas metering at the building level has until now been sparse. Metering is in place and data will be available going forward, but there are few historic data available. Detailed gas consumption data are available for the POWC from 2009 onwards, but data are only available for the Palm House and proximal buildings from mid-2012 and there are no data yet available for the Temperate House. Figure 5 below shows the gas consumption 248 vious simulation underestimated annual gas consumption by 15%, this simulation is overestimating consumption by 10%, in particular over-predicting in the latter half of the year. The more detailed simulation is a better representation of the reality and facilitates investigation of retrofit strategies which affect fresh air rates. and maximum ventilation rates for each zone are given in Table 1. Infiltration has been calculated and added to these ventilation rates to estimate the overall fresh air rate for each zone. The POWC is constructed from overlapping panes of glass with seals between the panes. Over time, these seals have degraded to an extent where the gaps between the panes are up to 8mm in places. For the infiltration calculation, an average crack depth, b, of 4mm has been assumed at each overlapping glazing joint, with an assumed length, L, of 0.1m. This gives rise to a maximum estimated infiltration rate for Zone 1 of 3.6ACH. 4.2 Palm House The Palm House has been simulated as a single zone with ventilation and infiltration modelled as described previously, but for this analysis vertical stratification of the temperature has also been considered, and the effects are illustrated below. As for the POWC, the building geometry has been taken from construction drawings where available, and also from site measurements. The setpoint temperature is 20oC during the day and 18oC at night and the ventilation set-point temperature has been set at 22oC, with a minimum ventilation rate of zero and a maximum of 60 ACH. The vegetation details are similar to those used in Zone 1 of the POWC, with the exception that the cultivated fraction of floor is slightly lower for the Palm House at 0.58, compared against 0.7 for the POWC Zone 1. The crack geometry for the Palm House is different from the POWC in that the glazing panes do not overlap, but are installed side by side, leading to primarily vertical gaps. For this geometry, the crack length, L, has been assumed to be 3mm, corresponding to the thickness of the glass, while the crack depth has been assumed to be an average of 0.5mm. This gives rise to a maximum estimated infiltration rate of 6.7ACH. Stratification has been simulated using the approach outlined previously, making use of measured temperatures which were taken over a 2 month period from March to May 2010. Based on the experimental data, the minimum gradient is assumed to be 0.3oC/m at dawn and dusk and 0.9oC/m in mid afternoon, hence parameters A and B in Equation 2 are 0.6 and 0.3 respectively. The volume of air is divided into three layers, and the air temperature is calculated at ground level and at elevations of 8.45, 10.5 and 20.5m. The temperatures are then averaged across each layer, and the results used to calculate the heat demand for each layer. Results for the analysis with and without temperature stratification included are given in Figure 7, with the metered data for 2012-2013 provided for comparison. Using this simple model of stratification results in a 18% drop in predicted gas consumption over the year. The effect is most significant in the winter months as in the summer the gas consumption is dominated by water heating. Compared against the metered data, the model without stratification appears initially to give a better agreement, but the me- Table 1. POWC zone set-point temperatures and maximum ventilation rates. Zone 1 2 3 4 5 6 Climate Wet tropics Winter garden Dry tropics Tropical ferns Temperate ferns Tropical orchids Internal set-point temperature (oC) Day Night 20 18 20 18 14 12 22 18 15 13 21 18 Ra_max (ACH) 53 65 104 77 36 43 Temperature stratification has not been included for this glasshouse as it is relatively low in height. The model outputs heat demand for these 6 zones. The predicted gas consumption for the whole building has been derived by factoring up to account for the remaining zones, and then assuming a seasonal boiler efficiency of 85%. A constant monthly contribution for water heating has also been estimated from the metered data and added in to the predicted consumption figures. A comparison of predicted and metered gas consumption data for 2009-2010 is shown in Figure 6. Also included in the figure are the results for the prior simulation assuming a constant 1.0ACH ventilation rate and no added infiltration. Figure 6. POWC gas consumption. The inclusion of step ventilation and infiltration in the simulation increases the predicted gas consumption by approximately 25%. Whereas the pre249 approximately 1cm, with an average gap between panes of 3mm, dimensions used in the infiltration calculation, giving rise to a maximum infiltration rate in Zone 3 (the main building) of 5.5ACH. Vertical temperature stratification has not been incorporated into the Temperate House model, as although it is as tall as the Palm House, monitored temperatures reveal substantially lower temperature differences between ground and walkway level than observed in the Palm House. The gas consumption predicted for the Temperate House, not including water heating, is shown in Figure 8. tered data includes consumption for additional cluster buildings which may amount to 10% of the total. Taking this into account, the model without stratification predicts a gas consumption 7.5% higher, and the model with stratification 10% lower than the metered data. More detailed temperature monitoring would facilitate further development of this approach and a greater quantity of metered data would provide further validation. Figure 7. Palm House gas consumption. 4.3 Temperate House Figure 8. Temperate House gas consumption. The Temperate House has been simulated as 5 zones, corresponding to the 5 connected parts of the building. Compared against the POWC, however, it is a much simpler model as the zones are connected linearly, and the temperature is more uniform throughout all zones, as detailed in Table 2 below. Ventilation and infiltration have been modelled in a similar way to the other buildings. The ventilation set-point temperature has been assumed to be 2oC above the daytime set-point temperature. The minimum ventilation rate is assumed to be zero throughout. The maximum ventilation rate, Ra_max, for each zone, given in Table 2 below, has been estimated based on the number of vents and the vent open area. However, it is known that at present not all vents are operable, a situation to be rectified in the proposed restoration. Table 2. Temperate House zone properties. Zone Climate Internal set-point temperature (oC) Day Night 1 South Block 13 7.25 2 South Octagon 7.25 7.25 3 Main building 7.25 7.25 4 North Octagon 7.25 7.25 5 North Block 10 7.25 In the absence of metered data, it is difficult to validate these results. However, it is informative to compare the results against those obtained for the Palm House and POWC which have been validated against metered data. Values are given in Table 3 for normalised gas consumption predictions for space heating. Also included in the table are results for Zone 3 of the POWC alone. This is the coolest zone in the POWC, maintained at 14oC daytime/12oC night-time. Results for two zones from similar glasshouses analysed at the Cambridge Botanic Gardens (CBG) are also included (Mortada 2012). These are very small zones, just 90 and 80 m2 floor area respectively, but they are maintained at relatively low set-point temperatures, and hence yield a useful comparison. The two zones comprise a ‘Mountain’ zone, maintained at a daytime temperature of 10oC dropping to 7oC at night, and a ‘Propagation’ zone, maintained at 10oC constantly. The results of the simulation of the CBG glasshouses have also been validated against metered data. As can be seen, the predicted total annual gas consumption is substantially less than that of the POWC or Palm House. Although the Temperate House is larger than the other two glasshouses, it is maintained at a considerably cooler temperature. Comparing normalised results for the cooler zones, it is clear that the set-point temperature has a significant effect on the gas consumption, as would be expected. The Temperate House results are similar to Ra_max (ACH) 44 87 70 87 44 The Temperate House is of more traditional construction than the Palm House, with overlapping glazing panes. The overlap has been assumed to be 250 those for the cool zones in the CBG glasshouses. These results give confidence that the analysis is appropriate. Temperate House, of not impacting on the appearance of the buildings in any way. The benefit is greatest for the Palm House, but may be harder to achieve owing to the nature of the joints between the glazing panes. Table 3. Comparison of gas consumption predictions. Glasshouse Annual gas consumption (not including water heating) Total Normalised by (x 106 kWh) volume kWh/m3 Temperate House 1.23 29 Palm House 2.08 118 POWC 2.62 113 POWC Zone 3 0.28 67 CBG ‘Mountain’ 0.01 32 CBG ‘Propagation’ 0.01 42 Table 4. Reduction in gas consumption after retrofit. Retrofit % reduction in gas consumption POWC Palm Temperate House House Increased boiler effi8 8 8 ciency Infiltration reduction 23 39 32 Low-E glazing 12 6 7 Night-time shading 10 5 7 Installation of low-emissivity glazing film and night-time shading are potentially beneficial, but costs and practicalities of installation must be taken into account. For example, the Temperate House is currently undergoing restoration, and it may therefore be cost-effective to implement these technologies, if deemed suitable, as part of the restoration work. For the POWC and Palm House, however, practical implications may preclude installation until a suitable time. Increasing boiler efficiency is an important retrofit as boiler replacement is likely to occur as part of ongoing estate asset replacement. It may be possible to justify early replacement if the savings can be demonstrated to payback the costs in a beneficial timescale. In addition, wider concerns may drive replacement with alternative technologies such as biomass boilers or CHP which could potentially further reduce costs and carbon emissions. It is also potentially viable to include glasshouses such as these in district heating and/or power schemes. The heat demand for these glasshouses is largest overnight, but in a network with domestic or office buildings for which the heat demand is highest during the day, further efficiencies could be made. For CHP to be a viable supply option, the night-time heat demand would have to be matched with a night-time power demand. This has been utilised in a parallel study investigating the energy supply options for a cluster of buildings around the POWC (Ward et al 2013). 5 RETROFIT OPPORTUNITIES The primary aim of developing the simulation model is in order to use it to predict the effects of retrofit technologies designed to reduce energy consumption. Retrofits need to address the thermal performance of the building without impacting on plant health, which means that levels of solar radiation must not be adversely affected and appropriate ventilation levels must be maintained. The suitability of available technologies is also constrained by the heritage nature of the site; in addition to being a UNESCO World Heritage Site, the Palm House and the Temperate House are recognised by English Heritage Grade 1 listing. This effectively means that potential retrofits must not impact on aesthetic aspects of the buildings and that any technologies deemed pertinent to the listing must be retained. With that in mind, the following retrofits have been considered for all three buildings: • • • • an increase in boiler efficiency from 85% to 92% reduction of infiltration, assuming 50% potential reduction in gap size application of low-emissivity glazing film, resulting in a reduction of emissivity from 0.84 to 0.22. night-time shading, assuming that during the hours of darkness the emissivity is reduced to zero. 6 DISCUSSION AND CONCLUSIONS The impact of the retrofits on predicted gas consumed to meet the annual heat demand is shown in Table 4. It should be noted that the potential retrofit benefits are not necessarily additive. For example, night-time shading will prevent low-emissivity glazing film having any effect during the night. As can be seen, the single most effective retrofit for all of the buildings is the reduction of infiltration. This is also the lowest cost option of the four and has the benefit, particularly for the POWC and the A simulation model has been developed for application to retrofit analysis for ornamental glasshouses. To this end, the original model has been adapted to simulate infiltration, ventilation and vertical temperature stratification. The approach has been applied to three specific glasshouses at the Royal Botanic Gardens, Kew, which vary in design and operation. The model predictions have been shown to be in 251 MPhil Thesis, Cambridge University Engineering Department, Cambridge, UK. Pieters J., and Deltour J. 2000. The Gembloux Dynamic Greenhouse Climate Model – GDGCM, Department of Biosystems Engineering, University of Gent, Belgium. Scurlock J.M.O., Asner G.P., and Gower S.T. 2001. Global leaf area index data from field measurements, 1932-2000. Data set. Available on-line [http://www.daac.ornl.gov] from the Oak Ridge National Laboratory Distributed Active Archive Center, Oak Ridge, Tennessee, USA. Ward R.M., Omu A., Rysanek A., Rainsford C., and Choudhary R. 2013. Analysis and optimisation of retrofit and energy supply strategy across a diverse urban building portfolio. Submitted to Building Simulation 2013. good agreement with metered gas consumption data where available. An assessment has been made of the potential impact on the gas consumption of four possible retrofit technologies, resulting in the reduction of infiltration being shown to be the single most effective retrofit for these particular glasshouses. In this paper, technologies deemed appropriate to the retrofit of heritage glasshouses have been considered, acknowledging that the heritage nature of the site restricts the use of approaches that may be appropriate elsewhere. However, it would be possible to use the tool to analyse a wide range of other retrofit scenarios, bearing in mind that it is important to ensure that plant health is maintained and that the requisite horticultural input is obtained for the design. Analysis of control strategy is also possible. For example, one aspect also considered was the changing of heating schedule times from fixed hours to dawn/dusk control, but this was found to have very little effect on the results as benefits gained in winter from a shorter heating day were outweighed by additional heating requirements in summer. Changes to ventilation control may also be beneficial and further work will address whether a more efficient strategy may be adopted without impacting on plant health. The model is potentially a useful tool to aid the development of a retrofit strategy for ornamental glasshouses, and its use may enable the avoidance of lengthy and expensive large scale trials. By obtaining estimates of the potential reduction in gas consumption attributable to different retrofit technologies, cost benefit analyses may be performed to identify the most cost-effective retrofit strategy. Further work will include an assessment of retrofit costs. More detailed validation of the models will also be performed, particularly for the model of the Temperate House once more extensive gas consumption data has been collected. In addition, the impact of rain on the thermal performance of the glass cover will be investigated. 7 REFERENCES Brown J., Ward R., Choudhary R., and Slater R. 2012 Algorithmic and declarative modelling of a greenhouse. 5th International Building Physics Conference, Kyoto Bunn R., 1986 Exotic Services, Building Services Journal, May. Eurocode 1, 2005 Eurocode 1: Actions on structures - Part 14: General actions – Wind actions EN 1991-1-4:2005 Hagentoft C.-E. 2001. Introduction to Building Physics. Studentlitteratur AB. Jolliet O., and Bailey B. 1994. The effect of climate on tomato transpiration in greenhouses: measurements and model comparison. Agricultural and Forest Meteorology, 58, 4362. Mortada A. 2012, Potential benefits of retrofits and alternative heating methods in Cambridge University Botanic Garden, 252 The benefits of FEM-SS-BES (Finite Element Method, State-Space, Building Energy Simulation) modeling exchange for building physics A.W.M. (Jos) van Schijndel, R.P. (Rick) Kramer Department of the Built Environment, Eindhoven University of Technology, Netherlands ABSTRACT: An overall objective of energy efficiency in the built environment is to improve building and systems performances in terms of durability, comfort and economics. In order to predict, improve and meet a certain set of performance requirements related to the indoor climate of buildings and the associated energy demand, numerical simulation tools are indispensable. In this paper we consider three types of numerical simulation tools: Finite Element Method (FEM), Building Energy Simulation (BES) and State-Space (SS) together. It is concluded that one of the main benefits of FEM-SS-BES modeling exchange is the possibility to simulate building energy performances with high spatial resolution and low computational duration times. 1 INTRODUCTION In this work FEM is just a method of solving Partial Differential Equations (PDEs), like Finite Volume methods (FVM) or Finite Difference methods (FDM). An overall objective of energy efficiency in the built environment is to improve building and systems performances in terms of durability, comfort and economics. In order to predict, improve and meet a certain set of performance requirements related to the indoor climate of buildings and the associated energy demand, numerical simulation tools are indispensable. In this paper we consider three types of numerical simulation tools: Finite Element Method (FEM), Building Energy Simulation (BES) and State-Space (SS). For each tool separately, there exist a vast number of references. Also on two tools combined, i.e. FEM-BES, BES-SS, FEM-SS, there is quite a lot of literature. However there is lack of research on an overall evaluation of the three tools FEM-SS-BES together. In this paper we present benefits of the FEM-SS-BES modeling exchange for building physics. The main reasons for converting models in each other are summarized in Table 1. We start with two combinations that are quite obvious and already commonly used. BES to FEM – BES is used to simulate the energy performance of buildings, using lumped parameter modeling. If local effects are important, FEM can be used to obtain high resolution results based on distributed parameter models and using BES simulation results as boundary values. FEM to SS – FEM based simulations can easily become computational time consuming. One of the methods to improve the computing time is to reduce the mathematical model to a lower order model by using for example a State-Space (SS) approximation. One of the main benefits of SS models is, that very efficient computation algorithms exist, that are able to almost completely reduce the computation time. If such a reduced order SS model is accurate enough, this method can be used for improving computation speed. Table 1. The main reasons for converting models in each other. To FEM BES * Global effects Lumped results SS From FEM BES SS Local effects High resolution results Computation Speed This paper comprehends an investigation of the remaining combinations of Table I. Each combination is presented in a separate Section, including background information and case studies. After these Sections the overall conclusions are provided. * Inverse Modeling * 253 Comsol (2013). In order to compare the Comsol 3D FEM model with the HAMBase (de Wit 2006 & HAMLab 2012) lumped model, an equivalent heat conduction of the air is used in Comsol instead of CFD. 2 FEM TO BES Commonly used within BES tools are zonal approaches of the volumes, assuming uniform temperatures in each zone, and 1D modeling of the walls. Due to the rapid development of Finite Element Method (FEM) software and Multiphysics approaches, it should possible to build and simulate full 3D models of buildings regarding the energy demand. Moreover, the 3D models would also provide detailed (i.e. high resolution) results of the indoor climate and the constructions. The main problem regarding the use of FEM for BES is how to compare a distributed parameter model (FEM) with a lumped parameter model (BES)? Because BES and FEM have quite different approaches, we used the following method: Step 1, start with a simple reference case where both BES and FEM tools provide identical results. Step 2, add complexity and simulate the effects with both tools. Step 3, compare and evaluate the results. For step 1, a suitable reference case was found at the current International Energy Agency Annex 58. It concerns a test box with overall dimension 120x120x120 cm³. Comsol was used to build a 3D model of the test box. In order to compare the Comsol 3D FEM model with the BES lumped model (using HAMBase(de Wit 2006 & HAMLab 2012)), an equivalent heat conduction of the air is used in the FEM model. This provides identical FEM versus BES results. Furthermore, these results are a first step towards more complex 3D FEM simulations including CFD, windows, ventilation, radiation, etc. In principle all these variants can be simulated in 3D using Comsol, with the notification that especially the CFD simulation could become quite time consuming. 2.1.1 Modeling Equation (1) shows the PDE: 𝜌𝑐 𝜕𝑇/𝜕𝑡 = ∇ ∙ 𝑘∇𝑇 (1) where T is temperature (K), t is time (s), ρ is density (kg/m3); c is specific heat (J/kgK) and k is heat conduction coefficient (W/mK) . Equation (2) shows the boundary values: qboundary = h(T – Te) + qirrad (2) where qboundary is the heat flux at a specific boundary (W/m2) , h is the heat transfer coefficient (W/m2K), Te is the external air temperature (K) and qirrad is the net radiation from the sun and sky to the surface (W/m2). The temperature distribution in the test box is simulated using Dutch weather data. As mentioned above the model was implemented and solved using Comsol. The default second-order Lagrange element type was used. The mesh contained 4414 tetrahedral elements and with average element quality of 0.7512. The number of degrees of freedom solved for was 6679 using the PARADISO algorithm with absolute and relative tolerances of 0.001. The temporal convergence error was less than 10-5 for each time step. After the solution was obtained with these settings, the grid dependency was evaluated by a grid refinement study. The latter showed no significant changes in the solution. 2.1.2 Results Figure 2 shows the 3D snapshots of the isosurfaces in simulated by the FEM software. 2.1 Step 1: Reference case For step 1, a suitable reference case was found at the current International Energy Agency Annex 58 (2012). It concerns a test box with overall dimension 120x120x120 cm³. Floor, roof and three of the four walls are opaque, one wall contains a window with opening frame. Details of the overall geometry with the exact dimensions can be found in figure 1. Figure 1. The reference case. We started to build a 3D model of the opaque test box, heavy weight, air change rate: ACH=0 using 254 line) and HAMBase (green line) during the first month. The verification result is satisfactory. Figure 3. Comparison of the simulated mean indoor air temperature using Comsol (blue line) and HAMBase (green line) during the first month for the opaque reference case. From figure 3, two important facts can be concluded: Firstly, these results can be used as an additional verification benchmark for both Comsol as well as HAMBase. And secondly, it is seems to be possible to accurately reproduce a BES simulation using a relative simple heat conduction based FEM model with a equivalent heat conduction coefficient for the indoor air, but so far without CFD and internal radiation. The next steps are to study effect of including windows and CFD. 2.2 Step 2. including windows The achievements of the first step i.e. the reference case were quite successful. Therefore we started to add more complexity in the form of solar irradiation. 2.2.1 Modeling A window was modeled at the south wall of the test box by the use of surface-to-surface radiation module of Comsol. We refer to the appendix for more details. Figure 2. 3D snapshots of the temperature isosurfaces. The main challenge now is how to match the high resolution distributed temperature results of Comsol with the lumped temperature results of the BES model. For this reference case (opaque test box, heavy weight, ACH=0) we were able to get a very good match by using a so-called equivalent heat conduction coefficient for the air inside the box in Comsol. keq = d/R = 1 / 0.34 = 2.9 2.2.2 Results A preliminary detailed result is presented in Figure 4. The solar irradiation inside the box is clearly visible. This figure shows the 3D temperature distribution in the test box with a window at a specific time. This was also modeled in HAMBase. (3) Figure 3 shows the comparison of the simulated mean indoor air temperature using Comsol (blue 255 been improved using newest techniques provided by recent MATLAB versions. Currently, the hourlybased model named HAMBase, is capable of simulating the indoor temperature, the indoor air humidity and energy use for heating and cooling of a multizone building. The physics of this model is extensively described by de Wit (2006). Also, the simplified building models are coupled to a PI-controller which maintains the indoor temperature at 20°C, yielding the characterization of the energy performance. 3.1 State space (SS) model One of many developed SS models is shown in Figure 5. The thermal model is a 3rd order model with 9 parameters. The hygric model is a 2nd order model with 5 parameters. Figure 4. The 3D FEM results of the reference case with a window. 2.3 Step 3, including CFD Currently we are working on the best way to compare these high resolution distributed temperature results of Comsol with the lumped temperature results of the BES model. This is by far the most challenging part of this part of the research. 3 BES TO SS Detailed modeling of the buildings itself may require much effort. Blueprints can be hard to find and destructive methods to obtain building material properties are not allowed. A simplified model with physical meaning is developed which is capable of simulating both temperature and moisture (Kramer 2012). The parameters of the model are derived by an inverse modelling technique which fits the output of the model to measured values of respectively temperature and relative humidity. The result is a combined 3rd-order thermal and a 2nd-order hygric model in State Space form (Linear Time Invariant). The inverse modelling with the developed simplified model is applied to four different case studies for validation purposes. Moreover, the case studies are used to compare the performance of the simplified hygrothermal model and HAMBase, an in-house developed Heat Air and Moisture simulation tool. This whole building model originates from the thermal indoor climate model ELAN which was already published in 1987 (de Wit et al. 1988). Separately a model for simulating the indoor air humidity (AHUM) was developed. In 1992 the two models were combined (WAVO) and programmed in the MATLAB environment (van Schijndel & de Wit 1999). Since that time, the model has constantly Figure 5. the developed thermal model (top) and hygric model (bottom). Thermal model inputs: i. Temperature outdoor; ii. Solar irradiation on vertical NORTH; iii. Solar irradiation on vertical EAST; iv. Solar irradiation on vertical SOUTH; v. Solar irradiation on vertical WEST; vi. Fixed temperature node for contact. plane oriented on plane oriented on plane oriented on plane oriented on modeling ground The reason for splitting up the solar irradiation is that the used model is represented in State Space form. The huge advantage of the State Space form is the very small calculation time. A State Space (SS) model belongs to the family of LTI-models (Linear Time Invariant), therefore the identified parameter fI (factor for solar gain) is time-invariant. Because the azimuth and elevation of the sun are not timeinvariant, it is impossible to derive a good State 256 Space model with only one input signal of the solar irradiation: nevertheless, the solar gain into the building is not a constant factor (fI) times Global Irradiation On Horizontal Plane. The solar model of Perez et al. (1987) is used to calculate the irradiance on a vertical surface. Hygric model inputs: i. Vapor pressure outdoor; ii. Fixed vapor pressure node. 3.2 Validation using a real building The castle of Amerongen (See Figure 6) situated in Amerongen the Netherlands, is selected as reference building. It is a 17th century building, surrounded by a canal, with thick massive walls, varying from 0.7 to 1.5 m thick. The building covers five floors. The main building materials are brick, wood and slate roof covering. The main part of the castle is free floating, i.e. no climate conditioning, and some rooms have limited dehumidification (7 kg/day) and limited heating. Figure 7. The indoor temperature (top) and vapor pressure (bottom), measured and simulated using HAMBase (BES) and SS. To be able to compare and rate how accurately the different models can reproduce a measured temperature or vapor pressure, three performance criteria are used: the MSE (Mean Squared Error), MAE (Mean Absolute Error) and FIT (goodness of FIT). The MSE is calculated according to, 𝑁 1 𝑀𝑆𝐸 = �(𝑦 ′ − 𝑦)2 𝑁 (4) 𝑘=1 where 𝑦 ′ is the measured signal and 𝑦 is the simulated signal. The MAE is calculated according to, 𝑁 1 𝑀𝐴𝐸 = �|𝑦 ′ − 𝑦| 𝑁 (5) 𝑘=1 Figure 6. Surrounding area (left) and exterior (right) of castle of Amerongen. The Goodness of Fit is calculated according to, The measurement data of the Grand Salon is used to identify the simplified model for this room. The identification procedure of the parameters of the state space is published in Kramer (2012). In this paper we show representative results and a summary of the parameter identification performance. Figure 7 presents the measured and the two simulated results i.e. HAMBase and SS for the indoor temperature and vapour pressure. (6) 𝑛𝑜𝑟𝑚(𝑦) is the Euclidean length of the vector 𝑦, also known as the magnitude. The above equation therefore calculates in the numerator the magnitude of the error between measured and simulated signal. This is divided by the denominator, calculating how much the measured signal fluctuates around its mean. Table 2 shows the fit performance indicators for the BES (HAMBase) and SS models. 257 Table 2. The fit performance indictors for the BES and SS models. The fixed vapor node was not used here, because no vapour source was measured. THERMAL MSE [°C2] MAE [°C] FIT [%] HAMBase 0.94 0.70 72.18 Simplified State Space 1.09 0.80 70.00 We summarize the main advantages of SS models: (1) The computation time is extremely fast (~0.02 sec. for an hourly based period of a year). (2) The parameters have, in contrast with black box approaches, physical meaning. This facilitates basic parameter studies using a single SS model. (3) The parameters can be obtained without detailed knowledge of the building itself by using relative simple measurements, i.e. inverse modeling. (4) The SS model can be used for BES purposes. The latter is presented in the next Section. HYGRIC MSE [Pa2] MAE [Pa] FIT [%] HAMBase 8.65e3 72.63 71.89 Simplified State Space 5.88e3 60.23 76.82 4 SS TO BES Furthermore Table 3 and 4 show the numerical values of the paramaters of the SS model. In Section we show how a similar SS model obtained by the approach of the previous section (now called thermal building SS model), can be used to simulate energy performances. Table 3. Numerical values of the thermal parameters. # Par. Value unit 1 Gw/Cw 4.10E-07 s-1 2 Gi/Cw 2.60E-05 s-1 3 Gi/Ci 4.48E+01 s-1 4 Gfast/Ci 2.94E+00 s-1 5 Gint/Ci 3.47E-01 s-1 6 Gint/Cint 1.82E+02 s-1 7 fI-N/Cint 3.55E-05 m2K/J 8 fI-E/Cint 4.99E-05 m2K/J 9 fI-S/Cint 5.27E-05 m2K/J 10 fI-W/Cint 4.17E-05 m2K/J 11 Gfixed/Ci 1.52E+00 s-1 12 Tfixed 3.21E-05 °C A simple control strategy is designed to maintain the indoor temperature of a thermal building SS model at 20°C: a PI-controller is connected with the thermal model and a closed loop is established from the output Ti(ndoor) to the PI-controller, see Figure 8. By connecting the systems, a new model object is generated. Figure 8. The energy model consisting of a PI controller and the original SS building model. The PI-controller is a transfer function with both the proportional gain and integral gain set to 0.01. These gains delivered the most steady indoor climate (20°C) with minimal power supply for thermal SS model of the building. Figure 9 shows the simulated power (ideal heating and cooling) of the controlled (SS) model. Table 4. Numerical values of the hygric parameters. # Par. value unit 1 Gw/Cw 9.22E-10 s-1 2 Gi/Cw 1.62E-05 s-1 3 Gi/Ci 2.56E+01 s-1 4 Gfast/Ci 1.11E+01 s-1 5 Pfixed - Pa 6 Gfixed/Ci - s-1 258 Due to the feedback loop, the Energy Model supplies the power one sample later (1 hr) than HAMBase. If the Energy Model’s output is delayed one hour, the similarity increases as shown in Figure 12. Figure 9. The power supply to maintain an indoor temperature of 20 oC all year. Also HAMBase itself is used to control the indoor temperature of the Simple Building at 20°C by allowing unlimited heating and cooling powers (HAMBase determines the needed maximum power). The resulting power [W] supply of HAMBase is compared to the resulting specific power [K/s] supply of the Energy Model, see Figure 10 (please mind the logarithmic y-axis). Figure 12. Idem as in Figure 11 but Energy Model output delayed 1 hour. Figure 12 shows that the combination of the SS model of Figure 5 with a PI controller can accurately reproduce the simulating heating of the BES (i.e. HAMBase) model. 5 CONCLUSIONS It is concluded that one of the main benefits of FEM-SS-BES modeling exchange is the possibility to simulate building energy performances with high spatial resolution and low computational duration times. Regarding FEM to BES – Firstly, these results can be used as an additional verification benchmark for both Comsol as well as HAMBase. Secondly, it is seems to be possible to accurately reproduce a BES simulation using a relative simple heat conduction based FEM model with a equivalent heat conduction coefficient for the indoor air, but so far without CFD and internal radiation. The latter is left over for future research. Regarding BES to SS - The paper presents case studies where SS models are successfully used for reducing computational times for BES models. Regarding SS to BES – Using this so-called inverse modeling approach, it is possible to obtain building energy performances from SS models. The FEM-SS-BES modeling exchange provides two alternative modeling approaches for each other. This may be beneficial if some specific limitations are encountered within one of the single FEM, BES, SS modeling methods. Figure 10. Necessary power to keep Tindoor of the thermal building SS model at 20°C. The overall comparison of the HAMBase power with the Energy model’s specific power reveals a remarkably accurate similarity. A more detailed comparison is given in Figure 11: the Energy Model’s specific power, multiplied by a factor of 63 (J/K), provides the same excitation level as the HAMBase power. This factor 63 can independently be obtained during a stationary simulation for which the thermal capacities play no role. Figure 11. Detailed view with specific power of Energy Model scaled (x 63). 259 6 NOMENCLATURE A B C Ci Cint Cp Cw D d fI Gfast Gi Gint Gw Irrad P Pe Pfixed Pi Pisim Psat Pvapour RH SG T Te Tfixed Ti Tint Tisim Tw u x y ε k ξ ρ state matrix, area [m2] input matrix output matrix indoor air capacitance [J/K] interior capacitance [J/K] specific heat capacity [J/kgK] envelope capacitance [J/K] transition matrix wall thickness [m] effective irradiance [m2] conductance from indoor air to outdoor air [W/K] conductance from indoor air to envelope [W/K] conductance from indoor air to interior [W/K] conductance from envelope to outdoor air [W/K] solar irradiation [W/m2] power [W] outdoor air vapour pressure [Pa] fixed vapour pressure [Pa] indoor air vapour pressure [Pa] simulated indoor air vapour pressure [Pa] saturation pressure [Pa] vapour pressure [Pa] relative humidity [-] and [%] solar gain factor of glazing [-] temperature [°C] outdoor air temperature [°C] fixed temperature [°C] indoor air temperature [°C] interior partitions temperature [°C] simulated indoor air temperature [°C] temperature of walls [°C] input vector state vector 7 REFERENCES Annex 58 (2013), http://www.ecbcs.org/annexes/annex58.htm Comsol (2013), www.comcol.com HAMLab(2013), http://archbps1.campus.tue.nl/bpswiki/index.php/Hamlab Kramer R. 2012. From castle to binary code: the application of inverse modeling for the prediction and characterization of indoor climates and energy performances, Eindhoven. Perez R. et al. 1987. A new simplified version of the perez diffuse irradiance model for tilted surfaces. Solar Energy, 39(3), pp. p.221-231. Schijndel A.W.M. van & Wit M.H. de, 1999. A building physics toolbox in MatLab, 7TH Symposium on Building Physics in the Nordic Countries Goteborg, pp81-88 Schijndel A.W.M. van 2007. Integrated Heat Air and Moisture Modeling and Simulation, PhD Dissertation, Eindhoven University of Technology Wit M.H. de, Driessen, H.H. 1988. ELAN A Computer Model for Building Energy Design. Building and Environment 23: pp285-289 Wit M.H. de 2006. HAMBase, Heat, Air and Moisture Model for Building and Systems Evaluation, Bouwstenen 100, Eindhoven University of Technology APPENDIX Radiative Heat Transfer Equations: J = ρG + εσT4 (A1) q = G - J = (1- ρ)G - εσT4 (A2) α= ε=1–ρ q = ε(G - σT4) output vector emission factor [-] thermal conductivity [W/mK] hygric capacity [kg/m3] density [kg/m3] (A3) (A4) 260 Experimental and numerical energy performance analysis of PCMenhanced building envelope products and systems J. Kośny, N. Shukla, A. Fallahi, and Fraunhofer Center for Sustainable Energy Systems, Cambridge, MA, USA E. Kossecka Polish Academy of Sciences, Warsaw, Poland ABSTRACT: Phase-transition properties of PCM-enhanced building materials are critical in designing and optimizing these materials for a given climatic conditions. These transition properties are latent heat, sub cooling, and hysteresis during melting and freezing. There are few whole-building simulation tools containing sufficient numerical algorithms of the detailed physics, which are necessary to provide accurate simulations of the thermodynamic behavior of these structures. In addition, there is a lack of reliable test methods that can accurately measure these phase-change properties. Well-developed Differential Scanning Calorimeter (DSC) testing method can be only used for relatively homogeneous materials. The DSC testing method excludes most of PCM-enhanced building products, PCMs with complex melting/solidification enthalpy profiles, and significant sub-cooling effects. In this study, we utilize a recently developed experimental method based on the existing heat flow meter apparatus and modified ESP-r whole building energy model. This numerical tool is capable of predicting temperature profiles and heat fluxes within different configurations of PCM applications with significant sub-cooling. 1 INTRODUCTION test data on PCM-enhanced building insulations and other PCM-integrated materials – Kośny et al. (2011), Shukla et al. (2012). Based on the extensive testing, we develop a testing procedure for accurate and reliable measurement of the enthalpy as a function of temperature. This paper presents exclusively results from the testing and analysis of the PCMenhanced fiber insulation. Dynamic phase-transition characteristics (latent heat, sub-cooling, hysteresis during melting and freezing, etc…) of the PCM-enhanced building materials are required to perform whole building energy simulations, design work, and energy code compliance analysis. In addition, dynamic test data is critical in optimizing the distribution and location of the PCM within a building to maximize the energy savings. Until recently, differential scanning calorimeter (DSC) has been the only available method to determine the dynamic properties of a PCM – Mehling, Cabeza (2008), Gunther et al (2009). Unfortunately, the DSC method is only valid for small and homogeneous specimens. This test method is incapable of capturing the complex non-uniform temperature distributions observed in large-scale building components due to presence of additives (fire retardants, conduction inhibitors, and adhesives). In this study, we use a recently developed experimental method based on an existing heat flow meter apparatus (HFMA) to measure the enthalpy changes during a phase-change event – Kośny et al. (2006), Tleoubaev and Brzezinski (2007), Kośny et al. (2007), and (2009). Conventionally, the HFMA is used to determine the steady-state heat transmission characteristics of building materials – ASTM (2006). We adopt a time-integrated dynamic approach on the HFMA to allow for measurement of enthalpy changes as a function of temperature. Using this dynamic HFMA (DHFMA) method, we generate dynamic The conventional HFMA method is based on the specification described in the ASTM C518 Standard Test Method for Steady-State Thermal Transmission Properties by Means of the Heat Flow Meter Apparatus - ASTM (2006). The DHFMA method is an upgrade to the previously developed rapid temperature ramp methodology based on HFMA that was developed to test PCM-enhanced fiber insulations – Kośny et al. (2006), and allowed for more accurate test data analysis. The upgrades improved the accuracy of the results with minimal modification to the existing equipment and required no costly hardware upgrades. In principle, a phase transformation event in a PCM-enhanced material system can be studied using the heat balance equation for the material system with the consideration of temperature-dependent specific heat – Kossecka & Kośny (2010). The onedimensional heat transport equation for such a case is: 261 ∂  ∂T ∂ ( ρ h ) = λ  ∂ x ∂ x ∂t are set to different temperatures to impose a temperature gradient on the specimen. Temperature, and top and bottom plate heat flow rates, QT and QB, are recorded at time interval, τ, by thermocouples and heat flow transducers, respectively, for each plate. Each temperature step is allowed to continue until the thermal equilibrium condition is reached. Considering a constant pressure and enthalpy, H (in terms of heat per unit square of surface area) is determined by integrating heat flow rates over time: (1) where ρ, λ, T, and h are the material density, thermal conductivity, temperature, and enthalpy per unit mass, respectively. Considering a constant pressure during the thermal event, the effective heat capacity, ceff, is defined as the derivative of the enthalpy (including latent and sensible heats) with respect to the temperature: ceff = ∂h ∂T [ where QTfinal and QBfinal are the residual heat flow signals from the upper and lower plates, respectively. At equilibrium the heat flow from the upper and lower plates are subtracted from the signal of interest to eliminate drift caused by small edge heat losses. ST and SB are the calibration factor for top and bottom plates, respectively. Effective volumetric heat capacity of the specimen, Ceff (in terms of heat per unit volume per unit temperature change), is defined as the derivative of the enthalpy with respect to the temperature. In other words, effective volumetric heat capacity of the PCM-enhanced building component can be determined by taking the slope of the enthalpytemperature curve as follows: (3) where α denotes the percentage of PCM, cgcarr is the specific heat of the gypsum carrier without PCM, and ceffPCM is the effective heat capacity of PCM. 𝐶𝑒𝑓𝑓 = 𝐿 𝑑𝑇 1 ≈ (𝐻𝑖+1 − 𝐻𝑖 )(𝑇𝑖+1 − 𝑇𝑖 ) 𝐿 (6) First, a small quantity of microencapsulated PCM in powder form was tested using DSC method. Initially, a relatively fast temperature change speed of 5°C/min was used in order to estimate temperature range of the phase change process. This “fast exploring” pre-test can be easily performed with a use of the HFMA as well. The relationship between the measured melting and solidification temperatures for different DSC heating rates is well-known – Günther et al. (2009). In general, it can be assumed that lower the heating/cooling rate during the DSC testing, the narrower the measured temperature range of the phase transition. Since DSC was used only as a pretesting tool, the temperature change rate was kept significantly faster than the natural processes taking place in most typical building envelope applications. Figure 1 shows heat capacity as a function of temperature during the melting and freezing (solidification) cycles for the microencapsulated paraffinic PCM typically used in European gypsum board products. Onset of melting and solidification occur (4) where cl represents the temperature independent specific heat in the liquid state. 2 DYNAMIC HFMA TESTING As mentioned previously, a conventional heat flow meter apparatus (HFMA) is used to measure steadystate thermal properties following ASTM C518 standard. In general, a HFMA consists of two isothermal plate assemblies with one or more heat flux transducers bonded to each plate. The plate temperatures are controlled using thermoelectric elements and water chillers. In the DHFMA, top and bottom plates are set to the same temperature, unlike the conventional HFMA where top and bottom plates 1 𝑑𝐻 where L is the thickness of the specimen. In the liquid state, the effective heat capacity of PCM does not show temperature dependence. The effective heat capacity of PCM can be represented as the sum of two terms: ceffPCM (T ) = cl + ( ceffPCM (T ) − cl ) , ]] (5) (2) For most PCMs, enthalpy profile with temperature is dependent on the direction of the phase change process, and the enthalpy profile during the melting process is often different than the solidification process. Therefore, the consideration of separate temperature-dependent specific heat functions for melting and solidification is important in the thermal design of the PCM-enhanced material. Effective heat capacity for a material which is a blend of gypsum carrier and PCM may be expressed as: 𝑐𝑒𝑓𝑓 = (1 − 𝛼)𝑐𝑔𝑐𝑎𝑟𝑟 + 𝛼 𝑐𝑒𝑓𝑓𝑃𝐶𝑀 [ H = ∑ H i + (QTi − QT final ) ST + (QBi − QB final ) S B τ 262 at 19 and 24 ºC, respectively, while peaks of melting and solidification cycles are observed at 23 and 22ºC. Relatively sharp end-point of the melting process (25 ºC) and onset of the freezing process (24 ºC) are defining a sub-cooling effect of 1ºC. Total phase change enthalpy or latent heat within temperature interval of 19−25ºC is determined to be 115 kJ/kg. However, a large number of PCMs have significant sub-cooling effects. Figure 2 shows, the DSC output for one of the bio-based PCMs used by the authors with a sub-cooling effect of about 5ºC. This material is made of agriculture/food industry waste, and is used because the material has higher enthalpy and lower flammability properties than paraffins. An additional advantage of this material is the significantly lower cost compared to the petroleum-based paraffinic products. research team sponsored by the U.S. Department of Energy initiated work on thermal insulations blended with microencapsulated PCMs - Kośny (2006/07). Kośny et al. (2007). These PCM– insulation mixtures function as lightweight thermal mass components reducing energy use in buildings and controlling peak-hour loads. Other anticipated advantages of the PCM-based products include improvement of occupant comfort, compatibility with traditional wood and steel framing technologies, and potential for application in retrofit projects. In this work, the earlier-developed rapid temperature ramp methodology, which utilized the Heat Flow Meter Apparatus (HFMA), for the testing of the PCMenhanced fiber insulations was further upgraded – Kośny et al. (2012), Shukla et al. (2012). For energy modeling purposes the testing equipment was modified to allow more accurate test data analysis and development of more detailed enthalpy curves. The main target of all these refinements was to allow for dynamic testing of PCM-enhanced products with minimal modification of existing equipment and without a need for costly hardware upgrades. PCM functional temperature range 50 Freezing temperature 40 30 Sub-cooling Freezing temperature range 20 0 2000000 -10 Melting temperature range -30 -40 Melting -50 16 18 1400000 Melting temperature Freezing 20 22 Freezing 1600000 24 26 Temperature[oC] Figure 1. Temperature-dependent heat capacity of microencapsulated paraffinic PCM measured using DSC method. Melting and freezing cycles show sub-cooling of `1°C. 1200000 1000000 800000 600000 400000 200000 0 10000.00 10 8000.00 DSC uW - 0.3C/min Bio-PCM - 29degC 6000.00 12 14 16 18 20 22 24 Temperature (°C) 26 28 30 32 34 Figure 3. Volumetric heat capacity as a function of temperature for cellulose-PCM insulation sample using bio-based microencapsulated PCM. 4000.00 DSC uW output Melting 1800000 -20 Heat Capacity (J/m3-K) ∆h [J/g] 10 2000.00 0.00 Figure 3 presents the volumetric heat capacity (J m−3 K−1) measured in the HFMA for PCMenhanced cellulose insulation of nominal PCM content of 25% by weight. Microencapsulated bio-based PCM was used in this experiment. This PCM product contains fire retardants, which is why enthalpy profile in this case is expected to be slightly different than DSC data presented in Figure 2. Usually an addition of PCM additives is primarily affecting solidification process. The measurements were performed as a function of temperature for an insulation specimen of 24.2 mm thick. -2000.00 -4000.00 -6000.00 -8000.00 37.0 35.0 33.0 31.0 29.0 27.0 25.0 23.0 21.0 19.0 17.0 -10000.00 Temperature [oC] Figure 2. DSC output for bio-based PCM. Melting and freezing cycles show sub-cooling of `5°C. Testing of PCM-enhanced materials is complex, since the transient characteristics of PCM-enhanced products depend additionally on the PCM content and quality of the PCM carrier. As stated earlier, DSC test method cannot be used for this purpose, because most of these materials are non-uniform. A We find that for PCM used in this experiment, volumetric heat capacity profiles are not similar for melting and solidification cycles. In fact, we observe 263 notable sub-cooling and hysteresis for the specimen. Onsets of melting and freezing were found to occur at ~19 and 24°C, respectively, and the phase transformation spreads over a broad temperature range of ~13°C. In addition, volumetric heat capacities of the PCM-enhanced cellulose insulation when PCM is in solid and liquid states were measured to be ~1.16 and 1.54 MJ m−3 K−1, respectively. The volumetric heat capacity tends to increase with temperature when PCM is in the solid phase, while heat capacity remains approximately constant when PCM is in liquid phase. By analyzing the step change in the enthalpy curve, the total volumetric heat capacity is found to be approximately 4.65 MJ m−3 for temperature range between 19 and 30°C. It is also interesting to note that in case of PCM-enhanced insulation a significant (5°C) sub-cooling was observed. In search for a whole building energy software with a built-in capability to model PCM sub-cooling effects with two different specific heat profiles (during melting and solidification), ESP-r was chosen. ESP-r is an advanced whole-building energy modeling software, extensively used by researchers to model multi-zone thermal, air, HVAC and other building-domain related phenomena. The software allows a detailed parametric study of the factors which influence the energy and environmental performance of buildings. ESP-r uses a finite volume discretization schemes to solve the corresponding conservation equations for mass, momentum, energy, etc. Within ESP-r, PCMs are modeled using the concept of special materials (Kelly 1998). Special materials were introduced to ESP-r as means of modeling active building elements that have the ability to change their thermo-physical properties in response to some external influence. The special material functions of ESP-r may be applied to a particular node within a multi-layer construction. Any node defined as a special material is then subjected to a time variation in its basic thermo-physical properties.s 4 WHOLE BUILDING ENERGY MODELING 4.1 Background As presented in Figure 3, PCM-enhanced building materials may have complex enthalpy curves with significant sub-cooling effects. This fact forms a requirement for an application of advanced transient numerical models taking into account separate enthalpy curves for melting and freezing. An important factor to predict the behavior of PCMs in numerical models is the calculation of corresponding thermal properties at each time step (specific heat Cp, or enthalpy H). Since the dependency of specific heat on temperature is highly non-linear during phase changes, it becomes critical for the numerical model to take into account accurate temperature-dependent specific heat. In addition to temperature dependency, specific heat is also dependent on whether the PCM is melting or solidifying (crystallizing). If the PCM temperature rises and exceeds its melting point, the specific heat profile would be different from the case when the temperature drops below the melting point. This temperature difference in between the two heat capacity profiles is called the sub-cooling effect (see Figure 1). In other words, for PCMs to solidify, their temperature needs to drop below the melting point temperature for the crystallization to start. In some PCMs the sub-cooling temperature range is relatively small, and is reasonable to define temperature-dependent specific heat with a single curve for simple systems (i.e. PCM-gypsum board containing microencapsulated paraffinic PCM). However, in the case of more complex PCM applications (i.e. inorganic PCMs, bio-based PCMs, PCM blends with insulations, or arrays of PCM containers), ignoring the sub-cooling may lead to major errors in predicting PCM thermal behavior. Therefore, in these applications, there is a need to have two separate curves to define temperature-dependent specific heat. 4.2 ESP-r Whole Building Energy Performance Analysis of PCM-Enhanced Building Enclosures A series of whole-building energy simulations was performed to analyze the energy performance of different configurations of PCM-enhanced attic insulation. The ESP-r computer model of a single story ranch house was used in this study. The numerical PCM algorithm was previously validated against the whole-scale field test results for frame wall systems containing PCM-enhanced cavity insulation (Fallahi et al. 2012). Computer code allowed multiple materials and time- and temperature-dependant thermal conductivities, densities, and specific heat to be specified for parametric analysis. ESP-r assumes one-dimensional heat transfer across the PCM layer(s). As reported by Heim and Clarke (2004) the differential equation of transient heat conduction with variable thermophysical properties is: 𝜕 𝜕𝑡 𝜌(𝑇) ℎ(𝑇) = ∇ ∙ [𝑘(𝑇)∇T(r → , t)] + q(r → , t) (7) where T is the temperature, ρ the density, h the enthalpy, k the conductivity and q the heat generation rate. ESP-r uses special material files SPMCMP53 through SPMCMP56 to simulate PCM’s thermal behavior, ranging in resolution. The SPMCMP56 model developed by Geissler, A. (2008) based on the Hoffmann, S. (2006) numerical model, is an enhanced model capable of taking into account the sub-cooling 264 effect. In addition to density, conductivity and phase change temperatures this model also uses the temperature-dependent specific heat of PCMs both during melting and solidification to describe in a mathematical way the material properties within the phase change temperature limits. In this method the stored/released latent heat, LH(T), during phase change is calculated from: 𝑇 𝑎+𝑐𝑇+𝑑𝑇 2 𝐿𝐻(𝑇) = ∫𝑇 2 𝑇𝑓 ( 1 1+𝑏+𝑒𝑇 2 ) 𝑑𝑡 tional attic located in Phoenix, AZ. The geometry is shown in Figure 4. The house was modeled as two separate zones: one zone for the living areas and one for the attic. The living areas are conditioned during cooling season with a set point of 22°C. Modeling material configurations on Figure 4 are as follows: a) Living space room air (bottom); 10-mm gypsum board; 0.30-m” PCM-enhanced cellulose; attic air (top) b) Living space room air (bottom); 10-mmj gypsum board; 0.22-m PCM-enhanced cellulose; 0.08-m cellulose insulation; attic air (top) c) Living space room air (bottom); 10-mm gypsum board; 0.08-m cellulose insulation; 0.22-m PCMenhanced cellulose; attic air (top) d) Living space room air (bottom); 10-mm gypsum board; 0.08-m cellulose insulation; 0.16-m PCMenhanced cellulose; 0.08-m cellulose insulation; attic air (top) (8) where a, b, c, d, e and f are curve-fitting parameters approximating specific heat capacity of the PCM as a function of temperature during phase change. T1 is the onset melting/solidification temperature and T2 is the temperature where melting/solidification ends. Outside those two limits (T1 & T2), the PCM stores/releases energy only in the form of sensible heat. Within the limits, the heat capacity of the PCM is a function of temperature. 4.3 ESP-r Energy Performance Analysis of PCMEnhanced Attic Insulation Since residential attics are subjected to greater temperature extremes than any other component of the building envelope, increasing the thermal capacitance of the attic can reduce diurnal temperature swings, and in turn, reduce both the total energy use and peak demand characteristics in moderate and coolingdominant climates. Figure 5. Ranch house geometry modeled in ESP-r program. a b c d Figure 4. Ceiling assemblies with PCM-enhanced cellulose. As a case study to demonstrate a practical application of the PCM model SPMCMP56, a residential ranch-style house in a hot climate with PCMenhanced cellulose ceilings was modeled. The energy savings were compared with the conventional cellulose insulation ceiling. As depicted in Figure 4, parametric analysis of four configurations of the attic floor insulation blended with 25% by weight of microencapsulated PCM was performed. Nominal enthaphy of the microencapsulated PCM was about 110 kJ/kg. The analyzed house is a lightweight residential building of approximately 143 m2 floor area and conven- Figure 6. Surface Temperature of underside of the ceiling for different PCM configurations. Mid PCM refers to configuration d), Top PCM to configuration b), Btm PCM to configuration c). Detailed modeled energy parameters of the house were accessible through the ESP-r central interface. 265 Figure 6 illustrates the temperature profile of the ceiling surface exposed to the living areas as simulated for the IWEC June climatic conditions. The No PCM configuration has the highest surface temperature fluctuations, while the “Top PCM” configuration was the lowest. The ranch house was simulated with the Phoenix, AZ climate over a course of one year. As the simulation result shows, PCM-enhanced cellulose yields whole-building cooling load energy savings from 3.6% to 5.7% depending on PCM configuration (Figure 7). Twelve inch PCM-enhanced cellulose insulation yields the highest saving of 5.6%. Placing 9” PCM on top of the attic insulation yields 5% savings, while on the bottom 4.7% savings. This is due to the fact that when PCM is on top of the attic insulation, it is exposed to higher temperature fluctuations of the attic as opposed to the constant room temperature. ent heat flux data for fiber insulation material containing microencapsulated PCM. The routinely used DSC method for dynamic thermal property measurement of a PCM is valid only for small quantities of pure PCM and is not appropriate for large-scale PCM-enhanced building components, where PCM-carrier blend can be nonhomogenous. In this work, we employed a novel method based on HFMA to measure dynamic thermal properties of a 24-mm. thick sample of PCMenhanced cellulose insulation containing nominal 25% by weight of a microencapsulated bio-based PCM with latent heat of ~110 kJ/kg. Volumetric heat capacity profile was determined by taking the slope of the enthalpy curve. Notable 5oC sub-cooling and hysteresis was observed for the tested specimen of the PCM-enhanced insulation. Using the DHFMA method it was found that the tested dynamic insulation containing a nominal about 25% by weight of a microencapsulated PCM had a total volumetric heat capacity of 4.65 MJ m−3 K−1 for temperature range between 19 and 30°C. If we assume 12% to 15% of the whole building cooling load is due to heat gains through the ceiling then savings from approximately 38.0% to 47.5% in ceiling-generated cooling load are achieved. Note that the above improved energy performance is due to reducing peak-hour cooling loads. Energy cost savings can be also obtained by PCM-enhanced cellulose insulation due to shifting peak-demand time. During the last several decades, simple PCM applications like PCM-gypsum boards have dominated the thermal storage market for building envelope applications. Today the focus has slowly begun to shift to more complex PCM applications (i.e. PCM blends with insulations, PCM containers, etc.). Therefore, it is critical to develop new energy models that can describe the true behavior of these PCMs used in these applications. In this work, we propose to use separate enthalpy curves to investigate the behavior of one such PCM sample. In this PCM study, we used earlier validated ESP-r PCM model SPMCMP56, for the energy modeling and PCM performance analysis. Simulation result showed PCMenhanced cellulose yields a whole-building cooling load energy saving from 3.6% to 5.7% depending on the PCM configuration for the Phoenix, AZ climate. This savings corresponds to approximately a 38.0% to 47.5% reduction in the attic-generated cooling loads. Figure 7. Modeled annual whole building cooling load energy of the rancher house with different PCM-enhanced ceiling configurations. A base case rancher house with no PCM ceiling is used for comparisons. 6 REFERENCES. American Society of Testing and Materials 2006. “Standard Test Method for Steady-State Thermal Transmission Properties by Means of the Heat Flow Meter Apparatus,” ASTM C518. Fallahi A., Shukla N., and Kosny J. 2012. “Numerical Thermal Performance Analysis of PCMs Integrated with Residential Attics”, SimBuild2012 Conference, Madison, WI. Gelissier A. 2008. SPMCMP56 subroutine in ESP-r Source Standard Code. http://espr.trac.cvsdude.com/espr/browser/branches/Joe_Clarke/src/esrubld/spmatl.F?rev=4 529 Günter E., Hiebler S., Mehling H., and Redlich R. - Enthalpy of Phase Change Materials as a Function of Temperature: Required Accuracy and Suitable Measurement Methods - 5 CONCLUSIONS In order to best optimize the building energy performance, it is critical to accurately characterize the dynamic thermal performance of the PCM-enhanced component. A new dynamic testing procedure utilizing symmetrical step changes of temperature, and whole building energy simulations using ESP-r model were utilized in this paper. A conventional heat-flow meter apparatus was used to obtain transi266 International Journal of Thermophysics, Volume 30, Number 4 / August, 2009 Heim D., and Clarke J. A. 2004. “Numerical modeling and thermal simulation of PCM gypsum composites with ESPr”, Energy and Buildings 36 (2004) pp 795–805. Hoffmann S. 2006. “Numerische und experimentelle Untersuchung von Phasenübergangsmaterialien zur Reduktion hoher sommerlicher Raumtemperaturen”. PhD Thesis. Bauhaus-Universität Weimar Kelly N.J. 1998. “Towards a design environment for building integrated energy systems: the integration of electrical power flow modeling with building simulation”, PhD Thesis. University Kośny J., Yarbrough D., and Wilkes K. 2006. “PCM-Enhanced Cellulose Insulation: Thermal Mass in Light-Weight Fibers,” presented at International Energy Agency and Department of Energy Ecostock 2006 Conference, May 31, 2006. Kośny J. 2006/07 Field Testing of Cellulose Fiber Insulation Enhanced with Phase Change Material, Oak Ridge National Laboratory report - ORNL/TM-2007/186; September 2008. Kośny J., Yarbrough D., Petrie T. W., and Syad A. 2007. “Performance of Thermal Insulation Containing Microencapsulated Phase Change Material,” presented at 2007 International Thermal Conductivity Conference, June 24–27. Kośny J., Kossecka E., and Yarbrough D. 2009. “Use of a Heat Flow Meter to Determine Active PCM Content in an Insulation” – Proceedings of the 2009 International Thermal Conductivity Conference (ITCC) and the International Thermal Expansion Symposium (ITES) August 29 - September 02. Pittsburgh, PA USA Kośny J., Kossecka E., Brzezinski B., Tleoubaev A., and Yarbrough D. 2012. “DYNAMIC THERMAL PERFORMANCE ANALYSIS OF FIBER INSULATIONS CONTAINING BIO-BASED PHASE CHANGE MATERIALS (PCMs)” - Reference: ENB3750, Energy & Buildings 2012 Kossecka E., and Kośny J. 2010 “Thermal balance of a wall with PCM-enhanced thermal insulation” - Central European Symposium on Building Physics, Cracow, Poland, 2010. Mahling H,, and Cabeza L.F. 2008. Heat and Cold Storage with PCM : An Up to Date Introduction Into Basics and Applications. Shukla N., Fallahi A., Kosny J. – “Performance Characterization of PCM Impregnated Gypsum Board for Building Applications” - 1st International Conference on Solar Heating and Cooling for Buildings and Industry (SHC 2012), Volume 30, Pages 1-1434 (2012).Page 95, Springer; 2008. Tleoubaev A., and Brzezinski A. 2007. “Thermal Diffusivity and Volumetric Specific Heat Measurements Using Heat Flow Meter Instruments for Thermal Conductivity” - presented at 2007 International Thermal Conductivity Conference, June 24–27. . 267 268 Influence of selected calculation tool on a design process: A case study M. Massetti Ramon Llull University, ARC Engineering and Architecture La Salle, Barcelona, Spain N. Morishita & T. Bednar Vienna University of Technology, Institute for Construction and Technology, Research Center for Building Physics and Sound Protection, Vienna, Austria ABSTRACT: Several tools exist to predict building energy use, but their application in current design practice has had limited penetration in the design process to influence the design evolution and decision making. Due to the complexity and singularity of each design process, the choice of the appropriate tool is not obvious and it is the critical moment which largely determines the success of energy calculations supporting building design. In this paper, a case study is analysed and some key factors are discussed to understand how far the chosen calculation tool was appropriate for the project and useful to assist the design process. The impact of calculation results on some design decisions is analysed. To analyse the case study, the project calculation is compared with a model calibrated with monitoring data from the actual building. The case shows that the calculation could inform some decisions regarding the envelope, but it was insufficient for providing information for other decisions regarding the ventilation system during the design process. calculation tools during different phases of the design process. Due to conflicting needs, it is highlighted that an acceptable trade-off in the choice of the tool is far from obvious. The doctoral thesis analyses this issue considering energy calculation in the specific context of the building design process. Attention is paid to the evolution of the process through different stages, the integration of different competences and the interaction of multiple design problems, which are not merely quantifiable. In fact, understanding the complexity of the design process is fundamental for an effective deployment of energy calculations. Relatively new studies exist on the building design process and the application in practice of design methodologies (Lawson 2006). They reveal that the design process is largely implicit for stakeholders, it can hardly be mapped and fully understood (Cross 2006). To show the application of energy calculation in this complex ambit and to support the investigations developed in the thesis, a case study was considered recreating a hypothetical design process, in which the use of specific modelling tools was analysed (Massetti 2011). In this paper, another case study is presented, which reconstructs a real design process. Here the actual use of the calculation tools in a real setting is observed. 1 INTRODUCTION Since the very beginning of the design process, fundamental decisions influence considerably the future energy use are taken which cannot be reconsidered after construction is complete. According to different experts an effective deployment of tools throughout the entire design process could substantially improve the future performance of buildings. Currently, there is a large variety of calculation methods to calculate energy and indoor environmental performance associated to building use. They range from simple calculation methods to advanced simulation tools. Nevertheless, research and professional experiences of architects and energy specialists reveal that ordinary professional activity rarely involves deep energy analysis and calculations to support the building design process despite the expected potentials of these tools (McElroy 2009, Clarke 2001, Hopfe 2005). More often than not, the application of energy simulation is restricted to exceptional cases, including fashionable iconic buildings (Kolarevic, 2003), and/or to late design stages (Hensen 2011). In the field of building energy simulation, McElroy (2009) and Clarke (2001) proposed to integrate energy simulation with a specific design methodology, but from their studies they recognized that several barriers still exist. In order to address this issue, it is fundamental to understand which factors must be considered to effectively integrate calculation tools at different phases of the design process. Few indications have been provided by previous investigations with this regards (ASHRAE 2009, ISO 13790 2008, Waltz 2000). A doctoral thesis is investigating the applicability of 2 SCOPE OF THE CASE STUDY AND METHODOLOGY OUTLINE To investigate use of energy calculation through the building design process, a case study is analysed. The specific situation of a real project is considered. In particular, the key factors in the choice of the cal269 culation method are considered to analyse how the energy calculation method adopted was appropriate for the specific project. The case of study intends to exemplify as far as possible an ordinary project as the aim is to consider the applicability of energy calculation in ordinary building design. For this reason, the case taken represents a common situation in Austria in terms of building use, size, budget, etc. Nevertheless, there were some uncommon conditions such as the fact that all consultants worked in the same office. The close cooperation of different specialists allowed the integration of different design aspects from the beginning of the project. Moreover, the project was the object of a research project, where some implemented measures were monitored. The scope of the case study is to address these questions: 1 Why is energy calculation used? 2 When is energy calculation used? Was it used in different phases? In which phases of the design process were calculations made? 3 Which kind of calculation method (and tool) was used? Why was the specific tool chosen? To address these questions, the design process was reconstructed a posteriori, identifying the energy calculation tools used. Then, the key factors for the choice of the tool in this specific design situation were discussed in detail. To reconstruct the design process, the design team was interviewed, the building site (Fig. 1) and the project documents (Figs 2 to 5) were analysed including the models set up by the design team at the design phase. Then, to discuss the key factors and understand how appropriate the tool was, the model used at the design phase was compared to a tailored model calibrated with monitoring data. Figure 2. Urban regulation plan (Design phase, 2007). Figure 3. Plan of the building site with urban alignments (Design phase, 2007). Figure 1. The site before the construction. Aerial view from the Bing website (av. 18 09 2012: www.bing.com). Figure 4. Plan of the 5th floor (Design phase, 2007). 270 Figure 6. Chronological reconstruction of the project from the design phase to the operation phase with main tools used indicated. Figure 5. Main sections (Design phase, 2007). 3 OVERVIEW OF THE PROJECT LIFE CYCLE The investigation focuses on the design phase starting at the beginning of the design process and ending with the submission of the project to wohnfonds wien in December 2007. In particular it is considered how energy calculation tools are used in this phase. The project life cycle was reconstructed. The identification of different phases is based on interviews and dates reported on delivered documents and models. During the design phase, the design team developed a proposal to submit to the public authorities of Vienna in order to qualify for the wohnfonds_wien funding for residential projects (av. 20 12 2012: www.wohnfonds.wien.at). The project was submitted in 2007 and the grant was received in 2008. As the designer affirmed, the land was granted with specific conditions such as limited construction costs, affordable rents, inclusion of social housing in the apartment building, and integration of several innovative technologies. These technologies were proposed by the design team within the design solution and included the use of an heat pump for heat recovery in the ventilation system, and monitoring of domestic hot water, cold water, electricity, and heat consumption (from district heating). The building was erected between the autumn of 2008 and spring of 2010. Tenants began occupancy in late spring, and the building is now fully occupied. Before the building came into operation, several aspects were assessed by the design team including cost, structure and building physics. Regarding building physics, the heating need, thermal comfort (summer overheating), and acoustical comfort were calculated. These different aspects were quantified separately with different tools. The tools OIB-hwb02h and ArchiPHYSIK were used at different phases to calculate heating demand: initial calculations were made at the design phase with the OIB-hwb02h tool and energy certification requirement were verified at a final stage with ArchiPHYSIK. Two in-house Excel tools for acoustic insulation and summer overheating were used to fulfil building performance requirements. The tool AnTherm was used to consider thermal bridges to fulfil performance requirements and provide input to ArchiPHYSIK (Fig. 6). 4 DESIGN PHASE According to the interview, different design team specialists started working closely together from the beginning of the project: architect, cost assessor, structural engineer, and later the energy, i.e. building physics assessor. Then also the building physics assessor intervened at the design phase making energy calculations. The OIB-hwb02h tool was used with the scope of verifying the heating demand goal that was previously agreed with the client. 4.1 OIB-hwb02h tool The calculation method of annual space heating demand is implemented in Excel. The OIB-hwb02h method was established by Austrian regulation (OIB-382-010/99 1999) for the purpose of the building energy certification. At the time of the project, the method was outdated for generating energy certificates. However, the designers chose this method with the different purpose of verifying their performance goals and also due to the tool’s simplicity and familiarity. The calculation of space heating demand is based on the energy balance of the building (1). A quasisteady-state (monthly-balance) method is used to calculate the heat balance for the heating season taking dynamic effects into account by an empirically determined gain utilization factor. Qh = (QT + QV ) − η × (Qi + Qs ) 271 [kWh/month] (1) In summary, the main project constraints included, − the architectural program − zoning regulations – building volume limit − energy performance requirements – space heating demand of max. 20 kWh/m²·a Total heat transfer by transmission (QT) and ventilation (QV) are obtained as a function of the heating degree days. Total solar heat gains (Qs) and internal heat gains (Qi) are obtained as a function of the accumulated solar radiation and internal heat gains over the given period applying the gain utilization factor (η). The gain utilization factor takes in to account the thermal inertia of the building and the relation between gains and losses considered in the balance equation. The calculation is carried out for the building as a single zone. Inputs are introduced in different Excel sheets: General information, Technical information, Building components, Calculations, Window types, Window areas, Door areas and Transmission heat transfer coefficients. The location is selected in the General Information sheet. Based on the selected location, the tool identifies the corresponding climate data from a library. In the Technical information, building use, weight of building construction, ventilation and thermal bridges are characterized for the whole building. In the Components sheet, opaque envelope is described specifying each type of construction. The modeller has the option to introduce the Uvalues of the construction or specify the properties of each layer. In the Calculations page, the geometrical dimensions of the building are specified for each floor. Based on that, the tool calculates areas and volumes. In the Window types sheet, each type of window is characterized by U and g-value. 4.3 Object of design decision Different design decisions affecting the energy performance of buildings were addressed through the design phase. Some of these decisions involved the use of energy calculations. According to the interview, they emerged at different points during the design process: − First, the building shape and window-wall ratio were considered. − Then, envelope components characteristics were selected with assistance from the space heating demand calculation − Later, the decision between natural ventilation or mechanical with heat recovery options was attained using the space heating demand calculation. In the calculation, only the heat recovery portion was modelled of the HVAC system. 4.4 Hypotheses for energy calculation at the design phase The OIB-hwb02h was used as an energy calculation tool. The calculation method determines the space heating demand according to a simple steady-state approach (the monthly balance method) based on heating degree days and the accumulated solar radiation (OIB-382-010/99 1999). Input variables of the tool include: − Exterior environment - standard input for energy certification − User-related factors - standard input for energy certification − Building features (shape and volume; window area; component characteristics) − Building systems features (only heat recovery efficiency) When the design solution was modelled, some aspects were already decided. Therefore, within the input of the energy model: − Some design variables were fixed (building volume, building shape, and window ratio) − Some design variables were open to explore the design solution (envelope and window component characteristics, ventilation heat recovery) − Other variables could not be represented by the inputs (shading factor from the green façade; building systems except heat recovery) 4.2 Project constraints at the design phase The building program consisted of a residential building combining apartments and a children’s group home. The program also provided for bicycle and pram storage, a garbage room, an underground garage, space for common activities, rooms for technical services, storage lockers, a laundry room, and cleaning services. The design team confirmed energy performance requirements with the client: the space heating demand could not be greater than 20 kWh/m2a. According to the interview, the client sought a low energy building, but was conservative with the space heating demand as it was feared that it would be too difficult to reach the lower energy standard of passive houses (15 kWh/m²a). The project is sited in a residential neighbourhood. New residential strips surround a park on two sides, and the project site is located on the third side facing the street (Fig. 1). The public park is intended for use primarily by the tenants residing in the new residential buildings surrounding it. The form of the site, and the distance and volume limits imposed by the zoning regulations provided some constraint to the building shape and volume (Fig. 1). 272 steady-state calculation few constants were needed as inputs; it was not necessary to specify any time schedules (e.g. occupancy schedules) which fit with the need of designers to simplify the input task. In addition, the OIB-hwb02h tool represents the building as a single zone. That way the task of modelling a multifamily apartment building is substantially simplified without affecting the accuracy of results. Level of complexity of the calculation algorithm. The calculation method is based on several simplifications of real phenomena compared with advanced simulation tools (e.g. EnergyPlus). In fact, the ambit of analysis is limited to space heating. The consideration of this single energy use, the exclusion of comfort aspects and systems modeling, considerably reduce the ambit of analysis and the complexity of solving simultaneously building, systems and plants. Moreover, the use of a quasi-steady-state calculation method dramatically reduces the extension of calculation process compared with the number of calculation iterations required in more faithful modeling approaches which repeat calculations for numerous time steps. Also, the simplification of the building volume as a single zone eludes the need of modeling multiple inter-zone heat flows. The simplicity of the calculation algorithm was commensurate to the quick response time and the transparency needed by the design team. Responsiveness to design decisions. The tool was selected to calculate space heating demand. The design team used the model to support the definition of construction components. All considered construction component options characterized by different thermal properties could be represented faithfully using the model. The calculation method illustrates the impact of selecting different components on the thermal performance. Therefore, components with suitable thermal properties could be selected. As the building envelope design evolved, ventilation system requirements were hypothesized. The first variant using heat recovery could be modelled, but the variant with the heat pump could not be modelled faithfully. It was necessary to compare it to an equivalent system (see discussion about flexibility to represent different scenarios). Moreover, the OIB-hwb02h model could only quantify space heating demand which alone was insufficient to decide if heat recovery with a heat pump in the ventilation system improved the energy performance. The effectiveness of this single indicator calculated by the OIB-hwb02h tool for the evaluation of this decision has been investigated. A calibrated model based on monitoring data from 2011 was used (Table 1). With that model multiple indicators were calculated including space heating demand, along with other indicators of energy consumption and energy cost. The actual performance of the building with the heat pump, indicated by O2007F was compared with the model of the ac- 4.5 Chronological reconstruction: generation and evaluation of design solution(s) The uncommon organization of the design team allowed the integration of different design aspects from the beginning of the project. The first proposal based on urban planning regulations was outlined by the architect together with the cost assessor and the structural engineer. The shape of the building and window-wall ratio were decided evaluating different design aspects including space distribution, form, structure and cost. Therefore, the energy assessment of the first proposal was conducted to verify the fulfilment of the space heating demand requirement. The design solution was modelled with the OIB-hwb02h calculation tool. Default values for envelope components were initially used which were modified to explore different options to fulfil the space heating demand. Triple glazing was hypothesised; however as the cost were higher than increasing the insulation thickness, thicker wall insulation was chosen. As the design requirement was not still achieved, mechanical ventilation with heat recovery was introduced. The resulting space heating demand of 19.79 kWh/m²·a was obtained by supplementing a heat pump to increase the heat transfer between exhaust and supply air. The implementation of a reversible heat pump was also intended to provide the possibility of some cooling to the supply air in summer using the same system. During the design phase, a green façade was considered to achieve summer comfort parameters. The effect of the green facade could not be modelled to fulfil summer comfort requirements using the OIBhwb02h tool. 4.6 Key considerations for selecting an energy calculation method In this section, the key factors for selecting the OIBhwb02h tool during the design phase are discussed. In particular the calculation method lying behind the tool is considered, rather than its implementation as a software application. Level of discretization. It is difficult to reconstruct the amount of information available and details when the model was initially set-up. Nevertheless, the level of detail reflects the time the building physics assessor had to reassess the heating energy demand for different iterations in a reasonable time to the design team. As the building physics assessor explained during the interview, the OIB-hwb02h tool was chosen as a heating energy demand could be calculated with a relatively small number of input data. For the same reason tools with more detailed input (such as ArchiPhysik, or advanced simulation tools such as EnergyPlus) were not considered at this stage. With OIB-hwb02h, the number of inputs necessary to describe the exterior climate and userelated factors was very limited. For the quasi- 273 evolved. Introducing heat recovery into the model was simple: the model inputs already defined were unchanged and an additional input was assigned, i.e. the heat recovery efficiency. Other design modifications would not have been reproduced in such a direct way with this tool. For example, rotating the building to modify its orientation is possible but very tedious in this model. Instead of directly modifying one angle parameter, it is necessary to modify the orientation of every single envelope surface modelled. The level of detail of that model, which represents individual windows, requires 65 inputs for modifications, one for each window. Flexibility to represent different scenarios. The tool was initially flexible enough to represent all different envelope variations and the first hypothesis of heat recovery. Nevertheless, when the heat pump was introduced for the heat recovery the model was unable to represent it. To emulate the heat pump effect in the calculations, the efficiency of heat recovery was increased from 50% to 75%. The model did not represent the design idea, but an equivalent solution intended to produce similar performance. This kind of assumption often occurs in practice. In this situation, the competence and experience of the modeller plays an essential role for detecting misleading results and avoiding incorrect interpretations. In this particular case, the assumption made to emulate the heat pump was probably a hazard. In fact, a heat pump for heat recovery in a centralized ventilation system of an apartment building is uncommon in Austria (Feist 2004), and the design team had no previous experience with this solution. Therefore, the utility of the model for this design decision was questionable. Accuracy. During the project, the designer introduces heat recovery with an efficiency of 50% (see D2007V1 in Table 2). The model indicated that the space heating demand was above the limit value of 20 kWh/m2·a. Heat recovery was improved with the introduction of the heat pump (see D2007F in Table 2). With the improvement, the model indicated that the limit value was achieved. To evaluate the accuracy of the model, a comparison based on actual building performance was conducted. The actual performance of the building (O2007F) with the heat pump was compared with the model of the actual building behaviour as if it was without heat pump (O2007V1). These calibrated models shows that in both cases the space heating demand fulfils the design requirement. Therefore, the accuracy of the model was insufficient to assess the heat recovery versus a realistic value limit. Nevertheless, it is interesting to compare the relative impact of this design decision according to the original design model and according to the actual building (-23% and -18% respectively for space heating demand). The trends are coherent. Therefore, the accuracy of the design model is sufficient for understanding the indicative tual building behaviour as if it was without heat pump shown in O2007V1 (Table 1). Table 1. Performance indicators of the model variants at the design phase (D2007V1 - D2007F) compared with the corresponding variants calibrated with the monitoring data of 2011 (O2011V1 - O2011F). Indicators not assessed by the calculation during the design phase are highlighted in grey. D2007V1 Heat recovery Space Hating Demand kWh/m²a Energy Consum. District Heating kWh/m²a Electricity kWh/m²a Energy Cost District Heating €/m²a Electricity €/m²a Total €/m²a 25.58 D2007F Heat recovery with heat pump 19.79 (-23%) O2011V1 Heat recovery 16.79 n.a. n.a. 61.39 n.a. n.a. 189.61 n.a. n.a. 4.30 n.a. n.a. 32.23 n.a. n.a. 36.53 O2011F Heat recovery with heat pump 13.73 (-18%) 56.49 (-9%) 195.57 (+3%) 3.95 (-9%) 33.25 (+3%) 37.20 (+2%) According to the calibrated model (right part of Table 1), space heating demand indicates that the solution with heat pump for heat recovery is more efficient (-18%), concurring with the results of the OIBhwb02h model used during the design phase (left part of Table 1). Considering the other indicators, such as energy consumption, the advantage of the heat pump is less evident. The heat consumption from district heating is slightly reduced (-9%), with no appreciable variation of electricity consumption of the building. Furthermore, considering the total energy cost of district heating and electricity, no appreciable improvement exists. It is observed that, with more comprehensive knowledge of different indicators the advantage of the heat pump appears to be less. Therefore the information provided by the OIB-hwb02h tool on the impact of the heat pump is insufficient and not meaningful to take a decision. Feedback immediacy. Rapid feedback of energy assessment was imperative to work in coordination with other design tasks and partners, and to develop the design phase according to the time and budget available. The tool provided rapid feedback for different options assessed regarding their thermal properties. From the interview, it was deduced that the tool was appropriate for the quick feedback needed by the design team. Flexibility to modify design. At the beginning of the design process, components’ characteristics were evaluated and the two hypotheses on heat recovery from ventilation were modelled. Throughout the design process, modifications to the design solution were reproduced editing the model as the design 274 process may also provide a benefit, strengthening the modeller’s insight of the project. Transparency. The calculation method was transparent, being very simple and based on widely used magnitude, the degree days. In particular the hwb02h method had broad diffusion in Austria and the building physics assessor was familiar with it. Therefore, results could be easily interpreted and the physical meaning of the project model was easily understood by the building physics assessor and communicated to the design team. Tables 3 and 4 summarize the key factors discussed in this section. trend of this design decision. The results indicate that the accuracy required from the tool depends on the question the model should answer: (1) is the space heating demand below the limit value? Or, (2) what is the heat pump impact on space heating demand? Table 2. Space heating demand of the model variants at the design phase (D2007V1 - D2007F) compared with the corresponding variants calibrated with the monitoring data of 2011 (O2011V1 O2011F). The option exceeding space heating demand limit is shown in grey. D2007V1 Heat recovery Space Heating Demand kWh/m²a 25.58 D2007F Heat recovery with heat pump 19.79 (-23%) O2011V1 Heat recovery 16.79 O2011F Heat recovery with heat pump 13.73 (-18%) Table 3. Synthetic judgment of the tool in relation to different design decisions. decisions on Level of discretization Level of complexity of calculation algorithm Responsiveness to design decisions Feedback immediacy Flexibility to design modification Flexibility to representation of different scenarios Accuracy Data coherence through the process Transparency Integrability in multiple design problems. The design was initially developed according to cost and structural considerations. Building physics assessments were integrated including the use of the energy calculation tool OIB-hwb02h. Different project aspects were developed collaboratively, as the consultants were in the same office. Energy performance was modelled separately from other performance aspects (i.e. with a dedicated tool for space heating demand calculations). With this work organization the simplicity of the tool played an important role (c.f. level of discretization, feedback immediacy, and complexity level of the calculation algorithm). It was important to provide rapid feedback to the design partners with understandable results for discussion. The simplicity of the calculation method facilitated the integration of multiple design aspects. Data coherence through the process. Information was produced using different dedicated models for energy, acoustics, summer thermal comfort, and thermal bridges analyses. Furthermore, two different tools were used to calculate energy performance: OIB-hwb02h during the design phase and ArchiPHYSIK during the construction phase. Using different tool required translating information several times through the OIB-hwb02h model and other tools. The increasing risk of input errors and the different inputs formats of each tool favors inconsistencies and asynchronies though different representations during the process. The use of different dedicated models to represent different phenomena may result in inconsistencies especially when different aspects of the analysis are strongly coupled. Inconsistencies are reflected in the accuracy of model outputs. If the accuracy demanded to answer design questions is fulfilled, these inconsistencies are acceptable. In this case, inconsistencies through different models are not investigated. The translation of information at different points through the design good acceptable scarce unacceptable building components good good ventilation system poor poor acceptable good unacceptable acceptable questionable poor questionable acceptable unacceptable questionable good / questionable Table 4. The table indicates how many times the tool is evaluated as good / unacceptable according to key factors in Table 3. decisions on good unacceptable building components 3 0 ventilation system 0 2 5 CONCLUSIONS The specific goal established for the project was formally achieved. In fact, the limit value of space heating demand was fulfilled assuming conventional hypotheses: quasi-steady-state conditions and conventional boundary conditions for climate and use. Therefore, the calculation method was useful to achieve this goal. Regarding effective energy efficiency improvements, the calculation method adopted provided no evident support for choosing between ventilation system options. Indeed, it is shown that the choice of the calculation method was not secondary. It had implications on the knowledge produced to drive design decisions. Observing different factors, it seems that the tool provided better support during a part of 275 the design phase but not for the whole design. It was more adequate at the beginning of the design phase then during the final stages. It offered an effective support to some design decisions (construction properties) and weaker or no support to other decisions (heat recovery and building systems). When designers addressed the former decisions (construction properties), an appropriate trade-off could be found in the choice of the tool. Instead, when building systems were chosen, the calculation method offered the immediacy needed by the designers. In turn, it was not flexible and adequate to represent the particular solution and able to quantify relevant performance indicators. Clearly, more comprehensive calculations not included in the hwb02h method were needed to inform these design decisions. But often comprehensive calculation methods hamper transparency and especially feedback immediacy – vital factor for the design team to exploit the tool. It is paradoxical that the calculation supported decisions where the designer had more experience (definition of construction components) and was weaker where more answers were needed on innovative elements of the project (the uncommon ventilation system). There are two possible explanations. (1) Possibly, the calculation method was chosen with the initial purpose to support the definition of construction components. Then, design evolution led to address the option of the heat pump to improve heat recovery. Often, a particular decision emerges through the process which can be quite unpredictable (Lawson 2006). It is quite possible that this option was not expected to be considered when the calculation method was chosen. (2) Design is constrained by time and budget limitations. HVAC modelling tends to be more complex and demanding than building envelope modelling. Therefore, designers did not assume that the task required complex and demanding calculation methods. The case study shows the difficulty to fully support different questions thorough the whole design process with a single tool. Although no radical changes occurred, design evolved through the design process (i.e. decisions addressed, relevant indicators needed, information produced, etc.) Therefore, the trade-off between different key factors for the choice of the tool was instable: the OIB-hwb02h calculation method that initially offered an acceptable trade-off became inadequate during the process evolution. The case study also shows that energy modelling is not necessary to address any design decision / question but it can be exploited when experience is insufficient to provide satisfying answers. In the case study, some key issues emerge which require further investigation: − One large barrier to energy modelling was extra time and cost. It is not clear how this cost is assumed. That involves economic, political and eth- ical considerations, but it is key point not separable by technical and design issues. − The use of simple energy modelling approach in ordinary projects similar to the case study may have a limited impact on individual buildings. It is not clear if the use of simple energy modelling approach in ordinary projects similar to the case study, involves a reduction of energy consumption at the scale of the building sector. − A comparison with other case studies should be addressed in the future. − During the last years the hwb02h method for calculating heating energy demand has been outdated in Austria and replaced by a full calculation of building and system performance (www.oib.or.at). Future studies could investigate the process of design decisions if the designers have a better focus on the total energy use. 6 REFERENCES ASHRAE 2009. ASHRAE handbook: fundamentals. Atlanta. Clarke J. A. 2001. Energy simulation in building design. Oxford: Butterworth-Heinemann. Cross N. 2006. Designerly ways of knowing. London: Springer. Feist W. (ed.) 2004. Arbeitskreis kostengünstige Passivhäuser Phase III. Darmstadt: Passivhaus Institut. Hensen J. L. M., and Lambert R. 2011. (ed.), Building performance simulation for design and operation. Oxon: Spoon Press. Hopfe C. J., Struck,C., Ulukavak Harputlugil G., Hensen J., and Wilde P. 2005. Exploration of using building performance simulation tools for conceptual building design. IBPSA-NVL conference: 8 ISO 13790 2008. Energy performance of buildings - Calculation of energy use for space heating and cooling. Kolarevic B. (ed.) 2003. Architecture in the Digital Age – Design and Manufacturing. New York: Spon Press. Lawson B. 2006. How designers think: the design process demystified. 