Sunday

How to Navigate the 

Technical Sessions 

There are four primary resources to help you 

understand and navigate the Technical Sessions: 

This Technical Session listing, which provides the 

most detailed information. The listing is presented 

chronologically by day/time, showing each session 

and the papers/abstracts/authors within each 

session. 

The Author and Session indices provide 

cross-reference assistance (pages 468-505). 

The floor plans on pages 42 and 43 show you where 

technical session tracks are located. 

Quickest Way to Find Your Own Session 

Use the Author Index (page 468) — the session code 

for your presentation will be shown along with the room 

location. You can also refer to the full session listing for 

the room location of your session. 

The Session Codes 

TA01 

The day of 

the week 

Time Blocks 

Sunday - Tuesday 

A — 8:00am - 9:30am 

B — 11:00am - 12:30pm 

C — 1:30pm - 3:00pm 

D — 4:30pm - 6:00pm 

Wednesday 

A — 8:00am - 9:30am 

B — 11:00am - 12:30pm 

C — 1:30pm - 3:00pm 

D — 3:20pm - 4:50pm 

Room number. Room locations are 

also indicated in the listing for each 

session. 

Time Block. Matches the time 

blocks shown in the Program 

Schedule. 

Room Locations /Tracks 

All tracks and technical sessions will be held in the 

Phoenix Convention Center, and Hyatt Hotel. 

Room numbers are shown on the Track Schedule and 

in the technical session listing. 

55 

Sunday, 8:00am - 9:30am 

■ SA01 

T ECHNICAL S ESSIONS 

01- West 101- CC 

GRASP and Tabu Search 

Contributed Session 

Chair: Celso Ribeiro, Universidade Federal Fluminense, Institute of 

Computing, Niteroi, RJ, 22410240, Brazil, celso@ic.uff.br 

1 - An Improved GRASP with Path Relinking for 

Commercial Districting 

Roger Rios, Professor, Universidad Autonoma de Nuevo Leon, 

Graduate Program in Systems Engineering, AP 111-F, Cd. 

Universitaria, San Nicolas de los Garza, NL, 66450, Mexico, 

roger@yalma.fime.uanl.mx, Hugo Escalante 

A commercial districting problem seeking to minimize territory dispersion under 

connectivity and multiple balancing constraints is addressed. A GRASP with path 

relinking (PR) is proposed. Both static and dynamic PR strategies are developed 

and assessed within a GRASP framework. Empirical work shows the effectiveness 

of the proposed method outperforming the best existing method in terms of 

solution quality. 

2 - A Heuristic for a Real-world VRP Arising in a Bottled-beverage 

Distribution Company 

Andrés Castrillón-Escobar, Student, UANL, Salvador Novo 105, 

Monterrey, Me, 64630, Mexico, cuatex1@hotmail.com, 

Roger Rios, J. Fabián Lopèz-Perèz 

A real-world problem arising from the bottled-beverage distribution industry is 

addressed. The problem is modeled as a multi-depot pick-up and delivery 

capacitated VRP with additional side constraints such as time windows, nonhomogeneous 

fleet, split delivery, and dock capacity, among others. To solve this 

problem, a GRASP is proposed. Empirical evidence on real-world very large 

instance is presented showing the proposed procedure improves current practice 

in terms of solution quality. 

3 - Redundancy Allocation Problem of Complex System using 

Metaheuristics: A Computational Study 

Jae-Hwan Kim, Professor, Korea Maritime University, Yeongdo-Gu 

Dongsam-Dong, Busan, Korea, Republic of, jhkim@hhu.ac.kr, 

Kil-Woong Jang 

Redundancy allocation problem (RAP) is to select redundancy levels at each 

subsystem in order to maximize the system reliability under several resource 

constraints(e.g., cost). The RAP is a well-known NP-hard combinatorial 

optimization problem. We present some metaheuristics (tabu search, etc.) for its 

solution. A computational study for the composed test problems is also analyzed. 

4 - Probabilistic Stopping Rules for GRASP Heuristics 

Celso Ribeiro, Universidade Federal Fluminense, Institute of 

Computing, Niteroi, RJ, 22410240, Brazil, celso@ic.uff.br, 

Reinaldo Castro Souza, Isabel Rosseti 

A drawback of most metaheuristics is the absence of effective stopping rules. We 

show experimentally that solution values obtained by GRASP fit a truncated 

Normal distribution and use this approximation to estimate the number of 

iterations needed to improve the best known solution. This estimation is used to 

implement stopping rules based on the trade-off between solution quality and 

computation time. This strategy is validated by results obtained for some GRASP 

heuristics. 

5 - Evaluation of Run-length Distribution for CUSUM Charts under 

Gamma Distributions 

Wenpo Huang, University of Macau, Faculty of Business 

Administration, Block1, B408, FBA, Taipa, Macau, China, 

wenpohuang@umac.mo, Lianjie Shu, Wei Jiang, Kwok Tsui 

Numerical evaluation of run-length distributions of CUSUM charts under normal 

distributions has received considerable attention. However, accurate 

approximation of run-length distributions under non-normal or skewed 

distributions is challenging but has been previously overlooked. This paper 

provides a fast and accurate algorithm based on the piecewise collocation method 

for computing the run-length distribution of CUSUM charts under skewed 

distributions such as gamma distributions. It is shown that the piecewise 

collocation method can provide a more robust approximation of the run-length 

distribution than other existing methods such as the Gaussian quadrature based 

approach, especially when the process distribution is heavily skewed. Some 

computational aspects including alternative formulation based on matrix 

decomposition and geometric approximation of run-length distribution are 

discussed. Design guidelines of such CUSUM chart are also provided.

SA02 

■ SA02 

02- West 102 A- CC 

Decision Analysis Applications in the Energy Industry 

Sponsor: Decision Analysis 

Sponsored Session 

Chair: Mazen Skaf, Partner & Managing Director, Strategic Decisions 

Group, 745 Emerson St, Palo Alto, CA, 94301, United States of 

America, MSkaf@sdg.com 

1 - A Decision Quality Approach to Natural Gas Resource 

Uncertainty 

Robert Stibolt, Managing Director, Galway Group, L.P., 

3050 Post Oak Blvd., Suite 1300, Houston, TX, 77056, 

United States of America, rstibolt@galwaygroup.com, Mazen Skaf 

The shale gas revolution started in North America has completely changed the 

dynamic of world natural gas markets. Implications are far-reaching, exciting, but 

also highly uncertain. For example, the shape of long-run supply curves depend 

on technology, geology, and geopolitics. This paper addresses these questions 

within a framework that embraces uncertainty while eschewing the illusion of 

precise prediction — a key to developing natural gas strategies that are robust. 

2 - Hydrocarbon Resource Allocation Strategy for Industry 

Development in an Energy-rich Nation 

Mazen Skaf, Partner & Managing Director, Strategic Decisions 

Group, 745 Emerson St., Palo Alto, CA, 94301, 

United States of America, MSkaf@sdg.com, Thomas Seyller 

We present a DA-based process and system for optimizing the allocation of 

supplies of a hydrocarbon resource across proposed industrial projects taking into 

account: i) uncertainty in future supply and demand, ii) various economic, social, 

and environmental value measures, and iii) existing commitments for current 

domestic uses and industries. The value-focused approach resulted in a paradigm 

shift in the industrial development strategy as well as in the future focus of E&P 

activities. 

3 - Solar Value Chain Decision Analysis in Oil and Gas Rich 

Countries 

Thomas Seyller, Sr. Engagement Manager, Strategic Decisions 

Group, 745 Emerson St., Palo Alto, CA, 94301, 

United States of America, tseyller@sdg.com, Dan Cornew 

We will explore how decision analysis can help oil and gas rich countries develop 

strategies around solar energy that maximize the total benefit to the nation – 

including economic value and costs, social benefits such as job creation, and 

environmental impacts. Strategies cover a wide range of possible incentives and 

mandates throughout the solar value chain, from manufacturing to deployment. 

■ SA03 

03- West 102 B- CC 

Risk and Uncertainty in Public Policy and Defense 



Chair: Jay Simon, Naval Postgraduate School, Defense Resources 

Management Institute, 699 Dyer Rd, Monterey, CA, 93943, 

United States of America, jrsimon@nps.edu 

1 - Uncertainty Model for Rapid Updating of Target Positions 

Eva Regnier, Naval Postgraduate School, 699 Dyer Road, 

Monterey, CA, 93943, United States of America, 

eregnier@nps.edu, Dashi Singham 

In many security contexts, forces must allocate their assets based on incomplete 

information about targets’ positions. They frequently receive human intelligence 

describing past or future events related to target locations. We present an 

uncertainty model that allows for rapid updating, conditioning on past or future 

events. 

2 - Ranking FDA Risks: The Medical Devices Case 

Robin Keller, University of California-Irvine, Irvine, CA, 

United States of America, lrkeller@uci.edu, Yitong Wang 

This talk focuses on a medical devices case study. Keller served as a decision 

analyst on the committee charged with suggesting how the FDA can compare 

risks across product categories (foods, cosmetics, drugs, etc.). The NRC book 

proposed a multiple objective decision analytic approach to ranking risks: “A Risk- 

Characterization Framework for Decision-Making at the Food and Drug 

Administration” is at http://dels.nas.edu/Report/Risk-Characterization- 

Framework-Decision/13156. 

INFORMS Phoenix – 2012 

56 

3 - Probability Segmenting: Managing Uncertainty in Military Drafts 

Jay Simon, Naval Postgraduate School, Defense Resources 

Management Institute, 699 Dyer Rd, Monterey, CA, 93943, 

United States of America, jrsimon@nps.edu, Jonathan Lipow 

When a nation chooses to enact a draft, this presents an interesting decision 

problem for those individuals eligible to be drafted: they can take actions which, 

at some cost, will reduce the probability of being selected. It is in society’s best 

interest to minimize these costly actions. We discuss “probability segmenting,” the 

idea of randomly assigning different probabilities to different individuals, as a 

possible method for improving the outcome at a societal level. 

■ SA04 

04- West 102 C- CC 

Advances in Decision Analysis 


Chair: Steven Kimbrough, Professor, University of Pennsylvania, 

3730 Walnut St., Philadelphia, PA, 19104, United States of America, 

kimbrough@wharton.upenn.edu 

1 - Extending the Multiscale Decision-making Model: From Two to 

Many Decision Alternatives 

Swathi Sudhaakar, Graduate Student, Virginia Tech, 116 Durham 

Hall, MC 0118, Blacksburg, VA, 24061, United States of America, 

swathips@vt.edu, Christian Wernz 

We generalize the multiscale decision making model by extending the agents’ 

action space from two to N decision alternatives. Multiscale decision theory 

(MSDT) models the interaction and decision interdependencies of agents in large 

enterprise systems. With our results, MSDT can now be applied to a wider range 

of applications and provides organizational and system designers with insights on 

how the number of decision alternatives affects agents’ decision processes. 

2 - On the Axiomatization of the Satiation and Habit Formation 

Utility Models 

Ying He, PhD Candidate, University of Texas at Austin, 

Department Information, Risk & Operational Mg, 1 University 

Station, B6000, Austin, TX, 78712, United States of America, 

ying.he08irom@utexas.edu, James Dyer, John Butler 

We propose a preference condition called shifted difference independence to 

axiomatize a general habit formation and satiation (GHS) model. Since the GHS 

model can be reduced to either a general satiation model (GSa) or a general habit 

formation model (GHa), our theory also provides approaches to axiomatize both 

the GSa and the GHa model. By adding extra preference conditions, we 

axiomatize a GHS model with a linear habit function and a linear satiation 

function. 

3 - Strategic Experimentation under Exclusivity 

Stefan Rampertshammer, Duke University, Fuqua School of 

Business, 100 Fuqua Drive, Durham, NC, 27516, 

United States of America, Stefan.Rampertshammer@duke.edu 

We study a model of strategic experimentation with exponential bandits in which 

we attach a prize and a level of exclusivity to the first success. We determine the 

socially optimal policy, examine equilibrium behavior, and show that the efficient 

outcome can be recovered in dominant strategy through a combination of 

exclusivity and prizes. 

4 - Solution Pluralism 

Steven Kimbrough, Professor, University of Pennsylvania, 3730 

Walnut St., Philadelphia, PA, 19104, United States of America, 

kimbrough@wharton.upenn.edu, Frederic Murphy 

Solution pluralism involves generating a plurality of optimal and near-optimal 

solutions for use by stakeholders. Modern heuristics and ever-advancing 

computational powers make solution pluralism a practicable approach to explore 

for near optimal solutions as a form of sensitivity analysis and multiple candidate 

solutions when the problem has multiple objectives. We discuss the concept and 

principles of solution pluralism and illustrate it with a real world application to 

electoral districting.

■ SA05 

05- West 103 A- CC 

Journal of Quality Technology Invited Papers 

Sponsor: Quality, Statistics and Reliability 


Chair: Daniel Apley, NorthWestern University, 2145 Sheridan Road, 

Evanston, IL, 60208, United States of America, 

apley@northwestern.edu 

1 - Bayesian Modeling and Optimization of Functional Responses 

Affected by Noise Factors 

Enrique del Castillo, Distinguished Professor, Department of 

Industrial & Manufacturing Engineering & Department of 

Statistics, Penn State University, University Park, PA, United States 

of America, exd13@psu.edu, Bianca M. Colosimo, 

Hussam Alshraideh 

We present a Bayesian predictive modeling approach for functional response 

systems. The goal is to optimize the shape, or profile, of the functional response 

under a robust parameter design scenario in which there are controllable and 

noise factors. The method is illustrated with real examples and model building 

and diagnostics aspects of the assumed mixed effects model are discussed. 

Alternative modeling strategies beyond those in the JQT paper (2012, 44(2)) will 

be presented. 

2 - Robust Leak Tests for Transmission Systems using Nonlinear 

Mixed-Effect Models 

Kamran Paynabar, University of Michigan, Ann Arbor, MI, 

United States of America, kamip@umich.edu, Judy Jin, 

John Agapiou, Paula Deeds 

Leakage of the transmission fluid or oil in powertrain systems can cause engine 

overheating and/or permanent damages. The inspection results of the amount of 

leakage at given testing times are sensitive to the tested part’s temperature, which 

varies from part to part but was not previously incorporated in the leak testing 

systems. The objective of this paper is to develop a robust leak testing system 

using nonlinear mixed-effect models, which is insensitive to the part temperature 

variations. 

3 - A Variable-Selection-based Multivariate EWMA Chart for 

Process Monitoring and Diagnosis 

Kaibo Wang, Associate Professor, Tsinghua University, 

Department of Industrial Engineering, Beijing, China, 

kbwang@tsinghua.edu.cn, Wei Jiang, Fugee Tsung 

Fault detection and root cause identification are both important tasks in 

Multivariate Statistical Process Control. This paper proposes a Variable-Selectionbased 

MEWMA chart. The new chart first locates potentially out-of-control 

variables via variable selection, and then deploys such information in the 

monitoring statistics with the reduction in dimensionality providing increased 

sensitivity to out-of-control conditions. Its performance is studied via numerical 

simulations and real examples. 

■ SA06 

06- West 103 B- CC 

Efficient Learning in Stochastic Optimization 

Sponsor: Simulation 


Chair: Ilya Ryzhov, University of Maryland, 4322 Van Munching Hall, 

College Park, MD, 21044, United States of America, 

iryzhov@rhsmith.umd.edu 

1 - Dynamic Pricing under a General Parametric Choice Model 

Paat Rusmevichientong, Associate Professor, University of 

Southern California, Marshall School of Business, 

rusmevic@marshall.usc.edu, Josef Broder 

We consider a stylized dynamic pricing model in which a monopolist prices a 

product to a sequence of customers, who independently make purchasing 

decisions based on the price offered according to a general parametric choice 

model. The parameters of the model are unknown to th 


57 

SA07 

e seller, whose objective is to determine a pricing policy that minimizes the regret. 

We characterize the regret profile for this problem. 

2 - Kernelized Approximate Linear Programming 

Vivek Farias, Massachusetts Institute of Technology, 

77 Massachusetts Avenue, Cambridge, MA, United States of 

America, vivekf@mit.edu, Ciamac Moallemi, Nikhil Bhat 

We present a practical, non-parametric ADP algorithm that enjoys graceful, 

dimension-independent approximation and sample complexity guarantees by 

‘kernelizing’ a recent mathematical program for ADP (the ‘smoothed’ 

approximate LP). A computational study on a controlled queueing network 

shows that our procedure is competitive with state of the art parametric ADP 

approaches that employ carefully tailored approximation architectures. 

3 - Sequential Filtering of Glaucoma Progression 

Jonathan Helm, Assistant Professor, Indiana University, Kelley 

School of Business, ODT, 1309 East Tenth Street, Bloomington, IN, 

47405, United States of America, jhelm@umich.edu, Mariel 

Lavieri, Mark Van Oyen, Joshua Stein, MD 

We develop novel filtering algorithms for monitoring the disease progression of 

glaucoma patients. The models are validated using patient data from two multicenter 

clinical trials. 

4 - Ranking and Selection with Unknown Correlation Structures 

Huashuai Qu, University of Maryland, College Park, MD, 

United States of America, huashuai@math.umd.edu, Ilya Ryzhov, 

Michael Fu 

We create the first computationally tractable Bayesian statistical model for 

learning unknown correlations among alternatives in fully sequential ranking and 

selection. We can learn the unknown values and unknown correlations 

simultaneously with the model. A Bayesian procedure is derived to allocate 

simulations based on the value of information. We test the model and algorithm 

in a simulation study motivated by the problem of optimal wind farm placement, 

and obtain encouraging empirical results. 

■ SA07 

07- West 104 A- CC 

Joint Session DM/HAS: Data Mining in Health Care 

Sponsor: Data Mining & Health Applications Society 


Chair: Shengfan Zhang, Assistant Professor, University of Arkansas, 

4207 Bell Engineering Center, Fayetteville, AR, 72701, 

United States of America, shengfan@mail.uark.edu 

1 - Analyzing Human Microbiome Data 

Paul Brooks, Assistant Professor, Department of Statistical Sciences 

and Operations Research, Virginia Commonwealth University, 

P.O. Box 843083, Richmond, VA, 23284, United States of America, 

jpbrooks@vcu.edu, Microbiome Consortium 

The human microbiome is the community of microbes that populate habitats on 

and around various body sites. Advances in high-throughput sequencing facilitate 

investigations into the impact of the microbiome on physiology and disease. 

Microbiome data presents unique challenges for analysis. In this talk, we present 

methods for analysis that seek to address some of these challenges. 

2 - Outcome and State Transition Modeling for Adaptive 

Interdisciplinary Pain Management 

Aera LeBoulluec, PhD Student/GRA, University of Texas- 

Arlington, 2205 Hunter Place Lane, Arlington, TX, 76006, 

United States of America, aera.leboulluec@mavs.uta.edu, Li Zeng, 

Victoria Chen 

In this research, we seek adaptive treatment strategies for interdisciplinary pain 

management using data from the Eugene McDermott Center for Pain 

Management at the University of Texas Southwestern Medical Center at Dallas. 

An adaptive dynamic programming approach is formulated to improve current 

and future pain outcomes. Outcome and state transition modeling handle the 

problem of endogeneity via an inverse probability of treatment weighting 

approach. 

3 - Machine Learning with Operational Costs 

Cynthia Rudin, Assistant Professor, Massachusetts Institute of 

Technology, 77 Massachusetts Avenue, Cambridge, MA, 

United States of America, rudin@mit.edu, Theja Tulabandhula 

We aim to merge theway estimation algorithms are designed with how they are 

used for a subsequent task. We consider the operational cost of carrying out a 

policy, based on a predictive model. The operational cost becomes a regularization 

term in the learning algorithm’s objective, allowing either an optimistic or 

pessimistic view of possible costs. Limiting the operational cost reduces the 

hypothesis space, and can thus improve generalization. We bound the complexity 

of such hypothesis spaces.

SA08 

4 - Predictive Modeling of Glycosylation Modulation Dynamics in 

Cardiac Electrical Signaling 

Hui Yang, Assistant Professor, University of South Florida, 

Tampa, FL, 33620, United States of America, huiyang@usf.edu 

The cardiac action potential is produced by the orchestrated functions of ion 

channel dynamics. This investigation is aimed at modeling the variations of 

cardiac electrical signaling due to remodeling of a K+ channels. This multi-scale 

modeling investigation reveals novel mechanisms of hERG channel modulation 

by regulated glycosylation that also impact cardiac myocyte and tissue functions. 

It can potentially lead to new pharmaceutical treatments and drug designs for 

cardiac arrhythmia. 

■ SA08 

08- West 104 B- CC 

Inventory Management for Global Health 

Sponsor: Public Programs, Service and Needs 


Chair: Jeremie Gallien, London Business School, Regent’s Park, 

London, NW14SA, United Kingdom, jgallien@london.edu 

1 - Improving the Public Distribution of Essential Medicines in 

Sub-Saharan Africa: The Case of Zambia 

Zachary Leung, Massachusetts Institute of Technology, 

77 Massachusetts Ave., Cambridge, MA, United States of America, 

zacleung@mit.edu, Jeremie Gallien, Prashant Yadav 

Despite remarkable and successful improvements efforts by the government and 

its partners, the current public distribution system of essential medical drugs in 

Zambia still results in low availability to patients relative to private sector 

standards. We present an alternative design involving mobile devices and 

optimization and evaluate this proposal via a simulation model built with field 

data. Our results suggest that this proposal would improve drug availability and 

reduce inventory costs. 

2 - Malaria Treatment Distribution in Developing World Health 

Systems and Application to Malawi 

Hoda Parvin, Research Analyst, CNA, 4825 Mark Center Dr., 

Alexandria, VA, 22311, United States of America, 

parvinh@cna.org, Shervin AhmadBeygi, Mark Van Oyen, 

Peter Larson, Jonathan Helm 

We present stochastic models to address malaria treatment distribution in 

developing world under demand uncertainty. We first analyze a two-stage 

stochastic programming approach to make aggregate-level nation-wide decisions. 

We then present a Markov decision model to develop operational-level 

transshipment strategies within a cluster of clinics. 

3 - Public Health Impact of the Global Fund’s Performance-based 

Financing Process: A Queueing Analysis 

Iva Rashkova, London Business School, London, United Kingdom, 

irashkova.phd2009@london.edu, Jeremie Gallien, Prashant Yadav 

The Global Fund is the largest financier of programs against HIV, Malaria and 

Tuberculosis. We study its disbursement process through (i) an econometric 

analysis of disbursement and procurement data between 2003 and 2012; and (ii) 

a queueing model predicting national stock levels of medicines. This model 

quantifies the impact of the current process on central drug availability in 48 

African countries and potential interventions that include bridge financing and an 

international buffer stock. 

4 - Inventory Management in Humanitarian Supply Chains: 

The Role of Schedules and Uncertainty in Funding 

Karthik V. Natarajan, University of North Carolina, Chapel Hill, 

NC, United States of America, karthik_natarajan@unc.edu, 

Jayashankar M. Swaminathan 

Motivated by the ready-to-use therapeutic food (RUTF) supply chain in Africa, we 

study the problem of managing inventory of a nutritional product in the presence 

of variable funding constraints. We derive the structure of the optimal inventory 

and allocation policy and computationally analyze the impact of different funding 

schedules on the operating costs andwaiting times in the system. 


58 

■ SA09 

09- West 105 A- CC 

Evolutionary Multi-Criterion Optimization - I 

Sponsor: Multiple Criteria Decision Making 


Chair: Kalyanmoy Deb, Professor, Indian Institute of Technology 

Kanpur, Department of Mechanical Engineering, Kanpur, UP, 208016, 

India, deb@iitk.ac.in 

1 - A Decentralized Bicriteria Timeshare Exchange Algorithm 

Bahriye Cesaret, PhD Student, Jindal School of Management, 

University of Texas at Dallas, 800 West Campbell Road, 

Richardson, TX, 75080, United States of America, 

bahriye.cesaret@utdallas.edu, Milind Dawande, 

Tharanga Rajapakshe 

Timeshare Exchange refers to the trading of vacation timeshare weeks among 

owners, so that they can interchange their respective vacation homes and thereby 

experience new destinations. We consider two objectives to capture the notions of 

“efficiency” and “fairness” in an exchange solution. Our main contributions 

include (i) a structural analysis of the resulting bicriteria timeshare exchange 

problem and (ii) a polynomial-time, decentralized algorithm for “good” bicriteria 

solutions. 

2 - Cooperative Surrogate Models Improving Multi-objective 

Evolutionary Algorithms 

Saúl Zapotecas Martìnez, CINVESTAV-IPN, Mexico, D.F., Mexico, 

saul.zapotecas@gmail.com, Carlos A. Coello Coello 

We present a multi-objective evolutionary algorithm (MOEA) assisted by two 

different meta-models: a local and a global one. The global model is constructed 

by using different surrogate models. The local model is generated by using the 

solutions in the current population of the MOEA. The training set for each model 

is updated along the search of the MOEA. The two different predicted values of 

the functions provided by the meta-models are used to estimate the final value of 

each function. 

3 - Diversity Enhancing Evolutionary Multiobjective Search 

Brian Piper, North Carolina State University, 2500 Stinson Drive, 

Raleigh, NC, United States of America, bepiper@ncsu.edu, 

Ranji Ranjithan, Hana Chmielewski 

Besides obtaining Pareto-optimal solutions, decision diversity is an important 

consideration in real-world multiobjective problems. Diversity can be defined as 

both the spread of solutions in the decision space and degree to which alternative 

solutions are available. We present algorithms for obtaining diverse Paretooptimal 

sets and compare results for several test problems to demonstrate the 

effectiveness of the algorithms. 

4 - Satisfying Multiple Objectives in Infrastructure 

Management Planning 

Hana Chmielewski, North Carolina State University, Campus Box 

7908, Raleigh, NC, United States of America, htchmiel@ncsu.edu, 

Ranji Ranjithan 

Evolutionary algorithms lend themselves to solving cross-disciplinary problems 

with multiple objectives. In infrastructure planning, for example, it may be useful 

to consider the resource and performance objectives of engineers, land-use 

planners, and emergency managers. An algorithm that offers maximally different 

non-inferior management plans by preserving the diversity of solutions in the 

decision space is applied to aid decision-making inwastewater treatment planning. 

■ SA10 

10- West 105 B- CC 

Risk in Stochastic Programming 

Sponsor: Optimization/Stochastic Programming 


Chair: Sitki Gulten, PhD Candidate, Rutgers University, Department of 

Management Science & Infor, 1 Washington Park, Newark, NJ, 07102, 

United States of America, sgulten@rutgers.edu 

1 - Two-stage Portfolio Optimization with Higher-order Conditional 

Measures of Risk 

Sitki Gulten, PhD Candidate, Rutgers University, Department of 

Management Science & Infor, 1 Washington Park, Newark, NJ, 

07102, United States of America, sgulten@rutgers.edu, 

Andrzej Ruszczynski 

Optimization of a portfolio with an option to rebalance over multiple periods 

from is a major problem in modern portfolio theory. In this research, different 

forward and backward scenario tree generation techniques are applied on 

multivariate GARCH-generated scenarios to construct scenario trees. Next, mean-

semideviation risk function with higher-orders is used to formulate the two-stage 

portfolio problem in which there are two periods and an option to rebalance 

between those two periods. 

2 - Complexity of Bilevel Coherent Risk Programming 

Jonathan Eckstein, Rutgers University, 640 Bartholomew Road, 

Piscataway, NJ, 08854, United States of America, 

jeckstei@rci.rutgers.edu 

We describe a bilevel programming approach to applying coherent risk measures 

to three-stage stochastic programming problems. This formulation technique 

avoids the time-inconsistency difficulties of naive models and the incomposability 

issues causing time-consistent formulations to have complicated, hard-to-explain 

objective functions. Unfortunately, we show that such bilevel formulations, when 

using the standard mean-semideviation and average-value-at-risk measures, are 

NP-hard. 

3 - Scenario Generation and Reduction Applied to Long-term 

Power Generation Expansion Planning with Risk 

Yonghan Feng, Iowa State Univseristy, 3024 Black Engr, Ames, IA, 

United States of America, yhfeng@iastate.edu, Sarah Ryan 

A stochastic program minimizes the combined expected cost and CVaR of long 

term power generation expansion planning. Scenarios are generated from 

stochastic process models of demand and natural gas price and then reduced by a 

new heuristic called forward selection in wait-and-see clusters (FSWC). 

Numerical results in a twenty-year case study indicate substantial computational 

savings while identifying similar solutions as those obtained by applying forward 

selection alone. 

■ SA11 

11- West 105 C- CC 

Advances in Stochastic Programming 


Chair: Alper Murat, Assistant Professor, Wayne State University, 4815 

4th Street, Detroit, MI, 48202, United States of America, 

amurat@wayne.edu 

1 - Simulation-based Two-stage Stochastic Programming with 

Recourse and Endogenous Uncertainty 

Tahir Ekin, PhD Candidate, The George Washington University 

Department of Decision Sciences, 2201 G St Funger Hall 415 NW, 

Washington, DC, 20052, United States of America, ekin@gwu.edu, 

Refik Soyer, Nicholas Polson 

We consider two stage stochastic recourse problems with decision dependent 

(endogenous) uncertainty. We develop an augmented probability simulation 

approach via Markov Chain Monte Carlo methods. Our approach is illustrated 

with a production planning problem involving continuous uncertainty and 

continuous first stage decision variables. 

2 - A New Algorithm for Moment-matching Scenario Generation 

with Application to Financial Optimization 

Diana Roman, Brunel University, Uxbridge, London, 

United Kingdom, diana.roman@brunel.ac.uk, Paresh Date, 

Ksenia Ponomoreva 

This method produces scenarios that match exactly the given first moment, 

covariance matrix, average marginal skewness and kurtosis of a random vector. 

Unlike previous approaches, it does not use optimization in the scenario 

generation process; thus it is computationally much simpler. We generate 

scenarios for a portfolio optimization model. Desirable properties are satisfied, e.g. 

in and out-of-sample stability. Optimal solutions vary only marginally with 

increasing number of scenarios. 

3 - Progressively Hedged Sample Average Approximation for 

Two-stage Stochastic Programs 

Alper Murat, Assistant Professor, Wayne State University, 

4815 4th Street, Detroit, MI, 48202, United States of America, 

amurat@wayne.edu, Nezir Aydin 

We present a novel hybrid heuristic method (PHSAA) for solving two-stage 

stochastic programs which integrates deterministic Progressive Hedging method 

with Sample Average Approximation. PHSAA improves over SAA by exploiting 

sample solution information and reducing the need to use large sample sizes. 

PHSAA is significantly more efficient than SAA while attaining same or better 

solution quality than SAA. We report on extensive results using capacitated 

reliable facility location problem (CRFLP). 


59 

■ SA12 

SA13 

12- West 106 A- CC 

Recent Advances in Integer Programming 

Sponsor: Optimization/Integer Programming 


Chair: Sebastian Pokutta, Friedrich-Alexander-Universitat 

Erlangen-Nürnberg, Cauerstrasse 11, Erlangen, 91058, Germany, 

sebastian.pokutta@me.com 

1 - Solving Mixed-Integer Semidefinite Programs 

Lars Schewe, Dr., FAU Erlangen-Nürnberg, Cauerstr. 11, Erlangen, 

91058, Germany, lars.schewe@math.uni-erlangen.de, Sonja Mars 

We present practical solution approaches to solve mixed-integer semidefinite 

programs: We show that by integrating standard semidefinite solvers into the 

mixed-integer solver SCIP, we can harness standard methods from integer 

programming. We present computational results on a variety of instances. 

2 - Comparing Symmetry Breaking Techniques 

James Ostrowski, University of Tennessee, 1425 S. Stadium Dr, 

Knoxville, TN, 37996, United States of America, jostrows@utk.edu 

There are a variety of ways to break symmetry in an MILP problem. We will 

discuss several of these methods, and test them on a variety of highly symmetric 

problems. 

3 - The Structure of Reduced Bases for Kernel Lattices 

Karen Aardal, Delft University of Technology, Mekelweg 4, 

Delft, 2628 CD, Netherlands, K.I.Aardal@tudelft.nl, 

Frederik von Heymann, Andrea Lodi 

We consider the structure of reduced bases for lattices consisting of all integer 

vectors belonging to the null space of a positive integer vector. We observe that 

when the length is measured as the Euclidean distance, as in the standard 

implementation of the LLL algorithm, then the reduced basis has an Identity 

matrix as a submatrix. We provide a theoretical analysis, implications, and 

computational illustrations. 

4 - The Structure of Reduced Bases for Kernel Lattices: 

Theoretical Analysis 

Frederik von Heymann, Delft University of Technology, Mekelweg 

4, Delft, 2628 CD, Netherlands, f.j.vonheymann@tudelft.nl, 

Karen Aardal, Andrea Lodi, Laurence Wolsey 

In this presentation we continue on the topic of the previous talk. Here we 

present the main ingredients of the theoretical analysis and give some 

computational examples. 

■ SA13 

13- West 106 B- CC 

Stochastic Integer Programming Applications 

Sponsor: Optimization/Linear Programming and Complementarity 


Chair: Burak Buke, The University of Edinburgh, School of 

Mathematics, King’s Buildings, JCMB, Edinburgh, United Kingdom, 

B.Buke@ed.ac.uk 

1 - A Novel Scenario Reduction Method for Two-stage 

Stochastic Programs 

Ali Koc, IBM Watson Research Center, 1101 Kitchawan Road, 

Yorktown Heights, NY, 10598, United States of America, 

akoc@us.ibm.com 

We present a novel scenario reduction method for two-stage stochastic programs. 

The objective is to minimize the bound on the worst case performance of the 

optimal value of the reduced problem under the original problem. The method 

considers both the uncertainty and the underlying optimization problem, as 

opposed the methods that just consider the uncertainty. We show that both the 

theoretical bounds and the numerical results obtained by the method are tighter 

than the methods in literature. 

2 - Reclaimed Water Network Design via Stochastic 

Integer Programming 

Weini Zhang, Systems and Industrial Engineering, University of 

Arizona, Tucson, AZ, 85721, United States of America, 

wzhang@email.arizona.edu, Guzin Bayraksan, Gunhui Chung, 

Kevin Lansey 

We model a reclaimed water distribution network for supplying treated water to 

the public users or agricultural areas using two-stage stochastic binary 

programming with random recourse. Both construction and energy costs 

expanded are considered. The network has spatial growth and reclaimed water 

demands are uncertain. A specialized algorithm is developed to exploit the

SA14 

structure of the network to efficiently solve the problem. Computational results 

are presented on a realistic problem. 

3 - Managing Capacity Flexibility in Make-to-Order 

Production Environments 

Fehmi Tanrisever, Eindhoven University of Technology, Den 

Dolech 2, Eindhoven, 5612, Netherlands, f.tanrisever@tue.nl, 

David Morton, Douglas Morrice 

We address the problem of managing flexible production capacity in an MTO 

manufacturing environment. We present a multi-period capacity management 

model where we distinguish between process and operational flexibility. Our 

results reveal that myopic policies may lead a firm to adopt more process 

flexibility. That is, process flexibility may be over-valued in the literature since it is 

assumed that a firm will operate optimally after the process flexibility decision. 

4 - Cross-training with Imperfect Training Schemes 

Burak Buke, The University of Edinburgh, School of Mathematics, 

King’s Buildings, JCMB, Edinburgh, United Kingdom, 

B.Buke@ed.ac.uk, Ozgur Araz, John Fowler 

Cross-training is an efficient way to achieve flexibility. We assume that training 

can be online as demand is revealed, in addition to the offline training before 

demand is realized. We show that if online training is as effective as offline 

schemes, offline training decreases as variability increases. If the effectiveness of 

online and offline schemes differ, offline training may increase as variability 

increases. We also provide a stochastic integer program to design cross-training 

schemes. 

■ SA14 

14- West 106 C- CC 

Advances in Project Management 


Chair: Bruce Pollack-Johnson, Associate Professor of Mathematics & 

Statistics, Villanova University, 800 Lancaster Avenue, Department of 

Mathematics & Statistics, Villanova, PA, 19085, United States of 

America, bruce.pollack-johnson@villanova.edu 

1 - Multi-player Project Planning and Scheduling under Risk 

Sharing 

Xin Xu, Rutgers University, 425 Mount Prospect Ave., Apt. 417, 

Newark, NJ, 07104, United States of America, 

xin.xu.2009@rutgers.edu, Yao Zhao 

We consider a product development project where the design work of subsystems 

is outsourced to suppliers by a manufacturer. We build a mathematical model to 

predict each firm’s behavior under a risk sharing partnership. We show that each 

firm tends to delay its task relative to what is the best for the entire project as a 

whole. We propose an alternative incentive scheme – fair sharing, to remedy the 

weakness of risk sharing while retaining its benefits. 

2 - Optimizing Team Staffing Decisions: Meeting Performance 

Expectations When Staffing One or Many Teams 

Alison Schroeder, Purdue University, 315 N Grant St., 

West Lafayette, IN, 47907, United States of America, 

schroe32@purdue.edu, Sara McComb, Deanna Kennedy 

Staffing project teams from a pool of candidates with similar skills and abilities is 

not trivial. Moreover, managers may want to staff one team or many teams from 

the pool. Herein we optimize the personality-performance relationship model to 

determine team assignments that fulfill managerial performance expectations. 

Implications are discussed. 

3 - The Impact of Reputation on Project Selection 

Ali Shafahi, Graduate Student, University of Maryland at College 

Park, Dept. of Civil & Environmental Eng., 1173 Glenn L. Martin 

Hall, College Park, MD, 20740, United States of America, 

ashafahi@umd.edu, Ali Haghani, Masoud Hamedi 

Contractor’s reputation is one of the contributing factors in project selection. This 

study takes advantage of an optimization model that considers monetary as well 

as non-monetary criteria. To investigate relative importance of reputation to profit 

on the project selection decision, the model is applied to a set of sample problems. 

The decision making framework and results of sensitivity analysis are presented. 


60 

4 - Project Planning and Scheduling to Maximize Quality in the 

Presence of Cost Overruns and Time Delays 

Bruce Pollack-Johnson, Associate Professor of Mathematics & 

Statistics, Villanova University, 800 Lancaster Avenue, Department 

of Mathematics & Statistics, Villanova, PA, 19085, 

United States of America, bruce.pollack-johnson@villanova.edu, 

Matthew Liberatore 

We present research designed to help deal with significant cost overruns and time 

delays. In such circumstances the overall quality of a project is most likely to 

suffer. We present strategies for dealing with such undesirable situations to 

minimize the reduction in the overall project quality by applying different project 

planning and scheduling models that evaluate the tradeoffs between quality, time, 

and cost. 

■ SA15 

15- West 202- CC 

Software Demonstration 

Invited Session 

1 - AMPL Optimization - Specifying “Logical” Conditions in AMPL 

Optimization Models 

Robert Fourer, President, AMPL Optimization, Inc, 2145 Sheridan 

Road, Evanston IL, United States of America, 4er@ampl.com 

Optimization modelers are often stymied by the complications of converting 

problem logic into algebraic constraints suitable for solvers. The AMPL modeling 

language thus allows various logical conditions to be described directly. 

Additionally a new interface to the ILOG CP solver handles logic in a natural 

way not requiring conventional transformations. 

2 - Palisade Corporation - Avoiding Failure: using @RISK to 

Identify, Analyze and Manage Risk More Successfully 

Rafael Hartke, Oil and Energy Industry Consultant, Palisade 

Corporation, 798 Cascadilla Street, Ithaca, NY 14850, 

United States of America, dsprague@schwartzmsl.com 

Projects fail. They often fail for a lack of effective planning, either because of 

excess optimism in terms of time, cost and performance or because of extra work 

in the plan in terms of scope, resources and unforeseen events. @RISK employs 

Monte Carlo based simulation to give insight through sophisticated statistical 

analysis of the various potential outcomes. Understanding range outcomes facilitates 

effective risk management, prioritizing critical risks and managing expectations. 

■ SA16 

16- West 207- CC 

Multiple Criteria Decision Making: Foundations and 

Some Approaches 

Cluster: Tutorials 


Chair: Murat Köksalan, METU, IE Department, METU, Ankara, Turkey, 

koksalan@ie.metu.edu.tr 

1 - Multiple Criteria Decision Making: Foundations and 

Some Approaches 

Murat Köksalan, METU, IE Department, METU, Ankara, Turkey, 

koksalan@ie.metu.edu.tr, Jyrki Wallenius 

In this tutorial we cover the foundations of Multiple Criteria Decision Making and 

define and demonstrate basic concepts. We classify approaches based on the 

problem structure as well as on the way decision maker’s preferences are elicited. 

We discuss some approaches, emphasizing the interactive approaches. We 

demonstrate the concepts and approaches together with illustrative examples and 

graphs and we provide information about further resources and references.

■ SA17 

17- West 208 B- CC 

Stochastic Programming with Random or 

Ambiguous Data 



Chair: Siqian Shen, Assistant Professor, University of Michigan, 

2793 IOE Building, 1205 Beal Avenue, Ann Arbor, MI, 48103, 

United States of America, siqian@umich.edu 

1 - Models and Algorithms for the Balance-Constrained Stochastic 

Bottleneck Spanning Tree Problem 

Jue Wang, Industrial and Operations Engineering, University of 

Michigan, 1205 Beal Avenue, Ann Arbor, MI, 48109, United States 

of America, juewang@umich.edu, Siqian Shen, Murat Kurt 

This paper studies chance-constrained formulations of the BCSBSTP with 

independently distributed arc weights, and reformulates them as MINLPs. We 

employ SOS1 and SOS2 variables to approximate nonlinearity with linear 

constraints, and compare the results with an SAA-based MIP reformulation of the 

orignal problem. In general, SOS1 computationally performs better than the other 

two approximation approaches, tested on random network instances with various 

distribution settings of arc weights. 

2 - Target Oriented Optimization in Inventory Management 

Zhuoyu Long, NUS Business School, Singapore, 

longzhuoyu@nus.edu.sg, Lucy Gongtao Chen, Georgia Perakis, 

Melvyn Sim 

In this study we investigate the impact of target on both one-period and multiperiod 

inventory management problems. We develop target-oriented decision 

criteria to model the effect of target. For the one-period (newsvendor) problem, 

the result from our theoretical model is consistent with experimental findings. For 

the multi-period problem, we incorporate the financing decisions, and report 

favorable computational results for using targets in regulating uncertain 

consumptions over time. 

3 - Data-driven Chance Constrained Stochastic Program 

Ruiwei Jiang, University of Florida, 411 Weil Hall, University of 

Florida, Gainesville, FL, 32611, United States of America, 

rwjiang@ufl.edu, Yongpei Guan 

Chance constraint is a convenient modeling tool of decision making under 

uncertainty. In practice, instead of knowing the true distribution of a random 

parameter, only a series of historical data sampled from the ambiguous 

distribution are available. In this talk, we develop exact approaches to reformulate 

the Data-driven Chance Constraints (DCC) based on historical data. 

■ SA18 

18- West 208 A- CC 

Applied Network Optimization Methodologies 


Chair: Isil Kirkizoglu, Middle East Technical University, Industrial 

Engineering Department, Ankara, 06800, Turkey, 

isilkirkizoglu@gmail.com 

1 - A New Branch & Bound Approach for the Traveling Repairman 

Problem 

Stefan Bock, Professor, Chair of Business Computing and 

Operations Research, University of Wuppertal, Gauflstrafle 20, 

Wuppertal, Germany, sbock@winfor.de 

This talk studies a special case of the well-known Traveling Repairman Problem. 

Specifically, while processing times at the customer locations are neglected, each 

customer has to be serviced not before a predetermined release date and, 

depending on the considered problem variant, not after a given due date. In order 

to solve the problem, a new best first Branch&Bound approach is proposed. Its 

efficiency is validated by means of computational tests. 

2 - Travelling Salesman Problems in Dynamic Environment 

Eran Simhon, Technion, Technion City, Haifa, HA, 32000, Israel, 

simhon@bu.edu, Shraga Shoval, Liron Yedidsion 

In the last 20 years there has been a growing interest in the field of Moving Target 

Traveling Salesman Problem (MT-TSP), in which targets move at fixed velocities. 

In this research we study the Resupply MT-TSP, in which the agent must visit the 

origin after each interception. We prove that the offline version is strongly NPhard 

and that the online version is unbounded. We also show the equivalence of 

special cases of the problem to scheduling problems with deteriorating jobs. 


61 

SA19 

3 - Benders Decomposition for a Splitter Location-allocation 

Problem in Fiber Optic Access Networks 

Sachin Jayaswal, Assistant Professor, Indian Institute of 

Management Ahmedabad, Vastrapur, Ahmedabad, Gu, 380015, 

India, sachin@iimahd.ernet.in, Astha Airan 

We propose a Benders Decomposition based algorithm to solve the Splitter 

Location-Allocation Problem in Fiber Optic Access Networks. Our numerical 

results are very encouraging: our proposed Benders Decomposition helps achieve 

the optimal solution for large problem instances in couple of seconds which 

CPLEX is unable to solve using the original MIP model even in an hour. 

4 - Parametric Programming Approach to Gene Expression 

Clustering using the Clique Partitioning Problem 

Victoria Ellison, North Carolina State University, Raleigh, NC, 

27605, United States of America, vmelliso@ncsu.edu, Yahya Fathi, 

Yoshitsugu Yamamoto, Amy Langville 

We introduce a parametric programming based algorithm to solving a binary 

integer linear programming (BILP) formulation of the Clique Partitioning 

Problem, for the purpose of gene expression clustering. We derive various 

structural properties regarding the existence and optimality of binary solutions to 

the linear relaxation of the BILP. Using linear programming sensitivity analysis we 

derive a measure of closeness between patients of similar and different disease 

subtypes. 

5 - Discretization for Heat Exchanger Network Design with 

Equipment Configuration 

Isil Kirkizoglu, Middle East Technical University, Industrial 

Engineering Department, Ankara, 06800, Turkey, 

isilkirkizoglu@gmail.com, Haldun Sural, Sinan Gurel 

This study presents an optimization approach to a chemical engineering process 

design problem. We propose a flexible solution approach for heat exchanger 

network design with detailed equipment configuration. It involves discretization 

of temperatures based on heat load equality and formulation of a shortest-path 

problem rather than dealing with nonlinearity at the network design level, a 

common method in the literature. Computational results are provided. 

■ SA19 

19- West 211 A- CC 

Joint Session Healthcare Logistics/SPPSN: Vehicle 

Routing in Healthcare Applications 

Cluster: Healthcare Logistics & Public Programs, Service and Needs 


Chair: Ali Ekici, University of Houston, Department of Industrial 

Engineering, Houston, TX, United States of America, 

aekici@central.uh.edu 

1 - A Constant-factor Approximation Algorithm for Multi-vehicle 

Collection for Processing Problem 

Lerzan Ormeci, Koc University, Rumeli Feneri Yolu, Sariyer, 

Istanbul, 34450, Turkey, LORMECI@ku.edu.tr, Esma Gel, 

Sibel Salman, Eda Yucel 

In multiple vehicle collection for processing problem, items accumulate at a set of 

sites and are transferred through a series of tours of multiple vehicles to a 

processing facility to be processed. We show that the problem with the makespan 

objective is NP-hard, and propose a polynomial-time, constant-factor 

approximation algorithm for its solution. 

2 - Managing Blood Collection Operations 

Ali Ekici, University of Houston, Department of Industrial 


aekici@central.uh.edu, Orsan Ozener, Gultekin Kuyzu 

In this paper, motivated by the practices in blood supply management, we study a 

variant of the vehicle routing problem. Considering processing requirements of 

donated whole blood in order to extract platelets, we analyze the pickup 

operations from donation centers and develop algorithms to maximize the platelet 

production. 

3 - A New Approach for Routing Courier Delivery Services with 

Urgent Demand 

Chen Wang, University of Southern California, University Park 

Campus, Los Angeles, CA, 90007, United States of America, 

Wang20@usc.edu, Fernando Ordonez, Maged Dessouky 

The courier delivery industry is faced with random demand, as well as urgent 

requests. An example of such application is the transportation of medical 

specimens. We propose better vehicle routing solutions that can efficiently satisfy 

random demand and urgent requests. We formulate a multi-trip vehicle routing 

problem, and devise a heuristic based on insertion and Tabu search. Simulations 

on randomly generated data and on real-world data show significant 

improvement from existing methods.

SA20 

■ SA20 

20- West 211 B- CC 

Joint Session: SPPSN/ENRE-Environment: 

Energy Policy and Sustainability 

Sponsor: Public Programs, Service and Needs & Energy, 

Natural Res & the Envi/ Environment and Sustainability 


Chair: Erin Baker, University of Massachusetts, Amherst, MA, 

United States of America, edbaker@ecs.umass.edu 

1 - Decommissioning and Repowering Decisions for Renewable 

Power Producers 

Chenlu Lou, Iowa State University, 3004 Black, Ames, IA, 50011, 

United States of America, clou@iastate.edu, K. Jo Min 

There have been substantial increases in the construction and operation of wind 

farms across the U.S.A. and the world, which will lead to a significant number of 

wind farm decommissioning and repowering decisions soon. We formulate and 

analyze mathematical models from a real options perspective with the operation 

and maintenance cost following the geometric Brownian motion process, and 

derive managerial insights for relevant government policy makers. 

2 - Hydrogen Production Facility Network Design from Stochastic 

Green Energy Supply Source 

Jorge Barnett Lawton, Zaragoza Logistics Center, C/ Bari 55, Plaza, 

Edificio Nayade, Portal 5, Zaragoza, 50197, Spain, 

jbarnett@zlc.edu.es, Mozart Menezes, Jarrod Goentzel 

Hydrogen has been identified as a clean alternative to store energy from volatile 

sources. We analyze the production and distribution network design problem of a 

firm generating electricity from wind and producing hydrogen by electrolysis, in 

the presence of stochastic energy prices and supply. We provide solution 

approaches using column generation, and use our model to evaluate a future 

hydrogen supply chain in Spain, under different distribution policies and 

government incentive schemes. 

3 - R&D Portfolio Analysis of Low Carbon Energy Technologies for 

Climate Change Mitigation 

Rose Zdybel, University of Massachusetts-Amherst, 120D Marston 

Hall, Amherst, MA, United States of America, rozoe@msn.com 

We analyze R&D portfolios of low carbon advanced energy technologies for 

climate change mitigation. We use the GCAM integrated assessment model to 

analyze the effects of combinations of low carbon energy technologies on CO2 

concentration stabilization costs and then combine the results with probabilistic 

data from expert elicitations. We also develop econometric relationships between 

variables to estimate results for input values other than those that are part of the 

elicitations. 

4 - A Decision Support System for Managing Wind-turbines with 

Storage in Smart-grids 

Frederic Murphy, Professor, Temple University, Alter Hall 435, 

Philadelphia, PA, 19122, United States of America, 

fmurphy@temple.edu, Fernando Oliviera 

We analyze the wind-turbine business model given the opportunities created by 

the smart grid and by efficient electric battery storage systems. We model the 

interaction between dynamic prices and wind-regimes; we analyze how batteries 

can be used to improve wind-turbines reliability and profitability. We study how 

forward and option contracts can be used by a wind-turbine to hedge risk and we 

use conditional value at risk to assess the wind-turbine risk exposure and as a 

basis for risk-hedging. 

■ SA21 

21- West 212 A- CC 

Applications of Networks and Graphs 


Chair: Zeynep Ertem, Texas A&M University, Industrial & Systems 

Engineering, 3131 TAMU, College Station, TX, 77843, 

United States of America, zeynep84@tamu.edu 

1 - Detecting Research Fronts using Citation Network Analysis 

Katsuhide Fujita, The University of Tokyo, 2-11-16 Yayoi, 

Bunkyo-ku, Tokyo, 113-8656, Japan, fujita@ipr-ctr.t.u-tokyo.ac.jp, 

Yuya Kajikawa, Junichiro Mori, Ichiro Sakata 

One of the important methodologies for detecting the research fronts is the 

citation network analysis. Especially, citation network analysis with 

combinational types or weights has possibilities of finding the new research fronts 

compared with non-weighted and single type of citation networks. We perform a 

comparative study to investigate the each type of citation networks in detecting a 

research front using visibility, speed, topological relevance and keyword similarity. 


62 

2 - Social Network Analysis at Wine and Food Festival 

Bomi Kang, Associate Professor, Coastal Carolina University, 

P O Box 261954, Conway, SC, United States of America, 

bkang@coastal.edu, Young Jae Kim, Taylor Damonte 

Social network analysis emphasizes the importance of constructing the 

interconnectedness among concepts and knowledge networks in respondents’ 

minds. The technique has gained popular use in social science over the past three 

decades. Using concept maps and degree centralities derived from social network 

analysis, authors demonstrated new insights in festivals in Myrtle Beach Area, 

through this novel technique. 

3 - Traffic Grooming in Optical Networks: Decomposition and 

Partial LP Relaxation 

Hui Wang, PhD Candidate, North Carolina State University, 2152 

Burlington Labs, Raleigh, NC, 27695, United States of America, 

hwang4@ncsu.edu, George Rouskas 

We consider the traffic grooming problem, a fundamental network design 

problem in optical networks. We propose a new decomposition approach with 

two subproblems: (1) virtual topology and traffic routing (VTTR), and (2) routing 

and wavelength assignment (RWA). We also propose an ascending utilization 

iterative algorithm based on partial LP relaxation to further improve the 

scalability of VTTR. Our approach delivers a desirable tradeoff between running 

time and quality of the final solution. 

4 - Graph Theoretical Analysis of Brain Networks 

Zeynep Ertem, Texas A&M University,Industrial & Systems 

Engineering, 3131 TAMU, College Station, TX, 77843, 

United States of America, zeynep84@tamu.edu, Sergiy Butenko 

Exploring the connectivity patterns of human brain is gaining more interest in the 

last decade. Correlated low frequency fluctuations in the blood oxygenation level 

dependent signal have been widely observed in connected brain regions. Resting 

state functional magnetic resonance imaging (R-fMRI) consists of hundreds of 

time periods for thousands of image voxels. Our aim is to model the brain as a 

complex network. The clusters are identified with graph theoretical analysis. 

■ SA22 

22- West 212 B- CC 

Topics in Real Time and Embedded Computing 

Sponsor: Computing Society 


Chair: Alex Mills, Indiana University Kelley School of Business, 1309 E. 

Tenth Street, Bloomington, IN, 47405, United States of America, 

millsaf@indiana.edu 

1 - Optimization-based Frameworks for Dynamic Configuration of 

Real-time Systems with Energy Constraints 

Eduardo Camponogara, Professor, Federal University of Santa 

Catarina, Cx. P. 476, Florianopolis, 88040, Brazil, 

camponog@das.ufsc.br, Riad Nassiffe, George Lima 

Embedded real-time systems powered by batteries require suitable support for 

energy-savings at the operating system level. Mechanisms to do so must take into 

consideration not only energy constraints but also schedulability since tasks must 

execute within predefined time windows. Further, it is desired that application 

quality of service (QoS) is optimized. This talk presents two optimization-based 

frameworks for maximizing application QoS subject to both schedulability and 

energy constraints. 

2 - Can Randomness Buy Clairvoyance? Stochastic Scheduling of 

Mixed Criticality Real-time Job Systems 

Bader Al-Ahmad, PhD Student, University of British Columbia, 

2205 Lower Mall, P.O. Box 70, Vancouver, BC, V6T1Z4, Canada, 

baderalahmad84@gmail.com, Sathish Gopalakrishnan 

We pose the problem of stochastically scheduling mixed criticality jobs when job 

execution time distributions at all criticality levels are given. By stochastic, the 

dynamical evolution of the system criticality level over time is captured by a 

stochastic process. We ask: is such information any beneficial in reducing the 

pessimism of current deterministic algorithms? Our objective is to match as 

closely the performance of a powerful oracle that knows the future system 

behavior realization. 

3 - Temporal Logic Testing for Cyber-physical Systems 

Georgios Fainekos, Arizona State University, 

699 S. Mill Ave., Tempe, AZ, 85281, United States of America, 

Georgios.Fainekos@asu.edu, Sriram Sankaranarayanan 

One of the major challenges in Model Based Development (MBD) of Cyber- 

Physical Systems (CPS) is the verification of real-time functional system 

properties. In general, the problem is undecidable due to the interplay between 

continuous and discrete system dynamics. In this talk, we present how the 

problem of verification can be posed as an optimization problem which can be 

solved using stochastic optimization techniques including Monte-Carlo methods 

and the Cross Entropy method.

■ SA23 

23- West 212 C- CC 

Survivable Networks 

Sponsor: Telecommunications 


Chair: Abdulah Konak, Pennsylvania State Berks, Tulpehocken Road, 

P.O. Box 7009, Reading, PA, 19610, United States of America, 

auk3@psu.edu 

1 - Survivable Capacitated Layered Networks with Spanning 

Tree Protection 

Tachun Lin, Cameron University, 2800 W Gore Blvd, Lawton, OK, 

73505, United States of America, tlin@cameron.edu, Zhili Zhou, 

Krishnaiyan Thulasiraman 

In this paper, we study the survivability problem in capacitated layered networks. 

We propose necessary and sufficient conditions for the existence of survivable 

routing with upper-layer protection, formulate this problem as a mixed-integer 

program, and design a novel upper-layer spanning tree protection algorithm 

which guarantees the upper-layer network connectivity and generates routings 

within lower-layer capacity. Computational experiments demonstrate the 

efficiency of our algorithm. 

2 - Resilience: A New Metric to Evaluate Network 

Reliability/Survivability 

Ozgur Kabadurmus, Auburn University, 3333 Shelby Center, 

Auburn, AL, 36849, United States of America, 

ozk0001@tigermail.auburn.edu, Alice E. Smith 

In telecommunication network design problems, survivability and reliability are 

used to evaluate quality of service. In this study, a new metric that combines 

network reliability with network resilience is presented to measure 

reliability/survivability more effectively. It is compared with well-known network 

reliability/survivability metrics, namely k-connectivity and two terminal 

reliability, and its benefits and computational efficiency are discussed. 

3 - Lexicographical Minimization of Routing Hops in Hopconstrained 

Node Survivable Networks 

Luis Eduardo Neves Gouveia, CIO and DEIO, Cidade Universiteria, 

Bloco C6, Campo Grande, Lisbon, 1749-016, Portugal, 

legouveia@fc.ul.pt, Pedro Patricio, Amaro de Sousa 

We address the following traffic engineering problem: how to route deterministic 

changes of original demands in a pre-dimensioned network, maintaining network 

survivability and QoS, while minimizing the number of hops of either i) all 

service paths or ii) the worst service path for each commodity, in a lexicographical 

sense. We present and discuss two classes of Integer Linear Programming models 

for this problem and computational results considering several traffic demand 

settings. 

4 - A Network Support System for Ad-hoc Networks using 

Autonomous Intelligence Agents 

Abdulah Konak, Pennsylvania State Berks, Tulpehocken Road, 

P.O. Box 7009, Reading, PA, 19610, United States of America, 

auk3@psu.edu 

This paper presents a decentralized approach to maintain the connectivity of a 

MANET using autonomous, intelligent agents. Concepts from the social network 

analysis along with flocking algorithms are utilized to maintain the connectivity 

of mobile nodes. Computational results are presented to demonstrate the effect of 

various local agent behaviors on the global network connectivity. 

■ SA24 

24- West 213 A- CC 

Advances in Modeling and Optimal Control of 

Infectious Disease Spread 

Sponsor: Health Applications Society 


Chair: Reza Yaesoubi, Post-Doctoral Research Fellow, 

Harvard Medical School, 641 Huntington Ave., Boston, MA, 02115, 

United States of America, reza.yaesoubi@gmail.com 

1 - A Model for Optimal Allocation of HIV Prevention and 

Treatment Funds 

Margaret Brandeau, Stanford University, Huang Engineering 

Center, Stanford, CA, 94305, United States of America, 

brandeau@stanford.edu, Sabina Alistar, Elisa Long, Eduard Beck 

We develop a model to determine the optimal allocation of resources between 

HIV prevention and treatment services to minimize the basic reproduction 

number. Our model includes multiple populations with different transmission 

modes, as well as production functions that relate investment in prevention and 


63 

SA25 

treatment to changes in transmission and treatment rates. We apply our model to 

examine HIV epidemic control in two different settings, Uganda and Russia. 

2 - Evaluating the Effectiveness of Interventions During an 

Influenza Pandemic 

Anna Teytelman, PhD Student, Massachusetts Institute of 

Technology, Operations Research Center, Cambridge, MA, 02139, 

United States of America, teytanna@mit.edu, Richard Larson 

With pandemic influenza, pharmaceutical and non-pharmaceutical resources are 

limited. We present dynamic, multi-regional, vaccine allocation schemes for the 

USA. For single communities with heterogeneous populations, we examine the 

effects of timing and targeting strategies on reaching herd immunity. We conclude 

with strategies for combining vaccines with non-pharmaceutical interventions 

such as hand-washing, public awareness campaigns and school closures to 

decrease intensity of an outbreak. 

3 - Resource Allocation for Optimal Control of Epidemics: 

Can Treatment Compete? 

Sabina Alistar, Stanford University, P.O. Box 17244, Stanford, CA, 

94309, United States of America, ssabina@stanford.edu 

Controlling infectious diseases such of HIV requires judicious resource allocation 

between treatment and prevention. We use an optimal control approach to 

determine how the allocation decision will change over time, as an epidemic 

evolves. Our model considers several epidemiological objectives relevant in 

practice, and non-linear epidemic dynamics. We characterize the optimal 

solution, identify conditions for optimality, and provide numerical examples. 

4 - A Framework for Real-time Decision Making During Epidemics 

Reza Yaesoubi, Post-Doctoral Research Fellow, 

Harvard Medical School, 641 Huntington Ave., Boston, MA, 

02115, United States of America, reza.yaesoubi@gmail.com, 

Ted Cohen 

We propose a novel framework for optimizing decisions during infectious disease 

epidemics based on real-time observations from the epidemic and resource 

availability. The framework is flexible and can incorporate diseases of varying 

natural history and a broad range of interventions. Using influenza and 

tuberculosis as examples, we demonstrate how the framework can be used to 

inform decisions to optimize population health within constraints imposed by 

resource availability. 

