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
rk
∆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