4th ed. Oxford: Architectural Press. Massetti M., and Corgnati S. P. 2011. Energy Simulation Supporting the Building Design Process. A Case Study at the Early Design Stage. SIMUL 2011. Barcelona. McElroy L. B. 2009. Embedding integrated building performance assessment in design practice. PhD Thesis. University of Strathclyde. OIB-382-010/99 1999. Leitfaden für die Berechnung von Energiekennzahlen. Waltz J. P. 2000. Computerized building energy simulation handbook. Monticello, NY: Marcel Dekker. 276 Generalizing roof geometry from minimal user input for building performance simulation K. Hammerberg, V. Jain, N. Ghiassi, A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: One strategy for increasing the use of energy performance simulation tools by primary building designers and novice users is to simplify the input process for complex building geometry. Detailed energy simulation of a building requires, among other things, specific data on the roof geometry. However, even in the case of simple roof forms, providing such data in a simplified user interface represents a major challenge. If this is circumvented by reducing the input requirements to the crucial information – e.g. angle, ridge height, floor plan, and typology – assumptions about the resulting roof geometry would have to be made. The purpose of this paper is to address these assumptions regarding the generalization of roof geometry in building performance simulation. The authors have developed an algorithm for auto-generating a set of potential roof forms based on a minimal set of input data. As a test of how well this set of solutions represents realistic and practical roof forms, we opted for the following methodology. We selected a number of existing buildings that represent, in the Austria context, basic typologies and floor plan arrangements. The floor plans and a set of basic parameters describing the roof are then provided to the algorithm, which produces an independent set of roof forms. The buildings are then simulated with both their existing roofs and the algorithmically generated roofs using an energy analysis tool. These results are compared regarding their equivalence in view of energy performance. 1 INTRODUCTION 1.1 The Complexity Roof Geometry The quantity and variety of building performance assessment tools attests to the large number of attempts that have been made to incorporate energy efficiency into the building design process. However, despite the increased effectiveness and savings in time, cost and effort associated with using these tools early in the decision making process (Domeschek et al. 1994), their use at these early stages of design remains very limited. External energy analysts and consultants are typically only involved after the design has been finalized (Hensen et al. 2004). Furthermore, simulation tools are not used for the creation of design alternatives or as a selection criteria for different design options (De Wilde 2004). Time and cost are often cited as the main barriers to the adoption of building simulation tools, particularly the time required to prepare the input information. This is can be attributed to several factors: the complexity of the data required for simulation, cumbersome data entry methods, and the often mediocre usability of available tools. With this in mind, it is important for the developers of such tools to address these sources of complexity and attempt to alleviate the burden of data entry on the end user without sacrificing the quality of results. Typical building floors can be easily represented in 2 dimensions and therefore even complex plans can be drawn with a simple 2D user interface. However, relatively simple building plans can result in complex roof geometries that cannot be represented in 2D. As a result, requiring users to input precise roof geometry necessitates a more sophisticated 3D user interface and places a larger data input burden on the user. Therefore, it is important, in view of simplifying the information input requirements on the user, to explore other alternatives. Our approach in this paper was to develop an algorithm that automatically generates a roof geometry from a few user inputs. Automatic roof reconstruction from sparse input data has been an important aspect in the field of remote sensing and photogrammetric research for the last two decades and many sophisticated algorithmic methods have been developed (Haala and Kada 2010). Within this field, the complexity of roof geometry is recognized as “an essential problem” (Milde and Brenner 2009) and although the end goal of these pursuits is different, the approach is generally the same: “to reduce the number of operator interactions per building” (Brenner 2000). 277 the width of the building wing and extruding this shape along the ridge line. The resulting geometries from each ridge line within a group are then intersected to create a single roof form. This intersection routine also doubles as a pre-selection filter. If it fails to generate a single solid geometry, then that roof option is discarded. The individual roof surfaces are then grouped by the horizontal component of their normal vector and windows are distributed accordingly. The window's size is determined by the fraction of the parent roof surface to the total surface area in a given direction multiplied by the user specified window area for that direction. The window shape is given by a rectangle that is proportional to the general shape of its parent surface. The window placement routine acts as another pre-selection filter. In the case that there is not enough surface area in one direction to accommodate the window area specified by the user, that roof option is discarded. In the case of our dataset we observed that this procedure predominately produced roofs that were significantly larger than the actual roof. Therefore, from the valid options created, the option with the smallest volume was selected as the final roof form. 2 METHODOLOGY 2.1 Algorithmic Generation Most of the complexity in the approaches described by Haala and Kada (2010) arises from the attempt to create a general solution for arbitrary building shapes and handling large and ambiguous data sets from remote sensing sources. For our algorithm, we assume that our problem is bounded more strictly. The constraint that simplifies the problem most significantly is the manual input of the building perimeter into an orthogonal grid by the user. Therefore, our generation algorithm assumes the input polygon will be a rectilinear polygon. In other words, the interior angle of all connected line segments will be either 90 or 270 degrees. 2.1.1 Input Parameters Our algorithm requires four inputs from the user. As mentioned above, the building perimeter is a key input. For any simulation tool that uses explicit geometric data, the building perimeter is a requirement for drawing the lower floor(s). As such, it doesn't represent any additional input burden on the user. Additionally, roof pitch angle, roof type, and the area of windows by direction are also required. 2.1.2 Procedure The input polygon is first decomposed into rectangles by using all combinations of line segment end point pairs and discarding invalid rectangles (i.e. no width or height) and rectangles which overlap the boundary polygon. Then all possible unique groupings of these rectangles are found that fill the polygon without internal overlapping. Each rectangle then represents two possible ridge line directions for each portion or wing of the building (the two lines of reflectional symmetry). Using combinatorics again, all unique combinations of ridge lines are found for each group of rectangles. This results in 2n combinations, where n is the number of rectangles. Each of these combinations results in a unique roof form. At the intersection of wings, these ridge lines are then selectively extended or joined with a fillet depending on the widths of the two wings. The ridge line of the smaller of the two wings is extended to meet the other. In the case that the two wings are of equal width, the ridge lines are joined with a fillet. These ridge lines are then used as the spine along which a roof geometry is constructed depending on the roof type. In the case of a gable roof, this is as simple as generating the profile shape from the user specified roof pitch angle and 2.2 Case Studies In order to test the equivalence of the algorithmically generated roofs against actual buildings, we selected 15 existing buildings that are representative of the typical residential housing stock in the rural areas of Austria. The geometric information and the input parameters mentioned above were extracted from the building plans provided in Prader and Fehringer (1987) and Hoffmann et al. (1987). These buildings were chosen for their orthogonal floor plans and typical roof forms. Therefore, they provide a good first test for the applicability of this approach. In other words, large discrepancies in the results within this simple set of buildings would reveal fundamental issues. For each of these buildings, a model was created with the original roof form as described in the building plans and another model created with the algorithmically derived roof form. Each building was given a label from A to O in the following figures. See Figure 1 for a comparison of the volumes of the original roofs to those that are automatically generated. Outside of a few outliers, the actual and predicted volumes display a high degree of equivalence. The mean error is ±4.0%. The surface area shows this same close approximation. However within the context of simplifying a 278 partitions explicitly, we included their contribution to the internal mass for more realistic thermal behavior modeling. Additionally, we made the assumption that all roof zones were occupied and conditioned spaces. Although this was not true of every building in our dataset, occupied attic spaces are a relatively common feature of single family homes in Austria. Furthermore, the eaves and roof overhangs are not included in the original or generated models. Therefore, the shading provided by these features are not taken into account. building performance tool, the impact of these geometric deviations on energy use is the primary concern. 625 B 550 Generated volume [m3] 475 400 O 325 250 N 175 100 100 Table 1. Input parameters. Variable Units Site Location Weather file Occupancy and internal loads Use Occupancy load persons EPD W·m-2 LPD W·m-2 Infiltration ACH Construction External Wall U-Value W·m-2·K-1 Roof W·m-2·K-1 W·m-2·K-1 Ground Floor Door W·m-2·K-1 Window W·m-2·K-1 HVAC System Type Heating SP °C Cooling SP °C A 175 250 325 400 475 550 625 Original volume [m3] Figure 1. Roof volume comparison – dashed line represents the ideal. 2.3 Simulation Protocol We chose Energy Plus (USDOE 2012) as the simulation environment for calculating the impact of algorithmically generated roofs on the building's energy performance. For the case of the generated roof, the algorithm was adapted to export the roof geometry directly in IDF format and we replaced the existing roof zone with this output. All other properties of the buildings were kept constant. Input Vienna, Austria Vienna Schwechat IWEC Residential 5 6 10 0.5 1.03 0.69 0.83 2.62 2.6 IdealLoadAirSystem 21 26 3 RESULTS 2.3.1 Inputs and Simplifications To aid in comparison of the results, we assumed all buildings had the same constructions. The overall U-values for these constructions were based on the default values for buildings of their age category from the OIB (2011). The exact layers of the constructions were determined by expert opinion on the Austrian housing stock and material properties from the database of IBO (2012). These values and the other relevant input parameters are shown in Table 1. Our primary simplification was to model each building with only two zones: one zone for the roof space and one for the lower portion of the building regardless of the number of floors or internal rooms. Instead of modeling these internal In the Austrian climate heating plays the largest role in overall building energy use. The simulation results for annual heating demand by floor area are shown in Figure 2. Note that the roofs that deviate from the ideal correspond to the deviations in volume in Figure 1. In other words, generated plans that underestimate the roof volume also underestimate heating energy. This is expected as it requires less energy to heat a smaller space. This relationship is also visualized in terms of percent error in Figure 3 below. The greater the deviation in volume, the greater the deviation in predicted heating energy. The mean error for the heating energy demand is ±5.6%. 279 strong during the summer months as shown in Figures 6 & 8. 600 90 A B 80 Solar gain by generated roof window area Heating demand (GG) [kWh·m-2·a-1] 100 O 70 60 N 50 40 40 50 60 70 80 90 100 Heating demand (OG) [kWh·m-2·a-1] A 525 D E H N 450 375 300 225 150 150 225 300 375 450 525 Solar gain by original roof window area [kWh·m-2·a-1] Figure 2. Comparison of original geometry (OG) and generated geometry (GG) heating demand per floor area per year. Figure 4. Comparison of solar gain in roof zone. Due to this increased heat gain, these generated roofs had correspondingly higher cooling energy demand (Fig. 5) and a mean error of ±14.5% for the predicted cooling energy demand. In some cases, this increased solar heat gain in the winter months translated into a lower heating energy demand (Fig. 6). N 25 O 15 R² = 0.96 14 A 10 Cooling load from generated geometry Heating demand error [%] 20 5 E 0 0 M 5 10 15 20 Volume generation error [%] Figure 3. The relationship between volume generation and heating demand errors. Although the cooling energy represents a smaller component of the total energy use in the climate of our study, our results regarding cooling and solar heat gain are worth discussing for the greater impact they would have if this approach were used in another climate or location. As mentioned above, the algorithm distributes window area across surfaces grouped by the horizontal component of their normal vector. This means the inclination of the windows are not taken into account by the roof generation algorithm. In our data set, this had the effect of creating more inclined skylights in the generated roofs. As a result, the majority of the generated roofs with windows had significantly higher solar gains. This can be seen in Figure 4, which compares solar heat gains (per unit window area) for original and generated models. This effect was particularly D A 12 N H E 10 8 6 4 4 6 8 10 12 14 -2 -1 Cooling load from original geometry [kWh·m ·a ] Figure 5. Comparison of cooling energy demand. Figures 6-9 show monthly values for four plans that illustrate how variations in both window placement and volume effect energy demand throughout the year. Case study A has a high percent error in both solar gain and volume. Case O varies significantly in volume, but has no windows and thus no difference in solar gain. Plan E has a very small error in volume, but a large amount of deviation in solar gain. Case study M has the same roof shape and no solar gain, in this case the values 280 for energy demand in the generated and original building are identical. vertically offset by knee walls (Fig. 11). Collecting user input for the ridge line direction and knee wall height should correct for a large part of this variation. Some of the deviation in energy demand is also caused by the distribution of windows on inclined roof surfaces without accounting for their original tilt angle. This caused a significant overestimation of the solar heat gain (Fig. 4). This would have an even larger impact in cooling dominated climates with 4 DISCUSSION The most significant cause of variation in energy demand in our study is the variations in roof form that result in volume differences. The generated roofs with the largest deviation in volume have ridge lines that run opposite to the ridge lines of the original roof forms (Fig. 10) or roofs that are 281 3000 2500 kWh 2000 1500 1000 500 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Original:Transmitted Solar Energy Generated:Transmitted Solar Energy Original:Heating Energy Generated:Heating Energy Original:Cooling Energy Generated:Cooling Energy Dec Figure 6. Case A – Monthly solar energy, heat demand, and cooling demand (roof zone). Inset: Original (left) and generated (right) geometries. 3000 2500 kWh 2000 1500 1000 500 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Original:Transmitted Solar Energy Generated:Transmitted Solar Energy Original:Heating Energy Generated:Heating Energy Original:Cooling Energy Generated:Cooling Energy Dec Figure 7. Case O – Monthly solar energy, heat demand, and cooling demand (roof zone). Inset: Original (left) and generated (right) geometries. 282 3000 2500 kWh 2000 1500 1000 500 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Original:Transmitted Solar Energy Generated:Transmitted Solar Energy Original:Heating Energy Generated:Heating Energy Original:Cooling Energy Generated:Cooling Energy Dec Figure 8. Case E – Monthly solar energy, heat demand, and cooling demand (roof zone). Inset: Original (left) and generated (right) geometries. 3000 2500 kWh 2000 1500 1000 500 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Original:Transmitted Solar Energy Generated:Transmitted Solar Energy Original:Heating Energy Generated:Heating Energy Original:Cooling Energy Generated:Cooling Energy Dec Figure 9. Case M – Monthly solar energy, heat demand, and cooling demand (roof zone). Inset: Original (left) and generated (right) geometries. 283 higher annual solar insolation. Therefore, in future work we recommend that a more sophisticated input method be used for the windows in the roof floor that captures inclination as well as direction and area. Alternatively, it may be possible to reduce this geometric variation through the use of a more sophisticated selection mechanism. With a thorough analysis of the housing stock under consideration, it should be possible to determine a ratio of volume to plan area for typical roofs of a certain type that could serve as a selection criteria. Our ongoing work has shown good results in this regard. regard to inclination caused an overestimation of solar heat gain in the generated models. Despite these variations in the generated roofs, the mean error for the heating energy demand was only ±5.6%. The error for cooling demand was higher at ±14.5%. However, we believe both would be reduced substantially with the inclusion of the additional input parameters previously discussed. This suggests that algorithmically generated roofs could be a viable option for reducing the input demand on users of building performance software without a significant reduction in accuracy. 6 REFERENCES Brenner C. 2000. Towards fully automatic generation of city models. International Archives of Photogrammetry and Remote Sensing 33(B3/1; Part 3): 84-92. De Wilde P.J.C.J., and Van der Voorden M. 2004. Computational support for the selection of energy saving building components. Delft: Delft University Press. Domeshek E.A., Kolodner J.L., and Zimring C.M. 1994. The design of a tool kit for case-based design aids. In J.S. Gero & F. Sudweeks (eds), Artificial Intelligence in Design vol. 94: 109-126. Netherlands: Kluwer Academic Publishers. Haala N., and Kada M. 2010. An update on automatic 3D building reconstruction. ISPRS Journal of Photogrammetry and Remote Sensing 65: 570–580. Hensen J., Djunaedy E., Radošević M., and Yahiaoui A. 2004. Building performance simulation for better design: some issues and solutions. In Wit, M.H. de (ed), Proceedings of the 21st PLEA International conference on Passive and low energy architecture: 1185-1190 (vol. 2). Eindhoven: Technische Universiteit Eindhoven. Hoffmann H., Tritthart M., and Kargl H. 1987. Architekten Planen für die Landwirtschaft. Graz: Ingenieurkammer für Steiermark und Kärnten. IBO 2012. Österreichisches Institut für Baubiologie und ökologie. baubook: Ökologische Bauprodukte. Retrieved from http://www.baubook.at/ Milde J., and Brenner C. 2009. Graph-based modeling of building roofs. In: Proceedings of the 12th AGILE Conference on GIScience. Hannover. OIB 2011. Österreichisches Institut für Bautechnik. Leitfaden Energietechnisches Verhalten von Gebäuden (OIB330.6-111/11-010). Vienna: Author. Prader H., and Fehringer F. 1987. Möglichkeiten positiver Einflussnahme auf typische Wohn- und Siedlungsformen im ländlichen Raum. Vienna: Author. USDOE 2012. Department of Energy: Energy Efficiency & Renewable Energy. EnergyPlus Energy Simulation Software. Retrieved from : http://apps1.eere.energy.gov/bu ildings/energyplus/ Figure 10. The different ridge lines of original (left) and generated (right) roofs of Case A result in significantly different volumes. Figure 11. A knee wall in the original roof (left) of Case N adds more volume compared to the generated roof (right). The roof typology within our data set should also be addressed. All of the roofs in our data set are from a ubiquitous, but narrowly prescribed building typology – the single family home in Austria. In order to more accurately capture unusual roof forms, the user may be required to input the roof profile in 2 dimensions. However, avant-garde roof shapes generally belong to buildings designed by professional architects and thus fall outside the scope of a simplified user interface. In the case of larger urban buildings, the exact form of the roof should play a smaller role in the overall building performance. 5 CONCLUSION The algorithm we developed performed well in the approximation of the geometry. However, even small variations in volume caused some deviations in the simulated energy demand. Additionally, the distribution of windows across surfaces without 284 Experimental hygrothermal study in wood and wood-based materials N. Shukla, D. Elliott, D. Kumar, C. Misiopecki and J. Kosny Fraunhofer Center for Sustainable Energy Systems, Cambridge, MA ABSTRACT: In-situ and continuous measurement of moisture in a wood assembly is critical in preventing condensation-related structural damages and failures from occurring, and also helps determine the system performance. In this study, we designed and built a practical, continuous and inexpensive electrical resistance based method to measure moisture content (MC) in wood and wood-based species in field conditions. The method is based on measurement of electrical resistance across a pair of metallic pin/screw sensor that is inserted into the wood specimen. Scientific literature contains calibration data between electrical resistance and MC determined using oven-dry method only for a handful of wood species and that too for moisture levels below fiber saturation point (FSP). In this paper, we determined calibration curves covering a wide MC range of ~0.1–1.0 for six different types of wood species. Further, we established best power fits for each of the wood species spanning the entire MC range studied in the paper. The oven dry data as a function of drying time allowed us to determine water diffusivity in all wood samples. was inexpensive and the sensors required little or no supervision. There are several methods available today to measure MC within a wood structure (ASTM D4442 2007, James 1963, Kupfer 1997). These methods can be broadly classified in two categories: 1) direct methods provide MC by directly measuring mass of the water present in the wood, while 2) indirect methods rely on the measurement of a physical property that is dependent on MC. Direct method can be further divided into thermogravimetric and analytical methods. In thermogravimetric methods, mass of moisture in a specimen is determined with application of heat i.e. drying of the sample with a variety of ways. Analytical methods are chemical methods where water is removed from the sample with the help of a chemical reaction. Indirect methods can be divided into mainly four methods: electrical, radiometric, optical and thermal methods. Among these, electrical methods are relatively inexpensive and easier to build. Compared to other methods, electrical methods are based on simpler physical relations (in this case electrical property) with MC, therefore, they tend to be more accurate and easy to implement. 1 INTRODUCTION Wood is a hygroscopic material and due to its porous structure it may absorb large quantities of water. Presence of water or moisture in a wood-based building component may compromise the structural and mechanical integrity of the building, in addition to causing health and safety issues (Trechsel 2006). Furthermore, in order to evaluate the energy performance of a wood-framed building structure, it is critical to know the level of moisture content (MC) within the wood structures of the building. Continuous measurement of MC is a challenging task especially for field projects where there are number of additional limitations such as the largescale of MC measuring device that prohibits the use in field conditions (X-ray, optical MC systems etc which are otherwise frequently used in lab setup), expensive nature of the MC device (optical, X-ray, neutron systems), complexity of the method (oven dry, infrared, thermal systems) or severe weather elements that may cause damage to the equipment and instrumentation. The main goal of this study was to develop a sensor based method that can accurately and reliably measure moisture content within wood structures in a continuous manner under field conditions. Other requirements were that the testing setup 285 and returned to the chamber at the same settings. To ensure no further moisture loss, the samples were weighed again approximately 16 hours later. Once the dry weight was established, the samples were immersed in water. The samples were left for several days and weighed incrementally. Once moisture content greater than 60−100% was reached, the samples were removed and air-drying began. The samples were stored vertically in a simple dish drying rack during this portion of the experiment. At least two measurements were taken throughout the day each weekday. Additional, yet less frequent, measurements were taken during the weekends. All three MC measurement methods below were performed on the samples at each reading. 2 SELECTION OF MOISTURE CONTENT MEASUREMENT METHOD After reviewing historical and current practices described in literature (Brischke 2008, Dill 2000, Glass 2010, Grinzato 1998, Hauschild 1998, James 1963, Maref 2009, Phillipson 2007, Said 2004, Stamm 1927, Sereda 1982, TenWolde 1987, Ye 2007), we determined that electrical resistance method will be most suitable choice for the purpose of this study since this method is inexpensive, accurate and easy to implement. At the same time, it is one of the only methods available that can be used for continuous MC monitoring in field projects. Calibration of resistance sensors involves finding a relationship between electrical resistance and moisture content as measured by a direct method. We decided to use oven dry weight direct method in this study since it is considered to be one of the most accurate and inexpensive direct methods. A handheld Delmhorst BD-2100 moisture meter was also included in this study. Although handheld meter is an easy to use device for a quick moisture content reading, it does not take into account the species of wood or the lack or presence of resin, which is integral to products such as OSB, and consequently may produce large errors in the measurement (ASTM D4444 2003, Forrer 1987, Garrahan 1988, James 1994). The electrical resistance method presents its own challenges, including the development and fabrication of an instrumentation scheme capable of measuring the range of resistance as the wood or wood-based sample dries. 3.1 Oven-dry or Weight Method Using a Pelouze 4.5 kg-capacity scale, the weight of the sample was measured and the moisture content, MC, was determined: 𝑀𝐶 = (𝑊𝑚 − 𝑊𝑑)/𝑊𝑑 (1) where Wm= the measured weight of the specimen, Wd= the dry weight of the specimen. Since oven-dry method provides the most accurate measurements out of the three methods used in the study, it was used for calibration of other methods. The additional weight of the sensors, described under “Electrical Resistance Method” was subtracted from the weight at each reading. 3.2 Handheld Moisture Meter Method Using a Delmhorst BD-2100 with 5/16” probes, ten measurements were taken and averaged over the 6” x 6” sample. Moisture content was observed to be non-uniform. Care was taken to avoid the edges, which typically had lower moisture content. The measurements were taken at locations that were close to the electrical sensors (used for the electrical resistance method) so that a comparison would be more accurate. 3 MEASUREMENT PROCEDURE In this study, moisture content was investigated in following six different species of wood and woodcomposite products: 1. Poplar 2. Red oak 3. White pine 4. OSB 5. 4-layer plywood (sample taken from discarded renovation waste) 6. 5-layer plywood Three MC measurement procedures were followed – the weight method, the hand-held moisture meter method, and the electrical resistance method. The three tests were conducted on the each wood product. Each wood product was cut into 6”x6” sized samples. These samples were dried on racks in a Thermotron 8200 climate chamber, where the temperature was maintained at 103±2°C and 5% relative humidity. After six hours, the samples were weighed 3.3 Electrical Resistance Method The electrical resistance method presents its own challenges, including development of instrumentation capable of measuring the range of resistance as the wood or wood-based sample dries. For lower resistance, e.g. high moisture content, a simple multimeter would have been sufficient. However, once the samples dried, it was expected that the resistance might approach GΩ range (Forsén 2000, Simpson 1999), so the decision was made to construct a custom circuit with a broader range using a National Instruments USB-6212 and a breadboard with a voltage divider circuit. This incorporated a 1 MΩ 286 reference resistance. The circuit diagram is shown in Figure 1. Two types of moisture sensors were constructed − one using metal nails or “pins” and one using metal screws. An exposed end of insulated 24 gauge solid copper wire was wrapped twice around the nails and secured by a heat shrink tube that extended from the point which was the desired depth of the nail in the sample would be reached, over the head of the nail until no bare wire was exposed. See Figure 2(a). The same type of wire was soldered under the head of the screws. The length of threading left unhindered by soldering was approximately 5/16”– the same length as the probes of the handheld moisture meter and the desired depth. Shrink tube was again used to cover as much exposed wire as possible. In this case, however, the head of the screw was not covered. The sensors were installed in the samples so that there was 1 inch of space between the pair and aligned with the wood grain direction. This is shown in Figure 2(b). Sensors were left in the samples for the drying process. (a) (b) Figure 2. (a) Metallic pin and screw sensors, (b) wood sample with sensors. 4 RESULTS Calibration for pin/screw sensors and correction for handheld moisture meter were performed with oven dry measurements. The readings were taken at least twice per day. Figure 3 shows a plot showing natural drying of the wood specimens once they were removed from water submersion step. It is observed that different specimen started with different levels of MC after water submersion test. As they were allowed to dry out in open atmosphere, MC reduced as wood tries to come to equilibrium with the MC of the surrounding known as equilibrium moisture content (EMC). On first look, it appears that MC is decaying exponentially. However, data cannot be fit with one unique curves exponential fit. It can be seen from Figure 3 that in higher MC region drying rate was faster. The evaporation rate seems to change around fiber saturation point (FSP), which is defined as MC when the wood cell wall is completely saturated with bound water, but no liquid water is present in the cell wall lumens. Above FSP, moisture movement is through capillary forces, while below FSP it is via diffusion. FSP for most wood species falls in the range of 25 to 30% MC (Glass 2010). Below FSP only bound water is present which is strongly attached to the cell wall and hard to remove. This explains slower drying rate in diffusion regime In diffusion regime, moisture diffusivity through wood, α, may be defined as the ratio of second power of length, L, to the time constant, τ, of the exponential decay curve: (a) (b) Figure 1. Electrical resistance measurement (a) circuit diagram, (b) instrumentation setup. 287 𝛼 = 𝐿2 /𝜏 (2) agreement with oven dry values, and follow 1:1 ratio line closely i.e. moisture meter does not require significant correction for OSB and red oak. On the other hand, polar and plywood meter readings are off by as much as 40% from oven dry measurements. Moisture diffusivity rates for all wood specimens are given in Table 1. It can be seen that moisture moves significantly faster in poplar, pine and OSB specimens compared to both ply specimens. Moisture Content (%) 90 Poplar Red Oak Pine Plywood 4Layer OSB Plywood 5Layer Moisture Content (Weight) 40,0 60 30 30,0 y = 0,9853x R² = 0,8979 20,0 10,0 ,0 ,0 10,0 20,0 30,0 40,0 Moisture Content (Handheld) 0 0 100 200 300 (a) 400 Time (hours) 40,0 Moisture Content (Weight) Figure 3. Moisture drying as a function of time for all wood specimens tested in the study. Red line represents fiber saturation point (FSP). 30,0 20,0 Table 1. Measured density, time constant and moisture diffusivity for wood specimens tested in the study. Wood Species Thickness (cm) Density (kg/m3) Time Const (τ) 10,0 Diffusivity (cm2/s) ,0 0 10 20 30 Moisture Content (Handheld) Poplar 1.975 392 16.5 1.9275 711 17.5 5.89E-05 Pine Plywood 4 Layers 1.91 340 15.9 6.38E-05 1.2775 588 17.4 2.61E-05 2.085 512 18.4 6.57E-05 1.25 505 18.7 2.32E-05 OSB Plywood 5 Layers (b) 6.56E-05 Red Oak 40 Poplar Red Oak Pine OSB 5-Layer Plywood Ideal +20% Boundary -20% Boundary +/-40% Boundary +/-40% Boundary Linear (Poplar) Linear (Red Oak) Linear (Pine) Linear (4-Layer Plywood) Linear (OSB) Linear (5-Layer Plywood) Figure 4. Handheld meter calibration curve for (a) red oak, (b) all species of woods tested in the study. Finally, we determine moisture sensor resistance calibration curves for all wood specimens. Figure 5 shows log of resistance as a function of true MC as measured from weight method for poplar specimen. The sample was dried from ~0.75 MC down to EMC of ~0.1. We find that resistance increases as MC reduces. This can be explained by the fact that water is much more conductive than wood. It is also found that resistance varies by ~5 orders of magnitude in the MC range observed during the experiment. It is interesting to note that resistance decreases by almost three orders of magnitude from FSP to EMC, while only a moderate drop by a factor ~60 is observed between 0.75 MC to FSP. Therefore, electrical resistance method is expected to more sensitive and accurate below FSP than above FSP. A best power equation is used to fit the entire range of data from 0.10−0.75 MC (Figure 5(a) and Table 2). A correlation coefficient of 0.985 was obtained for the power fit, suggesting very strong relation between resistance and MC data. Below FSP, electrical resistance seems to change linearly with MC. A linear fit is determined to fit the data as shown in Figure 5(b). Again a very strong correla- Next, we turn our attention to calibration of handheld moisture meter for the six wood specimens. The need for calibration of handheld meter arises from the fact that the meter uses same manufacturer-provided calibration for all wood species. Since calibration curves are expected to differ with wood species, this meter will provide incorrect readings unless it is corrected with a true MC measurement method such as oven dry method for every wood specimen. Figure 4(a) shows calibration result along with a best fit equation and correlation coefficient for red oak specimen. It is found that meter readings match well with weight measurements for this case. Figure 4(b) shows calibration data for all specimens. OSB and red oak readings are in good 288 tion coefficient of 0.984 is calculated, suggesting good quality of the linear fit. Rest of the samples show similar resistance calibration trends, and Table 2 lists the best fit along with the correlation coefficient for each sample. 4-L Ply 5-L Ply 5-L Ply (using nail sensors) -0.1437527 0.983 18.5912156119 -0.39473358326 0.943 8.080417 8.466054 -0.1724505761 0.993 7.348512 8.820771 -0.1392499 0.988 0.89 9.57335961590 -0.204455669854 0.901 7.84791820854 -0.110665901078 0.939 7.33219315574E -0.090854059133 0.930 5 CONCLUSIONS (a) The goal of this project was to develop a practical, continuous and inexpensive methodology to measure moisture content in wood and wood-based species under field conditions. To gain a good understanding of the existing moisture measurement procedure and methods, we conducted a detailed review of the scientific literature. Based on literature review, we found that the electrical resistance based method will be the most suitable method for this project since it relies on simpler physical relation between moisture content and electrical resistance. We decided to use pair of metal screws/nails as sensors because they are cheap, easy to make and instrument. Oven dry weight method was employed for true MC measurements and for calibration purpose. A handheld moisture meter was also used to provide quick MC readings. The main purpose of this effort was to see the accuracy of the meter and correct for any offset by comparing with oven-dry method data. For the purpose of the project, we fabricated testing setup for oven-dry and electrical resistance procedures. Pin and screw sensors were built in-house. Six different wood and wood-based species were tested. First, all the samples were completely dried at 103 ± 2°C in a climate chamber, and dry weight was measured. Next, samples were submersed under a water bath until they gained MC in range of 0.75−1. Samples were subsequently allowed to dry out in atmospheric conditions, and weight, resistance and handheld readings were taken at regular intervals. The experiments were completed once samples reached EMC. The oven dry data as a function of time was used to determine water diffusivity in all wood samples. Handheld meter readings were compared with oven dry data, and found to differ slightly for OSB and red oak, while deviations as high as 40% were found for poplar and ply samples. Correction relations were determined for all tested wood species and will be accounted for in future handheld meter measurements. (b) Figure 5. Moisture sensor resistance calibration for poplar using (a) a power fit covering the entire range of MC examined, (b) linear fit covering MC data below FSP. Table 2. Calibration relations for wood specimens. Screw sensor is used unless otherwise mentioned. CC is correlation coefficient. Wood Power Fit, Linear Fit, Specimen y=a*(x-b)^c y=a+bx Poplar Red Oak Pine OSB a b c CC 45.78859 -10.86970 -0.5317763 0.985 11.44312 5.322800 -0.2591502 0.988 20.22023 -2.500867 -0.3467718 0.994 7.813615E 5.307216 a b CC 10.2451297226 0.131521540670 0.984 9.60329583551 -0.197775784955 0.941 10.0842707969 -0.149093510381 0.953 6.37917764240 -0.046854267203 289 Electrical resistance as a function of true MC from oven dry method showed that resistance increased as samples became drier. Resistance change was found to be significant from FSP to EMC for all samples, indicating good sensitivity and improved measurement accuracy in this range. Power fits were determined for each sample covering the entire MC range observed during the experiment. Below FSP, data was found to follow a linear relationship, and a linear fit was also calculated for each sample. Both power and linear fits showed high correlation coefficients, suggesting strong relationship between resistance and MC data. Screw sensors were found to work reliably and continuously during the course of the study. They provided consistent and good quality data. Kupfer K. (ed). 1997. Material moisture measurements (in German). Expert Verlag, Renningen-Malmsheim. Lin R.T. 1967. Review of dielectric properties of wood and cellulose. Forest Products Journal. 17(7): 61. Maref W., Lacasse M.A., and Booth D.G. 2009. Experimental assessment of hygrothermal properties of wood-frame wall assemblies moisture content calibration curve for OSB using moisture pins. Journal of ASTM International (JAI), 7, (1), Second Symposium on Heat-Air-Moisture Transport: measurements and implications in buildings, Vancouver, B.C. April 19, 2009, pp. 1-22. Phillipson M.C., Baker P.H., Davies M. Ye Z., McNaughtan A., Galbraith G. et al. 2007. Moisture measurement in building materials: an overview of current methods and new approaches. Building Services Engineering Research and Technology, 28(4):303–16. Said M.N. 2004. Moisture measurement guide for building envelope applications. Research Report #190, Institute for Research and Construction. Available at http://irc.nrccnrc.gc.ca/ircpubs. Stamm A.J. 1927. The Electrical Resistance of Wood as a Measure of Its Moisture Content. Industrial and Engineering Chemistry 19(9), pp. 1021-25. 6 REFERENCES ASHRAE. 1985. Fundamentals Handbook, Chapter 23: Design Heat Transmission Coefficients. ASTM D4442 – 07. 2007. Standard Test Methods for Direct Moisture Content Measurement of Wood and Wood-Base Materials. ASTM D4444 – 08. 2003. Standard Test Method for Laboratory Standardization and Calibration of Hand-Held Moisture Meters. Brischke C., Rapp A.O., and Bayerbach R. 2008. Measurement system for long-term recording of wood moisture content with internal conductively glued electrodes. Building and Environment 43(10): 1566-1574. Dill M.J. 2000. A Review of testing for Moisture in Building Elements. CIRIA Report No. CIRIA-C538, Westminster, London SW1P 3AU, UK. Also available at www.ciria.org.uk. Forrer J.B., and Vermaas H.F. 1987. Development of an improved moisture meter for wood. Forest Products Laboratory Journal, 37(2), pp. 67-71. Forsén H., and Tarvainen V. 2000. Accuracy and functionality of hand held wood moisture content meters. Espoo: Techn. Research Center of Finland, VTT Publications 420. 79 p and App. 17 p. Garrahan P. 1988. Moisture meter correction factors. Proceedings of a seminar on in-grade testing of structural lumber. Forest Products Laboratory, Forest Service, US Department of Agriculture, Madison, Wisconsin, April 25-26. Glass S.V., and Zelinka S.L. (2010). Moisture Relations and Physical Properties of Wood. Wood Handbook Chapter 4, General Technical Report FPL-GTR-190. Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Products Laboratory: 4-1 - 4-19. Grinzato E., Vavilov V., and Kauppinen T. 1998. Quantitative infrared thermography in buildings. Energy and Buildings Journal, 29, pp. 1-9. Hauschild T., and Menke F. 1998. Moisture measurement in masonry walls using a non-invasive reflectometer. Electronics Letters, 34(25), pp. 2413-2414. James W.L. 1963. Electric Moisture Meters for Wood. FPL Note 08, Forest Products Laboratory, Madison, Wisconsin. James W.L. 1994. Fundamentals of hand held moisture meters: An outline. Proceedings ASTM Hand Held Moisture Meter Workshop. Forest Products Society, May 1993, Madison, Wisconsin, pp. 13-16. Sereda P.J., Croll S.G., and Slade H. F. 1982. Measurement of the time of wetness by moisture sensors. Atmospheric Corrosion of Materials, ASTM STP 767, pp. 267-285. Simpson W., and TenWolde A. 1999. Physical properties and moisture relations of wood. Wood Handbook: Wood as an Engineering Material. Madison, WI: USDA Forest Service, Forest Products Laboratory, General technical report FPL; GTR-113: Pages 3.1-3.24. TenWolde A., and Courville G.E. 1987. Instrumentation for Measuring Moisture in Building Envelopes. Forest Products Laboratory, HI-85-22 No3, Madison, Wisconsin. Trechsel H.R., and Bomberg M. 2006. Moisture Control in Buildings: The Key Factor in Mold Prevention: 2nd Edition. ASTM Stock Number: MNL18-2nd. Ye Z., Tirovic M., Davies M., Baker,P.H., Phillipson M.C., Sanders C.H., Galbraith G.H., and McLean R.C. 2007. The testing of two methods for the moisture measurement of building fabrics via comparisons with data from an X-ray system. Building and Environment 44(7) Pages 1409–1417. 290 Laboratory investigation of drying of built-in moisture in wood frame walls at passive house level A. Dalehaug & S. Geving & M. Gaare & K. Løtveit Norwegian University of Science and Technology, Trondheim, Norway J. Holme SINTEF Building and Infrastructure, Trondheim, Norway ABSTRACT: In this study a laboratory experiment has been performed, where drying of built-in moisture in the wooden members after closing the wall were monitored. Five wood frame wall elements in full height with 300 mm insulation were built between two climate chambers giving a specific indoor and outdoor climate. The bottom sills in the elements were wetted to different moisture levels, and instrumented with electrodes for measurement of wood moisture content. The top sills were also instrumented to see if there was any significant redistribution of moisture due to internal convection. The chosen drying climate was 23 °C and 60% RH at the interior side and 0 °C and 80% RH and the exterior side. The measurement period was two months. After the measurement period no visible mould growth was observed on any of the bottom sills. Using a mould growth model indicated however that mould growth theoretically should have taken place on the bottom surface of the bottom sills that originally had been put partly in water for 3-5 days. Mould growth were however not a theoretical problem for the bottom sills that either had been wetted by limited spraying of water or hygroscopically stabilized. Within two months all parts of the wood had dried below 20 weight%. The wettest parts were generally the bottom part of the bottom sills, especially in the middle of the wall and close to the sealed end of the sill. Some redistribution of moisture to the top sills due to internal convection was observed. 1 INTRODUCTION 2 LABORATORY TESTING When closing external wood frame walls during construction, it should be controlled that the wooden members are sufficiently dry so that mould or other problems may not occur. This may typically be done by measuring the moisture content before thermal insulation and vapour barrier are added. These measurements are then compared to a limit value, a critical moisture content, that should not be exceeded. Existing recommendations for critical moisture content in wood before closing the wall varies typically between 18-20 weight% in Europe and North America. A problem may however be that these recommendations are related to the standard constructions from a few years ago, and not modern constructions at passive house level. Some field investigations, e.g. (Nore and Clementz, 2011) indicate that ordinary passive house walls are not exposed to unfavourable moisture conditions when high levels of built-in moisture are avoided. Previous investigations have however also shown that built-in moisture dries slower in highly insulated wood frame walls, thereby increasing the risk of mould growth. There may therefore be a need for more strict requirements for the critical moisture content in such highly insulated constructions. 2.1 Introduction The laboratory test was carried out to simulate a realistic, but also critical situation that can occur on a building site. It was decided to do the test on simple bottom sills, because bottom sills are the construction parts that are most exposed to moisture loads in the construction period. In the laboratory experiment we are not copying the most extreme conditions, by letting the bottom sill dry for five days before closing the construction. The test is built up as a wood frame wall with normal dimensions, i.e. 2.40 m tall. The whole test wall consists of five different wall elements, where all the structural parts; bottom sill, top sill and studs are wooden. Before the constructions were tested between the climate rooms, the bottom sills were wetted in different ways to simulate the effect from rain and/or free water in shape of small dams on floors and other horizontal surfaces. The test wall was set up between two climate rooms with constant outdoor and indoor climate. The wall elements were mounted side by side in an opening in the wall separating the two climate rooms. The test sequence is as follows. • Hygroscopic moistening • Water bath and water spraying • Mounting and drying in the wall 291 • Closing of the structure and start measuring moisture • End of test convection with possible redistribution of moisture from bottom sill to top sill and studs. To be able to evaluate the risk of mould growth on the bottom sills, the moisture content in bottom sills was monitored. 2.2 Purpose The purpose of the laboratory test is to find out about how drying of moist bottom sills progress in highly insulated construction and to evaluate the risk of mould growth. I.e. to evaluate what moisture content that is acceptable in wooden bottom sills when closing highly insulated wood frame walls. The bottom sills were tested with different initial moisture contents, varying from relatively high levels to hygroscopic level. The tested walls were made at full height to be able to study the full effect of internal 2.3 Tested constructions 250 mm The tested wall consists of five wall elements. The wall elements are surrounded by a wooden framework insulated by mineral wool. Between the framework and the wall elements there is a continuous layer of 100 mm expanded polystyrene as shown in figure 1 and 2. Surrounding wall 100 mm EPS Top sill 48×300 mm Stud 24×300 mm Wall element A Wall element Wall element C D Bottom sill 48 mm Bottom sill 48 mm Wall element E Bottom sill 36 mm 2400 mm Bottom sill 48 mm Wall element B Bottom sill 48 mm Stud 24×300 mm Polythene foil 0.15 mm Stud 24×300 mm Bottom sill 48×300 mm 600 mm Bottom sill 36×300 mm Parqet polythene foam underlay 2 mm Polythene foil 0.15 mm Bottom sill 48×300 mm 400 mm 3000 mm 200 mm Figure 1. Installation of tested wall elements in wall between climate chambers. The details of the bottom of the framework are shown in figure 2. On top of the 100 mm expanded polystyrene layer there is a «Dummy»-bottom sill for practical reasons. To be able to moisten and mount the tested bottom sills in a short time, the main framework and most of the construction had to be in place before the testing started. The "dummy" can be looked upon as a replacement for a floor construction (for example a platform floor of 22 mm chip board). The polystyrene insulation will ensure independent temperature conditions around the perimeter. Each wall element is based on studs c/c 600 mm. To get the right percentage of wood in each wall element, the stud is split in two and each half (24 mm thick) were placed on each side of the mineral wool insulation. The studs were separated by a 0.2 mm thick polythene foil to prevent moisture transport sideways. 292 The measurements were done manually by using small stainless screws (dimension 3.5x9.5 mm) connected to isolated wires. Each measuring point consists of two screws 25 mm apart from each other. As stated above the tested sill is only 552 mm long. To get results that will be valid for longer sills, one end of the sills are sealed by epoxy paint. By positioning one row of measuring points 100 mm from the sealed end the measurements will be as from 452 mm from a normal end of a longer sill. Measuring points are also placed 10 mm from the open end to see if there is any difference. Asphalt impregnated fibre board, 12 mm 300 mm mineral wool 0.15 mm polyethene foil Bottom sill 48×300 mm Polyethene foam parqet underlay 2 mm 0.2 mm polyethene foil Measuring points are also placed in the top sill to see how much of the moisture that is redistributed by diffusion and convection in the cavity. The measuring points in the top sill are shown in figure 4. The measuring points are numbered 1 - 3, and the extra letter “t” for top sill. Measuring point St1 is at the midpoint of the outside of the sill and St2 and St3 are located on the underside and 10 mm from the side and 74 mm from the midpoint of the sill. Bottom sill 48×300 mm 100 mm ekspanded polystyrene Figure 2. Cross section of lower part of tested wall. On top of the "dummy" a vapour and moisture barrier of 0.2 mm polythene foil and a 2 mm layer of polythene foam were mounted. The latter to avoid openings around and in connections with wires from the measuring probes and avoid openings that would allow convection as the sill dries and may be change shape and dimensions. The upper bottom sill is the wetted "loose" bottom sill which was to be tested. After the open drying period, mineral wool and vapour barrier were mounted. Outdoor climate 2.4 Instrumentation To evaluate risk of mould growth, moisture content near the surfaces of bottom and top sills were measured by using a resistance measurement instrument that has weight-% as output. A Brookhuis Micro Electronics - FME moisture meter was used. The instrument can compensate for type of wood and temperature. In this construction spruce was used. The measurements were done on the surface of the underside of the bottom sill, because mould will grow on a surface and the underside will be the wettest place because of the smallest possibility to dry. The underside is also wetter because it was that side that was submerged in water. Measurements were also carried out at two points on the side of the bottom sills facing outdoor climate, and at one point on the topside of the bottom sills. See figure 3 for the nine measuring points (named S1 to S9) on each bottom sill. Indoor climate Figure 4. Measuring points on the surface of the top sills. Measuring points were also placed in the middle of the bottom sills to get better data to evaluate the drying process. This was done by using partly insulated electrodes hammered into pre drilled holes. The holes were drilled from the top to a depth of 22 mm (16 mm for bottom sill E). Then the electrodes were put into the holes, and knocked in another 3 - 4 mm. Distances were the same as for the screws. It was measured in six points in each bottom sill. The points were numbered from 1 - 6, with the letter «E» for electrode. The positioning of the measuring points are shown in figure 5. The internal points are positioned right above the measuring points on the underside of the bottom sills. Outdoor climate Outdoor climate Sealed end Indoor climate Indoor climate Sealed end Figure 3. Measuring points on the surface of bottom sills. Figure 5. Measuring points in the centre of the bottom sills. 293 2.5 Conditioning of the wooden material before the test All the framework for the construction were kept in a climate room until hygroscopic equilibrium at RH 84 % and temperature 30 °C. The operation took five weeks in the climate room. The bottom sills then had a moisture content of 17 - 18 weight-%, which was the maximum that could be achieved by plain hygroscopic moistening. The top sills and studs had a moisture content of 15 - 16 weight-%. The reason for higher moisture content in the bottom sills, was that they were split up to fit the wall elements. One end was sealed by to layers of a liquid membrane. The crossgrain of a piece of wood absorbs moisture much more effectively than the rest so by doing this, the rather short bottom sill will function more like a much longer sill. Figure 1. Bottom sill in water bath. Water spraying was done by using a spray bottle to simulate rain. The spraying was done until a thin film of water could be seen on the surface of the wood. That would be like simulating 0.3 mm of rain, see table 2 for number of water spraying for each bottom sill. These moistening operations were done in boxes stored at room temperature. Studs and top sills were only moistened hygroscopically, and were then built into the test rig without any additional moisture load. 2.6 Simulation of water dams and influence from rain Three of the bottom sills (A, B and C) were, after being hygroscopically moisturized, put in a 2 mm deep water basin and sprayed with water two times every day, see table 1. Table 2. Amount of water applied to bottom sills by spraying. Size of surface Size in Weight of Number of [mm] [mm3] water [g] sprayings [pcs]* Topside: 49 500 49.5 45 300 x 550 x 0.3 Short end: 4 320 4.32 4 300 x 48 x 0.3 Short end: 3 240 3.24 2 300 x 36 x 0.3 Long side : 7 920 7.92 7 550 x 48 x 0.3 Long side: 5 940 5.94 5 550 x 36 x 0.3 Table 1. Water applied to the bottom sills. BotDimension Moisture Moistening tom level sill A 48 x 300 Maxi2 mm water bath, 1 week mm mum + 2 daily water sprays + 5 days drying in wall B 48 x 300 High 2 mm water bath, 3 days mm + 2 daily water sprays + 5 days drying in wall C 48 x 300 Medium 2 daily water sprays all mm sides for 3 days + 5 days drying in wall D 48 x 300 Low Hygroscopic moistening mm + 5 days drying in wall E 36 x 300 High 2 mm water bath for 3 mm days + 2 daily water sprays + 5 days drying in wall 2.7 Mounting the test walls The frame around the wall elements was set up first in the climate room. The moistened framework, including top sill was mounted one week before the moistened bottom sills. During this week the temperature and RH on one side of the test wall was set to 23 °C and 70 % RH to avoid substantial drying of the wooden materials before mounting the bottom sills. The other side was covered by plastic foil. Then the wind barrier and the moistened bottom sills were installed in the test wall. See figure 7. Sill D was only moistened hygroscopically, and should therefore simulate a bottom sill that does not get in direct contact with water. This sill was built in with a moisture level of 15 - 16 weight-%. The moisture measurements were done after all bottom sills were built in. The water bath should simulate exposure of bottom sills to a body of water on a floor, because of for example unevenness or local slopes on the floor. When the bottom sills were in the water bath, the boxes were covered by plastic foil to avoid evaporation from the bath and drying of the sills, see figure 6. 294 3 RESULTS AND EVALUATION The moisture content in top and bottom sills were measured daily in all measuring points the two first weeks starting right after closing the wall elements. Later on it was measured roughly every two days for another two weeks because the moisture content was rather stable. 3.1 Correction of measured values because of varying temperature The electric resistance in wood is reduced by increased temperature (Geving & Thue, 2002). The resistance meter is set to 20 °C, but it will not be constantly 20 °C in the bottom and top sills. It is therefore necessary to make corrections for this temperature difference, at the different measuring points, to get the correct moisture. The correction is done by using the following formula. Figure 7. Mounted bottom sills in test walls. Then control measurements of moisture content in the framework in each wall element, including the top sills were done with hammer electrodes. The moisture level was measured at different depths in the framework and showed that moisture level was satisfying so no extra moisture had to be added. The bottom sills were left to dry in the test wall for five days to simulate more realistic conditions for a building. It is not realistic to believe that bottom sills will be built in if they have been heavily exposed to precipitation, without being allowed to dry for some days. The drying climate was 10 °C and 60 % RH on both sides of the wall. Then insulation and vapour barrier were mounted, and the measuring of moisture started, see figure 8. uk = corrected moisture content [weight-%] u = registered moisture content [weight-%] t = temperature in the wood material [°C] The temperature in the measuring points were simulated by using WUFI 2D, since no temperature are measured inside the test walls. 3.2 Overview of results from drying of bottom sills The main results from the drying process for all bottom sills are given in table 3. After the table a more detailed description will be presented. In the table the initial water content, water content after five days of open drying and finally after four weeks of drying of closed construction are displayed. The range of registered values and the average value is also presented. The measuring point with the highest water content is given for each bottom sill. Figure 8. Test wall seen from the warm side. When reading the table you should look at table 1 to see the water load and figure 3 for the numbering of the measuring points. To get a good moisture profile of the wetted bottom sills at the start of the test, a detailed experiment simulating the one week drying period was carried out followed by careful measurement of moisture content at many points in the sills. 295 Table 9. Results from the drying of bottom sills. BotMoisture content after apMoisture content aftom plying water and before ter five days of drysill drying [weight-%] ing [weight-%] A 30.3 – 53.5 (44.6) B 27.3 – 49.9 (40.0) C 14.0 – 42.1 (28.2) D 14.7 – 17.0 (15.8) E 26.8 – 48.7 (39.4) * Average values are in parentheses 21.5 – 46.0 (32.8) 18,1 – 42.3 (25.5) 11.9 – 25.1 (19.2) 13.8 – 15.5 (14.6) 18.5 – 40.3 (24.9) The wettest point after five days of drying E4 E5 S7 S7 E5 Moisture content four weeks after closing the wall* 11.6 – 28.9 (21.1) 14.1 – 21.3 (17,5) 11.5 – 18.6 (15.1) 12.2 – 15.3 (13.5) 11.5 – 21.9 (16.1) The wettest point four weeks after closing the wall S1 S7 S5 S5 S2 on from the warm side of the construction. Wood with high initial moisture located near the cold side will dry fast, but still have moisture content of 24 weight-% after four weeks. The wood in point S4 and S8, facing outdoor climate, dries faster than at the other points near the cold side. The cause for this might be internal redistribution caused by natural convection that makes the lower, outer part of the wall become drier. The bottom sill does not dry much in the centre towards out door climate and sealed end (E1). The registered drying is small probably because of moisture migrating from warm to cold side. Towards the inside the drying is slow because of the vapour barrier, but dries slowly out during the first four weeks. 3.2.1 Bottom sill A Bottom sill A was given the wettest treatment before the test: 2 mm water bath for 5 days + 2 daily water sprays. The treatment led to the highest moisture content of them all. Drying of sill A is slow and after 4 weeks the moisture content in parts of the sill is about 25 - 30 weight-%, see figure 9 and 10. This is mostly for the central parts of the underside of the sill; point S2 and S7. The centre of the sill dries faster than the underside and especially fast at the not sealed end. In the sealed end the membrane prevents moisture absorption and drying. The not sealed end will therefore absorb more moisture and in addition dry faster. After four weeks the moisture content is still higher on the not sealed end. 3.2.2 Bottom sill B Bottom sill B was exposed to 2 mm water bath for 3 days + 2 daily water sprays, i.e. about half the exposure of sill A. This led to high water content as listed in table 1, i.e as an average over 25 weight-%. Figure 9. Moisture content in bottom sill A. Outside Outside Figure 11. Moisture content in bottom sill B. Inside The water content is initially from 5 to 10% lower from the beginning than for sill A. There are less free water in the wood cells that make the open drying the first five days reach a lower water content than in sill A. The wettest points are S7 and S2 under the sill near the inside. It takes about 20 days to get the water content below 25 weight-%. In the middle of the sill it is only point E5 that starts above 25%. The typical development is the same as in sill A, the water content is highest at the not sealed end Inside Figure 10. Numbering of measuring points. S on surface, E in the middle between top and bottom of sill. Most of the measuring points indicate that the bottom sill is drying during the first four weeks. The wood in point E1 and in point S4 and S5 is facing outdoor climate and seem to be stabilized during the first day. A possible explanation can be condensation on the wind barrier because drying is still going 296 in the beginning, but dries out faster and is reduced to about 20% after three weeks. The graphs indicate that the drying continues. 3.2.3 Bottom sill C Bottom sill C was only exposed for 3 days to 2 daily water sprays to all surfaces, i.e. capillary absorption could only take place right after the spraying. This led to medium water exposure as listed in table 1, with an average water content of 28 weight-% when built in and reduced to 19% after the open drying. The water exposure was less than for sill A, B and E, as the simulation was not supposed to include so heavy rain that water dams were created. Figure 13. Moisture content in bottom sill D. The moisture levels seem to be varying a lot, but the scale in figure 13 is quite different compared to the other graphs. Still there are some variations that indicate that the points on the outer side of the construction dry faster than the rest. The wood dries fastest in the internal measuring points E3 and E6 towards inside climate and in S3 and S8 towards out door climate. This is the same development as in bottom sill C. 3.2.5 Bottom sill E Bottom sill E was exposed to 3 days in water bath with 2 daily water sprays, that is listed as high moisture load in table 1. The treatment was the same as for sill B. The sill had at closing a start moisture level at about 25 weight-%. I.e. about the same as sill B, but the thickness is 36 mm compared to the others at 48 mm. The measurements show that the sill is more moist at the surface than in the centre. The point with the highest water content after four weeks is S2, that is at the surface near the sealed end. The wood in point S7 has the highest moisture level after insulating and closing of the wall, but dries out very fast the first four weeks. The point is near the not sealed end. The sealed end reduces the rate of drying as for the other sills. Figure 12. Moisture content in bottom sill C. The moisture content is about the same on the surface as in the middle of the sill during the first four weeks. In the centre the moisture content is more reduced than on the underside of the sill. See figure 12. The measurements inside the sill indicate that the moisture balance is reached before the wall is closed. Only the surface in point S3, S7 and S8 continues to dry after about two days. Measurement of moisture levels in the top sill indicate that moisture is transported from the bottom sill to the top sill. This goes on for two days before both sills are drying. 3.2.4 Bottom sill D Bottom sill D was only exposed to hygroscopic moisture that is listed as low moisture load in table 1. The moisture content at the start was as an average about 14 eight-% after insulation and closing of the wall. The bottom sill has slightly higher moisture content on the surface compared to in the centre. See figure 13 for the moisture on the surface. Figure 14. Moisture content in bottom sill E. 297 5 REFERENCES Towards the outside the wood dries in measuring point E4, S3 and S8 during the first four weeks. The wood in the other points towards the outdoor climate is stabilized after a couple of days. Gaare M., and Løtveit K. 2012. Kritiske fuktforhold ved lukking av høyisolerte konstruksjoner i bindingsverk av tre. Master thesis; Norwegian University of Science and Technology, Trondheim, Norway Geving S., and Thue J. V. 2002. Fukthandboka. Norges byggforskningsinstitutt. Geving S., and Holme J. 2010. Increased insulation thickness and moisture. Proceedings of the 3rd Nordic Passive House Conference, Aalborg, Denmark, 7-8 October. Nore K., and Clementz C. 2011. Hygrothermal performance of highly insulated timber frame walls – field investigation and numerical validation. Proceedings of the 3rd Nordic Passive House Conference, Helsinki, Finland, 17-19. October. Time B., Geving S., Uvsløkk S., and Holme J. 2009. Gir en byggeskikk med mer isolasjon i bygningskroppen økt fare for fuktskader ? (Does a building tradition with more insulation in the construction increase risks for moisture damage?) Proceedings of the 2nd Nordic Passive House Conference, Gothenburg, Sweden, 27 – 29th April 4 CONCLUSION In this study a laboratory experiment has been performed, where drying of built-in moisture in the wooden members after closing the wall were monitored. Five wood frame wall elements in full height with 300 mm insulation were built between two climate chambers giving a specific indoor and outdoor climate. The bottom sills in the elements were wetted to different moisture levels, and instrumented with electrodes for measurement of wood moisture content. The top sills were also instrumented to see if there was any significant redistribution of moisture due to internal convection. The chosen drying climate was 23 °C and 60% RH at the interior side and 0 °C and 80% RH and the exterior side. The measurement period was two months. After the measurement period no visible mould growth was observed on any of the bottom sills. Using a mould growth model indicated however that mould growth theoretically should have taken place on the bottom surface of the bottom sills that originally had been put partly in water for 3-5 days. Mould growth were however not a theoretical problem for the bottom sills that either had been wetted by limited spraying of water or hygroscopically stabilized. Within two months all parts of the wood had dried below 20 weight-%. The wettest parts were generally the bottom part of the bottom sills, especially in the middle of the wall and close to the not sealed end of the sills. Some redistribution of moisture to the top sills due to internal convection was observed. 298 Wooden beam ends in masonry with interior insulation – A literature review and simulation on causes and assessment of decay D. Kehl, U. Ruisinger, R. Plagge, J. Grunewald Dresden University of Technology, Department of Architecture - Institute of Building Climatology, Dresden, Germany ABSTRACT: Decay of wooden beam ends in masonry walls is still a field of practice and scientific discussion. In literature many possible causes for decay are given. This paper delivers a comprehensive literature review about the hygrothermal performance of old masonry with and without interior insulation. In old walls without insulation the reasons for decay at wooden beam ends are often damaged downpipes, leaking roofs and lack of protection against wind driven rain. Furthermore the analyses of several interior insulation projects with wooden beam ends demonstrate that moisture load from outside by wind-driven rain (wdr) has an important influence on the behavior of moisture content of the beam end. In addition, hygrothermal simulations are carried out to detect and analyse the influence of the micro climate around the wooden beam end. A construction is simulated with and without convection from inside as well as with and without interior insulation. To interpret the results a simplified wood decay model based on the Viitanen model is used. The model considers the temperature dependency. This indicates that simple thresholds in national standards as 20 M-% moisture content is not suitable. 1 INTRODUCTION the wood and therefore the risk of decay increases (Promnitz 1881 in Krause & Ahnert 2009). This literature review demonstrates that the problems with wooden beam ends were early known but the causes for destruction were seldom mentioned. Only in (Anker 1919) the information was found that the wind driven rain (wdr) could cause high moisture content in masonry walls and decay within wooden beam ends. In recent literature (IBA 1985) (Hofmeister 1995) the main reasons for decay of wooden beam ends were found to be: The energy consumption of historical buildings with structured facades can often be improved only with interior insulation. A lot of buildings with interior insulation show that good results can be reached i.a. (Borsch-Laaks & Simons 2012) (Borsch-Laaks & Walther 2008) (Loga 2005). However architects, engineers and crafts men face confusion when handling wooden beam ends which penetrate the interior insulation and lie in the cold masonry. They are concerned that the wooden beam ends will increase in high levels of moisture content, resulting in destruction by decay. However, in most projects measurements of wooden beam ends demonstrate no problems with moisture content (see chapter 4). − damaged downpipes − leaking roofs and − lack of protection against wind driven rain This is confirmed by experts (Eßmann 2011) (Müller 2011). All reasons are caused by vacancy or lack of maintenance whereas the last one could also be caused by planning mistakes and poor material properties. Therefore, the approach of the outside of the wall is important for the survival of the wooden beam end. 2 REASONS OF DECAY OF WOODEN BEAM ENDS IN HISTORICAL BUILDINGS To understand the overall context it is important to know the reasons why wooden beam ends decay in historical buildings (without insulation). At the beginning of the last century wooden beam ceilings were the favorite construction in masonry buildings. In 1937 the proportion of wooden beam ceilings were estimated at nearly 80 % (Krause & Ahnert 2009). From this period of time you find information in literature about the construction details of wooden beam ends and even police regulations (Stade 1904) (Krause & Ahnert 2009). On one hand you should paint the wooden beam end with tar and build an air layer in front of the wooden beam end. On the other hand authors warn of the treatment with tar because the “vapour barrier” reduces the drying process of 3 RECENT STANDARDS In the sixties and seventies of the last century many investigations in Germany occurred (Reiher et al. 1960). They were about moisture load from outside and plaster properties and took place on different wall materials (i.a. bricks, aerated concrete). Therefore in 1981 national standards were made for protection against wind driven rain (DIN 4108-3). Dependent on the moisture load from wdr (wdr class I 299 – III) the standard imposes demands on the constructions and hygrothermal coefficients Aw-value (water absorption coefficient) and sd-value (vapour diffusion thickness) of the exterior plaster. These regulations cannot directly be used for masonry with interior insulation because the insulation reduces the drying potential of the wall. In (WTA 2013) the demands for plaster are a little bit higher than in the regulations. In other cases simulation is necessary. If the old plaster will replace it, it is not complicated to reduce the penetration of wind driven rain into the wall with a new water-repellent and vapour open plaster or coating. In regards to masonry without plaster, it could be necessary to treat the bricks with a water repellent finish. This is dependent on the amount of wind driven rain. The success of the treatment should be controlled. Otherwise there is a risk of moisture problems in the wall and at the wooden beam ends (Peper et.al. 2010) (Hasper et.al. 2010). of the gap between wooden beam and plaster has an influence on the moisture content at the beam end. 5 CONCLUSION OF THE LITERATURE REVIEW The conclusion of the literature review is: If the moisture load by wind driven rain is low or a water repellent and vapour open plaster protect the wall against the rain, the risk of decay at the wooden beam end is low. Bare brick masonry reacts with sensitivity on wind driven rain. If the wdr is high, high moisture content at the wooden beam end can be expected. In some cases a well-built water repellent treatment can reduce the moisture load form outside. The convection from inside to the air gap in front of the beam end causes higher relative humidity. 6 ASSESSMENT OF THE MOISTURE CONTENT AT THE BEAM END 4 WOODEN BEAM ENDS IN MASONRY WITH INTERIOR INSULATION (MEASURMENTS) As (Viitanen et. al 2010) (Maurice et.al. 2011) show the growth of decay is dependent on the temperature. Therefore the assessment of measurements and simulation results should take place with a temperature dependent borderline. 4.1 Moisture load by wind driven rain In the last 15 years a lot of long-term measurements were made at wooden beam ends in masonry walls with interior insulation. Some buildings have walls with exterior plaster, other have no plaster outside (bare brick walls). In the projects with exterior plaster (Loga 2005) (Bednar et.al. 2010) (Ruisinger 2012) (Gnoth et.al. 2005) the moisture content of the wooden beam ends range between 10 and 18 M-%. Except after renovation, where it was a little bit higher. In these projects no critical situations were found. Either the wind driven rain was low because of cardinal direction or the walls were protected with water repellent plaster or coatings. As before, in the projects with bare brick masonry the moisture load from outside was different. In (Stopp et.al. 2010) (Ratzlaff et.al. 2005) the wind driven rain was low due to cardinal direction. Only in (Hasper et.al. 2010), where the treatment of a bare brick wall with water repellent fluids were not sufficient, the moisture content of the beam ends was high over a long period of time. Figure 1. Simple temperature dependent borderline for decay (derived by Viitanen). The temperature and relative humidity of measurements and simulation results can evaluated. For engineering work a simple borderline is derived from Viitanen (figure 1). In the laboratory tests of Viitanen no decay of brown rot (serpula lacrymans and coniophora puteans) was established at spruce and pine sapwood after 12 month under these conditions. This indicates that simple thresholds in national standards as 20 M-% moisture content are not suitable. In Comparison to the measurement of wood moisture content by electrical resistance the assess- 4.2 Moisture load by convection In addition to wind driven rain, moisture comes from inside air, which flows from the room to the air layer in front of the beam end. If there is a connection from the room to the air gap in front of the beam end, the relative humidity is higher (Ruisinger 2012). So Ruisinger presented that the airtightness 300 The values for airtightness were taken out of (IWU 1995), where the airtightness of different materials and constructions at 50 Pa pressure difference were listed form the literature (table 1). ment of temperature and relative humidity has advantages. On one hand the measurement by electrical resistance is not very accurate (± 1,5 M-%) and on the other hand at simulation results in front of the beam end are almost independent of the sorption isotherm of the material in the software. Table 1. airtightness of different materials and constructions (IWU 1995). Construction / layer q50 [m³/m²h] Cement plaster 0,001 – 0,002 Lime cement plaster 0,002 – 0,05 „soft“ lime plaster (1928) 0,02 – 0,6 Plaster with distemper Airtight Bare masonry 2,1 – 15 Masonry with interior plaster 0,1 – 2,0 7 EXAMPLE In order to show the influence of the convection on the moisture content of the wooden beam end the following construction will be simulated by DELPHIN 5.8 (figure 2). The 36,5 cm thick masonry is west oriented. On the outside is a water repellent and vapor open plaster (Aw = 0,0083 kg/m²√s / sd-value = 0,22 m). Both the wall with and without insulation will be simulated. The moisture content of 5 mm of the beam end and the temperature and relative humidity will be evaluated. The masonry with plaster is nearly airtight. For the simulation the following assumption is made: plaster q50 = 0,1 m³/m²h / masonry q50 = 2,0 m³/m²h. A laminar airflow is assumed. 7.2 Assumption of pressure difference In order to calculate the convection into the air gap in front of the beam end the pressure difference by thermal buoyancy is needed (equation (1)). ∆𝑃 = 𝜌 ∙ ΔP ρ Te Ti g h 𝑇𝑒 −𝑇𝑖 𝑇𝑖 ∙𝑔∙ ℎ 2 (1) pressure difference [Pa] density of air (ρ = 1,3 kg/m³) [kg/m³] Temperature outside [K] Temperature inside [K] gravitational constant (g = 9,81 m/s²) [m/s²] height of the connected airspace in the building [m] From the stack-effect you get the following figure 3. It causes an over pressure in upper section of the building. The under pressure in summer is not taken in consideration. Figure 2. schematic figure of the simulated detail. The simulations take place with/without insulation and with/without convection. The details will be simplified by a 2D simulation. There is a great amount of wind driven rain in Essen’s outside climate. Inside, a normal moisture load of EN 15026 is used. In order to take account, the convection of the following assumptions is made. 7.1 Assumption for Airtightness of the masonry Recently there is no scientific evidence about the convection out of the room to the wooden beam end. Therefore measurements of convection at wooden beam ends will be executed at the TU Dresden. Now, for the simulation it is supposed that a pressure difference occurs in the upper section of the building by thermal buoyancy. This creates airflow from inside to the end of the wooden beam. Masonry is not airtight, therefore, it can gain an airflow through the brick and plaster layer (see figure 2). Figure 3. Calculated overpressure difference between in- and outside (stack-effect) by equation (1). 301 pendent borderline is better than only to consider the higher moisture content of figure 4. 8 RESULTS The simulations based on the mentioned assumptions are still subject to uncertainties. If one boundary condition will change (higher convection, water uptake of the exterior plaster or higher indoor climate) the construction can become higher in moisture content, resulting in decay. Figure 5. relative humidity and temperature at the wooden beam end embedded in the masonry without insulation and with convection. Figure 4. Moisture content of the wooden beam end: simulated with / without interior insulation and with / without convection. As a result of the interior insulation the moisture content of the wooden beam end increases. The convection leads to a further higher moisture content. Therefore both effects, interior insulation and convection, make sure that the construction will dry very slowly over the years (figure 4, dark dashed line). When the assembly of the interior insulation and airtightness of the penetrated beam improves, the moisture content decreases under 18 M -%. In addition, a better airtightness leads to lower energy consumption of the whole building. Already the amount of leakage is unknown. Therefore the TU Dresden will make measurements at wooden beams embedded in masonry. It remains to be seen whether the results of the measurements are in order to get new knowledge. Figure 6. relative humidity and temperature at the wooden beam end embedded in masonry with interior insulation and convection. If the airtightness is improved parallel to the reduction of the energy consumption with interior insulation (fig.7), the potential of damage is the same like in the masonry with convection which isn’t renovated (fig. 5); and that of even higher moisture content. In both cases the distance is nearly the same. 8.1 Evaluation with simplified decay model In figure 5 to 7 the analysis of the results is seen with the simplified decay borderline as a scatter diagram. In comparison to the non-insulated masonry with convection (fig. 5) the dotted values of the insulated masonry with convection move up too (higher moisture content) and left (colder) (figure 6). The potential of damage which is described by the distance of the dots to the borderline is only a little bit higher than before. To look at the temperature de302 11 REFERENCES Anker A. 1919. Naturbauweisen - Ein Ratgeber für Siedler und Baulustige, Deutsche Landbuchhandlung, Berlin Bednar T., Schneider U., Sima J., and Liebich H. 2010. Baudenkmale im Spannungsfeld von Energieeffizienz und Risikovermeidung, article in Bauphysik-Kalender 2010, Verlag Ernst und Sohn, Berlin Borsch-Laaks R., and Simons P. 2012. Wie dick darf Innendämmung sein? 7-Jährige Feuchtemessungen bei einem niedersächsischen Fachwerkhaus, article in Holzbau – die neue quadriga (edition 6-2012), Kastner-Verlag, Wolnzach Borsch-Laaks R., and Walther W. 2012. Innendämmung mit und ohne Dampfbremse – Erfahrungen aus Langzeitmessungen des Feuchtegehalts, article in Holzbau – die neue quadriga (edition 2-2008), Kastner-Verlag, Wolnzach Borsch-Laaks R. 2010. Zur Schadensanfälligkeit von Innendämmungen - Bauphysik und praxisnahe Berechnungsmethoden, proceedings Aachener Bausachverständigentage 2010, Aachen Hasper W., Kaufmann B., Pfluger R., and Feist W. 2010. Energetische Sanierung eines denkmalgeschützten Speichergebäudes mit aufgesprühter Zellulosedämmung, research report IEA SHC Task 37 – Subtask C, Eigenverlag, Darmstadt Eßmann F. 2011. Balkenköpfe in Außenwänden – Bauschaden durch Innendämmung?, proceedings 13. EIPOS Sachverständigentagen, Eigenverlag, Dresden Gnoth S., Strangfeld P., and Stopp H. 2005. Hygrothermisches Verhalten eingebetteter Holzbalkenköpfe im innengedämmten Außenmauerwerk, article in Bauphysik, Verlag Ernst und Sohn, Berlin Hofmeister G. 1995. Bauschäden an Holzbalkendecken in Feuchtraumbereichen, research report, IRB Verlag, Stuttgart IWU (Institut Wohnen und Umwelt); authors: Zeller, J.; Dorschky, S.; Borsch-Laaks, R.; Feist, W.: Luftdurchlässigkeitsmessungen mit der Blower Door in Niedrigenergiehäusern und anderen Gebäuden, Eigenverlag, Darmstadt 1995 IBA - Internationale Bauausstellung Berlin, Sanierung von Holzbalkendecken, Verlag Ernst und Sohn, Berlin 1985 Kehl D., Plagge R., and Grunewald J. 2012. Wann geht Holz kaputt? – Nachweistechnische Beurteilung von Holz zerstörenden Pilzen, proceedings 23. Hanseatischen Sanierungstagen, Heringsdorf, Fraunhofer IRB Verlag, Stuttgart (ISBN: 978-3-8167-8794-5) Krause K.H., and Ahnert R. 2009. Typische Baukonstruktionen von 1860 bis 1960 zur Beurteilung der vorhandenen Bausubstanz, Band 1-3, 7. Auflage, Verlag Bauwesen, Berlin (ISBN 978-3-410-21133-4) Künzel H., Zirkelbach D., and Schafaczek B. 2012. modelling the effect of air leakage in hygrothermal envelope simulation, proceedings BEST 3 Conference Atlanta Loga T. 2005. Energetische Modernisierung eines Gründerzeithauses in Wiesbaden, proceedings 6. Leipziger Bauschadenstag 2005, MFPA Leipzig GmbH, Eigenverlag, Leipzig Maurice S., Coroller l., Debaets S., Vasseur V., Le Floch G., and Barbier G. 2011. Modelling the effect of temperature, water activity an pH on the growth of Serpula lacrymans, article in Journal of Applied Microbiology (Dez. 2011), Wiley-Verlag (ISSN 1364-5072) DOI: 10.1111/j.13652672.2011.05161.x Müller U. 2011. Holzbalkenköpfe in historischem Mauerwerk – Zwischen Luftumspülung und Innendämmung, proceedings 2. Internationalen Holz[Bau]Physik Kongress in Leipzig - Holzschutz und Bauphysik, Eigenverlag, Aachen Figure 7. relative humidity and temperature at the wooden beam end embedded in masonry with interior insulation and without convection. 