■ SA25 

25- West 213 B- CC 

Optimization Methods in Radiation Therapy 



Chair: Timothy Chan, Assistant Professor, University of Toronto, 

5 King’s College Rd, Toronto, Canada, tcychan@mie.utoronto.ca 

1 - Imaging Based Treatment Plan Optimization for Glioblastoma 

Jan Unkelbach, Massachusetts General Hospital, 30 Fruit Street, 

Boston, MA, United States of America, junkelbach@partners.org, 

Bjoern Menze, Ender Konukoglu, Helen Shih 

Treatment planning for radiotherapy consists of two problems: 1) determining 

external beam parameters to realize a desired dose distribution, and 2) 

determining the desired dose distribution from clinical and biological data 

including dose-response relations, imaging data, or histology. Quantitative 

approaches to integrate these sources of data involve both mathematical modeling 

approaches a well as statistical machine learning methods. This approach is 

discussed for treatments of gliomas. 

2 - Spatiotemporally Adaptive Radiotherapy for Hypoxic Tumors 

Fatemeh Saberian, University of Washington, Seattle, WA, 

United States of America, saberian@uw.edu, Archis Ghate 

Lack of oxygen (hypoxia) decreases tumor radiosensitivity and hence can reduce 

treatment efficacy. We present a stochastic control approach to design treatment 

plans that adapt to spatiotemporal variations in hypoxia. 

3 - Beam Orientation Optimization in Radiation Therapy 

Treatment Planning 

Troy Long, Student, University of Michigan, 1205 Beal Ave., 

Room 1791, Ann Arbor, MI, 48109, United States of America, 

troylong@umich.edu, Edwin Romeijn 

Beam orientation optimization selects stationary beam positions to produce a both 

high quality and efficiently deliverable treatment plan, resulting in a large-scale 

combinatorial optimization problem. We study an efficient method for selecting 

beam orientations for IMRT treatments that explicitly incorporates the quality of 

the resulting optimal dose distribution.

SA26 

4 - Optimal Magnetic Sweeping Design for Metastatic Cancer 

Tumor Treatment 

So Yeon Chun, Assistant Professor, Georgetown University, 

McDonough School of Business, 37th and O Streets, Washington 

DC, 20057, United States of America, sc1286@georgetown.edu, 

Alek Nacev 

Metastatic cancer is cancer that has spread from the place where it first started to 

another place in the body. Patients with metastatic cancer have thousands of 

tumors spread throughout their body and it is recognized that many of these 

tumors are poorly vascularized compared to normal tissues. To improve poor 

chemotherapeutic efficacy in metastases, we control magnets to sweep 

chemotherapy. We analyze autopsy data of thousands of tumors and optimize 

magnetic pulling treatment parameters. 

■ SA26 

26- North 221 A- CC 

Inventory Models and Forecast Updating 

Sponsor: Manufacturing & Service Oper Mgmt 


Chair: Gad Allon, Northwestern University, 

2001 Sheridan Rd, Evanston, IL, United States of America, 

g-allon@kellogg.northwestern.edu 

1 - Is More Inventory Information Better? Impact of Downstream 

Data on Upstream Supply Forecast Accuracy 

Ruomeng Cui, PhD Student, Northwestern University, 2001 

Sheridan Rd, Evanston, IL, 60208, United States of America, rcui@kellogg.northwestern.edu, 

Jan Van Mieghem, Gad Allon, 

Achal Bassamboo 

We investigate the improvement of forecast accuracy of the upstream supplier’s 

shipment by using downstream retailer data for a beverage drink company. 

2 - Emergency Inventory Pooling 

Jordan Tong, Assistant Professor, University of Wisconsin-Madison, 

975 University Avenue, Madison, WI, 53706, United States of 

America, jtong@bus.wisc.edu, Jing-Sheng Song, Fang Liu 

Organizations contract with vendors to always maintain a minimum level of 

inventory for disaster preparedness. We consider the problem of how to manage 

this emergency inventory among multiple vendors. Instead of stockpiling, we 

show that a type of virtual pooling we call emergency inventory pooling can 

reduce costs. We identify the optimal policy and derive an algorithm for 

implementation. Finally, we examine for which firms and products emergency 

inventory pooling is most valuable. 

3 - Multi-item Closed-loop Inventory Control with Forecast 

Updates: A Model for DVD-by-Mail Services 

Li Chen, Assistant Professor, Duke University, 100 Fuqua Drive, 

Durham, NC, 27708, United States of America, li.chen@duke.edu, 

Safak Yucel, Kaijie Zhu 

In this paper, we present a multi-item forecasting-inventory model for a DVD-bymail 

service provider (such as Netflix). Our inventory model features a two-sided 

fixed cost for each item and a multi-item service constraint. We propose an 

innovative forecasting algorithm and an effective heuristic for this problem. 

Simulation studies show that a significant performance improvement is achieved 

by the new forecasting algorithm and the multi-item optimization heuristic. 

■ SA27 

27- North 221 B- CC 

Information Economics and Operations 



Chair: Wenqiang Xiao, Assistant Professor, New York University, 44 W 

4 Street, KMC 8-72, New York, NY, 10012, United States of America, 

wxiao@stern.nyu.edu 

1 - Efficient Supplier or Responsive Supplier? Analysis of Sourcing 

Strategies under Competition 

Fuqiang Zhang, Olin Business School, Washington University, 

St. Louis, MO, United States of America, Fzhang22@wustl.edu, 

Xiaole Wu 

Motivated by the recent trend of backshoring, this paper studies a sourcing game 

where competing firms may choose between efficient sourcing (e.g., sourcing 

from an international supplier) and responsive sourcing (e.g., sourcing from a 

domestic supplier). By characterizing the equilibrium outcome, we find some 

interesting results driven by the information structure of the sourcing game. 


2 - Signaling to Partially Informed Investors in the Newsvendor 

Model 

William Schmidt, Harvard Business School, Wyss House, 

Boston, MA, 02163, United States of America, wschmidt@hbs.edu, 

Vishal Gaur, Ananth Raman, Richard Lai 

We investigate the effect of short-term objectives and incomplete information on 

a firm’s operational decisions. We incorporate the newsvendor model using 

discrete stocking quantities into a signaling game and identify the circumstances 

under which the firm’s choice of stocking quantity purposefully does not 

maximize expected profits. Our model provides analytical support for the 

abundant anecdotal evidence that superior firms will knowingly under-invest in 

capacity under certain circumstances. 

3 - Stocking More versus Less: The Roles of Demand Volatility and 

Profit Margin under Market Valuation 

Wenqiang Xiao, Assistant Professor, New York University, 44 W 4 

Street, KMC 8-72, New York, NY, 10012, United States of America, 

wxiao@stern.nyu.edu, Guoming Lai 

We study the effects of asymmetric information of demand volatility on the 

inventory decision of a public firm who cares about not only its operational 

profits but also its market value. We find that the firm, when it faces a high 

demand volatility (less efficient), overstocks if the profit margin is low and 

understocks if the profit margin is high, while the firm with a low demand 

volatility (more efficient) may either overstock or understock in both cases. 

■ SA28 

28- North 221 C- CC 

Managing Manufacturing and Related Services 



Chair: Opher Baron, University of Toronto, 105 St George Street, 

Toronto, ON, Canada, Opher.Baron@rotman.utoronto.ca 

Co-Chair: Philipp Afeche, Rotman School of Management, University 

of Toronto, 105 St.George Street, Toronto, ON, M5S3E6, Canada, 

Philipp.Afeche@Rotman.Utoronto.Ca 

1 - Reverse Logistics and Warranty Matching in a Large Consumer 

Electronic Retailer 

Andre Calmon, Massachusetts Institue of Technology, 77 

Massachusetts Avenue, Bldg. E40-149, Cambridge, MA, 02139, 

United States of America, acalmon@mit.edu, Stephen Graves 

We describe, model and optimize the reverse logistics of our industrial partner, a 

Fortune 500 company that sells consumer electronics. Based on real data, we 

introduce a discrete-time model that captures the dynamics of this reverse 

logistics system. In addition, using an approximate dynamic programming 

approach, we propose a closed-loop data-driven policy for managing inventory. 

The performance of this policy is analysed theoretically and through numerical 

experiments. 

2 - The Assortment Packing Problem: Multiperiod Assortment 

Planning for Short-Lived Products 

Victor Martinez de Albeniz, Associate Professor, IESE Business 

School, Av. Pearson 21, Barcelona, 08034, Spain, 

valbeniz@iese.edu, Felipe Caro, Paat Rusmevichientong 

Retailers introduce new items to refresh product lines and maintain their market 

share. We study how a firm must decide, in advance, the release date of each 

product in a given collection over a selling season. Our formulation models the 

trade-offs among profit margins, preference weights, and limited life cycles. A key 

aspect of the problem is that each product is short-lived in the sense that, once 

introduced, its attractiveness lasts only a few periods and vanishes over time. 

3 - The Economics of Joint Production in Services 

Guillaume Roels, Assistant Professor, UCLA, 110 Westwood Plaza, 

Los Angeles, 90095, United States of America, 

guillaume.roels@anderson.ucla.edu 

In this talk, we propose a classification of services based on the degree of 

complementarity or substitutability between the provider’s and the client’s efforts. 

We study how efforts change when they become more complementary and when 

their distribution shifts from the service provider to the client. We identify nonmonotone 

effects of these changes, revealing a trade-off between operational 

efficiency, leading to more specialization, and complementarity, pushing for 

greater effort alignment.

4 - Understanding the Performance of the Long Chain and Sparse 

Designs in Process Flexibility 

Yehua Wei, Massachusetts Institute of Technology, 60 

Wadsworth,10B, Cambridge, MA, 02142, United States of 

America, y4wei@mit.edu, David Simchi-Levi 

This presentation develops theories to explain the effectiveness of the long chain 

and other sparse process flexibility designs in manufacturing systems. In the 

presentation, we first uncover a supermodular property for the long chain, and 

proceed to prove several theorems. Our results verify the effectiveness of the long 

chain, which has been observed empirically in both academia and industry. 

Moreover, the results provide new insights to how to create sparse designs. 

■ SA29 


Workforce Management in the Services 

Cluster: Workforce Management 


Chair: Jennifer Ryan, Associate Professor, RPI, 110 8th Street, CII, Troy, 

NY, 12180, United States of America, ryanj6@rpi.edu 

1 - Planning for the Future - A Model of Cardiac Surgeons in 

Canada 

Sonia Vanderby, Assistant Professor, University of Saskatchewan, 

57 Campus Drive, Saskatoon, SK, S7N 5A9, Canada, 

sonia.vanderby@usask.ca, Michael Carter 

We present a model simulating the workforce within a single specialty at a 

national level which includes students training to enter the profession, providing 

a tool that would help to inform future resource planning. This presentation 

includes details of the model, developed using system dynamics modeling, and a 

demonstration of it using the example of cardiac surgeons in Canada, and 

analyzes the effect of provider motivations on system outcomes. 

2 - Demand Management for Optimal Workforce Planning in 

Professional Service Firms 

Vincent Hargaden, University College Dublin, School of Mech & 

Materials Eng, Belfield, Dublin 4, Ireland, 

vincent.hargaden@ucd.ie, Amir Azaron, Jennifer Ryan 

Our previous work has focused on characterizing and modeling the workforce 

planning process in professional service firms. In this, we initially assumed 

customer projects, i.e., demand, have fixed start dates. However, this can lead to 

firms not being able to fulfil all demand due to insufficient supply of 

workers/skills at certain points in time. We now extend our model to allow firms 

better manage their demand and worker/skill availability by identifying the 

optimum set of project start dates. 

3 - Dynamics of Labor, Traffic, and Sales in Retail Stores 

Howard Hao-Chun Chuang, PhD Candidate, Mays Business School 

at Texas A&M University, College Station, TX, 77843-4217, 

United States of America, hchuang@mays.tamu.edu, Olga 

Perdikaki, Rogelio Oliva 

We perform an empirical study that comprises multiple modeling efforts. First, we 

develop a response function to quantify the impact of labor and traffic on sales. 

Second, we propose and econometrically estimate a generic staffing heuristic. 

Last, we conduct counterfactual analysis in which the heuristic performs closely 

to the idealized optimal and outperforms existing staffing levels. Our major 

contribution is to quantify the benefits of planning labor based on sales as well as 

traffic. 

4 - Maintenance Policies for Multiple Machines with Technician 

Constraints in a Stochastic Environment 

Raha Akhavan-Tabatabaei, Assistant Professor, Universidad de los 

Andes, Cra 1 Este #19A -40, Bogota, 1111, Colombia, 

r.akhavan@uniandes.edu.co, Juan Sebastian Borrero Angarita 

We consider the problem of dynamic allocation of technicians to perform 

preventive maintenance (PM) operations on degrading machines in order to 

minimize the expected PM, repair, and waiting work in process costs. We assume 

Markovian processes for the degradation and load, stochastic lengths for the PM 

and repair operations and that each technician is able to serve only a particular 

set of machines. Near-optimal policies are obtained using approximate dynamic 

programming techniques. 


65 

■ SA30 

SA31 


Operational and Financial Decisions of the Firm 

Sponsor: Manufacturing & Service Oper Mgmt/iFORM 


Chair: Jiri Chod, Boston College, 140 Commonwealth Ave, 

Chestnut Hill, United States of America, jiri.chod@bc.edu 

1 - Inventory Financing and Trade Credit 

Qi Wu, University of Texas at Austin, 2110 Speedway Stop B6500, 

Austin, TX, 78703, United States of America, 

qi.wu@phd.mccombs.utexas.edu, Kumar Muthuraman, 

Sridhar Seshari 

Trade credit is one of the common ways to finance inventory and is often offered 

via an early payment discount. In this talk, we present a model of trade credit, 

embedded in a stochastic inventory control problem. We compute the optimal 

joint inventory, cash management and trade credit decisions, based on which, we 

show that trade credit helps with smoothing cash flows and making better 

operational decisions. 

2 - The Impact of Budget Constraints on Flexible versus Dedicated 

Technology Choice 

Onur Boyabatli, Singapore Management University, 50 Stamford 

Road, Singapore, Singapore, oboyabatli@smu.edu.sg, Tiecheng 

Leng, Beril Toktay 

This paper analyzes the impact of financial constraints in a capacity investment 

setting. We model a two-product firm that decides on its technology choice 

(flexible versus dedicated) and capacity level under demand uncertainty, and the 

production quantities after this uncertainty is resolved. Differing from the 

majority of the stochastic capacity investment literature, we assume that the firm 

is budget-constrained both in the capacity investment and production stages. 

3 - The Provision and Impact of Trade Credit in Imbalanced Supply 

Chains: An Empirical Study 

Jianer Zhou, Assistant Professor, Boston College, Fulton Hall 454C, 

140 Commonwealth Ave., Chestnut Hill, MA, 02467, 

United States of America, jianer.zhou@bc.edu, Hsiao-Hui Lee 

Using a large dataset that links non-service U.S. suppliers with their major 

customers, we show trade credit to be a collaborative and competitive tool in 

supply chains. While suppliers with smaller market share extend more trade 

credit to be competitive, suppliers with weaker bargaining power extend less to 

mitigate financing risk. When offered at the industry-average level, trade credit 

increases suppliers’ sales, but when offered more aggressively, it reduces the 

entire supply chain sales. 

4 - Startegic Consumer Impact on Bankruptcy and Competition 

John Birge, Professor, University of Chicago, Chicago, IL, 

United States of America, john.birge@chicagobooth.edu, 

Rodney Parker, Song Alex Yang 

In a previous model of firm competition under the threat of one firm’s 

bankruptcy, we showed how that firm could use bankruptcy strategically to gain 

concessions from a dominant supplier. In this work, we consider the effect on the 

supply chain from consumers who anticipate future price discounts in the event 

of a bankruptcy. 

■ SA31 


Supply Chain Modeling under Uncertainty 

Cluster: Managing Disruptions in Supply Chains 


Chair: Joseph Geunes, University of Florida, 303 Weil Hall, Gainesville, 

FL, 32611, United States of America, geunes@ise.ufl.edu 

1 - A Capacitated Supplier Prioritizing Asymmetric Retailers 

Zohar Strinka, University of Michigan, 1205 Beal Ave., Ann Arbor, 

MI, 48109, United States of America, zstrinka@umich.edu, 

Edwin Romeijn, Christopher Tang 

We consider a capacitated supplier and a set of asymmetric retailers. The retailers 

are assumed to have complete information, face newsvendor-type costs due to 

uncertain demand, and have a common exogenous selling price. The supplier 

solicits bids of a per-unit price and order quantity from each retailer, and then 

allocates the fixed capacity based on those bids. We show structural properties of 

equilibrium solutions for two retailers when the highest bid receives highest 

priority.

SA32 

2 - Planning Product Availability for an Assortment of Vertically 

Differentiated Products 

Yalcin Akcay, Koc University, Rumelifeneri Yolu, Istanbul, Turkey, 

yakcay@ku.edu.tr, Hari Natarajan 

In many industries, firms offer a variety of substitutable products that customers 

can choose from. Because product demands are inter-related, managers must 

determine inventory policies jointly. Considering the case of a verticallydifferentiated 

product assortment, we model and analyze the problem of 

determining the optimal inventory policy for a product assortment. 

3 - Dynamic Pricing with Strategic Customers under Price 

Adjustment Policy 

Boxiao (Beryl) Chen, University of Michigan, 1205 Beal Avenue, 

Ann Arbor, MI, 48109, United States of America, 

boxchen@umich.edu, Xiuli Chao, Hyun-Soo Ahn 

We study the dynamic pricing strategy of a monopolist serving heterogeneous 

forward-looking customers. The firm offers a price adjustment policy, which 

guarantees current buyers a full refund of the difference if they find a lower 

future price within a certain time window. We establish the existence of a 

subgame perfect equilibrium pricing policy and show that the firm sets different 

equilibrium pricing patterns in different regions of system parameters. 

4 - Assortment Planning in Multiple Selling Channels 

Yiqiang Su, Phd Candidate, University of Florida, Department of 

Industrial and Systems Eng, 450 Weil Hall, Gainesville, FL, 32607, 

United States of America, ysu1987@ufl.edu, Joseph Geunes 

In this research, we consider a clicks-and-mortar retailer who sells goods to endcustomers 

through both physical and virtual (e-commerce) channels. The retailer 

has limited capacity, and would like to determine the set of products to offer, both 

in the local and online stores to maximize its profit from sales during a single 

selling season. MNL and Exogenous demand models are used to capture the 

customer choice behaviors, and a chance-constrained two stage model is built to 

solve this problem. 

■ SA32 

32- North 223- CC 

Quality Control and Learning in Manufacturing 


Chair: Gabriel Burnett, The Boeing Company, P.O. Box 3707, 

MC 13-98, Seattle, WA, 98124, United States of America, 

gabriel.a.burnett@boeing.com 

1 - Changing the Paradigm for Education in Quality 

Thong Goh, Professor, National University of Singapore, 

1 Engineering Dr 2, Blk E1A #06-25, Singapore, 117576, 

Singapore, isegohtn@nus.edu.sg 

Quality education has been “customer oriented” for many decades. Some ideas 

and techniques have become outmoded in the modern society. It is time that 

some paradigm shift was made in quality management education. This needs to 

be realized soon if the new generation of quality practioners are to find 

themselves relevant. 

2 - A Hybrid Inspection Strategy for Rapid Deployment of 

Automated Inspection Technology 

Patrick Spicer, General Motors, General Motors R&D Center, 

Warren, MI, 48093, United States of America, 

patrick.spicer@gm.com, Mike Wincek, Jeffrey Abell, Hui Wang 

New manufacturing processes for advanced automotive propulsion systems (i.e. 

batteries or fuel cells) may not have well-developed automated inspection 

technologies at first. In this presentation, a hybrid inspection strategy is described 

that enables a manufacturer to begin with human inspection and gradually 

phase-in automated inspection technology until the inspection task is completely 

automated. This strategy has been successfully implemented in an automotive 

assembly plant. 

3 - Continuous Improvement in Cellular Manufacturing: The Case 

of an Electronics Manufacturer in Japan 

Tomoaki Shimada, Associate Professor of Operations Management, 

Kobe University, 2-1 Rokkodai-cho, Nada-ku, Kobe, 657-8501, 

Japan, shimada@b.kobe-u.ac.jp, James Ang, Hisashi Kurata 

This study is case-based research that analyzes continuous improvement in 

cellular manufacturing at an IC recorder manufacturing plant in Japan. The study 

provides practical insight into how to make an effort to improve the productivity 

of one-worker cellular production over time. The implementation results indicate 

that workers reduced the takt time by about 19%, and confirm that continuous 

improvement plays an important role in cellular manufacturing from the 

perspective of human dimensions. 


66 

4 - Investigation of Learning Curves in Production Systems 

Gabriel Burnett, The Boeing Company, P.O. Box 3707, MC 13-98, 

Seattle, WA, 98124, United States of America, 

gabriel.a.burnett@boeing.com, Adam Graunke 

We investigate the behavior of learning curves in the context of large scale 

production systems. Learning curves are used extensively when planning 

production systems. We demonstrate the implications of a variety of commonly 

held assumptions. We then provide an empirical study to analyze which events 

(i.e. rate breaks, technology insertion, derivative products, etc) create a significant 

departure from the global learning curve and how such events could be modeled 

in the future. 

■ SA33 


Sustainable and Responsible Operations 

Sponsor: Manufacturing & Service Oper Mgmt/Sustainable 

Operations 


Chair: Jose Cruz, Associate Professor, University of Connecticut, 

School of Business, Storrs, CT, 06269, United States of America, 

jose.cruz@business.uconn.edu 

1 - Closed-loop Supply Chain with Quality Differentiation 

Patrick Qiang, Assistant Professor, Pennsylvania State University, 

Great Valley, Philadelphia, PA, United States of America, 

qzq10@gv.psu.edu 

We model the CLSC that captures consumer’s heterogeneous valuation on the 

remanufactured product and the new product. The model captures the interaction 

among various players, including manufacturers, retailers, and consumers. 

Numerical examples are used to provide managerial implications. 

2 - A Dynamic Model to Study Sustainable Business Policies in a 

Closed Loop Supply Chain 

Sudip Bhattacharjee, Associate Professor, University of 

Connecticut, Storrs, CT, United States of America, 

sbhattacharjee@business.uconn.edu, Jose Cruz 

The sustainability of a closed loop supply chain depends on the economic viability 

of all actors in the chain. We model the phenomenon analytically and create 

dynamic models to study key factors that impact profitability of partners in the 

chain. We find that product design, marketing and related actions of up-market 

producers significantly influence the ability of down-market recyclers to survive. 

We discuss environmental and revenue implications of several policy scenarios. 

3 - Operationalizing Sustainable Production Strategies Subject to 

Diminishing Resources 

Manuel Nunez, Associate Professor, School of Business, UConn, 

2100 Hillside Road Unit 1041, Storrs, CT, 06269, United States of 

America, mnunez@business.uconn.edu, Hsiao-Hui Lee 

We develop a two-stage dynamic model to analyze different sustainability 

strategic decisions such as starting a recycling program or developing alternative 

technologies to replace a diminishing resource. We investigate the impact of the 

parameters on the optimal decision and discuss how to operationalize the 

decisions in the supply chain. 

4 - Mitigating Global Supply Chain Risks through Corporate 

Social Responsibility 

Jose Cruz, Associate Professor, University of Connecticut, 

School of Business, Storrs, CT, 06269, United States of America, 

jose.cruz@business.uconn.edu 

We develop a decision model that captures supply side disruption risks, social 

risks, and demand side uncertainty within an integrated global supply chain and 

corporate social responsibility (CSR) modeling and analysis framework. We 

investigate the effects of CSR activities in a global supply chain and compute the 

equilibrium product flows, prices, and levels of social responsibility activities. The 

results show that CSR activities can potentially be used to mitigate global supply 

chain risk.

■ SA34 

34- North 224 B-CC 

Decisions in Product Development I 


Chair: Amir Sanayei, Wayne State University, 4815 Fourth St., Detroit, 

MI, 48202, United States of America, sanayei@wayne.edu 

1 - Decision Bias in Search for the Best Design Alternative 

Gulru Ozkan, Assistant Professor, Clemson University, 

101 Sirrine Hall, Clemson, SC, 29678, United States of America, 

gulruo@clemson.edu, David Hall, Jeremy Hutchison-Krupat, 

Fred Switzer 

In new product development, one key question is in which order to evaluate 

design alternatives and when to stop testing and iteration to develop a specific 

alternative further (search for the best alternative). Existing models analytically 

develop optimal search strategies for the managers while manager’s search 

behavior may deviate from these. Drawing from psychological literature in 

judgment and decision-making, such as the prospect theory, we analyze the 

manager’s decision making strategy. 

2 - Construction of a UX Component Model and its Application: 

A Case Study on Smart Phones 

Jun-yeon Heo, Pohang University of Science and Technology, Eng- 

Building 4-316, Hyoja-dong, Nam-gu, Pohang, Gyugbuk, Korea, 

Republic of, bluejy@postech.ac.kr, Min Jun Kim, Hyun-Jin Kim, 

Seungchul Shin, Hyo-in Ahn, Kwang-Jae Kim, Jinho Yim 

As the notion of User eXperience (UX) plays an important role in enhancing the 

competitiveness of products, UX evaluation has become an important issue in 

new product development. This talk presents a repository of UX components, 

called a UX component model, which is used for evaluating the UX of products. 

An application of the model is also presented using a case study on smart phones. 

3 - An Optimization of the Specification of BTO Technology System 

Yuji Sato, Mie Chukyo University, 1846 Kubo, Matsusaka, 515- 

8511, Japan, ysatoh@1988.jukuin.keio.ac.jp 

The objectives of this study were to develop a diagnosis procedure of user’s 

preference for a new BTO technology system, and to subsequently optimize the 

specification of the system. Determining the specification of a new system is 

complicated task, because of subjective factors entering into the evaluation of 

necessary and sufficient specification. By combining cost-benefit analysis and 

AHP, this issue was addressed. A case study was carried out to demonstrate the 

applicability of our approach. 

4 - Technology Selection: Performance, Cost and Time 

Considerations during Product Life Cycle 

Amir Sanayei, Wayne State University, 4815 Fourth St., Detroit, 

MI, 48202, United States of America, sanayei@wayne.edu, 

Leslie Monplaisir 

Technology selection is one of the most important and strategic decisions made in 

new product development process. We developed a mathematical modeling 

framework to select the best alternative technology in order to analyze the 

implications of time to market, performance and development cost. The model 

considers whole product life cycle in a competitive environment. It also 

determines optimal product launch time and the best managerial actions in order 

to maximize total profit. 

■ SA35 


Sustainability and Energy Cost 

Sponsor: Revenue Management & Pricing 


Chair: Parijat Dube, IBM, TJ Watson Research Center, Hawthorne, NY, 

United States of America, pdube@us.ibm.com 

Co-Chair: Genady Ya. Grabarnik, St. John’s University, Queens, NY, 

United States of America, grabarng@stjohns.edu 

1 - The Carbon Reduction Effect and Policy Choices of Adjusting 

Industrial Structure 

Zengkai Zhang, School of Management, Xiían Jiaotong University, 

No.28, Xianning West Road, Shaanxi, Xi’an, China, 

zengkaizhang@sina.com, Ju’e Guo 

Reducing carbon emissions and adjusting industrial structure are China’s two 

strategic tasks, while adjusting industrial structure is the essential way of carbon 

reduction. Based on China’s SAM of 2007, this paper develops China’s dynamic 

CGE model to evaluate the contribution of the industrial structure adjustment in 

carbon reduction, and proposes policy recommendations on adjusting industrial 

structure. 


67 

SA36 

2 - Medical Imaging Capacity and its Impacts on Energy and 

Sustainability 

Janet Twomey, Professor, Wichita State University, Industrial and 

Manufacturing Engineering, 1845 Fairmount St, Wichita, KS, 

67206, United States of America, janet.twomey@wichita.edu 

The research presented here highlights a new area for energy improvement in 

healthcare – the energy and material use link to patient outcome decisions and 

alternatives.The outcomes of a study investigating the impact of medical imaging 

capacity decisions on energy and sustainability will be presented for a medical 

center and the nation’s healthcare system. Research sponsored by: NSF CMMI 

1037961 and NSF CMMI 0946342. 

3 - SUITS: A Framework for Sustainable IT Services 

Parijat Dube, IBM, TJ Watson Research Center, Hawthorne, NY, 

United States of America, pdube@us.ibm.com, 

Genady Ya. Grabarnik, Laura Shwartz 

We introduce the Sustainable IT Services (SUITS) framework to develop models 

for operational performance of an IT Service Provider in a global environment 

with carbon taxation and carbon credit trading markets. The models are used to 

formulate a revenue optimization problem for the IT Service Provider in this 

environment. The model solutions can provide guidance for designing operational 

level of different IT components and for creating effective strategies for trading in 

carbon markets. 

■ SA36 


New Applications of Pricing 



Chair: Georgia Perakis, Massachusetts Institute of Technology, 

Cambridge, MA, United States of America, georgiap@MIT.EDU 

1 - Designing Subsidies with Industry Response Dynamics: 

Commitment vs. Flexibility 

Ruben Lobel, Assistant Professor, Wharton School, Philadelphia, 

United States of America, rlobel@wharton.upenn.edu, 

Georgia Perakis, Maxime Cohen 

Governments use consumer incentives to promote new technologies and 

stimulate investments from the private sector (eg. solar panels). We model the 

interaction between a government and an industry player in a multi-period 

setting under uncertain demand. We show how the timing of decisions will affect 

production and subsidy levels. In particular, a flexible subsidy policy will be on 

average more expensive for the government than a committed long-term policy. 

2 - Revenue Management of Reusable Resources with 

Advanced Reservations 

Cong Shi, PhD Candidate, Massachusetts Institute of Technology, 

77 Massachusetts Avenue, Cambirdge, MA, 02139, 

United States of America, shicong@mit.edu, Retsef Levi 

This paper studies a class of revenue management problems in systems with 

reusable resources and advanced reservations. A simple control policy called the 

class selection policy (CSP) is proposed based on solving a knapsack-type linear 

program (LP). It is shown that the CSP and its variants perform provably nearoptimal 

in the Halfin-Whitt regimes. The analysis is based on new approaches 

that model the problem as loss network systems with advanced reservations. 

3 - A Simultaneous Approximation for Many Choice Models 

Vineet Goyal, Columbia University, 500W 120th St., New York, 

NY, United States of America, vgoyal@ieor.columbia.edu, 

Jose Blanchet, Guillermo Gallego 

Modeling customer choice probabilities is a critical problem in assortment 

optimization problems. Several parametric models including MNL, Nested Logit, 

and Mixture of MNL are widely used in practice but the selection of optimal 

model is not usually clear. We present a non-parametric choice model that is a 

simultaneous approximation for a large class of choice models. Furthermore, we 

give a polynomial time algorithm to compute an optimal assortment under this 

choice model. 

4 - Demand-response in the Smart Grid: A Data-driven Pricing and 

Inventory Optimization Approach 

Pavithra Harsha, Research Staff Member, IBM Research, 

1101 Kitchawan Rd, 34-225, Yorktown Heights, NY, 10598, 

United States of America, pharsha@us.ibm.com, 

Dharmashankar Subramanian, Ramesh Natarajan 

Demand response schemes based on dynamic pricing are of considerable interest 

in the emerging smart grid. We formulate the demand management problem as a 

price-sensitive newsvendor and present a novel data-driven, distribution-free 

approach that is based on the use of quantile and mixed quantile regression to 

jointly estimate the optimal generation level and pricing signals in the presence of 

multiple drivers of demand. We illustrate the efficacy, robustness and accuracy 

over alternate methods.

SA37 

■ SA37 


Revenue Management and Pricing in Practice 



Chair: Tugrul Sanli, Director, R&D, SAS Institute Inc., 

SAS Campus Drive, Cary, NC, 27513, United States of America, 

Tugrul.Sanli@sas.com 

1 - Randomized Linear Programming Approach to Revenue 

Management 

Vijay Desai, Opertations Research Specialist, SAS Institute Inc., 

SAS Campus Dr, Cary, NC, 27513, United States of America, 

VijayV.Desai@sas.com, Altan Gulcu 

We provide a high-level overview of our approach to RM using RLP. Following 

are the key features of our methodology: a) Pricing of price-dependent products. 

b) Handles network-level overbooking as opposed to resource-level overbooking 

typically found in practice. c) Allows for imposition of service level constraints. d) 

Computes ancillary outputs such as occupancy forecast, revenue forecast and 

service levels, which are useful metrics for a revenue manager. 

2 - Dynamic Pricing and Revenue Management 

Maarten Oosten, SAS Institute, World Headquarters, 

SAS Campus Drive, Cary, NC, 27513, United States of America, 

maarten.oosten@sas.com 

New technologies allow for pricies to be changed continuously, and optimal 

dynamic pricing algorithms take this freedom explicitly into account. In this 

presentation we explore several directions in which this can be taken. 

3 - Estimation of Nested Logit Models using Data from a Single 

Firm 

Jeff Newman, Research Engineer, Georgia Institute of Technology, 

790 Atlantic Drive, Civil Engineering, Atlanta, GA, 30332, 

United States of America, jeff@newman.me, Laurie Garrow, 

Mark Ferguson 

We examine identification issues for estimating parameters for discrete choice 

models with censored data. Most work in revenue management using such 

models has been focused on simple multinomial logit models, because the unique 

IIA property of this model makes estimating and applying parameters convenient. 

However, IIA is not usually consistent with actual choice behaviors. We examine 

the benefits and drawbacks of employing a nested logit model, which does not 

exhibit IIA. 

4 - Improving the Forecasting Accuracy of Hotel Arrivals: 

A New Non-homogeneous Poisson Approach 

Misuk Lee, IHG, 3 Ravinia Drive, Atlanta, GA, 30327, 

United States of America, Misuk.Lee@ihg.com 

Demand forecast is one of key components of hotel revenue management. This 

study develops a new non-homogenous Poisson process approach to short-term 

demand forecast. By considering seasonality and heterogeneous booking curve 

patterns, we find better estimates for demand processes and thus improve 

forecasting accuracy. We present validation results on IHG hotel data along with a 

comparison to other benchmark models. 

■ SA38 


Impact of Customer Behavior on Service 

Sponsor: Service Science 


Chair: Margaret Pierson, Assistant Professor, Tuck at Dartmouth, 

Hanover, NH, United States of America, mpierson@hbs.edu 

1 - How Do Customers Respond to Service Quality Competition? 

Ryan Buell, Assistant Professor, Harvard Business School, 

Morgan Hall T37, Boston, MA, 02163, United States of America, 

rbuell@hbs.edu, Dennis Campbell, Frances Frei 

When does increased service quality competition lead to customer defection, and 

which customers are most likely to defect? We find that customers defect at a 

higher rate from the incumbent following increased service quality (price) 

competition when the incumbent offers high (low) quality service relative to 

existing competitors in a local market. We further show that it is the high quality 

incumbent’s most profitable customers who are the most attracted by superior 

quality entrants. 


68 

2 - Modeling Customer Behavior in Multichannel Customer 

Support Services: An Information Stock Approach 

Serguei Netessine, Professor, INSEAD, Boulevard de Constance, 

Fountainbleau, 77305, France, Serguei.Netessine@insead.edu, 

Kinshuk Jerath, Anuj Kumar 

We analyze multichannel customer support environment of a health insurance 

firm in the US. We build a structural probability model to understand the 

customers’ query arrival and channel choice processes. Our estimates suggest the 

telephone channel in our setting on average provides 20 times more information 

than the Web channel and that the Web on average provides information need 

instead of information gain – Web in our setting seems to confuse customers. 

3 - Customer Response to Short-Term Price Fluctuations 

Margaret Pierson, Assistant Professor, Tuck at Dartmouth, 

Hanover, NH, United States of America, mpierson@hbs.edu 

Transactions at 15 gasoline stations in Massachusetts over two years are used to 

estimate consumer elasticity with respect to price on a transaction basis. A 

significant proportion of the observed transactions display high sensitivity to 

price. This customer behavior results in congestion at pumps during high prices, a 

counterintuitive result which constrains stations’ price choices in practice. 

■ SA39 


Service Scripting and Value Co-Creation 



Chair: Gregory Heim, Associate Professor, Mays Business School at 

Texas A&M University, 320 Wehner Building, 4217 TAMU, College 

Station, TX, 77843-4217, United States of America, 

gheim@mays.tamu.edu 

1 - Scripting and Improvisation in Service Environments 

Enrico Secchi, University of Victoria, Peter B. Gustavson School of 

Business, Business & Economics Building Room 254, Victoria, BC, 

V8P 5C2, Canada, esecchi@clemson.edu, Rohit Verma, Aleda Roth 

Using the insights of the research on scripting and organizational improvisation, 

we develop an econometric model of the impact of different degrees of scripting 

and improvisation on services financial outcomes. We test our model using survey 

data and self reported financial performance from the hospitality industry. 

2 - Co-Creation of Value for IT-Enabled Services: 

A Case of Geocaching 

Tuure Tuunanen, University of Oulu, Department of Information 

Processing Science, P.O. 3000, 90014 University of Oulu, Finland, 

tuure@tuunanen.fi, Tero Vartiainen 

This study uses laddering interviews to explore how value is co-created for ITenabled 

services, more precisely geocaching. The results reveal that geocaching is 

distinctly hedonic in nature. The results also demonstrate how the consumer 

information systems framework can be applied for understanding which factors 

influence value co-creation for IT-enabled services. 

3 - Analysis of Role of Service Variety and Real Options in Resort 

Timeshare Service Value 

Gregory Heim, Associate Professor, Mays Business School at Texas 

A&M University, 320 Wehner Building, 4217 TAMU, 

College Station, TX, 77843-4217, United States of America, 


Although a major class of hospitality/travel experience service, little empirical 

research has examined timeshares or other fractional hospitality services. This 

paper examines whether service variety, in terms of breadth of vacation service 

experiences, or real options more strongly affect the perceived quality, valuation, 

and ownership of resort hotel timeshare services. We analyze hypotheses using 

both cross-sectional and longitudinal data sets.

■ SA40 


Dynamic Models for Sustainable Energy Policy 

Analysis 

Sponsor: Energy, Natural Res & the Environment/Energy 


Chair: Pedro Linares, Comillas P. University, Alberto Aguilera 25, 

Madrid, 28015, Spain, pedro.linares@upcomillas.es 

1 - Canadian Energy Supply and Demand with ENERGY 2020: 

Scenarios and Sensitivity Cases 

Matthew Hansen, Market Analyst, National Energy Board, 

444-7th Ave., SW, Calgary, AB, T3C 0T1, Canada, 

Matthew.Hansen@neb-one.gc.ca 

The National Energy Board periodically releases long-term projections of 

Canadian energy supply and demand as part of its Energy Futures series. Since 

2007, the ENERGY 2020 model has been central to this analysis. This 

presentation will describe how the National Energy Board uses ENERGY 2020 to 

create the scenarios and sensitivity cases featured in the Energy Futures reports. 

2 - Multi-stage Stochastic Optimization of an Energy System Model 

Joseph DeCarolis, Assistant Professor, North Carolina State 

University, 2501 Stinson Drive, Campus Box 7908, Raleigh, NC, 

27695, United States of America, jdecarolis@ncsu.edu, 

Kevin Hunter, Sarat Sreepathi 

We perform multi-stage stochastic optimization of a simplified energy system. The 

system consists of roughly 30 technologies used to supply primary energy 

resources, refine petroleum, generate electricity, and serve end use demands. The 

model incorporates crude oil, natural gas, and coal prices as stochastic parameters. 

Results indicate that the energy system requires little recourse as uncertainty is 

revealed over a multi-decadal time horizon. 

3 - Incorporating Endogenous Demand into Capacity Expansion 

Models for Developing Countries 

Rhonda Jordan, PhD Candidate, Massachusetts Institute of 

Technology, 143 Albany Street, #201A, Cambridge, MA, 02139, 

United States of America, rjordan@mit.edu 

Using a novel combination of approximate dynamic programming, system 

dynamics and linear programming, this research develops a power system 

capacity expansion model that incorporates both the technical details of power 

grid operation and the salient features of electricity demand in developing 

countries. Unlike traditional planning models, demand is represented 

endogenously. The capacity expansion strategies identified in this research are 

compared against that of more traditional approaches. 

4 - A Dynamic Model for Sustainable Energy Policy Making in 

Spain 

Alvaro Lopez-Peña, Comillas P. University, IIT, 

Santa Cruz de Marcenado 26, Madrid, 28015, Spain, 

alvaro.lopezpena@iit.upcomillas.es, Pedro Linares, 

Ignacio Pérez-Arriaga 

Today, there is an important need for sound sustainable energy policies. For 

assessing them, mathematical models are needed, which allow to represent 

energy systems’ evolution over the long term under different policy assumptions 

(trajectories). We present a dynamic model applied to Spain, which aims at 

representing realistic trajectories. By realistic, we mean that it takes into account 

important aspects of energy systems such as imperfect markets, investor’s myopia 

and risk aversion. 

■ SA41 


Models in Energy Investment 


Chair: Carlos Ruiz, École Centrale Paris and SupÈlec, Grande Voie des 

Vignes, Ch‚tenay-Malabry, 92295, France, carlosruizmora@gmail.com 

1 - On the Value of Improved Models for the Electricity Sector 

Alberto Lamadrid, Graduate Student, Cornell University, 

419 Warren Hall, Ithaca, NY, 14850, United States of America, 

ajl259@cornell.edu, Tim Mount, Ray Zimmerman, Robert Thomas 

This paper measures the value of the stochastic solution for a system with high 

penetration of renewable energy sources, contrasting it to a deterministic 

formulation with fixed reserves, as used by System Operators (SO). The 

performance analysis focuses on measures of the system economic costs. Our 

model has a SO minimizing the expected cost of providing both energy and 

ancillary services with a security constrained OPF. Uncertainty is reproduced 

using a Markovian transition probability matrix. 


69 

SA42 

2 - Mathematical Model for Planning Investments in Energy 

Sources for Argentina 

Julio Rolando Flores, Becario, INGAR, Eva Peron 2660, 

Santa Fe, SF, 3000, Argentina, jflores@santafe-conicet.gov.ar, 

Jorge M. Montagna, Aldo Vecchietti 

This paper presents a mathematical programming model for planning investment 

in energy sources for Argentina. The model considers the use of renewable and 

not renewable sources and uncertainties in demands, costs, and the amount of 

reserves of fossil fuels. The objective function of is to minimize the investment 

and operation costs of the energy supply. The results provide the visualization of 

the investments made: time periods in and their amounts and also how the 

energy matrix is affected. 

3 - Biomass Based Energy Logistics System 

Wooseung Jang, Professor, University of Missouri, E3437 Lafferre 

Hall, Columbia, MO, 65211, United States of America, 

jangw@missouri.edu, Rana Afzali Baghdadabadi, Cerry Klein, 

James Noble, Zhongwei Yu, Kurt Ehlers 

We develop an optimization model supporting the multi-stage decision making of 

infrastructure investment and operation of the biomass logistics system. In 

addition, we build a database for biomass availability, cost and other supply 

information. Our work is applied to the biomass power plants scheduled to 

operate in Missouri. 

4 - Transmission Expansion Planning using an AC Model: 

Formulations and Possible Relaxations 

Hui Zhang, Research Associate, Arizona State University, 1223 S 

Dorsey Ln, Apt 101, Tempe, AZ, 85281, United States of America, 

hui.zhang@asu.edu, Hans Mittelmann, Vijay Vittal, Gerald Heydt 

Starting with two MINLP models, this paper explores the possibility of applying 

AC-based models to the transmission expansion planning (TEP) problem. Two 

NLP relaxation models are then proposed by relaxing the binary decision 

variables. A reformulation-linearization-technique (RLT) based relaxation model 

is also presented in the paper. The simulation results show that by using proper 

reformulations or relaxations, it is possible to apply the AC models to TEP 

problems and obtain a good solution. 

5 - Robust Reliability Optimization by System Design and 

Components Allocation 

Carlos Ruiz, École Centrale Paris and Supèlec, 

Grande Voie des Vignes, Châtenay-Malabry, 92295, France, 

carlosruizmora@gmail.com, Rodrigo Mena, Enrico Zio, Yanfu Li 

We address the problem of power system design and components allocation 

under multiple sources of uncertainty. We consider a set of fixed buses with 

uncertain demand loads and we seek to identify the electrical lines and generating 

units that should be installed in order to minimize the system operational cost 

and maximize its reliability. The optimization relies on a bilevel model that is 

robust against generation capacity variability as well as unit and line failures. 

■ SA42 


Strategic Issues in Freight Transportation 


Chair: Zhougeng Lin, PhD Student, University of Portsmouth, Lion 

Gate Building, Lion Terrace, Portsmouth, PO1 3HF, United Kingdom, 

Zhougeng.Lin@Port.Ac,Uk 

1 - Mixed Fleet Dispatching Optimization in Truckload 

Transportation 

Hector A. Vergara, Visiting Assistant Professor, Oregon State 

University, 204 Rogers Hall, Corvallis, OR, 97331, 

United States of America, vergarah@onid.orst.edu, Sarah Root 

In a mixed fleet dispatching system for truckload transportation some truckloads 

are sent direct point-to-point while others are routed through a network of relay 

points where drivers are exchanged allowing them to return home more 

frequently. We propose a mathematical formulation for strategic relay network 

design and dispatch mode selection that incorporates several operational 

constraints within the variable definition. Computational results are presented 

along with areas for future research. 

2 - Freight Network Design with Pricing and Economies of Scale 

Panagiotis Ypsilantis, PhD Candidate, RSM, Erasmus University, 

Burgemeester Oudlaan 50, Rotterdam, Netherlands, 

pypsilantis@rsm.nl, Rob Zuidwijk 

The freight service network design in a strategic level is formulated as a bi-level 

mathematical program; the pricing, design and economies of scale are considered 

in a competitive environment. The net revenues of a TOC that operates barge and 

train connections to inland terminals according to the extended gate concept are 

maximized while the total costs faced by the users of the network are minimized. 

A solution approach for this problem and some numerical results are presented 

and discussed.

SA43 

3 - Freight Origin-destination Estimation using Bayesian Networks 

Based on Multiple Data Sources 

Yinyi Ma, Erasmus University Rotterdam, Burg. Oudlaan 50, 

Rotterdam, Netherlands, yma@rsm.nl, Roelof Kuik, 

Henk van Zuylen 

Freight origin-destination (OD) estimation becomes increasingly important. 

Traditional general traffic OD estimation methods based on loop detectors have 

the under-specification problem. This paper investigates a Bayesian Networks 

approach for estimating freight OD-matrices by Markov-chain-Monte-Carlo 

methods, thereby mitigating under-determinacy by exploiting proportionallycontributing 

multiple data sources, such as camera, Bluetooth and 

Weigh-in-Motion. 

4 - Simulation and Multi-objective Optimization Combining 

Framework for Container Terminal Optimization 

Zhougeng Lin, PhD Student, University of Portsmouth, Lion Gate 

Building, Lion Terrace, Portsmouth, PO1 3HF, United Kingdom, 

Zhougeng.Lin@Port.Ac,Uk, Dylan Jones 

This presentation proposes a combining framework to integrate simulation and 

multi-objective optimization(MOO) for better description of dynamic real-world 

systems in low cost and reasonable computational time. A case study on container 

terminal equipment optimization is discussed to demonstrate its applications. 

Results are compared by post-MOO structure and integrated MOO structure. 

■ SA43 


RAS Student Paper Award 

Sponsor: Railway Applications 


Chair: Juan Morales, BNSF Railway, TX, United States of America, 

Juan.Morales@BNSF.com 

1 - RAS Student Paper Award 

Juan Morales, BNSF Railway, TX, United States of America, 

Juan.Morales@BNSF.com 

Finalists of the 2012 RAS Student Paper Award will present their work in this 

session. Finalists were not determined prior to the abstract submission deadline of 

the printed programs. 

■ SA44 


Supply Chain Structures and Incentives 


Chair: Suleyman Demirel, PhD Candidate, University of Michigan, 

701 Tappan St., Ann Arbor, MI, 48109, United States of America, 

sdemirel@umich.edu 

1 - Agile Supply Chain Building via Semantic Supplier Search 

Jaehun Lee, POSTECH, 312, Eng4, POSTECH, Hyoja, Pohang, 

Korea, Republic of, jaehun_lee@postech.ac.kr, Taejong Yoo, 

Hyunbo Cho, Bohyun Kim 

The goal of a supply chain in manufacturing domain is to guarantee that the right 

amount of the right product is in the right place at the right time. To accomplish 

this, an agile supply chain formation is critical. We propose a semantic 

manufacturing supplier discovery framework that structuralizes buyer’s demand 

requirements and supplier’s capability information to assist in building such a 

supply chain. 

2 - Supply Network Structure and Firm Innovation: 

An Empirical Investigation 

Marcus Bellamy, Georgia Institute of Technology, 800 West 

Peachtree ST NW, Atlanta, GA, 30308, United States of America, 

marcus.bellamy@gatech.edu, Soumen Ghosh, Manpreet Hora 

We investigate the impact of supply network structure on firm innovation. Using 

a network-analytic lens, we develop and test our hypotheses linking firm level 

network properties and innovation. Our analysis considers how firm innovation is 

affected through the extended network it shares between each entity in the 

supply network. 


70 

3 - The Effect of a Temporary Product Distribution Channel on 

Supply Chain Performance 

Moutaz Khouja, Professor, University of North Carolina-Charlotte, 

9201 University City Blvd., Charlotte, NC, 28223, 

United States of America, mjkhouja@uncc.edu 

We analyze a supply chain of a manufacturer and two retailers. The first retailer 

always stocks the product. The second retailer is a deal-of-the day (DOTD) retailer 

who sells the product for a short time of few days at a discount. The DOTD 

retailer sells the product at a discount due to a price discount she gets for ordering 

a large quantity. We identify the discounted wholesale price and the discount 

order quantity. We find that he temporary channel can increase the profit of the 

manufacturer. 

4 - Revisiting Interorganizational Trust: Is More Always Better or 

Could More Be Worse? 

Veronica H. Villena, Assistant Professor, Pennsylvania State 

University, 410 Business Building, University Park, State College, 

PA, 16802, United States of America, vhv1@psu.edu, 

Thomas Choi, Elena Revilla 

We conduct a more comprehensive examination of trust by considering both its 

positive and negative effects and by investigating how they function differently 

under different types of uncertainties.We use data from a sample of 133 buyersupplier 

relationships. The results suggest that the negative effects of trust become 

more notable when there is high behavioral uncertainty between buyers and 

suppliers and less evident when there is high environmental uncertainty for 

buyers. 

■ SA45 


JFIG Paper Competition I 

Sponsor: Junior Faculty Interest Group 


Chair: Esra Buyuktahtakin, Assistant Professor, Wichita State 

University, IME Department, Wichita, KS, United States of America, 

esra.b@wichita.edu 

1 - JFIG Paper Competition 

Esra Buyuktahtakin, Assistant Professor, Wichita State University, 

IME Department, Wichita, KS, United States of America, 


Papers are submitted for this year’s JFIG paper competition, and each one is 

evaluated based on the importance of the topic, appropriateness of the research 

approach, and the significance of research contribution. In this session, finalistsselected 

in two rounds of review, will present their papers. For the selected 

finalists and the abstracts of the selected papers, please refer to the online 

program. 

■ SA46 


Organizational and Industry Renewal (I): Innovation, 

Industry Context, and Population-Level Learning 

Sponsor: Organization Science 


Chair: Andreas Schwab, , Associate Professor, Iowa State University, 

3315 Gerdin, Ames, IA, 50011, United States of America, 

aschwab@iastate.edu 

1 - Exploring Organizational Identity and Organizational Design in 

Technological Transitions 

Mary Tripsas, Associate Professor, Harvard Business School, 

219 Rock Center, Soldiers Field Rd, Boston, MA, 02163, 

United States of America, mtripsas@hbs.edu 

Innovation research suggests that a separate new technology unit with its own 

identity increases the odds of a successful technological transformation. Identity 

research, however, has shown that multiple identities can be harmful, posing a 

dilemma. Through a comparison of FujiFilm and Polaroid’s reactions to digital 

imaging, we propose that establishing a separate new technology unit can cause 

dormant identity conflicts to surface, and reconciling those conflicts is critical to 

success.

2 - Ties that Bind? The Differential Ability of de alio and de novo 

Firms to Leverage External Relations 

Glenn Hoetker, Associate Professor, W.P. Carey School of Business, 

Arizona State University, P.O. Box 874006, Tempe, AZ, 85287- 

4006, United States of America, Glenn.Hoetker@asu.edu, Pao-Lien 

Chen, Rajshree Agarwal 

External relationships may help a firm learn about disruptive technological 

changes, but may also impede the firm’s adopting new technologies. We examine 

the impact of a firm’s pre-entry experience on its dynamic capability to maximize 

learning and minimize inertia created by ties to suppliers. We find that de alio 

firms better harness information available through a larger network of suppliers 

than de novo firms, but are equally constrained by the length of prior 

relationships. 

3 - Whose Side Are You On? The Role of States in Influencing 

Healthcare Service Delivery Innovations 

Scott Feyereisen, 4666 East Camino Rosa, Tucson, AZ, 85718, 

United States of America, feyerbal@email.arizona.edu, 

Joseph Broschak 

We explore why all U.S. states would not adopt a seemingly rational, efficient and 

cost-reducing policy innovation. Using institutional and resource dependence 

theory, we develop and test hypotheses related to a state’s propensity to adopt 

contested healthcare legislation. Professional power and heterogeneity, state 

resources and prior state adoptions are hypothesized to affect a state’s likelihood 

of adopting a policy change. We test our predictions using data on anesthesia 

policy adoptions. 

4 - Population-Level Learning: Industry-Level Variability of 

Learning Outcomes from Innovations 

Andreas Schwab, Associate Professor, Iowa State University, 


aschwab@iastate.edu, Jay O’Toole, Anne Miner 

This study investigates industry-level variability in learning outcomes following 

the implementation of an innovation. We challenge the intuition that early 

periods are marked by more variation than later periods. Contingency factors that 

shape variability include varied imitation rules, deliberate innovation, and 

interactions across innovation features and levels of learning. We test hypotheses 

using historic data on the farm-system, an administrative innovation in baseball 

industry. 

■ SA47 

47- North 230- CC 

New Advances in Traffic Operations 

Sponsor: Transportation Science & Logistics/ Intelligent 

Transportation Systems (ITS) 


Chair: Carolina Osorio, Assistant Professor, Massachusetts Institute of 

Technology, Room 1-232, 77 Massachusetts Avenue, Cambridge, MA, 

02139, United States of America, osorioc@mit.edu 

1 - Combining Metamodel Techniques and Bayesian Selection 

Procedures to Derive Computationally Efficient Traffic 

Management Algorithms 

Hoda Bidkhori, Massachusetts Institute of Technology, 

77 Massachusetts Ave., Room 1-172, Cambridge, MA, 02139, 

United States of America, bidkhori@mit.edu, Carolina Osorio 

This talk introduces simulation-based optimization (SO) methods to address the 

problems in Traffic Management. It proposes an SO technique that uses 

metamodel information together with a Bayesian framework, where the 

parameters of the prior distributions are estimated based on probabilistic 

metamodel information. In order to derive an SO algorithm that achieves a good 

trade-off between detail, realism and computational efficiency, the metamodel 

combines information from a highresolution traffic simulator with information 

from a lower-resolution yet computationally efficient analytical queueing 

network model. We use this approach to address an urban traffic management 

problem using a detailed microscopic traffic simulator of the Swiss city of 

Lausanne. 

2 - Application of Online DTA Framework for Weather-related 

Traffic Management 

Hani S. Mahmassani, Northwestern University, Evanston, IL, 

60208, United States of America, masmah@u.northwestern.edu, 

Ying Chen, Jiwon Kim, Lan Jiang 

Weather-related traffic management through variable message signs, speed 

control and active demand management is an important application for predictive 

traffic estimation and prediction based on online DTA deployments. A framework 

and associated methodology is presented and demonstrated through several 

applications. 


71 

SA48 

3 - On Observable Chaotic Maps for Transportation 

Queueing Analysis 

Joseph Y.J. Chow, Assistant Professor, Ryerson University, 

350 Victoria Street, Toronto, ON, M5B 2K3, Canada, 

joseph.chow@ryerson.ca 

Chaotic mapping can capture a number of characteristics in queues: transient 

behavior, intermittency, steady state, and complex distributions. The proposed 

queueing model uses chaotic mapping of inter-arrival times to generate arrivals so 

that parameters can be calibrated with observable data. A sample mapping based 

on the logistic map is presented and shown to be ergodic. A joint parameter and 

state estimation algorithm is presented. The advantages are studied with two 

connected nodes. 

■ SA48 


Maritime Transportation 


Chair: Dimitri Papageorgiou, PhD Candidate, Georgia Tech, 

765 Ferst Ave, Atlanta, GA, 30332, United States of America, 

djpapag@gatech.edu 

1 - Evaluation of Intermodal Marine Container Terminal Gate 

Strategies via Simulation 

Jeffery Karafa, University of Memphis, 104 Engineering Science 

Bldg, 3815 Centr, 3815 Central Avenue, Memphis, TN, 38152, 

United States of America, jefferykarafa@yahoo.com, Mihalis Golias 

We present the development of a traffic simulation model to measure the impact 

of gate strategies on congestion levels at intermodal marine container terminal 

gates. The traffic model is used to quantify both travel time and delay and 

emission levels at the terminal gates before and after gate strategies have been 

implemented. 

2 - Equilibrium Model for Analysis on Opening and Improving 

Transportation Routes 

Kazuhiko Ishiguro, Kobe University, 5-1-1, Fukaeminami, 

Higashinada, Kobe, 658-0022, Japan, 

ishiguro@maritime.kobe-u.ac.jp, Pitu Mirchandani 

Transport cost reduction will trigger new transport demand. This study develops 

international trade model taking account of ocean carriers’ behavior based on 

multi-regional computable general equilibrium framework. Trade amount is 

formulated as the demand of goods produced in other region. The model is 

applied to opening and improving transportation routes and the impacts on trade 

are discussed. 

3 - Quickly Finding Good Solutions to Large-scale Maritime 

Inventory Routing Problems 

Dimitri Papageorgiou, PhD Candidate, Georgia Tech, 

765 Ferst Ave., Atlanta, GA, 30332, United States of America, 

djpapag@gatech.edu 

We describe aggregation and decomposition techniques for quickly finding good 

solutions to large-scale maritime inventory routing problems. Solving the LP 

relaxation of the “standard” mixed-integer programming formulation of these 

problems can require hours to solve, assuming they can be loaded into memory. 

Meanwhile, our goal is to find high quality solutions in minutes. We present 

methods that accomplish this objective and compare them to two benchmarks 

that have been used in the literature.

SA49 

■ SA49 


Joint Session TSL/SPPSN: Geographic, Building and 

Pedestrian Evacuation Modeling 

Sponsor: Transportation Science & Logistics & Public Programs, 

Service and Needs 


Chair: Douglas Bish, Assistant Professor, Virginia Tech, Department of 

Industrial and Systems Eng., 250 Durham Hall, Blacksburg, VA, 24061- 

0118, United States of America, drb1@vt.edu 

1 - Simulating Pedestrian Evacuation from Large Events 

Pitu Mirchandani, Arizona State University, 699 South Mill 

Avenue, Tempe, AZ, 85287, United States of America, 

pitu@asu.edu, Ning Wang, Zhuoyang Zhou 

At the end of a large event, attendees move from their seats to an exit, and then 

walk over to the vehicle, which they subsequently use to reach their ultimate 

destination. This paper develops a simulation model, using the pedestrian feature 

in VISSIM, to evaluate delays in normal and no-notice evacuation scenarios. 

Evacuation strategies are studied to make total evacuation time to leave the 

building or stadium and reach respective destinations as small as possible. 

2 - Pedestrian Route Choice Modeling in a Changing 

Physical Environment 

Lei Feng, University of Maryland, College Park, MD, 

United States of America, lfeng@umd.edu, Elise Miller-Hooks 

The selection of routes in a congested pedestrian network in which the physical 

environment is changing is explored. Models of route choice behavior in this 

environment are developed. The evolving physical conditions, improving 

pedestrian situational awareness, responsively changing goals, and individual and 

collective responses are explicitly recognized in these models of route choice. 

3 - Fleet Management and Equity Measures in Evacuation 

Transportation Planning for Multiple Hospitals 

Esra Agca, PhD Student, Virginia Tech, Department of Industrial 

and Systems Eng., 250 Durham Hall, Blacksburg, VA, 24061, 

United States of America, esra.vt@gmail.com, Douglas Bish, 

Roger Glick 

A coordinated response is required for to effectively evacuate multiple hospitals 

due to a regional threat (e.g., a hurricane). Organizing the evacuation response 

can be more complicated when it involves more than one hospital system (groups 

of hospitals under the same management). In this research we study optimal 

evacuation plans for multiple hospitals, including full centralized control, and the 

study of equity issues at various levels, including at the hospital system level. 

4 - Optimizing a Large Scale Evacuation Plan for Tucson, AZ 

Neema Nassir, Graduate Research Assistant, University of Arizona, 

Department of Civil Engineering and Engineering Mechanics, 

1209 E. Second Street, Tucson, AZ, 85721-0072, 

United States of America, neeman@email.arizona.edu, 

Hong Zheng, Mark Hickman, Yi-Chang Chiu 

Every year hundreds of chlorine tankers pass through downtown Tucson. This 

study aims to optimize evacuation under a hypothetical tanker spill. Current gas 

dispersion models indicate that such a spill may lead to deadly consequences for 

many persons in this large urban area. In our evacuation model, the objective is 

to minimize the total exposure of the citizens to the poisonous gas, taking into 

account the different concentration levels of gas over the space. 

■ SA50 


Military Applications 

Sponsor: Military Applications 


Chair: Sung-Pil Kim, Yonsei University / R.O.K Navy, Yonseiro 50, 

Seoul, Korea, Republic of, senseksp@gmail.com 

1 - Modeling Multi-echelon Inventories for Naval Repair Parts in 

the ROK Navy 

Sung-Pil Kim, Yonsei University/R.O.K Navy, Yonseiro 50, Seoul, 

Korea, Republic of, senseksp@gmail.com 

In comparison to the rapid modernization of defense equipment, government 

resources are restricted, necessitating an efficient solution for managing all 

national resources, including naval repair parts.This paper investigates the 

modeling of efficient multi-echelon inventory management of naval repair parts 

in the ROK Navy. 


72 

2 - Once More: Putting the “O” Back into Military 

Operations Research 

George Kuhn, LMI, 2000 Corporate Ridge, McLean, VA, 22102, 

United States of America, GKuhn@lmi.org 

Research results are described for two core military management topics: 

estimating Forces personnel attrition (casualty estimation), and forecasting unitlevel 

requirements for aircraft (helicopter) spare parts. While being clearly 

different planning domains on one level, at another level the difficulties that have 

long burdened estimating/forecasting such requirements share a common root: 

unless data collection and analysis are grounded on capturing and reflecting the 

inherent character of military operations, estimates/forecasts will inevitably fail to 

describe requirements accurately. Evidence is presented that, properly grounded 

on describing patterns of military operations, quantitative analysis can project 

requirements accuratelyóand these results differ markedly from standard 

portrayals that rely on conceptual and quantitative conventions incongruent with 

the underlying character of military operations. 

■ SA51 


Disaster Relief Planning and Operations 



Chair: Xiaofeng Nie, Assistant Professor, Nanyang Technological 

University, 50 Nanyang Avenue, Singapore, 639798, Singapore, 

xiaofengnie@ntu.edu.sg 

1 - Prepositioning of Supplies in Preparation for a Hurricane with 

Forecast Information Updates 

Rajan Batta, University at Buffalo (SUNY), Dept. of Industrial & 

Systems Engg, Buffalo, NY, 14260, United States of America, 

batta@buffalo.edu, Gina Galindo 

It is possible to use prediction models to obtain periodic forecasts about a 

hurricane’s main characteristics such as its path, wind force and time of landfall. 

Using this information we present a dynamic model for prepositioning supplies in 

such a setting. Our model has two distinguishing features: possible destruction of 

supply points during the disaster event, and the use of forecast information 

updates in order to strategically modify and improve the prepositioned quantities 

and locations. 

2 - A Stochastic Programming Model for Casualty Response 

Planning during Catastrophic Health Events 

Aakil Caunhye, PhD Student, Nanyang Technological University, 

758, Choa Chu Kang North 5, Singapore, 680758, Singapore, 

M090028@e.ntu.edu.sg, Xiaofeng Nie 

Catastrophic health events are natural or man-made incidents that create 

casualties in numbers that overwhelm healthcare response capabilities. We 

construct a three-stage stochastic programming model to locate alternative care 

facilities and transport casualties in response to such events. We also propose an 

L-shaped cut algorithm, to generate good solutions fast. We analyse the model 

and algorithm in the case of a high-impact earthquake in a subregion of San 

Bernardino, California. 

3 - Identifying the Role and Impact of Social Media in Evacuation 

Justin Yates, Assistant Professor, Texas A&M University, 4079 ETB, 

3131 TAMU, College Station, TX, 77843-3131, United States of 

America, jtyates@tamu.edu 

The next generation of evacuation planning and emergency response strategies 

needs to be built on the context of real-time information dissemination. 

The goal of this research is to build a new set of evacuation models capable of 

incorporating spatio-temporal data on extremely large scales and supported by 

new behavioral paradigms predicated on the use of social media. We discuss our 

initial efforts and insights in this presentation. 

4 - Heuristic Methods for Hazard Zone Determinations 

Ed Pohl, University of Arkansas, Department of Industrial 

Engineering, Fayetteville, AR, United States of America, 

epohl@uark.edu, Manuel D. Rossetti 

To better examine the use of optimization in emergency planning this research 

investigates the determination of hazard zones within a geographic area. A hazard 

zone is an area corresponding to a worst-case evacuation scenario and is a 

potential vulnerability that should be addressed by risk mitigation strategies. The 

research compares two heuristic approaches to determining hazard zones based 

on population and road connectivity within a spatial network.

■ SA52 


Electricity Markets: Game-theory Based Models for 

Strategic Planning 



Chair: Efraim Centeno, Institute for Research in Technology - ICAI, 

Comillas Pontifical University, Alberto Aguilera 25, Madrid, 28015, 

Spain, Efraim.Centeno@upcomillas.es 

1 - Study of Strategic Behavior Impact in the Spot Market on 

Capacity Expansion using a Linearised EPEC 

Efraim Centeno, Institute for Research in Technology-ICAI, 

Comillas Pontifical University, Alberto Aguilera 25, Madrid, 28015, 

Spain, Efraim.Centeno@upcomillas.es, Sonja Wogrin, 

Julian Barquin 

Capacity expansion decisions are represented using a bilevel approach (EPEC). In 

the upper level the competing generation companies maximize their individual 

profits, while the lower level represents the spot markets using a conjectured 

price-response approach. This model is linearised and solved to show the solution 

variation with respect to the level of competence in the spot market and that the 

market equilibrium can have multiple solutions with investments level that vary 

significantly. 

2 - Impact of Renewable Energy Sources on Generation Capacity 

Investments: A Stochastic MPEC Approach 

Sonja Wogrin, Institute for Research in Technology-ICAI, Comillas 

Pontifical University, Alberto Aguilera 25, Madrid, 28015, Spain, 

sonja.wogrin@iit.upcomillas.es, Julian Barquin, 

Efraim Centeno 

Strategic long-term generation capacity planning in liberalized electricity markets 

is a very complicated task, mainly because generation companies are confronted 

with a vast spectrum of uncertainties, one of the most prominent being demand 

fluctuations stemming from renewable energy sources. We propose a stochastic 

bilevel model formulated as an MPEC to analyze the impact that renewableinduced 

uncertainty has on capacity investment decisions and the optimal 

technology mix in the system. 

3 - A Three-level Model for Generation and Transmission 

Expansion Planning 

Javier Contreras, Professor, University of Castilla-La Mancha, 

E.T.S. de Ingenieros Industriales, Ciudad Real, 13071, Spain, 

Javier.Contreras@uclm.es, Enzo Sauma, David Pozo 

We present a three-level model for the expansion of an electric network. The 

lower-level model represents the equilibrium of a pool-based market; the 

intermediate level represents the Nash equilibrium in generation expansion, and 

the upper-level model represents the anticipation of transmission expansion to 

the investment in generation and the pool-based market equilibrium. The model 

is applied to a realistic power system in Chile to illustrate the methodology and 

proper conclusions are reached. 

4 - Dynamic Pricing of Peak Generation in Electricity Markets 

Heikki Peura, PhD Student, London Business School, 

Regent’s Park, London, NW1 4SA, United Kingdom, 

hpeura.phd2010@london.edu, Derek Bunn 

With reference to electricity, we study the problem of pricing peak production of 

an instantly perishable product in a dynamic setting with market power. In a 

repeated game with future demand uncertainty, peak producers’ profitability may 

hinge on their ability to implicitly collude on prices. Furthermore, diversification 

in generation technologies as well as financial constraints in the form of revenue 

targets may have significant impacts on peak producers’ pricing decisions. 


73 

■ SA53 


Optimization in Energy Procurement and 

Cost Allocation 



Chair: Ross Baldick, Professor, The University of Texas at Austin, 

1 University Station Stop C0803 ENS 502, Austin, TX, 78712, 

United States of America, baldick@ece.utexas.edu 

1 - Energy Supply Risk Due to Selling Over the Physical 

Generation Capacity 

Anderson Rodrigo de Queiroz, PhD, MC&E, Rua Silvestre Ferraz 

N.81, Itajubá, 37500-000, Brazil, ar_queiroz@yahoo.com.br, Josè 

W. Marangon Lima, Luana M. Marangon Lima, 

Verònica Etchebehere 

In Brazil the electricity market agents have to present full physical generation 

coverage, in other words only the generator’s assured energy can be negotiated. 

This work explores the possibility of financial leverage in terms of energy by 

allowing the negotiation of contracts that extrapolates the physical capacity. 

Short-term price simulations are used to value contracts and to study risk. The 

procedure gives more flexibility and liquidity to the market amplifying the 

business possibilities. 

2 - Game-theoretic Allocation of Transmission Costs Taking into 

Account Generation and Load Uncertainties 

Luiz Carlos Costa Jr, PSR, Praia de Botafogo 228/1701-A, 

Rio de Janeiro, RJ, 22250-145, Brazil, luizcarlos@psr-inc.com, 

Luiz Mauricio Thomè, Silvio Binato, Mario Veiga Pereira 

We extend the Aumann-Shapley allocation scheme to represent uncertainties in 

bus loads and energy production of renewables. Two criteria are compared: a 

marginal pricing scheme, in which the cost of each circuit is allocated to the 

scenario that led to the maximum use of that circuit; and a “tolling” scheme, in 

which the cumulative distribution of circuit usage is taken into account. The 

methodology will be illustrated for the 5,000 bus Brazilian transmission network. 

3 - Stochastic Model of International Fuel Procurement for Coalfired 

Plants in Hydrothermal Systems 

Fernanda Thomé, Research Engineer, PSR, Praia de Botafogo 

228/1701-A, Rio de Janeiro, RJ, 22250145, Brazil, 

fernanda@psr-inc.com, André Dias, Silvio Binato, Eduardo Faria 

This work presents an optimal strategy for international fuel procurement, 

transportation and storage in order to ensure continuous supply of coal-fired 

thermal plants. A stochastic programming problem takes into account the 

uncertainty associated with inflows in a hydrothermal system and other complex 

logistical and commercial aspects related to imported coal handling. The model 

results are shown by comparisons between different strategies when applied to 

different dispatch scenarios. 

■ SA54 

SA54 

54- Regency Ballroom A- Hyatt 

Nicholson Student Paper Prize I 

Cluster: Nicholson Student Paper Prize 


Chair: Ashish Goel, Stanford University, Stanford, CA, 

United States of America, ashishg@stanford.edu 

1 - A Polyhedral Study of Multiechelon Lot Sizing with 

Intermediate Demands 

Minjiao Zhang, Graduate Research Associate, The Ohio State 

University, Integrated Systems Engineering, 1971 Neil Ave., 

Columbus, OH, 43210, United States of America, 

zhang.769@osu.edu, Simge KüÁükyavuz, Hande Yaman 

In this paper, we study a multiechelon uncapacitated lot-sizing problem in series 

(m-ULS), where the output of the intermediate echelons has its own external 

demand and is also an input to the next echelon. We propose a polynomial-time 

dynamic programming algorithm, which gives a tight, compact extended 

formulation for the two-echelon case (2-ULS). Next, we present a family of valid 

inequalities for m-ULS, show its strength, and give a polynomial-time separation 

algorithm. We establish a hierarchy between the alternative formulations for 2- 

ULS. In particular, we show that our valid inequalities can be obtained from the 

projection of the multicommodity formulation. Our computational results show 

that this extended formulation is very effective in solving our uncapacitated 

multi-item two-echelon test problems. In addition, for capacitated multiitem, 

multiechelon problems, we demonstrate the effectiveness of a branch-and-cut 

algorithm using the proposed inequalities.

SA55 

2 - Robust Queueing Theory 

Chaithanya Bandi, PhD Student, Massachusetts Institute of 

Technology, Cambridge, MA, 02139, United States of America, 

cbandi@mit.edu, Dimitris Bertsimas, Nataly Youssef 

We propose an alternative approach for studying queueing systems by employing 

robust optimization as opposed to stochastic analysis. While traditional queueing 

theory relies on Kolmogorov’s axioms of probability and models arrivals and 

services as renewal processes, we use the limit laws of probability as the axioms of 

our methodology and model the queueing systems primitives by uncertainty sets. 

In this framework, we obtain closed form expressions for the waiting times in 

multi-server queues with heavy-tailed arrival and service processes. These 

expressions are not available under traditional queueing theory for heavy-tailed 

processes, while they lead to the same qualitative insights for independent and 

identically distributed arrival and service times. We also develop an exact calculus 

for analyzing a network of queues with multiple servers based on the following 

key principle: a) the departure from a queue, b) the superposition, and c) the 

thinning of arrival processes have the same uncertainty set representation as the 

original arrival processes. We show that our approach, which we call the Robust 

Queueing Network Analyzer (RQNA) a) yields results with error percentages in 

single digits (for all experiments we performed) relative to simulation, b) 

performs significantly better than the Queueing Network Analyzer (QNA) 

proposed in Whitt [1983], and c) is to a large extent insensitive to the number of 

servers per queue, the network size, degree of feedback, traffic intensity, and 

somewhat sensitive to the degree of diversity of external arrival distributions in 

the network. 

3 - New Functional Characterizations and Optimal Structural 

Results for Assemble-to-order M-Systems 

Emre Nadar, Carnegie Mellon University, Tepper School of 

Business, Pittsburgh, PA, United States of America, 

enadar@andrew.cmu.edu, Alan Scheller-Wolf, Mustafa Akan 

We consider an assemble-to-order M-system with multiple components, multiple 

products, batch ordering of components, random lead times, and lost sales. We 

model the system as an infinite-horizon Markov decision process and seek an 

optimal control policy: a control policy specifies when a batch of components 

should be produced, and whether an arriving demand for each product should be 

satisfied.We introduce new functional characterizations for submodularity and 

supermodularity restricted to certain subspaces. These enable us to characterize 

optimal inventory replenishment and allocation policies under a mild condition 

on component batch sizes via a new type of policy: lattice-dependent base-stock 

production and lattice-dependent rationing. 

■ SA55 

55- Regency Ballroom B - Hyatt 

Supply Chain and Operations Issues in 

Emerging Economies 

Cluster: Operations Research in Emerging Economies 


Chair: Mili Mehrotra, University of Minnesota, 321 19th Ave. South, 

Minneapolis, MN, United States of America, milim@umn.edu 

Co-Chair: Tharanga Rajapakshe, Assistant Professor, University of 

Florida, Warrington School of Business Administra, Gainesville, FL, 

32606, United States of America, 

Tharanga.rajapakshe@warrington.ufl.edu 

1 - Improving Agricultural Productivity: Impact of Minimum 

Support Prices 

Tharanga Rajapakshe, Assistant Professor, University of Florida, 

Warrington School of Business Administra, Gainesville, FL, 32606, 

United States of America, 

Tharanga.rajapakshe@warrington.ufl.edu 

A Minimum Support Price (MSP) for an agricultural crop is a guaranteed price, 

regardless of volume, at which a governmental entity agrees to purchase the crop 

from farmers. This work is an effort to understand the influence of MSP on a 

homogenous farming community cultivating a single crop. We analyze the impact 

of production cost and yield uncertainty on the MSP. Our results explain the role 

of MSP in improving productivity and social utility. 

2 - Improving the Milk Supply Chain in Developing Countries: 

Analysis under Competing Intermediaries 

Liying Mu, University of Texas at Dallas, 800 W. Campbell Road, 


muliying@utdallas.edu, Milind Dawande, Vijay Mookerjee 

Quality issues in milk – arising primarily from deliberate adulteration by 

producers — have been reported in many developing countries. High testing cost 

and competition between milk stations (which act as intermediaries by buying 

milk from producers and selling it to firms) are identified as the main causes for 

the low quality of milk. We propose a mechanism to incentivize high-quality milk 

with minimum testing, in the presence of competing milk stations. 


74 

3 - A Stochastic Optimization Based Decision Support System for 

Planning in a Process Industry 

Goutam Dutta, Professor, Indian Institute of Management, 

Production Quantitive Methods Area, Ahmedabad, Gujarat, Gu, 

380015, India, goutam@iimahd.ernet.in, Narain Gupta, 

Robert Fourer 

We introduce a multi-period, multiple scenario optimization based decision 

support system. The DSS is based on a two stage stochastic linear program with 

recourse for strategic planning in process industry. The model maximizes expected 

contribution subject to material balance, facility capacity, and constraints for nonanticipativity. 

We describe how the database structure for the stochastic 

optimization model is different in comparison to the multi-period deterministic 

model. 

4 - Agricultural Cooperatives for Alleviating Poverty: 

Models and Analyses 

Jaehyung An, PhD Student, UCLA Anderson School of 

Management, 11140 Rose Ave. #405, 90034, Los Angeles, CA, 


jaehyung.an.2013@anderson.ucla.edu, Soo-Haeng Cho, 

Christopher Tang 

In emerging markets such as Brazil, China and India, the agricultural sector plays 

an important role in their economic growth even though many farmers in these 

markets are trapped in the poverty cycle. To break this vicious cycle, government 

agencies, Non-Governmental Organizations, and social enterprises are helping 

farmers to establish agricultural cooperatives (coops). In this paper, we analyze 

the impact of these coops on the production quantities, as well as their impact on 

the market. 

■ SA56 

56- Curtis A- Hyatt 

Strategy and Strategic Planning 


Chair: Jay Lee, California State University Sacramento, 6000 J St., 

Tahoe Hall 2079, Sacramento, CA, 95819, United States of America, 

jlee@csus.edu 

1 - Applying Synthetic Control Methodology in Management 

Research: Analyzing the Impact of Government Assistance on 

Chrysler’s Performance 

Brian Richter, Assistant Professor of Business, Economics, and 

Public Policy, Richard Ivey School of Business, University of 

Western Ontario, 1151 Richmond Street North, London, ON, 

N6A3T9, Canada, brian.richter.@mccombs.utexas.edu, Adam 

Fremeth, Guy Holburn 

We introduce synthetic control to management research by demonstrating how to 

apply the technique to quantify the effects of managerial interventions when a 

single observation unit receives a treatment on a single date. The problem with 

such scenarios is that counterfactual data is typically lacking. The method, 

however, allows us to generate counterfactual data systematically from among 

untreated firms. In our application, we assess the impact of government assistance 

on Chrysler’s performance. 

2 - New Methodology for the Mandatory Minimal Outflow 

Representation on Hydrothermal Ssystems Planning 

Wellington Conceicão, Instituto Federal de Educacão e Tecnologia, 

Campus IFET, Juiz de Fora, Brazil, wellingtcc@yahoo.com.br, 

André Marcato, Joao Passos Filho, Rafael Brandi, Tales Ramos, 

Ivo Silva Junior 

This work presents an alternative methodology to supply the mandatory 

minimum outflow in the problem of long-term operation planning of 

hydrothermal interconnected systems. It was used the strategy of trying to 

maintain the storage reservoirs within safe levels through a minimum curve of 

stored energy level. The simulation results shows that when using this 

formulation a decrease in the minimum outflow deficit is obtained. 

3 - Game Theoretic Analysis of Inter-firm R&D Expenditure 

Distribution in Semiconductor Manufacturing 

Tom Heaps-Nelson, PhD Candidate, Massachusetts Institute of 

Technology, 77 Massachusetts Ave., Cambridge, MA, 02139-4301, 

United States of America, heaps@mit.edu 

Soaring R&D and capital investment costs has left the semiconductor industry 

with a handful of large chipmakers and semiconductor equipment suppliers. This 

paper employs non-cooperative game theory along with detailed empirical 

estimates of firm payoffs as a framework for understanding how the largest 

industry players are likely to share the burden of R&D funding for two ongoing 

industry-wide process technology transitions: EUV lithography and 450mm 

wafers.

4 - Market Partitioning and Competitive Dynamics in 

Network-based Industries 

Jay Lee, California State University Sacramento, 6000 J St., Tahoe 

Hall 2079, Sacramento, CA, 95819, United States of America, 

jlee@csus.edu 

This paper looks at how market partitioning framework is evolved at networkbased 

industry level where mature and emerging players compete with each 

other in a dynamic environment, and finally several propositions for further 

studies will be briefly discussed at the end. 

5 - A Persuasive Power to Accept Change 

Mohammad Z. Bsat, Associate Professor, National University, 

11736 Vail Court, San Diego, CA, 92131, United States of America, 

mbsat@nu.edu 

TQM principles made organizations more competitive. Higher education 

institutions should embrace TQM practices to undergo transformation & reverse 

the lagging effectiveness they suffer from. The efforts to improve education 

quality are hindered by the competition that controls higher education 

institutions. A collaborative approach to applying TQM principles can lead to a 

true transformation of higher education This paper addresses the issue of 

overhauling the overall quality of higher education. 

■ SA57 

57- Curtis B- Hyatt 

Bayesian Approach 


Chair: Douglas Jones, Rutgers Business School, 2000 Margerum 

Avenue, Lake Como, NJ, 07719, United States of America, 

dhjones@rci.rutgers.edu 

1 - The Economically Designed Multivariate Bayesian Control Chart 

for Short-run Production 

Pengwei Zhang, Xiían Jiaotong University, Xi’an, Shaanxi, 

P.R.China, Xiían, China, kimi.2004@stu.xjtu.edu.cn, Qin Su 

It has been known that Bayesian control charts are optimal tools for statistical 

process control but few results have been regarding the influence of the 

disturbance occurrence time on process cost. Based on new partition of process 

state, process transition matrices under different scenarios were constructed in 

this paper employing Markov chain method. Cost function was constructed 

iteratively, and finally optimization model for short-run production was 

formulated. 

2 - A Bayesian Approach for Model Selection in Fractionated Split 

Plot Robust Design Experiments 

Matthias Tan, Student, Georgia Institute of Technology, 251, 10th 

St. NW, Tenth and Home E TAE 208A, Atlanta, GA, 30318, 

United States of America, mtan6@gatech.edu, C.F. Jeff Wu 

This paper proposes a Bayesian method for model selection in fractionated robust 

design experiments that are run as split plot experiments. We employ a Bayesian 

hierarchical model that takes into account the split plot error structure. Forward 

selection and combined global and local search algorithms are proposed to find 

models with high posterior probabilities. The proposed method is used to analyze 

three real experiments. 

3 - An Optimization Approach to Approximate Bayesian Inference 

of Complex Systems 

Jose Lainez, Purdue University, 480 Stadium Mall Drive, West 

Lafayette, IN, United States of America, jlainez@purdue.edu, 

Linas Mockus, Seza Orcun, Gary Blau, Gintaras Reklaitis 

Monte Carlo approaches have been widely used for inference. The drawback of 

these methods is that they may be computationally expensive. Variational 

methods for approximate inference can provide efficient alternatives. However, 

the ones developed so far have been targeted for relative simple models. Here, we 

propose a variational approach that can handle models expressed as a set of 

differential-algebraic equations and whose posterior approximation is represented 

by a multivariate distribution. 

4 - Predicting Material Accounting Misstatements: 

A Bayesian Approach 

Feng Xu, Georgia Southwestern State University, 800 GSW State 

University Drive, Americus, GA, 31709, United States of America, 

fenghsu@hotmail.com, Zinan Zhu 

In this paper, we develop Bayesian models for predicting accounting 

misstatements. The outputs of the models are probabilistic descriptions for the risk 

of conducting misstatement and posterior comparison of the riskiness between 

different firms. Empirical analysis results suggest that, compared to classical 

models, the Bayesian prediction models improve in terms of reliability and 

sensitivity, which provides a useful tool to assess materials accounting 

misstatement risks. 


75 

5 - Bayesian Updating of Norming Tables 

Douglas Jones, Rutgers Business School, 2000 Margerum Avenue, 

Lake Como, NJ, 07719, United States of America, 

dhjones@rci.rutgers.edu, Marco Cors-Lozada 

The distribution of the observed test score is estimated by nonparametric Bayesian 

methods based on mixture distributions where the mixing weights are obtained 

from a Dirichlet Process. We introduce the Dirichlet Processes and related Markov 

Change Monte Carlo simulation. The predictive posterior distribution is used to 

estimate the norming table. Prior information is used in the base distribution of 

the Dirichlet Process. 

■ SA58 

SA58 

58- Phoenix East- Hyatt 

Dynamic Control Applications 

Sponsor: Applied Probability 


Chair: Mark Lewis, Professor, Cornell University, School of Ops 

Research and Inform. Engin, 226 Rhodes Hall, Ithaca, NY, 14853, 

United States of America, mark.lewis@cornell.edu 

1 - Dynamic Exploration with Alternative Technologies: When to 

Stop Surveying and Invest 

Nur Sunar, PhD Candidate, Stanford University, Graduate School 

of Business, Stanford, CA, United States of America, 

nsunar@stanford.edu, J. Michael Harrison 

We consider an oil exploration and production company that can use various 

costly technologies to survey a given site, thus reducing uncertainty about the 

value of the site. Using a continuous-time Bayesian framework, we characterize 

the optimal exploration policy, and the conditions under which it is optimal to 

stop surveying and invest in (or abandon) the site. 

2 - Optimal Hiring and Job Assignment 

Martin Reiman, DMTS, Alcatel-Lucent Bell Labs, 600 Mountain 

Ave., Murray Hill, NJ, 07974, United States of America, 

marty@research.bell-labs.com, Yue Jin 

We consider a model with two types of workers, novices and experts, who process 

a Poisson stream of jobs. The available controls involve hiring of novice workers 

and assignment of jobs to workers. Novices are promoted after sufficient work 

experience, and all workers eventually retire. This problem decomposes naturally 

into two time scales, motivating the use of results from nearly completely 

decomposable Markov decision processes in our analysis. 

3 - Ambulance Allocation using Multi-Class Queueing 

Mark Lewis, Professor, Cornell University, School of Ops Research 

and Inform. Engin, 226 Rhodes Hall, Ithaca, NY, 14853, United 

States of America, mark.lewis@cornell.edu, Kenneth Chong, 

Shane Henderson 

We consider the problem of allocating various types of emergency calls to 

emergency response vehicles. One set of vehicles is typically used for more serious 

calls while the other for less serious. However, when all of one kind are busy, the 

other may be allocated to an incoming call. The difficulty is in the fact that 

mismatching ambulances to calls leaves the system with the vulnerability of 

having to allocate an ambulance to a call that it is ill-equipped to handle. 

4 - Non-collaborative Systems with Setup Costs 

Tugce Isik, Georgia Institute of Technology, 765 Ferst Drive NW, 

Atlanta, GA, 30332-0205, United States of America, 

tugceisik@gatech.edu, Sigrùn Andradòttir, Hayriye Ayhan 

We study a system of two tandem queues with a finite buffer and two flexible 

servers. The servers are unable to collaborate and constant setup costs are 

incurred when the servers move between stations. We investigate how the server 

assignment policy that maximizes the long-run average profit depends on the 

magnitude of the setup cost. The optimal policy is characterized for different 

buffer sizes.

SA59 

■ SA59 

59- Phoenix West- Hyatt 

Many-server Queues: Approximations and Control 



Chair: Itai Gurvich, Northwestern University - Kellogg School of 

Management, Evanston, IL, United States of America, 

i-gurvich@kellogg.northwestern.edu 

1 - Impatience Differentiation in Large-scale Service Systems with 

Customer Reentrance 

Guodong Pang, Assistant Professor, Pennsylvania State University, 

University Park, PA, United States of America, gup3@engr.psu.edu, 

Weining Kang 

We consider large-scale service systems where customers need to reenter the 

system to fulfill their further service requests and the reentry may happen after 

some random delay time. We assume that the patience times of new and 

reentrant customers may have different distributions. Under mild assumptions on 

the general interarrival, service, delay and patience delay times, we investigate the 

impact of the impatience differentiation upon the service performance in such 

systems. 

2 - Staffing and Control for Call Center Outsourcing with Rework 

Jiheng Zhang, Hong Kong University of Science and Technology, 

Clear Water Bay, Hong Kong, Hong Kong-PRC, j.zhang@ust.hk, Jin 

Fang, Eser Kirkizlar 

We study a system with both in-house and outsourcer call centers. With certain 

probability, customers need to be rerouted to the in-house call center for further 

help after receiving service at the outsourcer. The pooling of the two call centers is 

allowed. Using a fluid approximation in a heavy traffic regime, we jointly 

determine the optimal control rule and the staffing policy that asymptotically 

minimize the staffing cost and long run holding cost subject to a service level 

constraint. 

3 - Non-FCFS Scheduling in Fluid Many-server Queues 

Petar Momcilovic, Associate Professor, University of Florida, 

303 Weil Hall, P.O. Box 116595, Gainesville, FL, 32611, 

United States of America, petar@ise.ufl.edu, Avishai Mandelbaum 

We consider non-FCFS scheduling policies in many-server fluid queues. In 

particular, we examine time-varying queues under the assumption that the 

scheduler knows realizations of customer patience/service times. Implementing a 

non-FCFS policy improves the overall system performance in terms of the number 

of customers that abandon during overloaded periods. 

4 - Scheduling Parallel Servers in the Non-degenerate Slowdown 

Diffusion Regime: Asymptotic Optimality 

Itai Gurvich, Northwestern University-Kellogg School of 


i-gurvich@kellogg.northwestern.edu, Rami Atar 

We consider the problem of minimizing holding costs in a parallel-server system 

operating in a many-server heavy-traffic regime with non-degenerate slowdown 

(NDS). The NDS many-server regime is distinct from existing heavy-traffic 

regimes. As in the conventional heavy-traffic regime, the asymptotics are captured 

by a Brownian control problem, the solution of which exhibits state-space 

collapse. We prove an asymptotic lower bound and construct a sequence of 

asymptotically optimal policies. 

■ SA60 

60- Remington- Hyatt 

Joint Session RM/Aviation: Robust Airline 

Optimization, Alliances and Revenue Sharing 

Sponsor: Revenue Management & Pricing & Aviation Applications 


Chair: Milind Sohoni, Indian School of Business, Gachibowli, 

Hyderabad, 500032, India, milind_sohoni@isb.edu 

1 - Reducing Flight Delays by Optimizing Block-time Allocation 

Mazhar Arikan, University of Kansas, School of Business, 

Lawrence, KS, United States of America, mazhar@ku.edu, 

Vinayak Deshpande, Milind Sohoni 

In this paper, we develop an approach to reallocate block-times across the 

network of an airline and analyze the trade-off between total network arrival 

delay minutes and network-wide block-time budget based on a stochastic traveltime 

model. Using data acquired from a major U.S. airline, we construct an 

efficient frontier by optimally allocating block-times while considering target ontime 

arrival probability for each flight in the network. 


76 

2 - Robust Gate Assignment to Reduce Tarmac Waiting Time 

Amy Cohn, University of Michigan, Ann Arbor, MI, 

United States of America, amycohn@umich.edu, Ryan Chen 

We study the problem of assigning flights to gates at an airport, with the goal of 

reducing delay associated with an inbound flight waiting on the tarmac because 

its gate is still occupied by the preceding outbound flight. This can happen 

because the outbound flight is late leaving, the inbound flight is early arriving, or 

both. We develop MIP models to assign flights to gates so as to recognize the 

inherent stochasticity of the system, and provide empirical analysis based on these 

models. 

3 - Revenue Sharing in Airline Alliances 

Xing Hu, Assistant Professor, University of Oregon, 484 Lillis Hall, 

1208 University of Oregon, Eugene, OR, 97401, United States of 

America, xingh@uoregon.edu, Rene Caldentey, Gustavo Vulcano 

We study codeshare, the practice of independent airlines collaboratively 

marketing and operating fights. We propose a two-stage hierarchical approach to 

study the revenue sharing rules and decentralized inventory controls from both 

cooperative and non-cooperative game approach. Through both analytical and 

numerical studies, we find that revenue sharing rule affects the system 

significantly. We identify the existence and uniqueness of an admissible and 

efficient revenue sharing. 

■ SA61 

61- Russell- Hyatt 

Aviation Support Operations 

Sponsor: Aviation Applications 


Chair: Wenwei Cao, Georgia Institute of Technology, 765 Ferst Dr NW, 

Atlanta, GA, United States of America, cww@gatech.edu 

1 - Cost Allocation in Fractional Jet Ownership Programs 

Wenwei Cao, Georgia Institute of Technology, 765 Ferst Dr NW, 

Atlanta, GA, United States of America, cww@gatech.edu, Ozlem 

Ergun, Lezhou Zhan, Ellis Johnson 

Higher charter and reposition costs have plagued the profitability of the fractional 

jet industry since its inception. Incentive programs and pricing adjustments are 

introduced to offset these common costs incurred to service the customers. A 

central problem in designing these programs is establishing a justifiable cost 

allocation scheme. Furthermore, accurate estimates of cost-to-serve are valuable 

information to other marketing and operational decisions. In this paper we 

develop different mechanisms for the cost allocation problem. The 

appropriateness and key trade-offs of the mechanisms are also discussed on the 

basis of certain principles of equity and computational performance. 

2 - An Integrated Optimal Decision Model for Ground Logistics and 

Air Cargo Scheduling 

Kwon Gi Mun, PhD Student, Rutgers Business School, #1009B, 1 

Washington Park, Newark, NJ, 07102, United States of America, 

kwongmun@eden.rutgers.edu 

The mathematical model of two-tiered generalized scheduling problem has been 

formulated. Thus the optimal assignment solution between airports and facilities 

will be computed with the minimum transportation cost at the first stage. And 

then the minimum total cost will be calculated with the optimal routing solution 

of aircrafts. This research will use a mathematical model with practical constraints 

and real data. 

■ SA62 

62- Borein A - Hyatt 

Joint Session Auctions/MSOM: Competitive Sourcing 

and Procurement 

Cluster: Auctions & Manufacturing & Service Oper Mgmt 


Chair: Cuihong Li, University of Connecticut, Storrs, CT, 06269, 

United States of America, cli@business.uconn.edu 

1 - Incentive Functions for Transportation Procurement Auctions 

Diwakar Gupta, University of Minnesota, 111 Church Street S. E., 

Minneapolis, MN, 55455, United States of America, 

guptad@me.umn.edu, Justin Azadivar, Yibin Chen 

State Transportation Agencies (STAs) use variants of the first-price sealed-bid 

auction mechanism to realize low cost, high quality, and fast completion of 

highway and bridge construction projects. STAs pay incentives based on the 

realized value of an attribute of the work, which may or may not be of value, but 

bids do not explicitly include this attribute. We present models to identify costminimizing 

incentive functions for agencies, net of any benefits that may accrue 

from the attribute.

2 - An Analysis of Scoring and Buyer-determined 

Procurement Auctions 

Natalia Santamaría Tobar, Universidad de Los Andes, 

Departamento de Ingenierìa Industrial, Bogotá, Colombia, 

n-santam@uniandes.edu.co 

I compare two procurement auction formats (in terms of the expected cost for the 

buyer) in which an incumbent competes with a group of entrants. A scoring 

auction in which suppliers compete on the adjusted bids or scores, and, a buyerdetermined 

auction in which suppliers compete on the price, and the buyer 

adjusts the bids with the non-price attributes after the auction to determine the 

winner. 

3 - Bidding Decisions under Costly Information Acquisition 

Brendan See, PhD Candidate, University of Michigan, 1205 Beal 

Ave., Ann Arbor, MI, 48104, United States of America, 

bdsee@umich.edu, Izak Duenyas, Damian Beil 

Procurement research typically assumes that suppliers know their cost to produce 

an item. While this may be true for an incumbent supplier, an entrant supplier is 

usually less informed. We model this scenario and allow the entrant to learn 

additional information at a cost. The entrant can learn prior to the auction or 

delay learning and make a less-informed bid. We study the entrant’s strategy and 

evaluate how the buyer can influence this strategy through subsidies and other 

methods. 

4 - Mechanism for Robust Procurements 

Yingqian Zhang, Assistant Professor, Erasmus University 

Rotterdam, Burg. Oudlaan 50, Rotterdam, 3062 PA, Netherlands, 

yqzhang@ese.eur.nl, Sicco Verwer 

We model robust procurement as an optimization problem. We show this 

problem is NP-complete, and propose a backtracking algorithm to find the optimal 

solution. We propose a multi-stage mechanism that is truthful, efficient, and postexecution 

individually rational. In the experiments, we compare our mechanism 

with an iterated greedy mechanism that represents the current practice in public 

procurements, in terms of the expected social welfare and the expected payments 

of the auctioneer. 

■ SA63 

63- Borein B- Hyatt 

Behavioral Perspective on Sales and Operations 


Chair: Fei Qin, PhD Candidate, University of Cincinnati, College of 

Business, 312 Carl H. Lindner Hall, Cincinnati, OH, 45221, 

United States of America, qinfi@mail.uc.edu 

1 - An Experimental Investigation of Information Utilization in 

Newsvendor Decisions 

M.Sinan Gonul, Assistant Professor, METU, Department of 

Business Administration, Middle East Technical University, 

Ankara, 06800, Turkey, msgonul@metu.edu.tr, 

Ayse Kocabiyikoglu, Itir Gogus 

In the current study, we aimed to investigate the utilization of information in the 

decision making process within classical newsvendor setting. Through an 

experimental design, three different types of information (i.e. demand, price and 

cost) were systematically withheld/made available and the resultant purchase 

decisions were recorded via software specifically tailored for the purpose. 

Behavioral patterns are examined, findings are discussed and future research 

directions are given. 

2 - On the Risk-attitude of Informed Newsvendors 

Torsten Gully, University of Cologne, Albertus-Magnus-Platz, 

Cologne, 50923, Germany, torsten.gully@uni-koeln.de, 

Ulrich Thonemann 

The behavior of Newsvendors has been extensively analyzed in the behavioral 

operations literature. In essentially all experiments actual order quantites deviate 

from the order quantity predicted by the classical Newsvendor model. We show 

that risk-attitudes influence the deviation if people can assess the involved risk. 

We formulate a model to explain the effect of risk-attitudes on Newsvendor 

behavior and report on the validation of our model through a controlled 

laboratory environment. 

3 - The Strategic Advantage of Behavioral Targeting: How Firm Can 

Benefit from Personal Data? 

Vincenzo Palermo, Georgia Institute of Technology, 

800 West Peachtree St., College of Management, Atlanta, 30308, 

United States of America, vincenzo.palermo@mgt.gatech.edu, 

Dan Breznitz 

We examine whether the access of personal data can create a competitive 

advantage for firms investing in behavioral targeting advertising. Specifically, we 

analyze a novel proprietary dataset on online investment strategies of 3,889 

companies from 2006 to 2011. We identify three strategies: paid search, 

behavioral targeting and natural search. We show that an investment strategy 


77 

based on personal data (behavioral targeting) is more effective than traditional 

advertising investment. Moreover, firms can experience positive synergies 

between behavioral and traditional investment strategies. Our results suggest that 

behavioral targeting lowers information asymmetries about consumers and 

therefore it increases advertising effectiveness. Despite its positive impact on firm 

performance, the use of personal data raises privacy concerns in that it is 

important to protect personal information without undermining the incentives to 

invest in online advertising. 

4 - Vertically Restrictive Pricing Contracts for Bounded Rational 

Customers: Does Equity Matter? 

Fei Qin, PhD Candidate, University of Cincinnati, College of 

Business, 312 Carl H. Lindner Hall, Cincinnati, OH, 45221, United 

States of America, qinfi@mail.uc.edu, Feng Mai, Amit Raturi, 

Michael Fry 

In a single-period experiment, we test and compare Retailer Fixed Markup (RFM) 

policy with Price-only. Through the lenses of RFM contract, we investigate SC 

Power (high vs. low markup) and Pricing power (who is the pricing leader, the 

manufacturer or the retailer) to see how they would affect subjects’ decision 

making process. We then attempt to update our analytical model to incorporate 

the findings from the experiment. 

■ SA67 

SA67 

67- Ellis East- Hyatt 

Data Fusion for Manufacturing System 

Performance Improvement I 



Chair: Jian Liu, Assistant Professor, The University of Arizona, Rm 221 

ENGR Building, 1127 E. James E. Rogers Way, Tucson, AZ, 85721, 

United States of America, jianliu@sie.arizona.edu 

Co-Chair: Ran Jin, Assistant Professor, Virginia Tech., 1145 Perry St., 

Blacksburg, VA, 24060, United States of America, jran5@vt.edu 

1 - Calibration and Adjustment of Nanomaterials Property 

Prediction Model using Experiment Data 

Chun Zhang, Georgia Institute of Technology, 755 Ferst Drive, NW, 

Atlanta, United States of America, czhang@fsu.edu, Ben Wang, 

Arda Vanli, Richard Liang, Kan Wang 

Carbon nanotube sheet, i.e. buckypaper (BP), is a multifunctional nanomaterial 

with extraordinary properties. Post treatment using poly (vinyl alcohol) (PVA) 

can further enhance the mechanical properties of BP. A finite element based 

mechanical model was developed to link PVA-treated BP’s modulus to its 

nanostructure characteristics. This model was then calibrated and adjusted using 

experimental data. The performance of the calibrated model was tested using the 

cross validation process. 

2 - Computational Geometry and Data Fusion in CAD/CAM for 

Manufacturing System Performance Improvement 

Yuan-Shin Lee, Professor, North Carolina State University, 400 

Daniels Hall, Ind. Eng., CBox 7906, North Carolina State 

University, Raleigh, NC, 27695, United States of America, 

yslee@ncsu.edu 

This paper presents methods of computational geometry and data fusion in 

CAD/CAM for manufacturing system performance improvement. Techniques of 

computational geometry processing and data fusion are presented for CAD/CAM 

decision optimization. Examples are provided on dealing with optimizing in 

CAD/CAM and CNC machining planes determination problems. The presented 

methods can be used to automate the traditional, experience-based tool selection 

tasks and to reduce the total machining time. 

3 - Ensemble Modeling for Manufacturing Scale-up through 

Experimental and Observational Data Fusion 

Ran Jin, Assistant Professor, Virginia Tech., 1145 Perry St., 

Blacksburg, VA, 24060, United States of America, jran5@vt.edu, 

Xinwei Deng 

In modern manufacturing scale-up efforts, both experimental and observational 

data are available for material, process and quality. Current research efforts 

limited their scope on modeling quality-process relationship based on single type 

of data. This paper presents an innovative method to model the quality-process 

relationship by the fusion of the two types of data. An ensemble modeling 

strategy is developed, and case studies in both simulation and wafer 

manufacturing are provided.

SA68 

4 - Detection of Approximate Periodicity in Categorical Time Series 

with Time Intervals 

Jian Liu, Assistant Professor, The University of Arizona, Rm 221 

ENGR Building, 1127 E. James E. Rogers Way, Tucson, AZ, 85721, 

United States of America, jianliu@sie.arizona.edu, Zhenrui Wang, 

Mingyang Li, Sean Dessureault 

Categorical time series are symbol sequences arranged in the order of occurrence. 

When time of occurrences of symbols is also available, order and time 

information should be fused to detect periodic patterns. In practice, symbols often 

repeat with only approximately equal time intervals. This paper employs 

statistical method for the detection of such approximate periodicity. A case study 

from mining industry demonstrates the effectiveness of the proposed approach. 

5 - Modeling, Experimental Design, and Analysis of 

Stereolithography Process for Direct 3-D Printing 

Jizhe Zhang, University of Southern California, 3715 McClintock 

Avenue, GER 240, Los Angeles, CA, 90089, United States of 

America, jizhezha@usc.edu, Qiang Huang, Arman Sabbaghi, 

Tirthankar Dasgupta 

We propose a systematic model to model and predict the shrinkage process to 

address the accuracy problem for Stereolithography (SLA, one technique for 3-D 

printing), instead of the traditional finite element simulation or experimental 

trial-and-error methods. A variant of Latin Square experimental design is 

employed to identify the process model and the experimental outcomes are 

analyzed for optimized compensation. One case study is presented to verify the 

shrinkage model and compensation. 

■ SA68 

68- Suite 312- Hyatt 

Finance: Portfolio Analysis 


Chair: Ludovic Cales, HEC - University of Lausanne, Bat. Extranef, 

Room 253, Lausanne, 1015, Switzerland, ludovic.cales@unil.ch 

1 - Absolute Return Portfolios 

John Beasley, Brunel University & JB Consultants, 

United Kingdom, john.beasley@brunel.ac.uk 

We consider the problem of deciding an absolute return portfolio, a portfolio of 

assets that is designed to deliver a constant return over time. We present a threestage 

mixed-integer zero-one program for the problem that explicitly considers 

transaction costs associated with trading. We extend our approach to the problem 

of designing portfolios with differing characteristics. Computational results are 

given for portfolios derived from universes defined by S&P international equity 

indices. 

2 - A New Approach to Asset Allocation: Constructing Portfolio 

Holdings Directly 

Nan Xiong, Carnegie Mellon University, Tepper School of Business, 

5000 Forbes Avenue, Pittsburgh, PA, United States of America, 

nxiong@andrew.cmu.edu, Javier Peña, Burton Hollifield 

In this paper, we propose a new approach to portfolio construction, which 

incorporates the estimation and optimization into a one-stage problem. We 

propose a new portfolio allocation strategy based on our one-stage formulation. 

We show that the resulting portfolio weights are less sensitive to the noisy data. 

Moreover, the proposed portfolio tends to perform better out-of-sample. The 

numerical results based on empirical and simulated data confirm our findings. 

3 - Interactive Portfolio Optimization using Mean-Gini Criteria 

Ran Ji, PhD Student, George Washington University, 2201 G St., 

Funger Hall 415H, Washington DC, DC, 20052, United States of 

America, jiran@gwu.edu, Miguel Lejeune, Srinivas Prasad 

We propose a bi-objective Mean-Gini portfolio optimization model. An interactive 

solution procedure is designed to elicit preferences among the efficient portfolios. 

Computational results describing the convergence of the interactive procedure 

and the out-of-sample performance of the portfolios will be presented. 

4 - A Bi-objective Portfolio Model for Supporting Services 

in a Hospital 

Bartosz Sawik, PhD, Department of Applied Computer Science, 

Faculty of Management, AGH University of Science and 

Technology, Al. Mickiewicza 30, Krakow, 30059, Poland, 

B_Sawik@yahoo.com 

This paper presents a bi-objective portfolio model for optimal allocation of 

supporting services in a hospital. The services include inventory, financial, 

operations management, etc. The optimality criteria of the problem are 

minimization of operational costs of supporting services subject to some specific 

constraints. The constraints represent specific conditions for resource allocation in 

a hospital. The computational experiments based on a real data from a hospital in 

Poland. 


78 

5 - A Rank-based Approach to Cross-sectional Analysis 

Ludovic Cales, HEC-University of Lausanne, Bat. Extranef, 

Room 253, Lausanne, 1015, Switzerland, ludovic.cales@unil.ch, 

Monica Billio, Dominique Guègan 

This paper studies the cross-sectional effects present in the market using a new 

graph theoretic framework. We model the evolution of a dynamic portfolio, i.e. a 

portfolio whose weights change over time, as a function of cross-sectional factors 

where the predictive ability of each factor is described by a variable. This 

modeling permits us to measure the marginal and joint effects of different crosssection 

factors on a given dynamic portfolio. Applications are presented. 

■ SA69 


Finance: Banking and Insurance 


Chair: Komlan Sedzro, Professor, School of Management, University of 

Quebec, 315 Ste Catherine Est, succ. Centre-Ville, Montreal, QC, 

H3C3P8, Canada, sedzro.k@uqam.ca 

1 - Loan Rate Optimization for Financing Platforms 

Wei Wang, IBM Research-China, Building 19 Zhongguancun 

Software Park, Haidian District, Beijing, 100193, China, 

Wangwcrl@cn.ibm.com, Yueting Chai, Hongwei Ding, Jin Dong 

Financing platform is an important channel for small and medium enterprises 

(SME) to get loan supports. When receiving a loan request from a client, the 

financing platform needs to made decision on the loan rate. We build an 

optimization model for decision support. The objective is to maximize the profit 

and minimize the financing risks. A local-search based heuristics algorithm is 

designed to find the optimal solution; the simulation model is built to evaluate 

each option. 

2 - B-spline Method of Estimating Term Structure of Interest Rate 

Yifeng Yan, Xi’an Jiaotong University, No.2269 Mailbox, Xi’an, 

710049, China, yanyifeng0123@163.com, Ju’e Guo 

This study implements the B-spline method to estimate the term structure of 

interest rate using data in Shanghai stock exchange. It shows that the B-spline 

method has fine goodness of fitting.The B-spline method can be used as an 

effective tool of monitoring interest rate level and formulating monetary policies. 

3 - The Informational Advantage of Local Investors: Evidence from 

Fund Managers Trades 

Kian Ming Tan, Associate Lecturer, University of New South 

Wales, Kensington, Sydney, Australia, eric.tan@unsw.edu.au 

The economic value of geography for money managers is now well supported in 

the literature by evidence of the profitability of local investments. Using data on 

both mutual fund equity holdings and fund returns, we find stocks with higher 

local investor participation to be associated with stocks abnormal returns during 

post event period. In addition, local ownership seems to predict funds increasing 

weighting on affected firms which experience lower credit risk. 

4 - Finite Precision Errors in Stochastic Financial Networks 

Ganesh Perumal, International Institute of Information 

Technology, 26/C Electronics City, Bangalore, India, 

ganesh_perumal@iiitb.ac.in, Abhilasha Aswal, 

G. N. Srinivasa Prasanna 

A financial network of money transfers involving multiple currencies may be 

erroneous due to the usage of IEEE 754 binary standards. The fluctuations in 

currency exchange rates make this a random graph. In this presentation we show 

how to exploit such graphs to maximize cumulative errors in favour of some 

entity. 

5 - Comparing the Performance of Microfinance Institutions in 

South America and Asia 

Komlan Sedzro, Professor, School of Management, University of 

Quebec, 315 Ste Catherine Est, succ. Centre-Ville, Montréal, QC, 

H3C3P8, Canada, sedzro.k@uqam.ca, Marie-Héléne Noiseux 

We apply multi-directional efficiency analysis to compare the performance of 

microfinance institutions (MFIs) in South America and Asia. Indeed, despite their 

supposedly importance in combating financial exclusion, many MFIs find it hard 

to attain financial independence. This fact in itself justifies the importance of 

studies on the performance of MFIs, especially interregional comparisons. The 

objective is to see if the MFIs in one region can copy the good practices of another 

geographic area.

■ SA70 


Decision Making over Networks 

Cluster: Computational Social Science 


Chair: Azarakhsh Malekian, Massachusetts Institute of Technology, 

Cambridge, MA, United States of America, malekian@mit.edu 

1 - Peers and Pundits: Correlated Opinion Dynamics 

Mohamed Mostagir, Massachusetts Institute of Technology, 

77 Massachusetts Ave., D32-632, Cambrdige, MA, 02139, 

United States of America, mosta@mit.edu, Daron Acemoglu, 

Asuman Ozdaglar 

We study the simultaneous evolution of the opinion profile and network topology 

of a system of N agents. Agents are more inclined to communicate with those 

who hold similar opinions, but are also attracted to popular individuals. This 

system exhibits strong correlation over time and is difficult to analyze. Regardless, 

we show that it converges to a consensus, describe the role of popular agents, give 

bounds on the speed of convergence, and characterize that opinion for large 

societies. 

2 - Striving for Social Status 

Greg Stoddard, Northwestern University, 2133 Sheridan Road, 


gregS@u.northwestern.edu, Rachel Kranton, Nicole Immorlica 

We consider a network of agents producing a positional good; an agent’s utility 

for the good is a function of both his intrinsic value for the good and his status, or 

production level relative to his neighbors. We show that the set of Nash equilibria 

of our game form a complete lattice. We then study the effects of status and 

network structure on society. We find that production increases and welfare 

decreases with status effects. Network structure can have arbitrary impacts on 

both quantities. 

3 - Network Security and Contagion 

Azarakhsh Malekian, Massachusetts Institute of Technology, 

Cambridge, MA, United States of America, malekian@mit.edu, 

Daron Acemoglu, Asuman Ozdaglar 

We consider the network security problem and compare the properties of the 

equilibrium and the socially optimal solutions. An infection percolates over the 

network. Agents protect themselves by investing in the security. We delineate 

conditions under which there is underinvestment in the equilibrium and also 

show which sorts of networks lead to more severe market failures. When attacks 

are strategically targeted,we show that there is a new mechanism for 

overinvestment. 

4 - Competitive Equilibrium in Two Sided Matching Markets with 

General Utility Functions 

Saeed Alaei, University of Maryland, 4315 Rowalt Dr, APT 201, 


saeed.a@gmail.com, Azarakhsh Malekian, Kamal Jain 

We present the first exact characterization of prices/payoffs at competitive 

equilibria in two sided matching markets with non-quasilinear utilities. The 

problem has been studied extensively (e.g. Demange, Gale, Quinzii, etc), however 

no constructive proof of existence and/or characterization was known for 

equilibria. We also derive a mechanism for Ad-Auctions which does not rely on 

the assumption that advertisers and search engine have the same belief about 

clickthrough rates. 

■ SA71 


Frontiers in Digital Business 

Sponsor: eBusiness 


Chair: Ali Tafti, University of Illinois at Urbana-Champaign, 

1206 S. Sixth Street, 350 Wohlers Hall, Champaign, IL, 61820, 

United States of America, atafti@illinois.edu 

1 - Effort and Quality in Online Crowdsourced Markets: Influence of 

Global Economic and Cultural Factors 

Jesse Bockstedt, University of Arizona, 1130 E. Helen St., Tucson, 

AZ, 85721, United States of America, 

bockstedt@email.arizona.edu, Cheryl Druehl, Anant Mishra 

One emerging consequence of crowdsourcing is increased competition in global 

labor markets for digital work. We analyze how economic and cultural factors 

influence the effort and quality of work contributed to these new virtual labor 

markets. Using data from a popular online logo design contest site we 

demonstrate that GDP per capita, broadband penetration, and Hofstede’s cultural 

dimensions are strong predictors of a global digital worker’s effort and work 

quality. 


79 

2 - Firm Capabilities, Entrepreneurial Behavior, and IT Sourcing 

Ali Tafti, University of Illinois at Urbana-Champaign, 1206 S. Sixth 

Street, 350 Wohlers Hall, Champaign, IL, 61820, United States of 

America, atafti@illinois.edu, Ramanath Subramanyam, Rui Guo 

We examine the relationship between firm capabilities and IT sourcing strategies, 

and the extent that this relationship is framed by specific entrepreneurial 

strategies. 

3 - Can Social Come to the Rescue? Monetizing Music in the 

World of Free 

Jui Ramaprasad, Assistant Professor, McGill University, 

1001 Sherbrooke West, Montreal, QC, H3H2V1, Canada, 

jui.ramaprasad@mcgill.ca, Genevieve Bassellier, Remi Desmeules 

We examine the role of social media enabled features and shared social 

responsibility (SSR) on willingness to pay (WTP) for online music. Through a 

survey-based experiment, we find that a user’s WTP increases if an online music 

retailer incorporates the ability to build status and includes concepts of SSR in the 

distribution of revenues. This has implications for online music providers, artists, 

labels and the music industry as a whole, as well as other digital goods providers. 

4 - A Theoretical Assessment of the Development and Success of 

Global Entrepreneurial Teams 

Joy Oguntebi, Assistant Professor, Rochester Institute of 

Technology, 108 Lomb Memorial Dr., Rochester, NY, 14623, 

United States of America, joguntebi@saunders.rit.edu 

The “globality” phenomenon has led to reduced national dominance and 

increased worldwide competition. As such, the growing ambitions of innovationminded 

individuals and organizations in this new global market are capitalizing 

on the value of global entrepreneurial teams. Such teams are becoming more of a 

necessity as their diverse nature enables them to be more responsive to the 

external environment. This study investigates the formation and influence of 

global entrepreneurial teams. 

■ SA72 

SA72 


Cloud Computing Research Trends 

Cluster: Cloud Computing 


Chair: Ilyas Iyoob, Director, Advanced Analytics, Gravitant - ChainOpt, 

4507 A Dorsett Oaks Cir, Austin, TX, United States of America, 

ilyas.iyoob@chainopt.com 

1 - Cloud Computing Research Survey – 2012 Update 

Ilyas Iyoob, Director, Advanced Analytics, Gravitant-ChainOpt, 

4507 A Dorsett Oaks Cir, Austin, TX, United States of America, 

ilyas.iyoob@chainopt.com 

In this presentation, we provide an update of the current research in Cloud 

Computing, specifically the business of cloud. Previously we had compared cloud 

computing to a pull supply chain in an effort to organize the research domain. 

Continuing along the same lines, we show the progress of research over the past 

year. At the same time, we identify key shifts in the cloud computing industry 

and provide insights on how they may impact future research. 

2 - Provisioning for Large Scale Loss Network Systems with 

Application in Cloud Computing 

Yue Tan, The Ohio State University, 210 Baker Systems Bldg., 1971 

Neil Ave., Columbus, OH, 43210, United States of America, 

tan.268@osu.edu, Yingdong Lu, Cathy Xia 

We present a stochastic modeling approach to guide the resource provisioning 

task for future service clouds as the demand grows large. This problem can be 

mapped to a capacity planning problem in a general multi-class Erlang loss 

network model under quality of service constraint. We give an improved 

optimization formulation that is not only easier to solve, but also yields 

asymptotically exact provisioning solutions with improved QoS guarantees. 

3 - Xpress-Mosel: Modeling Support for Distributed Computing in 

Physical and Virtual Networks 

Oliver Bastert, FICO, Starly Way, Birmingham, United Kingdom, 

OliverBastert@fico.com, Susanne Heipcke 

This talk presents the Xpress-Mosel environment and its functionality for 

distributed and remote modeling and solving of optimization problems. In 

particular, we describe new functionality for remote computing without any need 

for a local installation of Xpress that has recently been released. We discuss 

several examples of distributed applications implemented by some of our 

customers.

SA73 

■ SA73 


Economics 


Chair: Lorena Berumen, Professor, Universidad Panamericana, Augusto 

Rodin 498, Col. Insurgentes Mixcoac, Mexico, DF, 03920, Mexico, 

laberumen@up.edu.mx 

1 - Estimating Real Money Balances for a Selected Asian Country: 

A Simultaneous Equation Model 

Abul Jamal, Professor, Southeastern Louisiana University, College 

of Business, Hammond, LA, 70402, United States of America, 

ajamal@selu.edu, Yu Hsing 

Equilibrium real balances and interest rates have been the subject of extensive 

research. Most studies presume the supply of money to be determined by the 

central bank and that the interest rate is only determined by the demand for 

money In this paper we use both the demand for and supply of money to 

estimate real money balances in a selected Asian country. 

2 - The Volatility Characteristics of China’s GEM Market and the 

Countermeasures 

Jie Xiong, Xi’an Jiaotong University, Postbox 2230, School of 

Management, No.28 Xianning West Road, Shaanxi, Xi’an, China, 

xiongjie@stu.xjtu.edu.cn, Ju’e Guo, Dong Qian 

Select the data of China’s GEM index from June 1, 2010 to October 28, 2011. 

EMD method is used to analyze factors that influence the fluctuations of index. 

Three factors are found: trend, low-frequency component and high-frequency 

component. Than we use the AR-EGARCH-M model to analyze the dynamic 

relationship between the returns and volatility of the GEM market and give the 

corresponding countermeasures. 

3 - The Role of Product Variety and Maturity in the Market 

Valuation of IT Intensive Firms 

Wael Jabr, Assistant Professor, University of Calgary, 2500 

University Dr NW, Calgary, AB, T3A6L5, Canada, 

wjabr@ucalgary.ca, Eric (Zhiqiang) Zheng 

The pervasiveness and intensive use of Information Technology in every aspect of 

firm processes have enabled firms to maintain a portfolio of diverse products and 

allowed improved performance. In this study, we address the question of how 

companies should best manage their portfolio of IT products under disruptive 

environments. We hypothesize that diversity and maturity of products play an 

integral part in the firm’s performance and its market value. 

4 - A Panel Data Model to Explain Technical Change through IT in 

Latin America 

Lorena Berumen, Professor, Universidad Panamericana, Augusto 

Rodin 498, Col. Insurgentes Mixcoac, Mexico, DF, 03920, Mexico, 

laberumen@up.edu.mx, Jen Ai de la Cruz, Gilberto González 

Pèrez, Margarita Hurtado, Irma Glinz 

In this paper we evaluate the IT’s impact over the conformation of the export 

profile of Argentina, Brazil and Mexico through a dynamic panel data model. We 

established that the adaptation and learning process is a function of expenditure 

on education and innovation, infrastructure and the dynamic of consumption. 

Sunday, 11:00am - 12:30pm 

■ SB01 

01- West 101- CC 

Genetic Algorithms and Ant Colonies 


Chair: Maria Woodside-Oriakhi, Assistant Professor, 

The College of the Bahamas, Oakes Field, Nassau, GT2602, Bahamas, 

moriakhi@cob.edu.bs 

1 - Genetic Algorithm and New Heuristic in Block Relocation 

Problem - Fuel Reduction 

Mazen Hussein, PhD Candidate, University of Wisconsin, 

Milwaukee, 1559 N. Prospect Ave., Apt # 305, Milwaukee, WI, 

53202, United States of America, mhussein@uwm.edu, Matthew 

Petering 

This research considers a variation of the block relocation problem (BRP). A set of 

identically-sized items is to be retrieved from stacks in a specific order using the 

minimum fuel consumption. In this work we consider container weight in fuel 

consumption calculations and explicitly tracking the containers moves. We 

developed a new heuristic that is embedded in a genetic algorithm. Results show 

that the methodology is effective in identifying near-optimal parameter settings. 


80 

2 - Evaluation of Multi-objective Genetic Algorithms on a 

Preventive Maintenance Scheduling Model 

Kamran Moghaddam, Assistant Professor of Systems Engineering, 

Southern Polytechnic State University, 1100 South Marietta 

Parkway, Marietta, GA, 30060, United States of America, 

kmoghadd@spsu.edu 

A multi-objective optimization model to find optimal preventive maintenance and 

replacement schedules of a multi-component system is used to evaluate 

performance of different multi-objective genetic algorithms. This multi-objective, 

nonlinear, mixed-integer model is an NP-hard optimization model that cannot be 

solved using exact algorithms. The different performance metrics are used to 

evaluating closeness and diversity of non-dominated solutions found by multiobjective 

genetic algorithms. 

3 - Ant Colony Optimization for the Cardinality Constrained 

Portfolio Problem 

Maria Woodside-Oriakhi, Assistant Professor, 

The College of the Bahamas, Oakes Field, Nassau, GT2602, 

Bahamas, moriakhi@cob.edu.bs 

This paper extends the mean-variance model of Markowitz to include the 

cardinality constraints This limits the number of assets held in the portfolio and 

sets bounds on the proportion of an asset held. An ant colony optimization 

heuristic algorithm is presented using data from seven real world market indices. 