9 CONCLUSION AND FUTURE WORK The moisture content of wooden beam ends in masonry walls depends mainly on the protection of wind driven rain. If the outside surface of a wall is protected (e.g. water repellent and vapour open plaster, constructed as a cavity wall) or the wdr load is low (e.g. depending by cardinal direction), decay of wood is not expected. This is demonstrated by measurements in different projects. Another reason for higher moisture content at the wooden beam end is air flow from inside. The amount is currently not well known. Therefore the airflow at the wooden beam end in the masonry wall will be investigated at the laboratory of the Institute of Building Climatology in detail. The growth of decay depends on temperature (like mould growth). Therefore the assessment of measurements and simulations results should take place with the presented simplified engineering model. The simulation with and without convection as well as with and without interior insulation shows the influence on the moisture content of the beam end. 10 ACKNOWLEDGEMENT The research project “Energetisches Bewertungsverfahren für Bestandsgebäude mit Holzbalkendecken” (assessment of existing buildings with wooden beam ceilings) is supported by Germany's Federal Ministry of Economics and Technology (BMWi) in the research program “Energieoptimiertes Bauen (EnoB)” (Research for energyoptimised construction). 303 Peper S., Kaufmann B., Hasper W., Pfluger R. and Ochs F. 2010. Innendämmung und Wandfeuchte, article in Deutschen Bauzeitschrift (DBZ), Bauverlag, Gütersloh Ratzlaff M., and Schnieders J. 2005. Mehrfamilienhäuser in Ludwigshafen, article in Arbeitskreis kostengünstige Passivhäuser, Protokollband 32, passive house institute, Darmstadt Reiher H., Künzel H., Frank W., and Labus H. 1960. Wärmeund Feuchteschutz in Wohnbauten in Beitrag aus der Bauforschung, Verlag Ernst und Sohn, Berlin Ruisinger U. 2012. Innendämmung bei Holzbalkendecken, proceedings at 2. Tagung „Nutzer orientierte Bausanierung“, self publishing, Weimar (ISBN: 978-3-86068-4818) Stade F. 1904. 22. Die Schule des Bautechnikers XIII - Band Holzkonstruktionen, Reprint Verlag Leipzig, Holzminden (ISBN 3-8262-1901-5) Stopp H., Strangfeld P., Toepel T., and Anlauft E. 2010. Messergebnisse und bauphysikalische Lösungsansätze zur Problematik der Holzbalkenköpfe in Außenwänden mit Innendämmung, article in Bauphysik, Verlag Ernst und Sohn, Berlin Viitanen H., Toratti T., Makkonen L., Peuhkuri R., Ojanen T., Ruokolainen L., and Räisänen J. 2010. Towards modelling of decay risk of wooden materials; European Journal of Wood and Wood Products, Springer Verlag, BerlinHeidelberg WTA - Wissenschaftlich Technische Arbeitsgemeinschaft für Bauwerkserhaltung und Denkmalpflege e.V.: Merkblatt 6.5 - Nachweis von Innendämmsystemen mittels numerischer Berechnungsverfahren, Eigenverlag, München 2013 304 A hygrothermal analysis of international timber frame wall assemblies tested under temperate maritime climatic conditions L. Corcoran1, 2, A. Duffy2, 3, S. Rouholamin1, 2 1 Dublin Institute of Technology, School of Architecture, Ireland Dublin Energy Lab, Dublin Institute of Technology, Ireland 3 Dublin Institute of Technology, School of Civil and Building Services Engineering, Ireland 2 ABSTRACT: As the use of timber frame construction increases it is important that the assemblies specified have sufficient drying capacity to withstand any moisture loads that may occur over the life of the structure since decay is heavily dependent on the presence of moisture. The purpose of this study was to assess the hygrothermal performance of common timber frame wall assemblies under temperate maritime climatic conditions. Four timber frame wall assemblies were simulated using WUFI 2D. An additional moisture source was modelled to simulate wall performance under the presence of a construction defect. The results show that under normal conditions all four wall types behave in a similar manner. However once the moisture source is added, relative humidity in two of the wall types exceed thresholds above which the risks of mould growth are high. This effect is the result of the location of the OSB sheathing on the external side of the timber frame, which limits the escape of moisture from the wall assemblies. Wall types where the OSB is on the internal side have greater drying rates due to the increased rate of moisture mass transfer through the wall. This drying capacity is important in humid climates where the probabilities of high moisture loads are greater. 1 INTRODUCTION Hygrothermal analysis models allow realistic calculations of coupled heat, air and moisture transfer for one or multi-dimensional cases. There are many simulation tools available, each with their own level of mathematical sophistication and simplification of input parameters. This type of analysis is critical especially in temperate maritime climate zones where precipitation is more evenly dispersed throughout the year and relative humidities tend to be higher for longer periods of time. A survey was undertaken by the Wood Marketing Federation (2004) to increase the understanding of the timber frame construction sector in Ireland. The survey shows a steady rise in completed timber frame housing units from 1% in 1990 to 25% in 2004. With this increase in timber frame construction it is crucial that the most appropriate construction details and materials are specified to provide sufficient moisture tolerance over the life span of the building. The climate to which the building fabric will be exposed will have a significant impact on the long term durability of the construction. The thermal performance of the building envelope can also be significantly influenced by the interaction of moisture with building materials and components (Hagentoft, 2001). With timber, the potential for decay is heavily dependent on the presence of moisture (TRADA, 2008). Timber that is exposed to moist or high relative humidity levels for extended periods of time becomes susceptible to mould growth and fungal decay (Canadian Wood Council, 2004). In order to reduce the possibility of moisture related problems, designers must assess the possible moisture loads which may or may not occur over the lifetime of the building. This can be achieved through laboratory experiments, field tests and using hygrothermal simulation tools. Laboratory experiments and field tests can be expensive and are time consuming, for this reason the use of hygrothermal simulation tools is increasing. 2 AIMS OF THIS STUDY The aims of this study are: • To identify timber frame construction assemblies that are most commonly used on an international platform. • To use hygrothermal simulation software to assess how each wall assembly performs when exposed to temperate maritime conditions over a period of time. • To assess the drying capacity of each identified wall assembly by simulating an additional moisture load which may occur over the lifetime of the structure. 305 laboratory tests, field tests and simulation using computer software. Laboratory tests are the most difficult to set up. In order to test the impact of varying indoor and outdoor conditions on the assemblies, these conditions must be created on either side of the wall. Numerous measurement and control instruments are required in the setup of these experiments. These instruments are controlled and managed using a computer measurement and control system with a fitting control program. A detailed description of a typical experimental setup can be seen in Vinha (2007). Field tests involve monitoring the performance of actual building components under real conditions. Common sensors used in field tests include temperature, relative humidity, moisture content and air flow. These sensors log data at specified intervals. Examples of field tests can be found in (Geving and Uvslokk, 2000; Vinha, 2007; Gatland II et al., 2007). Many hygrothermal simulation tools have been developed over the last two decades. There are over 45 hygrothermal simulation tools in existence, each with their own level of mathematical complexity. These tools allow calculations that examine the thermal and moisture performance of building components relative to steady state or transient realistic boundary conditions. Calculations can predict how assemblies will perform over extended periods of time when exposed to natural climatic conditions. 3 LITERATURE REVIEW 3.1 Overview of common timber frame construction methods. There are many systems of timber frame construction available today, all of which serve different requirements and have their own advantages and disadvantages, these include; • Balloon frame construction • Platform frame construction • Open panel platform frame construction • Closed panel platform frame construction • Volumetric construction • Post and beam construction • Structural insulated panels. For the purpose of this research, the focus will be on the platform construction method. In platform construction the timber frame panels are floor to ceiling height, with the upper floors bearing on the panels or foundation below. The structural timber frame is braced as construction proceeds, allowing the walls and internal partitions of the next floor to be constructed in an efficient manner. Studies and reports by (Teasdale-St-Hilaire and Derome, 2007; Gatland II et al., 2007; Vinha, 2007; Geving and Uvslokk, 2000; Piot et al., 2008) show this method of timber construction as the typical representation from their respective geographic locations. According to a report compiled by the Irish Timber Frame Housing Consortium (2003) this method of timber frame construction accounts for 90% of timber houses built in Ireland. The principles of platform timber frame construction methods are similar internationally. External wall details consist of a number of common elements including; • external cladding • a ventilated cavity • breather membrane • sheathing board • timber frame with insulation • vapour control layer • plasterboard with an internal finish. Details that vary in timber frame wall constructions include; • type, grade and dimensions of materials, • presence, position and orientation of materials (Timber Frame Housing Consortium, 2003). This paper analyses the most common timber frame wall assemblies and compares their hygrothermal behaviour relative to temperate maritime climatic conditions. 4 REVIEW OF SIMULATION TOOLS 4.1 Overview hygrothermal simulation tools A report by Hens (1996) showed that 37 programs had been developed for HAM (Heat, Air and Moisture) analysis. A review of hygrothermal models suitable for building envelope retrofit analysis was conducted by the Canadian Mortgage and Housing Corporation (2003). This review identified 45 hygrothermal simulation tools. Delgado et al.(2013) expanded this figure to 57 hygrothermal modelling tools. This list was reduced to 14 hygrothermal models based on their availability to the general public. A critical review of these simulation tools and their level of complexity can be read in their book: Hygrothermal Numerical Simulation Tools Applied to Building Physics (Delgado et al., 2013). 4.2 WUFI (Wärme und Feuchte instationär Transient Heat and Moisture) 3.2 Overview of testing methods There are three main methods for testing the hygrothermal performance of building components: WUFI is a software family that allows realistic calculation of transient coupled one and two dimensional heat and moisture transport in multi306 layer building components exposed to natural climatic conditions (WUFI, 2012). Both one and two dimensional simulation tools have been validated numerous times using data derived from field and laboratory tests (Kunzel, 1995). In terms of heat transfer, WUFI takes into account the following: • Thermal conduction, • Enthalpy flows through moisture movement with phase change, • Short wave solar radiation and • Long wave radiation cooling. Vapour phase transfer is by vapour and solution diffusion. The liquid transport mechanisms taken into account are capillary conduction and surface diffusion. Convective heat and mass transfer is disregarded in the model. The governing equations for moisture and energy transfer in WUFI are, ∂w ∂ϕ = ∇(Dϕ ∇ϕ + δ p ∇(ϕPsat )) ∂ϕ ∂t (1) ∂H ∂T = ∇(λ∇T ) + hv ∇(δ p ∇(ϕPsat )) ∂T ∂t (2) The governing equations for moisture mass balance, air mass balance, salt mass balance and internal energy balance are, ∂ ( p wθ l + p vθ g ) = ∂t ∂ ( p w / v − j disp − j diff )θ l + ( p v / v + j diff )θ g − ∂x [ [ ∂ ( p aθ g ) = − ∂ ( p a / v − j diff )θ g ∂t ∂x ] ] (3) (4) [ ∂ ( p sθ l + p pθ p ) = − ∂ ( p s / v + j disp + j diff )θ l ∂t ∂x [ ] (5) ] ∂ pc pT + p p c ppTθ p + pl c plTθl + ( pv c pvT + pa c paT )θ g ∂t ∂ (6) pl c plT / vθl + ( pv c pvT + pa c paT ) / vθ g =− ∂x ∂  ∂T  + (hs − hw )( jdisp + jdiff )θl + (hv − ha ) jdiff θ g  − − λ ∂x  ∂x  [ ] Where pw is the liquid moisture partial density (kg/m3), θl is the volumetric content of the liquid phase (m3/ m3), pv is the mass density of water vapour (kg/m3), θg is the volumetric content of the gaseous phase (m3/ m3), v is the humidity by volume in the surrounding air (kg/m3), jdisp is the dispersive flux, jdiff is the diffusive flux, pa is the air partial density (kg/m3), pp is the mass density of the precipitated salt (kg/m3), ps is the mass density of the dissolved salt (kg/m3), θp is the volumetric content of the precipitated salt (m3/ m3), p is the density of the dry porous material (kg/m3), cp is the heat capacity of the solid material (J/kg K), cpp is the heat capacity of the precipitated salt (J/kg K), cpl is the heat capacity of the liquid water (J/kg K), cpv is the heat capacity of the water vapour (J/kg K), cpa is the heat capacity of dry air (J/kg K), λ is the thermal conductivity (W/mK), hv is the partial specific enthalpies of water vapour (J/kg K), hs is the partial specific enthalpies of salt (J/kg), hw is the partial specific enthalpies of water (J/kg) and ha is the partial specific enthalpies of air (J/kg). Where jH/jT is the heat storage capacity of the moist building material (J/kg), jw/jϕ is the moisture storage capacity (kg/m3), λ is the thermal conductivity of the moist building material (W/mK), Dφ is the liquid conduction coefficient of the building material (kg/ms), δp is the water vapour permeability of the building material (kg/msPa), hv is the evaporation enthalpy of the water (J/kg), Psat is the water vapour saturation pressure (Pa), T is temperature (K), φ is relative humidity (%). 4.3 Delphin Delphin is a commercial program used to analyse one and two dimensional transport of heat, air, moisture, pollutant and salt transport in porous building materials and assemblies (Delphin, 2008). Delphin has been validated several times with focus on different aspects of the software (Scheffler, 2008). A large number of variables can be obtained as functions of space and time including: • Moisture contents • Temperatures • Air pressures • Salt concentrations • Diffusive and advective fluxes of liquid water • Water vapour • Air • Heat • Enthalpy 5 METHODOLOGY 5.1 Methodology overview Four typical wall assemblies were chosen for simulation using WUFI in a temperate climate. WUFI was chosen because it has been validated in the field and laboratory more than other hygrothermal simulation tools. Before the main simulations were carried out, output parameters for one configuration were verified by simulating an identical model in Delphin. The wall assemblies were then tested to assess their performance in the chosen temperate climate. 307 Figure 1. Walls for simulation. *NOTE: Wall Type C has the same assembly as Type A – the difference is in the insulation material. applied in the model. A constant rate of air changes per hour is deemed satisfactory over the long term behaviour of the construction as demonstrated by Kehl et al.(2009) and also through a research project by Mayer and Künzel (cited in Kehl et al.). The figure of 25 ac/h was used based on validation tests by Karagiozis and Kuenzel, (2009) and suggested values by Kehl et al. (2009). 5.2 Choice of wall assemblies As mentioned in 3.1 timber frame external wall assemblies are very similar throughout the world. The main variables are: • type, grade and dimensions of materials • presence, position and orientation of materials. The following wall structures were chosen for this study (see Fig.1): 5.2.3 Breather membrane It is common practice to wrap the outer face of the timber frame with a breather membrane. The function of this membrane is to protect the timber frame from weather until the cladding is completed, and to provide a second line of defence against any wind driven rain that may penetrate the cladding. Breather membranes must combine a high degree of water resistance with low vapour resistance to allow any moisture that may become trapped behind to pass (or breathe) through. Some variations of timber frame assemblies do not require a breather membrane (wall type D). Wall Type A: 100mm Brick outer leaf, 50mm ventilated cavity, breather membrane, 11mm OSB, 140mm mineral wool, vapour control layer, 12.5mm plasterboard. Wall Type B: 100mm Brick outer leaf, 50mm ventilated cavity, breather membrane, 140mm mineral wool, 11mm OSB, vapour control layer, 12.5mm plasterboard. Wall Type C: 100mm Brick outer leaf, 50mm ventilated cavity, breather membrane, 11mm OSB, 140mm cellulose, vapour control layer, 12.5mm plasterboard. Wall Type D: 100mm Brick outer leaf, 50mm ventilated cavity, 35mm softwood fibreboard, 140mm wood fibre insulation, 15mm OSB, 25mm service void, 12.5mm plasterboard. A description of each component can be found below. 5.2.4 Sheathing The primary function of the sheathing is to provide the necessary panel stiffness to resist lateral forces due to wind loads. The most common materials used as sheating boards are glue bound boards such as plywood and OSB (Oriented strand board). Variations of timber frame walls include the use of more vapour permeable boards such as softwood fibre boards on the external side of the timber frame, moving the position of the OSB to the internal side (wall type D). 5.2.1 External Cladding The type of external cladding used in the simulations was brick, based on its popularity in temperate maritime climates. Other common cladding materials include timber and render systems. 5.2.5 Timber Studs With timber frame construction the depth of the studs usually dictates the thermal performance of the wall. Sizes for structural timber can be found in BS EN 336 – Structural timber. Sizes – permitted deviations. 140mm x 38mm studs @ 600mm centres will be selected as this reflects common practice (TRADA, 2008). 5.2.2 Ventilated Cavity A ventilated cavity of 50mm was simulated based on common practice and recommendations from TRADA (2008). A ventilation rate of 25ac/h was 308 5.2.6 Thermal Insulation The most common type of thermal insulation used in timber frame construction is mineral wool (glass or rock) (TRADA, 2008). This is generally in the form of semi rigid batts. Other insulation types include cellulose, hemp, phenolic, polyurethane, and wood fibre boards. 5.4.2 Computational Grid A fine numerical grid was selected to increase the accuracy of the calculations. The grid is set to increase in frequency toward component boundaries (see Fig.2.) and boundaries of different materials as steeper temperature and moisture profiles are to be expected which require an increase in grid element numbers. 5.2.7 Vapour control layer A vapour control layer is typically installed on the warm side of the insulation to control the flow of vapour from the internal space through the wall assembly. A typical vapour control layer consists of a 500 gauge 125 micron polythene sheet with an sd value of 50m (Wood Marketing Federation, 2007). For this study typical values will be used. 5.2.8 Internal lining and finish The internal finish typically comprises of 12.5mm plasterboard with a skim finish. A double layer of plasterboard may be necessary if a certain level fire resistance is required. For the purpose of this study one layer of 12.5mm plasterboard will be simulated. 5.3 Choice of climate data The study is concerned with temperate maritime climatic conditions. A maritime climate (also known as oceanic, marine and west coast) does not experience extreme temperature differences throughout the year. Summers are described as warm (not hot) and winters cool (not cold). Typical maritime climates lack a dry season as precipitation is more evenly dispersed throughout the year. Ireland and the UK are examples of countries with typical temperate maritime climates. Other countries that experience this type of climate (in parts) include: Norway, France, Denmark, Germany and Belgium. For the purpose of this study, an east coast Irish climate will be used as a typical representation of a temperate maritime climate. Climate data was derived from Meteonorm (2013) using Dublin airport as the weather station. A design reference year was created to give hourly values representing the most severe conditions likely to occur once every ten years. This complies with guidance set out in EN 15026 (BSI, 2007b). Figure 2. Computational grid used in WUFI. 5.4.3 Materials WUFI’s material database contains data for over 600 materials. Much of this data has been derived from laboratory tests or from calculated values and literature. European standards EN 12524 (BSI, 2000) and EN ISO 10456 (BSI, 2007a) contain tabulated data for hygrothermal properties for a number of building materials. A review of the default WUFI materials necessary for the simulations found that much of the data matched or was very close to the values recommended in EN 12524 (BSI, 2000) and EN 10456 (BSI, 2000). For this reason the WUFI database was used to provide material data for the simulations. 5.4.4 Initial Conditions Initial conditions are specified for each material in the model. Conditions for water content (kg/m3), relative humidity (%) and temperature (°C) are specified. Thermal conditions in the building component adapt very quickly to their surrounding conditions, usually within a few hours. An initial value of 80% relative humidity will be used in the simulations to examine the initial drying capacity of the materials. 5.4 Simulation Set up The following section describes the setup of the simulations. 5.4.1 Geometry The geometry used in the simulation was defined as a 600mm cross section. The width of the timber studs at each end is 19mm, half of the typical 38mm stud. The geometry was simplified to represent a typical symmetrical section from the wall assembly (see Fig.1.). 5.4.5 Boundary Conditions As described in 5.3, external conditions are representative of Dublin, Ireland. A design reference year representing the most severe conditions likely 309 8760 hours (1 year) using the same boundary conditions as in 5.4.5. One year was chosen as the time period for the verification test as both models were computationally intensive. Temperature and relative humidity values were simulated for the inside face of the OSB board and the results recorded for each time step. Temperature and RH time-series results for both simulations are shown in Figures 3 and 4 respectively. It was found that simulated temperature differences in the models ranged between 0.5°C and 3°C. Relative humidity values were more sensitive and these ranged from 0% to 5.9%. Although material parameters we made as close as possible for Delphin and WUFI, slight differences remained. Small variations in properties such as moisture storage functions may account for the differences in relative humidity profiles. A mean absolute percentage error (MAPE) analysis was performed to analyse the difference between the two sets of values for relative humidity. The MAPE analysis was performed for all 8760 time steps. The hourly MAPE (see Fig.5.) does not exceed 6.5% over the 12 month period and averages 2.1% over the 12 month period. This figure was deemed acceptable but lower errors may be obtained with further refinement of material properties. to occur once every ten years is used on the external side of the wall. Internal conditions are created as per EN 15026 (BSI, 2007b). Interior temperature and relative humidity levels are determined from the outdoor conditions. Details on this can be found in EN 15026 (BSI, 2007b) and in the WUFI Pro manual available at www.wufi-pro.com. Both symmetry axis are treated as adiabatic i.e. impermeable to heat and moisture exchange. 5.4.6 Sources An air change rate of 25 air changes per hour is assigned to the ventilated cavity (see 5.2.2). An additional moisture source equivalent to 1% of the wind driven rain is used to assess the drying capacity of each wall type and to mimic the effects of incomplete sealing of the breather membrane/OSB joints. This moisture source was applied to the outer 5mm of the insulation/stud zone. 5.4.7 Computational Parameters The simulation begins January 1st, the year is not important as a design reference year has been specified. The duration of the simulation is 26280 hours (3 years) using hourly time steps. All calculations include heat of evaporation, heat of fusion and capillary conduction. 5.4.8 Critical locations to be analysed Two points were chosen as critical locations to monitor temperature and relative humidity levels in the wall assembly (see Fig.1.). These points were chosen as they represent the most critical points where condensation may occur (BSI, 2011). As heat, air and moisture move through the wall assembly there is a chance that condensation may occur on the cold side of the insulation. This risk is increased when a board such as OSB with a high vapour resistance is positioned in this location. Similarly in the summer seasons when water vapour diffusion may travel in the opposite direction as a result of temperature differences, the critical point for analysis becomes the internal side of the insulation. This study will analyse both locations over the chosen 3 year period. Figure 3. WUFI and Delphin temperature profiles compared. 6 MODEL VERIFICATION 6.1 Overview of verification method Two hygrothermal models were used for verification purposes. Delphin (2008) was used to verify the WUFI (2012) model. The wall assembly used for verification was the equivalent of wall type A (see Fig.1.). It was decided that the material properties from WUFI’s database were to be used, based on the review in 5.4.3. The duration of the simulation was Figure 4. WUFI and Delphin relative humidity profiles compared. 310 It was decided to further analyse the drying capacity of the selected walls by simulating a moisture source equivalent to 1% of the wind driven rain deposited behind the OSB/breather membrane (the outer layer of the insulation). WUFI estimates wind driven rain using normal rain loads, wind velocity and driving rain coefficients (based on the location and orientation of the building element). ASHRAE 160P states: “In the absence of specific full scale test methods and data for the as-built exterior wall system being considered, the default value for water penetration through the exterior surface is 1% of the water reaching that exterior surface. The deposit site for the water shall be the exterior surface of the waterresistive barrier. If a water-resistive barrier is not provided then the deposit site shall be described and a technical rationale for its selection shall be provided.”(BSR/ASHRAE, 2008) The rationale behind this was to assess the drying capacity of each wall type and to mimic the effects of incomplete sealing of the breather membrane/OSB joints. This location is common to all wall types and so choosing this allowed changes in the performance of each wall type to be directly comparable. Figure 5. MAPE analysis for relative humidity. 7 RESULTS The initial results for analysis point ‘a’ show similar trends in relative humidity profiles (Fig. 6.) among the walls. Wall type D shows the lowest relative humidity values and therefore has the greatest rate of moisture mass transfer resulting in faster drying and lower risks of mould growth and interstitial condensation problems. Wall type A also performs well with relative humidity levels seldom greater than 81%. Wall types B and C experience the highest levels of relative humidity which reach 90% for small periods of time. Figure 6. Analysis point a (external side of insulation zone). Figure 8. Analysis point a after additional 1% wind driven rain moisture load. Figure.7 below shows relative humidity values for analysis point ‘b’. Overall, wall type B shows the lowest relative humidity values. All wall types are within ± 3% of each other. Relative humidity levels are seldom above 70% at this point. As a general rule, mould spores can germinate if the relative humidity at the surface exceeds 80% (BSI, 2011). Figure 7. Analysis point b (internal side of insulation zone). Figure 8. shows the relative humidity profiles of each wall at analysis point ‘a’ after the additional moisture load of 1% of the wind driven rain is added to the model. It can be seen that relative humidity levels in wall types A and C increase significantly while wall types B and D remain below the 80% threshold for the majority of the simulation time. It is evident from this that mass transfer is most efficient for wall types B and D resulting in faster drying. 311 dresden.de/delphin/index.php?aLa=en/ [accessed 16th December 2012] Gatland II S.D., Karagiozis A.N., Murray C., and Ueno K. 2007. The Hygrothermal Performance of Wood-Framed Wall Systems Using a Relative Humidity-Dependent Vapour Retarder in the Pacific Northwest. Thermal Performance of the Exterior Envelopes of Buildings X. Clearwater, Florida. 2-7 December. Geving S., and Uvslokk S. 2000. Moisture conditions in timber frame roof and wall structures: Test house measurements for verification of heat, air and moisture transfer models. Project Report 273. Norwegian Building Research Institute. Hagentoft C.E. 2001. Introduction to Building Physics, Lund: Studentlitteratur. Hens H. 1996. Heat Air and Moisture Transfer in Insulated Envelope Parts. Final report, 1, 90. Karagiozis A.N., and Kunzel H.M. 2009. The Effect of Air Cavity Convection on the Wetting and Drying Behaviour of Wood-Frame Walls Using a Multi-Physics Approach. Journal of ASTM International, 6, 15. Kehl D., Hauswirth S., Weber H., and HTL A. 2009. Is ventilation of timber facades essential: Research report. University of applied science, Switzerland. Kunzel H.M. 1995. Simultaneous Heat and Moisture Transport in Building Components. IRB-Verlag. Meteonorm. 2013. Software and Data for Engineers, Planners and Education: [Online] http://meteonorm.com/ [accessed 12th February 2013]. Piot A., Woloszyn M., and Brau J. 2008. Numerical Simulation Aided Design of an experimental protocol. Proceedings of the 8th Symposium on Building Physics in the Nordic Countries. Copenhagen, Denmark. 731-738. Scheffler G.A. 2008. Validation of hygrothermal material modelling under consideration of the hysteresis of moisture storage. Ph.D. Thesis, Dresden University of Technology, Dresden, Germany. Teasdale-St-Hilaire A., and Derome D. 2007. Comparison of experimental and numerical results of wood-frame wall assemblies wetted by simulated wind-driven rain infiltration. Energy and Buildings, 39, 1131-1139. Timber Frame Housing Consortium. 2003. Timber Frame Housing Report. [online] http://www.environ.ie/en/DevelopmentHousing/BuildingSt andards/TimberFrameHousing/ [accessed 12th November 2012]. TRADA 2008. Timber Frame Construction, Buckinghamshire, England, TRADA Technology Ltd. Wood Marketing Federation. 2004. Survey of Irish Timber Frame Manufacturers 2004 [online] http://www.coillte.ie/fileadmin/templates/pdfs/Timber_Fra me_Report%201-11.pdf [accessed 19th February 2013]. Wood Marketing Federation. 2007. Woodspec - A guide to designing, detailing and specifying timber in Ireland. Wicklow: Wood Marketing Federation. WUFI. 2012 WUFI Pro 2D Plus - Software for calculating the coupled heat and moisture transfer in building components. [online] http://www.wufi-pro.com/ [accessed 25th February 2013]. Vinha J. 2007. Hygrothermal Performance of Timber-Framed Walls in Finnish Climatic Conditions: A method for determining the sufficient water vapour resistance of the interior lining of a wall assembly. Ph.D. Tampere University of Technology, Finland. 8 CONCLUSION Two dimensional hygrothermal simulations were performed for four common variations of timber frame wall assemblies in typical temperate maritime climatic conditions. Initial results showed all four wall types have similar relative humidity profiles. Relative humidity levels fluctuate between 75 and 85% on the external side of the insulation (analysis point a). An additional moisture load was introduced on the external side of the stud/insulation zone to assess the drying capacity of each wall type and to mimic the effects of incomplete sealing of the breather membrane. Results show that under this additional moisture load, relative humidity conditions in wall types A and C increase to levels where mould growth is probable. However, wall types B and D reach stable conditions of relative humidity, seldom rising above the 80% threshold for mould growth. The wall types with the greatest drying capacity under the additional moisture loads were B and D. These wall assemblies both had OSB sheathing on the internal side, resulting in a greater level of moisture diffusion toward the ventilated cavity. Wall types A and C both had OSB sheathing on the external side, limiting the rate of moisture diffusion and mass transfer, thus reducing the drying capacity of the walls. This has important implications for existing, widely deployed assembly details which incorporate a highly vapour resistive material such as OSB on the external side of the stud. 9 REFERENCES British Standards Institution 2000. EN 12524: Building materials and products - Hygrothermal properties Tabulated design values. London: BSI. British Standards Institution. 2007a. EN 10456 - Building materials and products - Hygrothermal properties Tabulated design values and procedures for determining declared and design thermal values. London: BSI. British Standards Institution. 2007b. EN 15026 - Hygrothermal performance of building components and building elements - Assessment of moisture transfer by numerical simulation. London: BSI. British Standards Institution 2011. BS 5250: Code of practice for control of condensation in buildings. London: BSI. BSR/ASHRAE 2008. Standard 160P, Criteria for moisture control design analysis in buildings. Atlanta Canadian Mortgage and Housing Corporation. 2003. Review of hygrothermal models for building envelope retrofit analysis. Research Highlights. Technical Series 03-128. Ottawa: Canadian Mortgage and Housing Corporation. Canadian Wood Council. 2004. Managing Moisture and Wood. Ottawa: Canadian Wood Council. Delgado J.M.P.Q., Barreira E., Ramos N.M. M. and Freitas V.P.D. 2013. Hygrothermal Numerical Simulation Tools Applied to Building Physics. Springer. Delphin. 2008. Delphin Simulation program for the calculation of coupled heat, moisture, air, pollutant, and salt transport [online]. http://bauklimatik- 312 Long-term measurements and simulations of five internal insulation systems and their impact on wooden beam heads U. Ruisinger Graz University of Technology, Institute of Building Construction, Graz, Austria ABSTRACT: The project presented examines to what extent subsequently applied internal insulation can damage wooden beam heads. In a test house, five different internal insulation systems have been monitored by sensors that register heat and moisture conditions in and around the beam heads as well as in adjacent rooms and the outer climate. After three complete winter periods with different moisture loads from adjacent rooms a different drying and long-term reaction has become visible. Inside the wooden beams no critical moisture has been detected. The beam heads piercing the internal insulation form a three-dimensional constructional detail that cannot be calculated by purchasable 3D-software without putting huge effort into the generation of own models. Experiences in calculating these details with a two-dimensional software tool are also commented on. by micro biological growth. In 1885/86 the building was erected in neo-renaissance style as sanatorium, nowadays it is used as kindergarten. The existing construction consists of masonry which is thick 550 mm, plastered on both sides and south-west oriented. Inside and outside plaster as well as bricks have been measured and characterised by material functions at the Building Physical Research & Development Laboratory at the TU Dresden. It was detected that the bricks are extraordinarily porous. The insulation systems are: − 80 mm spray-on cellulose with 13 mm limecement plaster, − 80 mm perlite board with 6 mm adhesive mortar and 4 mm surface filler, − 120 mm insulation plaster (insulation aggregate: thermosilit) applied in two layers with 4 mm lime plaster covering, − 60 mm wood fibre insulation board with 4 mm clay adhesive mortar and 13 mm clay plaster and − 100 mm reed with 4 mm clay adhesive mortar and 20 mm clay plaster, applied in two layers. 1 INTRODUCTION Often historical buildings cannot meet modern demands: due to low temperature comfort, high heat energy losses next to diverse time-dependent damages they become more and more unattractive as property and for renting. To compete with recently constructed buildings and to reduce climate damaging emissions as well as heat energy costs historic buildings are insulated increasingly. Listed buildings with worth-keeping façades can be insulated inside only, so that the wooden beams of the ceilings pierce the plane of the internal insulation. In comparison with historic conditions, after refurbishment measures the wooden beam heads are located in an area that has become colder and more humid by the internal insulation. It is not known to what extent this change affects wooden beams at all. But the insecurity of this situation leaves planners and clients helpless. For this reason, it is often not taken advantage of the fully riskless potential of internal insulation or it is not insulated at all. Table 1. Hygrothermal material and insulation systems properties. Insulation Thermal Resissd–value Water uptake material tance sd,IS* in m coefficient** RIS* in m²K/W in kg/m²s0.5 Cellulose 1.59 0.27 1.20 Perlite 1.79 0.63 1.98 Insul. plaster 1.61 1.02 0.15 Wood fibre 1.49 0.49 0.01 Reed 1.70 0.34 < 0.01 * value for the whole insulation system (IS) with glue and covering ** water uptake coefficient of the insulation material 2 MEASUREMENTS IN A TEST HOUSE 2.1 Test house and insulation systems Currently, within the framework of the project OEKO-ID in Schönbrunngasse 30 in Graz five different insulation systems are examined. All of them are vapour diffusion permeable and partly described in companies’ self-portrayals “ecological”. For three full condensation and two drying periods the wooden beam heads have been examined with respect to the endangering of the building substance 313 beam, the fifth field with reed insulation also contains two wooden beams. As reference case, a wooden beam in a field without any insulation is monitored, as well. The treatment of the air gap between the bearing edge and the wooden beams is still a field of discussion. Two different ways of execution were carried out. At six of the eight wooden beams, the lower part of the bearings underneath the bulk material were sealed very carefully with plaster and sealing tape. Wood cracks were filled with wood plugs. In doing so, no air from the ceiling area between the beams could intrude into the air layer around the wooden beam in the bearings. At the reference beam and one beam insulated with reed no sealing measures were carried out. Above the bulk material, the air gap between wooden beams and brick wall at the beginning of the bearing were not sealed. So, the diffusion resistance between attic floor and bearings is comparably low. Table 1 shows hygrothermal characteristic properties of the insulation systems and the water uptake coefficients of the insulation materials. The insulation systems reduce the U-value of the entire construction of approximately 0.71 to 0.33 W/m²K. The insulation layers between both floors are interrupted by a bulk material mix on thin wooden boards between the beams consisting of historic building site waste and ashes (Figs 2, 3). 2.2 Experimental set-up Combined temperature and humidity sensors measure the internal and external climate. Surface temperature sensors and a probe for the global radiation complete the visible measuring devices (Fig. 3). The air gaps around the wooden beams are monitored by two temperature/humidity sensors in two different depths each. Other pairs of combined probes measure temperature and relative humidity on the cold side of the insulation systems, approx. 200 mm underneath the bottom side of the wooden beam. Two long sensors pierce every beam from the bottom side in the insulation plane angled upwards till the end grain (see Fig. 3 left). Furthermore, at two beams, insulated with cellulose and the non-insulated reference beam, the wood moisture is measured additionally in the plane of the insulation systems in two depths. Figure 1. View on external wall with five insulation systems and measurement technique (rear left) in the attic; insulation systems from left to right: cellulose (corner), perlite, insulation plaster, wood fibre board, reed. Figure 2. Section through the ceiling with view of the test area; distance between beam axes approx. 0.93 m. The eight monitored wooden beams are situated between a playroom of the kindergarten in the first floor and a room in the attic which is used as storage and where measurement technique is placed (Figs 1, 2). Each of the five insulation fields contains one wooden beam at least. The first field with cellulose insulation additionally contains the trimmer Figure 3. Sensors arrangement: vertical sections through external wall between two beams (left) and through wooden beam with view on the historic internal surface (right). 314 possibilities and limits of insulation systems this low moisture load was contrary to project aims. Therefore, the windows in the upper room were sealed and sufficiently dimensioned humidifiers in both rooms were installed in autumn 2011. The measuring of wood moisture by electric resistance depends on temperature. The temperature probes for the correction are located at the lower beginning of the moisture sensor, in the plane of the internal insulation. It is not clear where the long wood moisture sensors exactly measure. To be on the safe side, it was assumed that it is measured near the end grain. With the help of three-dimensional thermal simulations, all measured wood moisture data underwent temperature correction. Table 2. Mean values of room temperature, relative and absolute room humidities during winter months. Period First floor Attic floor [°C , % , g/m³] [°C , % , g/m³] Dec 2010 – Mar 11 23.4 , 21 2 , 4.4 18.6 , 33 , 5.2 Dec 2011 – Mar 12 22.2 , 33 3 , 6.5 20.3 , 37 , 6.6 Dec 2012 – Mar 13 21.0 1.0 , 42 , 7.7 19.6 , 57 , 9.7 2.3 Room climate measurements Courses of measured climate temperatures and humidities from Figure 4 are summarised in Table 2. Due to high room temperatures and unfavourable conditions in the building, during the first condensation period extremely low room humidities were discovered. As one aim of the project is to find out the In the following condensation period 2011/12 humidities in both rooms increased but not to the desired extent. After additional sealing measures, especially an additional well-locking internal door in the attic, high moisture loads by an average of 42 and 57 % could be reached finally. Figure 4. Daily mean values of 1st floor room temperature and room humidities as well as outside temperature at test house Schönbrunngasse 30 (parts of the data provided by the meteorological station of the Karl-Franzens University of Graz). Figure 5. Daily mean values of relative humidity on external resp. cold side of insulation systems. 315 moisture (above 95 % rel. humidity). Behind cellulose, wood fibre boards and with exceptions behind reed mats high moisture loads were detected. Transverse to grain direction, reed insulation nearly has no water conductivity and the vapour diffusion resistance is very low. Moisture from rooms can diffuse through the insulation and condensate at the external reed insulation side. Therefore, this system does not correspond to the state of the technology. Under these circumstances the moisture load on the cold side of the insulation was lower than expected. This favourable material behaviour has already been reported in another project (Wegerer & Bednar 2011). Unfortunately, both humidity sensors behind the insulation plaster broke down, probably due to long continuing high moisture. From the high course of wood moisture near the end grain (Fig. 7) and a manual measurement in November 2011 it can be concluded that the drying out process of the insulation plaster system wore on more than one year until second winter (see chapter 2.5). 2.4 Internal insulation systems performance Figure 5 shows the course of the relative humidity between former internal room surface and internal insulations system. In dependency of the amount of build-in moisture and hygric transport parameters, the time period for the drying-out process varied. At the monitored positions, 200 mm underneath the beams, reed, wood fiber insulation board and perlite board dried out very quickly. Behind the cellulose insulation of the trimmer beam, in the corner of the room (continuous black line), it took more than half a year until May 2011. The delayed application of internal plaster layers on the wood fibre board and the cellulose led to anew increases of moisture in these systems in December 2010. During the second cold period, liquid water formed behind the cellulose only. Between December 2012 and the middle of February 2013 the average rel. humidity in the lower room was 44 % rel. humidity. During this period as single insulation system the perlite boards prevented overhygroscopic Figure 6. Daily mean values of relative humidity at external end of bearings. Figure 6 reveals the effects of sealing the wooden beam bearings. With the aid of humidifiers, sealing measures in the rooms and the cooperation of kindergarten teachers, the relative humidity in the lower room was high enough to demonstrate these effects: during hibernal climate conditions and high humidity in the lower room the relative humidity in the two bearings not sealed (full lines in Fig. 6) take higher courses than in the sealed bearings (dash lines). Both graphs of the not sealed bearings react more quickly on climatic influences and therefore move more lively. In January and February 2012 and starting from November 2012 the relative humidity in the lower room increased. The moist air diffused through the ceiling covering that consists of lime 2.5 Hygrothermal behaviour in the bearing Time and time again liquid water at the boundary layer between internal insulation system and historic construction occurred (Fig. 5). Nevertheless, in the air gap between the end grain and the external side of the bearing overhygroscopic water formed much rarer (Fig. 6) in spite of lower temperatures in the bearing. Two reasons are responsible for this: the thermal conductivity of wood is approx. two to three times higher than of the insulation materials. So, the wooden beams act as thermal bridges and warm their immediate surroundings. The second reason is the sealing of the bearing towards the room in the first floor that avoids convective moisture flow. 316 The second combined temperature and relative humidity probe in the bearing was situated approx. in the middle of the depth (150 mm from the inner historic wall plaster surface, not documented here). In this area a relative humidity of 80 % was barely exceeded, and if yes then only on few days. In combination with Figure 6, this suggests a distinct horizontal decline of potential moisture damages in the bearings. A climate station of the Zentralanstalt für Meteorologie und Geodynamik (ZAMG) meteorological network is located at the University of Graz, approx. one kilometre away from the test house. By means of measured rain data from this station it could be found out that wind-driven rain has no significant influence on the measurements. plaster on wooden boarding and a single-layered reed mat as lathwork. In both bearings without sealing, beam seven in the field with reed insulation and beam eight in the field without insulation, the humidity courses run higher significantly. The open bearings at beams seven and eight revealed slightly higher temperature (< 2 K), as the air flux from the area between the wooden beams also transported energy, but the higher moisture storage capacity of the warmer air was not enough to reduce humidity. Comparing the situation at beams six and seven, both with identical insulation, it becomes visible that the moisture at beam six also rises with the time but much more unhurriedly. With higher external temperatures and especially with lower room humidities, the courses equalled, see end February 2012 in Figure 6. Figure 7. Measured daily mean values of wood moisture near the end grain of the wooden beams. buffer with respect to possible wood damages. All sensors have been calibrated several times during the measurements. While the wood moisture in nearly all beams stayed relatively dry in 2010 and 2011 the course of beam four with insulation plaster runs much higher than all others. This observation led to the assumption that the drying out process of the insulation plaster lasted until the second winter. Apart from the high built-in moisture it is assumed that the reasons for the delayed drying out are a higher diffusion resistance of the plaster itself and a dense coating between the two insulation plaster layers of 60 mm thickness each. From 2012 onward the wood moisture course has proceeded at an inconspicuous level. 2.6 Wood moisture Despite not minor insulation thickness resp. thermal resistance (Table 1), from Figure 7 it can be seen that the graphs of wood moisture stayed relatively low. As most courses run on a comparable height it is difficult to distinguish individual courses in Figure 7. The maximum wood moisture added up to 17 M-% (after temperature correction 17.5 M-%). After the drying out process in the second winter, the maximum moisture in beam one was 15.5 M-% (after temperature correction 16 M%). Actual guidelines or research reports (DIN 68800, Viitanen et al. 2010) state that these values are far from possible danger of wood damages. In DIN 68800, the limiting value of 20 M-% may be exceeded for maximum three months. Using a more precise model of prediction of wood degradation (Viitanen et al. 2010) higher wood moisture for longer periods are possible. Hence, the registered measuring data of the test house contain an ample Generally, it is astounding that the wood moisture stayed that dry despite − very high room moisture from December 2011 to February 2012 or between November 2012 and February 2013 (Fig. 4), 317 − humidity sensors in the boundary layer between insulation and historic construction partly measured condensate for months, only 200 mm away from the bottom side of some beams (Fig. 5), − surface condensate at the outer end of the bearing was detected (Fig. 6), separated only by an air gap from the end grain. 3.1 Comparison of simulation results with measurements From the test house the climate boundary conditions are known. Samples of the historic construction were measured and the material functions of the insulation systems are known as well. Together with detailed measurement data from inside the constructions it is possible to compare simulation results with measurements. In this ongoing process it soon became obvious, that especially the influence of air fluxes between bearing and rooms are difficult to model. Based on measurements of wall assemblies (Arbeitsgemeinschaft Mauerziegel 2010), air permeabilities were assigned to materials in a set of simulations and were varied. All simulations considered long-wave radiation exchange in air gaps resp. layers. The thermal conductivity of air varied in dependency of the geometry of the gaps and adjacent temperatures. The influence of build-in moisture also was taken into account. Figure 8 compares measurement data with simulation results at the wooden beam with perlite boards insulation. In the simulations, carried out with Delphin version 5.8.1, the models reproduced a cross section through the wooden beam. The courses indicate the relative humidity in the middle of the bearing near the beam cheek during the first year. These high moisture loads did not affect wood moistures in any notable way. The wooden beams did not react “quickly enough” to suffer from severe harm. At beams one (cellulose insulation) and eight (without insulations) two moisture sensors in 15 and 30 mm depth were fixed in the plane of the internal insulation. From beam one it became visible that moisture contents in the plane of the internal insulation run lower than near the end grain (not documented here). 2.7 Summary of previous measurements The low wood moisture at all insulation systems indicates that the wooden beams are not endangered by these internal insulation systems. Here, a greater risk probably would emanate from penetrating winddriven rain which is not the object of this research. It could be demonstrated that the sealing of the bearing reduces the relative humidity around the beam heads significantly. In spite of these favourable, not dangerously measuring data, the results of the test house should not be generalised. The very porous and capillary conducting bricks may be one key contributing factor for these positive data. Furthermore, the orientation of the test wall and the specific urban location could benefit these favourable data. Relative humidity in [%] 100 3 SIMULATIONS Currently, no commercially available software is able to calculate heat and moisture (gas & liquid) transport in three-dimensional constructions. Elaborate advancements of existing software tools, see Janetti et al. (2011), show that 3D-simulations with real climate conditions and adequate discretisation would take too long for practitioners. In the foreseeable future a well discretised 3D-calculation of a wooden beam under real climate conditions with three (heat, moisture, air) balance equations will take weeks. The interest in getting things straight concerning the hygrothermal coaction of wooden beams and internal insulation is actually very high. So, conventional 2D-software tools as Wufi (Künzel 1994) or Delphin (Nicolai 2007) are used in real building projects to compare different insulation variants with the existing construction. 90 No additional air flux models Constant air change rate n= 1 1/h Measurement Air pressure in dep. of vapour pressure 80 70 60 50 40 30 Okt Nov Dez Jän Feb Mär Apr Mai Jun Jul Aug Sep Figure 8. Comparison of different simulation variants with measurement data at beam 3 (perlite board): courses of relative humidity in the middle of the bearing (150 mm from edge of the bearing) from September 2010 till September 2011. The thick black line depicts the measurement data. The thin black line describes the result of a simulation how it is often done in practice, without additional air (flow) model. As mentioned, material parameters and climate conditions were well-known, a situation which is rarely given in practice. Despite this favourable starting position, both lines do not seem to derive from the measured detail nor identical climate conditions. 318 Another simulation model used a fixed air change rate of 1 h-1 between the air layers around the beam head and the climate of the upper room (thin grey line in Fig. 8). The bearing of this beam was wellsealed towards the lower room, that’s why, the air exchange with the room in the first floor was neglected. The improvements in comparison to the first simulation are marginal. The thick grey line shows the results of a simulation which used a laminar air flow model. In the absence of differential air pressures between (attic) room and outer climate, the differential vapour pressure was used to generate an air pressure gradient. Such procedure would enable designers and engineers to find an easy approach to consider air flux because with air temperature and humidity vapour pressures are known. The scaling approach used fitted quite well during warmer months but overestimated the air flux from inside to outside between end December 2010 and April 2011. Current work is about finding a more suitable scaling approach. 6 REFERENCES Arbeitsgemeinschaft Mauerziegel (im Bundesverband der Deutschen Ziegelindustrie e.v.) 2010. Luftdichtheit in Ziegel-Massivgebäuden, AMz-Bericht 1/2000 DIN 68800. Holzschutz – Teil 2: Vorbeugende bauliche Maßnahmen im Hochbau. Standard DIN 68800-2. Beuth Verlag, Berlin 2012 Janetti M.B., Ochs F., and Feist W, 2011. 3D Simulation of Heat and Moisture Diffusion in Constructions. http://www.comsol.com/papers/10921/ at 4.3.2013 Künzel H.M. 1994. Verfahren zur ein- und zweidimensionalen Berechnung des gekoppelten Wärme- und Feuchtetransports in Bauteilen mit einfachen Kennwerten. Dissertation, Universität Stuttgart Nicolai A. 2007. Modeling and Numerical Simulation of Salt Transport and Phase Transitions in Unsaturated Porous Building Materials, PhD Thesis, Syracuse University, NY, USA Viitanen H., Toratti T., Makkonen L., Peuhkuri R., Ojanen T., Ruokolainen L., and Räisänen, J. 2010. Towards modelling of decay risk of wooden materials. Eur. J. Wood Prod. (2010) 68: pp. 303–313 Wegerer P., and Bednar T. 2011. Long-term Measurement and Hygrothermal Simulation of an Interior Insulation Consisting of Reed Panels and Clay Plaster. Proceedings of the 9th Nordic Symposium on Building Physics NSB 2011, Tampere University of Technology, Department of Civil Engineering, Volume 1 (2011), pp. 331 - 3 4 CONCLUSIONS With hygrothermal probes the boundary layers between five insulation systems and existing construction, the air around the wooden beam heads in the bearing and the wooden beams themselves as well as adjacent climates have been measured. During the heating periods quite different humidity loads from the rooms developed. High moisture load in the room led to high humidity in the bearings and the insulation systems but not in the wooden beams. Obviously, the exposure time was not long enough. The measurements suggest that internal insulation does barely affect wooden beams of ceilings. However, other circumstances were favourable to the beam heads: very porous bricks and no remarkable driving rain load. The material properties of nearly all materials have been measured. Using two-dimensional software with heat and moisture transport the simulation results diverged a lot from the measurements. Only in due consideration of air movement concordance became better. The simulations are continued to reach better concordance. 5 ACKNOWLEDGEMENTS The project OEKO-ID is funded by the Austrian “Klima- und Energiefonds” of the FFG within the framework of „NEUE ENERGIEN 2020“(project number 818908). Thanks to the meteorological station of the Karl-Franzens University of Graz for providing measured climate data. 319 320 Derivation of an evaluation method for the hygrothermal and biohygrothermal behaviour of straw as insulation M. Klatecki & A. Maas Universität Kassel, Department of Building Physics, Kassel, Germany ABSTRACT: The durability of components consisting of straw bales depends substantially on its moistureproofing. Due to its suitability to serve as a substrate for fungal cultures, the integrity of straw can be compromised if a sufficient amount of water accrues in the material. In that case, fungal spores are able to germinate and decompose the straw, which is why appropriate measures must be adopted in order to prevent the attainment of a critical threshold for fungal growth. However universally valid criteria for the validation of the moisture proofing of straw bale constructions are not yet available. Up to now the prognosticated mildew growth in straw bale layers near the outside shows discrepancies between measurement and simulation. Therefore a new model, based on the works of Künzel and Sedlbauer, shall be developed. This is ascertained by metrological investigations of straw bale constructions and the subsequent adaption between the parameters of material characteristics and simulation. 1 INTRODUCTION In accordance with the scientific works of Künzel (Künzel 1994) and Sedlbauer (Sedlbauer 2001), it is intended to develop a technical dampness evaluation method for straw-insulated constructions on the basis of the current simulation programs, for instance WUFI. For this purpose, selected building elements are investigated in a double-climate system in order to determine the moisture’s technical behavior and subsequently they are compared with simulations based on the WUFI simulation program for hygrothermal analyses. On the basis of the results of the metrological investigation, an approximation between calculated and measured results is to be achieved by adapting the parameter, for example the material characreristics. The results of the comparison are validated on constructions freely exposed to the weather. The requirements for the growth of mildews within the constructions are determined by the evaluation method depicted in (Sedlbauer 2001) (if necessary, the substrate categories are adjusted for this purpose). With regard to these growth requirements, there are carried out additional investigations, which are however not yet brought to a close. In the following chapters the metrological investigations for determining the water transport processes within the straw layer and the practice of the simulation technical matching with the measured results are presented. By adapting the material characteristics of straw, for example the thermal conductivity or the moisture storage capacity, the matching between measured and simulated results at the relative humidity and temperature in different depths of the straw layer shall be determined. In Germany, first constructions were made of straw bales in the late nineties of the 20th century. Although most of them were built for experimental purposes, soon the first residential buildings made of straw appeared (for instance in the Rhön and in the „Ökodorf Sieben Linden“). The most prevalent technique in Germany is the pillar construction, in which the straw bale is used as an intercostal insulation (Rüger et al. 2004). First approaches concerning the load transferring construction method were investigated in a test object in „Sieben Linden“ and, more intensively, in the research project „Principles for the Technical Approval of Straw Bale Constructions“ (Danielewicz et. al. 2008), which was funded by the DBU. The durability of components consisting of straw bales depends substantially on its moisture-proofing. If a critically huge amount of water accrues in the material over a longer period, fungal spores are able to germinate and decompose the straw. Therefore, it must be prevented by appropriate measures that the thresholds for fungal growth are attained. By comparing the results of simulations based on common hygrothermal calculation methods with measurands of actual building elements, one can notice significant deviations concerning the degree of moisture within straw-insulated constructions, which is why a realistic evaluation of those constructions is possible only to a limited extent. The water contents inside the construction (especially in the exterior area), feature lower water contents than the results of the simulation, so that a higher risk of mildew growth is computationally revealed there. 321 bales of straw with interior and exterior rendering and planking. At the examined reference point close to the outside barrier within the layer of straw, no growth of mildews appeared. During all depicted investigations, the evaluation of the moisture proofing was conducted on the basis of the growth of mildews for a reference point inside the insulation layer consisting of straw (5 cm or 8 cm distance from the interface between exterior rendering and insulation) with the illustrated model of evaluation referred to (Sedlbauer 2001). At the moment, the status quo of the technique is marked by the general building approval (Z-23.11-1595) (DiBt 2009) issued by the German Institute for Constructional Engineering. The approval is based on the investigations (Krus & Saur 2005) and establishes limits for the acceptable average degree of moisture inside the insulation of straw as well as regulations for the permissible diffusion resistance inside the room. Furthermore, it contains specifications concerning the outside cover in form of a cladding and insulation. 2 BASIC PRINCIPLES Generally admitted criteria for the assessment of the moisture proofing of devices consisting of bales of straw are not yet available. The evaluation was hitherto made by means of the prognosticated mildew growth in a depth of 5 cm within the straw layer, which was determined on the basis of the attuned relative humidity and temperature. Additionally indicative results of exemplary investigations see (Danielewicz et. al. 2008), (Otto & Klatecki 2008), (Krus & Saur 2005) and (Ziegler & Gann 2004), metrological examination results of real building projects (Jolly 2000) and suggestions for the erection of walls see e.g. (Rüger 2004) and (Wimmer et. al. 2001) exist. Concerning (Danielewicz et. al. 2008), systematic hygrothermal simulations for walls, roofs, slabs and flat slab constructions were performed. The hygrothermal and biohygrothermal conditions were determined by applying the WUFI and WUFI-Bio programs. The climate data record „ibp 1991“ (Künzel & Schmidt 1999) is used for the external climate while the climate conditions within the building correspond to the residential property conditions with ordinary moisture load. On behalf of the Austrian Ecological Institute for Applied Environmental Research, the growth of mildews on straw and walls of straw was examined in (Ziegler & Gann 2004). The study assayed the hygrothermal characteristics and the risk of mildew growth at various devices. It specifically determined the boundary conditions that result in mildew growth in the laboratory and analyzed the emission of MVOC (Microbially Volatile Organic Compounds) from insulations of straw infested with mildews as well as the hygrothermal performance of structural components. During the experimental inquest for modeling the growth of mildews, a visual observation of the growth was conducted. In this context, it became apparent that neither substrate group 0 nor substrate group I (according to Sedlbauer’s scheme) (Sedlbauer 2001) was applicable without modifications. Comparing the examination results of the visible growth of mildews with substrate group I, it emerges that the isopleth model substrate group I after (Sedlbauer 2001) partly requires less time for the occurrence of visual mildew growth. However, the time measured during the examination partly correspond with that of the isopleth model substrate group 0. Concerning (Krus & Saur 2005) two different constructions with a shelter against the weather are investigated. The results of the calculation demonstrate that a vapour retarder (here OSB-plate) inside a room combined with a fibreboard fixed at the outside, can considerably improve the moisture technical behavior compared to a wall consisting of 3 BASIS FOR EXAMINATION The metrological examinations take place in a socalled double-climatic chamber. The double-climatic chamber is schematically illustrated in Figure 1. Figure 1. Sketch of the double-climatic chamber. The climate conditions can be independently regulated in both chambers and the attached specimens can be investigated metrologically. In doing so the temperature and the relative humidity within the specimen are gathered in different depths. The wall between both climate chambers, which contains the attached samples, is insulated and joint gaps are sealed. The hygrothermal conditions on different climates can be measured with the sensing elements previously built in the specimens. The assembly of the elements inside of the construction is illustrated in Figure 2. In the course of the examination the temperature and the relative humidity are investigated on the internal and external surface at a depth of 5 cm and 8 cm from the outside, in the midst of the construction and at a depth of 9 cm from the inside. 322 Figure 2. Structure of the sensing elements within the investigated construction. For the determination of the water transport process inside the straw layer, various constructions are examined. Table 2 for example depicts a construction with an internal and external width of vapour retarder with a diffusion resistance of 2 m. Figure 4. Sequence of the outdoor temperature and the relative humidity during the measuring cycle. Table 1. Illustration of a metrologically investigated constructition. Layer thickness [cm] Material 5 36 5 Mineral ultra-light-rendering Straw Loam rendering In order to evaluate the water transport processes within the straw layer, the last 24h after achieving the equilibrium moisture content are used for the comparison with the results of the hygrothermal simulations. A matching between the results of the measurement and the simulation shall be attained by adapting various parameters, for example material properties, with the program WUFI, and shall led to the knowledge of the hygrothermal behavior of straw. The validation takes place by using existing external wall construction at the location Sieben Linden near Magdeburg under the real climatic conditions of the year 2011. The considered wall structure is lying on the west side and consists of a wooden pillar construction with an infill of straw bales, a double-sided layer of loam rendering and a covered shuttering serving as a shelter against the weather. Due to the leaky shuttering, the resulting outside climate is almost identical to the external climate. Only the exposure to rain and solar radiation is prevented, which are excluded from the consideration. For this exterior wall, the temperature and relative humidity within the layer of straw are metrologically determined in the same way they were determined in the examination in the double-climatic chamber. The sensing elements keep a log of the conditions at the external boundary layer between exterior rendering and straw insulation and also in a depth of 5 cm within the straw layer. Additionally the developing indoor climate is recorded. The structure of the sensing elements is illustrated in Figure 5. During the metrological investigation, there are the climate conditions shown in Figure 3 and 4. Figure 3. Sequence of the indoor temperature and the relative humidity during the measuring cycle. 323 5 RESULTS OF THE EXAMINATION The results of examination are illustrated for the construction, which has achieved the equilibrium moisture content and is depicted in chapter 3. The temperatures and the relative humidity are approximately constant. Table 3 provides an overview on the resulting hygrothermal conditions after achieving the equilibrium moisture content within the construction. Figure 5. Structure of the sensing elements within the investigated construction. Table 3. Overview on the temperatures, the relative humidity at a steady state for specimen 2. Temperature Relative Humidity Place/Sensor θ ϕ [°C] [%] Room outdoor 0 88 9 cm depth from outside (A5) 1,4 85 12 cm depth from outside (A8) 2,1 77 Middle (M18) 11 50 13 cm from inside (I9) 16 33 Room indoor 20 40 The structure of the examined exterior wall construction is depicted in Table 2. Table 2. Illustration of the investigated external wall constructition. Layer thickness [cm] Material 4 46 4 Loam rendering, three-ply Straw Loam rendering, three-ply The results of the metrological investigations are edited for the program WUFI up to 8760 h. By aligning the measurements with the results of the simulation, using the material characteristics of straw from literature, there are substantial discrepancies. Position A5 for example, in a depth 9 cm from the outside, shows a variance of 6,5 % R.H. The warmer the construction, that concerns indoor positions, the more the variances decrease, see Table 4. Also by adapting the moisture dependent thermal conductivity and diffusion resistance, there is hardly an improvement. Through the increase of the sorption isotherm of the material straw of 10 kg/m3 starting from 70 % R.H., the variances between simulation and measurement decrease (Table 4). It is recognizable, that the variations of relative humidities at lower temperatures (10 °C) are considerably higher than at temperatures above 10 °C. Thus through the increase of the sorption isotherm, the variances can be reduced. 4 EXAMINATION METHOD In order to develop the hygrothermal simulation model, using the current simulation programs such as WUFI, metrological investigations are conducted on chosen straw-insulated constructions in a doubleclimate chamber. The metrological investigations are carried out on defined hygrothermal climatic conditions. The specimens are metrologically investigated in different depths within the construction by logging the temperature and relative humidity. Based on the results of the metrological investigation, a matching between simulated and measured results shall be achieved by adaption of the material characteristics in the program WUFI. For this purpose the last 24h of the metrological investigation after achieving the equilibrium moisture content are used. The results and the acting climatic conditions are edited for the simulation program up to 8760 h (1 year). In doing so especially the hygrothermal parameter of the moisture dependent thermal conductivity, of the diffusion resistance and of the sorption isotherm of the used materials are adapted. Additionally the moisture transport mechanisms are modified, so that a matching of the results can be achieved. At first the characteristics for the used materials, which are given in literature, are used for the simulation technical investigation. The validation of the ascertained material data and transport mechanisms takes place under actual climatic conditions on a metrological investigated exterior wall construction. Table 4. Average variances of measurements and simulated figure at a achieved equilibrium moisture content. Variant A5 A8 Variance at: Variance at: T R.H. T R.H. [°C] [%] [°C] [%] Initial point 0,34 - 6,78 0,41 - 3,19 Adaption 0,10 - 5,58 0,17 - 2,11 In Figure 6 and 7 the results of the comparative simulation are depicted with the measurement results. Figure 6 shows the comparison of the measurement and simulation with the approach of the material characteristics of straw from literature and 324 er, especially in spring there are still variances in the range of decreasing moistness. The months of summer and autumn were, concerning the range of the constant and increasing moisture behavior, not gathered by an increase of the sorption isotherm (increase starting from 70 %). Because of this fact, it is perceptible that the deviations emerge, if there are modifications concerning moistness, especially at moisture decrease, as one can see in Figure 8. Since the simulated results are almost congruent with the measurements, the depiction of the temperature profiles is expendable. depiction 7 shows the adaption of the sorption isotherm. 0 1460 2920 4380 5840 time [h] 7300 8760 0 1460 2920 4380 5840 time [h] 7300 8760 Figure 6. Depiction of Initial Point at a depth 9 cm from the outside (literature). 0 1460 2920 4380 5840 time [h] 7300 Figure 8. Rel. moistness in A5 at a depth 9 cm from the outside with increased moisture storage function in the layers of clay of about 10 kg/m3 water. 8760 6 CONCLUSION AND OUTLOOK 0 1460 2920 4380 time [h] 5840 7300 Concerning straw as material, the range of desorption and low temperatures within the construction cannot be described realistically without adapting the moisture storage function. Determining the water content with the simulation method would clearly lead to an overvaluation concerning the water content. In consequence of the considerably increased water content, it is an overvaluation of the potential risk of mildew growth in spring. The possibility to evaluate constructions insulated with straw with the simulation method in complex „temperature range zones“ with different moisture storage functions is presently being investigated in order to achieve a better elaboration. For this purpose, an external wall construction is, for instance, divided in 3 temperature ranges with their own moisture storage function according to their average 8760 Figure 7. Depiction of Adaption at a depth 9 cm from the outside. The validation of the adapted material data is effected by the comparison with the construction in chapter 2, which is actually exposed to the weather. The results show a good compliance between simulation and measurement in the months of winter and summer. By an increase of the sorption isotherm in winter the behavior of the relative humidity of above 80 % R.H. can be approximately reduced to 75 % R.H., thus it is almost an identical behavior. Howev325 yearly temperature. In this case, the exterior „temperature range zone“ obtains the highest moisture storage function, since the lowest average yearly temperature is in this range. During the process, the problem arises that the range of desorption can only be gathered to a limited extent. Alternatively, simulations depending on the season can be implemented (i.e. a simulation for each season). This requires up to 4 simulations with different moisture storage functions (depending on the season). 7 REFERENCES Danielewicz I., Fitz C., Klatecki M., Krick B., Krueger N., Krus M., Minke G., Otto F., Scharmer D., and Teuber W. 2008. Grundlagen zur bauaufsichtlichen Anerkennung der Strohballenbauweise – Weiterentwicklung der lasttragenden und Optimierung der bauphysikalischen Performance. DBU Abschlußbericht. Deutsches Institut für Bautechnik (DIBt). 2009. Allgemein bauaufsichtliche Zulassung: Wärmedämmstoff aus Strohballen “Baustrohballen”. (Z-23.11-1595). Jolly R. 2000. Strawbale moisture monitoring report submitted to Don Fugler, Canadian mortage and housing corporation. Unpublished manuscript. Krus M., and Saur A. 2005. Untersuchungsbericht IBP-Bericht RKB-09/2005: Feuchtetechnische Untersuchung an einer Wandkonstruktion aus Holzständerwerk mit Strohballenausfachung. Holzkirchen. Krus M., and Saur A. 2005. Untersuchungsbericht IBP-Bericht RKB-13/2005: Feuchtetechnische Untersuchung an einer Wandkonstruktion aus Holzständerwerk mit Strohballenausfachung. Holzkirchen. Künzel H.M. 1994. Verfahren zur ein- und zweidimensionalen Berechnung des gekoppelten Wärme- und Feuchtetransports in Bauteilen mit einfachen Kennwerten. Dissertation Universität Stuttgart. Künzel H.M., and Schmidt Th. 1999. Auswahl und Aufbereitung von meteorologischen Datensätzen für Feuchtetransportberechnungen. Tagungsband 10, Bauklimatisches Sympositum, S. 637-647. Dresden. Otto F., and Klatecki M. 2008. Hygrothermische und Biohygrothermische Simulationen an Bauteilen mit einer Dämmung aus Strohballen. Zentrum für Umweltbewusstes Bauen e.V. Kassel. Rüger B. 2004. Putzoberflächen auf Strohballenwänden. Fachverband für Strohballenbau Deutschland e.V. Rüger B., Scharmer D., and Rex M. 2004. Strohballenbau in Deutschland – Eine kurze Einführung. Fachverband für Strohballenbau Deutschland e.V. Sedlbauer K. 2001. Beurteilung von Schimmelpilzbildung auf und in Bauteilen. Dissertation Universität Stuttgart. Wimmer R., Hohensinner H., Janisch L., and Drack M. 2001. Wandsysteme aus nachwachsenden Rohstoffen – Wirtschaftsbezogene Grundlagenstudien. Haus der Zukunft Endbericht. Ziegler T., and Gann M. 2004. Österreichisches Ökologieinstitut für angewandte Umweltforschung: Wachstum von Schimmelpilzen auf Stroh und Strohwänden. Wien. 326 Smart vapour barriers in unventilated wooden roofs in a Nordic climate – laboratory study of drying effect under shaded conditions S. Geving and E. Thorsrud Norwegian University of Science and Technology, Department of Civil and Transport Engineering, Trondheim, Norway S. Uvsløkk SINTEF Building and Infrastructure, Trondheim, Norway ABSTRACT: Unventilated wooden roofs typically need a vapour barrier at the warm side to avoid interstitial condensation due to vapor diffusion and air leakages from the interior. A vapour barrier such as a polyethylene foil does however not allow drying of moisture to the interior, making the construction vulnerable to moisture damages. So-called smart vapour barriers or retarders could allow condensed moisture, built-inmoisture or moisture from minor leakages to dry to the interior. The drying to the interior will typically happen during summer conditions when the exterior surface is heated by solar radiation, thereby creating an inward directed vapour transport. A critical part of this concept may however be the parts of the roof that are shaded by various objects or surrounding buildings. The drying effect of applying four types of smart vapour barriers in unventilated wooden roofs during summer conditions in a Nordic climate – but shaded from direct solar radiation – is investigated in this laboratory study. The results showed that the drying rate decreased compared with flat roofs exposed for direct solar radiation, but also that the drying rate is at a relatively high level even under shaded conditions for some of the smart vapour barriers investigated. 1 INTRODUCTION not the vapour resistance was so low that the risk for interstitial condensation during winter conditions could be unacceptably high. While a vapour retarder has a given constant vapour resistance, some vapour barriers or retarders are sold (Note: “barrier” is used in the rest of this paper for simplification) on the European and North American market with adaptable vapour resistance in regard to what is actually needed. Popular terms for these products are “smart”, “intelligent”, “moisture adaptive” or “humidity dependent” vapour barriers. The physical behaviour of these products varies, but the main principle is that the vapour barrier should function as an ordinary vapour tight vapour barrier most of the time, preventing vapour diffusion into the construction from the indoor air – especially during winter conditions. If, on the other hand, the construction is wet, for example due to built-in moisture or leakages, so that the relative humidity (RH) on the exterior side of the vapour retarder gets high, the vapour resistance will be reduced so that there may be possibilities for drying inwards - especially during summer conditions. Another possibility may be that the barrier layer has capillary properties with possibilities to transport condensed water to the interior surface. Probably the first commercial product that can be put under the term “Smart Vapour Barrier” (SVB) was Hygrodiode that was developed in Denmark during the 80’s (Korsgaard, 1985). It consisted of synthetic fibres sandwiched between stripes of polyethylene, giving it a constant very high vapour resistance similar to the polyethylene film. If on the other hand moisture condensed on the exterior side In Nordic climates a vapour barrier is typically used on the warm side of the building envelope to avoid interstitial condensation during the heating season due to vapour diffusion and air leakages from the interior. Often a polyethylene foil is used, which has a very high vapour resistance. Such a vapour tight product does however not allow moisture to dry to the interior. A more vapour open product could allow condensed moisture, built-in moisture or moisture from minor leakages to dry to the interior. This could be especially useful for constructions with reduced or no possibility to dry to the exterior, such as unventilated flat roofs. It might allow cheaper solutions, making it possible to use organic (wooden) materials in such constructions or skip the use of wooden preservatives. The term “vapour retarder”, as opposed to “vapour barrier”, is often used on products that have a lower vapour resistance than recommended for vapour barriers. Such products may allow for some drying to the interior. The vapour retarders has a given constant vapour resistance, but different commercial products vary quite a lot in regard to their level of vapour resistance, typically somewhere between Sd = 2–10 m for products sold on the European market. Even though these levels of vapour resistances may give some drying to the interior during summer conditions, these vapour resistances may still be so high that the actual drying effect may be limited. Geving and Holme (2012) found for instance that ordinary vapour retarders had little effect on the total drying of ordinary wood frame walls – if 327 conditions was the cause of extensive mould and rot in unventilated wooden roofs with Hygrodiode. This study is investigating and comparing the performance of two newer products (Intello and AirGuard Smart) with Delta-Novaflexx (i.e. Difunorm Vario) and Hygrodiode – applied in unventilated wooden roofs under shaded conditions. In a prestudy of this investigation similar measurements for unventilated roofs exposed to direct solar radiation were documented (Geving, Stellander and Uvsløkk, 2013). of the vapour barrier, the water would be transported sideways through the fibres by the wicking action and be allowed to evaporate to the interior side. Research on the practical use of this product was performed both in Denmark and the US in the late 80’s and early 90’s (Korsgaard and Pedersen, 1990; Rode, Christensen and Vesterløkke, 1993; Pedersen et.al., 1992) and was found to give a significant drying effect of unvented lightweight roofs. In the mid 90’s another type of SVB was developed in Germany (Künzel, 1996). The product is named Difunorm Vario, but is also manufactured under other names, including MemBrain in the US, and Delta-Novaflexx in some parts of Europe. It consists of a polyamide film with no capillary properties, but with a pronounced difference in vapour resistance for high and low ambient RH. The Sdvalue is approximately 4-5 m when the RH is below about 40 % and Sd = 0,1-0,2 when the RH is above 80% (Künzel 1996). Also this product was first introduced as beneficial for unvented wooden roofs. Most research from then and till today has been focused on use of this product. There are a lot of studies on unvented roof systems with this product (Künzel, 1996; Künzel, 1998; Künzel, 1999; Künzel and Leimer, 2001; Ghazi Wakili and Frank, 2004, Kaufmann, Künzel and Radon, 2006; Bludau and Künzel, 2009; Bludau, Schmidt and Künzel, 2009). Since then and up till today there have been developed various similar products with RH-dependent vapour resistance. Some of these new products may have a higher resulting vapour resistance for winter conditions, compared to Difunorm Vario, reducing the risk for interstitial conditions. And some may have a lower resulting vapour resistance for summer conditions, resulting in higher drying rate to the interior. Documented scientific studies of applications for these new products are however limited although they may have some advantages compared to Hygrodiode and Difunorm Vario. The drying when using SVBs typically takes place during the summer season, when the external surface is heated by the sun or the outdoor air temperatures are generally high - and thereby the driving potential for inward vapour flow is highest. This inward flow will increase the RH to high levels, and even give condensation, at the interface between the insulation and the SVB. Thus the Sd-value of the SVB decreases and the inward directed vapour flow through the SVB increases. One problem may however occur if the roof is shaded, i.e. not receiving direct solar radiation, the roof is sloped to the north or if the external surface has a low solar absorptance. The external surface will then have a considerable lower temperature and the potential for inward vapour flow over the insulation cavity decreases. Brandt and Bunch-Nielsen (2006) reported for instance a number of projects where partly shaded 2 LABORATORY TESTS 2.1 General The purpose of these tests was to investigate the drying effect of applying four types of SVBs in unventilated wooden roofs with built-in moisture during summer conditions in a Nordic climate – for cases where the roof is shaded from direct solar radiation. Typical diurnal variations of external roof surface temperatures during summer, for different roof configurations shaded from direct solar radiation, were repeated during a period of 60 days, and the drying rates were monitored. The following type of compact unventilated wood frame roof was tested (from exterior side): • Roofing membrane • 21 mm plywood (spruce) sheathing or no sheathing • 200 mm glass wool • SVB • 13 mm gypsum board Even though the construction considered is an unventilated compact roof, the results may be applicable to assess the maximum possible drying rate to the interior for walls and ventilated roofs that are shaded from direct solar radiation. The measurements took place in the laboratories of Department of Civil and Transport Engineering, Norwegian University of Science and Technology, during autumn 2012. 2.2 Experimental set-up Six different configurations of the wood frame roof were tested, where the only difference was the type of SVB used or whether plywood sheathing was used or not. Four configurations had plywood sheathing and two had no sheathing. For the two configurations without plywood sheathing a thin cotton cloth was used instead to hold the added built-in moisture. Each configuration was tested with two specimens to get a better accuracy, i.e. a total of 12 specimens. The roof specimens were built up in 328 storage boxes of polypropylene; where the bottom of the boxes was used to imitate the roofing membrane and the rest of the materials were adjusted to fit the box, see Figure 1. The boxes had interior dimensions b = 0,28 m, l = 0,40 m and h = 0,22 m, and flanges onto which the SVBs and gypsum board were taped. The corner of the boxes were curved so the resulting “light-opening” of the boxes had an area of 0,105 m2. The boxes were placed in a lifted test rig with bottom up, insulation on all sides to get isothermal conditions and a heating foil that could be controlled to give the wanted surface temperature on the roofing membrane (i.e. the bottom of the polypropylene box). Each test box could be dismounted during the experiment to monitor the continuous weight loss. See details in Figure 2. 2.1 Materials The four types of SVBs are; • Delta-Novaflexx • Intello • AirGuard Smart • Hygrodiode 200A Delta-Novaflexx is a polyamide with an Sd-value ranging from 0,2–5 m. This product is manufactured under many different names, including MemBrain in the US, and Difunorm Vario in Germany. It was originally developed in Germany in the late 90’s (Künzel, 1996), and was the first commercial available relative humidity dependent vapour barrier product for building applications. Most scientific studies on use of SVBs in building applications are limited to this product only. Intello is a polyethylene copolymer, where the copolymer is an acrylic with hygroscopic properties. The Sd-value ranges from 0,25-26 m (Pro Clima, 2006). AirGuard Smart is a polyvinyl alcohol film (with spun bond polypropylene as reinforcement and protecting layer) with an Sd-value ranging from 0,05- >30 m (DuPont, 2012). Hygrodiode 200A is composed of a fabric of synthetic fibres with good capillary properties laminated with stripes of perforated PE-foil and polyamide-foil. It should be noted that this product is a further development of the original Hygrodiode mentioned earlier, i.e. the polyamide film giving the product drying posibilities also for conditions below 100% RH. The Sd-value ranges from 1-20 m (Icopal, 2005). The gypsum board has an Sd-value of approximately 0,1 m. Figure 1. Test box with gypsum board and measurement devices on top. Figure 2. Section of the test-rig showing the two parallel test boxes. 329 Table 2. Chosen diurnal cycles during test period of 60 days. Cycle no. Number of days Max. temperature (°C) 1 13 25 2 4 33 3 13 30 4 13 25 5 4 33 6 13 30 2.2 Boundary and initial conditions The external boundary conditions were given by the temperature controlled by the heating foil in the test rig. During sunny summer days the maximum temperature on flat roofs may reach 50-60°C. However, for shaded constructions this maximum temperature during sunny days will be considerably lower. It was decided to use diurnal cycles for external temperature over a measurement period of 60 days. The diurnal cycles were selected based on a parametric study with WUFI 1D Pro (Fraunhofer IBP, 2010). The external surface temperature of the roof was simulated for the summer period (May-Sept) for shaded conditions in Trondheim (and partly Oslo) in Norway. Since WUFI has no possibility to include the real shading effect, the parametric study included variations of sloped roof towards north (55 and 90°) and low solar absorption coefficient (α=0,3 and 0,5) for a flat roof. While a flat roof with solar absorption coefficient of 0,9 (no shading) gave a maximum temperature of 57 °C, the maximum temperature for the cases meant to simulate shading was between 15 and 24 °C lower, se Table 1. Based on case 4 and 5, the 60 days with highest external temperatures were identified (during a total of 5 months). Based on these 60 days three levels of diurnal variations were selected, and these three levels were repeated twice, see Table 2 and Figure 7. It should be noted that although the maximum temperature in case 4 and 5 differs from the maximum temperature in cycle 2/5, cycle 2/5 was designed so as to give the same drying potential as the average of the 8 hottest days. It was difficult to use the heating foil to recreate the exact temperature profile from the simulations, since the “cooling” period after turning off the heating foil after reaching the maximum temperature took many hours. It was therefore decided to use a somewhat lower maximum temperature that gave the same drying potential towards the interior summed up over the day. To simulate built-in moisture or an accidental leak the plywood sheathing was given an initial moisture content between 27-30 weight% (measured by gravimetric method) by immersion in water, except for one of the 12 specimens (Novaflexx 2A) that of some reason absorbed higher amounts of water (38 weight%). 2.3 Measurements The rig and test boxes were constructed so that the boxes could be dismounted and weighed regularly to follow the total drying of the configurations. In addition the temperature and RH at the interface between the SVB and insulation were measured (logged each sixth minute) in one of the two test boxes of each configuration. The moisture content of the plywood board was measured manually once a day by traditional resistance measurements, using screws (dimension 3,5 x 9,5 mm) with distance 25 mm as electrodes, one measurement point with the screws mounted from the top of the board and another point with the screws mounted from the bottom. The screws were sealed with epoxy on top to avoid extra capillary uptake during immersion in water. As for the RH measurements also the wood moisture content was only measured in one of the boxes for each roof configuration. The wood moisture meter was set at Scandinavian Spruce. Both the weighing of the boxes and the measurement of wood moisture content were made just before the heating foil was turned on, i.e. when the temperature of the boxes was uniform and close to laboratory temperature. The RH and temperature of the indoor climate were also continuously logged. Table 1. Maximum roof temperature from WUFI simulations. Case Description Max. temperature (°C) REF Flat roof, α = 0,9 57 1 Slope 90° (i.e. wall) to north, α = 0,9 33 2 Flat roof, α = 0,3 35 3 Flat roof, α = 0,5 42 4 Slope 55° to north, α = 0,9 36 5 Slope 90° to north, α = 0,9, Oslo 38 3 RESULTS The drying of the built-in moisture of the six roof configurations are illustrated in Figures 5 and 6. In Figure 5 is shown the total water content of the boxes during the test period, i.e. total water content divided on drying opening (area) at the SVB level. Each curve in Figure 5 represents the average of two test boxes. For all configurations, except for DeltaNovaflexx (with plywood sheathing) where specimen 2A had higher initial moisture content, the two parallel boxes had very similar drying curves. From The minimum temperature during night for these daily cycles in Table 2 was the same as the air temperature in the climatic chamber (23°C). The test rig was placed in a climatic chamber with constant indoor temperature and RH of 23°C and 49%. 330 the figure we see that the configurations with DeltaNovaflexx and AirGuard Smart have a much faster rate of drying than the rest of the configurations, followed by Hygrodiode and Intello. It should also be noted from Figure 5 that Hygrodiode and Intello have almost constant drying rates during the whole measurement period, while Delta-Novaflexx and AirGuard Smart have changing drying rates. This is probably due to the fact that both Hygrodiode and Intello experiences almost constant boundary conditions on the external side of the SVB (RH ≈ 100%), while Delta-Novaflexx and AirGuard Smart have varying RH on the external side of the SVB during the changing cycles of Table 2. This is shown in Figure 7 later on. Intello Hygrodiode Novaflexx w/cotton Total drying (kg/m2) 2 1,5 1 After 9 days 0,5 After 30 days After 60 days 0 Novaflexx AirGuard Smart AirGuard Smart w/cotton Figure 6. Total drying after 10, 30 and 60 days for the 6 roof configurations. Average values for two test boxes. Total water content (kg/m2) 2,5 When looking at both Figure 5 and 6 and comparing the drying rate and total drying for the two configurations without plywood sheathing (w/cotton) we see an indication of AirGuard Smart drying somewhat faster than Delta-Novaflexx. This is logical since AirGuard Smart has a lower Sd-value than Delta-Novaflexx at high RH-levels. This difference is however not big. If we compare the Delta-Novaflexx and Air Guard configurations with and without plywood sheathing we observe the total drying given in figure 6 is higher after 30 and 60 days for the configurations without plywood sheathing. This is probably due to the fact that the moisture release from the plywood sheathing is slowed down by both the moisture capacity and the vapour resistance of the material itself. In Figure 7 is shown the measured RH at the interface between the SVB and insulation level for the whole period of the test. We see that the configurations for Hygrodiode and Intello reaches 100% (condensation) within respectively 5 and 13 days – and stays at this level for the rest of the measurement period. This is not unexpected in regard to Hygrodiode, which has the highest Sd-value at high RH-levels (Sd ≈ 1 m according to manufacturer). When it comes to Intello this is on the other hand unexpected, at least in regard to the fact that Intello and Delta-Novaflexx has almost the same Sd-value at high RH according to the manufacturers, i.e. respectively Sd ≈ 0,32 m and 0,28 m at RH = 90%. We also see that the configuration with Novaflexx w/cotton reaches 100% during cycles 2, 3 and 5. From Figure 7 we see that the RH at the interface between the SVBs and the insulation is following a diurnal variation, following the diurnal temperature variation. When the diurnal temperature variation is low, the RH-variation is also low - and vice versa. 2 1,5 1 0,5 0 0 10 20 30 40 50 60 Days Figure 5. Total water content in the six roof configurations during the test period. We can also see from Figure 5 that the drying rate for the Delta-Novaflexx and AirGuard Smart configurations are relatively low during the first 13 days when the external surface temperature is only 25 °C, but is increasing when the external temperature is increased to 33 and 30 °C the next 17 days. In Figure 6 is shown the total drying from the start at several different points of times. We see that the configurations with AirGuard Smart and DeltaNovaflexx with plywood sheathing have almost the same total drying after 9 and 30 days. After 60 days Delta-Novaflexx has a higher total drying than AirGuard Smart, but this is probably due to DeltaNovaflexx starting at a higher moisture level and the specimens approaching moisture equilibrium. Intello and Hygrodiode show almost the same total drying during the whole period. 331 Novaflexx AirGuard Smart w/cotton Hygrodiode Temperature AirGuard Smart 100 45 90 40 80 35 70 30 60 25 50 12.10.12 0 Temperature (°C) RH (%) Intello Novaflexx w/cotton 20 22.10.12 10 12.11.12 30 2.11.12 20 23.11.12 40 3.12.12 50 60 Days Figure 7. RH at the interface between mineral wool and SVB shown together with the temperature at the roof surface (heating foil). high when measuring in plywood. Even if the wood moisture meter is set on a calibration curve for spruce which this plywood sheathing is made of, one will usually measure higher moisture contents than the real ones, this difference being higher for higher moisture contents (Geving and Holme, 2010). The results will however give an indication of the drying rate, even with these uncertainties. We see that the difference between the various SVBs is not big, especially the first 20 days. The last half of the measurement period we do however observe that the Delta-Novaflexx and AirGuard Smart has the highest drying rate, and of these two the AirGuard Smart configuration has the fastest drying rate down to below 30 weight%. In reality the difference between AirGuard Smart and the other SVBs may be higher, due to night-time undercooling of the roof surface caused by black-body radiation to the sky. In clear summer nights the exterior surface temperature will often drop below the outdoor air temperature. This will give an outward directed vapour flow during the night, which will probably be higher for those configurations with an RH close to 100% in the interface between insulation and SVB, such as Hygrodiode, Intello and partly Delta-Novaflexx. When comparing Delta-Novaflexx and AirGuard Smart with plywood sheathing we can observe from Figure 7 that the RH is lower for AirGuard Smart. This is probably due to the lower Sd-value of AirGuard Smart at high RH levels (e.g. Sd ≈ 0,07 m at RH = 88% according to manufacturer), i.e. approximately 4 times more vapour open than DeltaNovaflexx at that RH-level. It can also be observed that the configurations without plywood sheathing (w/cotton) have higher RH levels than the ones with plywood during the first part of the measurement period. This is probably caused by the fact that the moisture release from the plywood sheathing is slowed down by both the moisture capacity and the vapour resistance of the material itself, as was discussed earlier. The RH of the lower surface of the plywood will probably drop below 100% after some days of drying, thereby decreasing the vapour pressure difference and drying potential across the insulation cavity. The cotton layer however will maintain an RH of 100 % till most of the added water has dried, which happens after approximately 37 and 45 days for AirGuard Smart and Delta-Novaflexx respectively (as also can be seen from Figure 5). The moisture content in the plywood layer is shown in Figure 8. It should be noted that most of the measurements shows higher levels than 30 weight% which is normally considered the maximum level for achieving acceptable accuracy with this measurement method – and also very much higher than the actual built-in moisture content measured by gravimetric method (27-38 weight%). The difference between the initial moisture content measured with the electrical resistance method and the gravimetric method is probably caused by the fact that the measurement uncertainties are relatively 332 Moisture content (weight%) 70 60 Intello Novaflexx Hygrodiode AirGuard Smart The SVBs are generally used to reduce the risk for moisture damages such as rot and mould growth. The occurring RH within the construction during the drying period is therefore also of interest – not only the drying rate. From Figure 7 we could observe that the RH at the interface between the insulation and the SVB were 100% during most of the measurement period for Intello and Hygrodiode and almost half of the period for Delta-Novaflexx w/cotton. The RH for the AirGuard Smart and Delta-Novaflexx (w/plywood sheathing) however never reached so high humidity levels – thereby having lower risks for mould growth at the interface between insulation and SVB and at the bottom part of any wooden rafters. There are several limitations or shortcomings with this laboratory experiment. One is that the effect of undercooling of the external surface during the night caused by black-body radiation to the sky is not included. This overestimates the drying to the interior since in reality there may be some outward directed vapour flow during night. In addition the geometry of the experiment is a simplification, since rafters not are included in the test boxes. The rafters will represent an absorptive material in the insulation layer and may influence on the moisture distribution and diurnal fluctuations. How the SVBs function for shaded conditions is an important issue, since the inward driving potential for vapour flow may be considerably reduced for shaded parts of roofs (or walls). The results of this study however shows that one may have a considerable drying effect also under shaded conditions when applying SVBs with sufficiently low Sd-values at high RH-levels; such as Delta-Novaflexx and AirGuard Smart. 50 40 30 20 0 20 40 60 Days Figure 8. Moisture content in plywood board measured with electrical resistance method. Average of measurement points on top and bottom of the board. Note that the measurement values are higher than the real values due to high measurement uncertainty. 4 DISCUSSION Geving, Stellander and Uvsløkk (2013) performed laboratory measurements on similar roof configurations – but exposed to direct solar radiation. The maximum external temperature for the diurnal cycle (that was repeated 144 days) was 45°C - giving a much higher daily drying potential. In table 3 is compared the average drying rate over the first 30 days for both experiments. The experiment of Geving, Stellander and Uvsløkk (2013) also included a configuration with PE-foil (Sd ≈ 70 m), which is also included in the table for comparison. The drying rate for shaded conditions is reduced only with approximately 40% compared to the unshaded conditions for all the SVBs, except for Hygrodiode where the drying rate is reduced 75%. This high reduction for the Hygrodiode may be caused by its capillary properties – the capillary effect is probably functioning best under sunny conditions where the condensed layer at the Hygrodiode is thicker than under shaded conditions. Also if we compare the drying rate for the configuration with PE-foil (for unshaded conditions) with the SVBs for shaded conditions, we see that all SVBs have a considerable drying effect – though somewhat limited for Intello and Hygrodiode. 5 CONCLUSIONS Nowadays there exist several types of SVBs. They have various properties in regard to their range of vapour resistance at various RH-levels. In this study a laboratory experiment was performed to investigate the drying rate for various types of SVBs during summer conditions for an unventilated wooden roof shaded from direct solar radiation. The results showed that the various SVBs performed differently in regard to drying rate, depending on the vapour resistance and its dependency of RH. The results of this study showed that one may have a considerable drying effect also under shaded conditions when applying SVBs with sufficiently low Sd-values at high RH-levels Table 3. Comparison of average drying rate the first 30 days for this experiment (shaded conditions) and the experiment of Geving, Stellander and Uvsløkk (2013) (unshaded conditions). Type SVB Drying rate (g/m2day) Unshaded Shaded Delta- Novaflexx 46 29 Intello 12 7 AirGuard Smart 47 28 Hygrodiode 29 7 PE-foil 2 - 333 Pro Clima. 2006. pro clima INTELLO  System Brochure. www.proclima.com. Rode C., Christensen G., and Vesterløkke M. 1993. Moisture performance of flat, non-ventilated, wooden roof cassettes. Proceedings of the 3rd Symposium on Building Physics in the Nordic Countries, Copenhagen, Denmark, September 13-15, 623-631. 6 REFERENCES Bludau C., and Künzel H.M. 2009. Flat roofs in cold climates – Climatic limits for building flat roofs with a permeable vapour retarder. Proceedings of the 6th International Conference on Cold Climate HVAC, Sisimiut, Greenland, 2009. Bludau C., Schmidt T., and Künzel H.M. 2010. Hygrothermal effects in lightweight roofs shaded by PV-elements. Proceedings of Thermophysics 2010, Valice, Czech Republic, 3-5. November 2010, 56-62. Brandt E., and Bunch-Nielsen T.. 2006. Mold in building constructions in the Nordic countries – With emphasis on roofs. Proceedings of the RCI 21st International Convention and Trade Show, Phoenix, Arizona, March 1, 51-57.  DuPont. 2012. Tyvek Product Brochure. http://construction.tyvek.co.uk. Fraunhofer IBP. 2010. WUFI 1D (Version 5.1) [Computer Program]. Fraunhofer IBP, Holzkirchen, Germany. Geving S., and Holme J. 2010. The drying potential and risk for mould growth in compact wood frame roofs with builtin moisture. Journal of Building Physics, 33:249-269. Geving S., and Holme J. 2012. Vapor retarders in wood frame walls – do they have any effect on the drying capability? Proceedings of the 5th International Building Physics Conference, Kyoto, Japan, May 28-31. Geving S., Stellander M., and Uvsløkk S. 2013. Smart vapour barriers in compact wood frame roofs. Submitted to Buildings XII Conference – Thermal Performance of Ecterior Envelopes of Whole Buildings, December 1-5, 2013, Clearwater Beach, Florida. Ghazi Wakili K., and Frank T. 2004. A humidity dependent vapour retarder in non-ventilated flat roofs. In situ measurements and numerical analysis. Indoor and Built Environment, 13:433-441. Icopal, 2005. Hygrodiode  Product Brochure. www.icopal.dk. Kaufmann A., Künzel H.M., and Radoń J. 2006. Preventing moisture problems in retrofitted pitched roofs. Architectura, 5: 69-79. Korsgaard V. 1985. Hygro diode membrane: A new vapor barrier. ASHRAE/DOE/BTECC Conference: Thermal Performance of the Exterior Envelopes, Clearwater Beach, Florida, December 2-5. Korsgaard V., and Pedersen C.R. 1990. Moisture content and distribution in flat roofs with polyethylene or Hygro Diode vapour retarder. Proceedings of the 2nd Symposium on Building Physics in the Nordic Countries, Trondheim, Norway, August 20-22, 253-258. Künzel H.M. 1996. Humidity controlled vapour retarders reduce risk of moisture damages. Proceedings of the 4th Symposium on Building Physics in the Nordic Countries, Espoo, Finland, September 9-10.,447-454. Künzel H.M. 1998. More moisture load Tolerance of construction assemblies through the application of a smart vapor retarder. Proceedings of ASHRAE VII International Conference on Thermal Performance of the Exterior Envelopes of Whole Buildings, Clearwater Beach, Florida, USA, 129132. Künzel H.M. 1999. Flexible vapor control solves moisture problems of building assemblies – smart retarder to repace the conventional PE-film. Journal of Thermal Envelope and Building Science, 23: 95-102. Künzel H.M., and Leimer H-P. 2001. Performance of innovative vapor Retarders under summer conditions. ASHRAE Transactions, 107 (1). Pedersen C.R., Petrie T.W., Courville G.E., Desjarlais A.O., Childs P.W., and Wilkes K.E. 1992. Moisture effects in low-slope roofs: Drying rates after addition with various vapor retarders. Report, Oak Ridge National Laboratory, Tennessee, USA. 334 Computer modelling to evaluate the risks of damage to objects exposed to varying indoor climate conditions in the past, present, and future Z. Huijbregts, M.H.J. Martens, A.W.M. van Schijndel & H.L. Schellen Eindhoven University of Technology, Department of the Built Environment, Eindhoven, the Netherlands ABSTRACT: Within the European project Climate for Culture, we are trying to evaluate the risks for valuable historical objects exposed to changing indoor conditions due to outdoor climate change. We have used building simulation models to calculate the expected indoor climate in historic buildings subjected to a varying outdoor climate. This outdoor climate can be constructed from a historical data file (more than 100 years ago), recent meteorological data (less than 50 year ago) or predicted future data from outdoor climate scenarios (for the next 100 years). The simulated indoor climate is coupled with damage functions to predict the damage risk for objects directly. This paper deals with the modelling approach and shows the potential for damage risk evaluation. The historic, present and future indoor climate conditions in a characteristic historic building have been modelled and the damage risk to historical objects has been compared when the building is virtually placed on 468 different locations in Europe. In this way, the impact of future climate change on the indoor climate conditions in a building and damage risk to its collection can be assessed for areas all over Europe. compiled by meteorological institutes over the past 50 years. For example, Meteonorm (Meteotest 2009) provides hourly weather data files for over 100 weather stations across Europe that represent an average climate for that location over a period of one year. Up to now, these data have generally been used to predict the indoor climate and energy use of buildings in the design stage of a building, or later. A study that compared the future indoor climate within a historic house for several locations in Europe and coupled the results with damage functions for paper and salts can be found in (Lankester & Brimblecombe 2012). The authors of this article derived a linear transfer function between the indoor and outdoor climate for each month during their research period. However, linear transfer functions may not accurately take into account climate control systems and variable internal heat loads due to irregular use of shading devices and differences in building use. With hygrothermal building simulation models it is possible to vary internal heat loads, ventilation rate and climate control set points per hour to obtain an accurate prediction of the indoor climate. Two preliminary studies that used hygrothermal building simulation models to predict the impact of future climate change on two historic buildings in Western Europe and that compared the present indoor climate conditions in a historic church for different locations in Europe can be found in 1 INTRODUCTION Future outdoor climate scenarios indicate that the outdoor climate is changing and will continue to do so in the near future. This change in outdoor climate might greatly affect vulnerable historic buildings and their valuable interiors and objects. In the EU project Noah’s Ark, the impact of global climate change on European built heritage and cultural landscapes was analysed (Sabbioni et al. 2010). This project mainly focused on the damage potential of outdoor climate change to historic building facades. The results indicated that buildings might be at risk in many areas in Europe, due to increased amounts of precipitation as well as longer periods of consecutive precipitation. Within the European project Climate for Culture (CfC), we want to assess the impact of climate change on the indoor climate in historic buildings. We have used outdoor climate data from historical and recent weather files and numerical weather prediction models, provided by our partners in the CfC project, to analyse outdoor climate changes. The outcomes are data files with (hourly) values of the historical, present and future outdoor climate. We had already developed building physical computer simulation models to predict indoor climate of buildings, using (hourly measured) weather data files 335 (Huijbregts et al. 2012a) and (Huijbregts et al. 2012b), respectively. In the present paper, historical, present and future outdoor climate data are used as input for building simulation models to predict the changing indoor climate in a historic building all over Europe. The impact of the predicted indoor climate conditions on possible future damage to valuable objects has been estimated by using so-called damage functions. Damage functions have been derived from literature and are based on laboratory experiments. In these experiments, typical objects were subjected to artificial indoor climate conditions. The damage risk to the objects is based on the way the object responds to (or the damage caused by) certain indoor climate variations. To date, there is little literature providing a coherent approach from outdoor to indoor climate, microclimate and predicted damage to objects. In this paper we attempt both to provide and to evaluate such an approach. In Section 2, we will describe the computer simulation method that was used to predict the indoor climate as a result of the outdoor climate and building properties, the meteorological data that were used and the historic reference building that was investigated. In Section 3, we will describe the damage functions, used to evaluate the risk on deterioration of valuable objects that is induced by the indoor climate conditions. Furthermore, we will describe results of one case study using the simulation approach for the indoor climate evaluation in the past, present and future (Section 4). The conclusion and discussion will be given in Section 5. [%], wind velocity [m/s] and wind direction [°]. The indoor climate (derived from HAMBase) is characterized by three properties that are assumed to be uniform in the zone: air temperature, radiant temperature and RH. 2.2 Artificially generated historical climate data To evaluate the effects of climate change over a much longer period, we can make use of historical measured handwritten, and afterwards digitized, weather data. From the KNMI database, ancient climatic data from the 1850s can be obtained for six weather station locations in the Netherlands. These data consist of meteorological data that involve wind directions, wind pressure, temperatures, daily precipitation, surface air pressure, cloud cover and RH and that were manually recorded three times a day. However, to use these data in a simulation model, we need semi-continuous data with time intervals of an hour. So the manually recorded data had to be interpolated to hourly data. The interpolation was calculated based on hourly measured data from KNMI over the years 1971 to 2005: the recent past files. In general, the MATLAB interpolation function balances the smoothness of the missing data in the ancient files with the recent past files. As mentioned previously, the data in the ancient files are based on three time intervals, and the interpolation will estimate the values that are in-between these known data points to match the unknown missing data with known data from the recent past files. The data were selected on the basis of a best fit in a period in the recent past files, comparable with the period during the examined year, i.e. with comparable sun elevation and azimuth. The interpolation searched for the same value at a certain time within the same time interval of the given available data in the ancient climate file. In this way, historical outdoor climate data files for the years 1881 until 1896 were created. For example, on 1 January 1881, the cloud cover at one of three recorded times was 5. So, in the weather file from 1971 to 2005, also on 1 January, the interpolation function would look for the same value of cloud cover of 5 in a comparable time interval. This cloud cover value was used to calculate the ratio of the solar radiation in the historical weather file. The missing hourly values for temperature and RH were calculated by linear interpolation. 2 METHOD 2.1 Simulation model The indoor climate simulation model HAMBase (De Wit 2006) was used to evaluate the climate conditions within a historic building. With HAMBase, the thermal and hygric indoor climate and energy use for heating and cooling of multi-zone buildings can be simulated using building material properties, outdoor air temperature and relative humidity (RH), diffuse solar radiation on a horizontal plane, direct normal radiation and cloud coverage. The program makes use of a standard weather file with boundary conditions for air temperature, RH, wind velocity and direction, solar radiation and cloud cover. These data are usually derived from measured weather data from meteorological weather stations. In our case the data were obtained from KNMI (Royal Netherlands Meteorological Institute 2013). We made use of hourly measured data from 1960-2012 in a standard format, delivered by KNMI. For these so-called recent past files, the measured meteorological are: diffuse solar radiation [W/m2], air temperature outside [°C], direct solar radiation (plane normal to the direction) [W/m2], cloud coverage [1...8], RH outside 2.3 Present meteorological data Meteorological data from the regional climate model REMO of the Max Planck Institute for Meteorology (Jacob 2012) were used to analyse the outdoor climate all over Europe. The REMO model is based on the former Europe model: a numerical weather prediction model from the German Weather Service (Majewski 1991). REMO can be used for weather forecast and future climate simulations on a grid with a minimum horizontal resolution of approxi336 above the consistory. This study focuses on the results for the main zone in the church: the sanctuary. A comparison between the measured and simulated indoor temperature, RH and humidity ratio is presented in Figure 3. It should be noticed that in the simulation model, the number of visitors during ceremonies was kept constant and only ceremonies on Sunday morning were taken into account. Additionally, a constant value was assumed for the air exchange rate. The simulated temperature generally varied within ±2°C of the measured temperature, RH was predicted within ±10% of the measurements and the humidity ratio was predicted within ±2g/kg of the measurements. mately 10km. In REMO, climate data from the reference period 1961 to 1990 were used for a control run. The weather forecast, however, could not be predicted for individual days, but it is possible to generate an assumption for the average conditions for an area and the probability and magnitudes of the deviations from this average. The general averaging period is 30 years. REMO data for air temperature, surface temperature, RH, precipitation, wind speed and direction and global radiation were provided with a temporal resolution of one hour. For the evaluation of the present climate, a dataset was composed of climate measurements from multiple weather stations throughout Europe from 1960 until 1990 to represent characteristic climates for the different regions and locations. The meteorological data were thereafter interpolated on a regular grid over Europe. Figure 1 shows an overview of locations and altitudes of the 468 grid points which were used. Some weather locations are located at very high altitudes (>1000m). The weather data provided for these locations may considerably differ from weather data of nearby stations at lower altitudes. The Alps and Dolomites are examples of such locations. 2.4 Future outdoor climate scenarios REMO recently produced ‘future’ outdoor climate scenarios for two 31-year periods: near future (20202050) and far future (2070-2100) for all 468 grid locations over Europe. The scenarios were based on the IPCC A1B emission scenario for the period 2001-2100 (IPCC 2007). This emission scenario assumes a world of very rapid economic growth, a global population that peaks in the mid-century, and a rapid introduction of new and more efficient technologies that balance between fossil intensive energy sources and non-fossil energy sources. Figure 1. Overview of the locations and altitude of the modelled meteorological datasets on a uniform grid over Europe. 2.5 Reference building Three requirements have been defined for the reference building: it has to be a historic building, it has to represent a typical building style that can be found all over Europe and it has to be in use for its original function. A small church near Eindhoven, the Netherlands, was selected (Fig. 2). The church, which was built in the nineteenth century, has been registered as a state monument since 1968. Massive brick walls, slate roofing, and single glazing characterize the building. The church is frequently used for services, marriages and funerals. Continuous on-site measurements of the air temperature, surface temperature, and RH at various locations in and around the church were started in March 2011. In addition, measurements of the air exchange rate and heat flux through the walls were carried out. A HAMBase model was created of the entire church, consisting of four zones: the sanctuary, the consistory, the attic above the sanctuary and the attic Figure 2. The historical reference building. 337 3.1.2 Chemical degradation The concept of the Lifetime Multiplier (LM) is used to describe the time an object is usable, compared to a reference indoor condition (20°C and 50%RH). Apart from T and RH, the LM also depends on the activation energy, which is a material property (Michalski 2002). A small risk on chemical damage may occur when LM > 1, a medium risk may occur when 0.75 ≤ LM < 1 and a high risk is predicted when LM ≤ 0.75. 3.1.3 Mechanical degradation Hygroscopic materials react to changes in RH by absorbing or desorbing moisture from the air. The changes in moisture content imply dimensional changes of the materials. If these materials are not free to expand or contract, stresses occur in the object, which may lead to damage by mechanical degradation. As panel paintings are representative objects in many historic buildings, the hygroscopic and mechanical behaviour of panel paintings have been subject of a number of extensive studies, e.g. (Mecklenburg et al. 1998), (Rachwal et al. 2012a) and (Rachwal et al. 2012b). In this paper, therefore, panel paintings are chosen as reference objects for mechanical degradation. For this kind of paintings, two types of mechanical damage are important: damage to the wood support and damage to the pictorial layer. Damage to the wood support may occur when the entire object responds to a slow change of RH over time. The dimensionally changes of the object may be hindered by the construction of the object and lead to damage, such as cracks. Damage to the pictorial layer may occur when RH variations last longer than the response time of the panel. The moisture content within the panel changes and the object will swell or shrink. As the response of the gesso layer to RH variations is very fast, the mismatch in the response of gesso and the unrestrained wood support can lead to fracturing of the pictorial layer. Figure 3. Validation of the indoor climate simulation model of the sanctuary in the reference building. 3 DAMAGE FUNCTIONS Martens (2012) developed a new method to assess damage in objects due to the indoor climate. His method is based on the indoor climate an object is experiencing due to the response time of the object. The response time is defined as the time, needed for an object to react for 95% to a step change in RH. The RH of the object is derived from the measured indoor climate, using the response time of the object according to Equation 1: RH response,i = RH response,i − 1 + 3 ⋅ RH i n (1) 3 1+ n where n equals the number of measured data points in the response time and RHresponse of the object at time i is determined by taking the previous RHresponse at (i-1), adding a fraction of the current RH in the room and dividing by 1 plus that fraction. For a 95% reaction, the fraction equals 3/n. 4 RESULTS 4.1 Predicted historical indoor climate in the reference building A prediction of the historical indoor climate in the reference building was generated by combining the HAMBase model with the artificially generated historical climate data as was described in the Method section. In the simulation model, it was taken into account that the building remained unheated and 50 persons on average attended weekly ceremonies on Sunday morning. The estimated temperature, RH and humidity ratio in the year 1882 are shown in Figure 4. It is predicted that the minimum indoor temperature is slightly below freezing point and that 3.1.1 Biological degradation A method of Sedlbauer (Sedlbauer 2001) is used to determine biological degradation by fungal growth. Combinations of temperature and RH determine whether the fungus germinates or grows. 338 the maximum indoor temperature is around 25°C. High RH values are predicted: RH remains above 60% for most of the year and regularly exceeds 90% in winter. tween -5 and 25°C, while the mean humidity ratio varies between approximately 3 and 9g/kg. In Northern Europe, the average indoor climate in the 31-year period is characterized by low mean temperatures (-5 to 10°C) and mean humidity ratios varying between 3 to 6g/kg, which leads to high mean RH values (80 to 90%). The indoor climate in the Mediterranean area is generally warm (10 to 20°C) and has a mean humidity ratio of 4 to 7g/kg inland and 7 to 9g/kg in coastal areas, leading to a medium mean RH (40 to 60%). In between, the area around the United Kingdom and Ireland shows a temperate climate (5 to 15°C) and a medium mean humidity ratio (5 to 7g/kg) leading to a high mean RH (75 to 80%). 4.3 Future indoor climate in the reference building, virtually placed all over Europe The impact of future climate change on the indoor climate conditions in the reference building was predicted by calculating the difference in mean temperature and humidity ratio between the recent past, near future and far future. Figure 7 shows that in the near future, the average indoor temperature may increase by approximately 1°C in Western Europe and 1 to 2°C in Southern, Eastern and Northern Europe. A small increase in the mean humidity ratio is predicted in all areas, varying between circa 0 to 0.4g/kg in Southern Europe and 0.4 to 0.8g/kg in Eastern Europe (Fig. 8). Consequently, the mean RH may slightly decrease in most areas, in particular in Southern Europe. Larger changes are predicted in far future: the mean indoor temperature increase in Western Europe is approximately 2°C, while a mean indoor temperature rise up to 4°C may occur in Northern and Southern Europe (Fig. 9). The predicted mean humidity ratio change is highest in Eastern Europe and the coastal areas (1.2 to 1.6g/kg) and smallest in parts of Great Britain, Norway and in the inlands of Southern Europe (Figure 10). Figure 4. Predicted historic indoor climate conditions inside the sanctuary in 1882. 4.2 Present indoor climate in the reference building, virtually placed all over Europe The calculated meteorological data and damage risks were interpolated to a grid over Europe. The grid used has a resolution of 376x226 data points and covers the area between 30°N to 75°N and 28°W to 45°E. The mean indoor temperature and humidity ratio in the recent past (1960-1990) are shown in Figures 5-6. The mean temperature inside the church when it is virtually placed all over Europe varies be- Figure 5. Mean temperature between 1960 and 1990 inside the sanctuary when the church is virtually placed all over Europe. Figure 6. Mean humidity ratio between 1960 and 1990 inside the sanctuary when the church is virtually placed all over Europe. 339 Figure 7. Mean indoor temperature change from recent past to near future (a positive change means a temperature increase in near future). Figure 8. Mean indoor humidity ratio change from recent past to near future (a positive change means an increasing humidity ratio in near future). Figure 9. Mean indoor temperature change from recent past to far future (a positive change means a temperature increase in far future). Figure 10. Mean indoor humidity ratio change from recent past to far future (a positive change means an increasing humidity ratio in far future). creases in near future and far future, which causes an increased risk on chemical degradation of objects particularly in coastal areas in Southern and Western Europe (Fig. 12). No location in Europe has been found where the indoor climate conditions in the reference building may prevent mechanical degradation of the wood support or pictorial layer of panel paintings. Damage to the wood support is likely in some areas in Northern Europe in the three periods, but no consistency is found between the locations where this high damage risk is predicted (Fig. 13). Damage to the pictorial layer is likely in the recent past and near future in many areas in Northern, Eastern, and Southern Europe. In far future, damage is likely in almost all areas (Fig. 14). 4.4 Damage functions The previously described damage functions were used to predict the risks of biological, chemical, and mechanical degradation, based on the calculated indoor climate for the unheated reference building in the recent past, near future and far future. For the recent past weather data, a high risk on mould growth is predicted in Great Britain and the coastal areas of Western Europe and Scandinavia. The risk on mould growth may considerably increase in near future and far future in and around these areas, while the predicted mould growth risk in Southern Europe remains low (Fig. 11). In the recent past, LM > 1 for most areas in Europe, except for part of the coastal areas in Southern Europe. The LM gradually de- 340 Figure 11. Predicted mould growth risk in recent past, near future and far future. Figure 12. Predicted average annual lifetime multiplier in recent past, near future and far future. Figure 13. Predicted mechanical degradation risk of wood support in recent past, near future and far future. Figure 14. Predicted mechanical degradation risk of pictorial layer in recent past, near future and far future. Based on the predicted indoor climate conditions, the damage potential of biological, chemical and mechanical degradation was evaluated for the recent past, near future and far future. The preliminary results suggest that it could be possible to predict the indoor climate conditions and risk for damage in a building over a large area, using regional climate data from the past, present and future. Based on the applied future outdoor climate scenario, a small increase in indoor temperature and 5 CONCLUSION AND DISCUSSION This study presents a modelling approach to predict the historical, present and future indoor climate conditions in a historic building, when it is virtually subjected to an outdoor climate at various locations over Europe. The indoor climate conditions were calculated by a hygrothermal building simulation model. 341 Climate Change Pachauri, R. K., & Reisinger, A. eds., Geneva: IPCC. Jacob D. 2012. The Regional Model - REMO. Max Planck Institute for Meteorolgy. Available at: http://www.mpimet.mpg.de [Accessed January 31, 2013]. Lankester P., and Brimblecombe P. 2012. Future thermohygrometric climate within historic houses. Journal of Cultural Heritage, 13, pp.1–6. Majewski D. 1991. The Europa-Modell of the Deutscher Wetterdienst. In ECMWF Seminar on numerical methods in atmospheric models. pp. 147–191. Martens M.H.J. 2012. Climate risk assessment in museums: degradation risks determined from temperature and relative humidity data PhD Thesis., Eindhoven: Eindhoven University of Technology. Mecklenburg M.F., Tumosa C.S., and Erhardt D. 1998. Structural response of painted wood surfaces to changes in ambient relative humidity. In Painted wood: History and conservation: Proceedings of a symposium organized by the Wooden Artifacts Group of the American Institute for Conservation of Historic and Artistic Works and the Foundation of the AIC. Williamsburg, Virginia. Meteotest, 2009. Meteonorm dataset. Available at: http://www.meteonorm.com [Accessed February 21, 2013]. Michalski S. 2002. Double the life for each five-degree drop, more than double the life for each halving of relative humidity. In Preprints of 13th Triennal Meeting of ICOM Committee for Conservation. Rio de Janeiro, pp. 66–72. Rachwal B., Bratasz Ł., Krzemień L., et al. 2012a. Fatigue damage of the gesso layer in panel paintings subjected to changing climate conditions. Strain, 48(6), pp.474–481. Rachwal B., Bratasz Ł., Łukomski M., et al. 2012b. Response of wood supports in panel paintings subjected to changing climate conditions. Strain, 48(5), pp.366–374. Royal Netherlands Meteorological Institute, 2013. KNMI Data Centre. Available at: http://www.knmi.nl [Accessed January 31, 2013]. Sabbioni C., Brimblecombe P., and Cassar M., eds. 2010. The atlas of climate change impact on European cultural heritage: scientific analysis and management strategies, London: The Anthem-European Union Series. Sedlbauer K. 2001. Prediction of mould fungus formation on the surface of and inside building components, Holzkirchen: Fraunhofer Institute of Building Physics. De Wit M.H. 2006. HAMBase: Heat, Air and Moisture model for building and systems evaluation, Eindhoven: Eindhoven University of Technology. humidity ratio is predicted in the near future, while a considerable rise in temperature and humidity ratio may occur in far future. Damage evaluation shows that there are no places in Europe where no damage to objects is to be expected in recent past, near future and far future. In cold, humid climates, the risk for chemical degradation is regularly low, while the risk for mould growth and mechanical damage is rather high. In contrast, in warmer, dry climates, mould growth risks are rather low, while chemical and mechanical degradation are more important. Climate change may considerably increase the mould growth risk in Northern and Western Europe. Additionally, a higher risk on chemical degradation may occur particularly around coastal areas in Western and Southern Europe. No consistent impact of climate change on the predicted mechanical degradation of panel paintings was found. One of the most critical problems in using this approach is the uncertainty in people’s use of the building and its HVAC systems. Also the fact that materials will adapt to the long term local situation is not taken into account. Besides that, the current outdoor future climate scenario is based on only one IPCC emission scenario, which means that there is a high uncertainty in these data. In the near future, more generic building types for different areas in Europe will be selected to acquire more appropriate reference buildings for each location. Furthermore, the microclimate around objects could have an essential influence on the risk evaluation of objects and should be object of research in future. More objects will be included in the potential damage analysis, e.g. wooden organs, and the impact of climate control systems and climate adaptive measures will be investigated as well. The presented risk maps are not yet suitable for climate management in historic buildings, but should be seen as illustrative examples of potential impacts and risks. 6 ACKNOWLEDGEMENTS This work was supported by European Commission funding through the EU Climate for Culture project 226973 within FP7-ENV-2008-1. 7 REFERENCES Huijbregts Z., Kramer R.P., et al. 2012. A proposed method to assess the damage risk of future climate change to museum objects in historic buildings. Building and Environment, 55, pp.43–56. Huijbregts Z., Martens M.H.J., et al. 2012. Damage risk assessment of museum objects in historic buildings due to shifting climate zones in Europe. In Proceedings of the 5th International Building Physics Conference. Kyoto, pp. 1271–1278. IPCC, 2007. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on 342 Thermal comfort of individual rooms in the design of commercial buildings K. Nowak, K. Nowak-Dzieszko, M. Rojewska-Warchał Cracow University of Technology, Division of Building and Building Physics, Cracow, Poland ABSTRACT: This paper presents the results of the annual numerical simulations in the design of commercial buildings. The calculations were carried out in the Energy Plus, tightly integrated within the Design Builder program, which provides the simulations of both building envelope and building interiors. The simulations conducted for the Polish climatic conditions allowed the evaluation of the thermal comfort of the entire building and of the particular rooms. Influence of different solutions, insulation, shadings and occupancy density on the internal thermal conditions were analysed in this paper. rooms often become critical due to focusing only on energy needs. In order to avoid those critical conditions the thermal comfort in such south facing rooms should be analyzed to determine the risk of overheating. According to Polish regulations the building should be designed to avoid risk of overheating during the summer months. 1 INTRODUCTION It is very rare in the design process to take into consideration the requirements connected with the overheating effect. This issue is closely related to the thermal comfort of the building, especially during the summer months. The microclimate of the interior space is the combined effect of the design process, erection and the utilization of particular rooms. Thermal comfort is affected by human activity, clothing insulation and the environmental parameters such as air temperature, average radiation temperature, air flow speed and relative humidity. The evaluation of thermal comfort is based on the PMV (Predicted Mean Vote) and PPD (Predicted Percentage of Dissatisfied) indexes. Based on the simulations, the authors determined the influence of different materials, structural solutions, glazing, and sunshades on the thermal comfort of separate rooms during the summer season. Both new construction and renovated structures were evaluated. The article emphasizes the need to consider thermal comfort in the design process of individual rooms taking into account the orientation, location, and room usage. 3 THERMAL COMFORT Thermal comfort by ASHRAE 55 is “the condition of mind in which satisfaction is expressed with the thermal environment”. It is related to the thermal balance of the body which is affected by different parameters: personal (activity level or metabolic rate, thermal resistance of clothing) and environmental (air temperature, mean radiant temperature, relative air velocity, relative humidity). There are several different methods used for estimation of thermal comfort. International standard PN-EN ISO 7730 „Ergonomics of the thermal environment. Analytical determination and interpretation of thermal comfort using calculation of the PMV and PPD indices and local thermal comfort criteria” and ASHRAE 55 use Fanger’s method. The Fanger’s PMV (Predicted Mean Vote) method combines the following environmental features: air temperature, air velocity, mean radiant temperature and relative humidity and two personal variables: clothing insulation and activity level into the index that can be used to predict the average thermal 2 MICROCLIMATE OF THE ROOMS The national regulations regarding building design process draw attention to the expected energy consumption, heat losses and heat gains. In order to increase the solar gains and reduce the heating need, large window areas are located in the southern rooms. The thermal conditions in those specific 343 sensation of a large group of people. The ASHRAE thermal sensation 7 level scale with values between 3 and 3 describes the thermal sensation between “hot” and “cold”. In this model, all major modes of energy losses from the human body are taken into account and the person is assumed to be at the steady state condition. The most well known two-node thermal model is Pierce’s model based on two energy balance equations, one for core node and one for skin node. It takes into account the effects of heat accumulation, physical work done, conductive and convective heat transfer by blood, flow between the core and shell, heat loss caused by respiration, metabolic heat generated during physical activities and the heat exchange processes between the skin and its surroundings by means of convection, radiation and evaporation of moisture. The KSU (Kansas State University) two-node model is based on the changes that occur in the thermal conductance between the core and the skin temperature in cold environments, and in warm environments, it is based on changes in the skin wettedness. This model results in a thermal sensation vote (TSV) that uses a similar scale as the Fanger model PMV and the Pierce model. Fanger PMV method overestimates the warmth sensation at room temperatures above about 27°C and underestimates the cooling effect of increased air movement. All three referenced models are used by Energy Plus program to describe the thermal comfort. Figure 1. Model view of the building, south elevation. Percentage share of glazing areas at the elevations is as follows: N – 69% S – 44% E – 46%. W – 46% Exterior walls made of solid brick with plaster at both sides: U = 1,51 [W/m2K] (maximum U value allowed by polish standard requirements for external walls in this kind of building is 0,3 [W/m2K]), concrete flat roof, insulated with 5 cm of styrofoam: U=0,59 [W/m2K], slab on ground insulated with 5 cm of styrofoam: U= 0,50 [W/m2K], double glazing windows: U = 1,7 [W/m2K]. Building with the natural ventilation, gas heating system with convection heaters. Communication area located in the center part of each building story, office rooms located at the south side of building (similar at each level). The area of office rooms is relatively small but the ratio of the window area to the single room area is significant. Lecture classes located on the south, north-east and north-west sides of the building. The calculations were carried out in the Design Builder v.3. Program has been specifically developed around Energy Plus allowing the simulation of the building envelope and building interiors. It provides extensive databases of building materials, constructions, window panes, glazing units, shadings and blinds. The simulations conducted for the Polish climatic conditions (building located in Cracow) allowed the evaluation of the thermal comfort of the entire building and of the particular rooms. 4 BUILDING SIMULATION The aim of the building simulations was to assess the importance of different factors influencing the thermal comfort of the entire building, of separate rooms and possible overheating during summer months. The simulation was conducted for the academic building, plan area 72,9m x 21,7m; 10,5 m high with three levels. In the chosen building there are many south facing windows. Visualization of the south building elevation is presented in Figure 1. 344 4. Metabolic activity: factor 0,9, winter clothing – clo=1.0, summer clothing clo=0,5. 5. Ventilation requirements per polish national standards for commercial buildings, 20 m3/hour per person, based on occupancy overload assumed as follows: - Office rooms – one ac per hour - Lecture classes – five ac per hour 5 TESTS RESULTS 5.1 Entire building in the service conditions The simulation results, Figure 3, have shown that on a few days between 15th of May and 15th of September the average interior air temperatures of the entire building exceed 30°C and the PMV factor is higher than 2. Those microclimate building conditions exceed the optimal internal summer temperature of 25°C and recommended value -0,5 < PMV < +0,5. Figure 2. Building zones visualization – ground floor, first floor, second floor. The main aim of simulations was to determine the temperature of the building interior space and of the particular rooms during summer months. The Fanger comfort-model, Pierce two-node model and KSU two-node model were used to predict thermal performance of the building. In the first phase of simulations the building was modelled in the service conditions. The assumptions for the first phase of simulations (base case): 1. Heating system on from September to March (22°C), 7 days per week, 24 hours a day. 2. Occupancy density: - Office rooms – 0,1 person per m2 - Lecture classes – 0,85 person per m2 3. Operating Schedule: - Office rooms – July thru August 8:00 – 16:00, 5 days a week; September thru June 8:00 – 21:00, 7 days a week - Lecture classes - September thru June 7:30 – 21:00, 7 days a week, July thru August 7:30 – 21:00, 5 days a week Figure 3. Simulation results - base case a – temperature - air temperature inside, radiant temperature, operative temperature, outside dry bulb temperature b – PMV comfort factor (Fanger, Pierce, TSV Model). In the lecture class located at the south building side, on the second floor, the operative temperature for most of time is significantly higher than 25°C. The daily maximum interior temperature is 34,41°C and the PMV value is above 2. The number of discomfort hours in the assumed period of time is 1127. Those negative room conditions continue almost for the entire day and do not change significantly during the night. Due to the safety conditions rooms cannot be cooled during the night through the open windows. 345 5.2 Building rooms in the service conditions 5.3 Influence of thermal insulation The microclimate conditions of three different office rooms, 13 at ground level, 113 at the first floor and 213 at the second floor, were separately analyzed, Figure 4. The second phase of the modelling focused around the impact of insulation on the thermal comfort. Figure 4. Visualization of south building elevation. Windows in rooms 13, 113, 213. In all rooms the temperature reached well above the comfort conditions. Those results indicate that all rooms will have problems with overheating in the summer. Figure 5 below shows the number of overheating (discomfort) degree hours for temperatures above 25°C in different office rooms. Figure 5. Number of overheating (discomfort) degree hours for separate office rooms (13, 113, 213). For the validation of the building simulations in the service conditions, the test measurements of thermal comfort were conducted in one of the office rooms at the second floor of the building during the summer months. In the small office rooms the work conditions were uncomfortable. During the sunny days the internal temperature exceeds 28°C for most of the day and the PMV factor is above 1,5. The number of discomfort hours during the summer months in the analyzed rooms could be significantly lessened up to 41% if the building would be rotated 180o. Then the office room windows would be located at north elevation of the building. This solution however would have negative influence on the energy need during heating season and because of that no further analyzes were conducted. Figure 6. Simulation results for the office room 213 at the second floor and for lecture class 209 after thermal modernization. a – temperature - air temperature inside, radiant temperature, operative temperature, outside dry bulb temperature b – PMV comfort factor (Fanger, Pierce, TSV Model). Improvement of the thermal transmittance of the external walls up to U = 0,18 [W/m2K] was not a significant factor, as the percentage share of glazing at the south elevation is still the primary factor affecting thermal comfort. The results are presented in Figure 6. 346 The significant internal gains are affected by the high occupancy density especially in the lecture classes - 0,85 person per m2. A). Figure 7. Number of overheating (discomfort) degree hours for separate office rooms in building after thermal modernization, without internal shadings (13, 113, 213). Figure 7 shows discomfort hours for office rooms 13, 113, 213 in the building after thermal modernization. Comparing those data with the ones for the same rooms but in the building without thermal insulation presented in Figure 5, it’s noticeable that the number of discomfort hours has increased. The most unfavorable changes for 27/28°C. Improvement of building envelope insulation reduces the heat losses during both heating months and overheating rooms during summer season. The second aspect in a disadvantage of this solution. Due to the significant area of glazing generating solar gains, the internal shadings were installed in analyzed rooms. B). 5.4 Influence of occupancy density and shadings As the next modelling stage, a set of simulation runs was conducted for different occupancy density: 25% of total users density to decrease the internal gains. The results were compared with those for 100% users overload (the base case). The ventilation rate was adjusted to the reduced occupancy density and the separate set of simulations were conducted. Results shown in Figure 9. To lessen the unfavorable influence of glazing during summer months with the same occupancy density, the internal gains were reduced by internal solar shadings (blinds with medium reflexivity slats). Positive results of the internal shadings and reduced occupancy density can be noticed in the lecture class located in the south part of building, Figure 8, as well as in the office rooms. Figure 8. A) Simulation results for lecture class 209 (base case) a – Temperature - air temperature inside, radiant temperature, operative temperature, outside dry bulb temperature b – PMV comfort factor (Fanger, Pierce, TSV Model) B) Simulation results for lecture class 209 with the internal shadings and reduced occupancy density with reduced ventilation rate. a – Temperature - air temperature inside, radiant temperature, operative temperature, outside dry bulb temperature b – PMV comfort factor (Fanger, Pierce, TSV Model). Both inside air temperature and PMV comfort factor values are smaller than for the base case building, Figures 8A and 8B. Fanger PMV values differs from three others, the significant differences are especially noticeable for internal temperatures of about 20°C and lower. 347 6 CONCLUSIONS Most of the new office and commercial buildings are designed with extensive south facing window sections, which in most cases are poorly shaded from solar radiation. Glazing is the source of the excessive heat gains and may result in the overheating of the entire building but specifically of separate rooms located at the south side of the building. The work conditions in those specific rooms appear to be very uncomfortable. When using glass facades in building construction it is necessary to consider, at the designing stage, the thermal comfort conditions both for the entire building and for separate rooms. The simulation results presented in this article are the initial stage of the further analysis of the building modernization influence on the thermal comfort of rooms during the summer months for Polish climatic conditions. The conducted simulations result in the following conclusions: 1. Modernization of the building should be preceded by the extensive analysis of how the changes influence thermal comfort of the rooms. The priority aim is heating costs reduction in the winter season. The conducted analyses show that improving of the building envelope thermal insulation alone can unfavourably affect the internal conditions during summer season. 2. In the rooms with a significant glazing ratio the solar protection level and shadings appears to have the greatest influence on the thermal comfort. 3. Cooling of the rooms must be applied if none of the external or/and interior shading systems are used to avoid overheating. 4. None of the measures considered eliminated the overheating entirely however employing a combination of different measures reduced overheating significantly. 5. In the building modernization process, using of external shadings to reduce summer overheating, could be also taken under consideration. Those solutions are the subject of authors’ further researches. Figure 9. Number of overheating (discomfort) degree hours for lecture class number 209- base case, and 209* – with shadings and reduced occupancy density with reduced ventilation rate. Both in the lecture class and office rooms with the extensive windows area at south elevation the combination of internal shadings and reduced occupancy density improved the thermal comfort significantly. For the lecture class located at second floor, number 209, the comparison of the amount of overheating hours was presented in Figure 9. 5.5 Comparison of different options Figure 9 presents the number of discomfort degree hours for lecture class before modernization (base case) and after applying the thermal insulation, the internal shadings together with the reduction of occupancy density. The number of discomfort hours after modernization is much lower. Temperatures higher than 32oC were almost entirely eliminated and the number of degree hours above 28oC was reduced by almost 40%. A similar situation can be observed in three different office rooms located at different elevations, results shown below, Figure 10. Figure 10. Number of overheating (discomfort) degree hours for separate office rooms (13, 113, 213) with the internal shadings after thermal modernization. 7 REFERENCES Comparing Figure 10 with the findings shown in Figure 5 (for the same rooms) significant reduction of hours with the high operative temperature can be observed. The usage of internal shadings eliminated the temperature higher than 32oC and the number of degree hours above 28oC was reduced by almost 75%. ASHRAE 2001. Thermal comfort conditions for human occupancy, Standart 55-81. Atlanta. Holopainen R. (ed.) 2012. A human thermal model for improved thermal comfort. Espoo: VTT Technical Research Centre of Finland. 348 Lee J., and Strand R. An Analysis of the Effect of the Building Envelope on Thermal Comfort using the EnergyPlus Program Nouidui T. S., Phalak K., Zuo W., and Wetter M. 2012. Validation and Application of the Room Model of the Modelica Buildings Library. LBNL Report Number LBNL-5932E Nowak K. 2011. Modernizacja budynków a comfort cieplny pomieszczeń. Energia i Budynek, ISSN 1897-5879, 29-33 Orme M., Palme J.R, and Irving S. Control of Overheating in Well-Insulated Housing, CIBSE http://www.cibse.org/pdfs/7borme.pdf PN-EN ISO 7730 „Ergonomics of the thermal environment. Analytical determination and interpretation of thermal comfort using calculation of the PMV and PPD indices and local thermal comfort criteria” PN-83/B-03430 “Wentylacja w budynkach mieszkalnych zamieszkania zbiorowego i użyteczności publicznej Wymagania” 349 350 The aid of TRNSYS simulation for the conservation of an artwork. A case study L. De Backer & A. Janssens Ghent University, Department of Architecture and Urban planning, Ghent, Belgium M. Van Belleghem & M. De Paepe Ghent University, Department of Flow, heat and combustion mechanics, Ghent, Belgium ABSTRACT: Due to secularization more and more churches lose their original function, that is a place to worship, and become a place of cultural interest. To fulfil the comfort demands, a heating system or a HVACsystem was installed. These systems however impose new boundary conditions for the interior objects like paintings,… which may lead to damage. The case of the Saint-Bavo Cathedral in Ghent, Belgium is presented in this paper. In this church hangs the valuable polyptic ‘The adoration of the mystic Lamb’ painted by van Eyck. Use was made of a computer simulation to get insight in the working of different solutions to combine comfort and preservation. Because we want to be able in first instance to evaluate a whole year, a BESsoftware, was used. Using the dynamic simulation code TRNSYS the system performance in controlling thermal-hygrometric ambient parameters were obtained. number of modelling techniques have been developed to predict heat and moisture transport. Two main categories of mathematical models are in wide use. The first are the microscopic scale models, which use computational fluid dynamics (CFD) to calculate the values of all relevant parameters at closely-spaced points in all parts of the flow field with a high degree of resolution. The second are the macroscopic scale models (often called multi-zone models) which calculate flows between zones; usually whole rooms in a building (Feilden, 1994). The chosen model depends on the scope of the problem. The BES-software has a relatively short computation time in comparison with the CDF-models and is therefore more suitable for the simulation of longer time periods or for predicting the energy performance of a whole building. A large variety of multizone models for energy simulation exist. Common used macro scale models are TRNSYS, DOE-2, EnergyPlus, ESP-r. (Crawley, Hand, Kummert, & Griffith, 2008) 1 INTRODUCTION Historic buildings are a part of our history. The last decades, more and more historical buildings receive a different function than the one they have originally been built for. Churches, castles, abbeys are converted into museums, cultural centres, living accommodations,… Looking specific to churches, due to secularization more and more churches lose their original function: a place to worship. The church building which often contains valuable interior objects like organs, paintings, pulpit become a place of cultural interest. To fulfil comfort demands of the visitors and employees, a heating system or a more sophisticated HVAC-system (Heating, Ventilation and Air-Conditioning) is introduced. The introduced systems however impose other temperature and relative humidity conditions which may cause damage on the artworks, and more specific on panel paintings. There are different options to obtain insight in the indoor temperature and relative humidity: a measurement campaign and/or mathematical models. A measurement campaign gives insight into how the building temperature and humidity are related to the outdoor climate and heat sources. But a measurement campaign is quite time consuming and it only gives information of the actual setting of the building. The option of the use of a simulation study makes it possible to compare the effect on the indoor climate for different configurations. In the past, a 1.1 Case study In this paper a simulation study of the baptistery in the Saint-Bavo Cathedral in Ghent, Belgium is described. In this church hangs the famous polyptic ‘The adoration of the Mystic Lamb’ painted by the brothers Van Eyck in 1453 (Figure 1). Many tourist coming from the whole world visit the church to watch the artistic masterpiece of the 15th -century Flemish painting. 351 In 2008 a thorough examination performed by the Flemish cultural heritage guard service has proven that the present climate was not sufficient to preserve the polyptic. So in 2010 a project was set up, with financial support from the Getty Foundation, for the preservation of the altarpiece. Ghent University was commissioned to evaluate the current environment and also to propose measures to improve the preservation of the altarpiece. the single heating systems and what kind of modifications could improve the situation in a particular church. Also Schellen (Schellen, 2002) discussed the different heating systems in a church: their advantages and disadvantages. 2 METHOD Ceiling Steel frame:1m A multi-zone model was carried out with the commercially building software TRNSYS v17(Solar Energy Laboratory, 2010). The main strength of TRNSYS is its large libraries for HVAC and the building model that passed significant validation tests (Wetter & Haugstetter, 2006). With the aid of this software, the indoor conditions of the glazed chamber and baptistery could be predicted. First a model was made of the baptistery with his boundary conditions. For the calculations a time-step of one hour was used. Glass:3,4m Steel frame:1m Floor 2.1 Geometrical model of the baptistery Figure 1. The polyptic ‘The adoration of the mystic Lamb’ located in the Saint Bavo’s baptistery in a glazed chamber. The Saint-Bavo Cathedral is situated in the city Ghent in Belgium. The painting originally hung in the Vijdt chapel, but in 1986 the baptistery was selected as safe and new space for the exhibition of the painting. The baptistery is situated in the northwest tower of the church, and in this space the painting was enclosed in a bulletproof glazed chamber under a suspended concrete ceiling covered with a bitumen layer (Figure 2). As a results the baptistery became separated from the remaining part and there were no windows so that no sunlight felled into this space. The walls of the baptistery were raised with a local calcareous sandstone and they measured between 1 and 2 m thick. The vertical walls of the glazed chamber consisted of 3,4 m-high windows embedded in a steel framework. Because the structure of the church floor was unknown the assembly was based on literature (Fawcett, 2001): a plain square paving stone, laid on a bed of lime mortar. In the glazed chamber the floor level was raised 12 cm by the construction of a new floor on the old one. The main geometric and thermal characteristics of the model are shown in Table 1. One part was a measurement campaign which was started in February 2010. From these measurements was known that mainly the relative humidity fluctuations were too large. In 2011 some interventions were performed, partly to increase the comfort for the workers and partly to improve the preservation properties. The aim of this study was, to get insight in the working of the original climate and the new obtained climate with the aid of simulations. If the actual configuration is not satisfactory a new solution will be proposed. 1.2 Previous investigations To preserve a panel painting ASHRAE (Anon, 2011) gave general guidelines for the temperature and relative humidity and in particular, the allowed fluctuations around an average value. The plants placed to answer the comfort demand, creates a different climate than the ‘historic’ climate and sometimes other fluctuations (because, for example, there is only heated during a church service or during the opening. Past investigations of the indoor climate pointed out drastic effects on the artworks; fast deteriorating or damage owing to the climate changes. (Bratasz, Kozlowski, Camuffo, & Pagan, 2007; Troi & Hausladen, 14 – 17 October 2006; van Schijndel, Schellen, & de Wit, 2009). Therefore there was tried to find a solution to meet the demands of comfort and preservation. For example, Troi (Troi & Hausladen, 14 – 17 October 2006) compared the climate caused by the four most important heating systems in 26 churches in South Tyrol, Italy. By this way, she demonstrates what the strengthens were of Exterior Baptistery Zone 2:Baptistery ± 8m Vijdt chapel Panels Zone 1:Glazed chamber N Church ± 9,5m Figure 2. Floor plan of the baptistery and the shrine, located on the first floor of the West tower. 352 Table 1. Geometric and thermal characteristics of the model. Geometric values Floor surface Volume R-value Exterior wall Internal wall + window Ground floor Ceiling W/m²K W/m²K Baptistery 65 260 Baptistery 0.47 0.86 W/m²K W/m²K 0.26 1.67 m² m³ Glazed chamber 15 105 Glazed chamber 0.13 2.89 0.22 1.67 2.2 Boundary conditions of the model In the simulation only the zones dedicated to the conservation of the painting were considered. By consequence the building model consisted of two zones as shown in Figure 2: the baptistery and the glazed chamber. The floor and walls of the baptistery were assumed to be boundaries with the church and the outdoor environment. For the climate of the church Hobo data loggers of the H8 Pro Series were placed near the baptistery. Outdoor climate conditions were handed by the Sterrenwacht of Ghent University. Internal gains in the baptistery were the lights and the visitors. Lighting was minimal and was therefore assumed on 10W/m². The occupation rate of the baptistery was estimated on daily sold tickets. The visitors were allowed from 9h30 till 17h from Monday to Saturday and on Sunday from 13h till 17h in summer season. In the winter season the doors opened one hour later and closed one hour earlier: from 10h30 till 16h from Monday to Saturday and on Sunday from 13h till 16h. The thermal load for people was calculated based on an activity level 5 of the occupants (ASHRAE, 2001). With the aid of the results from a tracer gas measurement the ventilation and infiltration of the baptistery and glazed chamber was modelled making use of TRNFLOW. tion was chosen to simulate as it was one of the options but because of financial reasons not practicable. 3 RESULTS AND DISCUSSION The simulations results were compared with measured data in terms of indoor temperature and absolute humidity. In winter, simulation results were in good agreement with the measured data. In the summer there was an underestimation of the temperature. In addition there was also noticeable in the TRNSYS simulations that a certain lead time was needed, before the attenuated temperature was obtained. Further, deviations in the temperature and humidity course can be due to the estimations that were made for the simulation (see boundary conditions). 3.1 Visitor management The visitor management had a very limited impact on the allowable temperature and relative humidity course. Yet this management is a simple measure to apply and to please the visitors by giving the necessary space to experience the painting. 3.2 Actual intervention: heating and humidifier On Figure 3 hourly indoor temperature and relative humidity trends were shown for both the baptistery and glazed chamber when the heating works. The simulations pointed out that heating by the radiant panels caused a temperature variation that was higher than before. As a result, the temperature in the baptistery fluctuated between 12 and 16°C when the heating was in device. This was in line with the measurements. One might wonder whether these larger temperature fluctuations raised the damage risk. Following the most strict requirements of ASHRAE(Anon, 2011) temperature fluctuations of ±2 °C are permitted around a yearly average. Thereby, the fluctuations caused by the heating were smaller in the glazed chamber than in the baptistery. So, according to these standards the higher temperature fluctuations in the baptistery involved no risk for the painting. More dangerous was the influence of the heating on the relative humidity in the shrine. Although the heating ensures that the relative humidity was no longer too high (does not exceed the value of 80%), now the danger arose for a too low relative humidity. On the figure it can be seen that the daily relative humidity made large daily fluctuations, in which the absolute value of the relative humidity felled below 40%. To avoid this low value in RH, humidification was needed. Therefore in practice a mobile humidifier was placed in the glazed chamber. This however is not a solution to decrease 2.3 System configurations The following kinds of configurations have been considered: - No climate control is present. Only a visitor management was introduced; this means that the number of people in the baptistery is limited. There was calculated with the worst condition in which there were always 30 people present. - A radiant panel heating on the ceiling of the baptistery (set point: 12°C) in combination with humidifier (set point 40% RH). This type of configuration was chosen to simulate as it is the system that operates at the moment. - A single zone system. Temperature and relative humidity of the glazed chamber and baptistery were controlled by the baptistery. No cooling was provided. This type of configura353 the large RH- fluctuations and thus to improve the conditions for the preservation for the painting; for this a more advanced control system is needed. glazed chamber rad. heating baptistery rad. heating glazed chamber unheated baptistery unheated Figure 3. The effect on the temperature and relative humidity caused by radiant heating (values shown for one week when heating was needed). Figure 4. comparison of the relative humidity in the glazed chamber and the shrine for an allowable range of 40-80% and 50-70%. 3.3 A simple HVAC-system The system controls the temperature and relative humidity in the baptistery and this also indirect the glazed chamber in the baptistery. Measurements pointed out that annual average of the relative humidity in the baptistery and glazed chamber was 60%. To control the RH-fluctuations two options were studied: in the first the RH was allowed between 40 and 80%. The idea behind this was to avoid extreme values. Results pointed out that with these boundaries the RH fluctuated too much to obtain good conditions to preserve the painting. In the second options a smaller range for the RH was allowed: between 50-70%. Figure 4 shows that daily fluctuations were still present, but large differences were smoothed. The advantage of a simple HVAC system is that besides the heating and humidification fresh air was supplied. Till now, in the baptistery this is not the case and by consequence people complained in summer about headaches. 4 CONCLUSION AND PERSPECTIVES The simulations carried out in the paper a first attempt to create a model to look at the effect of specific interventions. Next to the originally and thus non-conditioned situation, three configurations were simulated: a visitor management, the actual heating system and a simple HVAC system. The simulation results of the current heating system were confirmed by the measurements: temperature variations became higher and the RH-value in the shrine does no longer exceed the value of 80%. On the other hand a new problem arose: danger for too low RH-values. This could be mended by humidification. Therefore in practice, a mobile dehumidifier was used in the glazed chamber. This provided the additional task for the supervisors to 354 follow up the devices. Not doing this could cause a dangerous drop in RH. Besides this, the problem of the RH-fluctuations which are too large, has not improved and thus the conditions for the preservation for the painting. So further measures are needed. Furthermore, a simulation was performed where the climate was controlled by a HVAC-system. To limit the RH-fluctuations a range of 50 till 70% RH was chosen for the simple HVAC system. The advantage of a simple HVAC system is that besides the heating and humidification fresh air is supplied. Therefore, it would be useful in further research to look at the cost price in addition to the thermohygric performance. Because a heating system was already placed there will be searched for a solution where the heating can be maintained in addition with fresh air supply and a device to control the moisture fluctuations. Also interesting is to study the option to allow seasonal variations in the relative humidity while limiting the short-term fluctuations. This seems at first sight a better system because with the simple HVAC system a tighter band was needed to avoid the large short-time fluctuations. Wetter M., and Haugstetter C. 2006. Modelica versus TRNSYS – A comparison between an equation-based and a procedural modeling language for building energy simulation. Paper presented at the Simbuild 2006, MIT, Cambridge, Massachusetts. 5 REFERENCES Anon. 2011. ASHRAE handbook: Heating, ventilating, and airconditioning applications, SI edition (pp. 23.21-23.22): American Society of Heating, Refrigerating and AirConditioning Engineers. ASHRAE. 2001. ASHRAE Handbook, 2001 Fundamentals. Atlanta, GA: ASHRAE. Bratasz, L., Kozlowski, R., Camuffo, D., and Pagan, E. 2007. Impact of indoor heating on painted wood - Monitoring the altarpiece in the Church of Santa Maria Maddalena in Rocca Pietore, Italy. Studies in Conservation, 52(3), 199210. Crawley D. B., Hand J. W., Kummert M., and Griffith B. T. 2008. Contrasting the capabilities of building energy performance simulation programs. Building and Environment, 43(4), 661-673. doi: 10.1016/j.buildenv.2006.10.027 Fawcett J. 2001. Historic floors: their history and conservation: Butterworth-Heinemann. Feilden B. M. 1994. Conservation of historic buildings. Oxford: Butterworth-Heinemann Ltd. Schellen H. L. 2002. Heating monumental churches : Indoor Climate and Preservation of Cultural Heritage. (PhD thesis), Technische Universiteit Eindhoven. Solar Energy Laboratory, U. o. W.-M. 2010. TRNSYS 17: A Transient System Simulation Programme. Madison, USA. Troi A., and Hausladen G. 2006. Indoor Climate & Damage Risk Due To Heating Systems In South Tyrolean Churches. Paper presented at the Conference And Brokerage Event The Construction Aspects Of Built Heritage Protection, Dubrovnik, Croatia. 14 – 17 October van Schijndel A. W. M., Schellen H. L., and de Wit M. H. 2009. Improved HVAC operation to preserve a church organ. Building and Environment, 44(1), 156-168. doi: 10.1016/j.buildenv.2008.02.003 355 356 Experimental measurements and analysis of the indoor conditions in Italian museum storerooms: A study case F. Cappelletti & P. Romagnoni University IUAV of Venezia, Department of Design and Planning in Complex Environment, Venezia, Italy A. Birra University IUAV of Venezia, Italy ABSTRACT: This paper presents a detailed analysis of the indoor microclimate in the storeroom of Oriental Art Museum of Venezia (Italy). Data of air temperature and air humidity collected during the years 20032004-2005 were elaborated to suggest which kind of materials can be safely stored in it. A dynamic simulation has been implemented in order to focus some interventions to improve the microclimatic conditions. Some different strategies were compared: (#1) the envelope improvement; (#2) the temperature and humidity control; (#3) the summer ventilation. The strategy #1 gives a good improvement during spring and autumn, but are not enough to mitigate the most severe external conditions. The strategy #2 is necessary to really improve the indoor microclimate even if the great variability of the objects inside the storeroom imposes to put in action different control strategies for the air humidity set-point. Strategy #3 is not a good solution because it contributes to decrease the cooling needs, but the energy need for dehumidification increases. particles, pollutant gases, bacteria and fungi that can deposit on the works of art (Gysels et al., 2004) In this paper an extensive number of measurements of air temperature and air humidity collected during the years 2003-2004-2005 was elaborated in order to describe the microclimatic conditions for the conservation of the collection in the Oriental Art Museum of Venezia (Italy) which is without any heating nor cooling system. The data analysed in this paper are related to the storeroom of the museum where a great part of the collection is still contained. For the microclimate measurements and analysis of environment conserving works of art we referred to the UNI 10829 (UNI,1999), which suggests some statistical descriptive parameters such as the frequency distribution and cumulated frequency of the variables measured in a long term monitoring. To represent the variability of the variables over the time the maximum, minimum, the median and the first and third quartile were used even if it is not commonly used in literature to represent this kind of monitoring. The second aim of the work was the attempt to find some potential intervention for the improvement of the indoor environment. To achieve this aim a dynamic simulation of the museum was carried out, trying to reproduce similar hourly profile of temperature and humidity. Finally the effects of some improvements on the envelope, of the introduction of a conditioning system and of summer ventilation have been evaluated. 1 INTRODUCTION The monitoring and the analysis of the indoor microclimate conditions inside museums has to be pursued in order to guarantee the best condition for the conservation of the objects. The determination of the thermo-hygrometric conditions is due to the combination of multiple factors such as the geometric configuration of the building, its geographical location, the number of occupant and their activity, the modality of use of the building, the setting of the HVAC system if present. To each perturbation the building will react to keep a new state of equilibrium which can require more or less time, depending on the entity of the perturbation. During this time the mass and energy exchanges between the environment and the objects in it occur and can determine some damages. A long-term microclimate monitoring is, for these reasons, very important not only to preserve objects from deterioration, but also for scheduling both the interventions and the management. Many examples of long-term monitoring in museums or exhibitions can be found in literature. Sometimes the purpose of these studies is the evaluation of the efficacy of the HVAC system to maintain the preservation conditions for the objects (Corgnati & Filippi, 2010), or to evaluate the effects on the indoor microclimate both of a specific kind of HVAC system and of the visitors (Baggio et alii, 2004); in other context the HVAC system is not present and the analysis of the monitored variables is aimed to assess the risk of damage for the works of art conserved (La Gennusa et al., 2005). Finally in some cases the monitoring of microclimate is not only focused on air temperature and air humidity, but can also consider the aerosol 1.1 The Oriental Art Museum of Venezia and the monitoring The Oriental Art Museum of Venezia in Italy is second only to the Museum of Oriental Art in Rome, 357 and preserves the valuable collection that Enrico di Borbone purchased during his voyage around the world in 1887. Starting from 1925, the collection was placed on the 2nd floor of Ca' Pesaro palace, as required by an agreement between the Municipality of Venice and the National Department of Education. This XVII century building, facing the Canal Grande, has a great historical and architectural value so that setting a technological system is not easy, according to the constraints dictated by the competent authority on safeguarding heritage and the inevitably invasiveness problems. The peculiarity of the case can be ascribed to the following factors: - the absence of any microclimate control system: the variation of the internal temperature and relative humidity is mainly due to external variation of these parameters; - the great variability of the materials stored up: metal, tissues, wood, paper, lacquers, parchments; - the climate of Venezia characterized of high percentage of air humidity. Considering these aspects a long-term monitoring of indoor temperature and relative humidity was carried out from 2003 to 2005. Five data loggers were installed in the storeroom. The attention of the further analysis will focus on the storeroom. The data loggers used can measure both air temperature and relative humidity with the characteristics specified in table 1. The data loggers recorded data every 10 minutes and they were located in different position in the storeroom. Two loggers (P2 and P3) were collocated on two different cupboards; three of them were located inside the cupboards (P4, P5, P6) which store tissues or paper. Figure 1 shows the position of the loggers. 2 INDOOR MICROCLIMATE ANALYSIS Figure 2 summarizes the distribution of the indoor air temperature and air humidity for each month from January 2003 to December 2005 as measured by the logger 3. The values represented are the descriptive statistical parameters: the maximum hourly value, the minimum hourly value, the median of the hourly values, the range between the first and the third quartile. The longer are the rectangular bars, the more variable are the data measured. The distance between the two lines represent the maximum monthly fluctuation. 