■ SB02 

02- West 102 A- CC 

Joint Session: DAS/ENRE-Energy: Expert Elicitation 

of Energy Technologies: Theory and Practice 

Sponsor: Decision Analysis & Energy, Natural Res & the 

Environment/Energy 


Chair: Erin Baker, University of Massachusetts, Amherst, MA, 


1 - Elicitations of Energy Penalties for Carbon Capture 

Technologies 

Karen Jenni, Principal, Insight Decisions, LLC, 2200 Quitman St., 

Denver, CO, 80212, United States of America, 

kjenni@insightdecisions.com, Gregory Nemet, Erin Baker 

We conducted elicitation interviews with 15 experts in carbon capture 

technologies, exploring the effects of different energy policies on the viability of 

and the energy penalties (EP) for 6 technologies, by 2025. Expert opinions varied 

substantially both within a specific technology and between technologies. A 

worldwide carbon pricing scenario leads to a decrease in the mean EP of 1% to 

10% across the technologies; a scenario of increased US government R&D leads to 

a decrease of 6% to 14%. 

2 - Online Expert Elicitation Tools: Development and Case Studies 

in Energy Technology Assessment 

Steve Davis, Senior Research Associate, Carnegie Institution of 

Washington, 260 Panama Strett, Stanford, CA, 94305, 

United States of America, sjdavis@dge.stanford.edu, Karen Fries 

We are developing innovative tools to facilitate online expert elicitation in order 

to reduce the time and resources required to characterize uncertainty and the 

sources/scope of disagreement among experts. Such characterization is critical to 

directing R&D efforts on technical topics. We focus on experts’ subjective 

judgments on energy technologies. Here we demonstrate the tools under 

development and present results of testing with experts on solar PV and airborne 

wind energy systems. 

3 - Uncertainty and Surprise in Energy Forecasts: 

The Cost of Photovoltaics 

Max Henrion, CEO, Lumina Decision Systems, 26010 Highland 

Way, Los Gatos, CA, 95033-9758, United States of America, 

henrion@lumina.com, Evan Sherwin 

Power system planners and other energy analysts would like to know the future 

performance and costs of renewable and fossil energy technologies. Several 

research groups have used expert elicitation to assess uncertainty using 

probability distributions for these quantities. We compare projections for 

photovoltaics. Recent dramatic and unexpected cost reductions suggest systematic 

overconfidence in expert assessments.

4 - Harmonization and Aggregation of Energy Technology 

Elicitations 

Erin Baker, University of Massachusetts, Amherst, MA, 


There have been a number of expert elicitations performed on energy 

technologies recently, including 5 studies on solar and 7 on CCS. In this project 

we are attempting to harmonize and then aggregate the data from disparate 

studies. Harmonizing the data across the elicitations has a number of specific 

challenges, as surveys differ along several dimensions, including the metrics 

elicited, the timing, and the levels of R&D funding considered. 

■ SB03 

03- West 102 B- CC 

Behavioral Decision Making 



Chair: Manel Baucells, Rand Corporation and University Pompeu 

Fabra, 1776 Main Street, Santa Monica, CA, 90401, 

United States of America, manel.baucells@upf.edu 

1 - Reference Price Comparison, Mental Accounting, and 

Anomalies in Consumption-Payment Decisions 

Woonam Hwang, PhD Student, London Business School, 

Regent’s Park, London, NW1 4SA, United Kingdom, 

whwang.phd2009@london.edu, Manel Baucells 

We propose a positive theory of consumer choice. When evaluating the hedonic 

benefits of consumption and payment decisions, consumers engage in a double 

comparison: one between the benefit of consumption and the reference price, and 

another between the reference price and the actual price. These comparisons are 

governed by proposed laws of mental accounting. Under standard assumptions, 

our model predicts various consumer choice anomalies such as sunk cost effects 

and the flat-rate bias. 

2 - Felicity during Anticipation and Recall 

Manel Baucells, Rand Corporation and University Pompeu Fabra, 

1776 Main Street, Santa Monica, CA, 90401, 

United States of America, manel.baucells@upf.edu, Silvia Bellezza 

We propose a set of psychologically plausible assumptions that yield the felicity, 

moment-utility, function of a given episode, including the anticipation and recall. 

The felicity function is U-shaped during anticipation. Shortening anticipation 

makes the event more surprising, and leads to an increase in utility from recall. 

The model has prescriptive recommendations to help cope for negative events. 

We provide empirical evidence in favor of the main implications of the model. 

3 - Model-based Correction of Overprecision Bias in Subjective 

Confidence Intervals 

Venkat Prava, Johns Hopkins University, 3400 N. Charles Street, 

313 Ames Hall, Baltimore, MD, 21218, United States of America, 

avadhanulu@gmail.com, Robert Clemen, Benjamin Hobbs, 

Melissa A. Kenney 

Elicited confidence intervals tend to be overly narrow. We propose a model to 

correct this bias based on the normal and bi-normal distributions. Model 

parameters are estimated by adjusting expert responses using a sufficient number 

of questions with known answers. We apply these models to datasets of expert 

elicitations, testing model validity with holdout analysis and quantile-quantile 

plots. The bi-normal model fits better, indicating an asymmetric overprecision 

bias. 

■ SB04 

04- West 102 C- CC 

Uncovering Risk and Probabilities 


Chair: Saurabh Bansal, Assistant Professor, Pennsylvania State 

University, 405 Business Building, University Park, PA, 

United States of America, sub32@psu.edu 

1 - Stakeholder Perspectives on Risk Scenarios for EMR 

Implementing 

Chi-Chang Chang, Assistant Professor, Chung-Shan Medical 

University Hospital, 110, Sec. 1, Chien-Kuo N. Rd., Taichung, 

Taiwan-ROC, changintw@gmail.com, Yu-Hung Cheng 

The purpose of this research is to elicit the risk scenarios and stakeholders’ risk 

priority during EMR project implementing. We adopt the intuitionistic fuzzy 

theory for risk management. We conducted a field study for a medical center of 

middle Taiwan to comment the risk of identification, analyses, measurement and 


81 

control, respectively. Based on the result of this study, we found that the risk 

measurement of the IFS be able to elicit the collective risk scenarios ranking of 

stakeholders. 

2 - A New Method for Valuing Probability Assessment 

Brad Powley, PhD Candidate, Stanford University, 

475 Via Ortega, Stanford, CA, 94305, United States of America, 

bpowley@stanford.edu 

A fundamental step in decision analysis is the encoding of probabilities. This 

paper introduces a method for valuing probability assessments on a continuous, 

uncertain quantity in a way that recognizes an expert’s inconsistency and the cost 

of the assessment. I will demonstrate this method using the example of a 

pharmaceutical company’s decision whether to market versus license a drug. 

3 - Caveats of Decision Rules for Comparing Alternatives under 

Incomplete Preference Information 

Antti Punkka, Systems Analysis Laboratory, Aalto University 

School of Science, Otakaari 1 M, Espoo, 02150, Finland, 

antti.punkka@aalto.fi, Ahti Salo 

We show that the recommendations provided by several decision rules may 

depend on the normalization of the additive value function. This means, for 

instance, that these rules may exhibit rank reversals if the two consequences that 

are selected for normalizing the value function are replaced by other 

consequences. As a remedy, we propose the analysis of all rankings generated by 

those value functions that are consistent with the stated incomplete preference 

information. 

4 - Profile-based Bayesian Method for Assessing Process Risk 

Chang-Ho Chin, Associate Professor, Kyung Hee University, 

Department of Industrial and Management, Yongin-si, Korea, 

Republic of, chin@khu.ac.kr, Aida Mercado Hernandez 

A process is a sequence of interdependent and linked activities to accomplish a 

predetermined goal. The early detection of process instances terminated without 

reaching the goal reduces the loss of resources. We propose a likelihood-based 

Byesian method to estimate the abnormal termination probability of process 

instances in real time. The estimated probability could be used as a precaution for 

the worst or a basis for go/no-go decision. 

5 - Probability Elicitation in the Presence of Elicitation Errors 

Saurabh Bansal, Assistant Professor, Pennsylvania State University, 

405 Business Building, University Park, PA, United States of 

America, sub32@psu.edu, Genaro Gutierrez 

We develop two analytical approaches to determine the parameters of a 

distribution using elicited fractile values in the presence of elicitation errors. Both 

approaches minimize the variance in the estimation error for the parameters of 

the distribution. We show that both these approaches should be preferred over 

approaches that ignore elicitation error and/or elicit only a specific set of fractiles. 

■ SB05 

SB05 

05- West 103 A- CC 

IIE Transactions 



Chair: Jianjun (Jan) Shi, The Carolyn J. Stewart Chair and Professor, 

Georgia Institute of Technology, 765 Ferst Drive, Room 109, Atlanta, 

GA, 30332-0205, United States of America, jianjun.shi@isye.gatech.edu 

1 - A Bayesian Framework for Online Parameter Estimation and 

Process Adjustment using Categorical Observations 

Kaibo Wang, Associate Professor, Tsinghua University, 

Department of Industrial Engineering, Beijing, China, 

kbwang@tsinghua.edu.cn, Jing Lin 

In certain manufacturing processes, low resolution categorical observations could 

be obtained as feasible and low-cost surrogates for accurate readings. In this work, 

a new online approach for parameter estimation and run-to-run process control 

using categorical observations is developed. The new approach is built in the 

Bayesian framework; it provides a convenient way to update parameter estimates 

when categorical observations arrive gradually in a real production scenario. 

2 - Multiscale Monitoring of Autocorrelated Processes using 

Wavelets Analysis 


United States of America, kamip@umich.edu, Judy Jin, 

Huairui Guo 

This paper proposes a new method to develop multiscale monitoring control 

charts for an autocorrelated process that has an underlying unknown ARMA 

(2,1) structure. The Haar wavelet transform is used to obtain effective monitoring 

statistics by considering the process dynamic characteristics in both time and 

frequency domains. Through several simulation scenarios, we show that proposed 

control charts outperform other existing charts in terms of detecting changes in 

an autocorrelated process.

SB06 

■ SB06 

06- West 103 B- CC 

Analysis of Steady-State Simulation Output 

Processes 



Chair: Emily Lada, Operations Research Specialist, SAS, SAS Campus 

Dr., Cary, NC, 27513, United States of America, Emily.Lada@sas.com 

1 - Evaluation of Batched Quantile Estimation Methods 

Dave Goldsman, Georgia Institute of Technology, School of ISYE, 

Atlanta, GA, 30332, United States of America, sman@gatech.edu, 

Christos Alexopoulos, James Wilson 

We study the performance of asymptotically valid confidence intervals (CIs) 

methods for steady-state quantiles computed from nonoverlapping batches. Our 

results form the basis for the development of fully sequential procedures that 

yield CI estimators of steady-state quantiles with user-specified absolute or 

relative precision. 

2 - Initialization of Finite-horizon Simulations 

Peter Glynn, Professor, Stanford University, Huang Engineering 


glynn@stanford.edu, Eunji Lim 

We introduce a new type of initialization problem that arises in real-time 

simulations. In such applications, the state needed to initialize the simulation is 

not completely observable, either because monitoring the physical system does 

not provide complete information or because the model includes artificial states 

that are not observable. We discuss the new initialization issues, make 

connections to the filtering literature, and present two algorithms to address these 

issues. 

3 - The Simulation Start-up Problem: Performance Comparison of 

Recent Solution Procedures 

James Wilson, Professor, Edward P. Fitts Department of Industrial 

Engineering, North Carolina State University, Raleigh, NC, 27695- 

7906, United States of America, jwilson@ncsu.edu, Anup Mokashi, 

Emily Lada 

We summarize the results of a comprehensive experimental performance 

comparison of some recent procedures for solving the simulation start-up 

problem, including the methods incorporated in MSER-5, N-Skart, waSSP, and 

SBatch. Substantial differences in performance are observed, especially in 

comparisons of MSER-5 with the latter three procedures. 

■ SB07 

07- West 104 A- CC 

INFORMS DM Student Paper Award Competition 

Sponsor: Data Mining 


Chair: Peter Qian, University of Wisconsin, Madison, WI, 

United States of America, peterq@stat.wisc.edu 

1 - INFORMS Data Mining Best Student Paper Competition 

Peter Qian, University of Wisconsin, Madison, WI, 

United States of America, peterq@stat.wisc.edu 

Four selected finalists will present their papers in the INFORMS DM Best Student 

Paper competition. The winner will be announced at the INFORMS DM business 

meeting and all finalists will receive an award certificate. 

■ SB08 

08- West 104 B- CC 

Joint Session Doing Good with Good OR/SPPSN: 

Doing Good with Good OR Competition: 

Finalist Presentations I 

Cluster: Doing Good with Good OR Student Competition & Public 

Programs, Service and Needs 


Chair: Sila Cetinkaya, Texas A&M University, College Station, TX, 

United States of America, sila@tamu.edu 


82 

1 - The Effect of Budgetary Restrictions on Breast Cancer 

Diagnostic Decisions 

Mehmet Ayvaci, Assistant Professor, The University of Texas at 

Dallas, Jindal School of Management, 800 West Campbell Road, 


ayvaci@wisc.edu, Elizabeth Burnside, Oguzhan Alagoz 

We develop a finite-horizon constrained Markov Decision Process to model 

diagnostic decisions after mammography where we maximize the total expected 

quality adjusted life years (QALYs) of a patient under resource constraints. 

Comparing to actual clinical practice, using optimal thresholds may result in 22% 

cost savings without sacrificing QALYs. Our modeling framework could be used 

for evaluating cost-effectiveness of diagnostic procedures. 

2 - Dynamic Monitoring of Chronic Disease 

Jonathan Helm, Assistant Professor, Indiana University, Kelley 

School of Business, ODT, 1309 East Tenth Street, Bloomington, IN, 

47405, United States of America, jhelm@umich.edu, Gregg Schell 

This research develops new methods for dynamically monitoring chronic disease 

to allow better detection of disease worsening with fewer tests. We developed a 

prototype and tested it on a major clinical trial with results that significantly 

outperform current practice. Our team, including engineers, a clinician, and an 

epidemiologist, is working toward integrating our algorithms into widely used 

glaucoma testing machines. 

3 - Optimal Distribution of Medical Backpacks and Health 

Surveillance Assistants in Malawi 

Amber Kunkel, Rice University, 6100 Main Street, Houton, TX, 

77005, United States of America, agkunkel@gmail.com, 

Elizabeth Van Itallie, Duo Wu 

The Malawian people face severely limited healthcare access. To address this 

issue, we used large-scale p-median and capacitated facility location problems to 

create a scalable, three-tiered plan for optimal allocation of Health Surveillance 

Assistants (HSAs), HSA designated medical backpacks, and backpack resupply 

centers in Malawi. This plan will be used to direct the Beyond Traditional Borders’ 

HSA backpack program scale-up. 

■ SB09 

09- West 105 A- CC 

Neurophysiology and Decision Making 



Chair: Niklas Ravaja, Professor, Aalto University, School of Economics, 

P.O. Box 21255, Helsinki, 00076, Finland, ravaja@mappi.helsinki.fi 

1 - Neurophysiological Methods in Decision Making Research 

Niklas Ravaja, Professor, Aalto University, School of Economics, 

P.O. Box 21255, Helsinki, 00076, Finland, 

ravaja@mappi.helsinki.fi, Pekka Korhonen, Outi Somervuori, 

Jyrki Wallenius, Murat Koksalan 

Neurophysiological or psychophysiological measures, such as 

electroencephalography (EEG), facial electromyography (EMG), 

electrocardiography (ECG), and electrodermal activity (EDA), have attained 

increasing attention in studying emotional and cognitive processes during 

decision making. This presentation will discuss the use of these measures. In 

addition, some recent decision making studies employing neurophysiological 

measures will be presented. 

2 - Purchase Behavior and Psychophysiological Responses to 

Different Price Levels 

Outi Somervuori, Aalto University School of Economics, 00076, 

Finland, outi.somervuori@aalto.fi, Niklas Ravaja 

The aim of the study was to examine emotional processes in purchase decision. 

The participants were presented purchase decision trials with 14 different 

products whose price levels were changed while their facial electromyography 

(EMG) and electrodermal activity (EDA) were recorded. The results suggest that 

low prices and national brand products induce higher positive emotions compared 

to high prices and private label products. Also, positive emotions are related to 

greater purchase intent.

■ SB10 

10- West 105 B- CC 

A Robust Optimization Approach to 

Stochastic Analysis 



Chair: Nataly Youssef, Graduate Student, Massachusetts Institute of 

Technology, 143 Albany St., Apt. 014C, Cambridge, MA, 02139, 

United States of America, youssefn@mit.edu 

1 - Robust Queueing Theory 

Chaithanya Bandi, PhD Student, Massachusetts Institute of 

Technology, 02139, United States of America, cbandi@mit.edu, 

Nataly Youssef, Dimitris Bertsimas 

We study queueing systems by employing robust optimization as opposed to 

stochastic analysis. We set the limiting laws of probability as the axioms of our 

approach and model queue primitives by uncertainty sets. We analyze single-class 

multi-server queueing networks and obtain closed form expressions for the 

waiting times with heavy-tailed arrival and service processes. Our approach yields 

accurate results compared to simulations and is fairly insensitive to network size 

and traffic intensity. 

2 - Network Information Theory via Robust Optimization 

Dimitris Bertsimas, Massachusetts Institute of Technology, 77 

Massachusetts Avenue, Cambridge, MA, United States of America, 

dbertsim@mit.edu, Chaithanya Bandi 

We provide via robust optimization (RO) inner and outer bounds for the capacity 

region of multi-user channels with interference for finite code length n and a 

code that matches the inner bound. As n increases,the bounds are tight and lead 

to an asymptotic characterization of the capacity region which is a well-known 

open problem. The RO problems take the form of rank minimization SDPs for 

Gaussian and MILPs for Exponential, Binary and Uniform channels. We report 

results for up to 200000 codewords. 

3 - Pareto Efficiency in Robust Optimization 

Nikos Trichakis, Assistant Professor, Harvard Business School, 

Boston, MA, United States of America, ntrichakis@hbs.edu, 

Dan Iancu 

We argue that the classical robust optimization (RO) paradigm need not produce 

solutions that possess the (suitably adapted) property of Pareto optimality. We 

illustrate how this could lead to inefficiencies and suboptimal performance in 

practice. We introduce and provide a basic theoretical characterization of Pareto 

robustly optimal solutions. Numerical studies demonstrate that such solutions 

have significant upside compared with ones obtained via classical RO, at no extra 

cost or downside. 

■ SB11 

11- West 105 C- CC 

Application of Stochastic Programming I 


Chair: Yang Tan, FedEx Services, 3640 Hacks Cross Rd., Memphis, TN, 

38125, United States of America, ytan0920@gmail.com 

1 - A Scenario Generation-based Lower Bounding Approach for 

Stochastic Scheduling Problems 

Lingrui Liao, Google Inc., 1550 Technology Dr., Unit 3085, 

San Jose, Ca, 95110, United States of America, lingrui@gmail.com, 

Hanif D. Sherali, Subhash C. Sarin 

We investigate scenario generation methods to establish lower bounds on the 

optimal objective value of stochastic scheduling problems. In contrast to the 

Sample Average Approximation approach, we use the idea of recursive stratified 

sampling to develop a discrete bounding method. Exact lower bounds for 

expectation and conditional value-at-risk objectives are developed, and are 

demonstrated using a single machine total weighted tardiness problem with 

experimental results. 

2 - A Stochastic Programming Approach for Decision-dependent 

Uncertainty in Production Planning 

Alwin Haensel, Université Paris Sud, Bât 650 Université Paris-Sud 

11, Orsay, 91405, France, haensel.alwin@gmail.com, 

Marco Laumanns 

Regulation pressure on production quality and standards is increasing. A 

production plan needs to consider risks of failing the inspections, whose 

realizations dependents on the production schedule. This dependency increases 

the complexity significantly. The uncertain inspection realizations are modeled by 

scenarios generated from given product-site hazards. We propose a general 

scenario based stochastic programming approach to compute solutions for riskneutral 

and risk-averse decision makers. 


83 

3 - Stochastic Approach for Resource Allocation and 

Facility Utilization 

Anirudha Kulkarni, Graduate Research Assistant, Rochester 

Institiute of Technology, Rochester, Rochester, NY, 14623, 

United States of America, apk2932@rit.edu, Scott Grasman, 

Shrikant Jarugumilli 

We formulate a stochastic resource allocation model that considers fluctuating 

demand over a long planning horizon. The stochastic program includes hiring, 

firing, overtime, shift-swapping and cross training. The rationale behind this is to 

reduce gaps in assignments and ensure efficient resource utilization. Results 

include discussion of the recourse, value of the stochastic solution, and 

implementation. 

4 - Stochastic Lot-Sizing Model for Deteriorating Items under 

Partial Backlogging 

Yang Tan, FedEx Services, 3640 Hacks Cross Rd., Memphis, TN, 

38125, United States of America, ytan0920@gmail.com 

This research is distinguished from most literature by considering the effects of 

deterioration and partial backlogging under stochastic customer demand. The 

optimal conditions under which a (s, S) policy holds are successfully derived and 

the explicit order quantities are obtained by reformulating the problem as a 

stochastic programming model. 

5 - Linear Profile Monitoring 

Azadeh Adibi, Arizona State University, 1265 E. University Dr, 

Tempe, AZ, United States of America 

In this study,a method is proposed for phase 2 monitoring of linear profiles. Also, 

performance of the proposed method is compared with an existing method based 

on average run length criterion under different shifts in the model parameters. 

■ SB12 

SB12 

12- West 106 A- CC 

Structure of Convex Integer Programs 



Chair: Marco Molinaro, Carnegie Mellon University, 5000 Forbes Ave, 

Pittsburgh, PA, United States of America, molinaro@cmu.edu 

1 - Closedness of Mixed Integer Hulls of Second Order Conic 

Programming Problems 

Diego Moran, Georgia Tech, 755 Ferst Drive NW, Room 422, Main 

building, Atlanta, GA, 30332, United States of America, 

dmoran@gatech.edu, Santanu Dey 

Given rational matrices $A$, $B$ and vector $b$, we show that there exists a 

polynomial time algorithm to check the closedness of the convex hull of the set 

$\{(x,y)|Ax +By -b \in L, x integral\}$, where $L$ is the Lorentz cone. Moreover, 

in the special case of pure integer problems, we present sufficient conditions for 

verifying the closedness of the integer hull of the feasible region of a more general 

second order conic programming problem (SOCP) that can also be checked in 

polynomial time. 

2 - Disjunctive Cuts for Convex Mixed Integer Nonlinear Programs 

(MINLP)s and Extended Formulations 

Mustafa Kilinc, Postdoctoral Researcher, University of Pittsburgh, 

3700 O’Hara Street, Pittsburgh, United States of America, 

mrk46@pitt.edu, Sina Modaresi, Jeff Linderoth, James Luedtke, 

Juan Pablo Vielma 

In this work, we study positive effects of extended formulations on generating 

cutting planes for convex MINLPs. First, stronger cuts can be generated in an 

extended formulation. Second, the performance of a pure LP-based cut 

generation method can be improved via extended formulations. 

3 - On the Relative Strength of Different Generalizations of Split 

Cuts 

Marco Molinaro, Carnegie Mellon University, 5000 Forbes Ave., 

Pittsburgh, PA, United States of America, molinaro@cmu.edu, 

Sanjeeb Dash, Oktay Gunluk 

In this paper we show the complete relationship between the split, cross, crooked 

cross and t-branch split closures. We also study the strength of cross cuts from 

basic and 2-row relaxations. This answers three open questions left open by Day, 

Dash and Gunluk (2010).

SB13 

4 - Lower-bound Analyses of the Lift-and-Project Ranks of 

Graph-based Polytopes 

Yu Hin Au, PhD Candidate, University of Waterloo, 200 University 

Ave., W., Waterloo, ON, N2L3G1, Canada, yau@uwaterloo.ca 

We’ll discuss some ongoing work on obtaining inapproximability results for 

Bienstock-Zuckerberg and Lasserre lift-and-project algorithms on graph-based 

optimization problems. This is joint work with Levent Tuncel. 

■ SB13 

13- West 106 B- CC 

Computational Issues on Solving Mixed Integer 

Second Order Cone Optimization Problems 



Chair: Julio Cesar Goez, PhD Candidate, Lehigh University, 18015, 

United States of America, jgoez1@gmail.com 

1 - Computational Effectiveness of Split Cuts for Second-order 

Conic Programming 

Sina Modaresi, University of Pittsburgh, 3700 O’Hara Street, 

Pittsburgh, United States of America, sim23@pitt.edu, 

Mustafa Kilinc, Juan Pablo Vielma 

Split cuts are one of the most effective cuts for linear MIPs and they are 

equivalent to MIR cuts. We show that this equivalency does not hold for secondorder 

conic MIPs by giving examples where split cuts strictly dominate conic MIR 

cuts proposed by Atamturk et al.. However, split cuts have to be added as 

nonlinear inequalities, while conic MIR cuts can be added as linear inequalities to 

the extended formulation. We compare the trade-off between the speed and the 

strength of these cuts. 

2 - Branch and Bound for Mixed Integer Second Order Cone 

Optimization: Impact of Disjunctive Conic Cuts 

Julio Cesar Goez, PhD. Candidate, Lehigh University, 18015, 

United States of America, jgoez1@gmail.com, Pietro Belotti, 

Ted Ralphs, Tamás Terlaky, Imre Polik 

We investigate the use of Disjunctive Conic Cuts (DCC) in a Branch and Cut 

framework when solving MISOCO problems. Various criteria are explored to 

select the disjunctions to build DCCs at different nodes of the tree. Additionally, 

we explore different criteria for node selection and branching rules. These 

experiments will help us to understand the impact that DCCs have in decreasing 

the size of the search tree and the solution time. 

3 - Restrict-and-Relax Search for 0-1 Integer Programs 

Menal Guzelsoy, SAS, 100 SAS Campus Drive, Cary, NC, 7513, 

United States of America, menal.guzelsoy@sas.com, 

George Nemhauser, Martin Savelsbergh 

We introduce restrict-and-relax search, an algorithm for 0-1 integer programs 

that explores a dynamic search tree by not only fixing variables (restricting), but 

by also unfixing previously fixed variables (relaxing). Starting by solving a 

restricted integer program, we may at any tree node selectively relax/restrict 

variables using dual/structural information. A proof-of-concept computational 

study demonstrates the effectiveness of the algorithm. 

4 - A Regularized Interior-Point Method for 

Semidefinite Programming 

Ahad Dehghani, McGill University, Montreal, Canada, 

ahad.dehghani@mcgill.ca, Jean-Louis Goffin, Dominique Orban 

Interior-point methods in semi-definite programming (SDP) require the solution 

of a sequence of linear systems which are used to derive the search directions. 

Safeguards are typically required in order to handle rank-deficient Jacobians and 

free variables. We propose a primal-dual regularization to the original SDP and 

show that it is possible to recover an optimal solution of the original SDP via 

inaccurate solves of a sequence of regularized SDPs for both the NT and dual 

HKM directions. 


84 

■ SB14 

14- West 106 C- CC 

Uncertainty in Project Management 


Chair: Bajis Dodin, Professor, University of California, School of 

Business Administration, Riverside, CA, 92521, United States of 

America, bajis.dodin@ucr.edu 

1 - Enumerating Feasible Task Sets with Ordered, Unmeasured 

Material Constraints 

Jordan Srour, Assistant Professor, Lebanese American University, 

P.O. Box 13-5053, Beirut, Lebanon, fjsrour@gmail.com, 

Walid Nasrallah 

We investigate the solution space for selecting a subset of project tasks under a 

resource shortage. The amount of resource required by each task is not known, 

but the tasks are fully ordered accordingly. We enumerate the ways to select a 

subset of tasks without exceeding the resource constraint. We highlight the 

relationship of this enumeration to published number sequences arising from 

regular Boolean functions. We provide both an enumeration algorithm and 

enumerations up to n=10. 

2 - Stochastic Approach for Project Scheduling Based on 

Fund Availability 

Yuvraj Gajpal, Assistant Professor, King Fahd University of 

Petroleum and Minerals, KFUPM Box 634, Dhahran, 

Saudi Arabia, gajpaly@gmail.com, Ashraf Elazouni 

The paper considers a finance based project where the contractors finance projects 

mainly through the owners’ progress payments supplemented by fund procured 

through establishing credit-line accounts. In this situation the best proactive 

approach for contractor is to schedule construction activities based on the 

available finance. A stochastic heuristic approach is proposed to device financebased 

schedules of multiple projects. 

3 - A Nonlinear Programming Model for Stochastic Project 

Crashing 

Ronald Davis, Associate Professor, San Jose State University, 

College of Business, One Washington Square, San Jose, CA, 

95192, United States of America, ronald.davis@sjsu.edu 

When beta distributions are used for every activity early start and finish time 

distribution in a project network, moment preserving beta approximations for the 

sum and product of beta cdfs can be coded in VBA to allow an analytic stochastic 

forward pass to be carried out, without simulation. Introduction of crashing 

variables and costs yields a nonlinear programming formulation to minimize 

crash cost subject to a constraint on mean project duration. Realistic example 

SOLVER results are shown. 

4 - Portfolio Management in a Highly Uncertain Environment: The 

Role of Interdependencies 

Olga Kokshagina, PhD Student, Centre for Scientific Management 

(CGS), CGS Mines ParisTech, 60 Boulevard Saint-Michel, Paris, 

75272, France, olga.kokshagina@mines-paristech.fr, Patrick Cogez, 

Pascal Le Masson, Benoit Weil 

This work deals with portfolio management strategies in high uncertainty. There 

exist strategies that consider projects separately or dependently. Literature shows 

first, the dependencies have a tendency to increase complexity and cost of the 

system. Second, there is a possibility to share risks in between projects, to 

highlight the effect of learning. Our paper investigates the contradictory role of 

interdependencies and the relevant management strategies in particular industrial 

situations. 

5 - Scheduling & Financial Planning in Probabilistic Projects 

Bajis Dodin, Professor, University of California, School of Business 

Administration, Riverside, CA, 92521, United States of America, 

bajis.dodin@ucr.edu, Abdelghani Elimam 

In probabilistic projects (PP) required duration and resources for some or all 

activities are given as random variables characterized by their own probability 

distribution functions (PDFs). Managing a PP requires dealing with several 

important issues. In this paper, we analyze the impact of the various stochastic 

variations on the duration, and cost of the project. Procedures for determining the 

PDFs of project cost and duration, and project schedule are developed.

■ SB15 

15- West 202- CC 



1 - Ziena Optimization LLC - New Developments in the KNITRO 

Optimization Solver 

Richard Waltz, President, Ziena Optimization LLC, 1801 Maple 

Avenue, Ste. 6305, Evanston, IL, 60201 United States of America, 

waltz@ziena.com 

This software demonstration will highlight the latest developments and 

improvements in the KNITRO optimization solver. Benchmark results will be 

provided for various problem classes. The demo will also provide an overview of 

how to effectively use KNITRO in a variety of environments and interfaces. 

2 - Artelys - Optimizing Military Attacks on Multi-Commodity 

Supply Chain Networks: a Min-Max Problem Solved with Artelys 

Kalis 

Sylvain Mouret, Product Manager, Artelys Kalis, 150 N. Michigan 

Avenue, Ste. 800, Chicago IL, 60601, United States of America, 

ain.mouret@artelys.com 

Learn how easy it is to solve general min-max problems arising in adversarial 

games with Artelys Kalis. Thanks to its hybrid CP/MIP architecture, it is possible 

to build an efficient branch and bound method specialized for such problems in 

just a few lines of code. We will demonstrate how Kalis has been embedded in a 

full-range operational solution based on the Artelys Network Designer GIS 

visualization platform. This solution is currently used to study supply chain 

network vulnerabilities with respect to military attacks. 

■ SB16 

16- West 207- CC 

Multi-agent Scheduling Problems 



Chair: Alessandro Agnetis, Professor, Universita di Siena, Dipartimento 

Ingegneria Informazione, via Roma 56, Siena, 53034, Italy, 

agnetis@dii.unisi.it 

1 - Multi-agent Scheduling Problems – Scheduling Problems 

Alessandro Agnetis, Professor, Universita di Siena, Dipartimento 

Ingegneria Informazione, via Roma 56, Siena, 53034, Italy, 

agnetis@dii.unisi.it 

In Multi-Agent Scheduling problems, a number of distinct agents, each owning a 

subset of all jobs, share common processing resources. The analysis in this case 

focuses on finding efficient and, possibly, fair solutions. After presenting some 

motivation, we review basic issues, illustrate the main complexity results, discuss 

related problems and point out future research needs. 

■ SB17 

17- West 208 B- CC 

SpORts I 

Sponsor: SpORts 


Chair: Dan Finkel, Technical Staff, Massachusetts Insitute of 

Technology, Lincoln Laboratory, 244 Wood Street, Lexington, MA, 

02420, United States of America, dfinkel@ll.mit.edu 

1 - Serving Strategy in Tennis 

Yigal Gerchak, Professor, Tel Aviv University, Ramat Aviv 69978, 

Israel, yigal@post.tau.ac.il, Marc Kilgour 

Common practice in Tennis is to hit a powerful first serve. Should it fault, the 

second serve is usually much weaker. Recently it has been argued that this 

strategy is flawed - both serves should be powerful. According to our analysis, 

however, the current practice of a weaker second serve appears to be roughly 

optimal. 

2 - On the (Non-Linear) Probability of Winning a (College) 

Football Game 

Eric Huggins, Associate Professor of Management, Fort Lewis 

College, Durango, CO, 81301, United States of America, 

huggins_e@fortlewis.edu 

In this paper, we estimate the probability that a football team wins a game based 

on the point spread and the over/under. This work is an extension of the Hal 

Stern’s 1991 paper with several key differences: 1) We analyze data for both 

college and professional football, using a much larger data set. 2) We factor in the 

over/under as well as the point spread. 3) We explore several non-linear models 

for the probability function. 


85 

3 - In-play Football Prediction 

Nils Rudi, Professor of Technology and Operations Management, 

INSEAD, 1 Ayer Rajah Ave., Singapore, 138676, Singapore, 

Nils.RUDI@insead.edu, Harry Groenevelt, Bhavani Shanker Uppari 

We model the scoring of two football teams as two Poisson processes. For a given 

pre-game predictive distribution, we show that there exists a unique pair of 

arrival rates, which facilitate in-play prediction as function of time and score. This 

result is extended to time and score dependent arrival rates, and comparisons 

with Logistic regression is provided. We test the models on an extensive data set 

from major European football leagues. 

■ SB18 

18- West 208 A- CC 

Applications of Network Optimization 


SB18 

Chair: Alireza Kabirian, Assistant Professor, California State University 

Northridge, 18111 Nordhoff St, Northridge, CA, 91330, 

United States of America, akabirian@csun.edu 

1 - Work Scheduling for Logistics Network 

Dening Peng, PhD Candidate, Arizona State University, 

1440 E. Broadway Rd., Tempe, AZ, 85282, United States of 

America, dening.peng@asu.edu, Pitu Mirchandani 

A network optimization model is built for work scheduling on a logistics network, 

that exploits features of minimum-cost-flow and scheduling models. Flows on a 

certain link can exceed the link capacity but will incur a flow cost that is much 

higher than regular flow cost. The objective of the problem is to maintain (or 

repair or expand) links within a given time horizon, while minimizing the total 

maintenance and flow costs. Experiments are reported on several test networks. 

2 - Transmission Network Expansion Planning under Demand 

Uncertainty and Risk Aversion 

Joao Claro, INESC Porto, Campus da FEUP, Rua Dr. Roberto Frias, 

378, Porto, 4200-465, Portugal, jclaro@fe.up.pt, Daniel Delgado 

Explicit incorporation of uncertainty in transmission network design can help 

improve the balance between important concerns such as network utilization, 

demand satisfaction, or dynamic sourcing. We present a mean-risk mixed integer 

linear programming model for transmission network expansion planning, and use 

it to search for network design insights, with a study of loss-averse design of three 

fundamental network building blocks - an independent design, a radial design, 

and a meshed design. 

3 - Modeling Wildfire Propagation using Stochastic Shortest Path 

Mohammad Hajian, PhD, Northeastern University, 39 Orient 

Street, Malden, MA, 02148, United States of America, 

mohammad.hajian@gmail.com, Emanuel Melachrinoudis 

In this paper, a stochastic shortest path algorithm is used to model the fire 

propagation on a landscape. The fire landscape is modeled as a network by using 

Delaunay triangulation. The time of fire traversal along each edge of the network 

is considered to be stochastic, depending on factors such as wind speed and 

direction, terrain, and weather conditions. The shortest path algorithm is used to 

find the probability distribution of fire arrival at a point of interest. 

4 - Role of Information Sharing in Interdependent Dynamic 

Infrastructure Networks 

Burak Cavdaroglu, Rensselaer Polytechnic Institute, 110 8th 

Street, CII Building Suite 5015, Troy, NY, 12180, 

United States of America, cavdab@rpi.edu, William Alan Wallace 

In case of disruptions in civil infrastructure networks, a timely restoration of these 

networks is achieved with an effective selection and scheduling of the network 

components to be restored. In today’s society, an infrastructure’s restoration plan 

has a greater impact on another’s due to the increasing amount of 

interdependencies among them. Using data representing the infrastructures of 

New Hanover County, NC, we depicted the impact of sharing the restoration 

plans among the infrastructures. 

5 - Expansion Planning of Natural Gas Pipeline Networks 

Alireza Kabirian, Assistant Professor, California State University 

Northridge, 18111 Nordhoff St., Northridge, CA, 91330, 

United States of America, akabirian@csun.edu 

We develop optimization models for expansion planning of an existing natural gas 

pipeline network over a long run horizon. The model is to mainly find where new 

pipelines and compressor stations of the network must be installed to minimize 

the net present worth of capital investments and operating expenses. We also 

develop an evolutionary solution algorithm to solve the model. Numerical results 

will be presented at the end showing how the model and the solution algorithm 

are applied.

SB19 

■ SB19 

19- West 211 A- CC 

Joint Session Healthcare Logistics/SPPSN: Routing 

Problems in Healthcare Operations I 



Chair: Burcu Keskin, University of Alabama, Alston Hall, Box: 870226, 

Tuscaloosa, AL, 3587-0226, United States of America, 

bkeskin@cba.ua.edu 

1 - A Sequential GRASP for the Therapist Routing and 

Scheduling Problem 

Yufen Shao, Research Engineer, ExxonMobil Upstream Research 

Company, 3120 Buffalo Speedway, Houston, TX, 77098, United 

States of America, yufen.shao@exxonmobil.com, Ahmad Jarrah, 

Jonathan Bard 

This talk presents a sequential GRASP for solving a weekly routing and 

scheduling problem for therapists. The problem contains both fixed and flexible 

patients with respect to appointment times, and two grades of therapists. In Phase 

I, feasible solutions are constructed one therapist and one day at a time. In Phase 

II, a high-level neighborhood search is proposed to obtain local optima. 

Performance is demonstrated using real and randomly generated data sets. 

2 - An Optimisation Model for Staff Planning in a 

Home Care Organization 

Pablo Andres Maya Duque, PhD, University of Antwerp/ 

Universidad de Antioquia, Ottoveniusstrat 26, box 6, Antwerp, 

2000, Belgium, pmayaduque@gmail.com, Peter Goos, 

Kenneth Sörensen, Marco Castro 

In this talk, we present the core optimization component of a decision support 

system that Landelijke Thuiszorg, a non-profit organization that provides home 

care services for several provinces in Belgium, will implement in order to assist 

the regional service planning. The optimization model takes into account 

assignment, scheduling and routing decisions simultaneously, while considering 

two objectives, namely the service level and the travelled distance. 

3 - Incorporating Patient, Nurse, and Agency Considerations in 

Home Health Care Routing 

Ashlea Bennett Milburn, Assistant Professor, University of 

Arkansas, 4207 Bell Engineering Center, Fayetteville, AR, 

United States of America, ashlea@uark.edu, Jessica Spicer 

Home health routing and scheduling problems can be modeled as multi-objective 

optimization problems, as home health agencies are often interested in creating 

nurse routes that achieve a variety of goals. We use a multi-objective tabu search 

heuristic to study the relationship among travel cost, nurse consistency (a 

measure of patient satisfaction), and balanced workload (an indicator of nurse 

satisfaction) objectives. Computational results for a number of realistic scenarios 

are presented. 

4 - A Multi-period Home Care Scheduling Problem with 

Work Balance 

Burcu Keskin, University of Alabama, Alston Hall, Box: 870226, 


bkeskin@cba.ua.edu, Shirley (Rong) Li, Charles Schmidt 

We consider a home health care scheduling problem for a local hospital with 

multiple types of caregivers. We determine the assignment of caregivers to the 

patients visited at their homes so that the total routing costs are minimized and 

the workload of caregivers are balanced over a planning horizon while satisfying 

synchronization, precedence, loyalty, and other practical constraints. We present a 

MILP formulation and a branch-and-price solution approach based on Dantzig- 

Wolfe decomposition. 

■ SB20 

20- West 211B- CC 



1 - American Optimal Decisions - Portfolio Safeguard (PSG): 

Advanced Nonlinear Mixed-Integer Optimization Package 

Stan Uryasev,American Optimal Decisions, 5214 SW 91 Way, Ste. 

#130, Gainesville FL 32608, United States of America, 

uryasev@aorda.com 

Portfolio Safeguard is an advanced nonlinear mixed-integer optimization package 

used in risk management, financial engineering, military, medical and other 

applications. Design and solve complex optimization problems with built-in 

functions (maximum, StDev, variance, probability, VaR, CVaR, cardinality, fixedcharge, 

recourse etc.). See real-life case studies in Windows and MATLAB at 

www.aorda.com/aod/psg.action. 


86 

■ SB21 

21- West 212 A- CC 

Advances in Networks and Graphs 


Chair: Sivan Altinakar, École Polytechnique de Montréal, C.P. 6079, 

Succ. Centre-ville, Montreal, QC, H3C 3A7, Canada, 

sivan.altinakar@gerad.ca 

1 - Approximating Precedence Network Structure with 

Incomplete Information 

Adam Graunke, The Boeing Company, P.O. Box 3707, MC 13-98, 

Seattle, WA, 98124, United States of America, 

adam.a.graunke@boeing.com, Gabriel Burnett 

Accurate precedence networks are highly useful for planning and analysis, yet in 

large scale production systems, they are difficult, if not impossible, to define. In 

this research we investigate precedence networks in the presence of incomplete 

precedence information, with the goal of estimating production-critical properties 

of the network. Specifically, we investigate critical path analyses and the level of 

confidence associated with the results. 

2 - Shortest Path with Secure Multipary Computation 

Abdelrahaman Aly, Universite Catholique de Louvain, 34, 

Voie du Roman Pays, Louvain-la-Neuve, 1348, Belgium, 

abdelrahaman.aly@uclouvain.be, Mathieu Van Vyve 

In various applications, i.e. elections, auctions, the computation of a global 

optimum requires input data from competing parties. A trusted third party to 

perform these computations is not guaranteed. Secure Multiparty Computation 

(SMC) is an encryption method which does not rely on such third party. The 

present research expands SMC’s scenario to Shortest Path problem. We describe 

SMC variants of Bellman-Ford and Dijkstra algorithms and compare their 

performance with the traditional variants. 

3 - Reconstruction of Three Dimensional Objects from Three 

Orthogonal Cartesian Bi-plane Projections 

Siddhartha Sampath, Arizona State University, 

849 W Elna Rae, Tempe, AZ, United States of America, 

Siddhartha.Sampath@asu.edu, Pavithra Ramamoorthy, 

Pitu Mirchandani 

This application describes a network flow model for reconstructing the threedimensional 

shape of a three dimensional object from biplane x-ray readings for 

all three orthogonal Cartesian planes.Each two-dimensional cross-section consists 

of one region or view of the object whose image is to be reconstructed.We can use 

apriori information to constrain the number of solutions, and then to obtain most 

likely reconstruction. 

4 - Breaking Symmetry in Consecutive Edge-coloring 

Sivan Altinakar, École Polytechnique de Montreal, C.P. 6079, 

Succ. Centre-ville, Montreal, QC, H3C 3A7, Canada, 

sivan.altinakar@gerad.ca, Alain Hertz, Gilles Caporossi 

Consecutive edge-coloring, a special case of edge-coloring in graph theory, aims to 

minimize the sum of the span of colors (integers) incident to each node. This has 

applications in scheduling. It is also NP-hard, and may be difficult to solve exactly 

even for a very small number of vertices. We compare different modeling 

approaches in Mixed Integer Programming and Constraint Programming, with an 

emphasis on the latter, and subsets of Lex-Leader constraints for efficient 

symmetry breaking. 

■ SB22 

22- West 212 B- CC 

COIN-OR Multi-Threading Software 



Chair: Bradley Bell, Senior Research Scientist/Enginer, University of 

Washington, IHME, 2301 Fifth Ave., Suite 600, Seattle, WA, 98121, 

United States of America, bradbell@seanet.com 

1 - Could We Use a Million Cores to Solve a Single Integer 

Program? 

Ted Ralphs, Associate Professor, Lehigh University, 200 West 

Packer Avenue, Bethlehem, PA, 18015, United States of America, 

ted@lehigh.edu, Thorsten Koch, Yuji Shinano 

Given the steady increase in cores per CPU, it is only a matter of time before 

supercomputers will have a million or more cores. In this talk, we investigate the 

opportunities and challenges that will arise when trying to utilize the distributed 

multi-core architectures that have recently become pervasive to solve a single 

integer linear optimization problem. We raise the question whether best practices 

in sequential solution of ILPs will be effective in massively parallel environments.

2 - A Multi-core Benchmark used to Improve 

Algorithmic Differentiation 

Bradley Bell, Senior Research Scientist/Enginer, University of 

Washington, IHME, 2301 Fifth Ave., Suite 600, Seattle, WA, 

98121, United States of America, bradbell@seanet.com 

We present an application of Newton’s method that finds all the zeros of a nonlinear 

function in an interval. This is being used as a benchmark to improve the 

multi-core performance of the COIN-OR open source AD software package 

CppAD. Different threading systems, and different versions of this software 

package, are easily compared for speed of execution. As an example of the use of 

this benchmark, we compare the results for three versions of the software 

package. 

3 - Parallelization of a Primal-dual IPM for SDP using OpenMP 

Brian Borchers, New Mexico Tech, Mathematics Department, 

801 Leroy Place, Socorro, NM, 87801, United States of America, 

borchers@nmt.edu 

CSDP is software package for semidefinite programming that uses a primal-dual 

interior point method. We discuss how the software was parallelized using 

OpenMP, and discuss the parallel performance of the algorithm on problems of 

varying size. 

4 - Application of SIMD Instructions in Solving Random 

3-dimensional Assignment Problems 

Mohammad Mirghorbani, The University of Iowa, 3131 Seamans 

Center, Iowa City, IA, 52242, United States of America, 

seyedmohamma-mirghorbaninokandeh@uiowa.edu, 

Pavlo Krokhmal 

We show how SIMD instructions can be utilized to vectorize an existing branch 

and bound method, BitCLQ, designed to enumerate k-cliques in special k-partite 

graphs. The k-cliques found map to guaranteed high-quality solutions of random 

3-dimensional assignment problems. We observed up to 100% improvement in 

the performance of BitCLQ on the same machine after vectorization. 

■ SB23 

23- West 212 C- CC 

Edelman 2012 Finalists II 

Sponsor: CPMS, The Practice Section 


Chair: Stephen Graves, Massachusetts Institute of Technology, Sloan 

School of Management, 77 Massachusetts Avenue, Cambridge, MA, 

02139, United States of America, sgraves@MIT.EDU 

1 - Carlson Rezidor Hotel Group Maximizes Revenue with 

Improved Demand Management and Price Optimization 

Pelin Pekgun, Assistant Professor, University of South Carolina, 

Moore School of Business, 1705 College Street, Columbia, SC, 

29208, United States of America, Pelin.Pekgun@moore.sc.edu, 

Suresh Acharya, Kathleen Mallery, Kyle Christianson, Ronald 

Menich, Phillip Finch, Frederic Deschamps, James Van Sistine, 

James Fuller 

Carlson Rezidor Hotel Group collaborated with JDA Software Group on a highly 

innovative revenue optimization project that started with enterprise demand 

forecasting in 2007. It was followed by a large-scale network optimization 

solution to dynamically optimize stay night rates based on price elasticity, 

competitor rates, remaining inventory, demand forecast and business rules. 

Compliant hotels consistently showed a 2-4% revenue improvement,and 

increased revenue by more than $16 million annually. 

2 - Optimizing Capital Investment Decisions at Intel Corporation 

Feryal Erhun, Stanford University, Stanford, CA, United States of 

America, ferhun@stanford.edu, Karl Kempf, Erik Hertzler, 

Timothy Rosenberg, Chen Peng 

Intel spends over $5B annually on manufacturing equipment. With increasing 

lead times from equipment suppliers and increasing difficulty in forecasting 

market demand, optimizing capital investment decisions is a significant 

managerial challenge. We describe the Velocity Program and the DMEP 

framework, which have benefited Intel hundreds of millions of dollars in 

documented cost savings and at least $2B in revenue upside for a manufacturing 

process transition during a period of economic crisis. 


87 

■ SB24 

SB24 

24- West 213 A- CC 

Joint Session HAS/SPPSN: Pandemic Modeling 

and Planning 

Sponsor: Health Applications Society & Public Programs, 



Chair: Dionne Aleman, University of Toronto, 5 King’s College Road, 

Toronto, On, M5S 3G8, Canada, aleman@mie.utoronto.ca 

1 - Quantifying the Value of Information for Inferring Contagion 

Patterns in Social Activity Graphs 

Lauren Gardner, Lecturer, University of New South Wales, UNSW, 

Sydney, 2052, Australia, laurengardner84@gmail.com, David 

Fajardo 

The model presented here seeks to predict the infection pattern which depicts the 

current state of the network for an ongoing outbreak of a communicable disease. 

This is accomplished by inferring the most likely path of infection through a 

contact network under the assumption of partially available infection data. The 

problem is formulated as a bi-linear integer program, and heuristic solution 

methods are developed based on iteration of two sub-problems which can be 

solved much more efficiently. 

2 - Developing Non-pharmaceutical Intervention Strategies for 

Pandemic Influenza Mitigation 

Tapas Das, Professor, Department of Industrial and Management 

Systems Engineering, University of South Florida, Industrial 

Engineering, Tampa, FL, United States of America, das@usf.edu, 

Dayna Martinez 

In this paper we examine the interactions between the parameters of virus 

epidemiology, social behavior, and non-pharmaceutical interventions (NPIs). This 

examination is conducted using an agent-based simulation model and designed 

statistical experiments. Results are used to design effective NPI strategies for 

pandemic influenza mitigation. Some numerical results are presented for various 

virus transmissibility scenarios. 

3 - Evaluation of Strategies to Mitigate Contagion Spread using 

Social Network Characteristics 

Mario Ventresca, University of Toronto, 5 King’s College Road, 

Toronto, On, M5S 3G8, Canada, mventresca@gmail.com, 

Dionne Aleman 

We study the effects of graph-based mitigation strategies with respect to their 

ability to contain disease spread. Evaluation is performed on a social network 

constructed from census data of the Greater Toronto Area of 5 million individuals. 

One outcome of the analysis highlights the pitfalls of betweenness and centrality 

measures for maximum mitigation given an unknown initial spreader. Our 

analysis focuses on pandemic disease, but the approach is applicable to any 

network contagion. 

4 - Prediction of Demand Surges for Immunization During 

Influenza Pandemics 

Michael Beeler, University of Toronto, 5 King’s College Road, 

Toronto, ON, M5S 3G8, Canada, mfbeeler@gmail.com, 

Michael Carter, Dionne Aleman 

A time-series model is constructed to predict surges in turnout at mass 

immunization clinics during influenza pandemics. Variables include media 

coverage, weather conditions, day of the week, and estimated weekly infection 

rate. The model is based on Toronto clinic turnout data during the H1N1 

pandemic in 2009.

SB25 

■ SB25 

25- West 213 B- CC 

Radiation Therapy Treatment Planning I 



Chair: Edwin Romeijn, Professor, University of Michigan, IOE 

Department, 1205 Beal Avenue, Ann Arbor, MI, 48109-2117, 

United States of America, romeijn@umich.edu 

1 - A Column-generation-based Technique for Multi-criteria Direct 

Aperture Optimization 

Ehsan Salari, Research Associate, Massachusetts General Hospital 

and Harvard Medical School, Francis H Burr Proton Therapy, 

55 Fruit Street, Boston, MA, 02114, United States of America, 

salari.ehsan@mgh.harvard.edu, David Craft, Jan Unkelbach 

Multi-criteria optimization (MCO) has proved to be a promising approach to 

radiation therapy treatment planning. MCO is typically employed in the Fluencemap 

optimization stage to find a Pareto-optimal solution that yields the desired 

trade-off between treatment evaluation criteria. In this study we investigate the 

extension of the MCO approach to the direct aperture optimization problem and 

develop heuristics to obtain a collection of apertures that can approximate the 

Pareto surface. 

2 - Determining Convex Objective Functions and Importance 

Factors in Multi-criteria IMRT Planning 

Taewoo Lee, University of Toronto, 5 King’s College Road, Toronto, 

Canada, taewoo.lee@utoronto.ca, Michael Sharpe, Timothy Chan, 

Tim Craig 

In multi-criteria optimization, objective function parameters are often determined 

by a trial-and-error process. Multi-criteria IMRT planning typically involves many 

objective functions, which leads to a large parameter space to search over. We 

develop an inverse optimization method to determine convex objective functions 

and parameters that are most critical in treatment planning. Results show the 

potential to both streamline the planning process and increase the treatment 

effectiveness. 

3 - The Effect of Tumor Repopulation on Fractionation Schedules in 

Radiation Therapy 

Jagdish Ramakrishnan, Massachusetts Institute of Technology, 77 


jagdish@mit.edu, David Craft, Thomas Bortfeld, Jan Unkelbach, 

John N. Tsitsiklis 

We consider optimizing the fractionation schedule and the number of treatment 

days for radiation therapy. The tumor control probability is maximized subject to 

a constraint on the normal tissue complication probability. We consider both 

exponential and gompertzian tumor repopulation between treatment fractions in 

the linear-quadratic formalism. Such a framework provides insights as to which 

types of treatment protocols (e.g., hypo-fractionation) are beneficial for various 

disease sites. 

4 - A Method for Improving the Dose Distribution Quality of 

Multi-criteria Radiation Therapy Plans 

Rasmus Bokrantz, KTH Royal Institute of Technology, SE-100 44, 

Stockholm, Sweden, bokrantz@kth.se 

This talk considers an approach to radiation therapy planning where possible 

treatment options are explored through realtime interpolation over precomputed 

solutions. A method is presented that improves the quality of interpolated 

solutions by minimizing a projective distance to the nondominated frontier under 

constraints on maintained dose distribution quality. Also, minimization of dose 

changes during conversion into deliverable machine settings are discussed in view 

of the presented method. 

■ SB26 


Emerging Topics in Supply Chain Management 



Chair: Li Chen, Assistant Professor, Duke University, 100 Fuqua Drive, 

Durham, NC, 27708, United States of America, li.chen@duke.edu 

1 - Integrating Inventory Replenishment and Cash Payment 

Decisions in Supply Chains 

Wei Luo, Duke University, 100 Fuqua Drive, Durham, NC, 

United States of America, wei.luo@duke.edu, Kevin Shang 

We provide a modeling framework that integrates financial flows into a two-stage 

supply chain where each location procures inventory based on cash available. We 

consider different payment schemes and derive joint optimal and near-optimal 


88 

inventory and cash policies. Our study demonstrates that an effective cash 

payment policy can mitigate the supply disruption risk and improve the overall 

supply chain efficiency. 

2 - On the Profitability of an Eco-Friendly Supply Chain 

Yang Li, PhD Student, Duke University, 100 Fuqua Drive, 

Durham, NC, 27708, United States of America, 

yang.li2@duke.edu, Fernando Bernstein, Kevin Shang 

We study a two-stage supply chain for eco-friendly problems. The production 

technology for eco-friendly products is more costly, but these products use 

components with less fossil-fuel content than regular products. In particular, ecofriendly 

products are less exposed to the price volatility of petroleum. We 

examine scenarios in which eco-friendly products are more profitable. 

3 - Competitive Quality Choice and Remanufacturing 

Adem Orsdemir, Kenan Flagler Business School, University of 

North Carolina, Chapel Hill, NC, 27599, United States of America, 

adem_orsdemir@kenan-flagler.unc.edu, Eda Kemahlioglu Ziya, 

Ali Parlakturk 

We consider an Original Equipment Manufacturer who faces competition from an 

Independent Remanufacturer. We explicitly characterize how OEM competes 

with IR in equilibrium. IR’s entry threat as well as its entry can decrease 

consumer and social surplus. We show either weak IR or strong IR is desirable for 

reducing the environmental impact. Comparing our results with benchmarks in 

which OEM remanufactures suggests that encouraging IRs to remanufacture in 

lieu of OEMs may not benefit environment. 

4 - Fixing Phantom Stockouts: A POS-Based Shelf Inspection 

Model 

Li Chen, Assistant Professor, Duke University, 100 Fuqua Drive, 

Durham, NC, 27708, United States of America, li.chen@duke.edu 

”Phantom stockout” is a retail stockout phenomenon caused by shelf execution 

failure and/or product shrinkage (e.g., theft and spoilage). In this paper, we 

propose a simple but effective partially-observable Markov decision process 

(POMDP) model to tackle this problem. We show that the optimal shelf 

inspection policy is a threshold policy based on the number of consecutive zerosales 

periods. We further extend the analysis to models with time-varying 

parameters. 

■ SB27 


Supply Chain Models with Multi-sourcing and 

Information Updates 



Chair: Eylem Tekin, Instructional Associate Professor, University of 

Houston, Department of Industrial Engineering, Houston, 

United States of America, etekin@central.uh.edu 

1 - Inventory Replenishment and Demand Allocation Decisions for 

Multi-Sourced Items 

Abhilasha Katariya Prakash, PhD Student, Texas A&M University, 

503 nagle st, apt 102, college station, TX, 77840, United States of 

America, abhilashapk@neo.tamu.edu, Eylem Tekin, Sila Cetinkaya 

We consider a manufacturer using multi-sourced parts from a contractual vendor 

(CV) and a spot market (SM). The CV holds inventory at a vendor managed hub 

and charges a fixed unit price under a quantity commitment contract. Inventory 

procured from the SM with volatile prices is held at an advance purchase hub. 

The overall problem deals with when and how much to order from each source 

and how to allocate demand to the alternative hubs. 

2 - Dynamic Inventory Replenishment Decisions with Bayesian 

Learning of Supply Yield Uncertainty 

Baykal Hafizoglu, Arizona State University, 699 S Mill Avenue, 

Tempe, AZ, 85281, United States of America, baykal@asu.edu, 

Sibel Salman, Esma Gel 

We consider a periodic-review inventory replenishment problem with yield 

uncertainty, where the quantity that the supplier ships in response to an order is 

random. We assume that the parameters that govern the supplier’s yield are not 

known with certainty in advance. We propose a Bayesian updating scheme that 

learns supplier’s yield parameters over time. We compare performance of 

Bayesian learning scheme with other strategies such as safety stock policy, and 

simple heuristics.

3 - Strategic Inventories and Dynamic Coordination with 

Production Cost Learning 

Xiuli He, Assistant Professor, University of North Carolina- 

Charlotte, 9201 University City Blvd, Charlotte, NC, 28223, 

United States of America, xhe8@uncc.edu, Tao Li, Suresh Sethi 

We consider a two-period decentralized supply chain in which the manufacturer’s 

production cost declines with cumulative production due to stochastic cost 

learning. We investigate the impact of learning curve effect on the value of 

strategic inventory to the channel members. We construct dynamic revenue 

sharing contracts that coordinate the channel. 

4 - Supply-side Rationale for Bundling 

Qingning Cao, University of Texas-Dallas, School of Management, 

800 West Campbell Road, Dallas, TX, 75080, United States of 

America, qingning.cao@utdallas.edu, Jun Zhang, Xianjun Geng, 

Kathryn Stecke 

We examine a firm’s capacity decisions and restrictions on a popular product 

when the market is uncertain and the firm has an option to bundle two products. 

Deriving the firm’s optimal bundling strategy, this talk sheds light on the interplay 

between a firm’s capacity (supply-side) and bundling (demand-side) decisions. 

■ SB28 


Customer Behavior and Operations Models 



Chair: Fuqiang Zhang, Olin Business School, Washington University, St. 

Louis, United States of America, Fzhang22@wustl.edu 

1 - Bounded Rationality in Service Systems 

Gad Allon, Northwestern University, 2001 Sheridan Rd, Evanston, 

IL, United States of America, g-allon@kellogg.northwestern.edu, 


The traditional economics and queueing literature typically assume that 

customers are fully rational. In contrast, in this paper, we study canonical service 

models with boundedly rational customers. We capture bounded rationality using 

a framework in which better decisions are made more often, while the best 

decision needs not always be made. 

2 - Consumer-driven Competition Sets 



Jun Li 

We use clickstream data from Expedia to construct and analyze hotel competition 

sets. We find that these competition sets are drastically different from the 

traditionally used attribute-based competition sets. 

3 - Competing for Socially Conscious Consumers: The Role of 

Supply Chain Structure 

Robert Swinney, Associate Professor, Graduate School of Business, 

Stanford University, 655 Knight Way, Stanford, CA, 94305, United 

States of America, swinney@stanford.edu, Ruixue Guo, Hau L. Lee 

We analyze a duopoly in which two competing firms choose whether to adopt 

socially responsible manufacturing practices. Doing so attracts demand from 

competing firms-those consumers who have a preference for socially responsible 

production-but also increases costs. We consider the impact of supply chain 

structure on the firms’ decisions, and discuss how the supplier and consumer 

population characteristics combine to influence the equilibrium degree of social 

responsibility in the industry. 

4 - Responsive Pricing of Fashion Products: The Effects of Demand 

Learning and Strategic Consumer 

Mike Wei, Olin Business School, Washington University, 

St. Louis, MO, United States of America, weim@wustl.edu, 

Fuqiang Zhang, Yossi Aviv 

We provide a stylized Bayesian demand learning model to investigate the value of 

responsive pricing under strategic consumer behavior. In particular, we find that 

increasing market uncertainty affects the value of responsive pricing adversely 

due to the spread effect and the manipulation effect. Furthermore, it can be 

shown that the price commitment can boost the seller’s revenue substantially, and 

the information can be much more valuable under strategic consumers than 

under myopic consumers. 