2.1 Analysis of the air temperature and air humidity For reason of space the temperature measured only by the logger 3 is shown in figure 2. The annual range of temperature is very high. In fact it is possible to see that the annual maximum range has been of 25.6°C during the 2003, of 23.4°C during the 2004 and the same during the 2005. Comparing the monthly fluctuations of temperature it is possible to find some critical months: in general it can be seen that the spring and autumn months are the ones with the higher fluctuation. For example during the 2003, in October the monthly temperature range was of 10.4°C, during the 2004 the maximum temperature range was in March with 9.6°C of variation; during the 2005 the maximum variation was of 10.3°C in November. Obviously since the indoor microclimate is not controlled by a system the internal fluctuations are the projections of the external ones. Considering the differences between the temperature values measured in different positions in the space, it can be underlined that the air temperature is quite homogeneous with a maximum standard deviation calculated hour by hour of 2.1 °C over the whole period (table 2). Table 1. Characteristics of the data loggers. Temperature Relative humidity Range 0 - 95% -10°C - 60°C (-20°C - 85°C) Accuracy ± 0.3 °C ± 3% (0°C - 70°C) Response time 10 s 10 s Table 2. Maximum hourly standard deviation of the temperature and humidity values collected by the different probes. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2003 T 1.1 1.2 2.1 1.1 1.4 1.2 1.1 1.1 1.8 1.3 1.0 1.1 [°C] RH 4.1 5.1 6.2 8.4 8.2 7.6 7.7 6.5 6.4 4.9 5.4 4.1 [%] 2004 T 1.0 1.6 1.7 1.1 1.0 1.0 1.0 1.0 1.1 1.2 1.2 0.8 [°C] RH 3.6 4.2 4.1 5.4 4.7 3.0 3.6 3.2 4.1 4.7 4.3 3.7 [%] 2005 T 1.1 1.0 1.2 1.2 1.1 1.4 1.3 1.6 1.4 1.4 1.7 0.9 [°C] RH 4.0 5.6 4.6 5.3 4.6 5.9 7.6 6.1 4.3 6.4 6.5 7.0 [%] Figure 1. The plan of Museum with the position of the data loggers in the storeroom. 358 The standard deviation for the air humidity values has been at maximum 8.4% that is quite high. The variability of the humidity can be ascribed to the different collocation of the data loggers: outside or inside cupboards. The probes inside cupboards adsorb and desorb the water vapor smoothing the air humidity fluctuations. For reason of space the air relative humidity measured only by the logger 3 is shown in figure 2. The annual range of air humidity is not so high. In fact from this representation it is possible to see the annual maximum range that has been of 27.4% during the 2003, of 26.5% during the 2004 and of 20% during the 2005. Comparing the monthly fluctuations of humidity it is possible to find some worst months which are not collocated in a particular period of the year. For example during the 2003, in October the monthly relative humidity range was of 17.3%, during the 2004 the maximum relative humidity range was in March with 14.7% of variation; during the 2005 the maximum variation was of 17.3% in November. During the three years the relative humidity never overcome the 71% and never falls down 42%. Concerning the variability of the humidity in the space it has been noted that inside the cupboards the fluctuations are mitigated if compared to the outside values. ted. Looking to the temperature gradients it can be seen that a part from the first half of 2003 and May June 2004 during the other months about 50% of the daily gradients doesn’t overcome 1.5 °C. Otherwise the maximum values are often above 2°C. The most synthetic parameter for this evaluation is the DI reported just for probe 3 and for the year 2003 in table 3 for the ranges and in table 4 for the gradients. Looking to table 4 it can be seen that in January, February and November the DI is 0% and in December is still very low (about 6%), but if we look to the DI for the air temperature (table 3) it can be seen that values lays outside the acceptable ranges for the 100% of the time. Table 3. Suggested values according to Italian Technical Standard (UNI,1999). RH ∆RHmax T ∆Tmax [°C] [%] [%] [°C] Parchments 13-18 50-60 6 Paper 18-22 1.5 40-55 6 Drawings 19-24 1.5 45-60 2 Tissues 19-24 1.5 30-50 5 Lacquers 19-24 1.5 50-60 4 Wood 19-24 1.5 45-60 4 Metals <50 - Concerning the air humidity gradients (table 4) the most stringent gradient of 2% is not respected for the most of the time during the 2003, while the higher gradients which are suitable for parchments, paper, tissues, lacquers and wood is maintained for the most of the time almost in each month. It must be said that these results concern to the probe 3 that is the probe located outside cupboards. Data collected from probes 5 and 6 demonstrate for the most of the time very little gradients for humidity: probably materials act like buffers and limit humidity fluctuations.Otherwise maximum values are higher than the acceptable value. Looking to the DI in table 3 it can be seen that in general the hygrometric conditions are most suitable for drawings and wood conservation but just from May to August. A general comment about the profile of temperature and humidity, here not included for reason of space: often the peaks in hourly gradients are due to the opening time and the closing time of the museum and they occur for a maximum of one or two hours. 2.2 Evaluation of the microclimate in relation to the conservation needs The Italian Technical Standard UNI 10829 (UNI, 1999) gives suggestions about the parameters to elaborate in order to better assess the thermohygrometric conditions inside museums. According to this Standard the temperature and humidity values have been evaluated in relation to the materials stored. The daily gradient of temperature and relative humidity were calculated for the all three years such as the hourly and two hourly gradients of the same parameters. Moreover, the frequency distribution of the temperature and humidity ranges that are suitable for particular materials have been used to assess the indoor environmental quality of the storeroom microclimate for the conservation of the objects. Finally a synthetic index, the Deviation Index, DI has been calculated every month: such parameter represents the percentage of the period of time during which the variations of air temperature and air humidity are outside the range of values allowed by the standard. In table 3 the optimal ranges and daily maximum gradients of temperature and relative humidity, according to (UNI,1999) for the materials preserved in the storeroom are reported for reference. In figure 3 the statistical distributions of the daily temperature range and the daily relative humidity range for each month for the probe P3 has been plot- 3 THE SIMULATION OF THE STOREROOM In order to evaluate some possible interventions to improve the indoor microclimate of the museum’s storeroom, a dynamic simulation was carried out trying to reproduce the indoor thermo-hygrometric conditions. The simulation code used is TRNSYS r.17. The museum was divided in three thermal zones: the storeroom, the offices and the museum. 359 temperature monthly gradient are quite similar with a maximum difference of 2°C, while the humidity gradient is very different. To set the simulation many unknown had to be solved: - the thermal characterization of the envelope; - the floor boundary temperature; - the air tightness of the building; - the ventilation rate; - the occupancy period. The external walls are 0.57m thick, and they are composed of bricks with an internal and external plaster; the overall estimated thermal transmittance is 1.03 W/(m2 K). Somewhere the external plaster is replaced with stone. The roof is a wooden roof with tiles and a layer of 0.025m of insulation material that was installed during the restoration in 1980; the overall estimated thermal transmittance is 1.31 W/(m2 K). The floor is composed of a wooden boards sustained by beams, and covered by the traditional “terrazzo alla veneziana” a typical kind of ancient venetian floor composed of fragments of marble mixed with plaster; the overall estimated thermal transmittance is 1.36 W/(m2 K). The windows have a single clear glazing (Ug = 5.7 W/ (m2 K)) and a wooden frame (Uf = 3.26 W/(m2 K)). Internal shades with a transmission coefficient of 0.7 are always on. The floor was, at a first attempt considered as adiabatic, but the comparison of the temperature with the measured data suggested to consider a profile of temperature at the floor boundary, in consideration to the fact that the Oriental Museum is located at the second floor and on the first floor the Modern Art Museum is heated and cooled by a system with a known temperature set-point hourly profile: these data were used. Concerning internal gains, considering that the inlet of people in the storeroom was very limited and sporadically, no internal gains were input for the simulation. The infiltration rate has been fixed in 0.3 h-1 which corresponds to a low permeability of the windows. In figure 4 the simulated temperature and relative humidity hourly profiles for four months during the 2003 have been represented (dotted black line) in comparison with the measured data (continuous black line) to show the level of approximation of the simulation. Concerning the temperature, the simulation results follow the measures trend quite well with a maximum gradient of 2-2.5°C. On the contrary the relative humidity simulated doesn’t fit with measured data: while the real data show a quite constant relative humidity the simulation output follows very much the external fluctuations. This can be probably due to the very important role of the furniture and of the works of art in adsorbing the water vapour and in maintaining the relative humidity constant. In table 4 the maximum and minimum monthly values for temperature and relative humidity obtained from simulation and from monitoring are compared. The Table 4. Maximum and minimum monthly temperature and relative humidity: comparison of measured and *simulated values. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2003 7.4 8.1 13.9 Tmin T*min 8.0 8.4 11.3 Tmax 13.5 14.4 19.7 T*max 13.9 12.7 17.2 RHmin 56.5 55.0 56.5 RH*min 36.1 21.1 17.8 RHmax 69.6 66.2 69.6 RH*max 87.5 65.4 71.9 12.9 12.3 21.8 21.3 48.3 17.3 60.3 71.9 20.5 19.2 28.0 27.0 47.6 20.8 59.9 77.0 25.4 25.4 33.0 31.7 46.0 43.0 53.4 76.0 25.8 26.7 21.1 13.4 14.0 10.8 26.4 27.2 20.1 12.0 12.1 8.2 31.9 33.0 26.8 23.8 17.7 16.6 31.1 31.5 27.0 21.9 16.3 14.7 43.7 43.9 43.3 48.4 60.3 56.4 33.0 39.7 27.4 18.7 34.3 16.6 52.8 55.0 58.4 65.7 70.7 70.1 81.0 83.0 79.7 94.0 98.0 80.8 4 EVALUATION OF MICROCLIMATE IMPROVING MEASURES To improve the microclimatic conditions inside the storeroom some actions are proposed, starting from the improvement of the envelope thermal resistance. However in order to obtain best results a heating and cooling system to control the temperature and the humidity is necessary. The alternatives evaluated are the following: - Intervention #1: roof insulation with 14 cm of rockwool (λ = 0.04 W/(m K)); substitution of the single glazing with a double glazing (Ug = 1.1 W/(m2 K); SHGC = 0.315) together with the substitution of the frame (Uf = 1.7 W/(m2 K)). The substitution of the windows determines also the reduction of the infiltration rate from 0.3 to 0.1 h-1. - Intervention #2: as Intervention#1 plus the control of the air temperature and humidity with a system. The temperature band is 18-26°C during the museum occupation time and 16-28°C outside this period. The humidity is controlled just for the maximum value with a setpoint for the de-humidification of 50% during the opening hours and 55% otherwise. - Intervention #3: as Intervention#2 plus the use of summer ventilation (from April to September) when the outside temperature is between 24 and 26°C. To decide the entity of the insulation layer and to choose the type of glazing the annual indoor air temperature gradient was used and the configuration that could minimize it, was applied. The efficacy of these measures was assessed looking to the new frequency distribution of the temperature and the humidity values and comparing to the suitable intervals for the objects (figure 5), while the comparison between the two solutions with the conditioning system was evaluated comparing the thermal energy need for heating, cooling and dehumidification as reported in table 5. Comparing the efficacy of Intervention #2 and #3 it can be seen that 360 over a wide range and sometimes can be very low (under 35%). Probably a humidification system could be necessary and the re-calculation of the monthly Deviation Indexes will be useful to assess the efficacy of the system control. when the summer ventilation is used even if the cooling needs decreases, the energy need for dehumidification increases much more: the cause is the entity of the outdoor relative humidity in Venezia that is quite high also during the summer. Looking to the frequency distributions in figure 5 it is possible to compare measured data, simulation results, simulation with Intervention #1 and with Intervention #2 during four months in 2003. In January the Intervention#1 don’t move the temperature above 18°C, while with Intervention#2 the temperature rises between 16°C and 19°C. During April and October the application of both the interventions maintain the temperature above 16°C. In July only the Intervention#2 keeps the temperature under 28°C. 5 REFERENCES Baggio, P., Bonacina, C., Romagnoni, P., and Stevan, A.G. 2004. Microclimate analysis of the Scrovegni Chapel in Padua – Measurements and simulations, Studies in Conservation Vol. 49, nr 3, pp. 161 -176 Corgnati, S. P., and Filippi, M. 2010. Assessment of thermohygrometric quality in museums: Method and in-field application to the “Duccio di Buoninsegna” exhibition at Santa Maria della Scala (Siena, Italy). Journal of Cultural Heritage, 11(3), 345–349. Gysels, K., Delalieux, F., Deutsch, F., Van Grieken, R., Camuffo, D., Bernardi, A., Sturaro, G., et al. 2004. Indoor environment and conservation in the Royal Museum of Fine Arts, Antwerp, Belgium. Journal of Cultural Heritage, 5(2), 221–230. La Gennusa, M., Rizzo, G., Scaccianoce, G., and Nicoletti, F. 2005. Control of indoor environments in heritage buildings: experimental measurements in an old Italian museum and proposal of a methodology. Journal of Cultural Heritage, 6(2), 147–155. La Gennusa, M., Lascari, G., Rizzo, G., and Scaccianoce, G. 2008. Conflicting needs of the thermal indoor environment of museums: In search of a practical compromise. Journal of Cultural Heritage, 9(2), 125–134. UNI (1999). Technical Standard 10829:1999 “Works of art of historical importance. Ambient conditions for the conservation. Measurement and analysis. Milano, Italy. Table 5. Energy needs for heating, cooling and dehumidification. Intervention#2 Intervention#3 Heating needs [kWh] 4795 4795 Cooling needs [kWh] 2256 2101 Energy need for de-humidification [kWh] 1541 2858 Temperature [°C] Considering the frequency distribution of relative humidity it can be seen once again that the actual situation is a narrow interval of oscillation around a particular value which depends on the month considered: above 60% for January, 55% in April and October, 45% in July. Looking to the simulation and the interventions the humidity values are distributed 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 80 70 Relative Humidity [%] 60 50 40 30 20 10 0 Figure 2. Statistical distributions of Temperature and Relative Humidity hourly values for each month for the probe P3: the upper black line represents the maximum, the intermediate dot the median, the under line the minimum and the rectangular box the interquartile range. 361 Gradient of Temperature [°C] 5 4 4 3 3 2 2 1 1 0 Gradient of Humidity [%] 14 12 10 8 6 4 2 0 Time [months] Figure 3. Statistical distributions of the Daily Temperature Range and the Daily Relative Humidity Range for each month for the probe P3. Table 6. Maximum value of the monthly Deviation Indexes (DI), expressed in percentages, according to the Italian Technical Standard UNI 10829 for the air temperature and the air humidity (year 2003). Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Air Temperature Paper 100 100 81.0 46.3 97.5 100 100 100 59.7 51.9 98.6 100 Lacquers, tissues, drawings, 100 100 90.7 62.2 31.0 100 100 100 12.4 51.3 100 100 wood Parchments Metals Relative Humidity Paper Tissues Drawings, wood Lacquers, parchments Metals 100 96.0 15.1 51.5 100 100 100 100 100 50.4 0.0 50.5 - - - - - - - - - - - - 87.5 100 87.5 100 100 71.4 100 100 87.5 86.2 97.9 1.7 40.0 36.0 0.0 0.0 41.5 0.0 15.0 19.3 30.0 2.7 10.6 5.4 26.4 83.1 23.1 78.3 98.7 31.4 100 100 100 100 100 80.8 87.5 71.4 87.5 3.7 37.0 58.5 95.7 92.1 28.5 32.7 100 80.8 100 100 100.0 97.9 63.0 41.5 19.3 10.6 83.1 98.7 100 100 Table 7. Maximum value of the monthly Deviation Indexes (DI) of the daily gradients, expressed in percentages, according to the Italian Technical Standard UNI 10829 for the air temperature and the air humidity (year 2003). Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Air Temperature Daily Gradient Parchments Paper, lacquers, tissues, draw0.0 0.0 53.1 20.0 53.1 69.7 75.0 64.5 41.9 15.2 0.0 5.7 ings, wood Metals Relative Humidity Daily Gradient Parchments Paper, tissues 0.0 0.0 0.0 0.0 3.2.0 0.0 0.0 0.0 3.3 0.0 0.0 12.9 Drawings 42.0 79.0 42.0 27.0 19.0 30.0 77.0 52.0 57.0 61.0 37.0 48.0 Lacquers, wood 0.0 0.0 0.0 6.7.0 3.2 3.0 0 6.5 13.0 16.0 3.3 12.0 Metals - 362 Relative Humidity 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 -2 -4 100 90 80 Relative Humidity [%] Temperature [°C] January 2003 Temperature Simulation Intervention#1 Intervention#2 External Temperature 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 -2 Measured Data Simulation Intervention#1 Intervention#2 External RH 100 90 80 70 60 50 40 30 20 10 Measured Data Simulation Intervention#1 Intervention#2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Time [days] External Temperature 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 -2 Measured Data Simulation Intervention#1 Intervention#2 External RH 100 90 80 Relative Humidity [%] Temperaure [°C] 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Time [days] 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Time [days] 70 60 50 40 30 20 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Time [days] Measured Data Simulation Intervention#1 Intervention#2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Time [days] External Temperature 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 -2 Measured Data Simulation Intervention#1 Intervention#2 External RH 100 90 80 Relative Humidity [%] Temperature [°C] 40 0 Relative Humidity [%] Temperature [°C] April 2003 50 10 Measured Data July 2003 60 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Time [days] October 2003 70 70 60 50 40 30 20 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Time [days] Measured Data Simulation Intervention#1 Intervention#2 External Temperature 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Time [days] Measured Data Simulation Intervention#1 Intervention#2 External RH Figure 4. Comparison between the temperature and relative humidity hourly profiles obtained by measured data, by simulation results and simulations with Intervention#1 and Intervention#2. 363 Temperature Relative Humidity 800 700 700 600 500 Time [h] 500 Time [h] January 2003 600 400 400 300 300 200 200 100 100 0 0 <16 18-16 18 19 Measured Data 20 21 Temperature [°C] Simulation 22 Intervention#1 23 24 24-26 <30 30-35 Measured Data Intervention#2 800 700 700 600 600 35-40 40-45 45-50 Relative Humidity [%] Simulation Intervention#1 50-55 55-60 >60 Intervention#2 500 Time [h] Time [h] April 2003 500 400 400 300 300 200 200 100 100 0 0 <16 18-16 18 19 Measured Data 20 21 22 23 Temperature [°C] Simulation Intervention#1 24 24-26 26-28 >28 <30 Intervention#2 30-35 Measured Data 800 700 700 600 600 35-40 40-45 45-50 Relative Humidity [%] Simulation Intervention#1 50-55 55-60 >60 Intervention#2 500 Time [h] Time [h] July 2003 500 400 300 300 200 200 100 100 0 0 <16 18-16 18 19 Measured Data 20 21 22 23 Temperature [°C] Simulation Intervention#1 24 24-26 26-28 >28 <30 Intervention#2 30-35 Measured Data 800 700 700 600 600 35-40 40-45 45-50 Relative Humidity [%] Simulation Intervention#1 50-55 55-60 >60 Intervention#2 500 Time [h] 500 Time [h] October 2003 400 400 400 300 300 200 200 100 100 0 0 <16 18-16 18 Measured Data 19 20 Simulation 21 22 Temperature 23 Intervention#1 24 24-26 26-28 >28 Intervention#2 <30 30-35 Measured Data 35-40 40-45 45-50 Relative Humidity [%] Simulation Intervention#1 50-55 55-60 >60 Intervention#2 Figure 5. Frequency distributions of temperature and relative humidity: comparison between measured data, simulation results and simulations with Interventions#1 and Intervention#2. 364 Variability assessment of summer comfort conditions in social housing using in situ measurements A. Curado Instituto Politécnico de Viana do Castelo, Escola Superior de Tecnologia e Gestão, Departamento de Ciências de Engenharia e Tecnologia, Viana do Castelo, Portugal V. P. Freitas & N. Ramos Universidade do Porto, Faculdade de Engenharia, Laboratório de Física das Construções – LFC, Porto, Portugal ABSTRACT: A social housing neighborhood with 4 blocks, 179 dwellings and about 450 inhabitants was recently retrofitted. An indoor climate measurement program was developed for the months of July, August and September of 2012, to evaluate the summer thermal comfort. The measurements were carried out in 24 dwellings. Indoor air temperature and relative humidity were continuously measured in the bedrooms and living rooms of the occupied dwellings. The purpose of this paper is to study thermal comfort according to the requirements presented in both EN15251 and ASHRAE-55, and the variability of the indoor air temperature using statistical analysis. The relation with parameters such as the period of measurements, the shape factor, and the occupancy was analyzed. In the end it is possible to conclude that the influence of the variability of the studied parameters on the indoor air temperature and thermal comfort, for short periods of analysis, follow a similar trend to what we have for a larger period. On the other hand the influence of the shape factor of the dwellings or the occupancy, on the indoor air temperature, reveals an effect on the measurements. 1 INTRODUCTION rooms are regulated by the occupants through opening and closing of windows. There have been some similar studies regarding the study of the variability on indoor air temperature and relative humidity in situ measurements: From Sept. 2009 to March 2011, Kalamees et al. [3], measured in one-hour intervals over one-year period in bedrooms, the indoor air temperature and relative humidity in 41 apartments in 29 apartment buildings in Estonia. The average indoor temperature during summer, from all measured houses was +24.6ºC (hourly values varied between +14.6ºC and +34.1ºC and standard deviation was 2.6ºC). According to Sung-Hyon H. [4], 3 to 4 week fuel consumption and temperature data were collected from some 1500 dwellings, over two successive winters in 2001/2002 an 2002/2003, under the government-funded British project named Warm Front. The results show that the Warm Front Scheme resulted in a mean increase of 1.6ºC in indoor temperature. There are, however, few studies of the variability on indoor air temperature in mild climates with warm summers, with particular focus on residential multi storey buildings in which the dwellings have very similar characteristics. The aim of this paper is to study the summer thermal comfort of a retrofitted social neighbourhood in Porto, Portugal, for the local climate during the months of July, August and September of 2012, with specific focus upon the possibility of overheating during this period. To assess indoor air temperature and relative humidity of the several rooms, a monitoring campaign was carried out for the summer months of 2012. Data loggers were installed in a set of 24 occupied dwellings to collect data every 10 minutes. The monitoring devices were place in bedrooms and living rooms for different periods of time; 4 dwellings were monitored for 3 months, and the other 20 dwellings were monitored for shorter periods of 2 and 3 weeks. The instrumented rooms (living rooms and bedrooms) were measured in operation conditions, with the residents doing the regular activities in their houses. The influence of the variability in variables such as the period of the measurements (3 months versus 2 or 3 weeks), the shape factor of the measured dwellings, and the occupation behaviour, on the indoor air temperature is analysed using statistic indicators The study of summer thermal comfort of the dwellings is set using an adaptive analysis presented both in EN15251:2007 [1] and ASHRAE55:2004 [2]. This approach is valid for the analysed dwellings, given they don´t have mechanical cooling systems in operation, and the thermal conditions in the 2 CASE STUDY “Lordelo do Ouro” is a social housing neighbourhood located in Porto, Portugal, that is part of a large social housing retrofitting program that is being conducted by local authorities. The program in365 volves the intervention in a considerable number of social housing neighbourhoods, in order to improve the quality of life of 35.000 persons living in 13.000 houses, owned by the city. The neighbourhood was originally built in 1978. It consists in 4 multi-storey housing blocks, presented in the plan in Figure 1, with a total of 179 dwellings with typologies that vary from T1 to T5, in which live about 450 inhabitants. The retrofitting works were concluded in 2011 [5]. The buildings were subjected to an envelope retrofitting (façades, roofs and windows), a common areas renovation, and a building infrastructures replacement, with the overall cost of 2.2 Million €. The envelope retrofitting could preserve the original appearance of the façade, and at the same time could increase the building´s energy efficiency and the thermal comfort of the residents. Additionally it was included the changing of the ventilation schemes of the buildings. Natural ventilation was adopted with improvement of the air admission points, and continuous mechanical extraction was applied in the kitchens [5]. This type of dwellings, due to climate and socioeconomic context, are not operated with continuous heating or cooling systems. The buildings aspect before and after retrofitting is presented in Figures 2-5. Figure 2. Pitched Roofs (Before and After Retrofitting) [6]. Building 14 Building 15 Figure 3. Room´s Windows (Before and After Retrofitting) [6]. Building 13 Building 12 Figure 1.Buildings location [5]. Figure 4.Kitchen´s Window (Before and After Retrofitting) [6]. The original solution before retrofitting included cavity walls without thermal insulation, pitched roofs covered with asbestos plaques without thermal insulation, and windows in wood frame with single glazing (the solar factor is 0.85). After retrofitting, the façade walls were kept cavity walls without thermal insulation, pitched roofs were covered with insulated aluminium panel sandwiches (50 mm of Mineral wool-MW) and insulated at the ceiling level with 80 mm of MW. The windows Figure 5. Ventilation (Before and After Retrofitting) [6]. were replaced by aluminium frames with double glazing (the solar factor is 0.63) [5].Table 1 summarizes the U-values of the building envelope, before and after retrofitting. 3 SUMMER INDOOR MEASUREMENTS An experimental campaign that lasted the months of July, August and September of 2012 was carried out to assess summertime thermal comfort in a set of 24 dwellings in Lordelo do Ouro´s neighbourhood. To evaluate the number of overheating hours continuous indoor air temperature and relative humidity measurements, with 10 minutes resolution, were per- Table 1. Envelope U-Values before and after retrofitting [5]. Building ElU-value before retU-value after retement rofitting (W/m2K) rofitting (W/m2K) Façade wall 1.3 1.3 Pitched roof 3.4 0.45 Windows 4.1 2.8 (Uwdn) 366 formed in living rooms and bedrooms in each of the 24 dwellings, using Hobo data loggers. As the measured rooms have the same geometry it is possible to evaluate the variability of the indoor air temperature under different boundary conditions (e.g. upper floor rooms, middle floor rooms and ground floor rooms), with different solar orientations (e.g. rooms facing Northeast, rooms facing Southeast), with different occupation behaviour (e.g. rooms with 1 old lady or rooms with 3 kids), in different periods of time (e.g. 3 months measurements versus 2 or 3 weeks measurements). Figures 6-8 represent the 24 measured dwellings in the building elevations (the dark colour square represents the dwelling measured for 1 year, the medium colour square the 3 dwellings measured for 3 months, and the light colour square the 20 dwellings measured for short periods of 2/3 weeks). Table 2 resumes the methodology adopted for the measurements. 4 SUMMER THERMAL COMFORT To simplify the summer thermal comfort analysis it is presented in this paper the study for one dwelling. The room chosen for analysis is the northeast bedroom placed in an upper floor dwelling (most critical position) in Building 14, measured for a whole period of 3 months. As previously referred, the comfort analysis is done according to both EN15251:2007 [1] and ASHRAE55-2004 [2] standards, applying an adaptive approach. The use of adaptive thermal comfort, assumes that given a long enough period of time (typically days or weeks), people will adjust their own clothing levels, behaviour and environment to suit external thermal conditions and therefore they are able to tolerate a wider range of thermal conditions [7] [8] [9]. Figure 9 shows the indoor air temperature (Ti) in the bedroom against the prevailing outdoor temperatures over the previous week (Trm -outdoor running mean temperature) for 2 categories of temperature limits: category II valid for Normal level of expectation and used for new buildings and renovations, category III valid for an acceptable, moderate level of expectation and may be used for existing buildings [1]. Figure 6.Measured Dwellings in Building 12. Figure 7.Measured Dwellings in Building 14. Use an A4 Figure 9. Overheating Analysis according EN15251:2007 [1]. Figure 8.Measured Dwellings in Building 15. Table 3 shows, both for categories II and III, that the number of hours of overheating is almost zero. Table 2. Measurements methodology. Build. Period of Time Measured Parameter 3 weeks T. (ºC) and RH(%) 14 4 Upper Floor 4 Middle Floor 4 Ground Floor 1 Dwelling for 1 year , 3 for 3 months, 8 for 3 weeks T. (ºC) and RH(%). CO2 measured in the dwelling instrumented 1 year. 15 6 Upper Floor 4 Middle Floor 3 weeks T. (ºC) and RH(%) 12 Dwellings 1 Upper Floor 1 Ground Floor Table 3. Summer thermal comfort according to EN15251 [1]. Thermal Comfort / EN15251 Category Hours of overheating/Total of Hours II 3/2208 (0,1%) III 0/2208 (0%) 367 Figure 10 shows the indoor air temperature (Ti) in the bedroom against the mean monthly outdoor temperature (Tmeanmonthly) for two sets of temperature limits —80% and 90% acceptability. The 80% acceptability limits are for typical applications and the 90% acceptability limits are used when a higher standard of thermal comfort is desired [2]. plying some statistical analyses over the data collected. The variability on the results is analysed for 3 different scenarios: a first one which compares, for a single dwelling (upper floor dwelling), the measurements results over a continuous period of 3 months, with the results obtained considering 4 periods of 3 weeks; a second one which compares, for 3 dwellings placed in vertical line (upper floor, middle floor and ground floor dwelling), with the same solar orientation, same occupation, and similar period of measurement, the influence of the shape factor on the results obtained; a third scenario which compares, for all the measured dwellings with the same solar orientation, placed in the upper floor level, with similar period of measurement, the influence of the occupation on the final results. So that it is possible to analyse a large amount of data collected in different periods of measurement, the results are presented in terms of the parameter “air temperature differences between indoors and outdoors (ºC)”, using statistics percentiles. Figure 10. Overheating Analysis according ASHRAE55:04 [2]. Table 4 shows for the 80% acceptability limits, a number of hours of overheating of 1.2% of the total hours. Despite being higher than the number of hours obtained using the EN15251 [1] criteria, the value assures a low overheating risk. Table 4. Summer thermal comfort according ASHRAE55 [2]. Thermal Comfort / ASHRAE 55 Acceptability Hours of overheating/Total of Hours 80% 27/2208 (1,2%) 90% 85/2208 (3,8%) 5.1 Variability versus measurement period (3 months versus 3 weeks) Figure 11 shows the hourly outdoor air temperature Tout (oC) for the months of July, August and September of 2012, in Porto. The temperature range varies between 12ºC and 39ºC. Figures 9 and 10 represent the indoor temperature (Ti) variation considering an adaptive approach for summer thermal comfort. This approach is valid when there is no mechanical cooling system installed, It is assumed that occupants will adjust their clothing, open windows and be more tolerant of higher indoor air temperatures (Ti) when the outdoor temperature (Tout) increase. The results show that according to the adaptive approach contained both in EN15251 [1] and ASHRAE55 [2] standards the summer thermal comfort is satisfied. According to ASHRAE Handbook—Fundamental [10], the indoor air temperature (Ti) may be used as a proxy for operative temperature (Toperative) under certain conditions described in Appendix C of the ASHRAE 55-2004 [2], which are fully fulfilled by the analysed dwellings. Figure 11. Hourly outdoor temperatures for summer months. Figure 12 and Table 5 shows the results in terms of “air temperature differences between indoors and outdoors, Ti-Tout (ºC)” for a bedroom facing northeast placed in the upper floor dwelling, considering a single period of 3 months of measurement, or 4 consecutive periods of 3 weeks. 5 VARIABILITY ON DWELLINGS UTILIZATION CONDITIONS The monitoring campaign carried out for the summer months of 2012 allowed the collection of a considerable amount of data. The dwellings that were instrumented have different solar orientations, boundary conditions, occupation, and were measured for different periods of time. It is possible to analyse the variability of the measurements done ap368 100% also possible to see for the percentile 10%, a negative correlation between Ti-Tout (ºC) and Tmean outdoor (ºC). Percentiles Ti -Tout (ºC) - Bedroom Facing Northeast in a UpperFloor Dwelling Measurement Period Variation Ti-Tout Ti-Tout Ti-Tout Ti-Tout Ti-Tout 90% 80% [OC]/3 Months [OC]/1st Period (1/7 to 23/7) [OC]/2nd Period (24/7 to 16/8) [OC]/3rd Period (17/8 to 8/9) [OC]/4th Period (9/9 to 30/9) Percentile 70% 60% 5.2 Variability versus dwellings shape factor 50% 40% Figure 14 and Table 6 shows the results in terms of “air temperature differences between indoors and outdoors, Ti-Tout (ºC)” for a bedroom facing northeast placed in 3 different dwellings in the same building and the same vertical alignment (an upper floor dwelling, a middle floor dwelling and a ground floor dwelling), considering a period of 3 months of measurements (1/7 to 30/9). The 3 dwellings have the same geometry, solar orientation and occupancy. 30% 20% 10% 0% -12.0 -10.0 -8.0 -6.0 -4.0 -2.0 0.0 2.0 Ti - Tout [ºC] 4.0 6.0 8.0 10.0 12.0 Figure 12. Graphical analysis for scenario 1 “3 months versus 3 weeks measurement”. Table 5. Statistical percentiles analysis for scenario 1“3 months versus 3 weeks measurement”. 100 90 50 10 0 3 Months 1st (1/7 to 23/7) 11.1 7,8 4,3 -1,0 -11.4 11,1 8,5 4,5 -1,2 -11,4 Ti-Tout [oC] 2nd (24/7 to 16/8) 9,7 7,8 4,6 -0,3 -4,8 3rd (17/8 to 8/9) 4th (9/9 to 30/9) 90% Ti-Tout [OC]/Upper Floor Dwelling 9,9 7,4 3,4 -1,7 -8,3 10,0 80% Ti-Tout [OC]/Middle Floor Dwelling 7,6 4,6 -1,1 -7,5 70% Percentiles Ti-Tout (ºC) - Bedroom Facing Northeast in a Vertical Line Shape Factor Variation 100% Percentile % Ti-Tout [OC]/Ground Floor Dwelling 60% 50% 40% 30% 20% The difference in considering a whole period of 3 months or small periods of 3 weeks is less than 1ºC for the percentile 50% (median). Differences even lesser than that, are obtained for percentiles 90% and 10%. Figure 13 shows the correlation between the “air temperature differences between indoors and outdoors, Ti-Tout (ºC)” and outdoor mean temperature, Tmean outdoor (ºC) in the measuring period. 10 10% 0% -14.0 Ti -Tout (o C) 6 Ti-Tout (90%) Ti-Tout (50%) Ti-Tout (10%) 20.00 20.50 21.00 Tmean outdoor(o C) 21.50 -6.0 -4.0 -2.0 0.0 Ti - Tout [ºC] 2.0 4.0 6.0 8.0 10.0 12.0 14.0 % Ti-Tout [oC]/UpperFloor Ti-Tout [oC]/Middle Floor Ti-Tout [oC]/Ground Floor 100 90 50 10 0 11,1 7,8 4,3 -1,0 -11,4 12,5 8,6 4,5 -1,0 -11,2 11,1 8,3 4,7 -0,8 -13,1 The results presented in Figure 14 and Table 6 don´t allow a definition of a general trend. The warmest dwelling with higher “air temperature differences between indoors and outdoors, Ti-Tout (ºC)” is the one placed in the middle floor. Also no extreme differences were found indicating that the variation of the shape factor can represent a determinant parameter to analyse the variability on the indoor air temperature. 4 -2 19.50 -8.0 Table 6. Statistical percentiles analysis for scenario 2 “shape factor variation in a vertical alignment”. 8 0 -10.0 Figure 14. Graphical analysis for scenario 2 “shape factor variation in a vertical alignment”. Correlation between Ti - Tout and Tmeanoutdoor (3 months measurements with 4 periods of 3 weeks) 2 -12.0 22.00 Figure 13. Correlation between Ti-Tout and Tmeanoutdoor (ºC). By the analysis of the results it is possible to assume that we can take a small period of indoor air temperature measurements and even so it can be approximately representative of a larger period. Except for 3rd period of measurements (“between” 17/8 to 8/9) which is represented, in Figure 13, by the three dots at right, it is possible to say that we have almost 3 horizontal lines approaching the dots. The exception was a very warm 3 weeks period (17/8 to 8/9), where the Tmean outdoor (ºC) deviated approximately 2ºC from the 3 months Tmean outdoor. In Figure 13 it is Figure 15 and Table 7 show the results in terms of “air temperature differences between indoors and outdoors, Ti-Tout (ºC)” for a bedroom facing northeast placed in an upper floor of 3 different buildings, considering a period of 2 weeks of measurements. The 3 dwellings have the same geometry and occupancy. The horizontal alignment of the bedrooms is the same (there is a bedroom placed next to the ga369 Table 8. Statistical percentiles analysis for scenario 3 “people occupation variation”. ble SE, another placed to the gable NW, and the third is placed in the centre of the building). % 10 90 50 10 0 Table 7. Statistical percentiles analysis for scenario 2 “shape factor variation in a horizontal alignment”. Ti-Tout [ C]/Gable NW Ti-Tout [oC]/Gable SE Ti-Tout [oC]/Center 100 90 50 10 0 9.8 7.8 4.6 0.1 -5.0 11.1 7.8 4.5 -1.4 -5.8 10.0 7.6 4.2 -0.1 -5.0 o B C 10,4 8,0 4,9 -0,8 -4,7 10,0 7,6 4,2 -0,1 -5,0 11,0 9,1 5,0 0,3 -4,7 9,8 7,8 4,6 0,1 -5,0 11,8 8,9 5,9 0,1 -4,4 G H I 11,7 8,7 5,5 -0,5 -4,6 11,1 7,8 4,5 -1,4 -5,8 12,1 9,2 6,1 0,3 -4,2 Figure 16 shows a spread created by the graphics for the 9 dwellings considered that is due to the occupancy. The spread for the percentile 50% (median) is of 1.7ºC, and for the percentile 90% is of 1.4ºC. The dispersion in the median value is the evidence that the difference in people occupation for similar dwellings, can bring out considerable differences in terms of the value for the indoor air temperature. Dwellings with higher level of occupation, great number of technical appliances (tv sets, personal computers, stereo equipments, etc.), and a higher number of household equipments and lighting levels have a higher indoor air temperature, as shown is Figure 15. The curves that are shifted to the right represent the bedrooms with higher internal gains. Figure 15. Graphical analysis for scenario 2 “shape factor variation in a horizontal alignment”. % A 10,4 8,2 4,5 0,2 -5,0 Ti-Tout [OC] D E F Once again the results presented in Figure 15 and Table 7 don´t allow a definition of a general trend. However it is possible to observe that the dwelling with the highest peak temperature is the one placed closer to gable SE. 6 CONCLUSIONS “Lordelo do Ouro” is a social housing neighbourhood in Porto, Portugal, recently retrofitted as a result of insufficient maintenance and natural decay due to its end of service life. To evaluate summer thermal comfort for the occupants after the retrofitting, it was carried out an experimental campaign in a set of 24 dwellings, during the months of July, August and September 2012. The main conclusions of the study are: 1) By the analyses of the graphic presented in Figure 11 it is possible to note that the monthly summer outdoor temperatures in Porto, during 2012, reached values above 30ºC in many days. The summer thermal comfort is an important issue to analyze. 2) According to the adaptive approach contained both in EN15251 [1] and ASHRAE55 [2] standards, the overheating risk for all monitored dwellings is very limited. The number of hours of overheating during summer months never exceeds 3, according to EN15251 [1], and 27 according to ASHRAE55 [2]. 3) By the analysis of the effect of variability in the period of measurements (3 months versus 3 weeks) it is possible to conclude that the spread on the indoor air temperature for a shorter period of measurement is less than 1oC. The correlation of the“air temperature differences between indoors and outdoors, Ti-Tout (ºC)” with the “outdoor mean tem- 5.3 Variability versus dwellings occupancy Figure 16 and Table 8 show the results in terms of “air temperature differences between indoors and outdoors, Ti-Tout (ºC)” for a bedroom facing northeast placed in 9 different dwellings with the same geometry, solar orientation and shape factor, located in the upper floor of different buildings, considering a period of 2 weeks of measurements. Figure 16. Graphical analysis for scenario 3 “occupancy variation”. 370 [8] McCartney K.J., and Nicol J.F. 2002. Developing an Adaptive Control Algorithm for Europe: Results of the SCATs Project. In Energy and Buildings 34(6): pp 623-635. [9] Brager G. S., and De Dear R. 2001. Climate, Comfort, & Natural Ventilation: A new adaptive comfort standard for ASHRAE Standard 55; University of California. Berkeley, USA. [10] ASHRAE 2005, Fundamentals Handbook, American Society of Heating, Refrigeration and Air-Conditioning Engineers. Atlanta, USA. perature, Tmean outdoor (ºC)” made possible to conclude with a certain approximation level, that we can take a small period of indoor air temperature measurements and even so it can be approximately representative of a larger period. 4) By the analysis of the effect of the shape factor on the indoor air temperature, it is possible to conclude that there is no evident trend when we analyze its variation in the building, along a vertical or a horizontal alignment. The effect of the roof thermal insulation is perhaps decisive on the obtained results. 5) By the analysis of the effect of variability in the occupancy of the dwellings, it is possible to conclude that for the same type of dwellings, we can have a difference of 1.7 oC on the indoor air temperature during the same period of time. The results show that the influence of the variability of the human behavior, on the indoor air temperature proved to be the most relevant parameter when we study the indoor climate of the dwellings. 6) To assess summer thermal comfort and variability analysis, sets of measurement data were collected and statistically treated. It is possible to observe differences of 1.7ºC for most of percentiles of indoor air temperature accumulated values. 7) According to the results presented in Section 4, the strategy adopted for the “Lordelo do Ouro” neighborhood’s retrofitting seems to be adequate to get summer thermal comfort, without the installation of cooling equipments in the dwellings. 7 REFERENCES [1] EN ISO 15251 2007-08. Indoor environmental input parameters for design and assessment of energy performance of buildings addressing indoor air quality, thermal environment, lighting and acoustics, Standard 15251-2007. Brussels, Belgium. [2] ASHRAE 2004. Thermal Environmental Conditions for Human Occupancy, Standard 55-2004. Atlanta, USA. [3] Kalamees T., Arumagi E., and Ilomets S. 2011. The Analysis of Indoor hygrothermal conditions in multi-storey wooden apartments buildings. In Annex 55 Working Meeting. Porto, Portugal. [4] Hong S.-H. 2011, Changes in Space Heating Energy Consumption Following Energy Efficient Refurbishment in Low-Income Dwelling in England. A Thesis submitted for the Degree of Doctor of Philosophy to the Bartlet School of Graduate Studies, University College London. London, England. [5] Freitas V.P. 2007-2009. Projeto de Reabilitação do Bairro de Lordelo do Ouro. Porto, Portugal. [6] Curado A. 2012., Conforto Térmico e Eficiência Energética nos Edifícios Residenciais Reabilitados. Projeto de Tese de Investigação apresentado à Faculdade Engenharia da Universidade do Porto, Porto, Portugal. [7] Nicol F., and Pagliano L. 2009. Allowing for Thermal Comfort in Free-Running Buildings in the New European Standard EN 15251. 371 372 Thermal environment in a room with dynamic infrared fireplace heater R. Ponechal University of Zilina, Department of Building Engineering and Urban Planning, Zilina, Slovakia M. Kanderkova University of Zilina, Department of Building Engineering and Urban Planning, Zilina, Slovakia ABSTRACT: Fireplace is considered to be a traditional symbol of comfort. It is a part of a dual heating system, together with a gas boiler, is the most popular way of heating in family houses. Both systems used to work simultaneously, therefore the air temperature of a room would not drop significantly in case of a fireplace burnout. This paper reports on the results from measuring and simulations in two selected freestanding houses equipped by a fireplace. The houses had different levels of thermal performance. This study focused on thermal comfort of the room where the fireplaces were situated. Indoor surface temperatures were measured very carefully and ESP-r (Energy Systems Research Unit) simulation tool was used afterwards to assume the mean radiant temperature for three phases: the first one simulated thermal comfort in a living room without a burned fireplace (thermal comfort was obtained by additional heating system). The second and the third phases simulated thermal comfort in a room with the fireplace, its behaviour in the initial phase of heating and in the phase of its normal functioning. Then there was studied the temperature variations of the external walls. 1 INTRODUCTION tures (Karjalalainen 2007, Lenzun 2008). Thermal comfort for people with different thermal sensitivity can be provided by changing the position in a room Even though usage of fireplaces has significant (Guidard 2000). The fireplace heating has potential downsides, they are still popular in central Europe to reduce total energy use by focusing on local heatmore than 40% of newly built free-standing houses ing systems (Sørlie 1993). When using 60 W heater have some. Fireplace heaters represent 20 % of all placed under the desk of a sitting person, he reported heaters sold between 2010 and 2011 in the Czech acceptable overall thermal comfort while ambient Republic (Blažíček 2012). In 2003 regional domestic room temperature was 17-18 °C. consumption of biomass in fireplace heaters, stoves and boilers was estimated at 70% of the total use of biomass in Lombardy (Caserini 2010). According to this study, more than 85% of domestic biomass heat2 FIELD MEASUREMENT ers are open or closed fire-place heaters or stoves. On the other hand, the low-emitting high-efficiency It has been already mentioned that the most popular pellet stoves are still in marginal use. A review about heating systems in family houses is the dual heating emission factors for different combustion technolosystem (gas boiler with a fireplace). There were two gies was performed by Department of Hydraulic Enhouses chosen for this study, with different level of gineering, Environmental, Transportation Infrastructhermal performance and thermal comfort, named tures and Detection. Due to the extreme variability house ’A’ and house ’B’. of emissions, caused by different chemical and physical characteristics of biomass, type and age of 2.1 Experimental house ´A´ stoves, their operative conditions and average emission factors according different pollutants and techHouse ’A’ was designed and built in 1960’s. Its exnologies, were identified (Tab. 1). ternal walls (300 mm thick) were made out of cinder concrete with an internal brick layer. The floor was renovated and well-isolated, with a polystyrene insuTable 1. Characteristics of combustion systems (Caserini). Combustion technology Efficiency CO emissions lation boards. The ceiling was a timber-wood con% g/GJ struction with plaster coating. The fireplace heater Open fireplace 18 5650 with maximal power 7 kW was situated in the corner Conventional wood stove 55 5650 of the living-room (4.9 x 3.1 m) (Fig. 2). The measLow emission wood stove 68 2260 urements were done from 15th to 19th February. The Pellet stove 82 620 interior air temperature and relative humidity was Natural gas domestic boiler 25 monitored, as well as the external air temperature and the interior surface temperature of selected exFireplace is not only a traditional symbol of comternal constructions, in a quarter-hour intervals. Exfort but also a task-ambient heating system. Females, ternal air temperatures during measurements fluctuchildren and elderly people prefer higher tempera373 and on the floor behind the fireplace. In the first case it was because of convection and in the second case it was because of the radiation from the fireplace. ated from -0.1°C to 6.9°C. Internal relative humidity on 15th February was around 50%. User mode was characterized by more frequently ventilation, caused by the smell of burnt gas from the stove. 30 Air temperature (°C) 28 26 24 22 15. 2. 2013 16:35 15. 2. 2013 16:20 15. 2. 2013 16:05 15. 2. 2013 15:50 Figure 1. The living-room with fireplace heater in the experimental house ´A´. 15. 2. 2013 15:35 15. 2. 2013 15:20 20 Air temperature - hight 1.1 m Air temperature - hight 0.1 m Air temperature - hight 2.4 m Figure 3. Air temperature stratification in house ´A´. Figure 2. Floor-plan of the living-room in the house ´A´. 2.2 Thermal environment in house ’A’ Thermal environment in the room was measured carefully with thermometers, airflow meters and infrared camera. The measurements were done on 15th February in these three phases: gas boiler heater ran (fireplace was not running), fireplace at the initial phase of heating and fireplace in the phase of its normal functioning. In the initial phase of heating the temperature of the fireplace heater body was around 250 °C and in the phase of its normal functioning it was 200 °C. Figure 3 visualized measurements of the air temperature in the living-room in three different heights. It is evident that the air temperature stratification grew dramatically when the heating systems were changed (from the gas boiler to the fireplace). Figures 4-9 show records of the internal surfaces during the measurement taken by the infrared camera. Internal surface temperatures of the gas stove and the fireplace are shown in one hour intervals. The biggest changes were seen on the ceiling over the fireplace Figure 4. The infrared camera picture shows the surface temperatures on the ceiling in house ´A´ within the gas boiler heating (fireplace do not runs) - the 1st phase. Figure 5. The infrared camera picture shows the surface temperatures on the ceiling in house ´A´ in initial phase of fireplace running - the 2nd phase. 374 Figure 6. The infrared camera picture shows the surface temperatures on the ceiling in house ´A´ during the phase of normal running of the fireplace - the 3rd phase. Figure 9. The infrared camera picture shows the surface temperatures on the floor in house ´A´ during the phase of normal running of the fireplace - the 3rd phase. 2.3 Fireplace operation in the house ’A’ The influence of the fireplace running on the temperature changes is demonstrated in Figure 10. 2.4 Thermal comfort in house ’A’ The ESP-r (Energy Systems Research Unit) simulation tool was used to calculate the mean radiant temperature from the obtained data (surface temperatures), recorded by the infrared camera and the infrared thermometer. The measurements were done at selected points on the walls, floor and ceiling of the living-room. There were 3-4 points of measurement in each of the 3 height-points on each wall. Figure 11 shows the simulation model of the mean radiant temperature. The surfaces were distinguished according to their temperature. Each surface represents a region with a united temperature with accuracy 1.0 K. Surface temperatures were calculated in the program ESP-r as boundary conditions for the construction with high conductivity and high coefficient of heat transfer by convection. The program is able to calculate the mean radiant temperature in selected position as “MRT sensor”. Obtained results have been summarized in the top view maps of the mean radiant temperature during three phases (Fig. 13-15). Visualization has been converted into a diagram, which shows the dependence of the mean radiant temperature and the distance from the fireplace (Fig. 12). Figure 7. The infrared camera picture shows the surface temperatures on the floor in house ´A´ within the gas boiler heating (fireplace do not runs) - the 1st phase. Figure 8. The infrared camera picture shows the surface temperatures on the floor in house ´A´ in the initial phase of fireplace running - the 2nd phase. 375 Figure 10. The temperatures in house ´A´. Figure 13. The top view map of the mean radiant temperature in house ´A´ during the 1st phase (fireplace did not run) measured in the height 1.1 m. Mean radiant temperature (°C) Figure 11. Simulation model of experimental room in house ´A´. 50 Fireplace heating - stage 1 45 Figure 14. The top view map of the mean radiant temperature in house ´A´ during the 2nd phase (at the initial phase of fireplace running) measured in the height 1.1 m. Gas boiler heating 40 Fireplace heating - stage 2 35 In this study we used the PMV index (Fanger 1973) to present the results and to indicate the local thermal comfort. The thermal comfort indices were calculated according to these conditions: clothing 1.0, air temperature 22.0 °C, activity 1.0 met, air speed 0.05 m/s, relative humidity 50%. The maps of PMV index were displayed in Figures 16-18. 30 25 20 0.5 1.0 1.5 2.0 2.5 3.0 Distance (m) Figure 12. Dependence of the mean radiant temperature on distance from the fireplace in high 1.1 m. 376 Figure 17. Top view map of predicted mean vote (PMV) index in house ´A´ during the 2nd phase (at the initial phase of fireplace running) measured in the height 1.1 m. Figure 15. The top view map of the mean radiant temperature in house ´A´ during the 3rd phase (the phase of normal running of the fireplace) measured in height 1.1 m. 2.5 Experimental house ’B’ The second experiment took place in house ´B´, witch was built between 2003 and 2006. External walls of the house (300 mm thick) are made out of clay blocks with an external polystyrene insulating board. The floor is well isolated with a polystyrene insulating boards. The ceramic ceiling is coated with plaster. Figure 16. The top view map of predicted mean vote (PMV) index in house ´A´ during the 1st phase (fireplace did not run) measured in height 1.1 m. Figure 18. The top view map of predicted mean vote (PMV) index in house ´A´ during the 3rd phase (the phase of normal running of the fireplace) measured in height 1.1 m. 377 50 Air temperature (°C) 45 40 35 30 25 1. 2. 2013 18:05 1. 2. 2013 17:50 1. 2. 2013 17:35 1. 2. 2013 17:20 1. 2. 2013 17:05 1. 2. 2013 16:50 1. 2. 2013 16:35 Figure 19. The living-room with fireplace heater in house ’B’. 1. 2. 2013 16:20 1. 2. 2013 16:05 20 Air temperature - hight 1.1 m The fireplace heater with maximal power 12 kW is situated in the corner of the living-room (4.9x2.95 m) (Fig. 20). The measurements were done from the 1st to 6th February. The interior air temperature and relative humidity was monitored, as well as the external air temperature and the interior surface temperature of selected external constructions, in a quarter-hour intervals. External air temperatures during the measurements fluctuated from -2.6°C to 5.6°C. Internal relative humidity on 1st February was around 27%. User mode was characterized by higher interior air temperature. Air temperature - hight 0.1 m Air temperature - hight 2.4 m Figure 21. Air temperature stratification in house ’B’. Figures 22-23 are show pictures of the internal surfaces during the measurement, taken by the infrared camera. The biggest changes were seen on the ceiling over the exhausts because of the hot air flow from the fireplace. Figure 22. The infrared camera picture showing the temperatures on the wall in house ’B’ with gas boiler heating system. Figure 20. Floor-plan of the living-room in house B. 2.6 Thermal environment in house ’B’ Thermal environment in the room was measured accurately with thermometers, airflow-meters and the infrared camera on 1st February in two stages: with a gas boiler heating and with a fireplace heating. Figure 21 documented the measurement of the air temperature in three different heights. Figure 23. The infrared camera picture showing the temperatures on the wall in house ’B’ with fireplace heating. 378 2.7 Fireplace operation in house ’B’ The effect of the fireplace operation on the temperatures can be seen in Figure 24. Figure 26. Top view map of the mean radiant temperature in house ´B´ in height 1.1 m. Figure 24. The temperatures in house ’B’. Thermal comfort indices were calculated according to these conditions: clothing 0.7 clo, air temperature 28.8 °C, activity 1.0 met, observed air speed 0.05 m/s, observed relative humidity 30%. The map of PMV index is documented in Fig. 28. 2.8 Thermal comfort in house ’B’ Mean radiant temperature (°C) The ESP-r simulation tool was used again to calculate the mean radiation temperature from the observed surface temperatures. Figure 25 shows a simulation model of the living-room in house B. The results are summarized by the top view map of mean radiant temperature (Fig. 26, 28) and the graph of functional dependence of mean radiant temperature on distance from the fireplace (Fig. 27). 50 45 40 35 30 25 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 Distance (m) Figure 27. Dependence of mean radiant temperature on distance from fireplace measured in height 1.1 m. Figure 25. Simulation model of the living-room in house ´B´. 379 with the PMV index from 1.0 to 2.0 (slightly warm warm) in the distance 2.0 – 3.0 m. This zone should be suitable for females, children and elderly people. 4 CONCLUSION Power and location of a fireplace should correspond with thermal protection of a house, as well as the function of the space. This paper makes a methodology how to determine the thermal comfort parameters in an existing room heated by a fireplace. Thermal comfort is documented in the maps of PMV index. It has been shown that the fireplace heating system can obtain sufficient thermal comfort parameters if the design is planned carefully and insufficient thermal comfort parameters if it is not (for example in case of oversizing heating system). Studied heating system can be used as the task ambient system to achieve thermal comfort for a specified group of people sitting near a fireplace. 5 REFERENCES Figure 28. Top view map of the predicted mean vote (PMV) index in house ´B´ measured in the height 1.1 m. Blažíček J. 2012. Sales of boilers in 2011. Information on: http://www.tzb-info.cz 18.6.2012 Caserini S. 2010. LCA of domestic and centralized biomass combustion. The case of Lombardy (Italy). In Biomass and bioenergy 34: 474–482. SinceDirect. Fanger P. O. 1973. Thermal Comfort. New York: Mc Graw Hill. Guidard S.E. 2000. Heat Radiation from Flames. Report prepared for Science and Technology Branch, Alberta Environment, ISBN 0-7785-1188-X, Edmonton, Alberta. Karjalalainen S. 2007. Gender Differences in Thermal Comfort and use of thermostats in everyday thermal environments. In: Building and Environment 42: 1594-1603 Lenzuni P. 2008. Classification of Thermal Environments for Comfort Assessment. The Annals of Occupational Hygiene 2009 53 (4): 325-332. Sørlie R. 1993. Maintaining acceptable thermal comfort with a low air temperature, by means of local heating. In Indoor Air Congress. Helsinki – Finland. 3 DISCUSSION 3.1 Experiment in house ’A’ The measurements in this study show changes in the temperatures from high entropic gas boiler heating system to low entropic fireplace heating system. Thermal comfort in the room with a fireplace heating system depends on the distance from the fireplace. For a man with an ordinary daytime clothing (0.7 clo), there was a PMV index 0.5 – 1.0 (slightly warm) in the distance 1.0 – 2.0 m from the fireplace. Further from the fireplace the thermal comfort was classified as neutral. Even though surface temperatures on the walls in the stage 3 (at the phase of its normal running) were higher than in stage 2 (fireplace in the initial phase of heating), the determining factor of the local thermal comfort was the temperature of the fireplace body. 3.2 Experiment in house ’B’ House ’B’ was well-isolated with a powerful fireplace. In the living-room the air temperature was as high as the well as high temperature stratification (more than 5.0 K between the height 0.1 m and 1.1 m). For the people with ordinary daytime clothing (0.7 clo), there was PMV index from 2.0 to 2.5 (warm) in the distance 1.0 – 2.0 from fireplace. Normally this would be a thermal discomfort zone. There is still a thermal discomfort behind this zone 380 Investigation of ceiling fans for improving summer thermal comfort K. Voss, T. Voß, J. Otto University Wuppertal, Dep. Architecture, Building Physics & Technical Services, Wuppertal, Germany M. Schweiker Karlsruhe Institute of Technology, Building Physics & Building Science, Karlsruhe, Germany E. Rodriguez-Ubinas Technical University Madrid, Dep. of Construction & Building Technology, Madrid, Spain ABSTRACT: Although most familiar in tropical climates and in Southern Europe, ceiling fans are also able to improve summer comfort in building interiors in Central Europe. This is achieved by improved air movement near occupants and modified heat transfer from building components. The paper reports on experimental characterisation of the air flow and the power consumption of a typical ceiling fan. Based on this, the influence on the heat transfer conditions at the ceiling and the indoor thermal conditions in a test cell were analysed. User surveys were carried out and analysed. Measurements and surveys confirm the significant effect of increased air flow due to the operation of the fan during the day. Although a change in ceiling heat transfer was measured for night operation, the effect is less pronounced than expected. Further detailed measurements are planned. The calculations with a single zone simulation model show that the peak room temperatures are slightly reduced. More detailed modelling is required to consider effects like stratification and local temperature gradients. 2006, DIN EN 15251 2007). In contrast to increased ventilation through windows, ceiling fans create air movement without introducing additional heat loads from the outside air. Air movement can be achieved even in rooms with deep floor plans, and can be controlled individually with desk fans or ceiling fans mounted over a desk. ○ Increased heat transfer to and from building components: When building components, particularly slab ceilings, are cooled at night by ventilation of the rooms with outdoor air, often the achieved effect is inadequate, despite high air flow rates. In combination with small temperature differences, the heat transfer rate is too low. When phase change materials are used in the ceiling zone, often the intended phase change at night does not occur (Eiker 2007, Engelmann 2010). Higher air flow rates usually cannot be achieved (free convection) or require a disproportionately large amount of electricity (fanassisted ventilation). Ceiling fans can specifically increase the convective heat transfer (night operation, air flow directed toward the ceiling). ○ Reduction of temperature stratification in high spaces such as atria due to the air circulation to improve thermal comfort in the occupied zone. ○ Increase in heat transfer to and from cooling and heating surfaces: The performance of heat-transfer surfaces can be increased significantly, locally and at specific times, by increasing convection due to the operation of ceiling fans (operation during the day, air flow directed downward). These benefits are partly offset by the electricity needed to drive the fans. Energy-efficient models operate with electronically commutated DC motors. Differences in the speed of rotation, number of vanes and construction principle cause differences in the 1 INTRODUCTION The application of ceiling fans to improve indoor thermal comfort in buildings is widespread in tropical and subtropical climatic zones. The fans there offer a technically simple, inexpensive, individually operable and, above all, effective method to increase air movement and thus thermal comfort in a room (Fairy 1986, Rohles 1983, Sekhar 1995, Aynsley 2005). Ceiling fans are also commonly used in Southern Europe, but they have seldom been used in Central Europe to date. The hot summers of 2003 and 2006 provoked discussion on summer comfort in building interiors and stimulated a number of research projects in Germany and critical adaptation of standards concerning protection against overheating in summer (Voss 2007). The predicted climate change and the anticipated temperature increase and heat waves have strengthened interest in this topic (Voss 2012). The focus is on non-residential buildings such as offices, schools and hotels. In principle, ceiling fans can create a number of effects in indoor rooms, of which the first two are the subject of the investigations presented in this paper: ○ Improvement of thermal comfort when indoor room temperatures are high: Air movement in the immediate vicinity of the human body increases convective heat transfer and the evaporation rate. Above a room temperature of 28 °C, heat transfer is dominated by evaporation, as the temperature difference between the body surface and either the air (convection) or the surrounding surfaces (radiation), which is needed for the heat transfer, decreases (Arens 2009, Schiavon 2008, DIN EN ISO 7730 381 for distances of up to 45 cm from the axis of rotation, the maximum in the longitudinal and transverse directions were found at a distance of about 75 cm from the fan axis. Thereafter, the speed decreases gradually and for the two high rotational speeds is still 80 % of the maximum at a distance of 1.6 m from the axis. During an experiment with the ceiling fan switched off but with the waste air unit operating at an air change rate of 3.5, the air speed near the ceiling was between 0.01 m/s and 0.05 m/s, orders of magnitude lower. ventilation efficacy and the floor area of a room which can be served by a single fan. Most models can be switched to operate at various rates, and the direction of rotation can be reversed. Remote operation controls are common today. 2 CHARACTERISATION OF A CEILING FAN An exemplary, four-vane fan with a DC motor and a vane diameter of 135 cm was selected as typical of the multitude of available fans and was positioned below the centre of the ceiling in an empty test room with a floor area of 15 m², a so-called btga box. There, the air flow profile and the electric power demand were measured for different operating points. The distance between the ceiling and the fan vanes was about 18 cm, while the room height was 285 cm. Of the six power settings, stages 2 (low), 4 (medium) and 6 (high) were used for the air flow measurements. The air speed field at the desk was measured under steady-state conditions (3-minute averages) along two axes in the empty test room with direction-sensitive hot-wire anemometer at heights of 60 cm and 110 cm above the floor, with the air flow from the fan directed downward (Fig. 1). The dominant vertical air speed component below the fan was measured. The measurements along two axes were necessary because the room is relatively small and not square. The local air speeds are subject to greater fluctuations, above all in the low rotational speed range of the fan. In the zone around the legs of a seated person (60 cm above the floor, room unfurnished), the maximum measured air speed varied between 0.4 m/s and 0.8 m/s, depending on the fan’s rotational speed, whereas at head height (110 cm) it was between 0.5 m/s and 1.3 m/s. Beyond 75 cm away from the fan axis, the air speeds decrease noticeably. Usually, a ceiling fan is operated at a low rotational speed when people are present to minimise draughts and noise (low setting, 6 W power demand). Thus, typical speeds lie between 0.25 m/s (60 cm) and 0.5 m/s (110 cm). This is 10 to 20 times higher than without a fan. According to DIN EN ISO 7730, increasing the air speed in the occupied zone of a room to 0.5 m/s already raises the acceptance range under high temperature conditions by 1.7 K (DIN EN ISO 7730 2006, DIN EN 15251 2007). To characterise the zone near the ceiling, the measurement of the air flow field was made 5 cm below the ceiling with the fan operating to direct the air flow upward. An omnidirectional hot-wire anemometer was used to this purpose. Depending on the rotational speed, the maximum air speed near the ceiling varied between 1.1 m/s (low), 1.3 m/s (medium) and 1.5 m/s (high). The maximum electric power demand during these measurements was 30 W, Fig. 4. While only moderate speeds were recorded Figure 1. Floor plan of the test room indicating the two axes for measuring the air flow at working height (person seated). The internal dimensions of the test room are 5.0 m (d), 3.0 m (w) and 2.85 m (h). The adjacent room on the right was used for the data acquisition system and the waste air unit. 1,6 air velocity [m/s] 1,4 1,2 1,0 0,8 long axis-high short axis - high long axis - med short axis - med long axis - low short axis - low 0,6 0,4 0,2 60 cm above the floor 0,0 0 1,6 30 60 90 distance from fan axis [cm] 120 air velocity [m/s] 1,4 1,2 1,0 long axis-high short axis - high long axis - med short axis - med long axis - low short axis - low 0,8 0,6 0,4 0,2 110 cm above the floor 0,0 0 30 60 90 distance from fan axis [cm] 120 Figure 2. Vertical component of air velocity of the downward directed air flow, for varying fan speeds, distance from centre and height above the floor: 60 cm height = upper diagram, 110 cm height = lower diagram. 382 1,6 air velocity [m/s] 3 EXPERIMENTAL INVESTIGATIONS IN ROOMS WITH CEILING FANS 1,4 1,2 Measurements were made in a test room on the grounds of the University of Wuppertal (2012) and also in real occupied rooms in office buildings in Kassel and Tübingen (2011) to investigate the effect of a ceiling fan on heat transfer to building components and indoor summer comfort. All locations are in Germany. The common feature of all rooms, apart from effective solar shading, is night ventilation in combination with the thermal buffering effect of non-suspended, steel and concrete slab ceilings. 1,0 0,8 0,6 0,4 0,2 0,0 0 30 long axis-high short axis - high long axis - med short axis - med long axis - low short axis - low 60 90 120 distance from fan axis [cm] 150 180 Figure 3. Air speed 5 cm below the ceiling as a function of the fan speed and the distance from the fan axis for downward directed air flow. 3.1 The test cell (btga box) power [W] 40 35 As a continuation of the fan characterisation measurements, the test cell was equipped with additional sensors to determine both the heat flux density and the surface temperatures at several positions on the ceiling as well the air temperature distribution in the room (see Fig. 1). The room is equipped with a waste air unit, which maintains an air change rate of 0.9 (40 m³/h) during the day and 3.4 (147 m³/h) during night operation as long as the boundary conditions concerning the room and outdoor temperatures were met. The inlet air then enters in the upper part of the façade via an outdoor air valve. The southfacing glazed façade consists of solar-control glazing. During night operation, the ceiling fan is operated on the highest setting with its air flow directed upward. air flow direction 30 downwards 25 night mode upwards 20 15 10 5 day mode 0 1 2 3 4 fan power setting 5 6 Figure 4. Power consumption of the exemplary ceiling fan at various fan power settings. The two curves indicate the rotation of the fan and the resulting air flow direction. Figure 5. Examples of measurements of heat flux density (upper diagram, 1-minute resolution) and temperatures (lower diagram) for two days with ceiling fan operation during the night, but without active night ventilation as the set points was not reached. 383 were up to 0.6 m/s (Fig. 7). The main trends in results from previous research (Fanger 1970, Candido 2008) are confirmed by these user surveys. During the day, the solar radiation results in a heat flux density of up to 40 W/m². Large differences were determined between the front and back zones of the room (a factor of 2), which are attributed to the radiative heat transfer between the ceiling and the floor, of which only the front area is irradiated. When the heat flux reverses direction during the night, the temperature differences and heat flows are much smaller. Operation of the fans leads to a) complete mixing of the temperature strata and b) doubling of the local heat flow rates (Fig. 5). 400 number I- 300 I+ 200 3.2 Field measurements 100 Field measurements were made in two office buildings. The selection and placement of sensors was similar to that for the measurements in the btga box. In each case, one office was equipped with a ceiling fan. The results were analysed in comparison to an office without a fan. In addition, user surveys were conducted (see Section 3.3). It was determined that ○ the heat fluxes at the ceiling were dominated by the temperature boundary conditions in the space above the monitored office. Adiabatic conditions did not prevail in either case (outdoor air or an office in the top storey and thus at a higher temperature level). ○ a major disadvantage of a comparative analysis concerns the conditions in the two rooms, which are not strictly comparable (presence of occupants, usage of solar shading systems, …). These differences are dominant compared to the effect of fan operation. For this reason, the results of these investigations will not be presented in greater detail here. 0 -2 -1 0 1 2 3 4 Perceived air velocity compared to preferred air velocity Figure 6. Histogram of difference between perceived and preferred air speeds. A value of 0 signifies that perceived and preferred air speeds coincide. Positive values stand for higher preferred air speeds than the currently perceived ones. 60 number 50 40 30 20 10 3.3 User surveys 0 Investigations with human subjects were carried out in the btga box during the summer of 2010 (Schweiker 2012). There were in total six different session types each subject had to participate in. During three of the session types (categorized as i-) the subjects were not allowed to open the window, to switch on the fan or to use the sun shading device. All these options were allowed in the other three session types (i+). Fig. 6 presents the results of the comfort ratings with regard to perceived and preferred air movements (N = 871 spot tests, obtained with 17 test persons). The results show that when a fan may be used (i+), the majority of the subjects did not desire any change in air movement, whereas most subjects wished to have more air movement when the option of using the fan is not available. In both cases, only a very small proportion of people wished to have less air movement. In user surveys conducted during the field measurements in the summer of 2011, the subjects indicated in 80 % of the cases (105 of 132) that they did not wish to change the air flow rate. The air speeds measured air velocity range in m/s Figure 7. Histogram of air speeds measured during field studies in the summer of 2011 for which occupants reported no preferred change of the air velocity. 4 SIMULATIONS A simulation model of the test room including its ventilation equipment was prepared for the TRNSYS program and validated experimentally. It includes modelling the convective heat transfer to the ceiling as a function of the temperature difference ( ). The factor f was doubled for fan operation from a standard value of 2.1 to a value of 4.2. The fan was activated simultaneously with night ventilation. The radiative heat transfer is modelled separately. Figure 8 shows the change of the operative room temperature with time for a selected peri384 7 REFERENCES od. As the radiative heat exchange dominates the convective component, the effect of increased convection is slight. On the other hand it has to be considered that the results are based on single zone modelling with fully mixed air temperature. The experimental results underline the significant differences between the thermal performance in the front, and the back of the room as well as the effect of stratification. So more detailed modelling is required to quantify the effects in detail (CFD). 32 Arens E., Turner S., Zhang H., and Paliaga G. 2009. Moving Air for Comfort. ASHRAE Journal, May 51 (25), 8 – 18 Aynsley R. 2005. Saving energy with indoor air movement. International Journal of Ventilation 4(2):167-75 Candido C., de Dear R., Lamberts R., and Bittencourt L. 2008. Natural Ventilation and Thermal Comfort: Air Movement Acceptability Inside Naturally Ventilated Buildings in Brazilian hot humid zone, In: Proceedings of Conference: Air Conditioning and the Low Carbon Cooling Challenge, Cumberland Lodge, Windsor, UK, 27-29 July 2008. London: Network for Comfort and Energy Use in Buildings, http://nceub.org.uk. DIN EN 15251: Eingangsparameter für das Raumklima zur Auslegung und Bewertung der Energieeffizienz von Gebäuden – Raumluftqualität, Temperatur, Licht und Akustik, Beuth Verlag, Düsseldorf, 2007 DIN EN ISO 7730: Ergonomie der thermischen Umgebung Analytische Bestimmung und Interpretation der thermischen Behaglichkeit durch Berechnung des PMV- und des PPD-Indexes und Kriterien der lokalen thermischen Behaglichkeit, Beuth Verlag, Düsseldorf, 2006 Eiker U., Seeberger P., Fischer H, Werner J., and Lude G.: Evaluierung eines im Passivhausstandard sanierten Bürogebäudes hinsichtlich des energetischen und raumklimatischen Verhaltens, HfT Stuttgart/Ingenieurbüro ebök, Schlussbericht, 2007, download: http://www.enob.info Engelmann P., Musall E., and Voss K. 2010. Sanierung eines Bürogebäudes der 60er Jahre zu einem Niedrigenergiehaus – EnSan REB, Schlussbericht, Universität Wuppertal, download: http://www.enob.info Fairey P., Chandra S., and Kerestecioglu A. 1986. Ventilative Cooling in Southern Residences: A Parametric Analysis, FSEC-PF-108-96, Florida Solar Energy Center, Cape Canaveral, FL. Fanger O. 1970. Thermal comfort. Danish Technical Press, Copenhagen. Rohles F.H., Konz S.A., and Jones B.W. 1983. Ceiling Fan as Extenders of the Summer Comfort Envelope, ASHRAE Transactions, Vol. 89, Pt. 1A., American Society of Heating, Refrigeration and Air Conditioning Engineers, Atlanta, GA., p. 51. Schiavon S. and Melikov A. 2008. Energy saving and improved comfort by increasing air movement, Energy and Buildings Volume 40, Issue 10, pp.1954-1960. Schweiker M., Brasche S., Bischof W., Hawighorst M., Voss K., and Wagner A. 2012. Development and validation of a methodology to challenge the adaptive comfort model. Building and Environment, 49, pp. 336-347 Sekhar S.C. 1995. Higher space temperatures and better thermal comfort a tropical analysis. Energy and Buildings 23:63-70. Voss K., and Künz C. 2012. Klimadaten und Klimawandel – Untersuchungen zum Einfluss auf den Energiebedarf, den Leistungsbedarf und den thermischen Komfort von Gebäuden, Bauphysik, Jg. 34,, Heft 5, S. 189-196. Voss K., and Pfafferott J. 2007. Energieeinsparung contra Behaglichkeit?, Schriftenreihe des Bundesamts für Bauwesen und Raumordnung, Nr. 121, Bonn, download http://www.bbsr.bund.de/nn_23494/BBSR/DE/Veroeffentli chungen/BMVBS/Forschungen/2007/Heft121.html operative temperature [°C] alpha low alpha high 30 28 26 24 5184 5208 5232 hours of the simulation year 5256 Figure 8. Change of the operative temperature with time for a period of 3 days in August due to a varying convective heat transfer coefficient by a factor of two (hourly data). 5 CONCLUSION On one hand, the reported work confirms the assumption that operating ceiling fans during the day in summer is a suitable measure to improve indoor comfort locally and thus increase user acceptance. By contrast, operating them at night to increase the efficiency of night ventilation does not result in a very significant effect, so that the additional electric power demand usually may not be justified. High energy efficiency is an essential factor in selecting an electrically powered ceiling fan. During daytime operation, the users were convinced by the immediately obvious benefit of the functionality, as a) there was no time delay and b) the function was visible and comprehensible. This favours ceiling fans for rooms such as hotel rooms, in which a changing group of occupants is addressed. In offices with several occupants (open plan), fans specific to each desk are of advantage, so that individual preferences can be accommodated. 6 ACKNOWLEDGEMENTS The presented analyses were carried out within the “Passiv Kühl” project, which was supported by the German Federal Ministry of Economics and Technology (BMWi). The authors thank the employees of ZUB in Kassel and ebök in Tübingen for their support during the field measurements. 385 386 Effect of temperature on water vapour transport properties J. Fořt, Z. Pavlík, J. Žumár, M. Pavlíková & R. Černý Czech Technical University in Prague, Faculty of Civil Engineering, Department of Materials Engineering and Chemistry, Prague, Czech Republic ABSTRACT: The temperature effect on water vapour transport properties of calcium silicate is studied, together with the influence of sample thickness. For material characterization purposes, the bulk density, matrix density, and total open porosity are measured at first. Sorption and desorption isotherms are measured using dynamic vapour sorption device in order to characterize the water vapour storage in researched material. The sorption process is analyzed for original material samples as well as for finely ground samples in order to evaluate the effect of inner porous space on water vapour storage. The steady state cup method is used for determination of water vapour transport properties, whereas the measurements are performed at several temperatures and for three different sample thicknesses. The obtained data show an important effect of temperature on water vapour transport and storage in studied material. The results also indicate a substantial influence of sample thickness on the calculated water vapour transmission properties. 1 INTRODUCTION versible thermodynamics and linear theory of mixtures are usually employed. The simplest models applied in the practice reduce the generally ncomponent system to only two components, water vapour and the porous skeleton of the solid phase, and account for only one generalized thermodynamic force which is the gradient of either partial pressure or concentration of water vapour in the porous space (Černý 2010). Then, two following relations can be used for description of water vapour flux jv (kg/m2/s) Water present in building materials and structures represents always a problem for their durability and functional properties. For example, excessive moisture significantly reduces the thermal insulation properties of insulation boards, panels and other lightweight materials. Here, especially condensed water represents a serious problem. The damp buildings and structures suffer also from structural decomposition of materials related to the loss of mechanical resistivity and frost induced damage. Highly humid interior environment of buildings is also accompanied by negative health effects, related especially to mould growth. On the other hand, dry environment is also not very convenient for building occupants because of respiratory problems. Therefore, building engineers and researchers must deal with the problem of water transport and storage within the porous structure of building materials, and investigate the material parameters characterizing this transport. Basically, in description of moisture transport and storage, two phases of water are considered, i.e. water vapour and liquid water. In this paper, we focused on water vapour transport and storage only. Diffusion of water vapour in air is presently a well understood phenomenon. On the other hand, in porous building materials diffusion is more complicated, and while it takes place in air (in the pore space), it is impeded by the reduction of the accessible cross-section, adsorption effects on the pore walls and the tortuosity of the pore paths (Krus 1996). In modelling the water vapour transmission in porous building materials, the methods of linear irre- jv = − D ⋅ gradρ v , (1) jv = −δ ⋅ gradpv , (2) where ρv (kg/m3) is the partial density of water vapour, D (m2/s) the diffusion coefficient of water vapour in the porous material, pv (Pa) the partial pressure of water vapour, δ (s) the water vapour diffusion permeability. Assuming water vapour to be an ideal gas, we can write the equation of state in the form pv = ρ v RT (3) , M where T (K) is the temperature, R ( 8.314 J/K/mol) the universal gas constant, M (18.02 g/mol) is the molar mass of water vapour. Under isothermal conditions we obtain from (1)-(3) the following relation between the water vapour diffusion coefficient and water vapour diffusion permeability D =δ 387 RT . M (4) Besides D and δ, several other coefficients are used, for the sake of better clarity for the building practice. Among them, the water vapour diffusion resistance factor µ (-) defined as phases is smaller than the temperature gradient across the vapour-filled pores. Water vapour moves primarily through the air spaces where the higher local temperature gradient provides a driving force for the diffusion of water vapour. For measurement of water vapour transport properties of porous media, the cup method is usually applied (Roels et al. 2004). Although this method was originally proposed for isothermal measurements only, its slight modification allows to access water vapour transport properties as function of temperature. For example Mukhopadhyaya et al. (2005) applied modified cup method for measurement of temperature dependency of water vapour transmission properties of two types of gypsum fireboard. They introduced simple and versatile technique that allows a user to vary the temperature condition of the cup test without altering the relative humidity. The obtained results demonstrate that there is a steady exponential increase of water vapour transmission rate through both the materials tested with temperature. Since the applicability of cup method for measurement in dependence on temperature was proved, we followed a similar procedure in our experiment. Da (5) D is the most often used material parameter for description of water vapour transport properties of building components as thermal insulation boards, vapour-tight layers etc. Here, Da (m2/s) is water vapour diffusion coefficient in air. µ= 2 EFFECT OF TEMPERATURE ON WATER VAPOUR TRANSPORT The above given equations characterize the water vapour transport under isothermal conditions only. However, the building envelopes are worldwide exposed to severe weather fluctuation, whereas the variation of temperature between + 50°C and -30°C is a realistic possibility. Temperature is one of the main forces for moisture movement and influences sorption characteristics of the both organic and inorganic building materials (Mukhopadhyaya et al. 2005, Pavlík et al. 2013a). Increase of temperature induces greater mobility of water molecules in any form of moisture, and it is widely accepted that water vapour transmission through any material is a function of temperature. Under non-isothermal conditions, such as those prevailing in real environment of buildings, diffusion of water vapour may be enhanced as compared to the isothermal conditions (Gibson 2000), (Wildenschild & Roberts 2001). Two main phenomena are responsible for this enhancement as described by Philip and deVries (Philip & deVries 1957). Normally, diffusive transport of water vapour is obstructed by the presence of liquid islands in the pore throats and diffusion is reduced at higher saturations. However, under a temperature gradient, a vapour pressure gradient develops in the gas phase and causes water to evaporate from one side of the liquid island, and diffuse in the gas phase to a liquid island of lower temperature where it condenses. Water flows through the liquid island as a result of differences in meniscus curvature between the two sides. The evaporation-condensation process repeats itself on the other side of the liquid island and the result is an enhanced diffusive flux through the medium (Philip & deVries 1957). The second enhancement mechanism proposed by Philip and deVries relates to the use of an average temperature gradient in Fick’s law. The thermal conductivity of the solid phase is greater than that of the liquid phase, which in turn is greater than the thermal conductivity of the air phase. Therefore, the mean temperature gradient averaged over all three 3 EXPERIMENTAL 3.1 Studied Material In the experimental study, calcium silicate thermal insulation board produced by Calsitherm, Germany, was tested. It is low-density product, mainly used as capillary active inside insulation, because of its very high capillary absorption coefficient and capillary moisture content. The raw materials are calcium oxide and silica, which react with water to form calcium silicates. Within the high temperature autoclaving, the induced high pressure replaces water in the pores, so that the fine porous structure is formed. Besides its beneficial thermal insulation properties, this material can be used also as fire protection board. 3.2 Basic Physical Properties Among the basic physical properties, matrix density ρmat (kg/m3), bulk density ρb(kg/m3) and total open porosity ψ (-) were measured. The experiments were done on cubic samples of side dimension 50 mm. The particular samples were firstly dried, and their dimensions and mass were precisely measured. In this way, the bulk density of the brick body was accessed. The matrix density was measured by helium pycnometry. On the basis of the knowledge of the matrix density and bulk density, the total open porosity of the brick body was calculated (Pavlíková et al. 2011). 388 3.3 Sorption and Desorption Isotherms Measurement 3.4 Cup Method The sample size for the cup experiments was 100 x 100 mm, whereas the samples thicknesses were 20, 30, and 50 mm. The measurement was carried under isothermal conditions at temperature levels 10, 20, 30, 40, 50°C. It was based on one-dimensional water vapour diffusion and measuring the diffusion water vapour flux through the specimen and measuring partial water vapour pressure in the air under and above specific specimen surface. Water vapour transmission properties of a studied material were found by placing a specimen of the material on the top of a stainless steel cup, whereas the specimen was on the contact with the cup sealed by technical plasticine. The cup contained sorption material, in our case silica gel. The sealed cup was placed in a controlled climate chamber and weighed periodically. The steady state values of mass gain or mass loss were utilized for the determination of the water vapour transfer properties. As one measurement for one temperature level was finished, the samples were removed from the cups, dried and the measurement was repeated for another chosen temperature. Within the measurement, the partial water vapour pressures above silica gel in particular cups and partial water vapour pressure in the climatic chamber were continuously monitored by combined temperature/relative humidity mini-sensors (see Figure 2 for the experimental arrangement). For measurement of sorption and desorption isotherms, dynamic vapour sorption device DVSAdvantage (Surface Measurement Systems Ltd.) was used, whereas the measurements were done at 25°C. Before the measurements, the sample of studied material was dried at first, and maintained in desiccator during cooling. Then, the sample was put into the climatic chamber of the DVS-Advantage instrument and hung on the automatic balances in the special steel tube. The particular samples were exposed to the following partial water vapor pressure profile: 0; 10; 20; 30; 40; 50; 60; 70; 80; 90; and 98% of relative humidity. During the experiment, the DVS-Advantage instrument was running in dm/dt mode (mass variation over time variation) to decide when equilibrium was reached. A fixed dm/dt value of 0.00004% per min was selected for all relative humidity segments (Pavlík et al. 2013b). This criterion permits the DVS software to automatically determine when equilibrium has been reached and complete a relative humidity step. When the rate of change of mass fell below the threshold over a determined period of time, the relative humidity set point proceeded to the next programmed level (Pavlík et al. 2012). Since we investigated also the effect of inner porous space on water vapour storage, the sorption and desorption isotherms measured for original materials samples were compared with data for finely ground samples. This evaluation was done for data measured at 25°C. Figure 1 shows the particle size distribution of studied calcium silicate powder measured on laser diffraction principle using the device Analysette 22 Micro Tec plus. Figure 2. View of the experimental setup. In the climatic chamber, almost constant values of chosen temperatures and 50% of relative humidity were maintained (see Figures 3, 4). Figure 1. Particle size distribution. 