89 

■ SB29 


Workforce Engineering 



Chair: Ruwen Qin, Assistant Professor, Missouri University of Science 

and Technology, 600 W. 14th St., 218 Engineering Management 

Building, Rolla, MO, 65409, United States of America, qinr@mst.edu 

1 - Cross-training of Team Members to Deal with Demand Mix 

Variation and Absenteeism 

Jordi Olivella, Assistant Professor, Universitat Politècnica de 

Catalunya, Esteve Terradas, 7, Castelldefels, 08860, Spain, 

jorge.olivella@upc.edu, David Nembhard 

The objectives of this work are: (1) To develop a method to determine the crosstraining 

goals of a work team in order to deal with certain levels of variation in 

the demand mix and absenteeism; and (2) Evaluating alternative policies of crosstraining 

of team members. The demand mix variation is defined by establishing a 

maximum time to devote to each product. To solve numerically the mathematical 

problem generated a selection constraint based procedure is developed. 

2 - Strategic Workforce Planning Models for 

Semiconductor Manufacturing 

Shrikant Jarugumilli, Sr. Operations Research Specialist, BNSF 

Railway, 2400 Western Center Blvd., Ft Worth, TX, 76131, 

United States of America, Shrikant.Jarugumilli@bnsf.com, 

Scott Grasman, Naiping Keng 

In this work, we propose an integrated framework that links the strategic 

workforce planning model to the operational workforce allocation models. The 

strategic workforce planning model enables the planners to make decisions to 

retain and hire workers, along with the design of worker cross-training programs 

based on the long-term requirements of the organization. 

3 - Development of a Human-based Metric Selection Protocol 

Gratchen Macht, Pennsylvania State University, 244 Leonhard 

Building, University Park, PA, 16802, United States of America, 

gam201@psu.edu, David A. Nembhard 

Relevant solutions to current complex problems require the incorporation of 

highly functional collaborative skills within a team-like framework. A humanbased 

selection protocol was developed through a mixture of technical skills and 

psychological profiles to address interdisciplinary teams. The results impact team 

formulation to improve the overall team performance. 

4 - Optimal Flow-line Conditions with Worker Variability 

Frank Bentefouet, Research Assistant, Pennsylvania State 

University, 1231 South Allen Street, State College, PA, 16801, 

United States of America, fub3@psu.edu, David Nembhard 

In this paper, we propose a framework to study the workforce scheduling 

problem in an assembly line also called flow line. We investigate the impact of 

within-worker and between workers variability and discuss the selection of a 

scheduling policy between fixed and work sharing system. We characterize the 

optimal switching time between workers in a system with two tasks and establish 

the maximum number of changes between assignments when considering a 

general floor shop. 

■ SB30 

SB30 


Supply Chain Risk Management 



Chair: Zhibin (Ben) Yang, Assistant Professor, University of Oregon, 

1208 University of Oregon, Eugene, 97403, United States of America, 

zyang@uoregon.edu 

Co-Chair: Volodymyr Babich, Assistant Professor, McDonough School 

of Business, Georgetown University,37th and O Streets, NW, 546 Rafik 

B. Hariri Building, Washington, DC, 20057, United States of America, 

vob2@georgetown.edu 

1 - Sourcing Strategies under Supply Yield Uncertainty 

Guang Xiao, Washington University in St. Louis, One Brookings 

Drive, Campus Box 1133, St. Louis, MO, 63130, United States of 

America, xiaoguang@wustl.edu, Lingxiu Dong, Nan Yang 

We consider a one-buyer-multi-supplier system in the presence of supply yield 

uncertainty. We explore the impact of various supply contractual arrangements 

and different yield uncertainty profiles on the equilibrium supply chain 

performance.

SB31 

2 - The Impact of Cost and Quality on Illegal Goods in the 

Supply Chain 

Maqbool Dada, Professor, Johns Hopkins Carey Business School, 

100 International Drive, Baltimore, MD, 21202, United States of 

America, dada@jhu.edu, Arkadi Seidscher, Stefan Minner 

A manufacturer of a branded product may deter competition from parallel 

imports, smuggled goods or counterfeits by investing in higher quality and 

lowering price. A four stage model of competition is developed to examine the 

impact of such competition. The analysis also has implications for legal means of 

competition; for example, from store brands in supermarkets or generic drugs in 

pharmacies. 

3 - How To (and How Not To) Manage Supplier’s Process 

Improvement: Delegation, Incentives, or Audit 

Mehmet Gumus, Assistant Professor, McGill University, 

1001 Sherbrooke West, Montreal, QC, H3A1G5, Canada, 

mehmet.gumus@mcgill.ca, Mohammad Nikoofal 

In this paper, we explore the effectiveness of process improvement as a supply 

disruption mitigation strategy in the presence of moral hazard and adverse 

selection. In order to address these issues, we develop a dyadic supply chain 

model where both the degree of supply disruption risk and supplier’s mitigation 

effort are unobservable from retailer. Our analysis shows that eliminating moral 

hazard via monitoring has a non-monotonic effect on the screening cost. 

4 - Intermediated Sourcing under Supply Disruption Risk 

Zhibin (Ben) Yang, Assistant Professor, University of Oregon, 1208 

University of Oregon, Eugene, 97403, United States of America, 

zyang@uoregon.edu, Volodymyr Babich 

We analyze a model where multiple buyers decide whether to procure goods from 

unreliable suppliers directly or using an intermediary firm. We show that lowrevenue 

buyers may enjoy spill-over benefits of procurement cost reduction 

when high-revenue buyers choose to use the intermediary. Intermediation may 

also reduce system-wide product waste, by coordinating diversification efforts 

across buyers. These benefits have to be weighed against intermediation costs, 

such as double-marginalization. 

■ SB31 


Supply Chain Risk and Resilience 



Chair: Kash Barker, Assistant Professor, School of Industrial and 

Systems Engineering, University of Oklahoma, 202 W. Boyd St., Rm 

124, Norman, OK, 73019, United States of America, 

kashbarker@ou.edu 

1 - Disruption Management during Supply Chain Disruptions 

Cameron MacKenzie, University of Oklahoma, 202 W. Boyd St., 

Norman, OK, United States of America, cmackenzie@ou.edu, 

Kash Barker 

Many firms suffered from supply disruptions due to the recent Japanese 

earthquake and tsunami. We analyze these disruptions by developing a model in 

which several suppliers’ production facilities are rendered inoperable. Each 

supplier must decide whether to move production to an alternate facility or wait 

for its facility to reopen. If suppliers do not produce at alternate facilities, firms 

will need to decide how to mitigate the impacts of a supply shortage. 

2 - Resilient Supply-chain Network Design with Multi-sourcing 

Haifei Yu, PhD, Northeastern University, No.11, Lane 3, 

WenHua Road, Heping District,Liaoning, Shenyang, 110004, 

China, hfyu@mail.neu.edu.cn, Shoufeng Ji 

With the uncertainty risk of supply disruptions, it is critical for the supply-chain 

system to be resilient. We study the problem of designing a resilient supply chain 

network with multi-sourcing, and propose a chance-constrained programming 

model, and use a mixed integer linear programming to solve it. We evaluate the 

benefit of multi-sourcing, balance operation costs, resilience, and disruption risk 

by the computational experiments. 

3 - Prediction and Prevention of Disruptions in Supply 

Chain Networks 

Xin Chen, Assistant Professor, Southern Illinois University 

Edwardsville, 3079 Engineering Building, Edwardsville, IL, 

62026-1805, United States of America, xchen@siue.edu 

The objective of this research is to apply the failure mode, effects, and criticality 

analysis (FMECA) and graph theory to predict and prevent disruptions in supply 

chain networks. A network may be disrupted due to node failures. A node 

represents a supplier, customer, or transportation route. This research applies 

graph theory to model supply chain networks and uses the FMECA to identify 

critical nodes. A case study of a construction supply chain is analyzed to validate 

the methodology. 


90 

4 - Supply Chain Resilience 

Ivan Hernandez, Stevens Institute of Technology, 167 8th Street, 

Apt 4, Hoboken, NJ, 07030, United States of America, 

ihernand@stevens.edu, Jose Ramirez-Marquez, David Nowicki 

We present a method that allows decision makers to ensure a Supply Chain (SC) 

returns to a high level of performance following a contingency. We introduce the 

concept of SC resilience as a function of the post-event investment resources 

needed for the SC to “bounce back”. The concepts can be used as a guide for 

companies to use their long-term SC plan as a starting point, then begin to ask 

themselves what they would do if various contingencies occur. 

■ SB32 

32- North 223- CC 

Assembly Line Design and Balancing I 


Chair: Slim Daoud, Université de Technologie de Troyes, 

32 Rue du Palais de Justice, Troyes, 10000, France, slim.daoud@utt.fr 

1 - Spreadsheet Based Optimization for Cell Formation Problem 

Gurkan Ozturk, Assistant Professor, Anadolu University, Faculty of 

Engineering, Department of Industrial Eng., Eskisehir, 26555, 

Turkey, gurkan.o@anadolu.edu.tr, Mumin Sonmez 

In this study, a new hybrid algorithm based on competitive neural network and 

particle swarm optimization approaches is proposed to solve cell formation 

problem with alternative part routings. This algorithm is implemented in two 

spreadsheet platform: MS Excel and Google Spreadsheet. Although MS Excel 

faster than the Google Spreadsheet, the property of internet access of google 

spreadsheet presents different advantages. 

2 - Neuro-fuzzy-genetic Algorithm for the Robotic Assembly 

Lines Balancing 

Slim Daoud, Université de Technologie de Troyes, 32 Rue du Palais 

de Justice, Troyes, 10000, France, slim.daoud@utt.fr, Yalaoui 

Farouk, Amodeo Lionel, Chehade Hicham, Duperray Philippe 

We are interested in an industrial application of a robotic assembly line problem. 

It consists of seizing the products on a moving conveyor and placing them on 

different deposit points. The goal is to optimize the efficiency of the line. As in our 

industrial application we are bounded by the execution time, we suggest a genetic 

algorithm combined with a neuro-fuzzy network to solve our problem. The 

experimental results show the advantages and the efficiency of the developed 

method. 

3 - Balancing Mixed Model Assembly Line with Capacity Buffer 

Jinlin Li, Xi’an Jiaotong University, No.28, Xianning West Road, 

Xi’an, China, ljl1019@163.com, Jie Gao, Linyan Sun 

This paper investigate the design of a mixed model assembly line with capacity 

buffer to satisfy unstable demands. The objective is to minimize the expected 

labor cost, including cost for normal operation and overtime work. A mathematic 

model is built and a method is proposed to estimate the cost lower bound. A 

branch, bound and remember algorithm is developed and the numerical 

experiment shows that the algorithm is effective and efficient. 

4 - Recognition of Freeform Surface Sheet Metal Features 

Saied Darwish, Professor, King Saud University, Riyadah, 

800/11421, Saudi Arabia, darwish@ksu.edu.sa, 

Hammouda Mousal 

This paper describes the design and implementation of a system for automated 

feature recognition of freeform surface CAD models of stamped and notched 

sheet metal parts, represented in STEP AP203 format through CATIA V5.

■ SB33 


Modeling Sustainability in Supply Chains 

Sponsor: Manufacturing & Service Oper Mgmt/ 

Sustainable Operations 


Chair: Krishnan Anand, Associate Professor, University of Utah, David 

Eccles School of Business, Salt Lake City, UT, 84108, United States of 

America, k.anand@utah.edu 

1 - Finding and Implementing Energy Efficiency Projects in 

Industrial Facilities 

Sam Aflaki, HEC, France, aflaki@hec.fr, Paul R. Kleindorfer 

There are many profitable Energy Efficiency projects in many industrial 

enterprises that are not implemented. One of the main barriers often cited as the 

main culprit is lack of an internal management framework to find, value and 

execute these projects. Using a conceptual approach, we rely on the existing 

sustainable operations tools to provide such a framework. A case example of a 

large manufacturing site illustrates the details of this framework. 

2 - Closed-loop Supply Chain Design under Uncertainty 

Vedat Verter, Professor, McGill University, 1001 Sherbrooke St. 

W., Montreal, QC, H3A 1G5, Canada, vedat.verter@mcgill.ca, 

Wenyi Chen, Beste Kucukyazici 

We present a mathematical formulation for designing a closed-loop supply chain 

under uncertainties with regards to demand, return rate and recovery rate. The 

model is applied in studying a realistic problem instance that involve the 

production and recycling of washers and tumble dryers in Germany. We will 

presents the managerial insights derived from a large set of computational 

experiments. 

3 - Consumer Markets for Remanufactured Products: 

Empirical Evidence and Economic Models 

James Abbey, Pennsylvania State University, 463A Business 


jda188@psu.edu, Dan Guide, Joseph Blackburn 

When considering remanufactured products, do consumers follow the typical 

principle of ‘cheaper is better’ or do other rules apply? Additionally, do consumer 

markets for remanufactured products demonstrate homogeneous price-taking 

behavior? This research addresses these and other questions in relation to 

consumer markets for remanufactured products. Insights derive from 

triangulation of empirical investigations and tightly coupled economic modeling. 

4 - Investing in Pollution Abatement Innovations under Emission 

Taxes and Cap-and-trade 

Francois Giraud-Carrier, University of Utah, David Eccles School of 

Business, 1645 E Campus Center Dr, Salt Lake City, UT, 84112, 

United States of America, f.giraud-carrier@business.utah.edu, 

Krishnan Anand 

We compare the incentives of polluting firms to invest in an innovative pollution 

abatement technology under two popular pollution governance structures— 

emission taxes and cap-and-trade. In our model, firms first decide whether to 

adopt the (costly) innovative technology, and then make production and 

abatement decisions. We compare the impact of each governance structure on 

firms’ pollution abatement strategies, output and profits, consumer surplus and 

welfare. 

■ SB34 

34- North 224 B - CC 

Product Achitecture and Design 


Chair: John Jung-Woon Yoo, Assistant Professor, Bradley University, 

1501 W. Bradley Ave., Peoria, IL, 61625, United States of America, 

jyoo@bradley.edu 

1 - Small Modular Infrastructure 

Caner Gocmen, Columbia Business School, 3022 Broadway, 

Uris Hall, 4V, New York, NY, 10027, United States of America, 

fgocmen13@gsb.columbia.edu, Garrett Van Ryzin, Eric Dahlgren, 

Klaus Lackner 

In many basic infrastructure industries, such as electric power generation, we 

have witnessed a trend of ever increasing unit size. We argue for a reversal of this 

trend - a radical shift to a world in which custom built technology of massive unit 

scale is replaced by small, modular and mass-produced units. To make this case, 

we develop a framework to evaluate the economics of unit scale and argue that 

many industries are already at a tipping point toward radically smaller unit scale. 


91 

2 - Strategic Linkages of Product Architecture Decisions: 

Organizational and Performance Implications 

Ujjal Mukherjee, Research Assistant, University of Minnesota, 

1261 Gibbs Ave., Apt G, Saint Paul, MN, 55108-1159, 

United States of America, mukh0067@umn.edu 

What are the strategic linkages to product architecture decisions and what are the 

organizational and performance implications of such linkages? We adopt an 

‘alignment’ perspective for the strategic positions, and the product architecture 

decisions. This frame-work is used as a basis to develop a proposed product 

architecture categorization and organizational architecture linkages. The 

framework includes the linkages ‘strategic positioning – product architectureorganization-performance’. 

3 - Social Network Effects on Product Line Design 

Dilek Gunnec, University of Maryland, 3330 Van Munching Hall, 


dgunnec@rhsmith.umd.edu, Subramanian Raghavan 

We model peer influence effects among the social network of users of products 

from the same product line. Influence among customers alters their utilities from 

the products, and therefore do not allow for an a priori preference ordering for an 

individual among different product profiles at the outset. We analyze product 

diffusion processes over the network and identify the set of products to maximize 

the total market share when each customer selects the product with the highest 

utility. 

4 - Performance Evaluation of Interface-based Solution Procedures 

for Modular Product Design 

John Jung-Woon Yoo, Assistant Professor, Bradley University, 

1501 W. Bradley Ave., Peoria, IL, 61625, United States of America, 

jyoo@bradley.edu 

In our earlier work we have proposed a collaboration system for modular product 

design and proposed three interface-based solution approaches, a mathematical 

approach based on Integer Programming, an algorithmic approach based on 

branch-and-bound method and, finally, a heuristic method. The first two 

approaches are for the optimal solution, while the last one is for a heuristic 

solution. In this talk, the experimental results are discussed and the performance 

of the three approaches is compared. 

■ SB35 

SB35 


Dynamic Pricing 



Chair: Ming Chen, Assistant Professor, California State University, 

School of Business Administration, Long Beach, CA, 90840, 

United States of America, Ming.Chen@csulb.edu 

1 - Robust Customized Pricing 

Elcin Cetinkaya, Doctoral Student, Lehigh University, 200 W. 


elcin.cetinkaya@lehigh.edu, Aurelie Thiele 

We study robust revenue management problems when companies request bids 

for services but their price-response function is not known precisely. The 

company bidding for these contracts only knows the previous bids it submitted to 

those businesses and whether they were accepted or not. We show how to derive 

tractable mathematical models for this problem and provide insights into the 

optimal solution. We also document the performance of the approach in 

numerical experiments. 

2 - Joint Pricing and Inventory Decisions under Poisson 

Decomposition with Insights on Optimal Assortment 

Bacel Maddah, Associate Professor, American University of Beirut, 

Bliss Street, Beirut, Lebanon, bm05@aub.edu.lb, Ebru Bish, 

Hussein Tarhini 

We consider substitutable retail products in a market with logit choice, demand 

generated by Poisson decomposition, and a newsvendor-type supply setting. We 

focus on the joint pricing and inventory problem for a given assortment, under a 

highly accurate Taylor-like approximation. We show that the expected profit is 

unimodal in the price and analyze the effect of inventory on pricing. Then, with 

no approximation, we analyze the joint assortment and inventory problem, and 

derive novel insights.

SB36 

3 - Survey of Dynamic Pricing Research on Selling a Given 

Inventory over a Finite Horizon 

Ming Chen, Assistant Professor, California State University, 

School of Business Administration, Long Beach, CA, 90840, 

United States of America, Ming.Chen@csulb.edu 

We present a review of existing research on dynamic pricing problems involve 

selling a given amount of inventory over a finite time horizon without inventory 

replenishment. We classify these problems into several classes based on a number 

of problem attributes and summarize main findings and managerial insights. 

■ SB36 


New Directions in Revenue Management 



Chair: Omar Besbes, Columbia Business School, 3022 Broadway, 

New York, NY, 10027, United States of America, ob2105@columbia.edu 

Co-Chair: Gabriel Weintraub, Columbia Business School, 3022 

Broadway, New York, NY, 10027, United States of America, 

gyw2105@columbia.edu 

1 - Intertemporal Price Discrimination: Structure and Computation 

of Optimal Policies 

Ilan Lobel, New York University, Stern Business, 44 W 4th St., 

New York, NY, United States of America, ilobel@stern.nyu.edu, 

Omar Besbes 

We consider the problem of a firm selling goods over time to customers with 

heterogeneous patience levels. We let customer valuations be correlated with 

their willingness-to-wait and look for a dynamic pricing policy that maximizes the 

long-term revenue of the firm. We prove several structural properties of the 

optimal pricing policy and develop a dynamic programming algorithm that 

computes the optimal policy in polynomial time. 

2 - ADP For (Very) High Dimensional Resource Allocation 

Vivek Farias, Massachusetts Institute of Technology, 77 


vivekf@mit.edu, Dragos Florin Ciocan 

Due to the heterogeneous nature of the resource being allocated (impressions), 

online Ad Display allocation presents a challenge to traditional revenue 

management approaches. We present a novel ADP inspired clustering technique 

that allows for real time resource allocation in such problems. Since this scheme is 

calibrated through solving a natural, but massive linear program, we also deploy a 

scheme that can solve effectively arbitrarily large LPs in a decentralized manner. 

3 - Optimal Timing of New Product Introduction using Preliminary 

Market Testing 

Rene Caldentey, New York University, 44 West Fourth Street, 

New York, NY, United States of America, rcaldent@stern.nyu.edu, 

Victor Araman 

Introducing new products into the marketplace is a risky bet that manufacturers 

must continuously take. As a result, it is not uncommon to witness major 

manufacturers discontinuing a product shortly after its introduction. In this talk, 

we consider a seller who first tests the market and learns more about the demand 

before deciding whether or not to launch a product (or set of products). We also 

discuss current e-commerce practices that have increase companies’ ability to do 

such market testing. 

4 - Auctions Design for Online Display Advertising Exchanges 

Santiago Balseiro, Columbia Business School, 

3022 Broadway, New York, NY, 10027, United States of America, 

srb2155@columbia.edu, Omar Besbes, Gabriel Weintraub 

Ad Exchanges are emerging markets where advertisers may purchase display ad 

placements directly from publishers via a simple auction mechanism. Using the 

novel notion of a Fluid Mean Field Equilibrium we study key auction design 

decisions that publishers face in these markets. In particular, we provide sharp 

prescriptions regarding the reserve price, the allocation of impressions to the 

exchange versus an alternative channel, and the disclosure of viewers’ 

information. 


92 

■ SB37 


Dynamic Pricing in Supply Chains 



Chair: Jian Li, Assistant Professor, Northeastern Illinois University, 

5500 N. St. Louis Ave, Chicago, IL, 60625, United States of America, 

j-li3@neiu.edu 

1 - The Inventory Billboard Effect on Information Sharing in 

Competing Supply Chains 

Zhengping Wu, Assistant Professor, Singapore Management 

University, 50 Stamford Road, Singapore, Singapore, 

zpwu@smu.edu.sg, Rong Li, Xin Wei 

The inventory billboard effect in operations management indicates that the 

increasing shelf-space allocated to a product has a positive effect on the product 

demand. This paper studies the billboard effect on the vertical information sharing 

strategy of competing supply chains in an environment with production 

diseconomies. We analyze how equilibrium information sharing strategy, 

wholesale price and retail quantity are affected by the billboard effect coefficient. 

2 - Price Trends in Dynamic Pricing Models 

Xiaowei Xu, Associate Professor, Rutgers Business School, 

1 Washington Park, Room 958, Newark, NJ, 07102-3122, 

United States of America, xiaoweix@business.rutgers.edu 

We consider a seller that has a limited amount of asset to liquidate in a finite time 

horizon. Demand signals are updated at preset time spots and the seller can adjust 

the product price accordingly. We offer the managerial insights of how 

probabilistic price trends are affected by time-varying and price-sensitive demand 

forms via exploring the recursive structure of the optimal pricing policy and the 

implied martingale decomposition. 

3 - Dynamic Pricing with Random Replenishment Capacity 

Jian Li, Assistant Professor, Northeastern Illinois University, 

5500 N. St. Louis Ave., Chicago, IL, 60625, 

United States of America, j-li3@neiu.edu, Xiaowei Xu 

In this research, we study the decision problem facing a retailer who sells a short 

seasonal product inventory in a stylized model. The retailer has an uncertain 

second chance to refill its inventory. We analytically derive the structure of the 

optimal dynamic pricing policy and procurement decisions. We also explore the 

properties of the optimal dynamic pricing and the expected profit function. 

■ SB38 


Service Innovations in the Global Marketplace 



Chair: Christian Wernz, Assistant Professor, Virginia Tech, 205 Durham 

Hall (0118), Blacksburg, VA, 24061, United States of America, 

cwernz@vt.edu 

1 - Service Engineering – Trends, Perspectives and the 

Contribution to Service Science 

Walter Ganz, Director, Fraunhofer IAO, Nobelstr. 12, Stuttgart, 

70569, Germany, Walter.Ganz@iao.fraunhofer.de, Thomas Meiren 

A large amount of work has been done in relation to the definitions, typologies of 

services. It is noticeable that the development of fundamental models for services 

has tended to be neglected. Discussing the options of model creation as well as the 

simulation of services was a central topic of our expert study “MARS”. It becomes 

obvious that the creation of models as well as simulation has gained an 

importance within the last years and it could make a relevant contribution for 

Service Science 

2 - Firm Value of Product Innovation in Service Ecosystems: 

A Study of Smartphones 

Rahul Basole, Associate Director and Senior Research Scientist, 

Tennenbaum Institute, Georgia Institute of Technology, 75 Fifth 

Street NW, Suite 600, Atlanta, GA, 30308, United States of 

America, rahul.basole@ti.gatech.edu, Hyunwoo Park 

We examine the asymmetric impact of product innovation on the market value of 

manufacturers (e.g. device manufacturers) and service providers (e.g. mobile 

network operators) in the mobile service ecosystem. Our analysis is based on a 

comprehensive dataset comprising over 1,500 smartphones and considers the 

influence of product characteristics, the frequency and timing of product 

innovation, and the network structure of the service ecosystem. Theoretical and 

managerial implications are discussed.

3 - The Design of Experiential Services 

Uday Karmarkar, Professor, UCLA, 110 Westwood Plaza, 

Los Angeles, CA, 90095, United States of America, 

uday.karmarkar@anderson.ucla.edu, Aparupa Das Gupta, 

Guillaume Roels 

We address sequencing and resource allocation of stages in a multi-stage service 

encounter where customer perceptions of experience are important, as in the 

hospitality industry. We employ evidence from behavioral studies about how 

people evaluate and retain memories of experiences. We model these phenomena 

parsimoniously, and use the models to design service processes so as to maximize 

ex-post customer satisfaction. 

4 - Service Convergence: A New Phenomenon? 

Christian Wernz, Assistant Professor, Virginia Tech, 205 Durham 

Hall (0118), Blacksburg, VA, 24061, United States of America, 

cwernz@vt.edu, Pooja Thakur, Kongkiti Phusavat 

We investigate the phenomenon of service convergence. Service convergence is 

defined as the tendency of previously separate service offerings to evolve towards 

an integrated product. We analyzed the business strategy of a Thai hospital and 

identified service convergence as a result of the firm’s quality improvement and 

innovation initiative. We discuss how the integration of medical services with 

hospitality services contributed to the hospital’s success in the medical tourism 

industry. 

■ SB39 


Service Quality Evaluation and Customer Retention 



Chair: Gregory Heim, Associate Professor, Mays Business School at 

Texas A&M University, 320 Wehner Building, 4217 TAMU, 

College Station, TX, 77843-4217, United States of America, 


1 - Development of a Service Quality Evaluation Scheme in a 

Service Laboratory Environment 

Ryeok-Hwan Kwon, Pohang University of Science and Technology, 

Engineering Buliding #4-316, Hyoja-Dong, Nam-Gu, Pohang, 

Kyungbuk, Pohang, Korea, Republic of, rh_kwon@postech.ac.kr, 

Hyun-Jin Kim, Kwang-Jae Kim 

A virtual reality (VR) laboratory for service testing has been constructed in Korea. 

One issue associated with its operation is how to evaluate the quality of a service 

tested in a laboratory environment. We present a service quality evaluation 

scheme for a service laboratory environment, consisting of two parts: a repository 

of service evaluation criteria and a guideline to select pertinent criteria for a 

particular case study. The result of a case study will also be presented. 

2 - Factors Affecting Students’ Perceptions of Quality in 

Higher Education 

Martha Ramírez-Valdivia, Assistant Professor, Universidad de La 

Frontera, Avenida Francisco Salazar 01145, Temuco 4780000, 

Chile, marthar@ufro.cl, Favia Catrileo, Horacio Miranda 

This study identified which background dimensions influences student 

perceptions of quality in higher education. A questionnaire was applied to 27,182 

chilean university students; 687 responses were received, 223 valid ones. A 

factorial analysis verified accuracy and validity. The main findings indicate that 

the lecturer¥s competencies and the lecturer¥s attitude and behavior are the main 

determinants in the provision of quality education. The perception of the quality 

service is medium to low. 

3 - Dynamic Customer Acquisition and Retention Management 

Greg King, PhD Candidate, University of Michigan, Department of 

Industrial and Operations, Ann Arbor, MI, United States of 

America, gjking@umich.edu, Xiuli Chao, Izak Duenyas 

In consulting, finance, and other service industries, customers represent a 

revenue stream, and must be acquired and retained over time. In this paper, we 

study the resource allocation problem of a profit maximizing service firm that 

dynamically allocates its resources towards acquiring new clients and retaining 

unsatisfied existing ones. We formulate the problem as a dynamic program and 

characterize the structure of the optimal acquisition and retention strategy. 


93 

■ SB40 


Stochastic Hydro-Thermal Scheduling I 



Chair: Steffen Rebennack, Assistant Professor of Operations Research, 

Colorado School of Mines, 1500 Illinois St., Golden, CO, 80401, 

United States of America, srebenna@mines.edu 

Co-Chair: Timo Lohmann, PhD Student, Colorado School of Mines, 

816 15th Street, Golden, CO, 80401, United States of America, 

tlohmann@mines.edu 

1 - A Dynamic Piecewise Linear SDDP Algorithm Applied to 

Stochastic Nonlinear Hydrothermal Scheduling 

Andre Luiz Diniz, Researcher, CEPEL-Brazilian Electric Energy 

Research Center, Av. Horacio Macedo 354, Cidade Universitária, 

Ilha do Fundão, Rio de Janeiro, RJ, 22220040, Brazil, 

diniz@cepel.br, Michel Ennes, Renato Cabral 

We propose dynamic piecewise linear models to represent nonlinear constraints 

and objective function in stochastic multi-stage convex optimization problems 

solved by stochastic dual dynamic programming (SDDP). The proposed approach 

provides valid cuts and convergence bounds and yields more accurate results as 

compared to standard piecewise linear models, in lower CPU times. We present 

results of its application for the generation planning of the real large-scale 

Brazilian hydrothermal system. 

2 - Representation of Nonconvexities in Stochastic Dual Dynamic 

Programming (SDDP) 

Fernanda Thomé, Research Engineer, PSR, Praia de Botafogo 

228/1701-A, Rio de Janeiro, RJ, 22250145, Brazil, 

fernanda@psr-inc.com, Mario Veiga Pereira, Sérgio Granville, 

Marcia Fampa 

This work describes an extension of the SDDP algorithm to represent 

nonconvexities such as variation of hydro production coefficient with storage and 

thermal unit commitment. The proposed methodology uses a Lagrangean 

relaxation on the water balance equation to generate convex Benders cuts during 

the backward recursion phase, with a special procedure to find the optimal 

multiplier. The approach will be illustrated with a realistic hydrothermal 

scheduling problem. 

3 - Optimizing Trading Decisions for Hydro Storage Systems using 

Approximate Dual Dynamic Programming 

Nils Löhndorf, Vienna University of Economics and Business, 

Nordbergstr 15/3/A, Vienna, 1090, Austria, 

nils.loehndorf@wu.ac.at, David Wozabal, Stefan Minner 

Optimal operation of hydro storage systems in electricity markets is a complex 

stochastic optimization problem. We formulate this problem as a Markov decision 

process where a stochastic mixed-integer program is solved in each time period. 

To find a near-optimal policy, we propose a solution approach that computes a 

polyhedral approximation of the value function. We describe how we applied the 

approach to optimize the operation of a real-world hydro storage system in the 

Austrian Alps. 

■ SB41 


Regulation in Energy 


SB41 

Chair: Eric O’Rear, Graduate Student, Purdue University, 

403 W. State Steet, West Lafayette, IN, 47907, United States of America, 

eorear@purdue.edu 

1 - A Decision Support Tool for Evaluating Regional Greenhouse 

Gas Emission Reduction Strategies 

Rick Olson, University of San Diego, 5998 Alcala Park, San Diego, 

CA, 92110, United States of America, r_olson@sandiego.edu, 

Scott Anders, Andrew Narwold, Nilmini Silva-Send 

Cities and counties planning policies to meet GHG reduction goals often focus on 

the direct impact of specific policy changes, e.g. increasing the use of mass transit. 

The interactions between factors, such as mass transit and telecommuting, are 

more difficult to understand. The paper presents a DSS that can be used by 

regional planners to develop strategies for meeting GHG reduction targets while 

considering the relationships between factors and the costs of the policies.

SB42 

2 - Regulated Design & Pricing in the New Vehicle Market 

William Morrow, Assistant Professor, Iowa State University, 

2104 Black Engineering Building, Ames, IA, 50011-2161, 

United States of America, wrmorrow@iastate.edu, Josh Mineroff, 

Kate S. Whitefoot 

Regulations influence vehicle design and pricing. However, few models of policy 

impacts include strategic design decisions in response to regulation. This talk 

describes simulation of equilibrium design and pricing decisions under regulation. 

Examples are given for the new vehicle market and associated fuel economy 

standards. Our results demonstrate that simulations that neglect strategic 

behavior may overestimate the effectiveness of current policy approaches. 

3 - The Rebound Effect and its Implications on Obama’s Plan for a 

Clean Energy Future 

Eric O’Rear, Graduate Student, Purdue University, 403 W. State 

Steet, West Lafayette, IN, 47907, United States of America, 

eorear@purdue.edu, Wallace Tyner, Kemal Sarica 

Our paper will compare the impacts of Obama’s modifications to CAF… 

standards, with a more efficient carbon tax generating a similar level of CO2 

reductions. Further, we intend to capture the impacts of the “rebound effect” 

phenomenon associated with a mandate for a higher fuel economy on the U.S. 

energy system. We utilize the U.S. EPA MARKAL model - a bottom-up, linear 

programming (LP), partial equilibrium model – to replicate and compare the 

policies aligned with our interests. 

4 - Optimal Number of Certificates to be Issued in Emission 

Markets 

Cristian Pelizzari, University of Brescia, Contrada Santa Chiara, 50, 

Brescia, 25122, Italy, pelizcri@eco.unibs.it, Falbo Paolo 

The price of allowances in the EU ETS is linked to the value of the expected 

penalty for the CO2 produced in excess by diverse economic players. Too severe 

environmental policies cause a rise in the price of allowances. The compliance of 

such objectives can increase production costs causing inflation and loss of 

competitiveness. A stochastic programming model is advanced where the 

regulatory authority trades-off economic growth with environmental targets, 

influencing the choices of the players by means of the number of allowances to be 

issued. 

■ SB42 


Tactical Issues in Freight Transportation 


Chair: Mihalis Golias, University of Memphis, 104 Engineering Science 

Bldg, 3815 Centr, Memphis, 38152, United States of America, 

mgkolias@memphis.edu 

1 - Model Slection for Shipper’s Mode Coice 

Subhro Mitra, Assistant Professor, University of North Texas, 

University Hills Blvd., Dallas, TX, 75241, United States of America, 

subhro.mitra@unt.edu, Ali S. Shaqlaih 

This paper presents discrete choice models developed to determine the probability 

of mode choice between railroad and truck, based on revealed preference data. 

Utility functions are formulated using the attributes of the modes, shippers and 

interaction between the two. The choice probability is initially estimated 

assuming the random component to have a logit distribution. The model is 

further developed assuming the stochastic component to have heteroscedastic and 

probit distribution. 

2 - Efficient Heuristic for Auto-carrier Loading Problem 

Saravanan Venkatachalam, Research Associate, Texas A&M 

University, 304A Fermier Hall, 3367 TAMU, College Station, TX, 

77843, United States of America, saravanan@entc.tamu.edu, 

Arunchalam Narayanan, Bimal Nepal 

This belongs to a special group of logistics problem, as it needs solution for two 

NP-hard subproblems, namely routing and loading. There are several efficient 

methods for the routing part, but very few for the loading aspect. Existing 

methods in literature concentrate on the cost of loading and reloading and 

approximate the geometric, weight, and mechanical constraints of the vehicles. 

We demonstrate the application of a new heuristic which incorporates all these 

loading constraints. 

3 - Load Mixing with Freight Due Dates 

Sarah Root, Assistant Professor, University of Arkansas, 

4207 Bell Engineering Center, Fayetteville, AR, 72701, 

United States of America, seroot@uark.edu, Crystal Wilson 

The underutilization in trucking leads to nearly 5 billion gallons of wasted fuel 

annually. One way to recapture part of this waste is to use collaborative logistics. 

This research focuses on one specific aspect of collaborative logistics: load mixing. 

Load mixing is the idea of mixing two or more items of different weights in the 


94 

same container to reduce the number of trucks needed. In this research we 

explore the impact of product due dates on the mixing of commodities to more 

utilize trucks. We propose heuristics to solve this problem, and present 

computational results from this work. 

4 - Cost-stable Truck Scheduling at a Cross-dock Facility with 

Unknown Truck Arrivals 

Mihalis Golias, University of Memphis, 104 Engineering Science 

Bldg, 3815 Centr, Memphis, 38152, United States of America, 

mgkolias@memphis.edu, Dincer Konur 

We study a cross-dock operator’s truck scheduling problem at inbound doors in 

case of unknown truck arrival times. Due to uncertainty of truck arrivals, a 

scheduling strategy is defined as a schedule with low variation levels. A biobjective 

optimization problem is formulated and we discuss a genetic algorithm 

based heuristic to find Pareto efficient schedules. The proposed approach is 

compared to first-come-first-served policies. 

■ SB43 


RAS Problem Solving Competition 

Sponsor: Railway Applications 


Chair: Kamalesh Somani, Manager Operations Research, CSX 

Transportation, 500 Water St, Jacksonville, FL, 32202, 

Kamalesh_Somani@csx.com 

1 - Movement Planner Algorithm Design for Dispatching on 

Multi-Track Territories 

Kamalesh Somani, Manager Operations Research, CSX 

Transportation, 500 Water St., Jacksonville, FL, 32202, 

Kamalesh_Somani@csx.com 

A dispatcher’s task is to give track authorities to trains to maximize overall system 

efficiency while abiding by a number of business rules. This is done by deciding 

which train takes the siding when two trains traveling in opposite directions meet 

each other, or when a faster train is to overtake the leading slower train. They 

must also decide on the departure and hold times at the yard where train 

originate terminate or stop for work events. The three finalists will compete at 

this session. 

■ SB44 


Topics in Supply Chain Management 


Chair: Tracy Johnson-Hall, Assistant Professor, Mason School of 

Business, College of William & Mary, P.O. Box 8795, Williamsburg, VA, 

23187-8795, United States of America, tdjohns@clemson.edu 

1 - Isomorphism Pressures and Management Support: An 

Empirical Investigation in Supply Chain Security 

Guanyi Lu, PhD Candidate, Mays Business School at Texas A&M 

University, 320N, 4217 TAMU, College Station, TX, 77843-4217, 

United States of America, glu@mays.tamu.edu, 

Xenophon Koufteros 

This study examines the antecedents and consequences of supply chain security 

practices. Data analyses based on 193 U.S. firms suggest that only normative and 

performance pressures have a sizable effect on top management support toward 

security (TMS). The results also show that TMS affects proactive and contingency 

planning which have strong positive effects on security performance. 

2 - Motivation of Time Compression Diseconomies on Transitory 

Network Collaborations 

Adrian Tan, PhD Candidate, Wilfrid Laurier University, 

75 University Avenue West, Waterloo, ON, N2L 3C5, Canada, 

tanx1410@mylaurier.ca, Hamid Noori 

Time-based competition is increasingly used across multiple industries as a basis 

for strategic advantage. However, the challenge of time compression diseconomies 

constraint companies from building internal resources for competitive purposes as 

rapidly as might be desirable. An empirical study to assess if companies can meet 

the challenge by engaging in transitory network collaborations is presented.

3 - Time to Recall: A Duration Analysis Model of Recall Strategies 

and Effectiveness 

Tracy Johnson-Hall, Assistant Professor, Mason School of Business, 

College of William & Mary, P.O. Box 8795, Williamsburg, VA, 

23187-8795, United States of America, tdjohns@clemson.edu, 

Manpreet Hora, Aleda Roth 

We study the role of supply chain characteristics and strategies adopted by firms 

during product recalls and their association with time to recall for perishable 

goods, specifically food products. We employ a duration analysis to examine this 

association. We discuss our results and their implications for firms, consumers and 

policy-makers. 

4 - Order Quantity Variability in Perishable Product Supply Chains 

Stefan Minner, Professor, Technische Universitat München, TUM 

School of Management, Arcisstr 21, Munich, 80333, Germany, 

stefan.minner@tum.de, Sandra Transchel 

We analyze orders for perishable products under customer reactions to stockouts 

and inventory issuing policies. Variability of order quantities is lower than 

demand variability and implies a reduced bullwhip effect. We show that different 

combinations of stockout and issuing policies exhibit different order quantity 

variability with increasing service level and discuss a choice game where supplier 

and retailer can choose between the available options of stockout and depletion 

policy. 

■ SB45 


JFIG Paper Competition II 

Sponsor: Junior Faculty Interest Group 


Chair: Esra Buyuktahtakin, Assistant Professor, Wichita State 

University, IME Department, Wichita, KS, United States of America, 


1 - JFIG Paper Competition 




Papers are submitted for this year’s JFIG paper competition, and each one is 

evaluated based on the importance of the topic, appropriateness of the research 

approach, and the significance of research contribution. In this session, finalistsselected 

in two rounds of review, will present their papers. For the selected 

finalists and the abstracts of the selected papers, please refer to the online 

program. 

■ SB46 


Organizational and Industry Renewal (II): Legitimacy, 

Social Norms, and Environmental Dynamics 



Chair: Andreas Schwab, Associate Professor, Iowa State University, 



1 - The Entrepreneur as David on Horseback 

Joseph Broschak, Associate Professor, University of Arizona, 

Eller College of Management, Department of Management & 

Organizations, Tucson, AZ, 85721, United States of America, 

broschak@email.arizona.edu, Kenneth Koput 

We investigate the role of social identity, information sources, and networks on 

new technology adoption. In the horse industry, the horseshoe is a pathdependent 

technology. Most horses are shod with the steel keg shoe that has been 

the dominant design since the 1800s, despite widespread changes in how horses 

are owned, kept, and used, and alternative shoe designs. Using survey data of 

horse owners, we explore how different social influence variables affected 

technology adoption decisions. 


95 

2 - When Practices Diffuse in a Bubble: Reverse Mergers and the 

Internet Wave 

Peggy Lee, WP Carey-ASU, 300 E. Lemon St., Tempe, 85287, 

United States of America, Peggy.Lee@asu.edu, Ivana Naumovska, 

Edward Zajac 

We theoretically and empirically examine the diffusion process in a bubble 

context. In doing so, we synthesize information-based theories of imitation with 

institutional theory to explain the growth and decline of reverse mergers (RM) of 

internet firms during the time of the internet wave (1998-2000). Our results 

show a clear two-stage pattern, with early RMs done by established firms, and 

late RMs done by fledgling firms. We also find evidence of an early mover 

advantage. 

3 - The Dynamic Mdel of Onership Hterogeneity and Corporate 

Social Responsibility 

Nai Hua Wu, Texas A&M University, 4221 TAMU, College Station, 

TX, 77843, United States of America, nwu@mays.tamu.edu, 

David Sirmon 

Research in corporate social responsibility (CSR) has focused mainly on 

managerial discretion or stakeholders. We intend to fill the gap by addressing CSR 

from the corporate governance perspective, specifically, the effect of ownership 

heterogeneity. Based on multiple agency theory, we predict that dedicated 

institutional owners and family owners are positively related to CSR performance, 

whereas transient institutional owners have negative influence. Our initial results 

support our predictions. 

4 - Redirecting Attention: Overcoming the Illegitimacy Discount 

Eric Yanfei Zhao, University of Alberta, 2-24 Business Building, 

U. of Alberta Business School, Edmonton, AB, T6G 2R6, Canada, 

ericy.f.zhao@gmail.com, Michael Lounsbury 

How can organizations spanning rigid categories solve the dilemma of sameness 

and difference and gain attention and favorable evaluation from audiences? In 

this paper, we argue that organizations can deploy cultural resources such as 

name to overcome audience inattention and legitimacy deficit due to category 

spanning. We tested and found support for this proposition using a sample of 

films released in the U.S. market between 1982 and 2007. 

■ SB47 

SB47 

47- North 230- CC 

Traffic Control Schemes and Application I 




Chair: Yang Cheng, Research Associate, University of Wisconsin- 

Madison, 2204 Engineering Hall, Madison, WI, 53706, 

United States of America, cheng8@wisc.edu 

1 - Equity Based Real Time Traffic Signal Control at Network Level 

using Historical Delay Information: An Application of 

Vehicle-to-Infrastructure (V2I) Communication 

H. M. Abdul Aziz, PhD Student, Purdue University, 150 Arnold 

Drive, Apt15, West Lafayette, IN, 47906, United States of America, 

haziz@purdue.edu, Satish Ukkusuri 

The efficiency of signal control schemes are often evaluated based on aggregate 

system measures only. However, efficiency does not guarantee the equity 

(fairness) of the system. This research proposes a signal control scheme that 

assures fairness (in terms of experienced intersection delay due to signal control) 

across the users of the system. The signal control scheme uses the historical delay 

information along the entire trip of the users and we propose a vehicle-toinfrastructure 

communication based framework for its implementation in the real 

world. 

2 - Large Scale Traffic Signal Control: A Simulation-based 

Approach 

Carolina Osorio, Assistant Professor, Massachusetts Institute of 

Technology, Room 1-232, 77 Massachusetts Avenue, Cambridge, 

MA, 02139, United States of America, osorioc@mit.edu, 

Linsen Chong 

We propose a simulation-based optimization algorithm that efficiently solves large 

scale traffic signal control problems. The method is based on a metamodel 

approach that combines information from a detailed yet inefficient traffic 

simulator with information from a highly efficient and scalable macroscopic traffic 

model. We efficiently solve a signal control problem for the entire Swiss city of 

Lausanne, and derive signal plans that reduce network-wide trip travel times.

SB48 

3 - Intersection Signal Optimization using Queue Length Estimation 

Yang Cheng, Research Associate, University of Wisconsin-Madison, 

2204 Engineering Hall, Madison, WI, 53706, United States of 

America, cheng8@wisc.edu, Jeff Ban, Bin Ran 

By modern traffic detection technologies, traffic condition data would have higher 

resolution in time and space, which is especially valuable for the interrupted 

traffic on signalized intersections. The queue length can be estimated more easily 

and accurately. Therefore, this study proposes a new signal control optimization 

method based on queue information to improve corridor coordination. The 

simulation results indicate that this approach has better performance than 

traditional control methods. 

4 - An Optimal Lane-based Signal Merge Control Model for 

Freeway Work Zone Operations 

Yue Liu, University of Wisconsin-Milwaukee, P.O. Box 784, 

Milwaukee, WI, 53201, United States of America, 

liu28@uwm.edu, Jing Mao 

This paper presents a dynamic control model for optimizing lane-based signal 

merge (LBSM) operations at freeway work zones. The control objective is to 

maximize the throughput. GA is employed to solve the model. Results reveal that 

the proposed method yields much better performance than traditional merge 

strategies under heavy traffic conditions. 

■ SB48 


Integrated Supply Chain Design 

Sponsor: Transportation Science & Logistics/ Freight Transportation 

& Logistics 


Chair: Maged Dessouky, Professor, University of Southern California, 

University Park Campus, Los Angeles, CA, 90007, 

United States of America, maged@usc.edu 

1 - Near-optimal Transportation and Fair Cost Allocation in an 

Agricultural Supply Chain with Perishable Products 

Christine Nguyen, University of Southern California, 

Los Angeles, CA, United States of America, nguyen7@usc.edu, 

Alejandro Toriello, Maged Dessouky 

We consider an agricultural supply chain where several suppliers consolidate 

transportation of a perishable product to achieve economies of scale. Using 

industry data from California flower growers, we compare a practical greedy onestep 

lookahead policy against optimal solutions computed from a dynamic 

program, and determine whether transportation costs can be allocated fairly 

among the suppliers. 

2 - Integrated Supply Chain Network Design: Location, 

Transportation, Routing and Inventory Decisions 

Mingjun Xia, Arizona State University, 1718 S Jentilly Lane, 

APT 204, Tempe, AZ, 85281, United States of America, 

Mingjun.Xia@asu.edu, Ronald Askin 

We propose a novel model to simultaneously optimize location, allocation, 

transportation, inventory and routing decisions in a stochastic multi-product 

supply chain system. Each producer provides a unique product. Each retailer has 

independent, random demand for multiple products. Consolidation and 

distribution facilities may be constructed to reduce shipment costs and increase 

the frequency of economical shipments. Heuristics are developed to solve largesize 

scale instances of the problem. 

3 - Integrated Planning of Supply Chain Layout and Transportation 

Infrastructure Rehabilitations 

Leila Hajibabai, University of Illinois at Urbana-Champaign, 205 N 

Mathews Avenue, Room 3142, Urbana, IL, 61801, United States of 

America, leila.hajibabai@gmail.com, Yun Bai, Yanfeng Ouyang 

Expansion of supply chains induces considerable freight shipments between 

supply/demand points and intermediate facilities, which significantly accelerate 

highway pavement deterioration. This paper presents an analytical method to 

simultaneously design the supply chain as well as the pavement rehabilitation 

plans, which determines the optimum number and location of facilities, shipment 

routes, and pavement rehabilitation intensity and frequency. 

4 - A Production Function-Based Empirical Analysis of Distribution 

Networks 

Donald Warsing, Associate Professor, North Carolina State 

University, 2346 Nelson Hall, Raleigh, NC, 27695, United States of 

America, don_warsing@ncsu.edu, Michael Kay, Christian Rossetti 

Using a number of public data sources, we compare distribution networks in the 

continental U.S. across several industry sectors. We gather data on, and/or 


96 

compute from base-level data, a number of independent variables that we use as 

inputs to a cost function for each DC in each industry data set. This cost function 

serves as an important means of analyzing the variance in the input data and cost 

data. We present insights regarding differences in distribution strategies across 

industries. 

5 - Designing LTL Load Plans to Mitigate Impact of Transit 

Time Variability 

Yu Zhang, PhD Student, Georgia Institute of Technology, 

765 Ferst Dr NW, Atlanta, GA, 30339, United States of America, 

zhangyu@gatech.edu, Alan Erera 

We develop a method to design robust load plans for LTL carriers in order to 

mitigate the cost impacts of transit time variability. Load plan candidates are 

generated based on optimizing operations assuming that transit times take 

expected values, and then evaluated and refined by Monte Carlo simulation. The 

proposed method is applied to a case study which is built based on terminal and 

shipment data from a national LTL carrier. 

■ SB49 


Joint Session TSL/SPPSN: Networks and 

Emergency Response 




Chair: Halit Uster, Industrial and Systems Engineering, Texas A&M 

University, College Station, TX, United States of America, 

uster@tamu.edu 

1 - A Multi-objective Integrated Approach for Emergency 

Response Network Design 

Jyotirmoy Dalal, Texas A&M University, 3131 TAMU, College 

Station, TX, 77843, United States of America, 

jyotirmoy.dalal@gmail.com, Halit Uster 

We consider an emergency response network design problem to determine supply 

locations and flows required in a large scale evacuation setting. We develop a 

multi-objective optimization model with generalized cost structures and 

proximity requirements. We present an epsilon-Costraint based approach for 

generating the efficient frontier and the computational results, demonstrating the 

efficiency of the method. 

2 - Network Design for Emergency Response 

Jianing Zhi, University of Alabama, 300 Alston Hall, Tuscaloosa, 

AL, 35487, United States of America, jzhi@crimson.ua.edu, 

Burcu Keskin, Sharif Melouk 

We investigate a multi-period ambulance location planning problem to minimize 

total cost while maintaining acceptable response times. The network consists of 

supply centers, responder and incident locations, and hospitals. We propose two 

models: 1) all incidents require service in the period in which they occur, and 2) 

incident servicing delays are allowed but with penalty. Through experimentation, 

we compare these two models in terms of service quality, response time, and cost. 

3 - Resilience in Passenger Transportation Networks 

Reza Faturechi, PhD Candidate, University of Maryland, 1173 

Glenn L. Martin Hall, College Park, MD, 20742, United States of 

America, reza@umd.edu, Elise Miller-Hooks 

A bi-level stochastic program with equilibrium constraints is presented for 

assessing and optimizing resilience of passenger transportation networks. At the 

upper level, preparedness actions are taken to enhance the network or facilitate 

response actions, and response actions are taken post-disaster to restore system 

capacity. At the lower level, a partial user equilibrium is sought in which 

passengers choose their routes to minimize travel time given post-disaster system 

performance. 

4 - Logistics Planning for Restoration of Network Connectivity 

After a Disaster 

Sibel Salman, Koc University, Rumeli Feneri Yolu, Sariyer, 

Istanbul, 34450, Turkey, ssalman@ku.edu.tr, Ayse Nur Kibar 

After a disaster, damaged roads should be repaired/unblocked by a fleet of 

machinery initially positioned in a depot to restore highway connectivity. We 

define a selective arc routing problem to find the walk of k vehicles with 

minimum time to reach connectivity. We provide a flow-based mixed integer 

programming model. We consider Istanbul’s highway and determine the number 

of machinery and their initial locations under different road damage scenarios for 

earthquake preparedness.

■ SB50 


Raytheon Military Defense Applications 



Chair: George Mayernik, Raytheon Integrated Defense Systems, 

225 Presidential Way, Mailstop 27/2430b, Woburn, MA, 01801, 

United States of America, george_e_mayernik@raytheon.com 

1 - Operations Analysis of a Precision Tracking Space System 

(PTSS) for Phased Adaptive Approach (PAA) 

Daniel Wilson, Senior Principal Systems Engineer, Raytheon 

Integrated Defense Systems, 225 Presidential Way, Mailstop 

27/2317D, Woburn, MA, 01801, United States of America, 

Daniel.D.Wilson@raytheon.com 

The Precision Tracking Space System (PTSS) is in concept development to provide 

tracking for the Phased Adaptive Approach (PAA). The Space Tracking 

Surveillance System (STSS) demonstrated the value of Low Earth Orbit (LEO) 

satellites to the Ballistic Missile Defense System (BMDS). Quality of Service (QoS) 

metrics are being used as analytics to develop traceability from performance 

specifications to the mission effectiveness, system risk, and affordability of 

alternative BMDS architectures. 

2 - Deployment Concerns and Battle Management Scheduling of 

Unmanned Aerial Systems 

Jeffrey Schvey, Raytheon, 235 Presidential Way, 

Woburn, MA, 01801, United States of America, 

Jeffrey_E_Schvey@raytheon.com, George Blaha 

In this talk, we present an overview of operations studies using a decentralized 

network of airborne sensors to provide missile defense engagement support. We 

consider deployment concerns such as basing logistics and target area coverage, as 

well as sensor resource scheduling during engagements to maximize system 

effectiveness. Results for various deployment strategies and sensor scheduling 

techniques are shown. 

3 - Additional Ballistic Missile Defense (BMD) Considerations in a 

Raid Environment 

George Mayernik, Raytheon Integrated Defense Systems, 225 

Presidential Way, Mailstop 27/2430b, Woburn, MA, 01801, 

United States of America, george_e_mayernik@raytheon.com, 

John Krasnadevich, Michael McGowan 

Recent USA Missile Defense Agency studies have involved defense of regions 

against multiple ballistic missile threats arriving within a specified period of time. 

This paper characterizes the key factors related to BMD against a raid of threats 

and subsequent impact on defense resources. The effect of raid size and density is 

shown including defense systems effectiveness in mitigating stressing raid 

conditions. Radar/engagement timelines and defended areas for a scenario are 

presented. 

■ SB51 


Technology Solutions for Soldier Support 



Chair: Tom Turner, Concurrent Technologies Corp., 747 River Road, 

Hollis Center, ME, 04042, United States of America, turnert@ctc.com 

1 - A Proposed Analytic Method for Predicting IED Emplacements 

Johnathon Dulin, Concurrent Technologies Corporation, 771 

Fairdale Ct, Castle Rock, CO, 80104, United States of America, 

dulinj@ctc.com 

The proliferation of Improvised Explosive Devices (IEDs) over recent years has 

spawned many efforts to mitigate their effects or disrupt networks; much less 

attention has been given to predicting emplacement. Observed environmental 

anomalies may suggest an IED, but improved operational planning can help keep 

personnel out of harm’s way. By combining the strategic planning of game theory 

with the policy optimization of MDPs, we have the framework for a new 

approach in predicting IED emplacements. 

2 - Representing the Power and Energy Needs of Today’s Soldiers 

in Modeling and Simulation 

Dan Rice, Technology Solutions Experts, Inc., 209 West Central St., 

Suite 300, Natick, MA, United States of America, 

daniel.rice@tseboston.com, Louisa Katlubeck, Adam Peloquin, 

Cortney Eldridge, Steven Yuhaski 

Each piece of equipment carried increases a Soldier’s load weight, potentially 

impacting performance and safety. Technology Solutions Experts, Inc. (TSE) used 


97 

simulation tools to study the tradeoff between Soldier encumbrance and task 

performance. TSE created a model and Monte Carlo simulation to study the 

power requirements related to a Soldier’s electronic equipment load weight to 

improve the ability to identify critical relationships between Soldier, equipment, 

and operational environment. 

3 - The Operational Readiness Level Metric for 

Technology Development 

Kenneth Donovan, Lockheed Martin Fellow, Lockheed Martin, 

1801 State Route 17C, MD 0532, Owego, NY, 13827, 

United States of America, ken.b.donovan@lmco.com 

Operational Readiness Level (ORL) is a proposed metric for assessing the 

operational maturity of a technology during the development cycle. ORL 

addresses a limitation of the Technology Readiness Level (TRL) metric that 

measures the technical maturity of an item, rather than the operational use. ORL 

requires early verification of the operational usage of a technology solution, 

thereby guiding technology development to produce items that benefit the user. 

Copyright 2012 Lockheed Martin Corporation. 

■ SB52 


ENRE Award Session 



Chair: Juan Pablo Vielma, University of Pittsburgh, 

1043 Benedum Hall, 3700 O’Hare Street, Pittsburgh, PA, 15261, 

United States of America, jvielma@pitt.edu 

1 - ENRE Student Award: More to Follow 

Juan Pablo Vielma, University of Pittsburgh, 1043 Benedum Hall, 

3700 O’Hare Street, Pittsburgh, PA, 15261, 


This session includes presentations by the winners of the 2012 student travel 

award and best publications in energy, natural resources and the environment. In 

addition the following 2011 awards will be given: -Bistra Dilkina, Best Paper 

Award in Forestry Sponsored Sessions; -Rodolfo Carvajal, Best Paper Award in 

Environment and Sustainability Sponsored Sessions; - Gonzalo Munoz and Donal 

O’Sullivan, Best Paper Awards in Mining Sponsored Sessions. 

■ SB53 

SB53 


Stochastic Programming in Energy 



Chair: Asgeir Tomasgard, Professor, Norwegian University of Science 

and Technology, Alfred getz vei 3, Trondheim, 7491, Norway, 

asgeir.tomasgard@iot.ntnu.no 

1 - The Optimization Problem for a SmartGrid Aggregator with 

Multiple Markets for Flexibility 

Stig Ottesen, PhD Candidate, Norwegian University of Science and 

Technology, Alfred getz vei 3, Trondheim, 7491, Norway, 

stig.ottesen@ncesmart.com, Asgeir Tomasgard 

More dynamics in the power systems lead to an increasing need for operational 

flexibility. Traditionally only generators and large consumers have access to 

markets where flexibility is valued. A SmartGrid aggregator may open such 

markets also for demand side participation. This paper describes the aggregator 

role and its possible business opportunities in current market regime. A 

mathematical formulation of the aggregator’s optimization problem is given 

covering different time horizons. 

2 - Long Term Capacity Equilibria under Risk: A Stochastic LP 

Approach Extending MARKAL 

Daniel Ralph, University of Cambridge Judge Business School, 

Trumpington Street, Cambridge, CB2 1AG, United Kingdom, 

d.ralph@jbs.cam.ac.uk, Yves Smeers 

Linear Programming, LP, is standard for modelling long term capacity equilibria, 

eg, MARKAL for electricity capacity, given Perfect Competition, PC. Uncertainty 

can be handled via expectations, ie, stochastic LP. We extend this to give an 

economic market interpretation of system capacity planning under risk aversion 

(Coherent Risk Measures). Namely, stochastic LP can be used to model capacity 

equilibria under risk assuming PC and financial securities are traded in a 

Complete Risk Market.

SB54 

3 - Optimal Fuel Switching and Contract Decisions in 

Electricity Generation 

Michael Rice, Colorado School of Mines, 816 15th St., Golden, CO, 

80401, United States of America, mrice@mines.edu 

The shale gas boom and proposed plant emissions regulations could accelerate a 

switch from coal to gas in electricity generation portfolios. We analyze optimal 

fuel contracts for power plants facing price and regulatory uncertainty using 

stochastic programming. 

4 - Capacity Expansion under Uncertainty with Decison 

Dependent Probabilities 

Asgeir Tomasgard, Professor, Norwegian University of Science and 

Technology, Alfred getz vei 3, Trondheim, 7491, Norway, 

asgeir.tomasgard@iot.ntnu.no, Lars Hellemo, Paul I. Barton 

We propose an investment problem modeled as a stochastic program with 

decision dependent probabilities. We assume there is an activity or technology 

available that will alter the probabilities of the discrete scenarios occuring. By 

investing in such technology or activity, it is possible to increase the probability of 

some scenarios, while reducing the probability of the remaining scenarios. We 

demonstrate the use of a specialized decomposition algorithm for this class of 

problems. 

■ SB54 


Nicholson Student Paper Prize II 

Cluster: Nicholson Student Paper Prize 


Chair: Ashish Goel, Stanford University, Stanford, CA, 

United States of America, ashishg@stanford.edu 

1 - American Optimal Decisions - Portfolio Safeguard (PSG): 

Advanced Nonlinear Mixed-Integer Optimization Package 

Yi-Hao Kao,Stanford University, Stanford CA, United States of 

America, edward@yhkao.com, Benjamin Van Roy 

We consider a problem involving estimation of a high-dimensional covariance 

matrix that is the sum of a diagonal matrix and a low-rank matrix, and making a 

decision based on the resulting estimate. Such problems arise, for example, in 

portfolio management, where a common approach employs principal component 

analysis (PCA) to estimate factors used in constructing the low-rank term of the 

covariance matrix. The decision problem is typically treated separately, with the 

estimated covariance matrix taken to be an input to an optimization problem. We 

propose directed PCA, an efficient algorithm that takes the decision objective into 

account when estimating the covariance matrix. Directed PCA effectively adjusts 

factors that would be produced by PCA so that they better guide the specific 

decision at hand. We demonstrate through computational studies that directed 

PCA yields significant benefit, and we prove theoretical results establishing that 

the degree of improvement over conventional PCA can be unbounded. 

2 - Optimal Queue-size Scaling in Switched Networks 

Yuan Zhong, Massachusetts Institute of Technology, 235 Albany 

Street, Cambridge, MA, 02139, United States of America, 

zhyu4118@mit.edu, Neil Walton, Devavrat Shah 

We consider a switched network model, an important instance of the stochastic 

processing network (SPN) model (cf. Harrison [9]). The SPN model has been used 

successfully to model a broad spectrum of networked systems, including those 

arising in communications, manufacturing, transportation, inventory 

management, etc. The main operational issue in this model is ecient resource 

allocation under service constraints. We focus on online resource allocation 

policies, where the allocation decisions are made based on the current state 

and/or past history of the system. In the main result of this paper, we provide a 

new class of online policies that achieve optimal queue-size scaling for a class of 

switched networks, including input-queued switches. In particular, it establishes 

the validity of a long-standing conjecture (documented in [20]) about optimal 

queue-size scaling for input-queued switches. 

3 - Heavy Traffic Approximation of Equilibria in Resource 

Sharing Games 

Yu Wu, Stanford University, Stanford, CA, United States of 

America, yuwu@stanford.edu, Ramesh Johari, Luc Bui 

We consider a model of priced resource sharing that combines both queueing 

behavior and strategic behavior. We study a priority service model where a single 

server allocates its capacity to agents in proportion to their payment to the 

system, and users from different classes act to minimize the sum of their cost for 

processing delay and payment. As the exact processing time of this system is hard 

to compute and cannot be characterized in closed form, we introduce the notion 

of heavy traffic equilibrium as an approximation of the Nash equilibrium, derived 

by considering the asymptotic regime where the system load approaches capacity. 

We discuss efficiency and revenue, and in particular provide a bound for the price 

of anarchy of the heavy traffic equilibrium. 


98 

■ SB55 


Revenue Management and Related Applications in 




Chair: Goutam Dutta, Professor, Indian Institute of Management, 


380015, India, goutam@iimahd.ernet.in 

1 - Designing Pricing Policies for Cloud Service Providers 

Megha Sharma, Assistant Professor, Indian Institute of 

Management Calcutta, Diamond Harbour Road, Joka, Kolkata, 

WB, 700104, India, megha@iimcal.ac.in, Sumanta Basu, 

Soumyakanti Chakraborty 

Infrastructure-as-a-Service (Iaas) is expected to grow at a rapid pace in the next 

few years because of the flexibility offered to online platform providers in terms 

of meeting demand fluctuations. Pricing this service naturally becomes a critical 

issue for IaaS providers. We explore different pricing options for IaaS providers 

and their acceptance conditions for online businesses. We conclude by comparing 

our findings from different pricing plans and offer some guidelines on pricing 

IaaS. 

2 - Revenue Management and Pricing for Freight Transport 

in Indian Railways 

G Raghuram, Professor, Indian Institute of Management 

Ahmedabad, Wing 15, IIMA Campus, IIMA Campus, Ahmedabad, 

380015, India, graghu@iimahd.ernet.in 

The pricing policy for freight transport in Indian Railways demonstrates a 

recognition of where it is monopolistically placed, and competing with other 

modes. This policy is a departure from the earlier one which considered the 

impact of rail transport cost on the final price of the commodity. There are also a 

variety of discounts and surcharges depending on the traffic flows. This paper 

brings out the implication of the pricing policy, organizational processes behind 

the same and a way forward. 

3 - A Mathematical Programming Approach with Revenue 

Management for Home Loan Pricing 

Goutam Dutta, Professor, Indian Institute of Management, 


380015, India, goutam@iimahd.ernet.in, Deepika Thakur 

We formulate a dynamic pricing model for home loans for a bank. The model 

optimizes the net present value of money available to the bank subject to pricing 

limits, cash flows. It also considers the demand price (interest rate) relationship 

and default probability as a function of interest rate. We then assume that 

different versions of demand function. We consider when demand function is 

given by a straight line, an exponential function and by rectangular hyperbola. 

■ SB56 


Technology Management 


Chair: Yuya Kajikawa, Associate Professor, Tokyo Institute of 

Technology, 3-3-6 Shibaura, Minato-ku, Tokyo, Japan, 

kajikawa@mot.titech.ac.jp 

1 - Characterizing Business Models using System Dynamics 

Kwanyoung Im, POSTECH, 326/4 Eng building, Hyoja dong, 

Pohang, Korea, Republic of, kwanyoung@postech.ac.kr, 

Jaehun Lee, Inyoung Koh, Hyunbo Cho 

To apply the concept of a business model in real business practices, the 

consolidation of domain knowledge and application tools is needed. This research 

is aimed at reviewing the related literature and at proposing a business model 

application tool based on system dynamics. 

2 - The Influence of Knowledge Based on Technology 

Collaboration 

Yan Liu, Xi’an Jiaotong University, Postbox 2230, Xi’an, China, 

liuyan0220@stu.xjtu.edu.cn, Hong Cai 

We explore how a firm’s knowledge base structure influences the formation of its 

technological collaborating relationship. We use patent data from Chinese 

electrical & electronic industry from 2001 to 2008 to study the effect of firms’ 

knowledge base on its propensity to establish technical cooperative relations. 

Counting data regression models show a strong link between these four 

properties of firm’s knowledge base and the propensity to form technological 

cooperating relationship.

3 - Participatory Challenge Platforms: Effective Integration of 

Public Expertise in Organization Efforts 

Tanya Kelley, Arizona State University, 411 North Central Ave., 

Suite 400, Phoenix, AZ, 85004, United States of America, 

tanya.musgrave@asu.edu, Erik Johnston, Chase Gordon 

Participatory challenge platforms facilitate the integration of external individuals’ 

personal interests and expertise into operational efforts of public and private 

organizations. We present results of NSF funded research on two such challenge 

platforms with distinct design considerations, 10,000 Solutions and The Policy 

Challenge. We focus on design tensions, outcomes, and how solutions are utilized 

by organizations. Findings suggest recommendations for future platform design 

and use. 

4 - Multilayer Analysis of Computational Intelligence for 

R&D Project Design 

Yuya Kajikawa, Associate Professor, Tokyo Institute of Technology, 

3-3-6 Shibaura, Minato-ku, Tokyo, Japan, 

kajikawa@mot.titech.ac.jp 

In this research, a methodology to analyze multilayer information is developed. 

We proposed and evaluated relatedness measures to comprehend semantic 

relationships between science, technology and social issues. Bibliometrics has 

been be used in previous works to comprehend trends of science and technology, 

however, have rather limited capability to support design than observation. We 

demonstrate that our methodology can support design of new R&D projects. 

■ SB57 


Topics in Queueing Models 


Chair: Jayakrishnan Nair, Caltech, 1200 E. California Blvd., Pasadena, 

CA, 91125, United States of America, ujk@caltech.edu 

1 - A Fluid Model for the Multi-class Network Queues with 

Time-varying Arrivals and Routing Paths 

Beixiang He, North Carolina State University, Raleigh, NC, 27695, 

United States of America, bhe@ncsu.edu, Yunan Liu 

We introduce and analyze a fluid queue network with multiple customer classes, 

non-Markovian patience and service distributions, time-varying arrival rates and 

staffing level, and deterministic routing paths. Unlike the traditional Markovian 

(or proportional) routing, customers are routed according to their own service 

paths inside the network in a deterministic fashion. We provide an efficient 

algorithm to compute all standard performance functions, such as the queue 

lengths and waiting times. 

2 - Modeling Cross Correlation with Markovian Arrival Processes 

Sunkyo Kim, Professor, Ajou University, San 5, Suwon, Korea, 

Republic of, sunkyo@ajou.ac.kr 

We consider a generalization of modeling cross correlation in decomposition 

approximation of queueing networks. The cross correlation between randomly 

split streams is accounted for in the second and third moments of the merged 

process which is approximated as a two-state Markovian arrival process. 

3 - Stabilizing Policies for Matching Portals 

Burak Buke, The University of Edinburgh, School of Mathematics, 

King’s Buildings, JCMB, Edinburgh, United Kingdom, 

B.Buke@ed.ac.uk, Miklos Rasonyi, Hanyi Chen 

The portals, which match the people who provide a specific service with the 

people who demand the service, are becoming increasingly popular. In this talk, 

we assume that customers and suppliers arrive at the portal according to 

independent Poisson processes. Each customer can match with a specific supplier 

with a given probability independent of other suppliers and customers. We show 

that such a system is unstable if uncontrolled and suggest control policies to 

ensure stability. 

4 - When Heavy-tailed and Light-tailed Traffic Compete: Response 

Time Tail versus throughput Optimality 

Jayakrishnan Nair, Caltech, 1200 E. California Blvd., Pasadena, 

CA, 91125, United States of America, ujk@caltech.edu, Krishna 

Jagannathan, Adam Wierman 

We study a setting in which a heavy-tailed flow and a light-tailed flow compete 

for service from a single server. In this setting, the classical max-weight policy, 

while throughput optimal, induces heavy-tailed delays for the light-tailed flow. 

We prove that by careful design of intra-queue and inter-queue scheduling 

policies, one can achieve throughput optimality and light-tailed delays for the 

light-tailed flow, without affecting the delay tail for the heavy-tailed flow. 


99 

■ SB58 


Network Science 



Chair: David Goldberg, Professor, Georgia Institute of Technology, 

Atlanta, GA, United States of America, dgoldberg9@isye.gatech.edu 

1 - Distributed Storage for Intermittent Energy Sources: Control 

Design and Performance Limits 

Yashodhan Kanoria, PhD Candidate, Stanford University, 350 

Serra Mall, Stanford, CA, 94305, United States of America, 

ykanoria@stanford.edu, David Tse, Baosen Zhang, 

Andrea Montanari 

The power output of renewable energy sources like wind and solar varies with 

time and location. Two strategies have been proposed to manage this variability: 

1) energy storage systems to average over time; 2) distributed generation to 

average over location. We introduce a network model to study the optimal use of 

storage and transmission resources, and propose a control design methodology. 

For specific network topologies we show asymptotic optimality and analytically 

quantify performance. 

2 - Temporal Load Balancing for Distributed Backup Scheduling 

Peter van de Ven, Postdoc, IBM, Yorktown Heights, NY, 

United States of America, petermvandeven@gmail.com, 

Angela Schoergendorfer, Bo Zhang 

We consider users that generate traffic to be transferred to a centralized backup 

server across a data network. Users have time-varying connectivity due to 

working hours, and backup attempts are initiated locally, without knowledge of 

the status of other users or the network load. The backup schedule is 

characterized by hourly backup probabilities, and we discuss how to choose the 

backup probabilities so as to stabilize the system and minimize the network costs. 

3 - Neighborly Measures for Sampling Independent Sets 

David Goldberg, Professor, Georgia Institute of Technology, 

Atlanta, GA, United States of America, 

dgoldberg9@isye.gatech.edu 

We study a family of measures which generalizes the well-known hardcore model 

for sampling the independent sets (i.s.) of a graph. In our model, the probability 

assigned to a given i.s. depends not only on its cardinality, but also on how many 

excluded nodes are adjacent to exactly k nodes belonging to the i.s. for each k. 

We give tight conditions for correlation decay in the infinite regular tree under a 

log-convexity assumption, and gain further insight through a perturbative 

analysis. 

■ SB59 

SB59 


Models for Online Content and Advertising Markets 



Chair: Assaf Zeevi, Columbia Business School, New York, NY, 

United States of America, assaf@gsb.columbia.edu 

1 - Buy-it-now or Take-a-chance: A Mechanism for Real-time 

Price Discrimination 

Hamid Nazerzadeh, Marshall School of Business, University of 

Southern California, Los Angeles, CA, United States of America, 

hamidnz@marshall.usc.edu, Elisa Celis, Greg Lewis, 

Markus Mobius 

Increasingly sophisticated consumer tracking technology allows advertisers to 

reach narrowly targeted consumer demographics. Such targeting improves the 

match quality between advertisers and users, but can also result in thin markets. 

Using historical bidding data from a large advertising exchange, we show that 

there is often a substantial gap between the highest and second highest 

willingness to pay. With this motivation, I present a simple sequential screening 

mechanism. 

2 - Learning Overlapping Community Structure in Social Networks: 

The Leader-follower Algorithm 

Tauhid Zaman, Massachusetts Institute of Technology, 77 

Massachusetts Ave., Cambridge, MA, United States of America, 

zlisto@mit.edu, Devavrat Shah 

We present the leader-follower algorithm (LFA) for learning overlapping 

community structure in social networks. It requires no input parameters and has 

a run-time linear in the network size. We prove that it has good performance on 

certain network structures which we show exist in real social data. We also find 

that the LFA can exactly recover the true overlapping community structure for 

many real social networks while more common methods fail.

SB60 

3 - Telling the “Hot” from the “Not”: Online Content 

Recommendations on the Internet 

Yonatan Gur, Columbia Business School, New York, NY, 

United States of America, ygur14@gsb.columbia.edu, 

Omar Besbes, Assaf Zeevi 

A new class of online services allows publishers to direct readers to other webbased 

content they may be interested in. We study this dynamic recommendation 

problem, focusing on challenges introduced by the massive stream of new content 

and the short shelf-life of articles, and focusing on topic-based popularity as basis 

for recommendations. Work is based on collaboration with a leading provider of 

online content recommendations. 

■ SB60 


Predictability and Resource Allocation Decisions in 

Air Traffic Management 



Chair: Shervin Ahmad Beygi, Lead Operations Research Analyst, 

Metron Aviation Inc., 45300 Catalina Ct, Dulles, VA, 20166, 

United States of America, shervin.ahmadbeygi@metronaviation.com 

1 - Performance-based Ground Delay Program Decision-making 

using Multiple Criteria : Single Airport Case 

Yi Liu, University of California, Berkeley, 107 McLaughlin Hall, 

Berkeley, CA, 94720, United States of America, 

liuyi.feier@gmail.com, Mark Hansen 

We develop performance metrics for capacity, efficiency and predictability, and 

show how these metrics may be traded off in the design of Ground Delay Program 

(GDP) under capacity uncertainty. We link the performance to the decisions on 

GDP clearance time and the scope of the program. This capability allows the 

service provider to make performance-based GDP decisions using multiple 

criteria. 

2 - Airline Voting Mechanisms for Parameter Selection for Air 

Traffic Flow Management (ATFM) Initiatives 

Vikrant Vaze, Member Research Staff, Philips Research North 

America, 345 Scarborough Road, Briarcliff Manor, NY, 10510, 

United States of America, vikrant.vaze@philips.com, 

Cynthia Barnhart 

Parameter selection for ATM initiatives is performed by the regulators while 

taking airline preferences into account. We design a voting scheme which can 

replace current ad-hoc methods. Strategic behavior is modeled using an integer 

programming formulation which explicitly accounts for tie-breaking rules. We 

solve for a Nash equilibrium using best response heuristics. We evaluate 

equilibrium properties including existence, uniqueness, convergence, system 

optimality and pareto optimality. 

3 - Flight Cancellation Behavior and Delay Savings Estimates 

Michael Seelhorst, University of California Berkeley, 1710 

Delaware St., Berkeley, CA, 94703, United States of America, 

mseelhorst@berkeley.edu, Mark Hansen 

A discrete choice model was used to infer airline flight cancellation preferences 

from a sample of on-time performance flight data. The predicted cancellation 

probabilities were then used as inputs to delay savings estimates, which were 

approximated with a continuous technique. The results from this work will be 

useful for incorporating flight cancellation decisions into system-wide delay 

forecasting models. 

■ SB61 


Airport/Airline Operations Management 



Chair: Ahmed Ghoniem, Assistant Professor, University of 

Massachusetts Amherst, 121 Presidents Dr., Amherst, MA, 01003, 

United States of America, aghoniem@isenberg.umass.edu 

1 - Enhanced Column Generation for Multiple-runway Aircraft 

Sequencing Problem 

Farbod Farhadi, Doctoral Student, University of Massachusetts 

Amherst, Isenberg School of Management, 121 Presidents Drive, 

Amherst, MA, 01003, United States of America, 

ffarhadi@som.umass.edu, Ahmed Ghoniem 

We address mixed mode multiple-runway aircraft sequencing problem. An MIP 

formulation is enhanced with valid inequalities and symmetry-defeating 


100 

constraints. An alternative set partitioning problem is solved by column 

generation using multiple algorithmic features that accelerate the convergence 

and tighten the duality gap. 

2 - Aircraft Rescheduling under Operation Disruptions 

Mohamed Kharbeche, Postdoctoral Associate, Qatar University, 

Mechanical and Industrial Engineering, Doha, 2713, Qatar, 

mkharbec@qu.edu.qa, Ameer Al-Salem, Ahmed Ghoniem, 

Hanif D. Sherali 

We address optimization models for multiple-runway aircraft sequencing 

problems under operation disruptions, namely, aircraft delays, new aircraft 

arrivals and flight cancellations. Extensive computational results are reported for 

exact as well as heuristic methods. 

3 - An Integrated Approach for Airline Flight Scheduling, Fleet 

Assignment, and Aircraft Routing 

Ki-Hwan Bae, Virginia Polytechnic and State University, 250 

Durham Hall, Blacksburg, VA, 24061, United States of America, 

kbae04@vt.edu, Hanif D. Sherali, Mohamed Haouari 

We propose a model that integrates certain aspects of the schedule design, fleet 

assignment, and aircraft routing processes, while considering various modeling 

features. Maintenance routing decisions as well as through-flight opportunities 

are additionally incorporated in our model, and we apply a series of different 

solution techniques to handle the large-scale model formulation, reduce its 

complexity, and enhance its solvability. 

4 - Environmental Value of Optimal Separation in Continuous 

Descent Arrivals 

Heng Chen, Isenberg School of Management, 121 Presidents 

Drive, University of Massachusetts, Amherst, MA, 01003, 

United States of America, heng@som.umass.edu, Senay Solak 

Optimal separation in continuous descent arrivals (CDA) refers to the use of 

spacing policies that are shown to be optimal based on stochastic dynamic models. 

These models involve optimization of different objective structures defined as 

functions of runway utilization and airline operating costs. Considering some 

potentially implementable objective structures, we derive both analytical and 

numerical results that quantify the environmental value of using optimal 

separation policies in CDA. 

■ SB62 


Joint Session Auctions/MSOM: 

Auctions and Procurement 

Sponsor: Auctions & Manufacturing & Service Oper Mgmt 


Chair: Gabriel Weintraub, Columbia Business School, 


gyw2105@columbia.edu 

Co-Chair: Sang Won Kim, Columbia Business School, 


skim14@gsb.columbia.edu 

1 - Production Cost Forecasts and Procurement Efficiency in 

Supply Chains 

Tunay Tunca, Associate Professor, University of Maryland, 

Robert H. Smith School of Business, College Park, MD, 20742, 

United States of America, ttunca@rhsmith.umd.edu 

We study the role supplier cost forecasting under upstream competition in supply 

chain procurement and examine the implications on market and channel 

performance. We derive a channel competitiveness measure that combines the 

forecast information structure in the supply chain with upstream channel 

fragmentation and study its relationship with supply chain efficiency. We show 

that such a measure can be a better indicator of supply chain performance 

compared to traditional measures. 

2 - Analysis of the Vickrey-Clarke-Groves (VCG) Mechanism in 

Combinatorial Auctions 

Sang Won Kim, Columbia Business School, 3022 Broadway, 

New York, NY, 10027, United States of America, 

skim14@gsb.columbia.edu, Jay Sethuraman, Gabriel Weintraub 

Recent theoretical works show that the VCG mechanism can have undesirable 

properties, such as low revenues, in combinatorial auction settings. In contrast, 

recent empirical work shows the opposite for a specific important application. In 

this work, we develop new theoretical results for VCG mechanism that explain 

this apparent paradox. More broadly, we show when VCG is expected to perform 

well in applications and when it does not.

3 - Bundled Procurement for (Free) Technology Acquisition and 

Future Competition 

Leon Chu, University of Southern California, Los Angeles, CA, 

United States of America, leonyzhu@usc.edu, Yunzeng Wang 

We study the effectiveness of a procurement mechanism of a buyer that bundles 

the product procurement with the technology acquisition. It turns out that each 

supplier has a dominant offering strategy for the technology provision that only 

depends on the technology difference of the suppliers and the ratio between the 

current project size and the future market size. We find that a ratio of 10% is 

sufficient for the suppliers to offer the best technologies for most reasonable 

market conditions. 

4 - Procurement Auctions with Error-prone RFQs 

Yan Yin, University of Michigan, Ann Arbor, MI, 

United States of America, yinyan@umich.edu 

Many requests for quotes (RFQs) contain errors that will trigger re-design (and 

windfall supplier profits) after production has begun. We study how the buyer’s 

payment changes in the error rate, supplier’s design knowledge, and competition 

intensity. Surprisingly, we find that a higher error rate can sometimes decrease 

the buyer’s total payment by mitigating supplier windfall profits. We also show 

“pre-paying” re-design costs can be an alternative strategy to stem windfall 

profits. 

■ SB63 


Worker and Consumer Behaviors 


Chair: Corey Kiassat, PhD Candidate, University of Toronto, Faculty of 

Applied Science & Engineering, 5 King’s College Road, Toronto, M5S 

3G8, Canada, ckiassat@mie.utoronto.ca 

1 - Hiring An Envy-free Secretary is Costly 

Nguyen Truong, IEOR PhD Student, University of California- 

Berkeley, 2414 Dwight Way #6, Berkeley, CA, 94704, 

United States of America, nltruong@berkeley.edu 

We consider the minimum expected rank secretary problem, where it is well 

known that the optimal stopping mechanism selects an applicant with rank 

$\approx 3.8695$. This mechanism, however, may not deliver intended results: 

early (late) applicants wish to delay (move to earlier) their interviews. In order 

for each applicant to not envy any other’s position, the optimal expected rank of 

the hired applicant is shown to be $\Omega(\ln(n))$. As such, it is costly to have 

envy-free participants. 

2 - Consumer Behavior Analysis in a Convenient Store Based on 

Shopping Path Data 

Chenqu Luo, Master Student, Department of Industrial 

Engineering, Tsinghua University, Building Shunde Room South 

607, Beijing, 100084, China, luocq11@mails.tsinghua.edu.cn, 

Lefei Li 

The consumers’ shopping path indicates much to their shopping behaviors. We 

develop a utility function that describes consumers’ preference on the next 

movement and use Beyesian method for estimation. Statistics show the shopping 

behavior in a convenient store is quite different from that in a supermarket. 

Moreover, we can identify typical path patterns that are closely related to the 

consumers’ target products and visiting time, which potentially support the 

managers’ category decisions. 

3 - Enhancing System Performance through Failure Risk Analysis 

from a Worker Characteristic Perspective 

Corey Kiassat, PhD Candidate, University of Toronto, Faculty of 

Applied Science & Engineering, 5 King’s College Road, Toronto, 

M5S 3G8, Canada, ckiassat@mie.utoronto.ca, Dragan Banjevic 

We want to perform a failure risk analysis to enhance system performance. We 

are specifically interested in the effect of worker characteristics on failure risk. In 

the case of failures caused by workers, the Decision Maker (DM) must intervene 

to mitigate the risk. We develop a revenue model, using the proportional hazards 

model, to provide the DM with a cost-benefit analysis, balancing the advantage of 

risk reduction against the direct cost of the intervention method. 

4 - Intrapersonal Heterogeneity in Intertemporal Choice 

Behavior: Evidence 

Sasha Bartashnik, Memorial Sloan-Kettering Cancer Center, 

405 Lexington Avenue, 3rd Floor, New York, NY, 10174, 

United States of America, bartasha@mskcc.org, Paul Kattuman 

We examine whether the traditional binary classification of individuals into 

dynamically consistent exponential or present-biased quasi-hyperbolic 

discounters is absolute using De Nederlandsche Bank Household Survey data. 

Most individuals are internally heterogeneous. Intrapersonal heterogeneity is 

significant and fairly stable over time. Our results have significant implications for 

enhancing individual and societal welfare via individualized choice architectures, 

particularly in healthcare. 


101 

■ SB66 

SB66 

66- Ellis West- Hyatt 

Joint Session DM/QSR: Sensor-based 

Health Informatics 

Sponsor: Data Mining & Quality, Statistics and Reliability 


Chair: Hui Yang, Assistant Professor, University of South Florida, 


1 - Spatiotemporal Change Detection Methods in Nonhomogeneous 

Change Size 

Sung Won Han, Fellow, Hoffmann-La Roche Inc., 

340 Kingsland St., Nutley, NJ, United States of America, 

sung_won.han@roche.com, Seoung Bum Kim, Kyu Jong Lee, 

Hua Zhong 

In the classical change point detection problem, the change pattern is 

homogeneous over time and space, and the change size is sustained after the 

change. However, in practical spatotemporal health care surveillance, the 

underlying assumptions of problems are more complicated. Change size can be 

various over space, so the change size can be large at the center of outbreak but 

small at the edge. We investigate the existing methods to see how they perform 

under the complex cases. 

2 - Localization of Myocardial Infarction using Vectorcardiogram 

Octant Transition Network 

Trung Le, Oklahoma State University, 322 Engineering North, 

Stillwater, Ok, 74075, United States of America, 

trung.le@okstate.edu, Satish Bukkapatnam 

Myocardial Infarction is a leading cause of death in the US. Detecting the location 

of MI in the early stages can significantly improve treatment efficiency and reduce 

mortality risks. We present an approach that uses the topological features of 

octant network reconstructed from VCG signals to detect and localize MI. 

Extensive tests using data of from Physionet indicate that transition features can 

differentiate healthy, inferior, anterior MI with >92% sensitivity and >93% 

specificity. 

3 - Multiscale Adaptive Basis Function Modeling of 

Spatiotemporal Signals 



Mathematical models facilitate the characterization and recognition of underlying 

biological patterns. However, there are practical issues pertinent to model efficacy, 

robustness, and generality. This paper presents a multiscale adaptive basis 

function approach to model space-time cardiac pathological behaviors, using the 

“best matching” projections of characteristic waves onto a dictionary of nonlinear 

basis functions. The model performance is experimentally evaluated with realworld 

signals. 

4 - Developing a Healthcare Cost Index using Advanced Analytics 

Dursun Delen, Professor, Oklahoma State University, 700 N. 

Greenwood Ave., Tulsa, OK, 74106, United States of America, 

dursun.delen@okstate.edu, James Hess 

Due to high variability, cost of healthcare has become one of the largest risk 

factors for today’s businesses. Establishment of a practical healthcare commodity 

index provides the pathway to transparency, which allows businesses to trade 

healthcare cost futures, and ultimately creates the open market environment 

necessary to encourage and support competitive pricing. This presentation focuses 

on the overall process of developing (based on data and analytics) such a 

healthcare cost index.

SB67 

■ SB67 


System Reliability and Maintenance 



Chair: Nan Chen, Assistant Professor, National University of Singapore, 

Department of Industrial and Systems Eng, 1 Engineering Drive 2, 

Singapore, 129789, Singapore, isecn@nus.edu.sg 

1 - Criticality Index for Degrading Components under 

Dynamic Environments 

Xiao Liu, liuxiaodnn_1@hotmail.com, Khalifa N. Al-Khalif, 

Dingguo Hua, Abdel Magid Hamouda, David Coit, Elsayed Elsayed 

Criticality index of a component depends on its degradation path and system 

configuration. This presentation investigates the criticality analysis for degrading 

components under dynamic environmental conditions. The component 

degradation is modeled by a k-dimensional Wiener process, and component fails 

when any of these k degradation paths attains its own threshold level. In the 

numerical examples, we illustrate the computation of component criticality in 

different system configurations. 

2 - A Quantile Regression Approach to Analyze Accelerated Life 

Testing Data 

Nan Chen, Assistant Professor, National University of Singapore, 

Department of Industrial and Systems Eng, 1 Engineering Drive 2, 

Singapore, 129789, Singapore, isecn@nus.edu.sg 

We propose a quantile regression framework to model the accelerated life tests 

(ALT) data. The quantile of the failure distribution at usage level can be easily 

estimated. Compared with traditional parametric regression methods, quantile 

regression is distribution free, efficient in presence of censoring, and is more 

flexible in modeling ALT relations. In addition, the method can be used naturally 

in the cases when the failure-free life depends on the accelerating factor. 

3 - An Integrated Approach to the Design of Statistically Precise 

and Energy Efficient Accelerated Tests 

Haitao Liao, Associate Professor, University of Arizona, Tucson, AZ, 

85721, United States of America, hliao@email.arizona.edu, 

Dan Zhang 

In this paper, an integrated experimental design approach is introduced with the 

objective of maximizing the reliability estimation precision while minizing the 

energy consumption of an accelerated test. Relevant statistical inference methods 

and required experimental needs and technologies are discussed. 

4 - Reliability and Maintenance for Dependent Competing Failure 

Processes with Shifting Thresholds 

Qianmei Feng, Associate Professor, Houston University, E206 

Engineering Bldg 2, Houston University, Houston, TX, 77204- 

4008, United States of America, qmfeng@uh.edu, Lei Jiang, 

David Coit 

We present reliability and maintenance models for systems subject to multiple 

dependent competing failure processes with a changing and dependent failure 

threshold. Three cases of dependency between shock process and hard failure 

threshold level are studied. Reliability models are developed based on degradation 

and random shock modeling. Two preventive maintenance policies are also 

applied and compared. Then an MEMS example is given to demonstrate the 

reliability models and maintenance polices. 

■ SB68 


Finance: Risk Management 


Chair: Woojin Chang, Seoul National University, Gwanak-gu 

Gwanak-ro 1, Seoul, Korea, Republic of, changw@snu.ac.kr 

1 - Continuous Dynamic Portfolio Choice with Bayesian Regret 

Shea Chen, University of California, Berkeley, 301 Main St., 

Unit 34E, San Francisco, CA, 94105, United States of America, 

sheachen@berkeley.edu, Andrew Lim 

We formulate a continuous-time portfolio choice problem in which the investor is 

uncertain about parameters of the model, and the objective function is a Bayesian 

version of relative regret. The optimal portfolio is characterized and shown to 

involve a “tilted” posterior, where the tilting is defined in terms of a family of 

stochastic benchmarks. 


102 

2 - Forecasting the Stress of Credit Card Portfolios 

Mee-chi So, University of Southampton, Southampton 

Management School, Southampton, SO17 1BJ, United Kingdom, 

m.so@soton.ac.uk 

This study aims to develop models to forecast the default rate of credit card 

accounts under different economic scenarios. Two modelling approaches were 

extended to include the impact of lenders’s actions within the model. The first 

approach was a regression model of the aggregate losses based on economic 

variables. The second approach was a set of vintage level models which 

highlighted the months on books effect on credit losses. A case study using the 

models was described. 

3 - Credit Rating Migration Model: Markov Switching Process 

Woojin Chang, Seoul National University, Gwanak-gu Gwanak-ro 

1, Seoul, Korea, Republic of, changw@snu.ac.kr, Sung Yeol Oh 

We develop a credit rating migration model based on a discrete-time discretevalued 

Markov process depending on hidden Markov chain. We also present a 

parameter estimation procedure for the proposed model using the EM algorithm. 

The likelihood ratio test shows that our model depending on hidden Markov 

chain better describes the credit rating dynamics than a benchmark Markov chain 

model. 

4 - A Model for Dynamic TaxCredit Utilization 

Nilofar Varzgani, PhD Student, Rutgers University, Rutgers 

Business School, Newark, NJ, 07102, United States of America, 

nilofarv@pegasus.rutgers.edu, Suresh Govindaraj, 

Michael Katehakis 

We develop a new stochastic dynamic programming model for dynamic tax credit 

utilization for a firm that provides maximum expected benefit over a rolling 

horizon. The model takes into account earnings and profits, net operating losses, 

and credit carryovers. We present computation to demonstrate the benefit the 

model provides. 

■ SB69 


Data Driven Portfolio Optimization 

Cluster: Optimization in Finance 


Chair: Victor DeMiguel, Professor, London Business School, Regent’s 

Park, London, NW1 4SA, United Kingdom, avmiguel@london.edu 

1 - Computing General Arbitrage Bounds for European Basket 

Options via Dantzig-Wolfe Decomposition 

Luis Zuluaga, Assistant Professor, Lehigh University, 

200 West Packer Ave., Bethlehem, United States of America, 

luis.zuluaga@lehigh.edu 

We study the problem of computing bounds on the price of a basket option given 

the only assumption of absence of arbitrage, and information about prices of 

other basket options. We provide a simple way to compute this type of bounds via 

a suitable Dantzig-Wolfe decomposition. We illustrate our results by computing 

upper and lower arbitrage bounds on gasoline/heating oil crack spread options, 

and Alternative Risk Transfer (ART) products. 

2 - Performance-based Regularization in Mean-CVaR Portfolio 

Optimization 

Gah-Yi Vahn, Assistant Professor, London Business School, 

Regent’s Park, London, United Kingdom, gvahn@london.edu, 

Noureddine El Karoui, Andrew Lim 

We introduce performance-based regularization (PBR), a new approach to 

addressing estimation risk in data-driven optimization, to mean-CVaR portfolio 

optimization. We assume the available log-return data is iid, and detail the 

approach for two cases: nonparametric and parametric (the log-return 

distribution belongs in the elliptical family). We show that the PBR methods 

produce efficient frontiers that are, on average, closer to the population efficient 

frontier with less variability. 

3 - Size Matters: Optimal Calibration of Shrinkage Estimators for 

Portfolio Selection 

Alberto Martin-Utrera, Mr, University Carlos III of Madrid, 

C/ Madrid, 126-28903, GETAFE, Sp, 28903, amutrera@estecon.uc3m.es 

We carry out a comprehensive investigation of shrinkage estimators in the 

context of portfolio optimization and we make several contributions to this field. 

First, we propose new shrinkage estimators for both the moments of asset returns 

and portfolio weights. Second, we propose novel calibration criteria to compute 

the optimal shrinkage intensity. Finally, we propose a parametric and a 

nonparametric approach to estimate the shrinkage intensity.

■ SB70 


Economics of Information Systems 

Sponsor: Information Systems 


Chair: Wael Jabr, Assistant Professor, University of Calgary, 2500 

University Dr NW, Calgary, AB, T3A6L5, Canada, wjabr@ucalgary.ca 

1 - Are Online Seller Recommendations Biased? An Empirical 

Investigation of OTA Hotel Listings 

Ori Marom, RSM Erasmus University, Burgemeester Oudlaan 50, 

Rotterdam, 3062 PA, Netherlands, omarom@rsm.nl, Roman Peskin 

Online travel agents such as Expedia and Priceline.com often have tremendous 

power to direct online buyers to a small subset of available products. Our paper 

investigates whether OTA selections of top-listed hotels can be regarded as biased 

from buyers’ viewpoint. Our empirical investigation indicates that that while OTA 

websites provide buyers with valuable search assistance by locating good-valued 

deals they also influence them to purchase relatively expensive products. 

2 - Social Media and Firm Equity Value 

Jie Zhang, University of Texas Arlington, 701 S West St., 

Arlington, TX, 76019, United States of America, jiezhang@uta.edu, 

Xueming Luo 

Companies have advocated social media to engage consumers and gauge product 

market performance. We empirically examine whether social media metrics are 

related to firm equity value. The results from the developed time-series models 

suggest that social media-based metrics (blogs and reviews) have higher predictive 

value for firm stock market performance compared to web traffic and search. We 

also find that firm stock performance responds faster to the social media-based 

metrics. 

3 - Learning under Ambiguity: An Experiment 

Yaroslav Rosokha, PhD Student, The University of Texas at Austin, 

2225 Speedway Stop C3100, Austin, TX, 78712-1690, United 

States of America, yaroslav.rosokha@gmail.com, Othon Moreno 

Gonzalez 

We design and conduct an economic experiment to compare the learning process 

of the agents under risk and ambiguity. Decisions are made in conjunction with a 

sequence of draws (with replacement), allowing us to track beliefs over time. We 

estimate: (1) Bayesian updating allowing for base rate fallacy; (2) Bayesian 

updating with multiple priors; and (3) Reinforcement Learning. Our findings 

suggest that new signal in is over-weighted more in risky than in ambiguous 

environments. 

4 - The Role of Online Reviews in Mitigating Product Uncertainties 

Wael Jabr, Assistant Professor, University of Calgary, 2500 

University Dr NW, Calgary, AB, T3A6L5, Canada, 

wjabr@ucalgary.ca, Mohammad Rahman 

User reviews are widely recognized as an important resource that customers rely 

on to help guide their purchases. This paper evaluates the efficiency of the market 

for information at reducing the uncertainty of consumers. We use Bayesian Belief 

and Information Gain to quantify such efficiency. We then show that the market 

design for information release plays a major role in affecting this efficiency. We 

investigate the evolution of reviews and corresponding efficiencies and market 

design issues. 

■ SB71 


Digital and Social Networks I 



Chair: Jui Ramaprasad, Assistant Professor, McGill University, 

1001 Sherbrooke West, Montreal, QC, H3H2V1, Canada, 

jui.ramaprasad@mcgill.ca 

1 - Celebrity Effect? An Empirical Study of Social Influence and 

Online Product Review Behavior 

Mingfeng Lin, University of Arizona, 1130 E. Helen St., Tucson, 

AZ, 85721, United States of America, mingfeng@eller.arizona.edu, 

Paulo Goes 

We study the effect of consumers’ online interactions on their product review 

behaviors. A popular form of users’ online interaction is the “subscription” model 

that gives priority to contents generated by the person subscribed to. Using a 

panel dataset from epinions.com, we examine how users’ product review 

behaviors evolve as their online followership – peers who are silent – grows. Our 

results have implications for research on word-of-mouth and user-generated 

content (UGC) sites in general. 


103 

2 - Are Paid Subscriptions on Music Social Networks Contagious? 

Ravi Bapna, Professor, University of Minnesota, 321 19th Avenue 

South, 55410, Minneapolis, MN, 55455, United States of America, 

rbapna@umn.edu, Akhmed Umyarov 

In this paper, we present a novel randomized experiment that tests the existence 

of causal peer influence in the general population of a particular large-scale 

online social network. Both simple t-test and logistic regression indicate that 

user’s adoption of a product causes her online friends to pay for it and adopt it as 

well. 

3 - Exploring the Role of Cultural Differences in Crowd-funder 

Contribution Decisions 

Gordon Burtch, Temple University, 1810 N 13th Street, Speakman 

Hall 201F, Philadelphia, PA, 19122, United States of America, 

gburtch@temple.edu, Anindya Ghose, Sunil Wattal 

Contrary to Thomas Friedman’s Flat World hypothesis, here, we argue that the 

world is by no means flat, as distance continues to exist along dimensions other 

than the physical. Many online marketplaces today offer participants the 

opportunity to engage in economic exchange with others the world over. As a 

result, the differences between parties are often striking. Bearing this in mind, we 

explore the impact of cultural differences on lending on the world’s largest microfinance 

website, Kiva.org. Leveraging an aggregate dataset of country-to-country 

lending, incorporating three million lending actions over a five-year period, we 

find that cultural differences are negatively associated with lending and, 

interestingly, that this association is decreasing in physical distance (we interpret 

this as an awareness effect). We consider the implications of this finding for the 

purveyors of globalized online marketplaces, making recommendations regarding 

information presentation and the application of institutional trust mechanisms. 

■ SB72 

SB72 


OR Applications in Cloud Computing 



Chair: Emrah Zarifoglu, Optimization Scientist, IBM, 

1 Franklin Parkway, Building 910, San Mateo, CA, 94402, 

United States of America, emrah.zarifoglu@utexas.edu 

1 - Pricing Differentiated Services in Cloud Computing Market 

Linna Du, PhD Student, University of Connecticut, School of 

Business, 2100 Hillside Road Unit 1041, Storrs, CT, 06269, 

United States of America, linna.du@business.uconn.edu 

The cloud computing market is featured as a large scale market for a non-storable 

commodity with instantaneously released capacity, differentiated services, 

stochastic demands and risk-averse vendor. We employ queueing network 

analysis to capture the system unique features and long run behavior. Our model 

helps business managers develop actionable pricing strategies for increasing the 

revenue rate, controlling the user’s service adoption and hedging against 

variability in realized services. 

2 - Stochastic Scheduling for Cloud Computing 

Laurent Pauwels, The University of Sydney, Rm 481, 

Merewether Building, Sydney, Australia, 

laurent.pauwels@sydney.edu.au, Dmytro Matsypura 

We consider a scheduling problem of a software-as-a-service provider (cloud 

user). The provider has access to virtually unlimited computing resources on a 

pay-as-you-go basis. The jobs that need to be scheduled have uncertain sizes and 

completion times. We present stochastic program formulations with respect to 

various measures of optimality, investigate structural properties of optimal 

solutions and analyze worst-case performance of various heuristics. 

3 - An Analytical Approach to Solving Load Balancing Problem in 

Cloud Computing 

Emrah Zarifoglu, Optimization Scientist, IBM, 1 Franklin Parkway, 

Building 910, San Mateo, CA, 94402, United States of America, 

emrah.zarifoglu@utexas.edu, Ilyas Iyoob 

The dynamic cost structure increases the importance of intelligent provisioning 

and management in cloud computing. We develop an optimization model to 

decide high-mark utilization and low-mark utilization values to turn on and off 

VMs while minimizing provisioning and maintenance costs. Solving the same 

optimization problem at different points in time provides us with a dynamic VM 

load balancing policy. The robustness of the policy relies on the agility of 

periodical model corrections.

SB73 

■ SB73 


Market Microstructure and High-Frequency Trading 

Cluster: Quantitative Finance 


Chair: Gerry Tsoukalas, University of Pennsylvania, 

The Wharton School, Philadelphia, PA, United States of America, 

gtsouk@wharton.upenn.edu 

1 - General State Space Models in Electronic Market Making 

Vibhav Bukkapatanam, PhD Candidate, Stanford University, 

141O Huang Engineering Center, Stanford, CA, 94305, 

United States of America, vibhav@stanford.edu 

In recent years, the dramatic growth in electronic trading has revolutionized 

financial markets and led to the proliferation of high frequency market makers. In 

this work, we analyze how general state space models can be used in several key 

components of a maket making system, including spot variance and correlation 

estimation, regime switching modeling of high frequency data and optimal 

inventory control based bid-ask spreads estimation. 

2 - Predicting the Impact of a Proposed Securities 

Exchange Regulation 

Waraporn Tongprasit, Stanford University, Stanford, CA, United 

States of America, Waraporn@stanford.edu, Benjamin Van Roy 

We propose an approach to counterfactual analysis of securities exchanges that 

can be applied to predict the impact of proposed regulatory changes. We conduct 

an empirical case study in which we predict the impact of a historical tick size 

change on exchange efficiency and compare predictions to what can be inferred 

from subsequent data. 

3 - Disentangling Price Impact from Alpha 

Mehmet Saglam, Columbia University, New York, NY, 

United States of America, MSaglam13@gsb.columbia.edu, 

Ciamac Moallemi, Michael Sotiropoulos 

Motivated by the performance measurement literature in active portfolio 

management, we are interested in attributing the price changes observed during 

an execution of a large trade between the client’s short term predictive ability, 

alpha view, and the broker’s price impact. Using execution data with a large 

universe of clients, we show that incorporating client’s alpha view drastically 

increases the explanatory power of existing models and enables to accurately 

estimate the price impact. 

4 - Optimal Trade Execution using Limit Order Book Information 

Rolf Waeber, Cornell University, 206 Rhodes Hall, Ithaca, NY, 

14853, United States of America, rw339@cornell.edu, 

Sasha Stoikov 

We consider an asset liquidation problem at the market microstructure level, 

given observations of the limit order book. The optimization is formulated in 

terms of a sequence of stopping times, at which we submit market sell orders. We 

describe the shape of the trade and no trade regions for different price and latency 

assumptions. In the empirical section, we show that our policy signicantly 

outperforms a benchmark TWAP algorithm on US treasury bonds. 

Sunday, 1:30pm - 3:00pm 

■ SC01 

01- West 101- CC 

Optimization in Data Mining 

Sponsor: Optimization/Global Optimization 


Chair: Onur Seref, Assistant Professor, Virginia Tech, 1007 Pamplin 

Hall, Blacksburg, VA, 24061, United States of America, seref@vt.edu 

1 - How to Reverse-Engineer Quality Rankings 

Cynthia Rudin, Assistant Professor, Massachusetts Institute of 


United States of America, rudin@mit.edu, Michael Cavaretta, 

Gloria Chou, Robert Thomas, Allison Chang 

A good or bad product quality rating can make or break an organization. 

However, the notion of “quality” is often defined by an independent rating 

company that does not make the formula for ranking products public. We provide 

a machine learning approach for “reverse-engineering” a rating company’s 

proprietary model as closely as possible. 


104 

2 - A Probabilistic Model for Assessing the Mortality Risk in 

Post-operative Patients 

Dmytro Korenkevych, 372 Maguire Village, Apt. 1, Gainesville, FL, 

32603, United States of America, dmitriy@ufl.edu, Petar 

Momcilovic, Panos Pardalos 

We used a probabilistic model based on Bayesian learning and optimization 

techniques to estimate the mortality risk in post-operative patients. The model 

incorporates discrete and continuous risk factors and provides a probabilistic score 

associated with the risk. The feature selection procedure was applied in order to 

discard irrelevant risk factors and 70/30 cross-validation analysis was performed 

to estimate the accuracy of the model. 

3 - Information-Theoretic Learning for Stimulus-specific 

Clustering with fMRI 

Chun-An (Joe) Chou, University of Washington, 3900 Stevens 

Way, Seattle, WA, United States of America, joechou@uw.edu, 

Kittipat “Bot” Kampa, Art Chaovalitwongse 

Multi-Voxel Pattern Analysis (MVPA) is a popular approach for cognitive 

recognition with functional MRI (fMRI), as it is consistent with a neural system 

that mental operations/representations are instantiated in a neural population 

code spanning multiple voxels. In this work, we employ an unsupervised 

clustering with an information-theoretic measure to identify the spatial patterns 

of voxels in response to stimuli. The results illustrate that there are specific 

clusters for different stimuli. 

4 - Optimization Models for Predicting the Antigenic Variants of 

Influenza A/H3N2 Virus 

Serdar Karademir, University of Pittsburgh, 1048 Benedum Hall, 

Pittsburgh, United States of America, sek73@pitt.edu 

We use the pairwise amino acid sequence comparison of influenza strains and the 

antigenic distance between each pair as input for a classification model to identify 

immunodominant positions that cause antigenic variety. The performance of the 

model is evaluated through cross validation where cross validation is modeled as 

an optimization program that minimizes the misclassification error. Comparison 

to results of existing classification methods in literature is also provided. 

■ SC02 

02- West 102 A- CC 

DA Entrepreneurs Speak: Challenges and 

Opportunities in Building Decision Analysis 

Based Businesses 



Chair: Chris Dalton, CEO, Syncopation Software, 6 State Street, 

Suite 402, Bangor, ME, 04401, United States of America, 

cdalton@syncopation.com 

1 - Tales from the Decision Analysis Industrial Complex 

Chris Dalton, CEO, Syncopation Software, 6 State Street, 

Suite 402, Bangor, ME, 04401, United States of America, 

cdalton@syncopation.com 

This talk will define what it means to be a Decision Analysis based business, and 

give examples of successful and unsuccessful DA business models. At the end of 

the session, we will have a panel discussion with the speakers and other members 

of the DA business community. 

2 - Building Decision Analysis Based Businesses 

Carl Spetzler, CEO, Strategic Decisions Group, 745 Emerson Street, 

Palo Alto, CA, 94301, United States of America, cspetzler@sdg.com 

As an early member of the original Decision Analysis Group at SRI Internaional 

and then one of the founders of SDG, I have participated in and observed about a 

dozen of start-up companies based on the power of DA. In this session, I will 

share my observations from the successes and failures of these entrepreneurial 

efforts. 

3 - Business Analytics, Information Technology, and Decision 

Analysis 

Don Kleinmuntz, Exec VP & Cofounder, Strata Decision 

Technology, 200 E Randolph St 49th Floor, Chicago, IL, 60601, 

United States of America, dnk@strata-decision.com 

Business Analytics is a huge wave that has overtaken a number of industries and 

is guiding the development of IT infrastructure in many others. There is a genuine 

opportunity to embed decision analysis tools and processes in that infrastructure 

and create tremendous impact.

4 - Panel Discussion: Challenges and Opportunities for 

DA Businesses 

Moderator: Chris Dalton, CEO, Syncopation Software, 6 State 

Street, Suite 402, Bangor, ME, 04401, United States of America, 

cdalton@syncopation.com, Panelists: Carl Spetzler, 

Don Kleinmuntz, Terry Bresnick, Jack Kloeber 

The speakers and other members of the DA business community will discuss the 

present business environment, taking questions from the audience as time allows. 

■ SC03 

03- West 102 B- CC 

Decision Models in Public Health 



Chair: Ozgur Araz, Assistant Professor, University of Nebraska Medical 

Center, Nebraska Medical Center, Omaha, NE, 68198-4365, 

United States of America, ozgur.araz@unmc.edu 

1 - Multiplicative Utilities for Health and Consumption 

Sam Bodily, John Tyler Professor of Business Administration, 

University of Virginia, Darden Business School, 100 Darden 

Boulevard, Charlottesville, VA, 22903, United States of America, 

BodilyS@virginia.edu, Casey Lichtendahl 

Two new multiplicative utilities offer correlation aversion and other advantages 

over an additive. They extend the QALY model. For constant health and 

consumption streams, they reduce to a double exponential utility in life duration, 

which can have risk aversion, decreasing absolute risk aversion, increasing 

relative risk aversion, as well as standard gamble probabilities (proportional time 

tradeoffs) that increase (decrease) in life duration and that exceed proportional 

time tradeoffs. 

2 - Determining Optimal Therapy for Diabetic Patients. Evidence 

from the U.S. Department of Veteran Affairs 

Vishal Ahuja, PhD Candidate, University of Chicago, Booth 

Business School, Chicago, IL, 60637, United States of America, 

vahuja@chicagobooth.edu, John Birge 

We investigate various treatments for Type 2 Diabetes in adults using data from 

the Veterans Health Administration. We analyze dynamic patterns of glucose 

lowering therapies, and evaluate the risks of associated diabetes outcomes. We 

propose a predictive model to evaluate effectiveness of various glucose-lowering 

therapies and recommend optimal treatment based on patient characteristics. The 

paper addresses confounding measures such as physician bias and unobservable 

patient characteristics. 

3 - Robust Design of a Cost-effective Diabetic Retinopathy 

Screening Program with Design of Experiments 

Irene Vidyanti, University of Southern California, Los Angeles, CA, 

United States of America, irenevidyanti@gmail.com, Shinyi Wu 

This paper demonstrates the use of Design of Experiments (DoE) coupled with 

simulation to design a robust and cost-effective public health intervention 

systematically, using the design of a Diabetic Retinopathy screening program as 

illustration. Optimum input factor settings are determined to find settings that 

achieve cost-effectiveness. Then, robust parameter design is used to find settings 

for a robust program by treating uncertain variables - such as screening 

compliance - as noise factors. 

4 - Aversion to Health Inequalities in Healthcare Prioritisation: 

A Multiobjective Perspective 

Alec Morton, London School of Economics and Political Science, 

Houghton Street, London, WC2A 2AE, United Kingdom, 

a.morton@lse.ac.uk 

In this paper we discuss the prioritisation of healthcare projects where there is a 

concern about health inequalities. Our analysis begins with a standard welfare 

economic model of healthcare resource allocation and we show how the problem 

can be reformulated as one of finding a particular subset of the class of efficient 

solutions to an implied multiobjective optimization problem. We demonstrate our 

approach through a worked example of treatment for clinical depression. 


105 

■ SC04 

04- West 102 C- CC 

Decision Makers 


SC04 

Chair: Rayan Hafiz, Engineering Specialist, Saudi Arabian Oil Company, 

Aramco, Dhahran, Saudi Arabia, rayan.hafiz@aramco.com 

1 - Customer Experience Optimization: Regaining Control 

of the Conversation 

Theresa Igharo, GPJ @ IBM, 11301 Burnet Rd., Austin, 

United States of America, igharot@us.ibm.com 

Today’s consumers have higher expectations of personalization while leaving 

tremendous volumes of seemingly unrelated digital ‘fingerprints’ across many 

locations. Organizations have been relying on ad-hoc, inefficient and 

disconnected set of processes to make crucial decisions; with often, no means to 

understand the impact of those decisions. Business analytics accessible to all 

decisions makers increases operational agility while providing organizations with 

actionable customer insights. 

2 - Deciding by Feeling or Calculation 

Mark Schneider, University of Connecticut, Uconn School of 

Business, 81 Cheney Drive, Storrs, CT, United States of America, 

mschneider@business.uconn.edu, Robin Coulter 

Affective and cognitive processes are widely recognized to play an important role 

in decision making. However, it remains unclear whether general features of a 

decision problem can systematically cue one set of processes over another. In this 

paper, we introduce and evaluate a mechanism for predicting whether people will 

rely on feeling or calculation in a decision task. 

3 - Optimizing Portfolios when Dealing with Ranges 

Siddhartha Sampath, Arizona State University, 849 W Elna Rae, 

Tempe, AZ, United States of America, 

Siddhartha.Sampath@asu.edu, Esma Gel, John Fowler, Karl Kempf 

We attempt to discuss the product portfolio selection process and demonstrate 

how to solve a non-linear knapsack problem when various products have their 

values interdependent on each other. When each value of benefit and cost is also 

a range the problem is made even harder. We discuss various strategies to address 

this problem. 

4 - Decoy Effects in Ethical Decision Making 

Wenjie Tang, Assistant Professor, IE Business School, Calle de 

Maria de Molina, 12, Piso 5, Madrid, Spain, wenjie.tang@ie.edu, 

Steffen Keck 

We study decoy effects in ethical decision making. Decoys are alternatives whose 

presence increases preference for another alternative in the choice set, though 

themselves rarely chosen. We consider a scenario where one is facing a choice 

between a monetary more beneficial alternative and a more ethical one. We 

propose that the presence of decoys make one more likely to choose the ethical 

alternative. We believe that such method can be used to induce ethical choice in 

various contexts. 

5 - Decision Analysis at Work: Making the Case for Strategic 

Technology Selection 

Rayan Hafiz, Engineering Specialist, Saudi Arabian Oil Company, 

Aramco, Dhahran, Saudi Arabia, rayan.hafiz@aramco.com 

In this work, we develop a decision analysis model to help the upper 

management make the strategic decision for an emerging technology’s 

companywide adoption. Such modeling will help understanding the impact of the 

decisions related to dealing with the targeted technology and will set the 

corporate direction dealing with it for the next 10 years.

SC05 

■ SC05 

05- West 103 A- CC 

Condition-based Maintenance Optimization 



Chair: Hao Peng, Assistant Professor, Eindhoven University of 

Technology, Den Dolech 2, Eindhoven, 5612WN, Netherlands, 

h.peng@tue.nl 

1 - A Condition-Based Maintenance Policy for Multi-Component 

Systems with High Setup Cost of Maintenance 

Qiushi Zhu, Ir.MSc, TU Eindhoven, De Lismortel, Paviljoen E.9a, 

Eindhoven, 5612 AR, Netherlands, q.zhu@tue.nl, 

Geert-Jan van Houtum, Hao Peng 

We propose a new CBM policy for multi-component systems with stochastically 

and continuously deterioration. To reduce the setup cost of maintenance, a joint 

maintenance interval is proposed. The optimal maintenance interval and control 

limits of components are determined by minimizing the average long-run cost 

rate related to maintenance and downtime. Moreover, a numerical study with a 

case of wind farm maintenance is performed based on a large number of nonidentical 

components. 

2 - Health Condition Prediction Accuracy Improvement and 

Condition Based Maintenance Optimization 

Zhigang Tian, Concordia University, 1515 Ste-Catherine Street 

West, Montreal, Canada, tian@ciise.concordia.ca, Bairong Wu 

Health condition prediction is a key part for condition based maintenance (CBM). 

It is observed that the prediction accuracy improves with the increase of the age 

of the component as it approaches the failure time in many applications. In this 

paper, we model the relationship between the prediction accuracy and the age of 

a component relative to its predicted failure time. A numerical method is 

presented to evaluate the cost of the CBM policy. 

3 - Condition-Based Maintenance for Degrading Systems under 

Dynamic Environments 

Xiao Liu, liuxiaodnn_1@hotmail.com, Khalifa N. Al-Khalif, 

Abdel Magid Hamouda, Dingguo Hua, David Coit, Elsayed Elsayed 

This talk presents an optimum condition-based maintenance policy for degrading 

systems under dynamic environments. The degradation is described by a Wiener 

process with both degradation rate and diffusion being influenced by system age 

and environment condition. We obtain the distribution of the remaining system 

life, based on which the dynamic CBM decisions are made. 

4 - Statistical Updating of Finite Element Model with Lamb Wave 

Sensing Data for Structural Damage Detection 

Arda Vanli, College of Engineering, Florida State University, 2525 

Pottsdamer St., Tallahassee, FL, 32310, United States of America, 

avanli@fsu.edu, Sungmoon Jung 

This study proposes a new probabilistic model updating method used to improve 

the prediction capability of a finite element analysis (FEA) model of structural 

damages with experimental observations from a sensor network. A controlled 

damage experiment on a composite panel showed that the proposed model 

updating approach achieves a prediction accuracy that is superior to existing 

calibration approaches. 

■ SC06 

06- West 103 B- CC 

Advances in Simulation Optimization 



Chair: Jie Xu, Assistant Professor, George Mason University, 4400 

University Dr, MS 4A6, GMU Engr Bldg RM2100, Fairfax, VA, 22030, 

United States of America, jxu13@gmu.edu 

1 - Integer-ordered Simulation Optimization using R-SPLINE 

Bruce Schmeiser, Professor, Purdue University, West Lafayette, IN, 

47905, United States of America, bruce@purdue.edu, 

Honggang Wang, Raghu Pasupathy 

We consider simulation-optimization problems where the decision variables are 

integer ordered. We develop R-SPLINE—-an algorithm that alternates between a 

continuous search and a discrete neighborhood search, to asymptotically identify 

a local minimum. R-SPLINE converges almost surely, with exponentially decaying 

likelihood of returning an incorrect solution. With no parameter tuning, 

R-SPLINE compares favorably with existing algorithms. 


106 

2 - Rare Event Simulation for Processes Generated via Stochastic 

Fixed Point Equations 

Jeffrey Collamore, Associate Professor, University of Copenhagen, 

Department of Mathematical Sciences, Universitetsparken 5, 

Copenhagen, DK-2100, Denmark, collamore@math.ku.dk, 

Anand Vidyashankar, Guoqing Diao 

Stochastic fixed point equations (SFPEs) arise in a wide variety of applications in, 

e.g., insurance, finance, and the analysis of algorithms. A general SFPE has the 

form V =_d f(V) (=_d denotes equality in distribution), where the objective is very 

often to infer the tail probabilities of V. We present a general importance sampling 

algorithm for computing these rare event probabilities, establishing efficiency, 

optimality, and sharp estimates for the running time. 

3 - The Gradient Oriented Polar Random Search 

Loo Hay Lee, Associate Professor, National University of Singapore, 

10 Kent Ridge Crescent, Singapore, Singapore, 

iseleelh@nus.edu.sg, Haobin Li, Ek Peng Chew 

Involving gradient information in a random search is believed to have better 

performance than both gradient-based and metaheuristic local search. However, 

the main difficulty is to incorporate and control randomness in a direction instead 

of a point. This paper makes use of generalized polar coordinates and 

corresponding random distributions, so as to design a brand new Gradient 

Oriented Polar Random Search (GO-POLARS) that is proved to satisfy the 

conditions for strong local convergence. 

4 - Gradient-based Adaptive Stochastic Search for 

Discrete Optimization 

Enlu Zhou, Assistant Professor, University of Illinois at Urbana- 

Champaign, University of Illinois, Urbana, 61801, 

United States of America, enluzhou@illinois.edu, Xi Chen 

We propose a stochastic search algorithm to solve discrete optimization problems 

by iteratively generating candidate solutions from a sequence of parameterized 

sampling distributions over the solution space. We convert the discrete 

optimization problem to an equivalent continuous problem on the parameter 

space so that a gradient method can be incorporated to update the parameter of 

the sampling distribution. We show the convergence of the algorithm via tools 

from stochastic approximation. 

■ SC07 

07- West 104 A- CC 

Joint Session Data Mining/Computational Stochastic 

Optimization: Towards Statistical Parsimony in 

Adaptive Dynamic Programming 

Sponsor: Data Mining & Computational Stochastic Optimization 


Chair: Victoria Chen, University of Texas at Arlington, Department Ind. 

& Manuf. Sys. Engr., Arlington, TX, United States of America, 

vchen@uta.edu 

1 - Adaptive Value Function Approximation for Continuous-State 

Stochastic Dynamic Programming 

Prashant Tarun, Assistant Professor, Craig School of Business, 

Missouri Western State University, St. Joseph, MO, 64507, United 

States of America, ptarun@missouriwestern.edu, Huiyuan Fan, 

Victoria Chen 

This study presents a sequential method of value function approximation in the 

high-dimensional, continuous-state dynamic programming framework. This 

method can adaptively determine both sample size and model structure. 

Furthermore, we introduce two new algorithms that help implement the adaptive 

value function approximation method efficiently. 

2 - Deterministic Sampling Designs with Local Methods for 

Optimization Problems 

Danilo Macciò, National Research Council of Italy, Via De Marini 

6, Genova, 16149, Italy, ddmach@ge.issia.cnr.it, Victoria Chen, 

Cristiano Cervellera, Diana Martinez 

The problem of numerically minimizing a functional cost involves the following 

issues: (i) the definition of a sampling of the domain where the functional is 

evaluated; (ii) the choice of a class of models to approximate the solution. This 

work presents a comparison of performances among several deterministic 

sampling designs (low-discrepancy sequences, orthogonal arrays and Latin 

hypercubes) in case local models are employed, in different functional 

optimization contexts.