389 equation (5). In these calculations, the temperature dependence of water vapour diffusion coefficient in air Da was accounted for according to the equation (7) formulated originally by Schirmer (Schirmer 1938) Da = 2.306 ⋅ 10− 5 p0 p 1.81  T  ⋅  ,  273.15 (7) where p0 (Pa) is standard atmospheric pressure (101 325 Pa = 760 mm of Hg), p the ambient pressure of the measured sample place into the climatic chamber, T (K) absolute temperature. The ambient pressure in climatic chamber was within the performed experiments continuously monitored by highly precise combined relative humidity, temperature and atmospheric pressure sensor produced by Ahlborn. It measures atmospheric pressure in the range of 700 1100 hPa with accuracy of ± 2.5 hPa. Figure 3. Temperature variations within the measurement. 4 RESULTS AND DISCUSSION Basic physical properties of researched calcium silicate board are given in Table 1. Studied material exhibited high total open porosity what is beneficial for its application as thermal insulation layer. Table 1. Basic properties of studied material. Bulk density Matrix density Total open porosity (kg/m3) (kg/m3) (-) 260 2 260 0.88 Figure 4. Relative humidity variation within the measurement at 20°C. Sorption and desorption isotherm measured at 25°C are presented in Figure 5. In the case of relative humidity variation in the climatic chamber, the partial water vapour pressure value corresponding to the steady state mass gain was used in the vapour transmission properties calculations. The water vapour diffusion permeability was accessed from the measured mass gain data according to the equation δ= ∆m ⋅ d , S ⋅ τ ⋅ ∆p p (6) where Δm is amount of water vapour diffused through the sample, d (m) the sample thickness, S (m2) the specimen surface, τ (s) the period of time corresponding to the transport of mass of water vapour, and Δpp (Pa) is the difference between partial water vapour pressure in the air under and above specific specimen surface measured within the particular experiments. Diffusion coefficient of water vapour D was calculated from the determined water vapour diffusion permeability according to equation (4). Water vapour diffusion resistance factor was accessed using Figure 5. Sorption and desorption isotherms. The obtained data showed high storage capacity of calcium silicate for water vapour molecules. On the other hand, the hysteresis of water vapour storage is very low, for lower partial water vapour pressures typically in the range of measuring uncertainty. See Table 2 for calculated hysteresis values. The shape 390 of sorption isotherm corresponds according to BET classification to II type (Brunnauer et al. 1938, Brunnauer 1943). nesses graphed in Figure 7. Here, the obtained data represent average value from the measurement of five samples. We can observe high influence of temperature on water vapour transport that was systematically enhanced with temperature increase. Water vapour diffusion resistance factor is presented in Figure 8. In this case, the highest resistivity to water vapor transmission was observed for lower temperatures what is in agreement with above given results and definition of this material parameter. Concerning the sample thickness within the measurement, the highest values of water vapor transport properties were calculated for samples having thickness of 5 cm. On the other hand, the lowest values were obtained for sample thickness of 2 cm. In case of water vapor diffusion resistance factor, the results were - in accordance with its definition - opposite. Table 2. Hysteresis in water vapour storage. Relative humidity (%) Change in mass (%) 0 0.23 10 0.33 20 0.30 30 0.32 40 0.40 50 0.58 60 0.77 70 0.98 80 1.20 90 1.52 The effect of inner porous space on water vapour storage is presented in Figure 6. Data measured at 25°C are given. Here, the milled samples exhibited higher storage capacity compared to original samples. This finding can be assigned to the assumed higher specific surface of the powder compared to the original material. Figure 8. Water vapour diffusion resistance factor in dependence on temperature. 5 CONCLUSIONS Determination of water vapour transport and storage properties of calcium silicate was presented. The proposed measuring procedure based on slight modification of cup method proved its applicability for measurement of water vapour transport parameters in dependence on temperature and sample thickness. The applied DVS Device gave highly precise results of moisture content changes at specific relative humidities. The obtained data revealed an important effect of temperature on water vapour transport in studied material. A similar performance can be expected also for other porous building materials. With the increasing temperature, the water vapour diffusion coefficient increased (contrary to the decreasing water vapour diffusion resistance factor but this was in accordance with its definition). The presented results can find use not only in building practice, but they can be applied also as input data for computational modelling of coupled moisture and heat transport, where such precise data are usually missing. Figure 6. Sorption and desorption isotherms measured for original and milled samples. Figure 7. Water vapour diffusion coefficient in dependence on temperature. Water vapour diffusion coefficient plotted as function of temperature is for specific sample thick391 6 ACKNOWLEDGEMENT Wildenschild D., and Roberts J. J. 2001. Experimental Tests of Enhancement of Vapor Diffusion in Topopah Spring Tuff. Journal of Porous Media, 4, 1-13. This research has been supported by the Czech Ministry of Education, Youth and Sports, under project No. SGS12/104/OHK1/2T/11. 7 REFERENCES Brunauer S., Emmet P. H., and Teller W. E. 1938. Adsorption of gases in multimolecular layers. Journal of American Chemical Society, 60, 309-319. Brunnauer S. 1943. The adsorption of gases and vapors. Princeton: Univ. Press. Černý R. (ed.) 2010. Complex System of Methods for Directed Design and Assessment of Functional Properties of Building Materials: Assessment and Synthesis of Analytical Data and Construction of the System. Prague: Czech Technical University in Prague. Gibson P. W. 2000. Effect of temperature on water vapor transport trough polymer membrane laminates. Polymer Testing, 19, 673-691. Krus M. 1996. Moisture Transport and Storage Coefficients of Porous Mineral Building Materials. Theoretical Principles and New Test Methods. Fraunhofer IRB Verlag. Mukhopadhyaya P., Kumaran M. K., and Lackey J. 2005. Use of the modified cup method to determine temperature dependency of water vapor transmission properties of building materials. Journal of Testing and Evaluation, 33, 316322. Pavlíková M., Pavlík Z., Keppert M., and Černý R. 2011. Salt transport and storage parameters of renovation plasters and their possible effects on restored buildings' walls. Construction and Building Materials, 25,1205-1212. Pavlík Z., Žumár J., Medveď I., and Černý R. 2012. Water vapor adsorption in porous building materials: experimental measurement and theoretical analysis. Transport in Porous Media, 91, 939-954. Pavlík Z., Fořt J., Žumár J., Pavlíková M., and Černý R. 2013a. Secondary Effects on Water Vapor Transport Properties Measured by Cup Method. World Academy of Science, Engineering and Technology, 73, 1020-1025. Pavlík Z., Medved´ I., Žumár J., Pavlíková M., and Černý R. 2013b. Theoretical and Experimental Analysis of Adsorption Isotherms of Building Materials. World Academy of Science, Engineering and Technology, 73, 1026-1031. Philip J. R., and de Vries D. A. 1957. Moisture movement in porous materials under temperature gradients. Transactions of the American Geophysical Union, 38, 222-232. Roels S., Carmeliet J., Hens H., Adan O., Brocken H., Cerny R., Pavlik Z., Hall C., Kumaran K., Pel L., and Plagge R. 2004. Interlaboratory Comparison of Hygric Properties of Porous Building Materials. Journal of Thermal Envelope & Building Science, 27, 307-325. Schirmer R. 1938. Die Diffusionszahl von Wasserdampf-LuftGemischen und die Verdampfungsgeschwindigkeit. Beiheft VDIZeitschrift, Verfahrenstechnik, 6, 170-177. 392 A new approach to measure liquid transport in capillary active interior insulation A. Binder, H. Künzel & D. Zirkelbach Fraunhofer Institute for Building Physics, Department of Hygrothermics, Holzkirchen, Germany ABSTRACT: The application of an interior insulation is not seen without reservations even by specialists. While it is a fact that interior insulation confronts installers as well as planners with a challenging moisture control task, it is, nonetheless, a task that can be resolved. This applies in particular to insulation systems that feature liquid transport in the hygroscopic region, the so-called “capillary-active” systems. Such systems utilize the materials’ ability to transport liquid moisture under non-isothermal conditions against the water vapor diffusion flux, in this way naturally limiting the moisture content inside a construction. The serviceability of the system and prevention of damage is ideally assessed beforehand via hygrothermal simulation. For this, it is vital to know the materials’ specific liquid transport characteristics. The description of liquid transport inside building materials is challenging because of many influencing -and partially counteracting- parameters. Conventional test methods are dominated by isothermal conditions and the transport in larger capillaries and therefore are not fully appropriate to quantify the liquid transport in interior insulation materials. The Fraunhofer Institute for Building Physics has lately developed a new laboratory method, the “Capillary Condensation Redistribution” (CCR-) test. In imitation of the specific boundary conditions of interior insulation systems, this test works with non-isothermal conditions and opposing liquid moisture and water vapor flow gradients, with humidification only by vapor diffusion. By closely monitoring the materials moisture content and distribution, the specific liquid transport properties in the hygroscopic region are determined via hygrothermal simulation. A newly installed outdoor-test at the facilities in Holzkirchen validates the promising results of the new laboratory test. A real-case scenario clarifies the significance of an exact and practical determination of liquid transport coefficients for interior insulation purposes even further. 1 INTRODUCTION active interior insulation systems in order to avoid the problems with conventional systems. Such systems work without additional vapor retarding layers, utilizing the materials ability to transport liquid moisture inside there pore system. In this way, they contribute to keeping moisture contents inside the construction below critical limits. To ensure for a proper function of the insulation and prohibit damage, these systems must be carefully planned beforehand, ideally by closely analyzing their hygric performance via transient hygrothermal building simulations. A framework requirement for respective software tools are detailed and reliable hygrothermal material parameters. Of particular importance here is the liquid transport ability, a material characteristic that is highly complex and, especially with respect to interior insulation, hard to measure. Experience shows that conventional testing methods are not fully appropriate here and sometimes cannot be used at all. Thus, the Fraunhofer Institute for Building Physics has developed a new laboratory method, considering the specific needs and boundary conditions of interior insulation materials. As laboratory measurements have been demonstrating promising results for some time now, an outdoor-test has recently been installed at the test-site in Holzkir- Rising fossil fuel prices and new building codes call for thermal retrofits of existing buildings. Not only preservation regulations, but also reasons like high building density or non-uniform ownership often demand the application of interior insulation. This strongly affects the thermal and hygric performance of the original wall structure. The accompanying circumstances -such as temperature reduction and reduced drying potential- consequently lead to rising moisture contents inside the construction and pose the challenging task to planners, manufacturers and builders to keep hygrothermal conditions below possibly critical conditions. Conventional systems work with additional vapor retarding layers, in this way constraining excessive water vapor diffusion into the wall system. Suchlike systems tend to react very sensible even to minor leakages, caused by flaws during installation or use. Moisture entering the system through these minor gaps is spreading inside the insulation layer with usually only minor potential to dry out. Not least due to this, interior insulation is seen with reservations by many still. Besides vapor retarders with variable diffusion resistance, there is a tendency to so-called capillary 393 chen, validating the new “Capillary Condensation Redistribution” (CCR-) test. A real-case scenario clarifies the significance of an exact and practical determination of liquid transport parameters for building simulation even further. 1.1 Capillary activity and moisture transport The commonly used term “capillary activity” denotes the characteristic of porous, hygroscopic materials to conduct liquid moisture inside their pore system as a result of surface tension and cohesion of solid and liquid matter. This transport effect is independent of orientation and works contrary to gravitation, too. In order for the process to initiate, sufficient amounts of (liquid) moisture need to be available inside the pore system. This moisture can either be already present (built-in moisture), or enter the system either by adsorption or water vapor diffusion. The velocity of liquid transport is hereby depending both on the manner of wetting, the distribution of water inside the pores and on the structure of the pore system. Large pores feature a high suction velocity, but low suction pressure, while with decreasing pore diameter, suction speed decreases, with increasing suction pressure. Consequently, liquid transport is dominated by large pores when high amounts of (liquid) moisture occur (driving rain etc.). After disruption of liquid supply, however, moisture is redistributed due to the higher suction pressure inside small pores. In pores not filled completely, another transport mechanism exists, for convenience commonly added to capillary activity, although technically not caused by capillary suction. This effect, called surface diffusion, is liquid transport inside multi-molecular moisture layers alongside a gradient of relative humidity. To account for the different transport velocities, the simulation software WUFI (Künzel 1994) is used. The software complies with respective standards (EN 15025 2007, ASHRAE 2009 & WTA 2002) and calculates liquid transport using two different sets of material parameters. Depending on boundary conditions (rain/no rain), the calculation is executed based on transport coefficients for water absorption (DWS) or redistribution (DWW). For a better understanding of the transport effects inside exterior walls, two basic boundary conditions have to be distinguished, as they evoke different moisture transport mechanisms. Figure 1. Moisture transport phenomena in pores of hydrophilic media under the influence of different boundary conditions. Isothermal conditions (figure 1, above), with equal temperatures on both sides of the material/wall, usually apply in summer. The gradients of partial pressure and relative humidity both take the same course, conducting vapor diffusion as well as liquid transport from moist to dryer zones. Under non-isothermal conditions (figure 1, below), usually applying in winter time, the exterior climate features lower temperatures and higher relative humidity levels than the interior air. As a result, vapor diffusion of the interior air is directed towards the exterior surface, leading to rising moisture inside the wall/insulation, as moisture is adsorbed at the pore surfaces. Thereby, the moisture level rises with decreasing temperature. The insetting liquid transport towards the interior surface can either be present in the form of surface diffusion or in the form of capillary suction in (usually smaller) pores with completely filled diameter. These effects are dominated mainly by smaller pores; the transport velocity is rather slow, thus. Nonetheless, further moisture accumulation is decelerated, eventually reaching dynamic equilibrium in time. 1.2 Conventional Testing and Characteristics For the determination of liquid transport, several standardized or approved procedures are available. A commonly used and very quick laboratory test method is the water uptake test (A-value). By monitoring the moisture gain of material samples wetted on one side, liquid transport coefficients primarily for water absorption, but also for redistribution, can be determined. Given that, as described, the suction process is dominated by the rather fast liquid transport inside large pores, liquid transport for re394 distribution tends to be overestimated for many materials. A test more appropriate for determining the redistribution coefficients is the drying test (Holm & Krus 1998, Scheffler et al. 2009). In this test, wet material samples are dried under steady-state boundary conditions, documenting the weight loss over time. Another method (Krus 1995) involves the onesided wetting of a material sample; after a certain moisture content is reached, the water supply is cut off, leaving the material to redistribute the absorbed moisture inside its pore structure. By nuclear magnetic resonance spectroscopy (NMR), the distribution of moisture across the sample is measured regularly. In both methods, the redistribution coefficients are determined by recomputing the measuring results via numerical simulation. The tests mentioned above have been developed for materials exposed to natural weathering. They are valid and adequate for their intended purpose. Subjected to wetting by driving rain, the liquid transport behaviour of the material is dominated by large amounts of water and the rather fast transport velocity in large pores. However, there is growing evidence that the existing methods may not be sufficiently accurate for characterizing the moisture behaviour of interior insulation materials. All tests involve wetting the material with liquid water. Because of the dominating effects inside larger pores, properties may generally not be transferred to lower moisture levels and different modes of wetting. Additionally, interior insulation materials are not designed for high amounts of moisture and may not tolerate them, entailing structural changes like agglutination or dilatation of fibrous materials. Furthermore, all tests are executed in isothermal conditions. As liquid, as well as vapour transport, have the same direction, a distinction between the two cannot be made. No information can be gained therefore about the quantity of each transport mode by itself. As described before, the relevant boundary conditions that apply to interior insulation materials are non-isothermal, however, this condition cannot be acknowledged by the tests so far. with non-isothermal conditions and opposing liquid moisture and water vapor flows, with humidification only by vapor diffusion. In the first phase of testing, the testing device was aligned horizontally, providing for horizontal moisture flows similar to real ones. Experience showed, however, that this may distort the moisture behavior (Binder et al. 2010), as a runoff of condensation moisture may occur. In continuation of the original idea, the device was modified to vertical orientation (figure 2). As liquid transport works independent of orientation (1.1), unaltered liquid transport characteristics can be expected. Under laboratory conditions, a dew-point undercut is applied to one side of a laterally sealed material sample. This produces a gradient of temperature and partial pressure, and, consequently, vapor diffusion into the material. The adsorbed moisture is condensing at the sealed back side of the material, where it causes an increase in relative humidity. Hence, the moisture content inside the material sample is rising. Due to the increasing gradient of relative humidity, a liquid transport back to the front surface of the sample sets in. Eventually, the opposing moisture fluxes will reach a dynamic equilibrium. To reveal the hygrothermal behavior of the material samples during testing, two modes of measuring are taken. Through periodic gravimetrical measurements, the moisture gain is analyzed and documented for the test period. The moisture distribution in the sample’s cross-section is measured periodically by using nuclear magnetic resonance spectroscopy. By recomputing both the measured moisture distribution as well as the moisture gain, via numerical simulation, highly detailed parameters for the moisture transport in the hygroscopic region are determined. The tests are executed in a climate chamber with steady-state conditions of 23 °C and 65 % RH. The dew-point temperature at these conditions is 16 °C. The dried and sealed material samples are applied to the carrier plate with heat conductive paste. All boundary conditions are measured and documented for the testing period. Regularly, the samples are removed from the test setup in order to measure moisture content; the moisture distribution is measured at the beginning and end of the test period, as well as on certain times during test period. A pecking motor is moving the samples one by one through the inductor with 12 mm per step, generating 15 measure points along each step. The resulting moisture distribution thus is very detailed. 2 CAPILLARY CONDENSATION REDISTRIBUTION TEST 2.1 Test concept and procedure In the desire to circumvent the difficulties and limitations depicted above, the Fraunhofer-Institute for Building Physics has developed a new laboratory method, considering the specific conditions of interior insulation (Binder et al. 2010, Zirkelbach & Binder 2011). In imitation of real-life boundary conditions of interior insulations, the CCR-test works 395 Figure 2. Relative humidity at the rear side of the mineral fibre interior insulation. 2.2 Test results The test results to be expected for capillary active materials are demonstrated exemplarily for a Calcium silicate (CS) material. Figure 3 states the gravimetrically measured moisture gain of three CS samples during the test period of 20 weeks. Starting from dry state, the samples gain weight fast at the beginning of the experiment, as vapor diffusion is taking place still to its full extent. With time advancing, the moisture gain is decelerated by insetting liquid transport. After 8 weeks, a weight maximum is reached at app. 39 kg/m³. Hereafter, weight decreases a little, with further fluctuations around a mean value of app. 35 kg/m³. Measurements show a good correlation between the tested samples at all times. Figure 4 shows the distribution of moisture across the samples at dynamic equilibrium. As expected, the highest increase of moisture can be detected at the sealed, cool back side of the sample. Towards the open-faced front surface, a comparatively even decrease of moisture is exhibited. Figure 4. Moisture distribution of calcium silicate samples during CCRtest. 2.3 Determination of transport coefficients The transport coefficients are hence reproduced by numerical simulation with WUFI® (Künzel 1994). For this purpose, the test results are processed for the implementation into WUFI®, exporting the moisture gain characteristics for the whole test period and the moisture distribution at dynamic equilibrium (or at the end of testing). The liquid transport coefficients can be generated in a very simple way based only on the A-Value (this method is also mentioned in Holm & Krus 1998 and then refined by a drying test). The simple values serve as initial approach. Step-by-step, the values are adapted accordingly to get a good correlation of calculated and measured moisture gain and distribution. Figures 3 and 4 show the resulting moisture gain and distribution calculated with liquid transport coefficients (DWW) based on CCR test or on the simple A-Value method and without liquid transport respectively, in comparison to measurement results. Figure 3. Moisture gain of calcium silicate samples during CCR test. 396 Calculation with DWW based on the simple A-value method shows large discrepancies to measurements, with generally low moisture gain and dynamic equilibrium being reached after one week already (for rain water absorption tests the method normally shows quite acceptable results). In comparison, calculation without DWW represents especially the rather fast moisture gain in the first weeks of testing better – however, its further linear increase does not reflect the measured materials behavior at all, neither does the calculated moisture distribution. Acceptable agreement with measurement results is reached with DWW according to CCR test. Here, moisture content at dynamic equilibrium correlates quite well with measured values; moisture distribution, too, reflects the measurements well. There is a certain discrepancy between calculated and measured moisture gain in the first weeks of testing. This initial behavior could be simulated with slightly lower liquid transport properties – the measured dynamic equilibrium of approx. 40 kg/m³ would then be exceeded however. Furthermore, the congruence of measured and calculated moisture distribution would decrease. Therefore, we focused on covering the long term behavior as well as the moisture distribution insde the material in an accurate way while accepting the discrepancies of the initial increase. The liquid transport coefficients determined via CCR test are stated in table 1 and figure 5. Figure 5. Liquid transport velocities for redistribution (DWW) acc. to CCR test. 3 OUTDOOR-TESTING FOR VALIDATION 3.1 Test concept and procedure To validate concept and procedures of the CCR-test, an outdoor experiment has been installed at one of the buildings (interior climate app. 25 °C, 50 % RH.) at the test-facilities in Holzkirchen. Therefore, a test façade facing west is used. This façade is partitioned into fields with dimensions of 500 x 500 mm, allowing the investigation of multiple wall compositions at the same time. The interior insulation assembly is installed into wooden framework with a thickness of app. 400 mm. For the experiment, several insulation materials, both hydrophilic and hydrophobic, are tested, out of which three are presented here. The basic composition for all is the following, listed from exteriorly: 20 mm lime-cement render, 240 mm solid brick, app. 20 mm plaster (to create an even surface for the insulation layer). As a representative of classical capillary-active materials, calcium silicate insulation (λ = 0,07 W/(mK), μ = 6,0), as shown above, is chosen. On top of the 90 mm thick insulation is a 15 mm thick layer of plaster. From the group of hydrophilic fibre materials, a cellulose fibre material (λ = 0,037 W/(mK), μ = 1,3) with 55 mm thickness, covered with 12,5 mm gypsum board, is installed. To serve as a reference case, a hydrophobic mineral fibre material (λ = 0,039 W/(mK), μ = 1,0) with 65 mm thickness, also with 12,5 mm gypsum board towards the interior, is installed. The respective liquid transport coefficients for redistribution are stated in table 1 and figure 5. Between the insulation layer and the plastered brick wall, an aluminium foil (sd-value 1500 m) is installed in order to avoid interference between water or vapour absorption of the original wall structure and the moisture behaviour of the insulation materials. At the interface between insulation and aluminium foil, both temperature and relative humidity are recorded continuously. Additionally, thin wooden plates, equipped with wood-moisture probes, are embedded. All exterior climate parameters are measured by the IBP´s weather station; for interior climate, temperature and relative humidity are measured. Table 1. Transport coefficients for redistribution (CCR- Test). Transport Relative Moisture velocity Type of Insulation Humidity content [m²/s] [%] [kg/m³] 80,00 3,0 3,40 x E-10 Calcium silicate 99,95 116,0 3,40 x E-10 100 899,0 2,90 x E-7 80,00 7,8 7,00 x E-10 Cellulose fibre 99,50 71,1 7,45 x E-10 100 430,0 2,30 x E-8 80,00 0,5 3,0 x E-11 Mineral fibre 100 220,0 3,0 x E-11 3.2 Test and calculation results The experiment started in November 2011. As the initial moisture conditions of the materials where not equal, a period of adjustment marks the measurements of the first winter period, featuring relative humidity and wood moisture courses only hardly comparable. Exhibited in the following therefore are 397 the measuring courses after a comparable state has been reached at the beginning of May 2012. For all assemblies, hygrothermal simulations where executed, implementing the measured temperature at the rear side of the insulation layer and the measured interior conditions as well as the liquid transport coefficients acc. to CCR test and other material parameters. Figure 6 states the measured temperature (hourly values) at the rear side of calcium silicate, for the period 01.05.2012 – 31.01.2013. Starting at mean temperatures of around 15 °C, temperatures fluctuate around 20 °C during summer month, with variations from app. 15 – 28 °C. At the end of August, temperatures decrease to app. 12 °C and reach minimum values of app. 7 °C in December. Temperatures of the other insulation assemblies were comparable. Depicted in figure 7 is the level of relative humidity at the rear side of calcium silicate. Measurement results show values ranging from app. 65 – 85 % RH during summer. With decreasing temperature, RH levels rise as expected, reaching 100 % RH at the beginning of November. The calculated RH level is higher than the measurements in May, following a somewhat slower calculated drying process. After app. 10 days, the values have aligned, however, showing qualitatively good concordance with measurements. In general, the calculation results fall below measurements up to app. 5 %, reaching only app. 98-99 % RH in December and hitting 100 % RH one month later in November. The relative humidity (rear side) of cellulose fibre is stated in figure 8. Compared to calcium silicate, RH levels are somewhat lower, only rarely excessing 80 % RH in summer. Fluctuations in this case are more noticeable than for calcium silicate, due to the somewhat higher liquid transport at low moisture levels and the lower diffusion resistance. In winter time, the increase is relatively slow, hitting 100 % RH only at the end of January. Here, the calculated level of relative humidity shows even better concordance with measurements, with slight transgressions of app. 1 %. The measurement results of the reference material mineral fibre are illustrated in figure 9. Explicit here is especially the wide range of fluctuations, due to the materials low diffusion resistance and moisture storage ability. Overall, RH here rises fastest, reaching 100 % RH during the second half of October. Here, too, a good agreement between calculation and measurement results is reached, showing only minor deviations. RH sometimes is hard to measure, especially in the higher moisture regions. Thus, wood moisture has been measured to validate RH measurements. The respective courses, shown in Figure 10, confirm the results stated. Here, too, Calcium silicate and mineral fibre measurements largely equal, with mineral fibre showing slightly higher fluctuations; cellulose fibre results depicting improved performance with lower moisture levels for the complete measurement period and especially in winter. Figure 6. Relative humidity at the rear side of the calcium silicate interior insulation layer in the outdoor validation test. Figure 8. Relative humidity at the rear side of the cellulose fibre interior insulation layer in the outdoor validation test. Figure 7. Relative humidity at the rear side of the calcium silicate interior insulation layer in the outdoor validation test. Figure 9. Relative humidity at the rear side of the mineral fibre interior insulation layer in the outdoor validation test. 398 Following is the RH development of the cellulose fiber, displaying maxima of around 96 % RH. Calcium silicate features lowest RH levels, staying beneath levels of 94 % RH even in winter. In summer (quarters II and III), all constructions are drying out to a certain extent, with only minor differences between minimum values to be detected. RH levels decrease to minima of app. 80 % RH. Figure 10. Wood moisture at the rear side of the interior insulation layers in the outdoor validation test. 4 SIGNIFICANCE IN PRACTICE To investigate the diverse results and their relevance for actual building situations, the hygrothermal behavior of a wall facing west, insulated with the materials shown above, is analyzed by numerical simulation. The simulated solid brick wall (dimension 30.0 cm, thermal conductivity λ = 0.51 W/(mK)) with external mineral-based render (A-value 0,1 kg/(m²√h) has a thermal resistance of 0.55 m²K/W, its U-value amounts to 1.36 W/m²K. The exterior surface is simulated without the influence of driving rain. At the interior side of the wall, insulation layers of different diameters are installed, each increasing the thermal resistance of the wall to the German energy code (ENEV 2009) level of 0,35 m²W/K. The interior surface is simulated with an additional sdvalue of 0.2 m, representing a layer of conventional paint or wall paper. For the exterior, the exterior climate conditions of Holzkirchen (IBP, moisture reference year) are applied. Indoor climate conditions, normal moisture load, according to WTA 6-201 (sinusoidal variation of indoor temperature and humidity between 20 °C, 40 % RH in winter and 22°C, 60% RH in summer), are taken as a basis for the interior. All simulations are executed until reaching dynamic equilibrium. Figure 11 displays the development of RH (mean monthly average) at the rear side of the insulation layer for one year at dynamic equilibrium, starting in October. As all simulations eventually reach dynamic equilibrium, no general moisture accumulation can be detected. Comparatively high levels of RH at the boundary layer of insulation and former surface are to be detected during heating period (quarters VI and I) nonetheless, however, as simulation results display. Not surprisingly so, as, indeed, a certain level of moisture inside the pore system has to be present in order for any noticeable liquid transport to get triggered at all. In contrast to the findings so far, however, highest RH is displayed by the mineral fiber insulation, reaching app. 97 % RH. in winter. Figure 11. Relative humidity (mean monthly average) at the rear side of different interior insulation materials in the outdoor validation test. 5 CONCLUSION AND PROSPECT In context with the swiftly developing market of capillary active interior insulation, liquid transport ability is important to secure an acceptable performance of the respective constructions. Experience shows, however, that the conventional tests, originally developed for materials under the influence of natural weathering, are not entirely appropriate in this respect. With the development of the Capillary Condensation Redistribution test, the Fraunhofer IBP tries to fill in that gap, providing a test especially designed for the specific questions and boundary conditions of interior insulation. Featuring nonisothermal boundary conditions, the CCR test manages without the manual addition of liquid moisture, in this way especially appropriate to measure liquid transport in moisture-sensitive materials like cellulose fibres. The transport coefficients are derived by numerically reproducing the development of moisture content and distribution. Due to the complex nature of the real pore structure and the inevitable simplifications necessary for numerical modelling, a perfect representation of calculation and measuring results is hardly possible. In favour of a good concordance of moisture content and distribution at dynamic equilibrium, slight discrepancies of the initial behaviour are accepted, hence. In the whole, with the new 399 CCR method, the materials specific liquid transport at lower levels of RH can be identified with significantly higher accurateness compared to conventional test methods. The promising results of the laboratory test are confirmed by the out-door test. The good concordance of measurement and calculation results shows the accuracy of the respective liquid transport coefficients derived by the laboratory test. Due to the impermeable material (aluminum foil) in-between brick wall and insulation, comparatively high amounts of moisture where expected. The relation of RH results of the single materials are surprising, though, in view of the fact that calcium silicate is generally thought to be one of the most capillaryactive insulation materials, an assumption that is not corresponding to test results. Here, not calcium silicate features lowest moisture levels, but the hydrophilic cellulose fiber. In particular the graphic representation shows that up to comparatively high levels of RH, transport velocity for redistribution is highest for cellulose fibre; only at highest RH levels, it is outrun by the liquid transport capability of calcium silicate. Findings in practical application of the liquid transport coefficients derived by numerical simulation indicate, however, best performance for calcium silicate, staying below 95 % RH even in winter time. This effect may be due to the relatively low vapour diffusion resistance of the cellulose material compared to calcium silicate. The definition of tangible design limits is very complex and still subject to discussions. The WTA (WTA 2009 & 2012 (in progress)) recommends not to exceed 95 % RH at the rear side of the insulation layer, a value that features a certain safety margin. Actual critical values are subject to many influencing parameters such as temperature, RH, possible gaps/leakages in construction and bio-degradability or frost resistance of the materials. The nature of liquid transport requires a certain amount of liquid moisture inside the pore system. Actual performance as well as design limits will consequently always be at close quarters. Regardless of design limits, thus, an evaluation of the hygrothermal performance of a capillary-active insulation system as exact as possible is indispensable. As even slight differences in the measured liquid transport can mean permissibility or failure of a system, an accurate measurement method is necessary, not in the least for quality management. The CCR-test promises to deliver very reliable and practical parameters. For interior insulation purposes, the respective coefficients also cover the materials drying process very accurately. For materials in use not only interiorly, but also exteriorly, the handling of both the slower liquid transport in lower moisture regions, and liquid transport in higher moisture regions (e.g. drying after a rain shower) is still under discussion. Here, a combination of the parameters of drying test and CCR-test may be required, containing an exact liquid transport behavior at low moisture regions and non-isothermal conditions as well as an exact and more rapid transport under the influence of high moisture, e.g. natural weathering. The practical consolidation of the respective measurement results and transport coefficients will be subject to future investigations. 6 REFERENCES Binder A., Zirkelbach D., and Künzel H. 2010. Test method to quantify the wicking properties of insulation materials designed to prevent interstitial condensation. Proceedings of the Buildings XI International Conference: 119-120. Clearwater Beach. ASHRAE ANSI Standard 160 2009. Criteria for moisturecontrol design analysis in buildings. DIN EN ISO 13788 2001-11. Wärme- und feuchtetechnisches Verhalten von Bauteilen und Bauelementen – Oberflächentemperatur zur Vermeidung von kritischer Oberflächenfeuchte und Tauwasserbildung im Bauteilinneren – Berechnungsverfahren. Berlin: Beuth. EnEV 2009. Energieeinsparverordnung. Verordnung über energie-sparenden Wärmeschutz und energiesparende Anlagentechnik bei Gebäuden. Künzel H. M. 1994 (1995). Simultaneous heat and moisture transport in building components. One- and twodimensional calculation using simple parameters. Dissertation Lehrstuhl Bauphysik, Fakultät für Bauingenieur- und Vermessungswesen der Universität Stuttgart. Krus M. 1995 (1996). Moisture transport and storage coefficients of porous mineral building materials – Theoretical principles and new test methods. Dissertation Lehrstuhl Bauphysik, Fakultät für Bauingenieur- und Vermessungswesen der Universität Stuttgart. Holm A., and Krus M. 1998. Bestimmung des Transportkoeffizienten für die Weiterverteilung aus einfachen Trocknungsversuchen und rechnerischer Anpassung. Bauinstandsetzen 4, No. 1, 33-52. Scheffler G., and Plagge R. 2009. Ein Trocknungskoeffizient für Baustoffe. Bauphysik, No. 31, 125-138. WTA-Merkblatt 6-2-01/D 2002. Simulation wärme- und feuchtetechnischer Prozesse. WTA-Merkblatt 6-4 2009. Innendämmung nach WTA I – Planungsleitfaden. WTA-Merkblatt 6-5, 2012 (in progress), Nachweis von Innendämmsystemen mittels numerischer Berechnungsverfahren. Zirkelbach D., and Binder A. 2011, Mit neuen Kennwerten genauer rechnen. In Bauen im Bestand, No. 34, 42-47. 400 Detailed heat, air and moisture transport modelling in cavity walls M. Van Belleghem & M. De Paepe Ghent University, Department of Flow, Heat and Combustion Mechanics, Ghent, Belgium M. Steeman University College Ghent, Faculty of Applied Engineering Sciences, Department of Construction, Belgium L. De Backer & A. Janssens Ghent University, Department of Architecture and Urban Planning, Ghent, Belgium ABSTRACT: Cavity walls are a widely used external wall type in North-western Europe. In this paper a cavity wall configuration was analysed with a heat and moisture transport model for porous materials integrated into a computational fluid dynamics solver. The studied wall is a wood frame wall with a brick veneer as outside leaf. Four cases were analysed in detail: drying of the outside or inside cavity leaf when initially saturated, both for summer and winter conditions. The model was compared with a widely used simulation tool for building envelope analysis (WUFI®). This simulation tool uses a simplified modelling approach for the convection in the cavity. The study showed that these simplifications are not always justified. The simplified model overestimates the drying and moistening rates of the cavity wall compared to the detailed model. Both models showed that the drying of the outer leaf is mainly determined by the outside conditions and the outside leaf dries out mainly to the outside and not to the cavity. The cavity ventilation in this case is of less importance. For the inside leaf this cavity ventilation is of mayor importance for drying. The largest discrepancies between both models were found for this inner cavity leaf. 1 INTRODUCTION and thermal performance for all. Furthermore, climate boundary conditions will determine to a great extent the performance of a cavity wall. A hygrothermal model could help researchers to gain a better understanding of the behaviour of cavity walls under different conditions and with varying configurations. The building envelope is constantly exposed to harsh environmental conditions. To prevent damage to the building and its envelope a sophisticated design is often needed. A specific class of multilayer wall systems is the cavity wall. The main durability problems in cavity walls are frost damage, rain penetration and mould development. Hens et al. (2007a) give a good overview of moisture related problems in cavity walls. Cavity walls have a good moisture tolerance especially against the largest moisture source: wind-driven rain. Besides acting as a capillary break and preventing moisture from outside to reach the inner wall, the ventilated cavity can also help to remove moisture from the inner cavity leaf. Salonvaara et al. (2007) conducted a literature review and found that not all authors agree on the effect of cavity ventilation. Hens et al. (2007a) state that there is no real benefit of cavity ventilation. Without special measures ventilation rates in the cavity will be low and drying of, for example, a brick veneer outside wall will mostly occur at the outside and less at the cavity side. Other studies indicate that cavity ventilation can increase drying (Straube 1998), if ventilation rates are high enough. Some studies even report negative effects of cavity ventilation. For instance when the absolute humidity of the air flowing through the cavity is high, the moisture in the air results in a hygroscopic moisture loading. This analysis clearly shows that the findings related to the benefits of cavity walls are contradictory. A wide range of building envelope types exists and an air cavity will not provide beneficial moisture A cavity wall combines a lot of transport mechanisms and sources for heat and moisture, which explains the difficulty in modelling such a configuration. Figures 1 and 2 illustrate some of the main transport mechanisms for heat and moisture in ventilated cavity walls. Figure 1 shows the heat fluxes, sources and sinks present in a cavity wall. For heat, three transport mechanisms can be identified: radiation, convection and conduction. At the outside, heat transport by radiation is very important. During sunny days solar radiation accounts for a large part of the heat gains of the wall. Solar radiation is referred to as shortwave radiation. Next, heat exchange by longwave radiation is possible. This is radiation emitted by the wall or received from the surroundings. The wall will also exchange radiant heat with the sky. During clear winter nights the temperature of the sky can be some 21°C lower than the environmental temperature (Hens 2007b). This can result in significant longwave radiant heat losses. Heat transfer by longwave radiation occurs in the cavity when the cavity leafs have a different temperature. If convection in the cavity is low, the long wave radiation will be the most important heat transfer mechanism between the cavity leafs. Also at the indoor environment, longwave radiation from sur401 rounding walls and objects can be of importance for the heat balance of the wall. Air flow along the wall will result in convective heat transfer. Transport through the (porous) solid cavity leafs is mainly by conduction. Conductive heat transport in the air is also present but will be small compared to convective transport. Figure 2 illustrates the main moisture fluxes and sources in a cavity wall. Moisture sources depicted in Figure 2 are wind-driven rain, rising damp and outdoor and indoor vapour. This vapour is transported to the wall or in the cavity by convection and diffusion and can be absorbed in the porous wall. Moisture in the building envelope can result in mould growth, deterioration of building materials, wood rot... In literature some simplified models for cavity ventilation can be found (e.g. Karagiozis et al. 2009, Ge et al. 2007, Sanjuan et al. 2011). In this paper it is investigated what the impact is of some of these simplifications on the heat, air and moisture transport in a cavity wall. Therefore a new model is developed. This model combines a detailed computational fluid dynamics model (CFD) for the air transport in the cavity with a Heat, Air and Moisture model (HAM) for the transport in porous materials. The new model is compared with a simplified commercially available model. 2 SIMPLIFIED VENTILATED CAVITY WALL MODEL To model the impact of cavity ventilation on the building envelope performance, Straube & Finch (2009) listed some possible modelling techniques. They used a commercially available model (WUFI® Zirkelbach et al. 2007) to calculate the heat, air and moisture transport in a cavity wall. The simplest approach according to them is ignoring ventilation. The cavity is modelled as still air. The thermal conductivity of the air layer is adapted to incorporate radiation, conduction and natural convection effects in a non-ventilated cavity. This results in a thermal resistance of the cavity as function of the inclination and cavity width. The vapour resistance factor of the air layer was adapted to include the effect of vapour diffusion and convection. Staube & Finch however found that this approach yields inaccurate results and state that the ventilation effect should be included in the modelling. A second approach reported by Straube & Finch (2009) is adjusting the vapour permeance of the exterior cladding. The user adapts the vapour permeance depending on the estimated ventilation rates. In some cases, the external cladding can be removed from the model. This is valid if the conditions in the cavity are the same as those of the outside. However the shielding effect of the external cladding for rain and solar radiation should still be included. Driving rain and solar radiation have a significant impact on the moisture transport in the cavity wall and these models tend to underestimate the moisture loading. This modelling approach can be improved by using measured cavity conditions as outside condition. Nevertheless the aforementioned modelling techniques tend to yield inaccurate results. Therefore Karagiozis et al. (2009) developed a simplified model for cavity wall ventilation and implemented it in WUFI®. The simplified model was able to capture the bulk performance of a cavity wall with reasonable accuracy and gave a rather good agreement with field data. To account for ventilation, heat and moisture sources and sinks were added to the air layer. The moisture and heat added to or extracted from the cavity is modelled as a well-mixed process. The heat source/sink is determined as the amount of enthalpy entering the cavity minus the amount leaving the cavity due to ventilation. The mass source/sink is the mass entering minus the mass leaving the cavity. This can be expressed by the following equations: Figure 1. Illustration of heat fluxes occurring in and around a cavity wall. Figure 2. Illustration of possible moisture fluxes occurring in and around a cavity wall. (1) (2) 402 where Sh and Sm are the heat and mass sources/sinks respectively, Qair is the volumetric air flow rate per volume of cavity [m³/sm³], hext and ρext are the enthalpy [J/kg] and air density [kg/m³] at the cavity entrance and hcavity and ρcavity are the enthalpy and density at the cavity outlet respectively. interesting to develop a model for heat and mass transport in a cavity, with a more detailed modelling of convection in the cavity. In the next section this model will be discussed before studying in more detail the impact of some of the simplifications used in WUFI®. The modelling techniques proposed by Karagiozis et al. (2009) strongly simplify the actual transport mechanisms in the cavity. In reality air will enter the cavity through one of the cavity ventilation openings. The driving forces for the ventilation are pressure difference due to wind pressure on the building façade and pressure difference due to buoyancy. These pressure differences can fluctuate strongly in time. Also changes in outside conditions over time such as outside temperature and radiation will affect the buoyancy forces in the cavity and thus alter the flow field in the cavity. At the same time heat and mass transfer from the cavity leafs to the cavity is determined by the flow field in the cavity. There is thus a strong coupling between the external conditions and the flow field in the cavity on the one side and between the flow field and the heat and mass transfer to the cavity on the other side. Karagiozis et al. neglect this coupling and state that assuming an averaged ventilation rate in the cavity often suffices. The assumption of constant flow conditions in the cavity is to some extent justifiable. Air velocity in the cavity is low and heat transport is mostly determined by radiation. If there is an initial difference in temperature between the cavity walls, this difference will disappear due to the radiant heat exchange between both surfaces. Since temperature differences are equalized in the cavity, the impact of buoyancy on the flow field will be less. Simultaneously the diffusion of water vapour from and to the cavity leafs is determined by the vapour diffusion resistance of the porous materials which is often larger than that of air. This again reduces the impact of the flow conditions in the cavity on the convective transport from cavity walls to cavity and explains why the simplifications introduced by Karagiozis et al. (2009) still result in reasonable agreement with measurements. 3 NEW VENTILATED CAVITY WALL MODEL Figure 3 shows an example of a cavity wall configuration. This configuration will be used for a more detailed study on heat and moisture transport. The cavity wall has an outside leaf of ceramic brick, an air cavity of 5cm and an inside leaf of wood fibre board, mineral wool insulation and gypsum board as inside finishing. Some material properties of these materials are listed in Table 1. The moisture transport properties such as vapour diffusion coefficient and moisture retention curve are found in Van Belleghem (2013). In this paper a conjugate modelling approach of the heat and moisture transport in the air and porous material is used. Heat and moisture transport in the air is solved together with the transport in the porous material. A special coupling procedure is used to assure heat and mass flux continuity and temperature and mass fraction continuity at the air-porous material interface. More details on the coupling procedure are found in Van Belleghem (2013). However there are some cases where the previously listed assumptions no longer apply, for example if a cavity wall is saturated with water and dried by convection. During the first drying stage, when the drying rate is constant, the moisture transport is determined by the convective boundary conditions. In order to accurately predict the drying of a wet cavity it is thus important to capture the convective boundary conditions in the cavity with a reasonable accuracy. It is not clear to what extend the simplifications proposed by Karagiozis et al. (2009) hold. It is thus Figure 3. Cavity wall configuration: wood frame wall with brick veneer as outside leaf. The heat and moisture transport equations in the air are given by Equation 3 and 4. (3) 403 The liquid moisture content and vapour moisture content can be linked to the total moisture by the open porosity ψ0, taking into account that w=wl+wv and ψ0=wl\ρl+wv\ρv. (4) With C the weighted heat capacity of moist air: (5) Equations 3 to 7 are solved using a finite volume solver. The equations are implemented into a commercial available CFD solver. Generally a calculation domain is divided into two zones: a porous material zone and an air zone. Four continuity conditions have to be fulfilled when air and material zone are coupled. − Continuity of temperature at the boundary: the temperature at the air side boundary Tsa should equal the temperature at the material side boundary Tsm. Tsm=Tsa=Ts − Continuity of the heat flux at the boundary: Heat conduction in the porous material to the surface equals the convective heat leaving the surface. − Continuity of mass fraction of water vapour at the boundary: The mass fraction at the material side of the air-material interface Ysm equals the mass fraction at the air side Ysa. Ysa=Ysm=Ys − Continuity of moisture flux at the boundary. Y is the mass fraction of water vapour in air [kg/kg], T is the temperature [°C], λeff is the effective conductivity [W/m²K] and is the sum of the turbulent and molecular conductivity, v is the velocity [m/s], ρ is the air density [kg/m³], Cv is the heat capacity of water vapour, Ca is the heat capacity of dry air, g is the mass flux of water vapour [kg/m²s]. Deff is the effective water vapour diffusion coefficient [m²/s]. For a detailed derivation of these equations the reader is referred to Steeman et al. (2009). Table 1. Hygrothermal material properties. Property Ceramic brick Wood fibre board ρ [kg/m³] 2087 270 Cmat [J/kgK] 840 1550 λmat [W/mK] 1+0.0047w 0.048 µdry [kg/m³] 24.79 6 Wcap [kg/m³] 130 162 ψ0 [-] 0.13 0.83 Property Mineral wool Gypsum board ρ [kg/m³] 60 690 Cmat [J/kgK] 1470 840 λmat [W/mK] 0.023 0.198 µdry [kg/m³] 10.68 Wcap [kg/m³] 295 ψ0 [-] 0.419 A velocity inlet at the bottom of the cavity is used and at the top a pressure outlet is assumed. Inlet temperature and mass fraction are based on the exterior weather conditions. As outside conditions the climate in Brussels is used, based on data from Meteonorm (Meteotest 2012). Two cases will be studied here, a warm summer day in June and a colder day in December. Temperature and humidity on the 20th of June 1995 in Brussels are used as summer condition, and the 17th of December for winter conditions. The solar radiation is taken from Hens (1997) and is the maximum solar radiation for a clear sky on a vertical west facade during June and December respectively. Figure 4 shows the daily variation of the solar radiation while Figures 5 and 6 shows the temperature and relative humidity for the 20th of June and the 17th of December respectively. These conditions are used as exterior conditions for the ceramic brick outside leaf and as inlet conditions for the air cavity. The convective heat and mass transfer coefficients at the exterior wall surface were taken to be constant. The exterior heat transfer coefficient is 19W/m²K and the mass transfer coefficient is 0.0217s/m (which is within the range suggested by ASHRAE (2009)). The mass transfer coefficient is based on mass fraction water vapour as driving force. As interior conditions for the cavity wall 21°C is used as constant room temperature and 50% RH as constant room relative humidity. The convective heat and mass transfer coefficients at the interior wall are 8W/m²K and 0.00915s/m respectively. The heat and moisture transport equations in the porous material are given by Equations 6 and 7. = (6) (7) In these equations Cmat [J/kgK] is the heat capacity of the dry porous material and ρmat [kg/m³] is the density of the dry material, wl [kg/m³] and wv [kg/m³] are the liquid and vapour moisture content respectively. Cl [J/kgK] is the heat capacity of liquid water. λmat [W/m²K] is the heat conductivity of the material including the moisture content effect. L is the latent heat of water [J/kg], RH is the relative humidity [-], pv is the partial vapour pressure, psat is the saturation vapour pressure. Kl is the liquid permeability and µ is the vapour resistance factor. pc is the capillary pressure [Pa]. 404 in Van Belleghem (2010) and Van Belleghem (2011) and gave good results. With the coupled CFD-HAM model it is possible to include convective transport in the cavity more accurately. However convective transport is not the only transport mechanism in a cavity. Figure 1 showed the different heat transport mechanisms present in a cavity wall. Air velocity in a ventilated cavity wall is generally low. As a result the convective heat transport is low and no longer the main heat transport mechanism. Heat transfer due to long wave radiation starts to play a major role at these low velocities. Therefore radiation cannot be neglected when studying coupled heat and moisture transport in cavity walls and a radiation (face-to-face) model is added to the heat transport model. Figure 4. Total solar radiation in winter(red) and summer(blue). 4 DRYING OF A VENTILATED CAVITY WALL To evaluate the performance of a simplified cavity model, a cavity wall configuration under specific boundary conditions was simulated for a period of one day with WUFI-2D® and compared with simulations performed with the coupled model. Figure 3 shows the cavity wall configuration that is used. The cavity itself has a width of 5cm. The simulations were performed under summer and winter boundary conditions. Figures 4, 5 and 6 show the outside temperature, relative humidity and radiation during a warm, sunny day in June and colder day in December. In total four simulation cases are studied: − Case 1: Summer conditions are used as boundary. The outside cladding, composed out of a brick veneer wall is assumed initially saturated with water. This mimics the situation after an intensive rain shower. Initial conditions for both simulation models (WUFI® and the coupled model) are listed hereafter. The brick layer was initially assumed almost saturated with water with a relative humidity in the brick of 99.99% and a corresponding moisture content of 129.7kg/m³. The initial temperature of the brick was assumed to be 18°C. The cavity air layer had an initial temperature of 25°C and a relative humidity of 50%. The inlet conditions of the cavity were the same as the outside boundary conditions. The wood fibre board had an initial moisture content of 17.8kg/m³ which corresponds with a relative humidity of 60%. The initial temperature of the wood fibre board plate was 25°C. The mineral wool and gypsum board also had an initial temperature of 25°C and a relative humidity of 60%. − Case 2: Winter conditions are used as boundary. Again the outside cladding is assumed initially saturated with water. Sky radiation during the night is neglected. Figure 5. Temperature and relative humidity for a summer day. Figure 6. Temperature and relative humidity for a winter day. Different time scales are present when heat, air and moisture transport in porous materials is modelled. Transport phenomena in the air have a much smaller time constant than phenomena in porous materials. It is thus not necessary to model all time variations in the air, since fluctuations in the air with high frequency will have no impact on the heat and moisture transport in the material. Therefore the air can be modelled as quasi-steady-state. This implies that flow unsteadiness is neglected. The flow field can be assumed constant during the time step if this time step is not too large. Changes in the flow field by changing boundary conditions are however still included. The time step size used for the simulations with the coupled model was 60 seconds. This value is based on earlier simulations with the coupled model 405 The brick had an initial temperature of 9°C and a moisture content of 129.7kg/m³, corresponding with a relative humidity of 99.99%. The inlet conditions of the cavity were again the same as the outside conditions. The initial temperature and relative humidity of the cavity air was 9°C and 80% respectively. The wood fibre board had an initial moisture content of 17.8kg/m³, a temperature of 9°C and a relative humidity of 60%. The temperature of the insulation was initiated at 15°C and the relative humidity in the insulation was 60%. The gypsum board facing the indoor environment had an initial temperature of 20°C and a relative humidity of again 60%. − Case 3: Summer conditions are used as boundary. This case is similar to case 1, only the moisture content of the wood fibre board differs. The wood fibre board is, similar to the brick, initially assumed saturated with water. The initial moisture content of the woof fibre board is 160kg/m³. This situation mimics for example rain penetration to the inside leaf or water leakage resulting in a wet inside leaf. − Case 4: This case resembles case 3, only here winter conditions are used as boundary condition. The same initial conditions are used as listed in case 2. The initial moisture content of the wood fibre board is 160kg/m³. imum difference of 9.4% is found between the coupled model and WUFI®. (a) (b) Figure 7. Case 1: the moisture content in the brick (a) and wood fibre board (b) for a summer day, starting from saturated brick veneer and relatively dry wood fibre board. Comparison of the coupled model for different cavity air flow rates (0.1m/s-, 0.2m/s , 0.3m/s---) and WUFI® (0.1m/s- -, 0.2m/s , 0.3m/s---). An inlet velocity of 0.2m/s was assumed. This velocity corresponds to a ventilation rate of respectively 288ACH. This air change rate was used as input for the simplified WUFI® model. Constant velocities at the inlet were assumed so the air change rate in the cavity is also constant over time. Temperature and moisture gradients in the cavity however result in a redistribution of the velocity profile in the cavity due to buoyancy. This distribution changes in time since the temperature and moisture distribution in the cavity change. This results in transfer coefficients which strongly vary in time and space. The simplified model does not take these variations into account. Figure 7 shows the comparison of the coupled model with the simulation result for the simplified model in WUFI® for case 1. In Figure 7(a) the moisture content in the brick veneer is depicted. Both models clearly show the same trends. Drying starts slow as the temperature of the surroundings and cavity is low and the relative humidity in the air is still high. At sunrise, the temperatures gradually rise and solar radiation further heats up the cavity, which increases the drying rate. This can be seen in the larger slope of the moisture content graph after 10am. However an overestimation of the drying rate by the WUFI® model compared to the coupled model can be noticed. At an inlet velocity of 0.2m/s, a max- -- -- Table 2 compares the maximum relative difference of the simulated moisture content in WUFI® and the coupled CFD-HAM model for all four cases. The relative difference is determined by dividing the absolute difference by the moisture content predicted by the CFD-HAM model. The table indicates that the difference in predicted moisture content in the wood fibre board is smaller than in the brick. This is because the brick starts from saturation while the wood fibre board only contains hygroscopic moisture. When a saturated material is dried, drying will take place in the first drying stage (constant drying rate period). During the first drying stage the drying rate is determined by the convection conditions. For the wood fibre board however moisture content is much lower and moisture is transported in the board by vapour diffusion. The moisture transfer from air to material and vice versa is in this case determined by the vapour diffusion properties of the porous material and less by the convection conditions in the air. In other words, the impact of convection is the largest for drying in the first drying stage. For hygroscopic 406 loading the impact of simplified modelling of the convection is less. Table 2. Relative difference of predicted tween WUFI® and the coupled model. Brick Case 1 9.4% Case 2 1.9% Case 3 9.1% Case 4 1.9% relative difference increased from 2.8% to 23.6% in summer and from 8.4% to 14.5% in winter. The WUFI model clearly predicts a faster drying at the cavity side. In the brick this difference was less pronounced since the drying of the brick took place at two sides. For the wood fibre board only drying at the cavity side is possible. moisture content beWood fibre board 2.8% 8.4% 23.6% 14.5% Figure 8 shows the simulation results for the drying of a cavity wall under winter conditions. Case 2 is similar to case 1 as they both start from a saturated brick wall. The winter conditions applied in case 2 will result in lower temperatures in the cavity wall which in turn results in lower drying rates. Also the relative humidity in the air is higher for these winter conditions as can be seen in Figures 5 and 6. The drying course of the brick is mainly determined by the relative humidity in the air. During a large part of the day the relative humidity is close to 100%. When the air is saturated the brick cannot dry out. The drying rate is no longer determined by the convection coefficients but by the humidity in the air. As a result there is a better agreement between WUFI® and the coupled model (only a difference of 1.9%). For the wood fibre board the difference between both models (8.4%) is larger than in case 1. Here the wood fibre board is hygroscopically loaded. During the whole day the mass fraction of water vapour in the cavity air is higher than in the wood fibre board and the moisture content of the board monotonically rises. In case 1 periods of hygroscopic loading were altered with periods of drying. This way, the too high moisture content during loading is compensated by the too high drying rate during drying and the overall difference between both models is less for this case. In Figure 7 differences of the predicted moisture content in WUFI® and the coupled model were larger for the saturated brick than for the unsaturated wood fibre board. However the impact of drying on the cavity side is partly masked since the brick dries out mainly to the outside, where convection is higher. This results in the WUFI® model performing reasonably well for brick drying (maximum deviation between WUFI® and coupled model around 9.4%). However, the situation worsens when drying of wood fibre board is considered. These simulation results are shown in Figure 9. In case 3 and 4 not only the ceramic brick was initially saturated with water but also the wood fibre board. The moisture content in the wood fibre board was compared for a simulation with WUFI® and a simulation with the CFDHAM model. Table 2 shows the remarkably higher maximum difference between both simulations. The (a) (b) Figure 8. Case 2: the moisture content in the brick (a) and wood fibre board (b) for a winter day starting from saturated brick veneer and relatively dry wood fibre board. Comparison of the coupled model (-) and WUFI® (- -). 5 CONCLUSIONS The analysis in this paper showed some of the capacities of the newly developed coupled model. The model allows a more detailed study of the complex heat and moisture transfer mechanisms in ventilated cavity walls. In the past convection in the cavity was often modelled in a simplified way (e.g. Straube & Finch 2009). This study showed that these simplifications are not always justified. To study the impact of the simplified convection modelling in a cavity a comparison was made between WUFI® which uses a simple convection model and the newly developed coupled model which models convection uncompromised. The comparison showed that the simplified model overestimates the drying and moistening rates of the cavity wall. Differences in predicted moisture content up to 23.6% are registered. Winter conditions resulted in less severe differences when drying of 407 saturated walls was modelled, because for these cases the high relative humidity in the air limits the drying rates. The largest discrepancies were found for simulations in summer conditions at the inside leaf when this leaf was initially saturated. Both models showed that the drying of the outer leaf (brick veneer) is mainly determined by the outside conditions and dries out mainly to the outside and not to the cavity. The cavity ventilation in this case is of less importance. For the inside leaf this cavity ventilation is of major importance for drying. The study showed that the largest discrepancies between both models were found for this inner cavity leaf. This study revealed that the simplified model cannot be used to evaluate the drying potential of a ventilated cavity. The simplified model systematically overestimates the ventilation effect. This has severe consequences when the model is used for cavity wall evaluation. Overestimating drying rates results in hazardous situations going unnoticed. The simulation would in such case indicate lower moisture contents than in reality and consequently lower risk for mould growth, wood rot or other structural damage. A correct evaluation of ventilated cavity walls is only possible if convection is modelled in detail. The newly developed coupled model allows such evaluation. 6 ACKNOWLEDGEMENT The results presented in this paper were obtained within the frame the research project IWTSB/81322/Van Belleghem funded by the Flemish Institute for the Promotion and Innovation by Science and Technology in Flanders. Their financial support is gratefully acknowledged. 7 REFERENCES Ge H., and Ye Y. 2007. Investigation of Ventilation Drying of Rainscreen Walls in the Coastal Climate of British Columbia. Proc. Thermal Performance of the Exerior Envelopes of Whole Buildings X. Florida, USA. Hens H., Janssens A., Depraetere W., Carmeliet J., and Lecompte J. 2007a. Brick cavity walls: a performance analysis based on measurements and simulations. Journal of Building Physics, 31(2):95-124 Hens H. 2007b. Building Physics – Heat, Air and Moisture: Fundamentals and Engineering Methods with Examples and Exercises. Berlin: Ernst & Sohn. Karagiozis A.N., and Künzel H.M. 2009. The effect of air cavity convection on the wetting and drying behavior of wood-frame walls using a multi-physics approach. Journal of ASTM International, 6(10). Meteotest, 2012. Meteonorm: global meteorological database. http://www.meteonorm.com Salonvaara M., Karagiozis A.N., Pazera M., and Miller W. 2007. Air cavities behind claddings – What have we learned? Proc. Thermal performance of the exterior envelopes of whole buildings X, Florida, USA. Sanjuan C., Jose Suarez M., Blanco E., and del Rosario Heras M. 2011. Development and experimental validation of a simulation model for open joint ventilated facades, Energy and Buildings, 43(12):3446-3456. Steeman H.J., Van Belleghem M., Janssens A., and De Paepe M. 2009. Coupled simulation of heat and moisture transport in air and porous materials for the assessment of moisture related damage. Building and Environment, 44(10):21762184. Straube J. 1998. Moisture control and enclosure wall systems. PhD, University of Waterloo, Ontario, Canada. Straube J., and Finch G. 2009. Ventilation Wall Claddings: Review, Field Performance and Hygrothermal Modeling, Technical report, Building Science Corporation. Van Belleghem M., Steeman H.-J., Steeman M., Janssens A., and De Paepe M. 2010. Sensitivity analysis of CFD coupled non-isothermal heat and moisture modelling. Building and Environment, 45(11):2485-2496. Van Belleghem M., Steeman M., Willockx A., Janssens A., and De Paepe M. 2011. Benchmark experiments for moisture transfer modelling in air and porous materials. Building and Environment, 46(4):884-898. Van Belleghem M. 2013. Modelling coupled heat and moisture transfer between air and porous materials for building applications. PhD. Ghent University, Ghent, Belgium. Zirkelbach D., Schmidt T., Kunzel H.M., Kehrer M., and Bludau C. 2007. WUFI 2D, Fraunhofer Institute for Building Physics. (a) Case 3: drying in summer conditions (b) Case 4: drying in winter conditions Figure 9. The drying course of a wet wood fibre board in a cavity wall under summer (case 3) and winter (case 4) conditions. Comparison between the coupled model (-) and WUFI® (- -). 408 On the use of the logarithmic of the capillary pressure for numerical simulation of moisture flow C. Rode Technical University of Denmark, Department of Building Physics, Lyngby, Denmark L. Juhl Technical University of Denmark, Department of Building Physics, Lyngby, Denmark ABSTRACT: Several choices exist for the selection of a driving potential for use in simulation of moisture transport in building constructions. The use of capillary pressure (Pc) as driving potential is challenging for numerical simulations since both the potential itself and the associated transport coefficient, the water permeability, vary by several decades between wet and dry areas of the materials. The hypothesis of this study is that the difficulties can be alleviated by replacing the capillary pressure with its logarithmic value (LPc). It reduces by orders of magnitude both the gradients in the driving potential and the huge variations in the transport coefficient. Consequently, the use of LPc entails a potential to use coarser discretizations, and thus less computational effort while achieving same accuracy. The study concludes that the use of LPc has the potential to outperform the use of Pc in terms of speed and accuracy, if the right precautions are made. ∂LPc ∂w = KPc ∂t ∂x 1 INTRODUCTION Numerical moisture calculations have become very important in order to analyse the performance of building envelopes. The first physical models accounting for vapour diffusion and capillary liquid transport in porous building materials were published in the late fifties and sixties (Philip & de Vries 1957; Luikov 1966). In many of the numerical models developed since, the capillary pressure has been applied as the moisture driving potential; e.g. Janssen et al. 2007 and Kalagasidis et al. 2004. However some researchers, e.g. Pedersen 1990 and Williams et. al. 2012, have applied its logarithmic value in order to reduce by orders of magnitude both the gradients in the driving potential and the huge variations in the transport coefficient. Experiences regarding the differences in terms of speed and accuracy have, to the authors’ knowledge, not been published. The present study investigates and compares the performance of using the two driving potentials. In total 250 simulations have been conducted. The performance of the two methods is likely to depend on the amount of applied control volumes. In order to clarify if the two methods will perform differently when applying a different number of control volumes, the number of control volumes has randomly been chosen between 5 and 273 for each of the 250 simulations. 2.1 Performance indicators The performance is assessed based upon: • Accuracy • Number of iterations • Computational effort 2.1.1 Accuracy In order to compare the simulation results with a reference solution (HAMSTAD 2002), five reference points have been selected to assess the accuracy of the calculations. The absolute error in moisture content [kg/m3] is evaluated at each point. The accumulated values state the precision of the simulations. 2 METHOD The study is conducted on the basis of a simplified 1-dimensional isothermal moisture model which applies either the capillary pressure (Pc), Eq. (1), or its logarithmic value (LPc) as driving potential, Eq. (2). ∂P ∂w =K c ∂t ∂x (2) 2.1.2 Number of iterations The number of iterations used when running a simulation is logged and applied as a performance indicator in order to assess which method requires the least iterations. (1) 409 2.1.3 Computational effort The calculation time is logged in order to assess the computational effort of the two methods. It should be noted that the study is conducted applying analytical material data based on capillary pressure (Pc). Hence, when conducting simulations applying LPc as the driving potential, transformations from Pc to LPc are required at each node point and each iteration step in order to evaluate moisture permeability and capacity of the material. This must be considered in the assessment of the two methods. 2.2.5 Material properties The sorption isotherm for the applied material is illustrated in Figure 2. The liquid permeability is illustrated in Figure 3. The shaded area illustrates the range applied during the analysis (RH: 0.45 - 0.95). 2.2 Case study Figure 2. Sorption isotherm. The study is conducted on the basis of the HAMSTAD Benchmark WP2#2 (HAMSTAD 2002). The result of the Benchmark case is applied as a reference solution, illustrated in Figure 4. The benchmark simulation data, construction and material data, boundary and initial conditions, are stated below. 2.2.1 Construction data The 1-dimensional simulations are conducted for a 0.2 m homogenous porous material. The reference points used for comparison of simulation performance are located at x = 0.00, 0.02, 0.10, 0.18 and 0.20 m. Figure 3. Liquid permeability. 2.2.6 HAMSTAD reference solution Figure 1. Construction. 2.2.2 Initial conditions The initial moisture content (t=0) is stated in Eq. (3). Figure 4. Hamstad benchmark solution. w 84.7687 = [kg / m3 ] ( RH 0.95 ) 2.3 The numerical model (3) The study is based on the use of a Finite Control Volume method (FCV) implemented in the MATLAB environment. The numerical model is a simple implicit FCV moisture transfer model, based on Eq. (7) applying Pc as the driving potential and Eq. (8) applying LPc. 2.2.3 Boundary conditions The relative humidity as interior and exterior boundary conditions (t>0) are stated in Eq. (4) and (5). = RH i 0.45 [−] (4) RH = 0.65 [−] e (5) 2.2.4 Surface transfer coefficients The surface vapour transfer coefficients are stated in Eq. (6). β=i β= 1.0 ⋅10−3 [m / s ] e (6) ∂P ∂w ∂Pc =K c ∂Pc ∂t ∂x (7) ∂LPc ∂w ∂LPc = KPc ∂LPc ∂t ∂x (8) Applying the notation stated in Figure 5 the numerical implementation of Eq. (7) and Eq. (8) can be written as stated in Eq. (9) and Eq. (10). 410 ∂w ∆xi ( Pc,i − Pold c,i ) = ∂Pc ∆t Kw ( Pc,i−1 − P ∆xw where win ,q represents the moisture content in con- ) + ∆Kx ( P e c ,i c ,i −P c ,i +1 ) trol volume i at iteration q, and win ,q +1 represents the new solution; the moisture content in i after iteration q+1. (9) e 3 RESULTS AND DISCUSSION ∂w ∆xi LPc ,i − LP old c ,i ) = ( ∂LPc ∆t K w Pc ,i ∆xw ( LP c ,i −1 − LP c ,i )+ K e Pc ,i ∆xe ( LP c ,i − LP c ,i +1 ) (10) As stated in Section 2 the two numerical approaches (Pc and LPc) are evaluated with respect to three performance indicators: accuracy, number of iterations and computational effort. In this section the results of the 250 (x2) simulations are assessed. 3.1 Accuracy and utilized iterations In Figure 7 the amount of iterations per simulation is plotted as a function of the number of applied control volumes. Evidently there is a strong advantage in using LPc as the driving potential since the number of necessary iterations is significantly reduced. The observation is valid for the entire range of applied control volumes. In Figure 8 the accuracy in moisture content of simulations is plotted in terms of the number of applied control volumes. The figure shows that the LPc approach is more precise when applying a coarse spatial discretization (few control volumes). By use of approximately 25 control volumes or more the difference in accuracy is insignificant. Figure 5. Illustration of the applied notations. 2.3.1 Linearization technique The system of equations is non-linear since the capacity and permeability both depend on the unknown capillary pressures, and thus it has to be iteratively linearized to allow solution. The Picard iterative scheme is applied. 2.3.2 Spatial discretization The spatial discretization is uniformly distributed, as illustrated in Figure 6. 3.2 Computational effort In the prior section it was illustrated that LPc outperformed Pc in terms of utilized iterations per simulation. In Figure 9 the calculation time per simulation is illustrated. As it can be seen, there is a tendency towards Pc performing better in terms of calculation time. The same deduction can be drawn by analysing Figure 10, which illustrates the calculation time per iteration. The explanation of this observation is seemingly straightforward: As mentioned in Section 2.1.3 the sorption isotherm and the liquid permeability are expressed in terms of Pc. This triggers a timeconsuming conversion of LPc to Pc in order to evaluate the material parameters. LPc would presumably outperform Pc in terms of calculation time if the material parameters beforehand were expressed in terms of LPc. It would eliminate the need for timeconsuming transformations at every individual simulation. Figure 6. Spatial discretization, FCV model. 2.3.3 Temporal discretization The model is configured to apply dynamic time stepping. The number of iterations (q) determines the new time step: if q < 10, ∆t new =1.1∆t old if q > 15, ∆t new = 0.9∆t old The time step, ∆t, is limited to a maximum of one hour. 2.3.4 Convergence criteria An iterative solution of the calculated moisture content is accepted when the convergence criteria stated in Eq. (11) is fulfilled for all control volumes: abs (1 − win ,q / win ,q +1 ) ≤ 1 ⋅10−8 (11) 411 Figure 7. Iterations per simulation. Figure 8. Accuracy: absolute error in moisture content [kg/m3]. Figure 9. Calculation time per simulation [sec]. Figure 10. Calculation time per iteration [sec]. The calculation time includes all simulation tasks. Figure 11. Calculation time per simulation [sec]. The calculation time includes computational effort utilized solving the Picard scheme. 412 Figure 12. FEM simulation, applying 4 elements. Figure 13. FEM simulation, applying 9 elements. Figure 14. FEM simulation, applying 14 elements. An illustration of the expected reduction in calculation time can be seen in Figure 11, which illustrates exclusively the calculation time spent solving the Picard Scheme. As it appears, the time consumption is similar for the two methods. Hence, if the material parameters are expressed in terms of LPc the amount of iterations required multiplied with the calculation time states that LPc will outperform Pc in terms of total simulation time. planation of this is presumably that the description of the material data, transport coefficient and the water permeability, are stated in terms of capillary pressures and conversion from Pc to LPc is required in order to evaluate the parameters. This operation is required for each control volume and iteration. Hence, the calculation time increases. However, this issue could presumably be addressed if the material parameters are converted to being functions of LPc. The overall conclusion of this study is that the use of LPc as driving potential has the potential to outperform the use of Pc if the right precautions are made. 3.3 Finite-Element-Method Applying the finite-element program COMSOL a superficial study of the difference between the two approaches (Pc vs. LPc) has been conducted. The COMSOL simulations were conducted on the basis of Eq. (7) and (8). The simulation results are illustrated in Figure 12 to 14. By analysing the figures it is evident that the observed tendency regarding accuracy while applying the FCV method also holds when applying the FEM method. A further analysis of the calculation time and number of iterations while performing simulations in COMSOL has not been addressed in this study. 5 REFERENCES HAMSTAD. 2002. Methodology of HAM modeling. Benchmark WP2#2. Chalmers University of Technology. Document. Chalmers 2002-h14. Kalagasidis A.S., Bednar T. and Hagentoft C.-E. 2004. Evaluation of the Interface Moisture Conductivity between Control Volumes - Comparison between Linear, Harmonic, and Integral Averaging, Proceedings of Buildings IX Conference. Luikov A.V. 1966, Heat and Mass Transfer in Capillaryporous Bodies, Pergamon Press, London. Janssen, H., Blocken, B. and Carmeliet, J. 2007. Conservative modelling of the moisture and heat transfer in building components under atmospheric excitation, International Journal of Heat and Mass Transfer 50 (2007) 1128–1140. Pedersen C.R. 1990. Combined heat and moisture transfer in building constructions, PhD thesis, Technical University of Denmark, Denmark. Philip J.R., and de Vries D.A. 1957. Moisture Movement in Porous Materials under Temperature Gradient, Transactions, American Geophysical Union, 38: 222-232. Williams Portal N.L., van Aarle M.A.P., and van Schijndel A.W.M. 2011. Simulation and Verification of Coupled Heat and Moisture Modeling. Proceeding of the European Comsol Conference, Stuttgart, 26-28 October 2011. 4 CONCLUSION The study confirms that the use of LPc reduces the amount of iterations required in order to reach the convergence criteria. Furthermore the performance in terms of accuracy is better when applying coarse spatial discretization. Applying a normal or fine spatial discretization the accuracy using Pc seems similar to that of using LPc. The computational effort/simulation time, applying LPc, was slightly higher compared to Pc. An ex- 413 Nomenclature w t ∆x K Pc LPc RH β total moisture content (kg/m3) time (s) spatial discretization (m) permeability (s) for Pc gradients capillary pressure (Pa) Logarithm of the capillary pressure (Pa) relative humidity surface vapour transfer coefficient (kg/m2 s Pa) Subscripts and superscripts int/ext i e/w q old / new 414 internal/external control volume identification east/west of “i” iteration number Previous / new time step The extent and implications of the urban heat island phenomenon in Central European region K. Kiesel, M. Vuckovic, A. Mahdavi Vienna University of Technology, Department of Building Physics and Building Ecology, Vienna, Austria ABSTRACT: Metropolitan areas worldwide display highly diverse microclimatic circumstances that are influenced by a variety of morphologies, structures, materials (particularly urban surface properties), and processes (mobility, industry, etc.). This diversity influences the intensity and extent of the urban heat island effect (UHI) in different cities. UHI may be understood in terms of emerging divergence between microclimatic conditions in the city proper versus the rural environs. Significantly higher temperatures are observed in the urban area as compared to the surrounding suburban and rural neighborhoods. A further rise in the appearance and intensity of UHI phenomena is to be expected in the coming years due to the on-going population increase in urban areas. Furthermore, the UHI effect is believed to be related to (and worsened by) the climate change. Thereby, the rise of global temperatures is likely to affect not only the health of the urban population (urban heat distress, pedestrian discomfort) but also the energy performance of the built environment (higher outdoor air temperatures lead to a significant increase in buildings' energy use for cooling). In this context, this paper presents the results of an on-going EU-supported research project, which investigates the urban heat island phenomena in a number of urban regions in Central European countries (Stuttgart, Warsaw, Prague, Padua, Ljubljana, Modena, and Budapest). Toward this end, we pursue a twofold approach. First detailed information regarding urban and rural climate in a 7-day period for each of the participating cities was collected and analysed. The results show a considerable variance, which, if ignored, would lead to major uncertainties in inferences made based on performance simulation. Secondly, long term data on rural and urban climate was obtained for all participating cities and included in the analyses. 