3 - Sequential MARS for Optimization Problems 

Diana Martinez, The University of Texas at Arlington, 415 Summit 

Ave., Apt # 213, Arlington, TX, 76013, United States of America, 

diana.martinezcepeda@mavs.uta.edu, Victoria Chen, 

Jay Rosenberger 

To find a solution for optimization problems, in some cases an iterative method to 

sequentially reduce the solution space is required. It is possible then to have 

different design configurations for testing that involve the estimation of a 

response value. MARS modeling technique has been used as a surrogate method 

in such cases. An algorithm that proposes different MARS sequential routines has 

been developed and tested. 

4 - Global Optimization for a Piecewise Linear Regression 

Spline Function 

Nadia Martinez, The University of Texas at Arlington, 415 Summit 

Ave. Apt. No. 213, Arlington, TX, 76013, United States of America, 

nadia.martinezcepeda@mavs.uta.edu, Jay Rosenberger, 

Victoria Chen, Diana Martinez 

In this research, we optimize a piecewise linear version of a Multivariate Adaptive 

Regression Spline (MARS) function subject to constraints that include both linear 

regression models and piecewise linear MARS models. A mixed integer 

programming (MIP) model is developed and solved using branch-and-bound. The 

MIP is adjusted to enable flexibility to handle both continuous and categorical 

variables. This method is demonstrated on designing an automotive safety system 

for a major US automaker. 

■ SC08 

08- West 104 B- CC 

Joint Session Doing Good with Good OR/SPPSN: 

Doing Good with Good OR Competition: Finalist 

Presentations II 

Cluster: Doing Good with Good OR Student Competition & Public 

Programs, Service and Needs 


Chair: Susan Martonosi, Harvey Mudd College, Department of 

Mathematics, Claremont, CA, United States of America, 

martonosi@hmc.edu 

1 - Resource-based Patient Prioritization in Mass-casualty 

Incidents 

Alex Mills, Indiana University Kelley School of Business, 1309 E. 

Tenth Street, Bloomington, IN, 47405, United States of America, 

millsaf@indiana.edu, Nilay Argon, Serhan Ziya 

Mass-casualty triage currently relies on a fixed priority ordering among different 

classes of patients and does not explicitly consider resource limitations. The 

nominated study separates medical classification from prioritization and 

incorporates survival probabilities and resource limitations. Based on analytical 

and simulation results, additional lives can be saved by using the proposed 

policies to support decisions made by medical providers. 

2 - Improving Patient Access to Surgical Care at the Juravinski 

Hospital through Informed Decision Making 

Daphne Sniekers, PhD Candidate, University of Toronto, 

5 King’s College Road, Toronto, ON, M5S 3G8, Canada, 

daphne.sniekers@mail.utoronto.ca 

A generic simulation-based perioperative decision support tool was implemented 

at the Hamilton Health Sciences Juravinski Hospital to demonstrate how changes 

to the operating room schedule and resource availability affect patient flow. The 

tool was used in negotiations with the various stakeholder groups in order to 

come to consensus on changes to implement. The tool provided invaluable 

support in their decision-making. 

3 - using Simulation Methods to Guide Decision Making of the 

Kentucky Cabinet for Health and Family Services 

Russell Harpring, University of Louisville, Louisville, KY, 

United States of America, russ.harpring@gmail.com 

The purpose of this study is to find potential improvements to the intake process 

at the Kentucky Cabinet for Health and Family Services (KCHFS) main office in 

Louisville using simulation software. KCHFS deals with hundreds of clients every 

day, with each client in the system having different needs. By simulating the 

process with Arena Software, experiments can be run with real-time data and 

solutions found to complex problems involving process flow, staffing, and layout. 


107 

■ SC09 

09- West 105 A- CC 

Evolutionary Multi-Criterion Optimization Tutorial 



Chair: Kalyanmoy Deb, Professor, Indian Institute of Technology 

Kanpur, Department of Mechanical Engineering, Kanpur, UP, 208016, 


1 - Statistical Stopping Criteria for a Multiobjective Genetic 

Algorithm 

Natalia Viktorovna, Analytics Consultant, SAS Institute, 100 Sas 

Campus Drive, T6059, Cary, NC, 27513, United States of America, 

natalia.viktorovna@sas.com, Reha Uzsoy, Juan Gaytan 

In this research we explore the data structure at iterations of a genetic algorithm. 

Main objective is the generation of a statistical stopping criterion. We fit a 

multivariate distribution to objective function values and estimate the probability 

of creating new non-dominated solutions. This probability is used as a stopping 

criterion in a modified NSGA II aimed at solving the Multicriteria Project 

Scheduling Problem. Computational experimentation was developed to test the 

criteria performance. 

2 - A Tutorial on Evolutionary Multi-criterion Optimization (EMO) 

Kalyanmoy Deb, Professor, Indian Institute of Technology Kanpur, 

Department of Mechanical Engineering, Kanpur, UP, 208016, 


For the past two decades, multi-criterion optimization problems have been 

increasingly being solved using evolutionary optimization methods, simply due to 

their parallel search and ability to find multiple trade-off solutions in a single 

simulation. In this tutorial, we shall present principles of EMO, results on a 

number of challenging problems and highlight some recent research challenges. 

■ SC10 

SC10 

10- West 105 B- CC 

Probabilistic Constrained Problems with Continuous 

Random Variables 



Chair: Andras Prekopa, Professor, RUTCOR, Rutgers University, 

640 Bartholomew Road, Piscataway, 08854, United States of America, 

prekopa@rutcor.rutgers.edu 

1 - Convex Approximations in Stochastic Programming by 

Semidefinite Programming 

Istvan Deak, Professor, Department of Computer Science, Corvinus 

University of Budapest, Hungary, istvan.deak@uni-corvinus.hu, 

Imre Polik, Andras Prekopa, Tamás Terlaky 

The following question arises in stochastic programming: how can one 

approximate a noisy convex function with a convex quadratic function that is is 

optimal in some sense. Using several approaches for constructing convex 

approximations we present some optimization models yielding convex quadratic 

regressions that are optimal approximations in L_1, L_2 and L_infinity norm. 

Extensive numerical experiments to investigate the behavior of the proposed 

methods are also performed. 

2 - Serially Linked Water Reservoir Network Design 

Olga Myndyuk, RUTCOR, Rutgers University, 

United States of America, olik.myn@gmail.com 

In this paper, we find optimal reservoir capacities such that at least k consecutive 

periods the demands are met in the course of a given number of periods with a 

probability which is greater than or equal to the prescribed probability. 

Formulation, mathematical properties and the solution methodology of the model 

are given. Numerical example for 2 reservoirs with 8 consecutive out of 24 

periods is presented for the model with demand and inflow being normally 

distributed random variables. 

3 - Solution of a Probabilistic Constrained Stochastic Programming 

Problem with Random Technology Matrix 

Andras Prekopa, Professor, RUTCOR, Rutgers University, 640 

Bartholomew Road, Piscataway, 08854, United States of America, 

prekopa@rutcor.rutgers.edu 

We formulate and solve a probabilistic constrained problem, where some of the 

technology coefficients are normally distributed random variables. The method of 

solution combines the supporting hyperplane method and one, originally 

designed to solve problems with discrete random variables. Closed form formulas 

for the gradient and numerical results will be presented.

SC11 

4 - Simultaneous Confidence Intervals for Future Values of a 

Stochastic Process 

Jinwook Lee, RUTCOR, Rutgers University, 640 Bartholomew 

Road, Piscataway, NJ, 08854, United States of America, 

jinwook@rci.rutgers.edu, Andras Prekopa 

Given a stochastic process with known finite dimensional distributions, we 

construct lower and upper bounds for the probability that future values of the 

stochastic process run within prescribed limits. For the solution of the problem we 

use the technique of the univariate and multivariate discrete moment problem. 

Applications and numerical example will be presented. 

■ SC11 

11- West 105 C- CC 

Application of Stochastic Programming II 


Chair: Aaron L. Nsakanda, Associate Professor, Carleton University, 

1125 Colonel By Drive, Ottawa, ON, K1S5B6, Canada, 

aaron_nsakanda@carleton.ca 

1 - Application of Multistage Stochastic Programming Modeling 

and Solution Method on Capacity Planning 

George Kolomvos, European Dynamics S.A, 209, Kifisias Av., 

Athens, Greece, kolomvos@gmail.com, Georgios Saharidis 

We address the problem of capacity planning in a gas distribution network under 

uncertain demand and prices. We consider a large number of scenarios claiming 

to describe as good as possible the space of demand and prices for gas. We model 

the problem as a multistage stochastic problem, where the decision on the first 

stage variables - capacities - will have a critical impact for all future stages. The 

problem is solved using the nested decomposition method. 

2 - Solving the Drift Control Problem 

Melda Ormeci Matoglu, Ozyegin University, Kusbakisi Cad No 2, 

Altunizade, Istanbul, Turkey, melda.ormeci@ozyegin.edu.tr, 

John Vande Vate 

We model managing capacity as a Brownian drift control problem. We formulate 

an LP and develop a bound for the continuous problem from a dual solution to 

the discrete problem. Showing the equivalence between strongly feasible bases 

and deterministic unichain policies, we combinatorialize the pivoting process and 

solve the LP without computing its coefficients. We develop scheme analogous to 

column generation to drive the gap between the discrete approximation and the 

continuous problem to zero. 

3 - Time Consistency and Risk Averse Dynamic Decision Models: 

Interpretation and Practical Consequences 

Davi Valladão, IBM Research/PUC-Rio, Av. Pasteur, 138/146 

Botafogo, Rio de Janeiro, Brazil, davimv@br.ibm.com, 

Alexandre Street, Birgit Rudloff 

We discuss a CVaR based portfolio selection problem and compare a time 

consistent formulation to an inconsistent one. We develop a new way of 

measuring the impact of an inconsistent policy on the objective function. We 

present a sensitivity analysis by computing this impact for different planning 

horizons and risk aversion levels. Finally, we develop a suitable economic 

interpretation for its recursive objective function based on the certainty 

equivalent of the related preference function. 

4 - Contracts with Options in Loyalty Reward Programs Supply 

Chain 

Aaron L. Nsakanda, Associate Professor, Carleton University, 

1125 Colonel By Drive, Ottawa, ON, K1S5B6, Canada, 

aaron_nsakanda@carleton.ca, Yuheng CAO, Moustapha Diaby 

We explore the issue of whether an option mechanism is a viable alternative to 

help hedge against demand uncertainties in a Loyalty Reward Programs 

enterprise-led supply chain. A two-stage stochastic linear programming with 

simple recourse model is developed as well as a solution procedure based on the 

sampling average approximation scheme. Our preliminary numerical result will 

be reported. 

5 - Optimal Control of Production-inventory Systems with 

Multiple Facilities 

Mohsen Elhafsi, Professsor, University of California, School of 

Business Administration, Riverside, CA, 92521, 

United States of America, mohsen.elhafsi@ucr.edu, Xi Chen 

We study a single-product production-inventory make-to-stock system with 

identical production facilities with no order cancellations. Demand is backlogged 

if not satisfied immediately. The objective is to control the number of production 

facilities to be used at as well as the inventory level so as to minimize the total 

production, holding and backlog costs. We show that the optimal policy is defined 

by a set of inventory levels and a set of facilities activate-up-to levels for facilities. 


108 

■ SC12 

12- West 106 A- CC 

Approximate Dynamic Programming and Discrete 

Optimization 



Chair: Alejandro Toriello, Assistant Professor, Industrial and Systems 

Engineering, University of Southern California, 3715 McClintock Ave, 

GER 240, Los Angeles, CA, 90089, United States of America, 

toriello@usc.edu 

1 - Patient Admission and Scheduling under Multiple 

Resource Constraints 

Christiane Barz, Assistant Professor, UCLA Anderson School of 

Management, 110 Westwood Plaza, Los Angeles, United States of 

America, christiane.barz@anderson.ucla.edu, Kumar Rajaram 

We consider a patient admission problem to a hospital with multiple potential 

bottlenecks (e.g. beds and OR). While there is an uncontrolled arrival process of 

emergency patients, the hospital has the freedom to postpone or even reject the 

admission of non-emergency patients. We model the control process as a Markov 

Decision Process and suggest price-directed heuristics. 

2 - Dynamic Software Release Management 

Angelo Mancini, PhD. Student, University of Chicago Booth 

School of Business, 5807 South Woodlawn Ave., Chicago, United 

States of America, amancini@chicagobooth.edu, Dan Adelman 

Software release management is the process by which firms decide on the timing, 

composition and development schedule of their future releases. We construct a 

semi-Markov decision process (SMDP) that models these decisions in the 

presence of a stochastically improving competitor. We then formulate the static 

problem as a mixed integer nonlinear program, whose parameters we obtain by 

making a value function approximation in our SMDP. We report numerical 

results. 

3 - Computing Bounds and Solutions to the Traveling Salesman 

Problem with Approximate Linear Programming 

Michael Poremba, PhD. Student, Industrial and Systems 

Engineering, University of Southern California, 3715 McClintock 

Ave., GER 240, Los Angeles, CA, 90089, United States of America, 

poremba@usc.edu, Alejandro Toriello 

We study the asymmetric traveling salesman problem (TSP) using approximate 

linear programming (ALP) to restrict the dynamic programming formulation. The 

second author has previously applied ALP to the TSP to obtain a nested family of 

polyhedral lower bounds. We discuss two computational methods for solving 

these formulations, constraint generation and a primal-dual procedure. We also 

present a price-directed tour generation algorithm that uses ALP solutions as 

approximate costs-to-go. 

4 - A Dynamic Traveling Salesman Problem with Stochastic 

Arc Costs 

Alejandro Toriello, Assistant Professor, Industrial and Systems 

Engineering, University of Southern California, 3715 McClintock 

Ave., GER 240, Los Angeles, CA, 90089, United States of America, 

toriello@usc.edu, William Haskell, Michael Poremba 

We propose a dynamic TSP with stochastic arc costs and use approximate linear 

programming to obtain a semi-infinite linear programming lower bound. The 

bound requires only expected costs and the support set, and is valid for any 

distribution with these parameters. Though NP-hard for general support sets, we 

show that the bound can be solved efficiently for polytopes with polynomially 

many extreme points and hyper-rectangles. We analyze the price-directed policy 

and derive performance guarantees.

■ SC13 

13- West 106 B- CC 

Mathematical Programming and Applications 



Chair: Hande Benson, Associate Professor, Drexel University, 

Department of Decision Sciences, 3141 Chestnut Street, Philadelphia, 

PA, 19104, United States of America, hvb22@drexel.edu 

1 - High-contrast Imaging via Fast Fourier Optimization 

Robert Vanderbei, Professor, Princeton University, Operations 

Research & Financial Engineer, 209 Sherrerd Hall, Princeton, NJ, 

08544, United States of America, rvdb@princeton.edu 

Many interesting and fundamentally practical optimization problems involve 

constraints on the Fourier transform of a function. Because the fast Fourier 

transform is recursive, it is difficult to embed into a linear optimization problem. 

However, one can adapt the main idea behind the fast Fourier transform so as to 

make it encodable as constraints in an optimization problem. We demonstrate on 

a real-world problem from the field of high-contrast imaging. 

2 - Incorporation of Delivery Times in Gamma Knife(R) 

Perfexion(TM) Treatment Optimization 

Hamid Ghaffari, PhD Candidate, University of Toronto, 5 King’s 

College Rd, MIE Department, Toronto, ON, M5S 3G8, Canada, 

ghaffari@mie.utoronto.ca, Mark Ruschin, Dionne Aleman, 

David Jaffray 

Although the use of mathematical optimization techniques can greatly improve 

the quality of treatment plans in various radiation therapy treatment settings, one 

complication is the potentially clinically unrealistic nature of optimized 

treatments. The difficulty arises from two factors: (1) machine limitations that 

govern the minimum amount of radiation delivery time, and (2) long treatment 

times due to the complexity of optimized treatments. In the first scenario, if a 

particular configuration of the radiation delivery device is used, then it typically 

must deliver radiation for a minimum length of time. Incorporation of such 

requirements in a mathematical model generally requires additional constraints 

and binary variables, increasing the difficulty of the optimization. In the second 

scenario, mathematically optimized treatments commonly assign (small amounts 

of) radiation to be delivered from many configurations, drastically increasing the 

time needed to deliver the treatment (beam-on time). We examine these two 

issues within the penalty-based sector-duration optimization model for Gamma 

Knife(R) Perfexion(TM) (Elekta, Stockholm, Sweden) and the combined sectorduration 

and isocentre optimization model to reduce beam-on time and to ensure 

that machine limitations regarding delivery times are met. 

3 - Portfolio Optimization with Cone Constraints and 

Discrete Decisions 

Umit Saglam, Drexel University, LeBow College of Business, 

Philadelphia, 19104, United States of America, us26@drexel.edu 

We consider a portfolio optimization problem where the investor’s objective is to 

choose a trading strategy that maximizes expected return penalized by transaction 

costs. We include portfolio diversification constraints in our single and 

multiperiod models. The overall problem is a mixed-integer second-order cone 

programming problem, which we solve with the Matlab-based solver MILANO. 

This talk will focus on the solution and warm-start of the second-order cone 

programming subproblems. 

4 - MILANO: Mixed-Integer Linear and Nonlinear Optimizer 

Hande Benson, Associate Professor, Drexel University, Department 

of Decision Sciences, 3141 Chestnut Street, Philadelphia, PA, 

19104, United States of America, hvb22@drexel.edu 

We will present the primal-dual penalty interior-point method that is 

implemented in MILANO for providing warmstart capabilities and robustness in a 

linear and mixed-integer linear programming framework. Extensions to cone 

programs and nonlinear programming problems will be explored. 

■ SC14 

14- West 106 C- CC 

Applications of Project Management 


Chair: Hiroko Nakamura, Project Researcher, The University of Tokyo, 

7-3-1 Hongo Bunkyoku, Tokyo, Japan, techhn@mail.ecc.u-tokyo.ac.jp 

1 - Change’s Timing and Cost Discrepancies in Public 

Project Procurement 

Jane Park, The George Washington University, 2201 G Street NW 

Duques Hall, Washington, DC, 20052, United States of America, 

janepark@gwu.edu 

The purpose of this study is to help manage changes better in public project 

procurement. Using the data from transportation projects in Florida, this study 


109 

classifies the reasons of change order and investigates how differently each reason 

affect the cost discrepancy between the original contract amount and the final 

payment depending on its timing of occurrence over the project duration. 

2 - A Study to Identify Key Factors Influencing Cost Overruns in 

Medium-scale Infrastructural and Social Projects 

Mohamed Salem, Senior Civil Engineer Specialist, Islamic 

Development Bank-Special Assistance Division, P.O. Box: 5925, 

Jeddah, 21432, Saudi Arabia, msalm@isdb.org, Alaa Elwany, 

Hamdy Elwany 

The accuracy of cost estimation for infrastructural and social projects is one of the 

key challenges that faces governments and funding organizations. Cost estimates 

during feasibility studies involve many assumptions and inherent uncertainties. 

The study analyzed the key factors that contribute to cost escalation. In particular, 

the study used multiple regression analysis to develop an empirical formula for 

accurately predicting cost overrun/ underrun during the project implementation. 

The study is based on analyzing data from 110 real-world projects implemented 

over the past decades. 

3 - Solving the Resource Investment Problem with Time Lags 

(RIP/max) via Mixed Integer Programming 

Juan D. Palacio, Instructor, Universidad de los Andes, 

Departamento de Ingenieria Industrial, Carrera 1 Este # 19 A-40, 

Bogota, Colombia, jd.palacio34@uniandes.edu.co, 

Sebastian Brunal, Andrés L. Medaglia 

The RIP/max is an NP-Hard problem that has been solved mainly using heuristics. 

We tackle this problem using a mixed integer programming model with proven 

success on variants of the project selection and resource constrained scheduling 

problems. We found new best known solutions and optimality certificates for 

instances of up to 30 projects. We also report new lower bounds and show how 

this formulation adapts to the resource investment problem (RIP). 

4 - Survey and Evaluation of Bibliometrics Methods for Research 

Planning and Technology Management 

Hiroko Nakamura, Project Researcher, The University of Tokyo, 

7-3-1 Hongo Bunkyoku, Tokyo, Japan, techhn@mail.ecc.utokyo.ac.jp, 

Yuya Kajikawa, Shinji Suzuki 

Managers and researchers of research institutes and industries were surveyed to 

determine the needs and expectation of bibliomentrics development for research 

planning and technology management. Based on actual problems, patent analysis 

methods to produce technology overviews of an industry and detection of 

emerging technology and of technology links between automobile and aviation 

industries were studied and evaluated. 

■ SC15 

SC15 

15- West 202- CC 



1 - GAMS Development Corporation – Deploying Your Application 

Built Around GAMS 

Michael Bussieck, GAMS Development, 1217 Potomac Street NW, 

Washington, DC, 20007, United States of America, 

mbussieck@gams.com, Lutz Westermann 

We will demonstrate the development of an application built around GAMS and 

its effortless deployment. Furthermore, we will show recent enhancements to the 

GAMS system. 

2 -Salford Systems - Real-World Data Analysis 

Mikhail Golovnya, Senior Scientist, Salford Systems, 9685 Via 

Excelencia, Ste. 208, San Diego CA 92126, United States of 

America, dsprague@schwartzmsl.com 

This demo, intended for the modeler wanting to apply data mining methodology, 

will emphasize real-world data analysis. The main concepts behind well-known 

data mining algorithms will be discussed: CART®, MARS® automated non-linear 

regression, TreeNet® stochastic gradient boosting, RandomForests®, GPS 

Generalized Pathseeker, ISLE Importance Sampled Learning Ensembles and the 

Rulefit Rule Extraction Engine. We will discuss what is novel in the software, 

cover implementation, and show where the software fits in terms of other data 

mining software methods.

SC16 

■ SC16 

16- West 207- CC 

Humanitarian Logistics 



Chair: Ozlem Ergun, Associate Professor, Georgia Institute of 

Technology, ndustrial and Systems Engineering, Atlanta, GA, United 

States of America 

1 - Humanitarian Logistics 

Ozlem Ergun, Georgia Institute of Technology, 765 Ferst Dr NW, 

Atlanta, GA, United States of America, oergun@isye.gatech.edu, 

Pinar Keskinocak, Julie L. Swann 

The increase in natural and man-made disasters in the recent years has 

highlighted challenging problems in humanitarian operations, in addition to 

major long-term humanitarian development challenges. Humanitarian logistics 

plays an important role in preparing for, responding to, and recovering from 

sudden-onset disasters and addressing long-term development issues. The 

management of humanitarian logistics operations involves many challenges such 

as conflicting objectives from multiple stakeholders, coordination and 

collaboration, high uncertainty, and scarcity of resources. In this tutorial, we 

provide an introduction to humanitarian logistics and its main application areas 

while outlining current research trends and the major challenges faced in the area 

today. We also present four decision aid tools addressing typical applications of 

humanitarian logistics: network design for donated human breast milk delivery in 

South Africa, port simulation and shipment scheduling for a large international 

agency, demand estimation and emergency procurement for CARE, and postdisaster 

medical response under accessibility issues. In presenting these decision 

aid tools, we aim to underline the differences in issues arising from sudden-onset 

disasters and long-term development problems, various stages of the life cycle of 

disasters, and different decision levels, while also pointing to general aspects in 

modeling humanitarian logistics problems. 

■ SC17 

17- West 208 B- CC 

SpORts II 

Sponsor: SpORts 


Chair: Dan Finkel, Technical Staff, Massachusetts Insitute of 

Technology, Lincoln Laboratory, 244 Wood Street, Lexington, MA, 

02420, United States of America, dfinkel@ll.mit.edu 

1 - Optimizing an NBA Team’s Approach to Free Agency using the 

Knapsack Problem 

Sam Kirshner, Queen’s School of Business, 143 Union Street, 

Kingston, Canada, skirshner@business.queensu.ca 

This paper models an NBA team’s approach free agency as a knapsack problem. 

The free agents comprise the items and the team’s cap space corresponds to the 

size of the knapsack. As free agents are signed, both the talent pool to fill the 

team’s needs and the money in the market decreases. Thus, the optimal free 

agency strategy must considers the trade-off of losing a player to a competitor and 

acquiring the player at a price that preserves as much cap space as possible to sign 

additional players. 

2 - Bicycle Tours: Modeling the Perceived Exertion of a Daily Path 

Katherine Carl, Graduate Associate, University of Arizona, 1130 E. 

Helen Street, P.O. Box 210108, Tucson, AZ, 85721, 

United States of America, kcarl@email.arizona.edu, Moshe Dror 

Bicycling has garnered interest in recent years. Perceived exertion is an important 

consideration when selecting bicycling tours. We propose a model for generating 

measures of perceived exertion for bicycling tours for 5 categories of riders. We 

present results of a survey designed to verify the accuracy of our model and 

demonstrate a good fit with exertion values reported by cyclists. 

3 - Determining NHL Player Types and Point Shares 

Timothy Chan, Assistant Professor, University of Toronto, 

5 King’s College Rd, Toronto, Canada, tcychan@mie.utoronto.ca, 

David Novati 

In this talk, we present a methodology using clustering, projection and regression 

analyses to determine distinct NHL player types and measure players’ individual 

contributions to overall team performance. We apply our method to historical 

NHL player data at the season level. Initial results suggest that our point share 

values are aligned with perceptions of “goodness” of a hockey player. In 

particular, players with the highest point share values tend to win major NHL 

awards. 


110 

■ SC18 

18- West 208 A- CC 

Methodologies for Network Optimization 


Chair: Paul Grigas, Graduate Student, Massachusetts Institute of 

Technology, 23 Gorham St., Somerville, MA, 02144, 

United States of America, pgrigas@mit.edu 

1 - Self-learning Injection-locked Laser Network for Solving 

NP-complete Problems 

Kai Wen, Edward L. Ginzton Laboratory, Stanford University, 

Stanford, CA, 94305, United States of America, 

kaiwen@stanford.edu, Kenta Takata, Shoko Utsunomiya, 

Yoshihisa Yamamoto 

NP-complete problems are important and difficult problems. Here we show that a 

novel computation machine based on injection-locked laser network and selflearning 

algorithm, driven by random phase noise, can solve NP-complete Ising 

problems in polynomial time. The stochastic simulation on MAX-CUT problems 

on cubic graph and the two-layer lattice problems demonstrates its polynomial 

time complexity up to problem size of 800 nodes. 

2 - A Parallel Decomposition-based Methodology for 

Network Optimization 

Dia St. John, INEG Department, University of Arkansas, 

Fayetteville, AR, 72701, United States of America, 

destjohn@uark.edu, Chase Rainwater, J. Cole Smith 

This work focuses on the development of parallel techniques that provide 

computational capabilities to solve large-scale network interdiction models. By 

dictating the interdiction state of a subset of system arcs, each parallel iteration 

focuses on determining the most disruptive decisions for a subset of the network. 

The use of message passing between these parallel structures facilitates efficient 

communication between simultaneous iterations and contributes to the 

algorithmic effectiveness. 

3 - Circuit Bases and Multicommodity Flows 

J N Hagstrom, Professor, University of Illinois at Chicago, 

Information & Decision Sciences, 601 S Morgan, Chicago, IL, 

60607-7124, United States of America, hagstrom@uic.edu 

We show that a basis for a capacitated multicommodity flow on a graph can be 

identified as a set of spanning trees, one for each commodity and one for the 

slacks, plus a circuit basis of the graph, with each circuit assigned to a single 

commodity or to the slacks. Work is in progress to develop a simplex algorithm 

implementation using this new insight. 

4 - An Algorithm for Convex Non-separable Network Flow 

Problems 

Trung Hieu Tran, Research Fellow, National University of 

Singapore, Industrial and Systems Engineering Dept, 1 

Engineering Drive 2, Singapore, 117576, Singapore, 

isetth@nus.edu.sg, Kwong Meng Teo 

While algorithms for multi-commodity network flows often assume linear or 

convex separable costs, non-separability is often a more realistic assumption. We 

propose a modified gradient projection-based algorithm for this more general class 

of problems, and present its performance using standard test cases. Empirical 

studies also show that a convex non-separable network flow problem can be 

handled if the same problem, but with linear cost, can be solved using CPLEX. 

5 - Proximal Subgradient and Dual Averaging for Sequential 

Decision-making and Non-smooth Optimization 

Paul Grigas, Graduate Student, Massachusetts Institute of 

Technology, 23 Gorham St., Somerville, MA, 02144, 

United States of America, pgrigas@mit.edu, Robert Freund 

We analyze and show interconnections between prox subgradient and dual 

averaging methods for both sequential decision making and non-smooth convex 

optimization. Our sequential decision-making problem generalizes the problem 

addressed in the Hedge Algorithm as studied by Baes and Burgisser, and our 

framework provides a new interpretation and extensions of the algorithm 

AdaBoost. Lastly, we examine connections between various first-order method, 

and propose new first-order methods as well.

■ SC19 

19- West 211 A- CC 


Problems in Healthcare Operations II 



Chair: Burcu Keskin, University of Alabama, Alston Hall, Box: 870226, 


bkeskin@cba.ua.edu 

1 - A Multi-level Location-allocation Problem in Human Milk 

Banking Network Expansion 

Melih Çelik, Georgia Institue of Technology, 765 Ferst Dr NW, 

Atlanta, GA, 30332, United States of America, 

melihcelik@gatech.edu, Wenwei Cao, Nadia Viljoen, 

Julie L. Swann, Ozlem Ergun 

Motivated by a project on a human milk bank reserve, we study a large-scale 

multi-level location-allocation problem for supply chain network expansion. With 

limited resources, equitable access to breast milk for newborn infants is one of the 

main concerns. We incorporate several fairness metrics into the model. Under 

certain metrics, the MIP becomes intractable. We develop efficient heuristics to 

find good solutions and conduct comparative computational analysis of different 

fairness metrics. 

2 - Enabling Greater Access to Home Meal Delivery 

Maciek Nowak, Associate Professor, Loyola University Chicago, 

1 E. Pearson, Chicago, IL, 60611, United States of America, 

mnowak4@luc.edu, Mike Hewitt, Leonardo Gala 

Meals on Wheels relies on a volunteer workforce to deliver meals to 

approximately one million homebound citizens in the US. One strategy for 

accommodating these requests is to deliver multiple (frozen) meals at a time. 

However, the funding agencies and the volunteers value the relationships that are 

developed by having a client interact with the same volunteer. We present an 

approach to quantify the efficiencies associated with delivering frozen meals while 

maintaining volunteer consistency. 

3 - An Integrated Patient Transportation and Operating Room 

Scheduling Model 

Sepehr Nemati Proon, PhD Student, University of Pittburgh, 3700 

O’Hara Street, Pittsburgh, PA, 15261, United States of America, 

sen12@pitt.edu, Oleg Shylo, Andrew Schaefer, Oleg Prokopyev 

We consider a model that integrates the optimal transportation of outpatients 

from their residences to a hospital and back using a capacitated fleet of vehicles, 

and surgery scheduling in an operating suite. The objective is to minimize the 

overall time that patients spend on the road and in the hospital. We formulate 

this problem as a mixed integer programming and present a Branch and Price 

algorithm to solve it. 

4 - Integrated Dispatch and Relocation Model to Minimize 

Emergency Vehicle Response 

Ibrahim Capar, University of Alabama, Tuscaloosa, AL, 

United States of America, icapar@cba.ua.edu, Burcu Keskin, Sharif 

Melouk 

To prepare for emergency events, EMS managers must address issues related to 

emergency vehicle dispatching and relocation. We use an MIP formulation to 

model emergency vehicle dispatching and relocation with the goal of minimizing 

incident response time while still maintaining adequate network coverage. We 

conduct extensive experimentation to examine model performance and glean 

managerial insights. 

■ SC20 

20- West 211 B- CC 



1 - Spry, Inc. – The Analyst Workbench: Better Data, Faster 

Decisions 

Todd Jones, Director of Business Solutions, Spry, Inc., Director of 

Business Solutions, Sparks, MD, 21152, United States of America, 

tjones@spryinc.com 

A single hub of integrated data, one that can be sliced however an analyst wants. 

The ability to quickly add and integrate new data sources; the ability to use R, 

Tableau and other familiar tools. These are the demands of today’s analyst. This 

presentation will introduce the role semantic technologies and big data play in 

developing this data hub. 


111 

2. - Oracle - Learn About Oracle Crystal Ball 

Michael Franden, Academic & Higher Education Sales Mgr, Oracle, 

500 Oracle Parkway, Redwood Shores, CA, 94065, United States of 

America, michael.franden@oracle.com 

Oracle’s Crystal Ball software is a spreadsheet-based software suite for predictive 

modeling, forecasting, Monte Carlo simulation and optimization. Crystal Ball is 

used in universities and schools worldwide to teach financial risk analysis, 

valuation, engineering, portfolio allocation, cost estimation, and project 

management. Join us for an overview of Crystal Ball and our teaching resources. 

■ SC21 

SC21 

21- West 212 A- CC 

Modeling and Optimization Techniques for Network 

Robustness 

Sponsor: Optimization/Networks 


Chair: Vladimir Boginski, Assistant Professor, University of Florida, 

1350 N. Poquito Rd, Shalimar, FL, 32579, United States of America, 

boginski@reef.ufl.edu 

1 - Reliable Routing under Independent Random Arc Failures and 

Algorithms with Scenario Reduction 

Qipeng Zheng, Assistant Professor, West Virginia University, 

Industrial & Management Systems Engineer, P.O. Box 6070, 

Morgantown, WV, 26505, United States of America, 

qipeng.zheng@gmail.com, Daniel Simmons 

Routing problems have been widely researched in literature. Of particular interest 

to this talk is on the stochastic cases of routing, which consider the risk of route 

failures. This talk proposes stochastic programming models for routing problems 

under independent arc failures. We propose methods which study the dual of the 

stochastic programs when the routes are given, and generate efficient cuts for the 

original problem to avoid complete enumeration of all outcomes. 

2 - Modeling and Analysis of Generalized Path-Restricted Robust 

Network Clusters 

Alexander Veremyev, NRC Research Associate, Air Force Research 

Laboratory, Building 13, 101 West Eglin Boulevard, Eglin AFB, FL, 

United States of America, averemyev@gmail.com, 

Vladimir Boginski 

Short path connectivity between each pair of nodes is an essential network 

property in different real-life settings. A k-club may not reflect certain pathconnectivity 

requirements, i.e, some important pairs of network components may 

be required to have shorter communication paths than others, or to be dependent 

on the cluster size. Several possible generalizations of a k-club concept and new 

compact linear 0-1 programming models for identifying such network clusters are 

discussed in this talk. 

3 - Optimal Design and Augmentation of Strongly Attack-Tolerant 

Two-Hop Clusters in Directed Networks 

Grigory Pastukhov, PhD Student, University of Florida, 

303 Weil Hall, Gainesville, FL, 32611, United States of America, 

gpastukhov@ufl.edu, Eduardo Pasiliao, Vladimir Boginski, 

Alexander Veremyev 

We consider the problems of minimum-cost design and augmentation of directed 

network clusters that have diameter 2 and maintain the same diameter after the 

deletion of up to R nodes and/or arcs anywhere in a cluster. We present provably 

tight theoretical bounds and a heuristic algorithm for the considered class of 

problems. Computational experiments suggest that the proposed heuristic does 

identify high-quality near-optimal solutions. 

4 - Transmission Switching in Electric Grids with Uncertain 

Line Failures 

Oleg Shirokikh, University of Florida, 303 Weil Hall, Gainesville, 

FL, 32611, United States of America, olegshirokikh@ufl.edu, 

Alexey Sorokin, Vladimir Boginski 

We consider the problem of identifying a cost-efficient strategy for transmission 

line switching in electric grids under risk considerations associated with uncertain 

failures of multiple lines (arcs) in the network. Conditional Value-at-Risk 

constraints are utilized to restrict potential losses associated with failure scenarios.

SC22 

■ SC22 

22- West 212 B- CC 

Service and Security 



Chair: Hu Hao, Chinese University of Hong Kong, Room 943, 9/F, 

Cheng Yu Tung Building, Shatin, N.T., Hong Kong, 999077, China, 

huhao@baf.msmail.cuhk.edu.hk 

1 - A Portfolio-based Approach for the Analysis of Integrated 

Telecommunications Service Providers 

Steven Powell, Professor Emeritus, California State Polytechnic 

University, Pomona, CIS Department, 3801 West Temple Avenue, 

Pomona, CA, 91768, United States of America, 

srpowell@csupomona.edu 

As their existing domestic and foreign markets mature and emerging market 

opportunities dwindle, telecommunications service providers (TSPs) are seeking 

growth from consolidating their fixed line and mobile platforms to offer 

integrated data, video, and voice services, often in multiple play bundles. This 

paper presents a multidimensional technique for analyzing integrated TSPs 

strategically based on the degree of market attractiveness and competitive 

strength their business portfolios possess. 

2 - Analysis of a Threshold-based Node Sleep Scheduling Policy 

for Energy Harvesting Sensor Networks 

Arupa Mohapatra, PhD Student, Texas A&M University, 3131 

TAMU, College Station, TX, 77843, United States of America, 

arupa@tamu.edu, Natarajan Gautam 

Autonomous wireless sensor networks with energy harvesting nodes are gaining 

prominence in long-lived remote surveillance applications. Node sleep scheduling 

is a popular approach to ensure “energy-neutral” operation in these networks. 

We present a fluid model based analysis of performance of such a network under 

a threshold based sleep scheduling policy. 

3 - Fast Approximate Choice of Record Length for Large 

Encrypted Database 

David Shallcross, Applied Communication Sciences, Piscataway, 

NJ, 08854, United States of America, dshallcross@appcomsci.com 

To conceal actual record lengths in encrypted databases, we consider choosing a 

fixed record length. Actual records are to be split and padded to form multiple 

records of the chosen length. We want to minimize the total memory consumed. 

An exact optimum can be easily be found in time quadratic in the maximum 

record length. We now present an approximate algorithm that runs in less time, 

and also some results on the sensitivity of this approximation to database updates. 

4 - Social Capital and Service Continuity: An Investigation of 

Telecom Service 

Hu Hao, Chinese University of Hong Kong, Room 943, 9/F, Cheng 

Yu Tung Building, Shatin, N.T., Hong Kong, 999077, China, 

huhao@baf.msmail.cuhk.edu.hk, Jianmin Jia, Yuho Chung 

Telecom service forms social network naturally. Consumer’s perception of its 

value depends on social capital in the network. If consumers appreciate the value, 

they will pay more attention to service continuity and recover soon after 

abnormal status occurs. Using telecom service data, we examine the role of social 

capital. The results indicate that network structural properties and reciprocal 

social norm have significant impact on consumer’s attention to service continuity 

in social network. 

■ SC23 

23- West 212 C- CC 

Edelman 2012 Finalists I 



Chair: Stephen Graves, Massachusetts Institute of Technology, Sloan 

School of Management, 77 Massachusetts Avenue, Cambridge, MA, 

02139, United States of America, sgraves@mit.edu 

1 - Transformation of HP Business Model through Advanced 

Analytics and Operations Research 

Suresh Subraman, Hewlett-Packard, San Diego, CA, 

United States of America, Suresh.Subramanian@hp.com 

So that Hewlett-Packard customers can shop anywhere anytime, HPDirect.com 

developed an ECommerce value chain across financial, marketing, warehouse 

planning and operations departments. Key drivers impacting online traffic were 

identified and quantified by time series forecasts and regression models. Their 

holistic customer targeting engine doubled purchase conversion rates and 

increased order sizes up to 30% -thanks to Bayesian modeling, Markov analysis 

and Linear Discriminant analysis. 


112 

2 - Delivering Profitable Growth for HPDirect.com using 

Operations Research 

Arnab Chakraborty, Director, Global Analytics, HP, No. 66/2, Ward 

No. 83, Bagmane Tech-Park, Embassy Prime, CV Raman Nagar, 

Bangalore, Ka, 560093, India, arnab.chakraborty@hp.com, 

Manav Shroff, Rohit Tandon, Girish Srinivasan 

HP’s online business, is meant to build robust multi-channel presence given the 

evolution in shopping behavior. The business benefitted from OR Solutions that 

overcame problems faced along the e-commerce value chain -Traffic Driver 

Identification(impacts marketing & budget management),Purchase Pattern 

Prediction(effective campaign design) AND Order Forecasting(Optimize 

warehouse inventory holding & order fulfillment).Deploying these solutions since 

‘09 have driven additional revenue of $117Mn. 

3 - Centers for Disease Control and Prevention: Advancing Public 

Health and Medical Preparedness 

Greg Burel, Center for Disease Control and Prevention, 

Washington, DC, United States of America, Greg.Burel@cdc.gov, 

Eva Lee 

CDC public health experts teamed with operations researchers to devise 

sophisticated modeling and computational strategies to improve mass dispensing. 

With the powerful information decision-support suite, RealOpt©, expanded 

capabilities are identified for those with limited resources and stressed 

environments. Over 4,000 public health and emergency directors can now make 

rapid and effective interventions for at-risk populations, lessening the subsequent 

disease mortality and save money. 

■ SC24 

24- West 213 A- CC 

Data Driven Results and Models in Healthcare 

Operations 



Chair: Hari Balasubramanian, Assistant Professor, University of 

Massachusetts, 160 Governors Drive, Amherst, MA, 01003, 

United States of America, hbalasubraman@ecs.umass.edu 

1 - Designing Clinical Trials for Cancer: An Analytics Approach 

Stephen Relyea, Massachusetts Institute of Technology, 77 

Massachusetts Avenue, Bldg. E40-149, Cambridge, MA, 02139, 

United States of America, srelyea@mit.edu, John Silberholz, 

Dimitris Bertsimas, Allison O’Hair 

We present a data-driven approach for designing new clinical trials for cancer 

chemotherapy treatments. Our approach combines (i) text-mining to develop a 

database of existing trial results, (ii) statistical modeling to predict outcomes of 

new drug combinations, and (iii) optimization models to select novel treatments 

that strike a balance between maximizing patient outcomes (exploitation) and 

learning new things about treatments that may be useful in the future 

(exploration). 

2 - Appointment Reminder Systems and Patient Preferences 

Nan Liu, Columbia University, 600 W 168th St., Room 603, New 

York, NY, 10032, United States of America, nl2320@columbia.edu, 

Stacey Finkelstein, Beena Jani, David Rosenthal, 

Lusine Poghosyan 

We utilized a cross-sectional design where patients ranked various appointment 

reminder systems and indicated their usage of technology and familiarity with 

other service providers contacting them via text message and e-mail. We found 

that patient usage and familiarity of technology are the best predictors of 

perceived effectiveness and responsiveness to text message and e-mail reminders. 

After these variables are accounted for, age and other demographic variables are 

not significant predictors. 

3 - using Emprical Data Analysis and Simulation to Improve 

Capacity Planning for Inpatient Beds 

Asli Ozen, PhD Student, University of Massachusetts Amherst, 160 

Governors Drive, Amherst, MA, 01003, United States of America, 

aslozen@gmail.com, Hari Balasubramanian 

In hospitals nationwide, misaligned bed capacity results in ED crowding and 

PACU holds. We used patient admission rates, surgery schedules, bed placement 

rules, length of stays and capacity information from Baystate Medical Center 

(Springfield, MA) to develop computer based optimization and simulation model 

to improve capacity allocation of inpatient beds. Our goal is to provide guidelines 

on how hospitals should manage their inpatient bed capacity in the presence of 

demand and LOS variability.

4 - using Empirical Data to Classify Patients for Improved 

Appointment Scheduling 

Hari Balasubramanian, Assistant Professor, University of 

Massachusetts, 160 Governors Drive, Amherst, MA, 01003, 

United States of America, hbalasubraman@ecs.umass.edu, 

Hyun-Jung Oh, Ana Muriel 

We analyze patient flow data for 400 patients collected at a family medicine 

practice in Massachusetts. We then present an stochastic integer programming 

model to schedule appointments to minimize a weighted combination of patient 

waiting and provider idle time. The model sequences patient types with different 

nurse and physician time requirements and staggers their appointment start times 

appropriately while keeping the basic slot structure used by the schedulers at the 

practice. 

■ SC25 

25- West 213 B- CC 

Radiation Therapy Treatment Planning II 



Chair: Edwin Romeijn, Professor, University of Michigan, IOE 

Department, 1205 Beal Avenue, Ann Arbor, MI, 48109-2117, 

United States of America, romeijn@umich.edu 

1 - Optimizing the Scenario Positions for Robust Radiation Therapy 


Albin Fredriksson, KTH Royal Institute of Technology, Department 

of mathematics, Royal Institute of Technology, Stockholm, Sw, 100 

44, Sweden, albfre@kth.se 

Worst case optimization over a set of scenarios is a way to achieve radiation 

therapy plans that are robust to uncertainties. However, the multiple scenarios 

exacerbate the conflict between target coverage and healthy tissue sparing, and 

may lead to unnecessary sacrifices of plan quality even in scenarios with no 

conflicts. We therefore optimize the scenario positions while enforcing minimum 

requirements, thereby determining scenario positions where the goals can be 

simultaneously satisfied. 

2 - A Hierarchical Aperture Shape Optimization Method for VMAT 


Chunhua Men, PhD, Elekta, 13723 Riverport Drive, Suite 100, 

Maryland Heights, MO, 63043, United States of America, 

chunhua.men@elekta.com, Xiao Han 

Volumetric Modulated Arc Therapy (VMAT) is a promising yet complex modern 

radiotherapy treatment technique. In this talk, we present a hierarchical aperture 

shape VMAT treatment plan optimization method in which an arbitrary 

deliverable aperture (e.g., beam-eye-view) is created at each control point, and 

each aperture is then modified using a two-step approach based on the “price” of 

each beamlet. Clinical tests show the effectiveness and the efficiency of the 

proposed method. 

3 - Homogeneous Radiation Therapy Treatment Plans for Gamma 

Knife Perfexion 

Kimia Ghobadi, University of Toronto, 5 King’s College Road, 

Tornto, ON, M5S 3G8, Canada, kimia@mie.utoronto.ca, 

David Jaffray, Mark Ruschin, Dionne Aleman 

In this work the feasibility of fractionated radiation therapy (RT) for Gamma 

Knife Perfexion is studied. In RT, the targets are typically large and overlap with 

healthy organs, and require homogeneous dose coverage. We use a fast 

geometry-based algorithm to select isocentres to provide quality RT treatments. 

After isocentres selection, a sector duration optimization is used to find the final 

shot shapes. This approach was tested on seven cases and yielded clinically 

satisfactory treatments. 

4 - A Robust Re-optimization Scheme for Intensity Modulated 

Radiation Therapy (IMRT) 

Fei Peng, PhD Candidate, University of Michigan, IOE Department, 

1205 Beal Ave., Ann Arbor, MI, 48109-2117, United States of 

America, feipeng@umich.edu, Marina Epelman, Edwin Romeijn 

Errors in patient setup position can bring uncertainty into radiation therapy 

treatments. While robust planning before the treatment in some way accounts for 

this uncertainty, it is unable to foresee variations in the actual delivered dose. We 

propose a robust re-optimization scheme that adjusts the model according to the 

delivered dose and the number of fractions remaining, and periodically reoptimizes 

the plan. Comparison with the one-shot optimization is presented with 

real-patient cases. 


113 

■ SC26 


Behavioral Operations 



Chair: Tava Olsen, Professor, University of Auckland, 12 Grafton Road, 

Auckland, 1071, New Zealand, t.olsen@auckland.ac.nz 

Co-Chair: Valery Pavlov, University of Auckland, 12 Grafton Road, 

Auckland 1071, New Zealand, v.pavlov@auckland.ac.nz 

1 - When Does the Devil Make Work? An Empirical Study of the 

Impact of Workload on Server’s Performance 



Fangyun Tan 

We analyze a large, detailed operational data set from a restaurant chain to shed 

new light on how workload. We find that when the overall workload is low, an 

increase in the workload leads to higher server performance. However, there is a 

saturation point after which any further increase in the workload leads to 

degradation of performance. 

2 - Product Launches and Buying Frenzies: A Dynamic Perspective 

Javad Nasiry, Assistant Professor, Hong Kong University of Science 

and Technology, Room 4345, ISOM Dept, HKUST, Clear Water 

Bay, Kowloon, Hong Kong, Hong Kong-PRC, nasiry@ust.hk, Pascal 

Courty 

Buying frenzies in which a firm intentionally undersupplies a product during its 

initial launch phase are a common practice within several industries such as 

electronics. We develop a dynamic model of buying frenzies based on social 

coordination among customers and characterize the conditions under which they 

are optimal. We also propose a measure of “consumer desperation” and 

demonstrate that buying frenzies can have a signififcant impact on a firm’s profit. 

3 - Stuctural Properties of Supply Chain Contracts under 

Bounded Rationality 

Valery Pavlov, University of Auckland, 12 Grafton Road, 

Auckland 1071, New Zealand, v.pavlov@auckland.ac.nz, 

Tava Olsen, Elena Katok, Ernan Haruvy 

Recent empirical studies on the performance of coordinating contracts made it 

clear that the standard theory makes strong predictions that are not always 

fulfilled. Our study aims to characterize structural properties of such popular 

mechanisms as wholesale pricing, two-part tariffs, and the minimum order 

quantity contract using the framework of quantal response equilbirium for 

boundedly rational players. 

4 - Newsvendor Decisions when Stakes Matter 

Mirko Kremer, Pennsylvania State University, State College, PA, 

United States of America, muk22@smeal.psu.edu 

Research on decision-making in the newsvendor problem has documented 

systematic order biases and a persistent inability to learn. We argue that these 

behavioral anomalies persist simply because making many small stakes decisions 

(the common implementation) allows simple decision heuristics to overpower the 

motive to make an optimal decision. In a controlled laboratory experiment, we 

find that subjects in fact learn how to place optimal newsvendor orders when 

stakes matter. 

■ SC27 

SC27 


Operations and Logistics Analytics 



Chair: Georgia Perakis, Massachusetts Institute of Technology, 

Cambridge, MA, United States of America, georgiap@MIT.EDU 

1 - Allocating Subsidies as Network Design - Tractable Fairness 

Constraints for the System Flow Problem 

Gonzalo Romero, Massachusetts Institute of Technology, 100 Main 

Street, E62-461, Cambridge, MA, 02142, United States of America, 

gromeroy@mit.edu, Retsef Levi, Georgia Perakis 

We study the problem faced by a central planner allocating subsidies to competing 

firms that provide a commodity, to minimize its market price subject to a budget 

constraint and upper bounds on the allocation to each firm. We consider two 

types of subsidies, co-payments and technology subsidies, and obtain structural 

results and near optimal solutions in various important cases. Additionally, we 

characterize a new kind of tractable fairness constraints for the system optimum 

flow problem.

SC28 

2 - Stochastic Optimization for Resource Allocation with 

Random Emergencies 

Joline Uichanco, Massachusetts Institute of Technology, 

Cambridge, MA, United States of America, uichanco@mit.edu, 

Mallik Angalakudati, Siddharth Balwani, Jorge Calzada, 

Georgia Perakis, Nicolas Raad 

We study the problem of scheduling jobs to resources, assuming a random 

number of emergency jobs arrive in the future. We model the problem using 

stochastic optimization. We propose decomposition techniques and heuristics, for 

which we prove analytic performance guarantees. We apply these to a major US 

gas utility company to schedule its services. Using actual data, we show that our 

model reduces overtime by 65%. The company is currently piloting a web-based 

planning tool using our models. 

3 - Maintenance and Flight Scheduling of Low Observable Aircrafts 

Retsef Levi, Massachusetts Institute of Technology, 100 Main 

Street, Building E62-562, Cambridge, MA, United States of 

America, retsef@mit.edu, Vivek Farias, John Kessler, 

Thomas Magnanti, Yaron Shaposhnik, Yaron Zarybnisky 

We study maintenance scheduling issues that are unique to the Air Force’s lowobservable 

aircraft. We classify this problem as a modified restless multi-armed 

bandit problem, which allows for multiple types of actions (different maintenance 

actions). Using an LP relaxation, we are able to develop index policies that allow 

us to decide the timing and type of maintenance to perform on which aircraft, as 

well as select which planes should be flown for operations each day. 

4 - Integrated Sales and Operations Strategies for Differentiated 

Fresh Products 

Zhengliang Xue, IBM Research, 1101 Kitchawan Rd., Route 134, 

Yorktown, NY, 10598, United States of America, 

zxue@us.ibm.com, Markus Ettl, David Yao 

We study the coordination of pricing and inventory strategies contingent on 

product freshness. A retailer sells both fresh and frozen products where the 

fresher provides a higher quality but perishes quickly. Customers are segmented 

as high-end and low-end, and high-end customers are more sensitive to 

freshness. Customers might upgrade or downgrade their purchase if a stock-out 

takes place. We analyze the joint pricing and inventory decisions to maximize the 

expect profit under uncertain demand. 

■ SC28 


Empirical Research in Supply Chains 



Chair: Vishal Gaur, Professor, Cornell University, Johnson School, 

Ithaca, NY, 14853, United States of America, vg77@cornell.edu 

1 - Deriving Supply Chain Metrics from Financial Statements 

Robert Bray, Assistant Professor, Kellogg/Northwestern, 

311 Sheridan Ave., Palo Alto, CA, 94306, United States of 

America, robertlbray@gmail.com, Haim Mendelson 

We develop an empirical supply chain model that enables empiricists to derive 

operational metrics from “demand signal processing” how demand information 

flows through supply chains and perform corresponding counterfactuals. The 

metrics illuminate previously-obscure supply chain aspects, and the 

counterfactuals permit related “what if” studies. We demonstrate our 

methodology with six studies and firm-level Compustat data. 

2 - Managing Global Sourcing: Inventory Performance 

Karan Girotra, INSEAD, Boulevard de Constance, Fontainebleau, 

France, Karan.GIROTRA@insead.edu, Nitish Jain, 

Serguei Netessine 

The shift towards sourcing products from global suppliers has been a key 

economic trend over the last three decades. This study provides the first rigorous 

firm-level empirical evidence, using over half a million sea shipments from global 

suppliers to US firms, on the operational impacts of global sourcing and provides 

actionable guidance for managing global sourcing. 

3 - When Disruptions Matter: An Exploration of Disruption 

Types and Moderators 

William Schmidt, Harvard Business School, Wyss House, Boston, 

MA, 02163, United States of America, wschmidt@hbs.edu, 

Ananth Raman 

We research and identify several factors that drive a differential impact of supply 

chain disruptions on company value. Our research reveals that the market 

distinguishes between disruptions not only based on the characteristics of the 

disruption, but also on the characteristics of the company which are under 

management’s direct control. Insight on this issue can help managers structure 

their operations and target risk mitigation efforts toward those disruptions that 

destroy the most value. 


114 

4 - Does Inventory Turnover Predict Future Stock Returns? 

Yasin Alan, Cornell University, Sage Hall, Room 301R, Ithaca, NY, 

14850, United States of America, ya47@cornell.edu, Vishal Gaur, 

George Gao 

We examine the impact of inventory turnover performance of publicly listed U.S. 

retailers on their stock returns using the traditional approach in the asset pricing 

literature with annual test portfolios. We find that high inventory productivity is a 

strong predictor of future stock returns, and investigate potential reasons. 

■ SC29 


Queueing Control 



Chair: Eser Kirkizlar, State University of New York - Binghamton, 

School of Management, Binghamton, NY, United States of America, 

eser@binghamton.edu 

1 - Optimal Admission Control Policies for Two Station Tandem 

Queues with Loss 

Daniel Silva, Georgia Institute of Technology, Atlanta, GA, 30363, 

United States of America, dfsi3@gatech.edu, Hayriye Ayhan, 

Bo Zhang 

We study a system of 2 queues in tandem, the 1st station has a buffer of size one, 

the 2nd has an arbitrary buffer. There is a gatekeeper and losses incur when a job 

is turned away at the first station or when the second is full upon service 

completion at the first. We fully characterize the optimal policy to minimize long 

run average cost for systems with small to moderate buffers at the 2nd station. 

We also propose an easily implementable heuristic policy that yields near optimal 

performance. 

2 - Dynamic Control of a Closed Two-stage Queueing Network for 

Ship Production with Outfitting 

Fang Dong, University of Michigan, Ann Arbor, MI, 

United States of America, ppfang@umich.edu, Mark Van Oyen, 

Soroush Saghafian, David Singer 

As an essential and complex part of ship construction, outfitting represents at 

least 50% of ship construction cost and time. Ship block outfitting is most cheaply 

done at the steel block fabrication (stage 1), but at increased cost and time, it can 

be done at the final ship assembly (stage 2). We develop a closed two-stage 

queueing network MDP model with decisions as to where outfitting is performed 

to gain insights into optimal scheduling control policies. 

3 - Parametric Analysis of Optimal Admission Policies in Tandem 

Queueing Networks 

William Millhiser, Assistant Professor, Baruch College, City 

University of New York, One Bernard Baruch Way, Box B9-240, 

New York, NY, 10010, United States of America, 

william.millhiser@baruch.cuny.edu 

We derive the parametric analysis of optimal admission policies in tandem queues 

with or without blocking after service. Our method is event-based dynamic 

programming. Such analysis allows insights into how admission policies are 

sensitive to price, demand, service rates, and workforce staffing choices; it is 

generally applicable to systems where service is delivered in stages and advance 

reservations are not taken. We give a counterintuitive result, an open problem, 

and a conjecture. 

4 - Skill and Capacity Management in Large-scale Marketplaces 

Eren Cil, University of Oregon, 1208 University of Oregon, 

Eugene, OR, 97403-1208, United States of America, 

erencil@uoregon.edu, Gad Allon, Achal Bassamboo 

We characterize the optimal skill screening mechanism of a firm moderating a 

large-scale service marketplace where the ability of a service provider to cater 

customers, who can be of two classes, varies. We show that when the values that 

a service provider generates for each customer class are independent, the firm 

may need to refuse some of the service providers via its screening mechanism 

whereas this is never optimal when these values are perfectly correlated.

■ SC30 


Risk Mitigation in Supply Chain Management 



Chair: Burak Kazaz, Associate Professor, Syracuse University, Whitman 

School of Management, 721 University Ave., Syracuse, NY, 13244, 

United States of America, bkazaz@syr.edu 

1 - Quick Response Rules for Updating Inventories of Substitutable 

Products with Short-Term Forecast Updating 

Saurabh Bansal, Assistant Professor, Pennsylvania State University, 

405 Business Building, University Park, PA, 

United States of America, sub32@psu.edu, James Dyer 

This paper presents results for determining optimal inventory levels for a 

newsvendor who (i) makes a joint-inventory decision for multiple substitutable 

products for the next replenishment period using latest forecasts, and (ii) knows 

the forecasts and the corresponding optimal inventory levels for the current 

replenishment period. These results are extremely easy to implement when faced 

with sudden demand shocks. 

2 - Corn or Soybean: Optimal Farm Space Allocation under 

Yield and Price 

Cerag Pince, cerag.pince@the-klu.org, Onur Boyabatli, 

Javad Nasiry 

We consider farm space allocation decision between corn and soybean in a multiperiod 

setting. In each period the farmer decides the allocation due to yield and 

spot price uncertainties given the cross-farming benefit, i.e, the yield is 

stochastically larger when a product is not grown in the same farm space for two 

consecutive periods. We investigate the impact of yield and price variability for 

each product, and the correlation among these uncertainties on the optimal farm 

allocation decision. 

3 - Wine Futures and Advance Selling under Quality Uncertainty 

Tim Noparumpa, Assistant Professor, Providence College, 

Providence College School of Business, Koffler Hall, Providence, 

RI, United States of America, tnoparum@syr.edu, Burak Kazaz, 

Scott Webster 

We examine the use of wine futures as a form of operational flexibility to mitigate 

quality-rating risk in wine production. As the wine ages in barrels, the firm has to 

determine what proportion of its wine should be sold as futures, and what 

proportion should be sold in retail. While advance selling enables the firm to 

recuperate the investment early and pass on the risk of holding inventory that is 

uncertain in value to consumers, it can also lead to smaller returns than retail 

revenues. 

4 - Agricultural Supply Chain Planning under Yield Uncertainty 

Nur Cavdaroglu, Post-doctoral Researcher, Whitman School of 

Management Syracuse University, Whitman School of 

Management Syracuse Un, Syracuse, NY, 13244, United States of 

America, ncavdaro@syr.edu, Burak Kazaz, Scott Webster 

We consider the production planning problem in an agricultural supply chain 

with a single retailer and multiple farmers. The quality level of the product is 

directly proportional with the improvement effort (e.g. fertilization) during the 

cultivation phase and the price of the end-product changes accordingly with 

farmers’ efforts. We analyze the problem and develop pricing mechanisms that 

coordinate the system and induce farmers’ to invest in the same level of effort 

with the centralized system. 

■ SC31 


Models of Risk and Uncertainty in Supply Chains 



Chair: Ed Pohl, University of Arkansas, Department of Industrial 

Engineering, Fayetteville, AR, United States of America, 

epohl@uark.edu 

1 - A Cargo Prioritization Model for Inland Waterway Disruptions 

Jingjing Tong, University of Arkansas, 4207 Bell Engineering 

Center, Fayetteville, United States of America, tong@uark.edu, 

Heather Nachtmann 

For an ongoing project funded by the U.S. Department of Homeland Security, we 

are developing a cargo prioritization model that can provide timely knowledge 

regarding what barges should be prioritized for offloading when the inland 

waterway is closed or reduced due to a disruption. Our presentation focus is the 

formulation of a non-linear binary integer cargo prioritization model that 

considers terminal capacity, barge draft, cargo value, and commodity type. 


115 

2 - Impact of Random Yield in Decentralized Supply Chains with 

Competing Suppliers 

Fei Qin, PhD Candidate, University of Cincinnati, College of 

Business, 312 Carl H. Lindner Hall, Cincinnati, OH, 45221, United 

States of America, qinfi@mail.uc.edu, Haresh Gurnani, Uday Rao 

We investigate supply chain (SC) performance when suppliers have unreliable 

service rate, defined as a random proportion of the planned production quantity. 

We look at single sourcing as well as dual sourcing, where both suppliers 

compete. We check the SC agents’ optimal solutions when the suppliers differ 

from one another in terms of their yield distributions, their procurement costs, 

and their risk correlation levels. 