1 INTRODUCTION cities (Harlan et al. 2011). Additionally, higher air temperatures have a direct effect on the energy use due to increased deployment of air conditioning (Akbari 2005). In this context, this paper presents the results of an on-going research project that investigates the urban heat island phenomena in the Central European area (Mahdavi et al. 2013). Within the framework of the aforementioned UHI project, we collected a large set of data concerning the extent of the UHI effect in multiple cities in Central Europe. Analysis of the data reveals the extent of the UHI effect and a considerable variance in its manifestations. Toward this end, we pursued a two-fold approach. First, detailed information regarding urban and rural climate in a 7-day period for each of the participating cities was collected and analysed. The results show a considerable variance, which, if ignored, would lead, amongst other things, to major uncertainties in inferences made based on thermal performance simulation. Secondly, long term data on rural and urban climate was obtained for all participating cities and included in the analyses. Recently, a number of research efforts have been initiated to better understand the variance in microclimatic conditions due to factors such as urbanization, presence and density of industrial or commercial buildings, green areas, bodies of water, etc. (Grimmond 2007, Alexandri 2007). The geometry, spacing, and orientation of buildings and surrounding open areas greatly influence the microclimate in the city (Kleerekoper et al. 2012). Looking on the smaller scale, microclimate can vary significantly across an area consisting of even a few streets. On a greater scale, this deviation is observed in terms of significantly higher urban temperatures than that of the surrounding rural environment. This circumstance (see, for example, Voogt 2002) is referred to as the urban heat island phenomenon (UHI). It is usually quantified by the term Urban Heat Island Intensity (∆θ), which is the difference between urban and background rural temperatures. Furthermore, the UHI effect is thought as being directly related to (and worsened by) the climate change. Increase in average temperatures is believed to adversely affect the health of people living in 415 2 URBAN HEAT ISLAND QUANTIFICATION island effect, for the purposes of presented analysis, we operate with ∆θ as a generic indicator. The specific aim of this paper is to identify and evaluate the extent of the UHI effect and its variance in the broader geographical context of the participating cities. Table 1 includes some general information about our research project's participating cities in terms of area, population, latitude, longitude, and altitude. Additional information concerning cities' location and topology is provided in Table 2. Numerous studies have been carried out discussing and quantifying the UHI phenomenon (see, for example, Arnfeld 2003, Blazejczyk 2006). Efforts have been made to describe the characteristics and patterns of UHI (Voogt 2002, Hart and Sailor 2007). Observations have shown that the UHI phenomenon shows different characteristics during different seasons (Gaffin et al. 2008) and that is pronounced differently during the night and the day (Oke 1981). Furthermore, the intensity of urban heat islands is believed to rise proportionally to the size and population of the urban area (Oke 1972). More recently, Gaffin et al. (2008) performed a detailed spatial study of New York city’s current UHI and concluded that summer and fall periods were generally the strongest UHI seasons, consistent with seasonal wind speed changes in the area. The UHI most often refers to the increase of urban air temperature when compared to rural. Generally, heat island intensities are quantified in the range of 1–3 K (Voogt 2002). Furthermore, Voogt also noted that under certain atmospheric and surface conditions, the maximum observed heat island magnitudes can be as high as 12 K. The UHI phenomenon has also been extensively studied in terms of the effect on the urban microclimate and energy use for heating and cooling of buildings (Stewart and Oke 2012, Kolokotroni et al. 2007). Furthermore, material properties of urban surfaces can result in higher urban temperature compared to that of rural area (Grimmond et al. 1991, Akbari et al. 2001). Taha (1997) examined the impacts of surface albedo, evapotranspiration, and anthropogenic heat emission on the near-surface climate and found out that increases in urban albedo or increase in vegetation in urban areas can reduce air temperature up to 2 K. Table 1. Information about the participating cities. City Area Population Latitude Longitude Altitude [km2] [millions] range[m] Budapest 525 1.74 47° 30' N 19° 3' E 90-529 Ljubljana 275 0.28 46° 3' N 14° 30' E 261-794 Modena 183 0.18 44° 39' N 10° 55' E 34 Padua 93 0.21 45° 25' N 11° 52' E 8-21 Prague 496 1.26 50° 5' N 14° 25' E 177-399 Stuttgart 207 0.60 48° 46' N 9° 10' E 207-548 Vienna 415 1.73 48° 12' N 16° 22' E 151-543 Warsaw 517 1.70 52° 13' N 21° 00' E 76-122 Table 2. Information about the urban topology. City Vienna Vienna is located in north-eastern Austria, at the eastern most extension of the Alps in the Vienna Basin. Stuttgart Stuttgart's center lies in a Keuper sink and is surrounded by hills. Stuttgart is spread across several hills, valleys and parks. Padua Padua is located at Bacchiglione River, 40 km west of Venice and 29 km southeast of Vicenza. The Brenta River, which once ran through the city, still touches the northern districts. To the city's south west lie the Euganaean Hills. Budapest The Danube River divides Budapest into two parts. On the left bank the Buda is located, with over 20 hills within the territory of the capital, and on the right bank the flat area of Pest is located with its massive housing, as well as commercial and industrial areas. Prague Prague is situated on the Vltava river in the center of the Bohemian Basin. Modena Modena is bounded by the two rivers Secchia and Panaro, both affluents of the Po River. The Apennines ranges begin some 10 km from the city, to the south. Warsaw Warsaw is located some 260 km from the Baltic Sea and 300 km from the Carpathian Mountains. Furthermore, Warsaw is located in the heartland of the Masovian Plain. Ljubljana Ljubljana is located in the Ljubljana Basin between the Alps and the Karst Plateau. 3 METHODOLOGY The definition, description, and quantification of the UHI effect rely on a large body of both short-term and long-term measurement results (Gaffin et al. 2008). In this context, we were particularly interested in quantifying the frequency, magnitude, and time-dependent (diurnal and nocturnal) UHI intensity distribution in a course of a reference week. Long-term development of urban and rural temperatures was another point of interest. The magnitude of the UHI effect can be expressed, amongst others, in terms of Urban Heat Island intensity (∆θ). This term denotes the temperature difference (in K) between simultaneously measured urban and rural temperatures. While there may be more detailed and informative means of expressing the urban heat 416 Topology 4 RESULTS The assessment of current UHI intensity in observed urban areas has been derived from data sets in a course of a reference week. The reference week was chosen by each participating city independently, in order to provide the most reliable input information. Each participating city provided data (including air temperature, wind speed, and precipitation) from two representative weather stations (one urban and one rural). Data was recorded on hourly basis. These data sets needed to be suitable for the UHI analysis. This presumed that the air temperatures during the whole period should be considerably high, while the wind speed should preferably be below 5 m/s for most of the time. From the hourly values of UHI intensity the cumulative frequency distribution for the reference week period was calculated. Moreover, the weeklong data for each city was processed into mean hourly urban temperature and UHI values of a reference day. To obtain a long-term impression of the urban and rural temperature development in the participating cities, mean annual (urban and rural) temperatures and UHI values were derived for a period of 30 years. With two exceptions (Modena, Warsaw), the record set was obtained from the same two representative weather stations (urban and rural) used for the short-term analysis. Table 3 provided an overview of the time periods used for both the short-term and the long-term analyses. 4.1 Short-term (reference week) analyses Figure 1 shows the cumulative frequency distribution of UHI values for the participating cities for the reference week. Figures 2 and 3 show, for a reference summer day (representing the reference week), the hourly values of urban temperature and the mean hourly UHI values respectively. 100 90 Cumulative frequency [%] 80 VIENNA 70 STUTTGART 60 PADUA 50 BUDAPEST 40 PRAGUE 30 MODENA 20 WARSAW 10 LJUBLJANA 11 8 9,5 6,5 5 3,5 2 0,5 -1 -2,5 -4 -5,5 0 UHI intensity [K] Figure 1. Cumulative frequency distribution of UHI intensity for a one week summer period. VIENNA STUTTGART PADUA BUDAPEST PRAGUE MODENA WARSAW 23:00 21:00 19:00 17:00 15:00 13:00 11:00 09:00 07:00 05:00 03:00 LJUBLJANA 01:00 θ [0C] Table 3. Overview for the data sets used for the analysis. Long-term Climate Data Reference URBAN RURAL Week STATION STATION Budapest 20-26.8.2011 2000-2011 2000-2011 Ljubljan 20-26.8.2011 1980-2011 1980-2011 a 20-26.8.2011 1980-2010 1980-2009 Modena Padua 18-24.8.2011 1994-2011 1994-2011 Prague 8-14.7.2010 1980-2011 1980-2011 Stuttgart 20-26.8.2011 1981-2011 1980-2011 Vienna 20-26.7.2011 1994-2011 1994-2011 Warsaw 9-15.6.2008 1980-2011 1980-2011 39 37 35 33 31 29 27 25 23 21 19 17 15 Time Figure 2. Mean hourly urban temperature for a reference summer day. 7 6 VIENNA UHI intensity [K] 5 STUTTGART 4 PADUA 3 BUDAPEST 2 PRAGUE 1 MODENA WARSAW 0 LJUBLJANA 23:00 21:00 19:00 17:00 15:00 13:00 11:00 09:00 07:00 05:00 03:00 -2 01:00 -1 Time Figure 3. Mean hourly UHI intensity distribution for a reference summer day. 417 4.2 Long-term analyses 5 DISCUSSION Figures 4 and 5 show for the participating cities the (mean annual) urban and rural temperatures respectively over a period of 30 years. Figure 6 shows the long-term UHI intensity trend over the same period. The reference week data clearly demonstrate the existence and significant magnitude of the UHI effect in participating cities, especially during the night hours (Figures 1 and 3). However, the timedependent UHI patterns vary considerably across the participating cities. In Warsaw, for example, UHI intensity level ranges from around 2 K during daytime to almost 7 K during the night, while in Stuttgart levels are rather steady, ranging from 1 K to 2 K. The UHI pattern differences are also visible in the cumulative frequency distribution curves of Figure 1. In this Figure, a shift to the right denotes a larger UHI magnitude. The historical temperature records suggest an upward trend concerning both urban and rural temperatures (see Figures 4 and 5). Consistent with regional and global temperature trends, a steady increase in rural temperatures of up to about 2.5 K can be observed in all selected cities (with the exception of Budapest, for which data was available only for a rather short period). In the same 30-years period, the mean annual urban temperature rose somewhere between 1 K (Stuttgart) and 3 K (Warsaw). A number of factors may have contributed to this trend, namely increase in population, energy use, anthropogenic heat production, and physical changes in the urban environment (e.g., more high-rise buildings, increase in impervious surfaces). Note that, while both rural and urban temperatures have been increasing, the value of the UHI intensity has been rather steady. Our data suggest increasing UHI intensity trends in Warsaw and Ljubljana, whereas a slight decrease can be discerned from Stuttgart and Prague data (see Figure 7). 16 15 14 VIENNA 13 STUTTGART θ [0C] 12 PADUA 11 BUDAPEST 10 PRAGUE 9 MODENA WARSAW 8 LJUBLJANA 7 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992 1990 1988 1986 1984 1982 1980 6 Year Figure 4. Development of (mean annual) urban temperatures over a period of 30 years. 16 15 14 VIENNA 13 STUTTGART θ [0C] 12 PADUA 11 BUDAPEST 10 PRAGUE 9 MODENA WARSAW 8 LJUBLJANA 7 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992 1990 1988 1986 1984 1982 1980 6 Year Figure 5. Development of (mean annual) rural temperatures over a period of 30 years. 1,5 1,5 1,4 VIENNA UHI intensity [K] 1,4 STUTTGART PADUA BUDAPEST PRAGUE MODENA PRAGUE 1,1 MODENA WARSAW 1,0 WARSAW LJUBLJANA LJUBLJANA 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992 1990 1988 1986 1984 1982 1980 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992 1980 0,9 0,9 1990 1,0 BUDAPEST 1988 1,1 PADUA 1,2 1986 1,2 STUTTGART 1984 1,3 1,3 1982 UHI intensity [K] VIENNA Year Year Figure 7. Long-term UHI intensity trend over a period of 30 years. Figure 6. Long-term development of the UHI intensity over a period of 30 years. 418 6 CONCLUSION 8 REFERENCES We presented the initial results of an EU-supported project concerned with the extent of the UHI phenomena in a number of Central European cities. The objectives of this project are to provide a common understanding of the UHI effects and to conceive and evaluate appropriate mitigation and adaptation measures. We presented both short-term and long-term data with regard to urban and rural temperatures in the participating cities (Stuttgart, Warsaw, Prague, Padua, Ljubljana, Modena, Vienna, and Budapest). The analysis results demonstrate the existence and significant magnitude of the UHI effect in all participating cities. A time-dependent (diurnal and nocturnal) pattern could be observed implying larger UHI intensities during the night hours. However, the hourly based observations show a significant variation in UHI intensity in different cities, especially in terms of peak values. These results imply the need for further studies concerning UHI as a variable phenomenon over space and time and especially in a broader geographical context. Finally, the findings stress the importance of assessment and modeling approaches that would establish a link between UHI intensity and salient urban variables such as urban density and morphology, block layout, canyon geometry, surface properties, vegetation, bodies of water industrial sites, transportation systems and infrastructures. The development of a systematic UHI assessment and modeling framework (Mahdavi et al. 2013) represents a critical component of our ongoing project. Akbari H., Pomerantz M., and Taha H. 2001. Cool surfaces and shade trees to reduce energy use and improve air quality in urban areas, Solar Energy, Volume 70, Issue 3: 295–310. Akbari H. 2005. Energy Saving Potentials and Air Quality Benefits of Urban Heat Island Mitigation, Lawrence Berkeley National Laboratory, Berkeley, CA. Alexandri E. 2007. Green cities of tomorrow?, Sustainable Construction, Materials and Practices, Portugal SB07: 710-717. Arnfeld A.J. 2003. Two decades of urban climate research: a review of turbulence, exchanges of energy and water, and the urban heat island, International Journal of Climatology, Volume 23, Issue 1: 1-26. Blazejczyk K., Bakowska M., and Wieclaw M. 2006. Urban heat island in large and small cities, 6th International Conference on Urban Climate, Göteborg, Sweden, June1216 2006: 794-797. Gaffin S. R., Rosenzweig C., Khanbilvardi R., Parshall L., Mahani S., Glickman H., Goldberg R., Blake R., Slosberg R. B., and Hillel D. 2008. Variations in New York city’s urban heat island strength over time and space, Theoretical and applied climatology, Volume 94: 1-11. Grimmond C.S.B. Cleugh, H., and Oke, T.R. 1991. An objective urban heat storage model and its comparison with other schemes, Atmospheric Environment, Volume 25B, Issue 3: 311-326. Grimmond C.S.B., 2007. Urbanization and global environmental change: local effects of urban warming, Cities and global environmental change, Volume 173, Issue 1: 83-88. Harlan S L., and Ruddell D.M. 2011. Climate change and health in cities: impacts of heat and air pollution and potential co-benefits from mitigation and adaptation, Current Opinion in Environmental Sustainability, Volume 3, Issue 3: 126-134. Hart M., and Sailor D.J. 2007. Assessing causes in spatial variability in urban heat island magnitude, Seventh Symposium on the Urban Environment, San Diego, CA, 9 September 2007. Kleerekoper L.,van Esch M., and Salcedo T.B. 2012. How to make a city climate-proof, addressing the urban heat island effect, Resources, Conservation and Recycling, Volume 64: 30-38. Kolokotroni M., Zhang Y., and Watkins R. 2007. The London Heat Island and building cooling design, Solar Energy, Volume 81,Issue 1: 102–110. Mahdavi A., Kiesel K., and Vuckovic M. 2013. A framework for the evaluation of urban heat island mitigation measures, SB13 Munich Conference, Germany, to be published. Oke T.R. 1972. City size and the urban heat island, Atmospheric Environment, Volume 7, Issue 8: 769-779. Oke T.R. 1981. Canyon geometry and the nocturnal urban heat island comparison of scale model and field observations, Journal of Climatology, Volume 1: 237–54. Stewart I.D., and Oke T. R. 2012. Local Climate Zones for Urban Temperature Studies, Bulletin of the American Meteorological Society, Volume 93: 1879–1900. Taha H. 1997. Urban climates and heat islands: albedo, evapotranspiration, and anthropogenic heat, Energy and buildings, Volume 25, Issue 2: 99-103. Voogt J.A. 2002. Urban Heat Island, Encyclopedia of Global Environmental Change, Volume 3: 660-666. 7 ACKNOWLEDGEMENTS This project was funded in part within the framework of the EU-Project "Development and application of mitigation and adaptation strategies and measures for counteracting the global Urban Heat Island phenomenon" (Central Europe Program, No 3CE292P3). 419 420 Urban energy and microclimate: Wind tunnel experiments and multiscale modeling J.Allegrini1,2, P. Moonen1,2, S. Saneinejad1,2, V. Dorer1 & J. Carmeliet1,2 1 Swiss Federal Laboratories for Materials Science and Technology (Empa), Laboratory for Building Science and Technology, Dübendorf, Switzerland 2 Swiss Federal Institute of Technology Zürich (ETHZ), Chair of Building Physics, Zürich, Switzerland ABSTRACT: An urban microclimate model including air flow, heat and moisture transport in porous urban surfaces and solar and longwave radiation is presented, validated with wind tunnel experiments and used to study the effect of evaporative cooling on the thermal comfort in a street canyon. has to be coupled with an UMC model (Bouyer et al. 2011, Rasheed 2009). The urban microclimate is modeled using a computational fluid dynamics (CFD) model for the urban air domain, a radiation model including solar radiation and longwave radiative exchange between urban surfaces (figure 1). This urban microclimatic model interacts with building energy simulation models and heat and moisture (HAM) transfer models for evaporative cooling processes situated at the porous building envelope and porous urban surface. The UMC model has to be linked to a regional meteorological meso-scale model to cover urban heat islands. 1 INTRODUCTION A major part of the final energy consumption in our nowadays society is due to buildings and cities. For the future, we have to find new urban energy concepts based on the integration of renewable energy supply, conversion, storage, distribution and management on neighbourhood or urban scale, where buildings will become interconnected, harvesting, exchanging and storing energy. The objective might be that local neighbourhoods become energy selfregulating, minimizing the additional supply from national or international energy systems. This challenging new concept has the potential to substantially decentralize the energy sector. The energy demand of neighbourhoods depends not only on their building energy systems, but also on the microclimate created around the buildings, which can differ substantially for rural, suburban and urban areas (Watkins et al. 2002). The development and analysis of these new urban energy scenarios has to be supported by advanced multiscale simulation methods covering the urban microclimate (UMC), including urban heat island effects and building energy systems on city quarter level. Even for buildings in an urban context, common practice in detailed building energy simulation (BES) still relies on stand-alone building configurations, not accounting for the influence of neighbouring buildings, except perhaps for shading. However, the urban climate and microclimate can strongly affect the building energy demand (Santamouris et al. 2001, Kolokotroni et al. 2006, Stromann-Andersen et al. 2011). Compared to stand-alone buildings, buildings in urban context experience higher ambient temperatures due to urban heat island effects and local heat rejection from other buildings, an altered radiation balance, due to the presence of surrounding buildings, and changed convective heat exchange, due to the different wind flow pattern. Thus, BES Figure 1. The Urban Microclimate Model (UMC) consists of an air flow model solved by CFD, a heat and moisture transport model taking into account evaporative cooling effects by porous urban surfaces, a radiation model and a building energy simulation model. A detailed study regarding the influence of the urban microclimate in street canyons on the energy demand for space cooling and heating of buildings can be found in Allegrini et al. (2012a). It was shown that the space cooling demand for the stand-alone build421 and Reynolds numbers at which the measurements are conducted, together with the corresponding Froude numbers. To reach the same Froude numbers in wind tunnel as in full-scale, rather high surface temperatures are needed for the wind tunnel measurements. Further, it is assumed that the flow is in a Reynolds independent flow regime. More details can be found in Allegrini et al. (2013). ings is much lower compared to the buildings situated in the street canyons, while the heating demand decreases for the urban case. In Allegrini (2012), the UMC model including the urban heat island effect is explained in detail and the reader is referred to this work. In this paper, we focus on two aspects related to the urban microclimate model on street canyon scale: the influence of buoyancy on the flow field in a street canyon and the influence of evaporative cooling on the thermal comfort in a street canyon. Table 1. Surface temperatures and Reynolds numbers with corresponding Froude numbers Re 70 °C 90 °C 110°C 130°C 9000 1.49 1.04 0.80 0.65 14200 3.68 2.58 1.99 1.62 19200 6.75 4.74 3.65 2.97 24600 11.11 7.79 6.00 4.88 30700 17.29 12.13 9.34 7.59 2 WIND TUNNEL EXPERIMENTS AND VALIDATION 2.1 Wind tunnel setup 2.2 Wind tunnel results This study is conducted in the ETHZ / Empa atmospheric boundary layer wind tunnel in Dübendorf, Switzerland. The flow in an urban street canyon is modelled by building up a cavity of 0.2 x 0.2 x 1.8 m (W x H x L; W: width; H: height; L: length) in the wind tunnel test section (Figure 2). Time averaged streamlines for isothermal cases and cases with windward wall and leeward wall heating are given in Figure 3 for three different Reynolds numbers. Figure 2. Street canyon model installed in the ETH / Empa wind tunnel. For the measurements, the wind tunnel floor and the street canyon surfaces are made out of aluminium plates, thus forming a low roughness surface. With such low roughness, low-Reynolds number modelling can be applied for the validation study (see below). Heating mats are attached at the back of each aluminium plate forming the street canyon to increase the surface temperatures. Due to the high thermal conductivity of aluminium, the temperature is uniform over the whole heated surface. The flow field in a cross-section of the street canyon is measured with particle image velocimetry (PIV). 300 double frame images are recorded at a frequency of 4 Hz and time averaged temperature and turbulent kinetic energy (TKE) fields are determined. For some specific cases, the air temperatures inside the street canyon are measured with thermocouples on an equidistant grid of 10 x 10 measurement points. The flow direction is normal to the street canyon axis and the inflow temperature is 23 °C. Table 1 gives the different surface temperatures Figure 3. Streamlines for isothermal cases and cases with windward wall and leeward wall heating. If no surface is heated, one main vortex in the centre of the street canyon and corner vortices in the bot422 yon and has a similar size as the main vortex at lower Reynolds numbers. For high Reynolds numbers (>24600) the size of the secondary vortex decreases and at the highest Reynolds numbers (30700) the flow structure becomes similar to the vortex structure of the isothermal case. For leeward wall heating, the main vortex becomes strengthened by the buoyancy effect (Figure 3), mainly for low Reynolds numbers. In Figure 5 normalised velocity profiles are given on the vertical centreline of the street canyon for different Reynolds numbers and different leeward wall temperatures. At low Reynolds numbers a strong increase of the velocities with increasing leeward wall temperature can be observed. For a leeward wall temperature of 130 °C the velocities inside the street canyon are more than double compared to the isothermal case. For high Reynolds numbers no strong dependency of the velocity profiles on temperature can be noticed, because the flow is in a forced convective flow regime. tom corners are formed. With leeward wall heating the structure is the same, but the velocities inside the street canyon are increased, because buoyancy strengthens the main vortex. For windward wall heating with low Reynolds numbers the flow structure changes significantly. Instead of one main vortex, two vortices with similar size can be found inside the street canyon. For higher Reynolds numbers the flow structure is similar to the structure of the isothermal case (forced convection). The velocities in the street canyon with windward wall heating are significantly lower compared to the other cases. The changes of the flow field, caused by buoyancy, show the importance of studying the detailed flow fields in urban areas for the determination of the convective heat transfer at building facades. Figure 5. Normalised velocity profiles on the vertical centreline of the street canyon for different Reynolds numbers and different leeward wall temperatures. The results of the wind tunnel measurements show the importance of buoyancy on the flow in an urban street canyon. This does not only influence the thermal comfort of pedestrians, but has also an impact on the convective heat transfer at building facades. Here the convective heat transfer was not measured. 2.3 CFD models and simulations For this validation study, steady 2D RANS CFD simulations using ANSYS-Fluent 12.0 are conducted with the standard and the realizable k-ε model. A 2D geometry was chosen to save computational power for future coupled BES-CFD simulations. At the near-wall regions the boundary layers are either resolved with low-Reynolds number modelling (LRNM) or modelled with wall functions (WFs) (Allegrini et al. 2012b). The dimensions of the computational domain are set according to the wind tunnel dimensions (Figure 6). The grid is built based on a grid sensitivity analysis and consists of 5500 cells. It is refined towards the walls. The same mesh is used for simulations with LRNM and WFs. Figure 4. Streamlines for cases with windward wall heating and five different Reynolds numbers. For the cases with windward wall heating, buoyancy significantly changes the vortex structure. Streamlines for the five different Reynolds numbers are presented in Figure 4 (windward wall temperature: 130 °C). Besides the main vortex a second counter rotating vortex is formed. This secondary vortex is located in the lower part of the street can423 The following two CFD model combinations are used: (i) LRNM with a realizable k-ε model (ii) nonequilibrium wall functions with a standard k-ε model. Further simulations are conducted with the same models as the simulations with NEWFs, but the extension of the upstream domain in front of the street canyon is only 0.25 H (NEWF short). With this short upstream domain, the boundary layer cannot develop and therefore the approach flow velocities do not decay as in the case with the long upstream domain. In short upstream domain case, the flow velocities at the entrance of the street canyon will be more accurate. In the boundary layer above the street canyon the simulation with NEWFs predict too low velocities. With LRNM and NEWFs with the short domain a very good agreement between the simulations and the measurements can be found. In the shear layer at the top plane of the street canyon the measured velocities are strongly decreasing and at H = 1 the slope of the profiles changes strongly. At this point the best agreement for the velocity can be found for the NEWFs, while the other two models overestimate the velocity. The same trend can be observed close to the ground of the street canyon. Therefore it can be concluded that the velocities inside the street canyon are mainly influenced by the velocities in the shear layer. Even with too low velocities above the street canyon the flow inside the street canyon can be predicted well (NEWFs). For the TKE all CFD models can capture the general trend. Inside the street canyon the NEWFs (long and short domain) perform better than LRNM that underestimates the TKEs. Figure 6. Computational domain. At the inlet, the measured boundary layer profiles for the velocity and turbulent kinetic energy (TKE) are imposed (Figure 6). All surfaces are modelled without roughness, because no roughness can be defined with LRNM. This may be justified since the wind tunnel model was made out of aluminium with a very low surface roughness. For the surfaces inside the street canyon, the surface temperatures obtained from the experiments are used as a temperature boundary condition. The surfaces outside the street canyon are modelled adiabatic. Symmetry boundary conditions are imposed at the top boundary and outflow boundary conditions at the outlet. To account for buoyancy the density, the specific heat, the thermal conductivity and the viscosity are approximated with polynomial functions as a function of the temperature in the Navier-Stokes equations. Radiation is not considered in the CFD simulations, because for the wind tunnel measurements the surface temperatures were kept constant with heating mats and therefore radiation did not influence the flow field. 2.4 Validation results In Figure 7 normalized centreline profiles of the horizontal velocity and TKE are given for the case where all surfaces are heated inside the street canyon. Figure 8. Trajectories of the centre of the main vortex as a function of the freestream velocity for different windward wall temperatures. For windward wall heating the vortex structure is strongly influenced by buoyancy. The information gained from comparing centreline profiles is rather Figure 7. Normalised centreline profiles of the horizontal velocity for cases with all surfaces heated inside the street canyon (70 °C). 424 limited, because already small changes of the centre of the vortices strongly change the profiles. Therefore here the trajectories of the centre of the main vortex are compared (Figure 8). As for the measurements also for the CFD simulations the centre of the main vortex is moving from the leeward wall side towards the windward wall side and then to the centre of the street canyon with increasing Reynolds numbers. The trajectories are similar for the simulations with LRNM and NEWFs. For LRNM the flow is more in the forced regime than for NEWFs for all simulations. It can be concluded that with CFD the general development of the vortex structure can be predicted for cases with windward wall heating. In Figure 9 contour plots (measured, LRNM and NEWFs) of the air temperature are given for the case with a Froude numbers of 6.75, 17.29 and 7.59. It is expected that the most accurate results would be achieved by LRNM, where the boundary layer at the walls can be accurately resolved. With LRNM the temperatures are underestimated for all profiles by about 10 °C. The results can be improved by using NEWFs. In the near-wall regions (ground, windward and leeward walls) the temperatures are overestimated by NEWFs due to the overestimation of the thickness of the temperature boundary layer. In the centre of the street canyon a good agreement for the temperatures with NEWFs can be found. A reason for the better agreement of the temperatures with NEWFs compared to LRNM could be that the velocities in the shear layer correspond better to the measured velocities and therefore the predicted air exchange rate is more accurate. Another possible reason is that the convective heat transfer predictions are more accurate due to a different near-wall treatment. This cannot be evaluated here, because the wall heat fluxes were not measured. 2.5 Discussion Due to buoyancy there is a strong coupling between the flow field inside the street canyon and the heat fluxes at the boundaries of the street canyon. Therefore already small errors in the prediction of the fluxes of the flow field can significantly change the results of the CFD simulations. In the case of a street canyon the important fluxes are the convective heat fluxes at the three surfaces, the heat flux out of the street canyon and into the street canyon (red arrows in Figure 10). Figure 10. Sketch of the heat fluxes in a street canyon. The main driver of all the fluxes is the heat exchange at the top plane of the street canyon. There is a strong coupling between the heat fluxes and the flow structures inside the street canyon. For example an increase of the leeward wall temperature can increase the heat exchange at the top plane due to increased buoyancy effects. This increase can lead to lower temperatures at the windward wall and a significant change in the whole flow structure inside the street canyon. Therefore it can be concluded that accurate flow predictions at the top plane of the street canyon (blue box in Figure 10) are needed to get accurate flow predictions inside the street canyon with CFD. 3 EVAPORATIVE COOLING IN A STREET CANYON A second example of the Urban Microclimate (UMC) model is the influence of evaporative cooling on the thermal comfort in a street canyon. The air flow due to forced convection and buoyancy, including heat and vapour transport in the air domain, is solved using CFD, the effect of solar and longwave radiation including multiple reflections is solved using a radiation model, while evaporative cooling in the porous urban materials is modelled using a heat and moisture transport model (Saneinejad et al. 2012). The 2D street canyon of 10 x 10 m is exposed to two days during June taken from a typical meteorological year (TMY) of Zürich. The air temperature varies between 13.5°C and 19°C and the relative Figure 9. Contour plots of the air temperature inside the street canyon. Wind tunnel measurements and CFD simulations for cases where all surfaces are heated using two different turbulence models and different near-wall modelling approaches (LRNM: Low-Reynolds number modelling with realizable k-ε; NEWF: non-equilibrium wall functions with standard k-ε). 425 humidity varies between 62% and 86%. The windward wall is assumed to be wet after a rain shower (RH ≈ 100%) while the other surfaces are dry. The walls of the canyon are made of ceramic brick (0.09 m tick) and have albedo of 0.4. The soil is covered with 0.1 m concrete layer, assumed to be dark colored with albedo of 0.1. The reference wind speed at 10 m height is 5 m/s. To evaluate the effect of evaporative cooling, we compare the surface temperature at the top point (9.5 m), and bottom point (0.5 m) of the windward wall, as well as the average air temperature and relative humidity in the street canyon for two cases: (i) when the windward wall is initially wet, and (ii) when the windward wall is initially dry (Figures 11a-d). It can be seen that evaporative cooling results in a maximum drop of the wall surface temperature of 15°C at the top location(Figure 11a) and 13°C at the bottom location (Figure 11b) of the wall, during the first day of drying. The material at top and bottom is in its first drying phase, where water evaporates from the surface of the wall, and a lot of latent heat is extracted. On the second day the temperature difference at the top location (Figure 11a) is less because the wall at this location has reached the second drying phase, where the moisture front recedes into the material and therefore less energy is required for evaporation. The average air temperature in the street canyon (Figure 11c) is approx. 3°C lower in the case with the wet wall, due to evaporative cooling. The average relative humidity (Figure 11d) is approx. 15% higher, compared to the dry case, due to mixing of moisture evaporated from the wall with the air in the canyon. Further we evaluate the effectiveness of evaporative cooling by studying the comfort conditions of a person standing in the street canyon. For this study, a 1.8 m tall person standing 1 m away from the windward wall is considered. To study the comfort of this person, we use the universal thermal climate index (UTCI) (Fiala et al, 2012). The parameters used for determining the UTCI are the air temperature, the vapor pressure, the wind speed and the mean radiant temperature. Figure 12a shows the UTCI for a person standing 1 m away from the windward wall, for two cases, i.e. with and without evaporative cooling. It can be seen that the person can feel up to max. 2.5°C cooler during the warmest part of the day on the first day and 2.7°C on the second day due to evaporative cooling. To understand this phenomena better, we look at the four parameters which influence the UTCI at the studied location, being the air temperature, the mean radiant temperature (Tmrt), the vapour pressure (or RH), and the wind speed. The evolution of these parameters during the 48 hr drying, for cases with and without moisture, is shown in Figures 12b, c, d, e and f. It can be seen that for the case with moisture, the air temperature and Tmrt at the studied location are approx. 2°C and 8°C lower a b c d Figure 11. Comparison of cases with and without moisture (a) surface temperature at the height of 9.5 m, (b) surface temperature at the height of 0.5 m on the windward wall, (c)average air temperature and (d) average relative humidity in the street canyon. 426 a b c d e f Figure 12. Comparison of cases with and without moisture for a) UTCI b) air temperature c) mean radiant temperature d) vapour pressure e) relative humidity and f) wind speed, at the position of a person standing 1 m from the windward wall. is low, since in this case the removal of heat by wind is less efficient. in comparison to the case without moisture, respectively, while the vapor pressure is approx. 2 hPa (relative humidity approx. 15%) higher and the wind speed is approx. 0.3 m/s lower. Lower air temperature and radiant temperature due to evaporative cooling result in an improved thermal comfort, while a higher relative humidity and lower wind speed result in a lower thermal comfort. The more comfortable conditions during the second day is that some parts of the wall are already in the second drying phase and dry slower, resulting in a lower vapor pressure in the street canyon. Further a parametric study learns that the evaporative cooling is most effective when the wind speed 4 CONCLUSION 427 An urban microclimate model is presented, which consists of an air flow model solved by CFD, a heat and moisture transport model taking into account evaporative cooling effects by porous urban surfaces, a radiation model taking into account solar and longwave radiation including multiple reflections and a building energy simulation model. For a detailed description of the this model and results describing the influence of the urban microclimate on the building energy we refer to our accompanying paper at this conference (Modelling the urban micro- Santamouris M., Papanikolaou N., Livada I., Koronakis I., Georgakis C., Argiriou A., and Assimakopoulos D.N. 2001. On the impact of urban climate on the energy consumption of buildings. Solar Energy 70 (3): 201-216. Stromann-Andersen J., and Sattrup P.A. 2011. The urban canyon and building use: urban density versus daylight and passive solar gains. Energy and Buildings 43: 2011-2020. Watkins R., Palmer M., Kolokotroni P., and Littlefair P. 2002. The London Heat Island - results from summertime monitoring. Building Services Engineering Research and Technology 23 (3): 97-106. climate and its impact on the energy demand of buildings and building clusters). In this paper we first present wind tunnel results showing the effect of buoyancy on the air flow and temperature in a street canyon. These results are used to validate our CFD model. The results show that CFD can predict the general flow structures and the influence of buoyancy. The detailed flow field inside the street canyon is strongly dependent on the flow structure inside the shear layer at the top of the street canyon. Therefore to get accurate results for the flow profiles inside the street canyon, the flow inside the shear layer has to be predicted correctly. Then, the urban microclimate model is used to study the effect of evaporative cooling on thermal comfort in a street canyon. It is shown that evaporative cooling can enhance the thermal comfort at low wind speeds by cooling down the urban surfaces and the air in the street canyon. However, due to the evaporation the relative humidity of the air in the street canyon increases, which results in a lowering of the thermal comfort. 5 ACKNOWLEDGEMENTS Funding by the Swiss Federal Office of Energy (project no. 154 143) and CCEM (Project Urban Multiscale Energy Modelling) is gratefully acknowledged. 6 REFERENCES Allegrini J. 2012. Urban Climate and Energy Demand in Buildings. PhD thesis no. 20848, ETH Zürich. Allegrini J., Dorer V., and Carmeliet J. 2012a. Influence of the urban microclimate in street canyons on the energy demand for space cooling and heating of buildings. Energy and Buildings 55: 823-832. Allegrini J., Dorer V., and Carmeliet, J. 2012b. An adaptive temperature wall function for mixed convective flows at exterior surfaces of buildings in street canyons. Building and Environment 49: 55-66. Allegrini J., Dorer V., and Carmeliet J. 2013. Wind tunnel measurements of buoyant flows in street canyons. Building and Environment 59: 315-326. Bouyer J., Inard C., and Musy M. 2011. Microclimatic coupling as a solution to improve building energy simulation in an urban context. Energy and Buildings 43: 1549-1559. Fiala D., Havenith G., Bröde, P., Kampmann, B., and Jendritzky G. 2012. UTCI-Fiala multi-node model of human heat transfer and temperature regulation. International Journal of Biometeorology 56 (3): 429-441. Kolokotroni M., Giannitsaris I., and Watkins R. 2006. The effect of the London UHI on building summer cooling demand and night ventilation strategies. Solar Energy 80: 383-392. Rasheed A. 2009. Multiscale Modelling of Urban Climate. PhD Thesis 4531, EPF Lausanne. Saneinejad S., Moonen P., Defraeye T., Derome, D., and Carmeliet J. 2012. Coupled CFD, radiation and porous media transport model for evaluating evaporative cooling in an urban environment. Journal of Wind Engineering & Industrial Aerodynamics 104-106: 455-463. 428 Spatial distribution of wind-driven rain in the urban environment A. Kubilay, D. Derome, J. Carmeliet Chair of Building Physics, ETH Zurich, Switzerland Laboratory for Building Technologies, Empa, Switzerland B. Blocken Building Physics and Services, Eindhoven University of Technology, The Netherlands ABSTRACT: Wetting by wind-driven rain (WDR) refers to rain droplets impinging on building facades due to the co-occurrence of wind and rain. WDR is one of the most important moisture sources with potential negative effects on the hygrothermal performance and durability of buildings. WDR should be considered for energy-retrofitting of building envelopes, planning of energy efficient cities, and assessment of moisture transport through the building facade. Research in the past has shown that the wetting distribution on a building is strongly influenced by its neighboring buildings. Therefore, modelling the complex air flows and rain trajectories in an urban environment is needed to estimate the WDR distribution accurately. In this study, numerical simulations of WDR with an Eulerian multiphase (EM) model are performed for an arrangement of cubic buildings representing a small part of an urban neighborhood. For validation, a 1:2 scaled field experiment setup is built on the Empa campus in Dübendorf, Switzerland consisting of 9 cubes arranged in an urban configuration. The experimental geometry consists of an arrangement of nine cubic buildings, which are located in an urban roughness layer. A weather station is positioned near the experimental site in order to measure the horizontal rainfall intensity and the approach-flow wind profile. A total number of 18 WDR gauges are positioned in such a way that a high spatial resolution is achieved for two of the nine building models. With a more detailed understanding of WDR, this work may lead to improvements such as modifications of surface properties of building materials or improvements in rain control strategies. Three methods exist for estimating the WDR load on building surfaces: (1) measurements, (2) semiempirical methods, and (3) numerical simulations with Computational Fluid Dynamics (CFD). Generally, measurements are difficult, timeconsuming and prone to errors. They are also confined to the meteorological conditions present at the time of experiments and are case-specific. Semiempirical methods on the other hand are fast and easy to use but they only give approximations and they cannot provide detailed information in space and time. Furthermore, both measurements and semi-empirical techniques are limited in terms of spatial detail and resolution. As a result, the moisture source due to WDR is often considered to be uniform across large parts of the facade. This, in turn, may lead to large errors in moisture transport simulations, as in reality, WDR loading is far from uniform over the facade. CFD simulations are a strong option to achieve a higher spatial and temporal resolution. In the literature, the steady-state numerical solution technique developed by Choi (1991), which is based on Reynolds Averaged Navier-Stokes (RANS) equations and Lagrangian Particle Tracking (LPT), has been extensively used. Blocken & Carmeliet (2002) extended Choi’s simulation technique by adding the temporal component, allowing the determination of both the spatial and temporal distributions of driving rain for transient rain events. This approach was further 1 INTRODUCTION Wind-driven rain (WDR) is the type of rain, which has a horizontal velocity vector due to the effect of wind. WDR is one of the most important moisture sources on building facades (Blocken & Carmeliet 2010). The deposition of rain on building facades may lead to damage mechanisms, such as surface soiling due to runoff, weathering, algae growth, salt damage and frost damage and may also be a source of moisture leading to mold growth on the inside in uninsulated massive masonry walls. Moreover, several authors have indicated structural and performance problems in photovoltaic modules related to moisture-related soiling, ingress and degradation (Meyer & van Dyk 2004; Jorgensen et al. 2006). The WDR intensity is governed by several parameters, such as building geometry, environment topography in the urban context, position on the building facade, wind speed, wind direction, rainfall intensity and raindrop-size distribution (Blocken & Carmeliet 2002). WDR is an important boundary condition in building-envelope heat-air-moisture (BE-HAM) transport models, which analyse the hygrothermal performance of buildings. Therefore, it is crucial to have accurate information on the spatial distribution of surface wetting instead of using an average value across the facade. 429 validated with full-scale measurements on standalone buildings (Abuku et al. 2009; Blocken & Carmeliet 2002; Blocken & Carmeliet 2006a; Blocken & Carmeliet 2007; Briggen et al. 2009; Tang & Davidson 2004; van Mook 2002). Recently, it has been shown that the Eulerian Multiphase (EM) model gives encouraging results in terms of WDR calculations on the windward facade of a stand-alone building (Huang & Li 2010). Kubilay et al. (2013) showed that the EM model allows less computational complexity by decreasing the user time spent for the simulation by at least a factor of 10, while achieving comparable results with the LPT models in terms of the convective transport of rain phase. Moreover EM model gives WDR results over the total domain, where in LPT model, only WDR on chosen surfaces is targeted. In urban environments, the wind flow is influenced by other buildings or objects in the vicinity, which will have an impact on the WDR intensities. In past research, it has been shown that the effect of turbulent dispersion becomes important for cases where the smaller droplets travel parallel to the building facade at low wind speed values (Blocken et al. 2010). In an urban environment with an array of low-rise cubic buildings, there are regions where the flow between the buildings is weakly coupled with the free stream flow. These are the recirculation regions where the wind speed values are low. Figure 1 shows the main characteristics of such flows in a simple geometry. In this paper, the study on WDR in an urban environment is presented. The RANS standard kepsilon model (Launder & Spalding 1972) is used to calculate the wind-flow numerically around the buildings. A spatially scaled field measurement is conducted in order to validate the calculations. Section 2 describes the field measurement setup and presents some of the recorded meteorological data. In section 3, the numerical method to predict the WDR amount is explained and the experimental setup is numerically reproduced. In section 4, WDR simulation results are presented. Finally, section 5 provides a general discussion and conclusion. 2 FIELD MEASUREMENTS 2.1 Geometry In order to model the scaled representative urban environment, an array of three rows of three low-rise cubic buildings are placed on the Empa campus in Dübendorf, Switzerland in an orthogonal grid, as shown in Figure 2. The cubes each have dimensions H x H x H = 2 x 2 x 2 m³ and they are spaced H = 2 m apart from each other, leading to a canyon aspect ratio of 1. The scaling to real buildings is aimed to be close to 1:2. Figure 3 shows one of the cubic buildings in the array. The cubes are made of wood panels on a wood structure, finished with paint. The roofs are flat and covered with a polymeric membrane as roofing. The cubes are placed on an asphalt parking lot, and positioned on pavement blocks and wooden bars for further protection and convenience of relocation. The total roof height including the support beneath the cubes is 2.17 m. Note that the gap between the bottom of the cubes and the ground is covered in order to prevent any airflow underneath the cubes, which would distort the flow characteristics. The measurement setup is located in a suburban area with an open field towards the west-south wind directions. To the west, the field is about 50 H long, and further upstream there are high trees and buildings present. To the southwest, the field is about 75 H long, and further upstream there are high-rise buildings and a motorway. From northwest to east, there are several low-rise and high-rise buildings nearby. The main wind direction, from where the most WDR is obtained, is west due to the surroundings and the local meteorological data. According to the Davenport specifications, the surface roughness needs about 10 km of upstream length to adapt (Wieringa 1992). The location of the measurement site is depicted in Figure 4. The measurement field is within an urban roughness layer with an aerodynamic roughness length, y0 = 1.0 m. WDR measurements are conducted on the windward facades of the cubes A and B, which look towards west, indicated as shaded on Figure 2. In addition to the WDR data, the reference wind speed, the reference wind direction, and the horizontal rainfall intensity are recorded in order to provide boundary conditions for the actual CFD simulations. The weather station is positioned 3H = 6 m upstream of the array of cubes, as determined using preliminary CFD simulations. The position is chosen as near as possible to catch the internal boundary layer information and, at the same time, outside the region influenced by the cubes. Figure 1. Flow characteristics in street canyons. 430 upstream wind flow profile. For wind that approaches from west, the influence from the surrounding buildings and trees is estimated to be minimum. The horizontal rainfall intensity is measured by a rain gauge positioned near the weather mast. The rain gauge has a horizontal opening and a tipping bucket mechanism. All data are gathered on a 1-minute basis and afterwards averaged over a time interval of 10 min. The effective volume of the tipping bucket is 2 ml and the opening is 0.02 m2, which results in a resolution of 0.1 mm per tip. For rainfall intensities lower than 0.6 mm/h, the tipping bucket data will not be accurate, as 10-min rainfall intensity, on average, will be smaller than the rain gauge resolution. The WDR measurement on the building facades is conducted by WDR gauges with tipping bucket mechanisms. There are 18 WDR gauges in total, 9 are installed on cube A, and 9 on cube B (see Figure 3). Collectors of the WDR gauges are manufactured at Empa according to the guidelines of Blocken & Carmeliet (2006b). They have ordinary glass sheet surfaces, which are known to promote runoff. The collector area is 0.2 x 0.2 m2 and the effective tipping bucket volume is 1 ml, which result in a resolution 4 times higher than the horizontal rain gauge, or 0.025 mm. The rain gauge and the WDR gauges have been calibrated in July 2012 and have been recording since November 2012. For validation, the 10-minute-averaged WDR data will be compared to the simulation results using various modeling approaches. Figure 2. Schematic of the measurement setup. Figure 3. Positions of the WDR gauges on the facade. Figure 4. Surroundings of the measurement site in a 10 km radius. Figure 5. Collector of the WDR gauges. 2.3 Accuracy of WDR measurements 2.2 Measuring instruments The measurements are being conducted following experience from previous experiments on small and large buildings (Blocken & Carmeliet 2002, 2004, 2005, 2006a; Briggen et al. 2009). The studies indicate the following possible errors that should be taken into account when quantifying WDR on the In order to measure the reference wind direction and the reference wind velocity, the weather mast is equipped with an ultrasonic anemometer at 8 m height. Two additional cup anemometers are installed at 2 and 5 m height to measure the 431 In Figure 6, the wind rose of data recorded at 8 m height (4H) between 10/11/2012 and 30/11/2012 is given. During this period, the wind direction was distributed evenly between northwest and south. Figure 7 shows the meteorological data record of a rain event on 28/11/2012. During the rain event, the wind direction was mostly northwest. Note that the data record shows some preliminary measurements, which do not comply with all the guidelines mentioned. As the given guidelines must be strictly followed in order to keep the discrepancies due to measurement errors small, only few rain events suitable for CFD validation can be recorded over a year. building facades: (1) evaporation of droplets adhered to the collection area of the WDR gauge, (2) error due to rest-water that remains in the tipping bucket, (3) splashing of drops from the collection area, (4) condensation on the collection area, and (5) wind errors due to the disturbances of the wind field and the raindrop trajectories near the WDR gauge by the presence of the gauge itself. The most important cause of measurement errors is stated to be adhered droplet evaporation. During a rain event, there is always a certain amount of water adhered to the collection area, which is not collected in the reservoir. A numerical study by Blocken Carmeliet (2006b) shows that all the impinged droplets adhere to the surface until a certain threshold of impinged amount of WDR is reached. After the threshold value the total volume of adhered droplets remains constant, thus the error is considered to be important especially for light to moderate rainfall intensities. Figure 7. Measured data for the rain event on 28/11/2012. 3 NUMERICAL SIMULATIONS Figure 6. Wind rose (m/s) of the period between 10/11/2012 and 30/11/2012 at the measurement site. In the EM model, rain is regarded as a continuum as is the air (wind). Each class of raindrop size is treated as a different phase, as each group of raindrops with similar size will interact with the wind-flow field in a similar way. The total WDR impingement is calculated on all building facades at once with a summation over all classes of raindrop sizes by using the raindrop size distribution as the weighting factor. In order to keep the measurement errors as small as possible, firstly, rain events with large amounts of WDR must be selected. This way, the loss of water due to evaporation of the adhered droplets can be kept at a relatively limited amount. Secondly, during a rain event, there must not be many dry periods in between the rain showers, which could lead to restwater error in the tipping bucket. As a result, the rest-water would be recorded at the end of the dry period, or possibly could never be recorded as a result of evaporation. Thirdly, the selected rain events must have reference wind speed values lower than 10 m/s and with horizontal rainfall intensities lower than 20 mm/h in order to keep splashing error small. Finally, during the selected rain events, the wind direction must be approximately perpendicular to the building facade in order to limit the wind error. In the present study, this condition is even stricter, as the measurement field is only open in the west-southwest direction. 3.1 Governing equations For the air phase calculations, 3D steady RANS with the standard k-ε model (Launder & Spalding 1972) is used in the present study. The governing equations of the air phase can be summarized by the continuity and momentum equations, Eq. (1) and (2), respectively. ρa denotes the density of air, p the pressure, τij the Reynolds stresses. ∂u j ∂x j 432 =0 (1) ∂ρ a ui ∂ ( ρ a ui u j ) ∂p ∂τ ij + = − + ∂t δ xj ∂xi δ x j (2) Rain phase calculations are one-way coupled with the air phase, with the effect of raindrops on the fluid flow ignored. This is a valid assumption as the volumetric ratio of rain in air is below 1x10-4 (Huang & Li 2010). The continuity and the momentum equations for each rain phase are shown in Eq. (3) and (4), respectively. αk denotes the phase fraction of phase k, d the raindrop diameter, uki the velocity component of phase k, ui the velocity component of wind, ρw the density of the raindrops, g the gravitational acceleration, and Cd the drag coefficient. Note that turbulent dispersion is neglected in this study. δα k δ (α k ukj ) + = 0 δt δ xj Figure 8. Raindrop size distribution through a horizontal plane. 3.3 Model geometry and computational domain (3) 3µ Cd Re R δα k uki δ (α k uki ukj ) + = αk g + αk ( ui − uki ) δt δ xj ρ wd 2 4 (4) In Eq. (4), ReR denotes the relative Reynolds number, which is calculated with Eq. (5). ρ a d   (5) = u − uk Re R µ The computational domain includes the array of 9 cubic low-rise buildings, as described in section 2.1. The computational domain has dimensions of 35H x 15H x 6H, where H=2 m is the edge length of a cubic building. The blockage ratio of the domain is about 3.33% and the distance of the building group from the domain boundaries satisfies the guidelines stated in Tominaga et al. (2008) and Franke et al. (2011). The grid is shown in Figure 9, and consists of 1,761,056 hexahedral cells. The cell height on the ground surface is 0.1H, and the cell height on the building surfaces is 0.03H. 3.2 WDR parameters The WDR intensity is related to the unobstructed horizontal rainfall intensity by the specific catch ratio, ηd(k), and the catch ratio, η. The specific catch ratio is related to the rain phase k, which is a class of raindrop diameters, d. The catch ratio is related to the entire spectrum of raindrop diameters. The quantities of the WDR parameters can be obtained after the calculation of the rain phase volumetric ratio and velocity fields by the following equations: = ηd ( k ) Rwdr ( k ) α k Vn ( k ) = Rh ( k ) Rh f h ( k ) Figure 9. Computational grid. (6) 3.4 Boundary conditions η = ∫ f h ( Rh , d )ηd dd The inlet profile of mean wind speed is defined with the typical log-law expression: (7) d where Rwdr denotes the WDR intensity, Rh the horizontal rain intensity through the horizontal plane, fh(Rh,d) the raindrop size distribution through the horizontal plane (Blocken & Carmeliet 2004) and |Vn(k)| the velocity magnitude of the rain phase in the direction normal to the building facade. In the present study, the raindrop size distribution through the horizontal plane (see Figure 8) is based on the work of Best (1950). This work was based on a wide bibliographical survey and measurements for a large number of rain events at various locations. U ( y) =  y + y0  ln   κ  y0  u*ABL (8) where U(y) denotes the mean streamwise wind speed at height y above the ground plane, u*ABL the ABL friction velocity, κ the von Karman constant (0.42 in the present study), and y0 the aerodynamic roughness length. In the present study, an aerodynamic roughness length of 0.1 m is chosen. The ABL friction velocity, u*ABL, is chosen such that 433 boundaries are undisturbed and far away from the object in the domain. The rain phase velocity component along the direction of gravitational acceleration is set equal to the terminal velocity for that phase. The rain phase velocity components perpendicular to the direction of gravitational acceleration are set equal to the air phase velocity components, so that the relative velocity between wind and rain is zero at the boundary. The boundary conditions for the rain phases at the building walls, on the ground and at the outlet are set in such way that the normal gradient of the volumetric ratio, ∂αk/∂n, equals zero when the normal wind velocity vector is pointing out of the domain, and the values of the volumetric ratio, αk, are equal to zero when the normal wind velocity vector is pointing into the domain. With these boundary conditions, the interaction between the raindrops and the walls are not modelled and the raindrops leave the domain as soon as they hit a wall boundary, avoiding any inflow of rain phase into the domain due to possible recirculation regions. the desired reference wind speed at the height of the cubes, UH, is obtained. The inlet profiles of turbulent quantities are defined by Eq. (9) and (10). k ( y) = u*2 ABL Cµ ε ( y) = u*3 ABL κ ( y + y0 ) (9) (10) For wall treatment, the standard wall functions by Launder & Spalding (1974), with appropriate roughness modification (Cebeci & Bradshaw 1977), are used. The equivalent sand-grain roughness height, ks, and the roughness constant, Cs, is determined by the following relation (Blocken et al. 2007): ks = Ey0 Cs (11) where E is an empirical constant with a value of 9.793. In the present study, for the ground surface, ks is taken 0.1 m, and Cs is set as 9.7. The building surfaces are assumed to be smooth. For the top boundary, constant values are set for U, k, and ε by using the values from the inlet profiles at the same height as suggested by Blocken et al. (2007). This is done in order to limit the horizontal inhomogeneity, as other top boundary conditions, such as symmetry condition, can cause streamwise gradients. Although imposing constant values does not allow fluid to enter or exit the domain, the top boundary is far enough from the buildings to not cause a problem. A constant static pressure of 0 Pa is used at the outlet boundary. Symmetry conditions are applied on both sides of the domain. With the definitions established in section 3.2, the volumetric ratio of the rain phase k can be calculated as: αk = Rh f h ( Rh , d ) Vt ( d ) Figure 10. Terminal velocity of raindrops. 3.5 Solution strategy OpenFOAM® 2.0 is used in this study as the CFD code. It is an open-source, implicit, segregated, and double precision CFD toolbox. An additional solver has been implemented into the code by the authors for solving the governing equations of the rain phases. This solver gives the rain phase velocity, volumetric ratio, and specific catch ratio distributions in a one-way coupled fashion. The pressure-velocity coupling for the wind flow field solution is taken care of with the Semi-Implicit Method for Pressure Linked Equations (SIMPLE) algorithm. Second order discretization schemes are used for both the convection terms and the viscous terms of the governing equations. For the air phase calculations, the simulations were terminated when all the scaled residuals reached 10-6. For the rain phase calculations, variable values at various (12) where Vt(d) represents the terminal velocity of fall of a raindrop with diameter d. The terminal velocity of water drops was studied by Gunn & Kinzer (1949) and shown in Figure 10. The implemented drag coefficients should be in agreement with the terminal velocity data, otherwise some artificial acceleration towards the ground can be experienced during simulations even in regions with undisturbed flow field with zero vertical component. The drag coefficients measured by Gunn & Kinzer (1949) take into account changes in the shape of falling raindrops. The value of volumetric ratio of the kth phase, αk, is imposed at the inlet and top boundaries. For the rain phase velocity, uk, it is assumed that the 434 locations have been monitored for determination of convergence. In the present study, the following steps have been followed: a. The wind-flow field around the building is solved for UH = 3 m/s. The wind-flow fields for other values of the reference wind speed (UH = 1, 2, 4, 5, 6 m/s) are obtained by linear scaling. Such scaling is allowed for flows around sharp-edged bluff bodies, where the positions of flow separation are independent of the Reynolds number. b. Using each reference wind-flow field, specific catch ratio distributions are calculated for several raindrop diameter values (diameters ranging from 0.3 to 1 mm in steps of 0.1 mm, from 1 to 2 mm in steps of 0.2 mm, and from 2 to 6 mm in steps of 1 mm). c. Catch ratio distributions are obtained for a series of reference horizontal rainfall intensities Rh = 0, 0.1, 0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12, 15, 20, 25, and 30 mm/h using the droplet size distribution as input. Figure 11. Streamlines of the air (light) and rain (dark) phases at reference wind speed, UH = 4 m/s. 4 NUMERICAL RESULTS Figure 11 shows the steady-state simulation results of the streamlines of air and rain phases at a reference wind speed UH = 4 m/s for raindrop size d = 0.3 mm. The streamlines for the air phase have the characteristics as schematically depicted in Figure 1. The recirculation regions before and in-between the cubes result in different stream-wise rain phase velocities at the facades. Note how the streamlines of the rain phase go more parallel to the windward facade of cube B compared to cube A. The resulting surface wetting distribution is shown in Figure 12 for cubes A and B. Note that the catch ratio values are about the same on the upper quarter of the cubes. This is due to the fact that the wind velocity fields are very similar above that height. However, the effect of different flow regimes in front of the cubes can be observed by the difference of catch ratio values on the lower three quarters of the cubes. Cube B has a larger vertical wetting gradient although the maximum catch ratio is about the same as cube A. For the lower side corner of both cubes, the catch ratio chart is plotted in Figure 13 for different reference wind speed and horizontal rainfall intensity values. The catch ratio values tend to increase with increasing wind speed, and stay almost constant with respect to horizontal rainfall intensities except for the lower intensities. For higher reference wind speed values, there is a large difference in surface wetting between cubes A and B. At wind speed values higher than 5 m/s, the catch ratio values tend to decrease for cube B, which is against the general assumption that WDR is approximately proportional to the product of wind speed and horizontal rainfall intensity. Figure 12. Catch ratio distributions for the windward facades for cubes a) A, and b) B, at a reference wind speed, UH = 4 m/s. Figure 13. Catch ratio charts for cubes A and B at the lower side corners on the windward facades. 5 CONCLUSION In the present study, experimental and numerical work has been presented in order to analyze the surface wetting due to WDR in an urban neighborhood. The measuring equipment has been described and some meteorological data have been presented. The numerical model to calculate the WDR has been described and results have been shown. During the rain event on 28/11/2012, the wind direction was mostly northwest, which results in a 435 Blocken B., and Carmeliet J. 2010. Overview of three state-ofthe-art wind-driven rain assessment models and comparison based on model theory. Building and Environment, 45(3), 691-703. Blocken B., Deszö G., van Beeck J. and Carmeliet J. 2010. Comparison of calculation methods for wind-driven rain deposition on building facades. Atmospheric Environment, 44(14), 1714-1725. Blocken B., Stathopoulos T., and Carmeliet J. 2007. CFD simulation of the atmospheric boundary layer: wall function problems. Atmospheric Environment, 41(2), 238-252. Briggen P. M., Blocken B., and Schellen H. L. 2009. Winddriven rain on the facade of a monumental tower: Numerical simulation, full-scale validation and sensitivity analysis. Building and Environment, 44(8), 1675-1690. Cebeci T., and Bradshaw P. 1977. Momentum Transfer in Boundary Layers. New York: Hemisphere Publishing Corporation. Choi E. C. C. 1991. Numerical simulation of wind-driven rain falling onto a 2-D building. Paper presented at the Asia Pacific Conf. on Computational Mechanics, Hong Kong. Franke J., Hellsten A., Schlunzen K. H., and Carissimo B. 2011. The COST 732 Best Practice Guideline for CFD simulation of flows in the urban environment: a summary. International Journal of Environment and Pollution, 44(14), 419-427. Gunn R., and Kinzer G. D. 1949. The terminal velocity of fall for water droplets in stagnant air. Journal of Meteorology, 6(4), 243-248. Huang S. H., and Li Q. S. 2010. Numerical simulations of wind-driven rain on building envelopes based on Eulerian multiphase model. Journal of Wind Engineering and Industrial Aerodynamics, 98(12), 843-857. Jorgensen G. J., Terwilliger K. M., DelCueto J. A., Glick S. H., Kempe M. D., Pankow J. W., Pern F. J., and McMahon T.J. 2006. Moisture transport, adhesion, and corrosion protection of PV module packaging materials. Solar Energy Materials and Solar Cells, 90, 2739-2775. Kubilay A., Derome D., Blocken B., and Carmeliet J. 2013. CFD simulation and validation of wind-driven rain on a building facade with an Eulerian multiphase model. Building and Environment, 61(0), 69-81. Launder B. E., and Spalding D. B. 1972. Lectures in mathematical models of turbulence. London, New York: Academic Press. Launder B. E., and Spalding D. B. 1974. The numerical computation of turbulent flows. Comput. Methods Appl. Mech. Eng., 269-289. Meyer E. L. and van Dyk E. E. 2004. Assessing the reliability and degradation of photovoltaic module performance parameters. IEEE Transactions on Reliability Tang W., and Davidson C. I. 2004. Erosion of limestone building surfaces caused by wind-driven rain: 2. Numerical modeling. Atmospheric Environment, 38(33), 5601-5609. Tominaga Y., Mochida A., Yoshie R., Kataoka H., Nozu T., Yoshikawa M., and Shirasawa T. 2008. AIJ guidelines for practical applications of CFD to pedestrian wind environment around buildings. Journal of Wind Engineering and Industrial Aerodynamics, 96(10-11), 1749-1761. van Mook F. J. R. 2002. Driving rain on building envelopes. PhD Thesis, Eindhoven University of Technology, Eindhoven, The Netherlands. Wieringa J. 1992. Updating the Davenport roughness classification. Journal of Wind Engineering and Industrial Aerodynamics, 41(1–3), 357-368. disturbed approach-flow wind profile due to the buildings in that direction. A measurement dataset suitable for CFD validation requires west-southwest wind direction. The WDR simulations show that for wind approaching from west, the surface wetting profiles are quite different for the cubes A and B. The catch ratio distribution for the cube B does not agree with the general assumption that WDR is approximately proportional to the product of wind speed and horizontal rainfall intensity. The recirculation region and the sheltering effect in front of the cube B leads to a decrease in catch ratio values at wind speed values higher than 5 m/s. Furthermore, the facade of the cube B has larger vertical wetting gradients. The assumption that the moisture source due to WDR is uniform across large parts of the facade may lead to large errors in further analysis. An accurate link between the surface wetting due to WDR and the building envelope performance using heat, air, moisture transport (BEHAM) models can be formed only after careful analysis of impinging droplet characteristics. 6 ACKNOWLEDGEMENTS The research was supported through the Swiss National Science Foundation (SNF) - Project no. 135510. 7 REFERENCES Abuku M., Janssen H., Poesen J., and Roels S. 2009. Impact, absorption and evaporation of raindrops at building facades. Building and Environment, 44(1), 113-124. Best A. C. 1950. The size distribution of raindrops. Quarterly Journal of the Royal Meteorological Society, 76(327), 1636. Blocken B., and Carmeliet J. 2002. Spatial and temporal distribution of driving rain on a low-rise building. Wind and Structures, 5(5), 441-462. Blocken B., and Carmeliet J. 2004. A review of wind-driven rain research in building science. Journal of Wind Engineering and Industrial Aerodynamics, 92(13), 10791130. Blocken B., and Carmeliet J. 2005. High-resolution winddriven rain measurements on a low-rise building experimental data for model development and model validation. Journal of Wind Engineering and Industrial Aerodynamics, 93(12), 905-928. Blocken B., and Carmeliet J. 2006a. The influence of the windblocking effect by a building on its wind-driven rain exposure. Journal of Wind Engineering and Industrial Aerodynamics, 94(2), 101-127. Blocken B., and Carmeliet J. 2006b. On the accuracy of winddriven rain measurements on buildings. Building and Environment, 41(12), 1798-1810. Blocken B., and Carmeliet J. 2007. Validation of CFD simulations of'wind-driven rain on a low-rise building facade. Building and Environment, 42(7), 2530-2548. 436 The use of vegetation for social housing renovations: A case study in the City of Palermo L. Pastore & R. Corrao University of Palermo, Department of Architecture, Palermo, Italy P. Heiselberg Aalborg University, Department of Civil Engineering, Aalborg, Denmark ABSTRACT: The paper shows the first results of a research carried out at the University of Palermo, which aims to indentify sustainable strategies for the renovation of social housing in the Mediterranean Basin, with focus on the use of vegetation as strategy for the enhancement of urban microclimate and thermal comfort. To achieve these goals the social housing complex of Medaglie d’Oro in the city of Palermo was chosen as case study in order to carry out some analyses for the assessment of outdoor comfort parameters of the area. By means of numerical simulations performed with the software ENVI-met, the actual state of the area was evaluated and four scenarios of renovation strategies through the integration with different vegetal species were proposed. The simulations show the effectiveness of the vegetation for the mitigation of the outside temperature and demonstrate that more accurate analyses on building indoor comfort can be performed taking into account the factors that influence its surrounding environment. 1 INTRODUCTION (Akbari 1992, Givoni 1998, Huang et al. 1987), especially in temperate and hot areas, such as the Mediterranean Basin, where energy consumptions are mostly due to the use of refrigerating systems (Margani 2010). Research on urban climatology demonstrated that evapotranspiration from trees and shrubs can reduce temperatures and buildings energy use for whole neighbourhoods and that vegetation brings, at the same time, other benefits including the reduction of air pollution and greenhouse gas emissions, the protection from harmful exposure to ultraviolet (UV) rays, the decrease of storm water runoff and a higher quality-of-life in general. Nevertheless, although indoor comfort and building energy performance simulations, through the use of proper software tools, have been acquiring more and more importance in the last years for sophisticated building energy assessments, they still tend to be disjointed from the elements that characterized the surrounding environment, such as the presence of vegetation, types of soil, the nearness of other buildings, their configuration and the albedo values of their surfaces. Starting from these assumptions, the research that is being conducted at the University of Palermo aims to identify efficient strategies for the recovery of social housing complexes in the Mediterranean Basin. The study intends to move towards two complementary directions: firstly, the analyses of the influence of vegetation, properly designed, on urban microclimate to enhance outdoor (and then indoor) thermal comfort and, secondly, the assessment of different solution to improve buildings energy performance taking mainly into account passive retrofit systems. In the present paper the results obtained from the first stage of the research are described. Traditional architecture produced a large repertory of buildings typologies derived from the adaptation to the available resources -materials, technologies, knowledge and competence- and deeply tied to the environment and the climatic context. The integration of vegetation with fabrics, for example, has always been a common architectural practise: whether simply low-tech way of providing shade or part of more sophisticated building components, plants were often considered as an integral design element because of the environmental benefits they offer. Since the second half of the XX century, because of the improvement of building techniques and the large availability of fossil fuels, traditional technologies for buildings comfort were generally disregarded and substituted by systems related to the engineering of single buildings components, acting most of the time independently one from the other. These design practises had as a consequence the diffusion of constructive practices characterized by the negation of a holistic approach at the basis of the project origination and where «…nature is held at bay in the interests of notions of “precision” of control and “efficiency” of operation…» and «…machines have taken over the environmental function…» (Hawkes 2006). This phenomenon, together with the increasing urbanization and industrialization, had led in the last decades to an exacerbation of the heat island, deeply affecting global energy costs and the quality of urban life. In the last decades several researches have demonstrated that effective ways of mitigating heat islands exist, and that vegetation is one of the simplest ways to cool our communities and save energy 437 Table 1. Input data for simulations in winter - Stage 1 Winter Day Feb 11th 2012 Feb 27th 2012 Wind speed 2.7 [m/s] 5.4 [m/s] Wind direction SW: 225 NNW: 357.5 Temperature Atm. 8.3°C 11.4°C Specific Humidity 4.49 [gH2O/kgAir] 4.2 [gH2O/kgAir] Relative Humidity 66% 50% Albedo Walls 0.22 0.22 Albedo Roofs 0.22 0.22 Shortwave adj. 0.75 0.75 2 SIMULATION METHODOLOGY The social housing complex Medaglie d’Oro of the city of Palermo was chosen as case study in order to assess to what extent vegetation and types of soil can affect the outdoor comfort of a micro urban area and, as a consequence, the users’ comfort inside the buildings located in that area. The research is being conducted according to the following steps: • Stage 1 - Meso-level simulation of an urban area of 470x740 m, enclosing the Medaglie d’Oro complex, in order to provide boundary conditions for the microclimate simulation within the residential area to be examined, in the absence of local measurements.x • Stage 2 - Simulation of an area of 110x375 m corresponding to the Medaglie d’Oro complex, starting from the microclimate data acquired through the simulation at stage 1, in order to evaluate the effects of vegetation on the near-buildings meteorology. • Stage 3 - Simulation at the apartment scale in order to assess the indoor thermal and visual comfort, starting from the microclimate data acquired through the simulation at stage 2, and proposal of some retrofit strategies at the building scale. The urban simulations were performed by means of the software ENVI-met which is a 3D microclimate model, based on the fundamental laws of fluid dynamics and thermodynamics, designed to simulate the surface-plant-air interactions in urban environment with a typical resolution of 0.5 to 10 m in space and 10 sec in time (Bruse 2013). The numerical model simulates aerodynamics, thermodynamics and the radiation balance in complex urban structures taking into account the location of the area, the sun position, the urban geometry, vegetation, type of soil and various construction materials. Figure 1. Aerial photo of Medaglie d’Oro complex. Figure 2. Layout of the residential complex with the identification of vegetation and soil typologies. tion of the area for the vehicular access and park. For this reason the site presents an extended use of asphalt and a scarce presence of vegetation (Fig. 2). The buildings are characterised by a reinforced concrete frame structure, visible from the outside, and external wall made by (from outside to inside) 13 cm pumice blocks separated from internal 8 cm perforate bricks by a 6 cm air cavity and finished with different thickness plaster layers on both sides. The apartment blocks present considerable alterations of their original configuration as proved by the closure of many loggias and the addition of several air conditioning units hung to the buildings façades (Fig.3). 2.1 Area of study Palermo (Sicily) is one of the southern cities of Italy and one of the warmest in Europe. Its Mediterranean climate is characterized by mild winters and hot and dry summers with an average annual ambient air temperature of 18.5 °C, and approximately 2530 hours of sunshine per year. The Medaglie d’Oro complex is located in a southwest peripheral area of the city that went through a progressive urbanization process starting from the ‘70s. The residential complex, designed in 1972 by B. Colajanni, is constituted by two rows of three sevenstorey apartment blocks, facing each other, and a three-storey building situated between them, occupying a total area of around 29,500 m2. Despite its dimension and configuration, the complex is lacking of open spaces for residents’ leisure and social activities (conversely foreseen in the original layout), due to the almost exclusive utiliza- Figure 3. The actual state of the residential blocks. These interventions demonstrate the inadequacy of the apartments to satisfy the contemporary users’ needs related to the necessity of increased minimum living spaces and indoor comfort levels. 438 energy fluxes, depending on locations, sometimes occurs using ENVI-met (Bruce 2012). For this reason, a feature supplied by the program was used to adjust the shortwave solar radiation calculated to the given situation. As recommended (Bruse 2012), 48 hours simulations were used to pass the initial transient time in order to obtain results as reliable as possible. By comparing the results obtained from the four 2.2 Weather data The city of Palermo has several weather monitoring stations displaced in its territory. For running the preliminary simulations with ENVI-met the data registered by the weather station installed at the University of Palermo (UniPa) were used, also considering its proximity to the Medaglie d’Oro area (c.a.1200 m). In particular, the measurements referring to the year 2012 were considered, taking into account the coldest and the warmest months of the year, respectively February (average temperature of 9,81°C) and August (average temperature of 27,7°C). In light of the weather data analyzed, for both months two different “sample days” were selected for the simulations. The choice was made taking into account the weather conditions in typical days with average values of temperature, wind speed and solar radiation and the variation of local climate when wind speed and solar radiation are above the monthly average values. Table 2. Input data for simulations in summer - Stage 1. Summer Day Aug 05th 2012 Aug 15th 2012 Wind speed 0.91 [m/s] 1.64 [m/s] Wind direction SSW: 212.5 SW: 225 Temperature Atm. 30.51°C 26.86°C Specific Humidity 8.15 [gH2O/kgAir] 16.4 [gH2O/kgAir] Relative Humidity 27.33% 68.41% Albedo Walls 0.22 0.22 Albedo Roofs 0.22 0.22 Shortwave adj. 0.82 0.85 simulations to the weather data registered at the University of Palermo, correspondence of values emerges in relation to temperature and direct solar radiation, while significant differences can be noticed for the wind speed values in correspondence to the four receptors (Figure 5). In fact, as show in Figure 6, wind speed progressively decreases in correspondence of buildings, although differences are of the order of around 1m/s. 2.3 Stage 1 - Preliminary simulations for the assessment of the input data In the absence of local measurements, a more detailed analysis on the trend of microclimatic conditions in the proximity of the residential lot was needed in order to evaluate the boundary conditions depending on the surrounding built environment and open spaces. For this purpose an area of 470x740 m, including part of an urban park in the north, open spaces in the south-south west sides and a denser urban area in the east side, was first considered for simulation. The ENVI-met model was built according to a 3D-nesting grid of 74x47x10 (x,y,z grid size = 10 m) and four receptors were set into the boundary of the Medaglie d’Oro area (Figure 4). Figure 4. The urban area considered for the preliminary simulations and position of the four receptors. The main input data entered are shown in Table 1 and Table 2. Over or underestimation of the solar Figure 5. Comparison between temperature, wind speed and direct radiation registered at UniPa and those generated for the four receptors. 439 Figure 6. Wind speed values performed by ENVI-met for August 15th. 2.4 Stage 2 - Simulations at the scale of Medaglie d’Oro complex For this stage of the research results obtained from the previous simulations were taken into account. In particular, average values were calculated by weighing up the results registered by the four receptors and the values registered at the weather station of the University of Palermo. Furthermore new input data were entered such as walls and roofs thermal transmittance. The new ENVI-met model was built according to a 3D-nesting grid of 117x21x10 (x,y,z grid size = 5 m). Five receptors (A, B, C, D, E) were taken into account for these simulations in order to extract exact values of temperature, wind speed and solar radiation: four receptors in correspondence of two apartments situated in two different buildings (two receptors for each apartment positioned on the two façades of the building) and one placed in the open space between the two buildings (Figure 7). Figure 8. Simulation models for the current situation and the four greening scenarios. Table 3. List of vegetation typologies species considered for the simulations. Vegetation Typology H (m) LAI (m2/m2) A Tree 10 4.6 B Tree 10 1.175 C Tree 6 4.6 D Tree 3 1.75 Sk Tree 15 6.5 Pl Tree 15 2.6 Bu Hedge 2 5.9 G Grass 0.05 0.5 Gr Grass 0.10 3 Figure 7. Individuation of the receptors in the ENVI-met model. In addition to the current situation, four different “greening scenarios” were simulated, considering progressive addiction of vegetation with different LAI - Leaf Area Index values (Figure 8). Hence, a list of vegetation typologies was provided in order to identify the existing plant in the site and those foreseen in the “greening scenarios” (Table 3). Input values for stage 2 are shown in Table 4 and Table 5. ENVI-met is capable to generate climate data at different heights. In this case, for receptors A, B, C and D the generated data refer to 7,5 m a.g.l. (above ground level representing the hight of a typical apartment while for receptor E a height of 1,5 m was considered, corresponding to the pedestrian zone. 440 Table 4. Input data for simulations in winter - Stage 2. Winter Day Feb 11th 2012 Feb 27th 2012 Wind speed 2.7 [m/s] 5.4 [m/s] Wind direction SW: 225 NNW: 357.5 Temperature Atm. 8.4°C 10.3°C Specific Humidity 5.3 [gH2O/kgAir] 5.4 [gH2O/kgAir] Relative Humidity 78% 76% Albedo Walls 0.22 0.22 Albedo Roofs 0.22 0.22 Heat Tran. Walls 0.827 [W/m2K] 0.827 [W/m2K] Heat Tran. Roofs 2.27 [W/m2K] 2.27 [W/m2K] Shortwave adj. 0.75 0.75 Table 5. Input data for simulations in summer - Stage 2. Summer Day Aug 05th 2012 Aug 15th 2012 Wind speed 0.88 [m/s] 1.51 [m/s] Wind direction SSW: 212.5 SW: 225 Temperature Atm. 29.88°C 26.86°C Specific Humidity 12.65 [gH2O/kgAir] 18.5[gH2O/kgAir] Relative Humidity 50.43% 68.41% Albedo Walls 0.22 0.22 Albedo Roofs 0.22 0.22 Heat Tran. Walls 0.827 [W/m2K] 0.827 [W/m2K] Heat Tran. Roofs 2.27 [W/m2K] 2.27 [W/m2K] Shortwave adj. 0.82 0.85 By comparing the data generated during the winter days, irrelevant differences are recorded as for wind speed and ambient temperature while in scenarios 2, 3 and 4, considerable reduction of direct radiation is registered due to the shade provided by the trees. Regarding the summer days data generated show analogous results for the two days. For all four receptors, wind speed does not present significant variation comparing the current situation with the proposed scenarios. Benefits derived from the presence of hedges and lawn of scenario 1 are almost imperceptible for receptors B and D. The same situation occurs for scenarios 2 unless for the shading effect of the trees foliage on the building façades. Also in scenarios 3 and 4, in which dense crown trees are used as shading elements, solar radiation is obviously very reduced in respect to the original layout. Relevant results are obtained in correspondence of receptors A and C, positioned near the buildings walls overlooking the internal area of the complex, and for receptor E, positioned in the open area. In general what emerges is that, comparing the scenarios, highest variations of temperature imputable to the presence of vegetation occur during the warmest hours of the day (1-3 P.M.) and during the night (11 P.M.- 4 A.M.). In particular, in scenario 4 the following situation occurs: • receptor A -southwest exposed- registered a decrease up to 1°C in the early morning and in the afternoon and up to 1,5 °C during the night; • receptor C -northeast exposed- registered a decrease up to 1,5°C in the early morning, up to 1°C in the morning-afternoon and up to 2,5 °C during the night; • receptor E registered a decrease up to 1,5°C in the early morning, up to 1°C in the morningafternoon and up to 3°C during the night. Figure 9. Temperature, wind speed and direct radiation generated for receptor C in all scenarios. Figure 10. Temperature, wind speed and direct radiation generated for receptor E in all scenarios. 441 This is attributable to the fact that the greater trees densification of the complex common area has a higher influence on the microclimate as shown in Figure 11. Figure 11. Comparison between the current scenario and the scenario 4 of the temperature values on Aug 5th, 11 AM. 3 CONCLUSION Data generated in this first stage of the research by means of the software ENVI-met demonstrate how effective the adoption of vegetation can be for the control of the outdoor comfort levels during the design and the renovation process of micro urban areas. In fact the results of the measurements gave important information about the microclimatic differences among the five scenarios analysed (current situation and four “greening scenarios”) related to the social housing complex Medaglie d’Oro and prove that the cooling effect of the area is higher in correspondence of the internal common area of the complex where vegetation is denser. Hence, this part of the research draws attention to the importance of deeper approaches for high accuracy investigations on real urban and building scale thermal comfort levels. Starting from generated meteorology of this first stage, further analyses are being held at the building and apartment scale to assess the effect of shading, radiant interactions and evapotranspiration on the indoor comfort levels and to evaluate how these strategies can be combined with other retrofit interventions to further reduce the buildings energy consumptions. 4 REFERENCES Akbari H. 1992. Cooling our Communities. A Guidebook on Tree Planting and Light-Colored Surfacing, Washington D.C.: United States Environmental Protection Agency. Akbari H. 2002.Shade trees reduce building energy use and CO2 emissions from power plants, Environmental Pollution 116: 119-126 Bruse M., and Fleer H. 1998. Simulating surface-plant-air interactions inside urban environments with a three dimensional numerical model. Environmental Modelling and Software 13, n. 3: 373-384. Bruse M. 2012. ENVI-met online manual. Available on www.envi-met.com. Accessed on February 2013. Fahmy M., Sharples S., and Eltraposi A. 2009, Dual stage simulations to study the microclimatic effects of trees on ther- 442 mal comfort in a residential building, Cairo, Egypt. 11th International IBPSA Conference, Glasgow, Scotland, July 2730, 2009. Givoni B. 1998. Climate consideration in urban building design. New York: Van Nostrand Reinhold. Grosso M. 2011. Il raffrescamento passivo degli edifici in zone a clima temperato. Rimini: Maggioli Editore. Hawkes D. 2006. The Selective Environment: Environmental Design and Cultural Identity, in Bay J. and Ong B. L., Tropical Sustainable architecture. Social and Environmental Dimensions: 29-44. Oxford: Elsevier Ltd. Huang L., Akbari H., Taha H., and Rosenfeld A.H. 1987. The Potential of Vegetation in Reducing Summer Cooling Loads in Residential Buildings. Journal of Climate and Applied Meteorology 26:1103-1116. Jesionek K., and Bruse M. 2003. Impacts of vegetation on the microclimate: Modelling standardized building structures with different greening level. ICUC5, Lodz 1-5- September 2003 Margani G. 2010. L’edificio passivo nel clima mediterraneo, Costruire in laterizio 141: 46-49. Impacts of building performance monitoring on integrated energy management K

RELATED PAPERS

Naunyn-Schmiedeberg's Archives of Pharmacology

Adenosine 3?,5?-monophosphate in cultured neuroblastoma cells: Effect of adenosine, phosphodiesterase inhibitors and benzazepines

1973 •

Joachim Schultz

View PDF

Anais do XX Congresso Brasileiro de Engenharia Química

Estudo De Simulação Da Inversão Água/Óleo Com a Utilização Do Método Level Set

2015 •

Marintho Quadri

View PDF

Geological Quarterly

Analiza palinologiczna i diatomologiczna osadów interglacjału mazowieckiego w Adamówce (Kotlina Sandomierska)

2013 •

Krzysztof Bińka

View PDF

Vie sociale

Inclusion scolaire et action publique, entre contradictions et inachèvement

2015 •

Mazereau philippe

View PDF

Revista Cubana De Anestesiologia Y Reanimacion

Dolor crónico o persistente

2013 •

lisbeth lesly chavez alvarez

View PDF

Jurnal Sain Peternakan Indonesia

Pengaruh Pemberian Pupuk Feses Sapi dengan Dosis yang Berbeda Terhadap Pertumbuhan dan Produksi Rumput Odot (Pennisetum purpureum Cv.Mott) di Kabupaten Kepahiang

2019 •

suteky tatik

View PDF
The Psychological Implications of Creative Activities: An investigation into how painting affects stress levels

2010 •

Celeste Combrinck

View PDF

Diabetes

Early Endothelial Dysfunction Severely Impairs Skin Blood Flow Response to Local Pressure Application in Streptozotocin-Induced Diabetic Mice

2004 •

Berengere Fromy

View PDF

Energy Conversion and Management

A semi-theoretical model for energy efficiency assessment of air source heat pump systems

2021 •

wenyuan zhao

View PDF

RihumSo : Revista de investigación del Departamento de Humanidades y Ciencias Sociales

Las representaciones sociales presentes en novelas televisivas sobre la gestación por sustitución en Argentina. El caso de Pequeña Victoria

2020 •

Elizabeth Ormart

View PDF

arXiv (Cornell University)

Quantum Inverse Semigroups

2022 •

Eliezer Batista

View PDF

Arquivo Maaravi: Revista Digital de Estudos Judaicos da UFMG

Desterritorializando e reterritorializando um shtetl nos trópicos. Leitura do romance A guerra no Bom Fim, de Moacyr Scliar

2013 •

Soraya Lani

View PDF

Frontiers in Psychiatry

Violence and Cannabis Use: A Focused Review of a Forgotten Aspect in the Era of Liberalizing Cannabis

2020 •

Stéphane Potvin

View PDF

Cadernos Saúde Coletiva

A necessidade de cuidado na percepção de pessoas idosas em processo de fragilização

2022 •

Karla Giacomin

View PDF

Extended Abstracts of the 2011 International Conference on Solid State Devices and Materials

Effects of Excimer Laser annealing of Oxide Semiconductor Films

2011 •

Ryoichi Ishihara

View PDF

Studia Warmińskie

„Mirabilia animalia” w „Nocach attyckich” Aulusa Gelliusza. Wybrane przykłady

2017 •

Maria Piechocka-Kłos

View PDF

Archivos de Zootecnia

Efecto de tres dietas en el crecimiento de crías de cocodrilo americano Crocodylus acutus

2012 •

Olimpia Farnés

View PDF

Versión. Estudios de Comunicación y Política

Activismo social, TIC y democratización en dos escenarios de disputa

2013 •

Laura Beatriz Montes de Oca Barrera

View PDF

Cambridge University Press eBooks

Marriage Registrars, Same-Sex Relationships, and Religious Discrimination in the European Court of Human Rights

2018 •

Christopher McCrudden

View PDF

PeerJ

Large-scale differences in diversity and functional adaptations of prokaryotic communities from conserved and anthropogenically impacted mangrove sediments in a tropical estuary

2021 •

Fábio Chinalia

View PDF

Revista de Nutrição

Dietary reference intakes: aplicabilidade das tabelas em estudos nutricionais

2006 •

Renata Maria Padovani

View PDF

In: Bucholc, M. (eds) Established-Outsiders Relations in Poland: Reconfiguring Elias and Scotson. Palgrave Studies on Norbert Elias. Palgrave Macmillan, Cham. https://doi.org/10.1007/978-3-031-49523-6_3

Anti-Gender Preaching on IKEA Intranet: Catholic Church as a Besieged Fortress (2019–2021)

2024 •

Agnieszka Laddach

View PDF

Frontiers in Immunology

Editorial: HLA and KIR Diversity and Polymorphisms: Emerging Concepts

2021 •

Ramit Mehr

View PDF

BC Studies: The British Columbian Quarterly

Reframing Nikkei Histories: Complicating Existing Narratives

2016 •

Andrea Geiger

View PDF
Swift, Jonathan: A Modest Proposal for Preventing the Children of Poor People of Ireland from Being a Burden to Their Parents or the Country, and for Making Them Beneficial to the Public

2020 •

Emer O'Sullivan

View PDF

RELATED TOPICS

Sustainable Building DesignBuilding Physics