3 - Securing Global Medical Nuclear Supply Chains through 

Economic Cost Recovery and Risk Management 

Dong Li, Univerity of Massachusetts-Amherst, 121 Presidents 

Drive, Amherst, MA, 01003, United States of America, 

dongl@som.umass.edu, Anna Nagurney, Ladimer Nagurney 

In this paper, we develop a new generalized network model for the optimization 

of the complex operations of medical nuclear supply chains in the case of the 

radioisotope molybdenum, with a focus on minimizing the total operational cost, 

the total waste cost, and the risk associated with this highly time-sensitive and 

perishable, but critical, product used in healthcare diagnostics. A case study for 

North America demonstrates how our model and computational framework can 

be applied in practice. 

■ SC32 

32- North 223- CC 

Assembly Line Design and Balancing II 


SC32 

Chair: Jeonghan Ko, Ajou University and University of Michigan, 

Department of Industrial Engineering, Suwon, Korea, Republic of, 

jko@ajou.ac.kr 

1 - Energy Considerations in a Two Machine Flowshop Scheduling 

Problem with Sequence Dependent Setups 

Afshin Mansouri, Senior Lecturer in Operations and Supply Chain 

Management, Brunel Business School, Brunel University London, 

Uxbridge, UB8 3PH, United Kingdom, 

Afshin.Mansouri@brunel.ac.uk, Emel Aktas 

A two-machine flowshop permutation scheduling problem is addressed to 

minimize Cmax and energy consumption as conflicting criteria. We develop a 

mathematical formulation and propose a Multi-Objective Genetic Algorithm 

(MOGA) to find approximations of the Pareto front. Performance of the MOGA is 

compared with Cplex on various problems. The MOGA can support informed 

decision making via trade-off analysis between Cmax as a measure of service and 

energy consumption as a measure of sustainability. 

2 - Mathematical and Heuristic Approaches for Simultaneous 

Balancing and Sequencing Problem 

Rifat Gurcan Ozdemir, Istanbul Kultur University, Atakoy Campus 

D100 Yanyol, Istanbul, Turkey, rg.ozdemir@iku.edu.tr, 

Selin Kapcak, Tugce Hanci 

Mixed model assembly line balancing and sequencing problems are solved 

simultaneously by using a mathematical model and a heuristic algorithm. In 

balancing problem, tasks are assigned to the stations for minimizing unbalanced 

workload. In sequencing problem, launching sequence of products is determined 

for minimizing total amount of uncompleted work. A case problem is solved for 

illustrating the performances of the developed mathematical model and the 

heuristic algorithm. 

3 - Exploring the Efficient Frontier of the Multi-objective General 

Assembly Line Balancing Problem 

Bryan Pearce, Clemson University, Industrial Engineering, 

110 Freeman Hall, Clemson, SC, 29634, United States of America, 

bpearce@clemson.edu, Kavit Antani, Kilian Funk, Mary E. Kurz, 

Maria Mayorga, Laine Mears, Alireza Madadi 

Methods for the general assembly line balancing problem are presented for 

environments with mixed-model flow, multi-manned stations, and a variety of 

task assignment constraints. Several figures of merit can be considered, with 

metaheuristic and heuristic methods to support exploring alternative solutions.

SC33 

4 - Lot-streaming in Assembly Shops 

Niloy Mukherjee, PhD Candidate, Virginia Polytechnic and State 

University, 114 Durham Hall, Blacksburg, VA, 24060, 

United States of America, niloym@vt.edu, Subhash C. Sarin 

We address a two-stage, multiple-lot assembly shop, lot streaming problem. While 

lot streaming reduces the makespan, it also increases handling costs. Given a 

sequence in which to process the lots, we develop a polynomial time algorithm to 

obtain optimal sublot sizes that minimize a weighted sum of the makespan and 

transportation costs. A method for sequencing the lots is also proposed. 

5 - Assembly Decomposition for Supply Chain and Subassembly 

Planning Considering Component Quality 

Jeonghan Ko, Ajou University and University of Michigan, 

Department of Industrial Engineering, Suwon, Korea, Republic of, 

jko@ajou.ac.kr, Hui Wang 

We present a decomposition model to divide the assembly to subassemblies to be 

joined in the final assembly process. This new graph-theoretic approach evaluates 

the effect of subassembly and supply chain structures on the final product quality. 

The evaluation also considers the defect rates of components and assembly tasks 

as well as inspection errors. The results suggest that the defect rate of the final 

product can be managed better by proper subassembly planning and supplier 

selection. 

■ SC33 


Sustainable Operations and Supply 

Chain Management 


Operations 


Chair: Tim Kraft, University of Virginia, Darden School of Business, 

Charlottesville, VA, 22902, United States of America, 

kraftt@darden.virginia.edu 

1 - Environmental and Economic Assessment of Remanufacturing 

Strategies for Product+Service Firms 

Anton Ovchinnikov, Darden School of Business, University of 

Virginia, 100 Darden Blvd, Charlottesville, VA, 22903, United 

States of America, Ovchinnikova@darden.virginia.edu, Gal Raz, 

Vered Blass 

This paper provides a data-driven assessment of economic and environmental 

aspects of remanufacturing for product+service firms. We present a model, 

conduct a behavioral experiment to estimate consumer demand, and based on the 

realistic parameter values obtained from an industial partner examine when and 

why remanufacturing results in higher profit and lower environmental damage. 

2 - Making use of Dual Sourcing to Reduce Carbon Emissions 

in Transport 

Tarkan Tan, Eindhoven University of Technology, Den Dolech 2, 

Pav F-7, Eindhoven, 5612AZ, Netherlands, T.Tan@tue.nl, 

Kristel Hoen, Geert-Jan van Houtum, Jan Fransoo 

We consider a carbon-aware company that orders several products from suppliers, 

which can be fulfilled via a fast or a slow channel. We study a multi-item dualsourcing 

problem subject to an emission constraint. We use a single-index order 

policy that specifies two order-up-to levels and we find solutions for the multiitem 

setting that meet the emission constraint. Making use of actual emission 

figures, we elaborate on factors that determine when significant emission 

reductions can be achieved. 

3 - To Sell and to Provide? The Implications of the Auto 

Manufacturer’s Involvement in Car Sharing 

Yannis Bellos, Georgia Institute of Technology, 

800 West Peachtree St. NW, Atlanta, GA, United States of America, 

Ioannis.Bellos@mgt.gatech.edu, Beril Toktay, Mark Ferguson 

We study the auto manufacturer’s choice regarding whether to provide mobility 

service (e.g., car sharing) in conjunction with the traditional sales channel. We 

explicitly model the consumer’s choice between purchasing a vehicle, benefiting 

from the mobility service or relying on an outside option (e.g., public 

transportation) as well as the vehicle usage decisions. We characterize the benefit 

to the manufacturer of providing mobility service and the environmental 

implications of this strategy. 

4 - The NGO’s Dilemma: How to Influence Firms to Replace a 

Potentially Hazardous Substance 

Tim Kraft, University of Virginia, Darden School of Business, 


kraftt@darden.virginia.edu, Yanchong Karen Zheng, Feryal Erhun 

In this paper, we analyze an NGO’s decisions when a substance within a product 

is identified as potentially hazardous (e.g., BPA). We determine under what 

market and regulatory conditions an NGO should target the industry versus a 


116 

regulatory body in order to influence firms to replace a substance. We consider 

the perspectives of both an antagonistic NGO that maintains an arms-length 

relationship with firms and a pragmatic NGO that takes into consideration firm 

costs when making its decisions. 

■ SC34 

34- North 224 B - CC 

Collaboration and Innovation in Product Development 


Chair: Celestine Aguwa, Visiting Assistant Professor, 

Wayne State University, 4815 Fourth Street, Detroit, MI, 48201, 

United States of America, celestine.aguwa@wayne.edu 

1 - Effect of Bargaining Power and Information Asymmetry on 

Product Quality in Outsourcing 

Narendra Singh, PhD Candidate, Georgia Institute of Technology, 

800 West Peachtree Street NW, Atlanta, GA, 30308, United States 

of America, Narendra.Singh@mgt.gatech.edu, Stylianos Kavadias, 

Ravi Subramanian 

We examine the effect of outsourcing on product quality when the OEM has a 

backup option of manufacturing the product herself. We characterize the 

situation as a game-theoretic model consisting of sequential decisions made by 

the OEM and the contract manufacturer (CM) in two scenarios - the CM offering 

the contract to the OEM, and vice-versa. We contrast the results for the two 

scenarios in the absence and presence of information asymmetry about cost 

structure. 

2 - Choice of Equity Partnerships for New Product Development 

Niyazi Taneri, Singapore University of Technology and Design, 

20 Dover Drive, Singapore, Singapore, niyazitaneri@sutd.edu.sg, 

Arnoud De Meyer 

We develop two games of equity partnerships for Joint Ventures and Licensing 

Agreements. We find that operational constraints have opposing impacts on the 

value from each partnership structure and play a role in the choice among the 

two alternatives. Analysis of alliance choice data from the pharmaceutical 

industry supports our analytical results. 

3 - Effect of Correlated Beliefs on the Survival of Collaborative R&D 

Projects 

Xiang He, University of Cambridge, Emmanuel College, 

Saint Andrews Street, Cambridge, CB2 3AP, United Kingdom, 

xh232@cam.ac.uk, Nektarios Oraiopoulos 

Owing to the remarkable uncertainty in drug R&D, biotech and pharmaceutical 

firms often engage in exploratory R&D within multiple alliances. Case evidence 

suggests that contracts are often writ with neither party having an accurate 

estimate of the value of the drug candidate. Accordingly, we study the 

effectiveness of R&D collaboration on the project level when both firms have 

imperfect but correlated beliefs about the quality of the drug candidate. 

4 - The Impact of Exploration and Exploitation on Innovation 

Success and Innovation Failure 

Jennifer Bailey, Georgia Institute of Technology, 

800 W. Peachtree Road, NW, Atlanta, GA, United States of 

America, Jennifer.Bailey@mgt.gatech.edu, Manpreet Hora 

We investigate the impact of exploration and exploitation learning activities on 

the innovation process. Specifically, our two-pronged approach examines how 

these activities can be effectively harnessed to first, achieve innovation success 

and second, to mitigate innovation failure. We utilize patent data in the biotech 

industry to test our hypotheses. 

5 - Effect of Ratings Modification on a Fuzzy-based Modular 

Architecture for Medical Device Design 

Celestine Aguwa, Visiting Assistant Professor, Wayne State 

University, 4815 Fourth Street, Detroit, MI, 48201, United States 

of America, celestine.aguwa@wayne.edu, Leslie Monplaisir, 

Prasanth Sylajakumari 

This research is to determine the effect of customer ratings on the optimal 

number of modules for medical device design. To address this issue, a typical 

glucometer is used as proof of concept to demonstrate the methodology and 

analyze the impact of changing the customer ratings on the optimal number of 

modules and minimum deviation. The implication is to generate scholarly work 

and to reduce the number of potential failure points in medical devices by 

determining the optimal number of modules.

■ SC35 


Ticket Pricing and Demand Estimation 



Chair: Ozge Sahin, Johns Hopkins University, 100 International Dr., 

Baltimore, MD, 21202, United States of America, ozge.sahin@jhu.edu 

1 - Should Event Organizers Prevent Resale of Tickets? 

Yao Cui, PhD Candidate, University of Michigan, 701 Tappan 

Street, Ann Arbor, MI, 48109, United States of America, 

cuiyao@umich.edu, Izak Duenyas, Ozge Sahin 

We are interested in whether preventing consumer resale of tickets benefits 

capacity providers for sporting and entertainment events. Consumers’ purchasing 

decisions take into account their inter-temporal valuations as well as their 

potential opportunity to resell the tickets. We compare the revenues from fixed 

pricing, dynamic pricing and pricing with options, and consider how the capacity 

provider’s revenues change with respect to the transaction cost of resale under 

each pricing strategy. 

2 - Semi-parametric Estimation of Ticket Purchases at a Theater 

Necati Tereyagoglu, Assistant Professor, Georgia Institute of 

Technology, College of Management, Atlanta, GA, 

United States of America, necati.tereyagoglu@scheller.gatech.edu, 

Senthil Veeraraghavan 

We develop a semi-parametric competing proportional hazard framework that 

models the dynamic effects of an organization’s show and time related pricing 

decisions on the customer’s propensity to purchase a ticket for a variety of shows 

during a specified timeframe. We test our model on ticket sales and seat pricing 

data for a season of 21 concerts from a leading performing arts organization. We 

suggest a discounting policy using these estimates. 

3 - Revenue Management of Consumer Options for Tournaments 

Caner Gocmen, Columbia Business School, 3022 Broadway, 

Uris Hall, 4V, New York, NY, 10027, United States of America, 

fgocmen13@gsb.columbia.edu, Robert Phillips, Guillermo Gallego, 

Santiago Balseiro 

We analyze tournament options where the event manager sells team-specific 

options for a tournament final, such as the Super Bowl, before the finalists are 

determined. These options guarantee a final game ticket to the bearer if her team 

advances to the finals. We develop an approach by which an event manager can 

determine the revenue maximizing prices and amounts of tickets and options to 

sell for a tournament final. We show that offering options can increase expected 

revenue and social welfare. 

4 - Endogeneity and Price Sensitivity in Customized Pricing 

A. Serdar Simsek, PhD Candidate, Columbia Business School, 

3022 Broadway, Uris Hall, 4L, New York, NY, 10027, 

United States of America, asimsek13@gsb.columbia.edu, 

Robert Phillips, Garrett Van Ryzin 

Endogeneity occurs in customized pricing when a seller uses unrecorded 

customer characteristics correlated with price-sensitivity in setting the price. 

Endogeneity can lead traditional regression approaches to under-estimate price 

elasticity. We use two sources of data from auto loan pricing: one from an on-line 

lender and one from an indirect lender to test for endogeneity. We present our 

results as well as recommendations for how to detect and control for endogeneity 

in the estimation process. 

■ SC36 


Revenue Management and Pricing Applications 



Chair: Ahmet Kuyumcu, President, Prorize LLC, 3399 Peachtree Rd NE, 

STE 400, Atlanta, GA, 30326, United States of America, 

akuyumcu@prorize.com 

1 - Revenue Optimization for Rental Industry 

Hossam Zaki, Vice President, Zilliant, Austin, TX, 

United States of America, hossam.zaki@zilliant.com 

This presentation will provide insights and comparisons between the revenue 

management (RM) problem for the One-Way Truck Rental industry and other 

traditional RM problems (e.g. equipment rental, car rental, airline and air cargo). 

We will also describe the main components of a typical RM system and several 

design and deployment options. 


117 

2 - Optimal Pricing for Student Housing 

Jian Wang, Vice President, The Rainmaker Group, 

4550 North Point Parkway, Suite 400, Alpharetta, GA, 30022, 

United States of America, jwang@letitrain.com 

The application of revenue management to conventional multi-family housing 

has been successful, but it is new to student housing. In this presentation, we 

describe some unique characteristics of student housing, and point out the 

drawbacks of prevalent pricing methods. We then propose an optimal pricing 

model, and illustrate the application to a real case. 

3 - Luxury Yacht Rentals – How to Forecast Demand and 

Optimize Prices? 

Pelin Pekgun, Assistant Professor, University of South Carolina, 

Moore School of Business, 1705 College Street, Columbia, SC, 

29208, United States of America, Pelin.Pekgun@moore.sc.edu, 

Ronald Menich 

Luxury yacht rentals is a niche market with a unique business model that 

presents rich revenue opportunities but also makes it vulnerable in a challenging 

economy. Traditional forecasting approaches do not work due to data sparsity and 

low volume of transactions, while pricing becomes difficult with hard-to-identify 

price elasticity from customers that already pay high prices. We present a 

forecasting and optimization framework that can address the unique challenges in 

this environment. 

4 - Market Share Modeling in Revenue Management 

Ahmet Kuyumcu, President, Prorize LLC, 3399 Peachtree Rd. NE, 

Ste. 400, Atlanta, GA, 30326, United States of America, 

akuyumcu@prorize.com, Utku Yildirim 

Market share modeling could be effectively used when demand and price data for 

majority of the competing products is available. This presentation briefly reviews 

market share modeling and provides successful real-world applications. 

■ SC37 

SC37 


Location Models 

Sponsor: Location Analysis 


Chair: Oded Berman, Professor, University of Toronto, 105 St. George 

Street, Toronto, ON, M5S 3E6, Canada, Berman@Rotman.Utoronto.Ca 

1 - Linearization Techniques for the p-Median Problem with 

Unreliable Facilities 

Serigne Gueye, Assistant Professor, University of Avignon, 339 

chemin des Meinajariès, Agroparc BP 91228, Avignon, 84911, 

France, sgueye@zlc.edu.es, Mozart Menezes 

The p-Median problem with unreliable facilities consists in locating m unreliable 

facilities to n customers in such a way to minimize the expected cost of serving 

the customers. Each facility has a disruption probability, independent to the site 

location and to other facility. The problem is expressed as a bilinear program for 

which some linearization techniques are applied. Numerical tests are then 

performed. 

2 - The Downside Risk-averse Center on a Network 

Jiamin Wang, Long Island University, Long Island, NY, 

United States of America, Jiamin.Wang@liu.edu 

We consider finding a location that minimizes the maximum expected weighted 

distance to the nodes of a network with random demand weights. This problem is 

similar to the classical center problem except VAR, a downside risk-averse 

measure, is adopted as a constraint. We derive a set of dominant points and 

develop algorithms for demand weights of discrete and continuous probability 

distributions. This study provides a better understanding of the impact of risk 

attitude on location decisions. 

3 - Minimax Regret Covering Location on a Network 

under Uncertainty 

Oleksandr Shlakhter, University of Toronto, Joseph L. Rotman 

School of Management, 105 St. George Street, Toronto, ON, M5S 

3E6, Canada, alex.shlakhter@Rotman.Utoronto.Ca, Dmitry Krass, 

Oded Berman 

We consider p-cover problem on a network with uncertain weights of nodes. For 

each node only interval estimate of its weight is known. The problem is to find 

the “minimax regret” covering location, i.e. to minimize worst case loss in 

objective function because a location decision is made without knowing the 

actual weights at nodes. We present the algorithm for this problem on a general 

network. We also consider two special cases, when the solution algorithm is 

polynomial.

SC38 

4 - Resource Allocation for a New Product Introduction: 

Joint Facility Location and Marketing Strategies 

Vahideh Sadat Abedi, PhD Candidate, University of Toronto, 

105 St. George Street, Toronto, ON, M5S 3E6, Canada, 

VahidehSadat.Abedi07@Rotman.Utoronto.Ca, Oded Berman, 

Dmitry Krass 

We analyze the problem of joint design of the network of retail facilities and 

marketing strategies of a firm introducing an innovative product or service to a 

network of markets. After introducing a framework for the joint study of these 

decisions, we study the effect of different types of marketing media and their 

substitutive behaviour. We also investigate allocation of resources of the firm 

between different types of media and opening facilities. 

■ SC38 


Service Operations and Retailing 



Chair: Olga Perdikaki, Assistant Professor, Mays Business School at 

Texas A&M University, 320 Wehner Building, College Station, TX, 

77843-4217, United States of America, operdikaki@mays.tamu.edu 

1 - Timining of Service Investments under Competition and 

Demand Uncertainty 

Olga Perdikaki, Assistant Professor, Mays Business School at Texas 

A&M University, 320 Wehner Building, College Station, TX, 

77843-4217, United States of America, 

operdikaki@mays.tamu.edu, Jayashankar M. Swaminathan, 

Dimitris Kostamis 

We study how competition and demand uncertainty affect firms’ timing of service 

level decisions relative to demand realization. We model a symmetric duopoly in 

which firms compete on price and service levels and analytically characterize the 

Nash equilibrium in the timing of service level decisions. 

2 - Retail Execution and Store Associate Incentive Design 

Nicole DeHoratius, Zaragoza Logistics Center, Edifico Náyade 5, 

C/ Bari 55-Plaza, Zaragoza, Spain, NDeHoratius@zlc.edu.es, 

Saravanan Kesavan, Adam Mersereau 

We determine the extent to which a change in the design of store associate 

incentives at several pilot stores of a major retail chain drive store associate effort 

allocation and consequently store performance. Our results yield insights useful to 

a retail chain seeking to maximize long-term profitability rather than short-term 

sales when designing store associate incentive plans. 

3 - Inventory Record Inaccuracy: Operational Causes and 

Labor Effects 


at Texas A&M University, College Station, TX, 77843-4217, 

United States of America, hchuang@mays.tamu.edu, Rogelio Oliva 

Inventory record inaccuracy (IRI) can be attributed to shrinkage, misplacement, 

and transaction errors. We develop a system dynamics model that captures the 

physics of in-store operations while integrating multiple sources of error. By 

performing Monte-Carlo simulation, we confirm the conjecture that labor-related 

errors are roor causes of IRI. We then empirically test the impact of store labor 

and environments on IRI using Bayesian shrinkage estimation and panel data 

modeling. 

4 - Optimal Dynamic Return Management of Fixed Inventories 

Mehmet Sekip Altug, Assistant Professor, George Washington 

University, 2201 G. Street, NW, Washington, DC, 

United States of America, maltug@gwu.edu 

We consider a retailer that sells a fixed amount of inventory over a finite horizon. 

We assume that return policy is a decision variable which can be changed 

dynamically at every period. While flexible return policies generate more 

demand, it also induces more returns. We characterize the optimal dynamic 

return policies based on two costs of return scenarios. We find out how these 

policies change with inventory and time. We extend our model to multiple 

competing retailers. 


118 

■ SC39 


Service Science & Innovations with Analytics 



Chair: Haluk Demirkan, Professor, Arizona State University, 

P.O. Box 874606, Tempe, AZ, 85287, United States of America, 

haluk.demirkan@asu.edu 

1 - Transitioning to a Service-oriented Business Model: 

A Technology Roadmapping Approach 

Robert Harmon, Professor of Marketing & Technology 

Management, Portland State University, School of Business, 

Portland, OR, 97207, United States of America, 

corevalue@comcast.net, Bill Hefley, Haluk Demirkan 

For technology firms, the evolution from manufacturing to service-dominant 

business models as a primary driver of innovation has become an important 

trend. This research reviews the literature on the key dimensions, frameworks, 

and organizational considerations for the service transition process. It utilizes a 

technology roadmapping approach to illustrate the issues and strategy dimensions 

faced by a technology manufacturer as it navigates the service transition process. 

2 - Quality of Service Modeling for Internet Service Providers 

Seyedreza Mousavi, Arizona State University, BA 319, WPC School 

of Business, Arizona State University, Tempe, AZ, 85287, 

United States of America, smousav4@exchange.asu.edu 

According to the literature, there are three main contributors to quality of service 

of internet service providers: traffic and network engineering, network 

architecture, and interconnection. The current study attempts to develop a model 

in which market and quality factors are linked together. A panel of experts rated 

the strength of each relationship in the model. The results revealed how the 

contributors to the QoS significantly impact the ISP’s profit. 

3 - Effects of Electronic Payment Services on Consumer Behavior 

in Electronic Retailing: A Research Prop 

Irfan Kanat, Arizona State University, W.P. Carey School of 

Business, Tempe AZ 85287, United State of America 

ikanat@asu.edu, Ajay Vinze, Paul Steinbart 

e-payment services advanced rapidly in the recent years. Major e-retailers like ebay 

are now players in the market, traditional financial institutions are trying to 

defend their positions with new products. Each service offers a different set of 

features suited for different transactions. The effects of the features sets on 

consumer behavior has not been investigated in the literature. We present a 

review of the literature and propose a quasi experimental research design to 

address this gap. 

■ SC40 


Wind Energy, Optimal Location and 

Effective Utilization 



Chair: Golbon Zakeri, University of Auckland, 70 Symonds Street, 

Auckland, New Zealand, g.zakeri@auckland.ac.nz 

1 - Mitigating Real Time Price Uncertainty Induced by Wind Power 

through Stochastic Unit Commitment 

Anthony Papavasiliou, University of California at Berkeley, 4141 

Etcheverry Hall, IEOR Department, Berkeley, CA, 94720-1777, 

United States of America, anthony.papavasiliou@gmail.com, 

Yi He, Alva Svoboda 

Wind power supply uncertainty introduces significant uncertainties in day-ahead 

trading due to its impact on real-time electricity prices. In particular, in the day 

ahead market utilities face the challenge of self-commitment, self-scheduling and 

cycling decisions on slow units. We present a stochastic unit commitment model 

and solution methodology for addressing these challenges and demonstrate 

savings relative to current practice which is based on deterministic models.

2 - The Effects of Stochastic Market Clearing on the Cost of Wind 

Integration: A Case of New Zealand Electricity Market 

Javad Khazaei, Princeton University, Department of ORFE, 

Princeton, NJ, United States of America, jkhazaei@princeton.edu, 

Geoff Pritchard, Golbon Zakeri 

Implementing a stochastic electricity market clearing, as a suggested alternative to 

conventional clearing mechanisms, can impose some extra cost on the market. 

Therefore, it is essential to estimate the efficiency gain resulting from 

implementing a stochastic market clearing mechanism. We describe the result of 

an empirical study to quantify value of a stochastic clearing mechanism for the 

New Zealand electricity market. We extend our analysis for possible larger wind 

integration in the future. 

3 - Modeling the Economics of Variable Generation Resources 

James Merrick, Electric Power Research Institute (EPRI), 3420 

Hillview Avenue, Palo Alto, CA, 94304, United States of America, 

jmerrick@epri.com, Geoffrey Blanford 

The US-REGEN model is a general equilibrium model of the United States 

economy, with a detailed regional representation of the power sector. The power 

sector component is formulated as a quadratic programming problem. This talk 

will provide an overview of the optimization formulation, focusing on how the 

spatial and temporal distributions of variable generation resources are 

represented. The formulation of the sub-problem to choose and weight 

representative hours will be additionally presented. 

4 - Capacity Expansion in an Electrical System with Intermittent 

and Weather-dependent Generation 

Phil Bishop, Senior Economist, Electricity Authority, P.O. Box 

10041, Wellington, 6143, New Zealand, phil.bishop@ea.govt.nz, 

Bruce Smith 

The electricity system in NZ is dominated by generating plant based on highly 

variable hydrological resources yet conditions currently favor investment in 

geothermal and intermittent plant. Capacity expansion in this setting entails 

challenges for the efficient and reliable operation of the system and imposes costs 

beyond those incurred by the generation investor. A capacity expansion model 

with a detailed treatment of firming requirements is described and simulation 

results are presented. 

■ SC41 


Renewable Energy Issues 


Chair: Paulo Correia, Professor, Unicamp, Energy Department, 

Campinas, Brazil, pcorreia@fem.unicamp.br 

1 - Costs and Environmental Attributes of Bulk Energy 

Transportation Options for Renewable Energy 

Elisabeth Gilmore, University of Maryland, 2101 Van Munching 

Hall, College Park, MD, United States of America, 

gilmore@umd.edu, Andrew Blohm 

We develop a multi-attribute decision model to evaluate options for long distance 

transport of bulk energy from renewable sources, focusing on the cost and 

environmental trade-offs. We find that the preferred option depends on distance, 

the amount of energy to be transported, and the existing infrastructure and 

markets. We apply the model to an ocean thermal energy conversion (OTEC) 

installation where we include moving the energy as ammonia, either as an energy 

carrier or as fertilizer. 

2 - Competition, Welfare and Green Energy 

Talat Genc, University of Guelph, Department of Econ, 

Guelph, N1G2W1, Canada, tgenc@uoguelph.ca 

We estimate the behavior in hourly wholesale electricity auctions and address 

some policy questions relevant to intermittent technologies and air pollution in a 

dynamic game-theoretic analysis. We examine the Ontario wholesale electricity 

market and find that our competition model has a high predictive power. MAE 

for our price estimations is much lower than the one for ISO estimations. We also 

calculate the welfare loss generated in the market. 

3 - Wind Farm Layout Optimization Considering Multiple Objectives 

Jose Espiritu, Assistant Professor, The University of Texas at El 

Paso, 500 West University Avenue, El Paso, TX, 79902, United 

States of America, jfespiritu@utep.edu, Carlos Ituarte-Villarreal, 

Nicolas Lopez, Heidi Taboada 

This work presents a new multiple objective evolutionary algorithm to obtain 

optimal placement of wind turbines in a wind farm, the objective functions 

considered are the maximization of power output, the minimization of wind farm 

cost and the maximization of system reliability. In the present work, we solve 

three well known problems. The final solution to this multiple objective problem 

is presented as a set of Pareto solutions. 


119 

4 - Studying Various Optimal Control Problems in Biodiesel 

Production in a Batch Reactor under Uncertainty 

Pahola Benavides, Graduate Student, University of Ilinois at 

Chicago, Industrial Engineering, Chicago, IL, 60607, 

United States of America, ptatibg@uic.edu, Urmila Diwekar 

Optimal control problems encountered in biodiesel production are formulated 

using various performance indices like maximum concentration, minimum time, 

and maximum profit. The problems involve determining optimal temperature 

profile so as to maximize these performance indices. This paper presents these 

three formulations with and without variability and uncertainty in feed 

composition. 

5 - Trading Energy from Wind Plants with a Portfolio Approach 

Paulo Correia, Professor, Unicamp, Energy Department, Campinas, 

Brazil, pcorreia@fem.unicamp.br, Laura Gunn 

The growing environmental impacts associated to the generation of electric 

energy highlights the need to expand the use of renewable energy sources. These 

sources use natural resources whose availability depends on climate behavior 

which normally entails a certain intrinsic risk. Therefore, the evaluation of an 

alternative energy source project always involves a commitment between the 

expected benefit and the risk incurred, which can be conveniently dealt with a 

Markovitz benefit-risk model. 

■ SC42 


Routing and Travel Time Prediction 


SC42 

Chair: Behzad Aghdashi, Research Assistant, North Carolina State 

University, 2510-203 Avent Ferry Rd, Raleigh, NC, 

United States of America, saghdas@ncsu.edu 

1 - Improved Tour Determination for the Dynamic Traveling 

Salesman Problem 

Taesu Cheong, Assistant Professor, National University of 

Singapore, 1 Engineering Drive 2, Singapore, 117576, Singapore, 

isects@nus.edu.sg, Alan Erera, Chelsea (Chip) White 

After each stop of a multi-stop trip, we assume that the vehicle chooses the next 

stop to visit, based on current traffic conditions. Travel time along each arc in the 

network is modeled as a random variable, and we assume that network 

congestion dynamics can be described by a stationary Markov chain. We propose 

an efficient algorithm for determining the next stop to visit in a multi-stop tour, 

using action elimination procedures and a rolling forecast horizon. 

2 - A Case Study for Time-dependent Routing Planning with 

Time Windows 

Qian An, PhD Candidate, University of Southern California, 

Los Angeles, CA, 90007, United States of America, qan@usc.edu, 

James Moore, Ugur Demiryurek, Cyrus Shahabi 

In this paper we study the importance of time-dependent route planning as 

opposed to static planning. A time-dependent network dataset of the entire Los 

Angeles County highways and arterial streets is used. It includes 6300 traffic 

sensor data with update rate of every 1 minute covering approximately 3000 

miles. With extensive experiments we show that time-dependent planning could 

save 9% as compared to static planning in uniform time window cases, and 20% 

in random time window cases. 

3 - A Hierarchical Agent-based Simulation for Traveler Behaviors 

Yiheng Feng, University of Arizona, 1127 E. James E. Rogers Way, 

P.O. Box 210020, Tucson, AZ, 85705, China, 

yihengfeng@email.arizona.edu, Hui Xi, Sojung Kim, 

Young-Jun Son, Larry Head 

A hierarchical agent-based framework is proposed to model the travelers’ 

behaviors with route choice on the upper level and tactical driving on the lower 

level. The route choice level considers learning from previous experiences, 

heterogeneity of different travelers, and incomplete network information, while 

the tactical driving level concerns car-following and lane changing behaviors. 

Results from the proposed models are shown to reach the same user equilibrium 

solutions as in the literature. 

4 - Path and Multi-step Ahead Freeway Travel Time Prediction 

Wenxin Qiao, University of Maryland-College Park, 1173 Glenn 

Martin Hall, College Park, MD, 20742, United States of America, 

wxqiao@umd.edu, Ali Haghani 

This research utilized multiple data sources to predict freeway travel times 

through an integrated model based on large historical traffic and weather 

information dataset, along with real time traffic information. This model is 

proposed to perform path and multi-step ahead travel time prediction. Numerical 

experiments are conducted and comparison results of different models’ 

performances are presented as well as sensitivity analysis on the efficient size of 

the historical dataset.

SC43 

5 - Scenario Generation in Freeway Reliability Analysis 

Behzad Aghdashi, Research Assistant, North Carolina State 

University, 2510-203 Avent Ferry Rd, Raleigh, NC, 

United States of America, saghdas@ncsu.edu 

Scenario Generation is primary step in evaluating freeway’s travel time reliability. 

In this research different mathematical modeling challenges are discussed. The 

main focus of this research is to minimize the error and bias imposed by modeling 

incidents and weather events. 

■ SC43 


Joint Session RAS/CPMS: Panel Discussion: 

Analytics and Intermodal 

Sponsor: Railway Applications & CPMS, The Practice Section 


Chair: Bruce Patty, Vice President, Veritec Solutions, 824 Miramar 

Terrace, Belmont, CA, 94002, United States of America, 

bpatty@veritecsolutions.com 

1 - Roundtable on Analytics and Intermodal 

Moderator: Bruce Patty, Vice President, Veritec Solutions, 824 

Miramar Terrace, Belmont, CA, 94002, United States of America, 

bpatty@veritecsolutions.com, Panelists: Gino Phillips, Craig Littzen, 

Vernon Prevatt 

This roundtable will take place during two sessions. During this first session, there 

will be three panelists from the Intermodal industry (Gino Phillips from APL, 

Craig Littzen from Swift, and Vernon Prevatt from CSX) who will share their 

thoughts about current and potential applications of analytics to the intermodal 

industry. After their opening presentations, there will be an opportunity for the 

audience to ask questions and participate in an open discussion. 

■ SC44 


Contracts in Supply Chain 


Chair: Arnab Bisi, Associate Professor of Operations Management, 

Indian Institute of Management Calcutta, Joka, Kolkata, 700104, India, 

abisi@iimcal.ac.in 

1 - Optimization of Supplier and Customer Strategies under 

Options Contract and Unknown Pricing/Demand 

Matan Shnaiderman, Bar-Ilan University, Ramat-Gan, 52900, 

Israel, shnidem@biu.ac.il, Avi Ceder 

This work develops optimal strategies of a supplier and customers when using 

options contract. One supplier sells goods for two customers. The supplier uses 

option contracts, for extending profits, such that the customers have the 

possibility to purchase options in advance and being the first to receive the 

production. The stochastic demands at the customers’ sites are continuously 

distributed. We show that customers who receive high profits, are interested to 

buy the options and exercise them. 

2 - Operations and Marketing Equilibrium Strategies under 

Wholesale Price and Revenue Sharing Contracts 

Fouad El Ouardighi, Professor, ESSEC Business School, Av. B. 

Hirsch, Cergy Pontoise cedex, 95021, France, elouardighi@essec.fr, 

Dieter Grass, Steffen Jorgensen, Gary Erickson 

The paper studies the relative merits of two specific contracts, the wholesale price 

and the revenue sharing contract in a dynamic vertical channel with one 

manufacturer and one retailer. We consider decisions that develop from Nash 

equilibria. Two different strategies are available to the players, i.e., open-loop and 

feedback Nash equilibrium strategies. We compare how these different strategies 

affect the players’ decisions and payoffs under wholesale price and revenue 

sharing contracts. 

3 - The Impact of Supplier Innovation on Contract Efficiency 

Xiaolu Zuo, PhD Student, City University of Hong Kong, 

Tat Chee Avenue, Kowloon, Hong Kong, Hong Kong-PRC, 

zuoxiaolu29@gmail.com, John Liu 

Consider in a one supplier-one retailer supply chain setting, the supplier 

experiences process improvement, reflected by the production cost reduction, 

which follows a discrete dynamic way. We analyze the impact of supplier 

innovation on the efficiency of supply chain contract and give an explanation for 

the incentive of supplier innovation. 


120 

4 - Wholesale-price Contracts with Postponed and Fixed 

Retail Price 

Arnab Bisi, Associate Professor of Operations Management, Indian 

Institute of Management Calcutta, Joka, Kolkata, 700104, India, 

abisi@iimcal.ac.in, Yanyi Xu 

We study wholesale-price contracts with retail price-postponement in a supply 

chain consisting of one manufacturer and one retailer. For additive and 

multiplicative demand models, we establish sufficient conditions for the 

unimodality of profit functions and derive the unique optimal solutions. We also 

extend existing results on the fixed retail price case and a revenue management 

problem. 

■ SC45 


Joint Session MI/HAS: Operations Research in 

Patient Safety and Care Quality 

Sponsor: Minority Issues & Health Applications Society 


Chair: Laila Cure, Assistant Professor, Western Michigan University, 

4601 Campus Dr., Kalamazoo, MI, 49008-5336, United States of 

America, laila.cure@wmich.edu 

1 - A Multiple-drawer Medication Layout Problem in Automated 

Dispensing Cabinets 

Jennifer Pazour, Assistant Professor, University of Central Florida, 

4000 Central Florida Blvd, Orlando, FL, 32816, United States of 

America, Jennifer.Pazour@ucf.edu, Russell Meller 

We investigate the problem of locating medications in automated dispensing 

cabinets to minimize human selection errors. We formulate the multiple-drawer 

medication layout problem as a quadratic assignment problem. As a way to 

evaluate various medication layouts, we develop a similarity rating for medication 

pairs. To solve industry-sized problem instances, we develop a heuristic approach. 

2 - A Decision Support Model for the Location of Hand Sanitizer 

Dispensers in Hospitals Units 

Laila Cure, Assistant Professor, Western Michigan University, 

4601 Campus Dr., Kalamazoo, MI, 49008-5336, 

United States of America, laila.cure@wmich.edu 

Noncompliance with hand hygiene guidelines is a cause of infection in hospitals. 

To improve compliance, hand sanitizer dispensers should be placed in convenient 

locations, but fire safety regulations constrain their number and distribution. We 

develop a model to incorporate guidelines, regulation, and hospital characteristics 

into the dispenser installation decision. The proposed model can help in 

evaluating the current locations of dispensers and in determining the best possible 

configuration. 

3 - A Hub Infrastructure to Reduce Infusion Pump Medication 

Errors in Hospitals 

Kenneth Musselman, Strategic Collaboration Director, Regenstrief 

Center for Healthcare Engineering, Purdue University, Mann Hall, 

Suite 225, 203 Martin Jischke Drive, West Lafayette, IN, 47907, 

United States of America, kmusselm@purdue.edu 

Reliable, timely and low cost data collection is necessary to achieve a significant 

reduction in hospital medication errors. The Infusion Pump Informatics (IPI) 

System, which was developed using a hub infrastructure, allows infusion pump 

data to be readily accessible, meaningful and actionable. Participating hospitals are 

now forming virtual communities that exchange information, share analysis 

results, trigger research, and collaborate on ways to improve patient safety. 

4 - Prediction, Non-clinical Factors and Intervention in 

Unnecessary Hospital Readmissions 

Brandon Pope, Purdue University, 315 N. Grant Street, 

West Lafayette, IN, United States of America, popeb@purdue.edu 

Unnecessary readmissions waste $17 B per year in U.S. hospitals on Medicare 

patients alone. In this presentation we focus on calibrating predictive models 

using nearly 400,000 discharges over a 3.5 year span from a network of 11 U.S. 

hospitals. We also examine the relative importance of non-clinical factors in 

readmission rates. Motivated by financial reform, we model the discharge 

intervention problem, highlighting the possibility of unintuitive optimal policies, 

and opportunities for research.

■ SC46 


Organizational and Industry Renewal (III): Leadership, 

Team Management, and Organizational Learning 



Chair: Andreas Schwab, Associate Professor, Iowa State University, 



1 - Schematic Inertia: A Dynamic and Complexity Pespective on 

Sared Cognition 

Benjamin Herndon, Assistant Professor, Georgia Institute of 

Technology, 800 W. Peachtree St., NW, Atlanta, GA, 30308, United 

States of America, benjamin.herndon@mgt.gatech.edu, Kyle Lewis 

This paper extends theories of socially shared cognition by forwarding a 

framework to explain how – over time – teams are affected by dynamic and 

residual schema and dimensions of shared cognition. We articulate these 

dimensions and consider how they form, accrue and decay as team tasks and 

boundaries change over time. Finally, we posit relationships between these 

dimensions and a team’s ability to realize performance synergy or dissonance on 

the task at hand. 

2 - Exploring the Role of Culture in Psychological Safety and 

Team Learning 

Mary Zellmer-Bruhn, Associate Professor, University of Minnesota, 

321 19th Ave., S, Minneapolis, 55455, United States of America, 

zellm002@umn.edu 

Psychological safety is vital to internal and external team learning. Given 

differences in behavioral expectations across cultures, evidence that work-related 

constructs may carry different meaning across cultures, and increases in globally 

distributed work, examining psychological safety in different cultural settings is a 

valuable avenue for research. This paper presents an exploratory framework for 

the role of culture and language differences in team psychological safety. 

3 - An Exploration of Inter-Organizational Heterogeneity in 

Organizational Goal Setting 

Cynthia Penaflor, Research Assistant, Brigham Young University, 

2112 Margalene Way, Austin, TX, 78728, United States of 

America, cynthiapenaflor@gmail.com, Peter Madsen 

Limited availability of data on organizational goals has necessitated the 

imputation of organizational goals in empirical tests of the behavioral theory of 

the firm; but this approach cannot account for inter-organizational heterogeneity 

in goal setting. Our project plans to overcome this challenge by analyzing newlyavailable 

data from all companies that have been members of the S&P 500 index 

from 2006-2011. We test how closely imputed goals approach self-reported 

organizational goals. 

4 - Shared Leadership: The Impact of the Occupy Movement on 

Decision Making in Organizations 

Adina Lav, Doctoral Student, George Washington University, 

2480 16th Street, NW #445, Washington, DC, 20009, 

United States of America, adinalav@gwmail.gwu.edu 

Social and political movements around the globe witnessed over the past year are 

great examples of the changes taking place in how people learn from each other 

and express their leadership at the societal level. This oral presentation describes 

the Occupy movement and its organic development of consensus building and 

shared leadership and calls for organizations to broaden their perspectives of 

decision making by understanding new ways in which leadership and consensusbuilding 

takes place. 

■ SC47 

47- North 230- CC 

Travel Time Estimation and Applications 




Chair: Xuesong Zhou, University of Utah, 110 Central Campus Dr, 

MCE 2000, Salt Lake City, UT, 84112, United States of America, 

zhou@eng.utah.edu 

1 - Predicting Corridor-level Travel Time Distributions Based on 

Stochastic Flow and Capacity Variations 

Hao Lei, Phd Student, University of Utah, 110 Central Campus Dr, 

MCE 2000, Salt Lake City, UT, 84112, United States of America, 

hao.lei@utah.edu, Xuesong Zhou 

This research aims to establish a point-queue based end-to-end travel time 


121 

prediction method on a corridor with multiple merges and diverges. A set of 

analytical equations is developed to calculate the number of queued vehicles 

ahead of the probe vehicle and further capture many important factors affecting 

end-to-end travel times. 

2 - On the Link Travel Time Imputation Methods 

Lei Zhu, Graduate Research Assisant, University of Arizona, 1209 

E 2nd street, RM 324J, Tucson, AZ, 85721, 

leizhu@email.arizona.edu, Yi-Chang Chiu 

Link travel time imputation is important for shortest path algorithms under a 

certain circumstance. Based on traffic flow conservation law and the average flow 

theorem, the missing travel time can be estimated by flow of adjacent upstream 

and downstream road segments. In this research, the missing link travel times are 

categorized into several conditions and the proposed imputation methods have 

been shown to deal with each condition effectively. 

3 - ParkPGH: Pittsburgh’s Predictive Parking App 

Tayo Fabusuyi, Lead Strategist, Numeritics, 5907 Penn Ave., Suite 

313, Pittsburgh, PA, 15206, United States of America, 

Tayo.Fabusuyi@numeritics.com, Robert Hampshire, 

Katsunobu Sasanuma 

ParkPGH is a smart parking solution that provides information on the availability 

of parking spaces within Pittsburgh’s downtown. The application uses an event 

calendar, historical data, multiple regression model and a Bayesian updating 

process in making both long and short term predictions about parking availability. 

Parking information is delivered through channels that include websites, SMS 

text, voice, iPhone app and a mobile version of the website that is optimized for 

other smartphones. 

4 - Quantification of the Benefits of Real-time Information for 

Individual Travelers 

Dongyoon Song, NEXTRANS Center, USDOT Region V Regional 

UTC, School of Civil Engineering, Purdue University, 


song50@purdue.edu, Srinivas Peeta 

This study seeks to quantify the benefits of real-time travel information in terms 

of both conventional travel time savings as well as psychological effects for 

individual travelers. Revealed preference (RP) data from an experiment with a 

driving simulator and stated preference (SP) data from a survey will be utilized in 

behavioral analysis to estimate the value of the real-time information. 

■ SC48 

SC48 


Traveling Salesman Problem with Time Windows 

Sponsor: Transportation Science & Logistics/ Freight 

Transportation & Logistics 


Chair: Amelia Regan, University of California, Irvine, CA, 

United States of America, aregan@uci.edu 

1 - Hybrid Parallel Metaheurustics for the Traveling Salesman 

Problem with Time Windows 

Amelia Regan, University of California, Irvine, CA, United States 

of America, aregan@uci.edu, Dmitri Arkhipov 

Our work examines a novel way to parallelize the TSPTW problem. 

2 - The Probabilistic Travelling Salesman Problem with Time 

Windows 

Stacy Voccia, University of Iowa, Iowa City, IA, United States of 

America, stacy-voccia@uiowa.edu, Barrett W. Thomas, 

Ann M. Campbell 

We introduce the Probabilistic Traveling Salesman Problem with Time Windows 

(PTSPTW), where in addition to stochastic customer presence, each customer has 

an associated time window during which deliveries must be made. We present a 

recourse model and a Variable Neighborhood Search (VNS) algorithm to solve 

problem instances. Using computational experiments, we demonstrate the value 

of incorporating stochasticity into the model. 

3 - A Branch-and-Price Approach for Vehicle Routing with Soft 

Time Windows and Stochastic Travel Times 

Duygu Tas, TU/e Eindhoven University of Technology, School of 

Industrial Engineering, P.O. Box 513, Eindhoven, 5600 MB, 

Netherlands, D.Tas@tue.nl, Michel Gendreau, Tom van Woensel, 

Nico Dellaert, Ton de Kok 

We study a vehicle routing problem with soft time windows and stochastic travel 

times. In our problem setting, we have for each node, except the depot, a known 

demand, a fixed service duration and a soft time window. Our objective is to 

minimize the sum of transportation costs and service costs. We apply a column 

generation procedure to solve our model. To generate an integer solution, we 

embed our column generation procedure within a branch-and-price method.

SC49 

4 - A Divide-and-conquer Analysis of Very Large Asymmetric 

Vehicle Routing Problems 

Dauwe Vercamer, Ghent University, Ghent, 9000, Belgium, 

Dauwe.Vercamer@ugent.be, Birger Raa, Dirk Van den Poel 

When working with extremely large datasets in real-life VRPs, solution 

neighborhoods must be kept simple in order to obtain reasonable process times. 

By dividing the set into smaller samples, more elaborate neighborhoods can be 

explored or the total process time can be reduced. However, this also limits the 

optimization. Our research investigates the impact of using these smaller samples 

on the solution quality and provides approaches for overcoming the downsides of 

the division. 

5 - A Multi-terminal Truckload Pickup and Delivery Problem with 

Time Windows in a Dynamic Context 

Hossein Zolfagharinia, Wilfrid Laurier University, 1207-125 

Lincoln Road, Waterloo, ON, N2J 2N9, Canada, zolf1380@wlu.ca, 

Michael Haughton 

We study a dynamic full truckload pickup and delivery problem with time 

windows. In this work, we explicitly address the role of terminals in designing a 

dispatching rule in which driving hours regulations are taken into account. We 

show how the model can be mathematically formulated and efficiently solved. 

■ SC49 


Joint Session TSL/SPPSN: From Emergency 

Preparedness to Debris Cleanup 




Chair: Changhyun Kwon, Assistant Profesor, University at Buffalo, 

SUNY, 400 Bell Hall, Buffalo, NY, 14260, United States of America, 

chkwon@buffalo.edu 

1 - Innovative Cycle Canceling Algorithm for Routing of Evacuees 

in Networks with Multiple Threat Zones 

Neema Nassir, Graduate Research Assistant, University of Arizona, 

Department of Civil Engineering and Engineering Mechanics, 

1209 E. Second Street, Tucson, AZ, 85721-0072, United States of 

America, neeman@email.arizona.edu, Mark Hickman, 

Yi-Chang Chiu, Hong Zheng 

An exact algorithm is developed and tested for finding the optimal routing of 

traffic flows, to evacuate a network with several threat zones, where the threat 

level may depend on the exposure or risk in each zone. A network 

transformation is proposed that facilitates detecting flow-augmenting paths, in 

which the cycles with the highest negative costs are quickly identified. 

2 - Dual-Toll Pricing and Drivers’ Preferences in Hazardous 

Materials Transportation Regulation 

Changhyun Kwon, Assistant Profesor, University at Buffalo, SUNY, 

400 Bell Hall, Buffalo, NY, 14260, United States of America, 

chkwon@buffalo.edu, Tolou Esfandeh, Masoumeh Taslimi, 

Rajan Batta 

A dual-toll setting policy is used as an effective and flexible regulation tool to 

encourage the carriers to route shipments on low risk paths. The proposed model 

not only aims to minimize the total risk on the network but also to consider the 

drivers’ preferences and priorities in route selection based on toll prices and travel 

time. Risk measurement model entails the risk in terms of overall accident 

probability and the consequences of the accident occurrence. 

3 - Robust Shelter Location in Geographic and 

Building Environments 

Elise Miller-Hooks, Professor, University of Maryland, 1173 Glenn 

L. Martin Hall, College Park, MD, 20742, United States of America, 

elisemh@umd.edu, Shabtai Isaac, Lei Feng, Reza Faturechi 

A bilevel, integer stochastic program is presented for the robust shelter location 

problem in geographic and building environments. At the upper level, the 

optimal location and size of shelters is chosen to support evacuation given an 

event requiring building occupants or individuals in the region to seek egress or 

protective covering. Individuals choose egress routes to these safe locations in the 

lower level to minimize disutility (travel time, exposure, geometric 

considerations, ...). 

4 - Post-disaster Debris Recycling and Disposal 

Alvaro Lorca, Georgia Institute of Technology, 765 Ferst Dr NW, 

Atlanta, GA, United States of America, alvarolorca@gatech.edu, 

Pinar Keskinocak, Ozlem Ergun 

The large volumes of debris generated by disasters can generate severe long term 

financial and environmental problems to the affected communities if not handled 

appropriately. We study the impact of strategic decisions in the collection and 


122 

disposal phases of post-disaster debris management such as recycling, collecting 

debris separately and incinerating debris, on the tradeoff between financial costs 

and environmental issues. 

■ SC50 


Raytheon (USA) and Brazilian Navy Defense 

Applications 



Chair: George Blaha, Senior Principal Systems Engineer, Raytheon 

Integrated Defense Systems, 2461 S. Clark St. #1100, Arlington, VA, 

22202, United States of America, gblaha@raytheon.com 

1 - Data Association using Fuzzy Knowledge 

Cleber Oliveira, DSc, Brazilian Navy, Ilha de Mocangue s/n, 

Niteroi, 24040-300, Brazil, cleber.almeida@casop.mar.mil.br 

This paper presents a maritime tactical picture, which was assembled by the 

association of the reports from surface targets provided by multiple sensors of a 

Brazilian vessel. The decision-making process conjugates the gating technique and 

the conceptual structure of decisions in a fuzzy environment to solve the data 

association problem considering four dimensions. 

2 - Leveraging Mission Analysis and Modeling to Investigate 

Conceptual CG VOI Location Decision Support 

David Procter, Raytheon Integrated Defense Systems, 

225 Presidential Way, Woburn, MA, United States of America, 

david_procter@raytheon.com, Fred Mott 

This paper presents the approach and results of a joint Raytheon - U.S. Coast 

Guard mission analysis and experiment to evaluate mission analysis-derived 

decision support concepts supporting USCG missions. The objective was to 

identify and investigate, via mission analysis and modeling, the potential of 

candidate decision support aids and CONOPS to enhance maritime security. 

Raytheon’s Mission Profiling (Patented) approach was applied to a conceptual 

vessel of interest (VOI) location application. 

3 - Sensor Scheduling for Space Event Identification 

and Characterization 

George Blaha, Senior Principal Systems Engineer, Raytheon 

Integrated Defense Systems, 2461 S. Clark St. #1100, Arlington, 

VA, 22202, United States of America, gblaha@raytheon.com, 

Christopher Cox, Michael Schmidt 

Timely identification and characterization of space events (e.g. breakups, 

maneuvers, conjunctions, dockings) requires effective scheduling of Space 

Surveillance Network (SSN) sensors. We present techniques for modeling event 

characterization peformance and quantifying the need for improved data 

collection. Combined with needs-based scheduling, these techniques may be used 

to develop effective observation schedules. Application to the current and 

planned SSN architecture is considered. 

■ SC51 


Joint Session MAS/SPPSN: Defense Operations and 

Homeland Security I 

Sponsor: Military Applications & Public Programs, Service and 

Needs 


Chair: Anna Doro-on, Sr. Consultant/Project Engineer, Water, Tunnel & 

Infrastructure Concepts, Inc., 120 S. Euclid Avenue, Pasadena, CA, 

91101, United States of America, anna@annaforhomelandsecurity.com 

1 - Military Support for Homeland Security: Iranian WMD 

Anna Doro-on, Sr. Consultant/Project Engineer, Water, Tunnel & 



anna@annaforhomelandsecurity.com, Martin Hershkowitz 

Iranian weapons of mass destruction (WMD) is a threat against United States, that 

has created the government to initiate both military and civil support. This paper 

demonstrates risk assessment based on cumulative prospect theory, analyzes the 

alternatives for effectiveness of military operations in the event of attacks for risk 

acceptability.

2 - Optimization Models for Improvements to Civil Infrastructure 

Systems Subject to Natural Hazards 

Brian Piper, North Carolina State University, 2500 Stinson Drive, 

Raleigh, NC, United States of America, bepiper@ncsu.edu, 

Ranji Ranjithan 

We present optimization models to improve the resilience of a civil infrastructure 

system (CIS) subjected to natural hazards. The engineering improvements 

considered reduce the risk to the CIS by either limiting consequences or reducing 

their associated probabilities. The uncertainties inherent to future natural hazards 

are represented by hazard scenarios. The models are applied to realistic data from 

a coastal community to demonstrate the value of a system-wide approach to the 

CIS improvements. 

3 - Dividing a Territory with Obstacles 

Raghuveer Devulapalli, Research Assistant, University of 

MInnesota, 310 8th St SE #205, MInneapolis, MN, 55414, India, 

devul002@umn.edu, John Carlsson 

We consider the problem of dividing a territory into sub-regions so as to balance 

workloads of $n$ fixed facilities over that territory. We assume that the territory 

is a simply connected polygon containing a set of simply connected obstacles. We 

give a fast analytic center cutting plane method that divides the territory into $n$ 

compact, connected sub-regions, each of which contains a facility, such that the 

workloads in each sub-region are balanced. 

4 - Optimal Selection of Countermeasures in IT Security Planning 

Tadeusz Sawik, Professor, AGH University of Science & 

Technology, Department OR & IT, Al.Mickiewicza 30, Krakow, 

30059, Poland, ghsawik@cyf-kr.edu.pl 

A mixed integer programming and conditional value-at-risk are applied for 

selection of countermeasure portfolio in IT security planning to prevent or 

mitigate cyber-threats. Given a set of potential threats and a set of available 

countermeasures, the selection of countermeasures is based on their effectiveness 

of blocking different threats, implementation costs and probability of potential 

attack scenarios. 

■ SC52 


Renewable Energy Integration Strategies 



Chair: Seyed Madaeni, PhD Candidate, The Ohio State University, 

210 Baker Systems Engineering Building, 1971 Neil Avenue, 

Columbus, OH, 43210, United States of America, madaeni.1@osu.edu 

1 - Optimal Location of Wind Power Capacity: A Point-estimate 

Solution Approach 

Salvador Pineda, Technical University of Denmark, DTU Electrical 

Engineering, Elektrovej, Kongens Lyngby, 2800, Denmark, 

spmo@elektro.dtu.dk, Juan Miguel Morales 

We present a tool for a generating company to optimally decide on the 

investment in new wind power capacity. The tool is formulated as a stochastic 

mathematical program with equilibrium constraints, which is hard to solve. We 

propose a point-estimate solution approach that effectively captures the 

information embedded in the probability distributions of the uncertain 

parameters, while remarkably trimming down the computational burden of the 

model. 

2 - A Nash Equilibrium Method to Control Plug-in Electric Vehicle 

Charging with Wind Integration 

Xiaomin Xi, The Ohio State University, 1971 Neil Avenue, 

Columbus, OH, 43210, United States of America, xi.12@osu.edu, 

Ramteen Sioshansi, Vincenzo Marano 

We apply a decentralized method using Nash Equilibrium to control plug-in 

electric vehicle (PEV) charging so as to reduce the negative effect of wind 

uncertainty. A economic dispatch model and a cost minimization problem are 

solved iteratively, in which SO sends scenario-based prices to a load aggregator 

(LA) that represents PEV owners, and LA submits bids by minimizing expected 

charging cost plus a penalty term. Convergence is assured under certain rules but 

is not necessarily social optimum. 

3 - Quantifying Variability and Uncertainty Metrics for Wind and 

Solar Power 

Patrick Sullivan, National renewable Energy Laboratory, Golden, 

CO, 80401, United States of America, Patrick.Sullivan@nrel.gov 

Using the ReEDS capacity-expansion model as a backdrop, we present statistical 

methods for quantifying variability and uncertainty of wind and solar power 

across an interconnected grid. Algorithms that estimate capacity value, operating 

reserve requirements, and curtailment levels describe the impact of resource 

variability on the electric sector and inform model investment decisions. 


123 

4 - Capacity Value of Photovoltaic Power 

Seyed Madaeni, PhD Candidate, The Ohio State University, 

210 Baker Systems Engineering Building, 1971 Neil Avenue, 

Columbus, OH, 43210, United States of America, 

madaeni.1@osu.edu, Paul Denholm, Ramteen Sioshansi 

In this study we focus on capacity value of Photovoltaic (PV) power and compare 

techniques that can be used for this purpose. These techniques consist of either 

reliability and statistical methods used to estimate the probability of a system 

outage event or simpler approximation techniques. We further investigate the 

effect of including subhourly data in our analysis. 

■ SC53 

SC53 


Real Options in the Energy Sector 



Chair: Afzal Siddiqui, University College London, Gower Street, 

London, United Kingdom, afzal.siddiqui@ucl.ac.uk 

1 - Economics of High-temperature Reactors for Industrial 

Cogeneration: A Utility’s Perspective 

Reinhard Madlener, Professor of Energy Economics and 

Management, RWTH Aachen University, Mathieustrasse, Aachen, 

52074, Germany, RMadlener@eonerc.rwth-aachen.de, 

Jona Hampe 

We study the economics of high-temperature reactors (HTR) for nuclear 

cogeneration, based on a business model developed from a utility’s perspective. 

For the economic evaluation, we perform a discounted cash flow (DCF) analysis, 

followed by a real options analysis (ROA) used to determine the value of the 

utility’s flexibility to react to market changes. We investigate both the option to 

delay the investment and the option to switch power plant operation from 

cogeneration to power-only mode. 

2 - Switching Options: Empirical Evidence from the 

Electricity Industry 

Stein-Erik Fleten, Professor, Norwegian University of Science and 

Technology, Industrial Economics and Technology Mgmt., 

Alfred Getz v 3, Trondheim, NO-7491, Norway, 

stein-erik.fleten@iot.ntnu.no, Carl Ullrich, Erik Haugom 

We examine empirically the real options to shutdown, startup, and abandon 

peaking power plants. The unique dataset details information for 1,337 individual 

plants for the period 2001–2009. Reduced form regressions provide evidence of 

real options effects for shutdown and abandonment decisions. Moreover, 

uncertainty about the outcome of ongoing deregulation in retail electricity 

markets significantly decreases the likelihood of shutting down operating plants. 

3 - Sequential Investment, Capacity Sizing, Dividing Flexibility 

Ryuta Takashima, Chiba Institute of Technology, Chiba, Japan, 

takashima@sun.it-chiba.ac.jp, Kimitoshi Sato, Yuta Naito 

We consider the problem of a typical investor who has discretion over not only 

the timing, but also the sizing of a new plant in sequential manner. We contrast 

the sequential investment strategies for different stage numbers in order to the 

value of flexibility. Additionally, we analyze the sequential investment for a case 

in which there exists a fixed cost in the investment one. The optimal stage 

numbers of sequential investment are obtained for various fixed coasts. 

4 - Investment in Alternative Energy Technologies under Physical 

and Policy Uncertainty 

Afzal Siddiqui, University College London, Gower Street, London, 

United Kingdom, afzal.siddiqui@ucl.ac.uk, Ryuta Takashima 

Policymakers have often backed alternative energy technologies, e.g., nuclear 

power, due to their relatively low operating costs and emissions. However, they 

have also been quick to respond to public perceptions about the safety of such 

plants by suspending construction or even decommissioning existing facilities. We 

address public concerns about physical plant risks along with stochastic market 

prices for energy in modelling investment and decommissioning of alternative 

energy technologies.

SC54 

■ SC54 


Memorial Sloan-Kettering Cancer Center - 2012 

INFORMS Prize Winning Organization 

Cluster: 2012 INFORMS Prize 


Chair: Michael Gorman, University of Dayton, Dayton, OH, United 

States of America, michael.gorman@udayton.edu 

1 - Memorial Sloan-Kettering Cancer Center - 2012 INFORMS Prize 

Winning Organization 

Sasha Bartashnik, Memorial Sloan-Kettering Cancer Center, 

405 Lexington Avenue, 3rd Floor, New York, NY, 10174, 

United States of America, bartasha@mskcc.org 

Memorial Sloan-Kettering Cancer Center (MSK) is the oldest and one of the 

largest cancer hospitals in the world - over 25,000 patients come to MSK for 

treatment every year. The changing health care landscape presents a unique 

challenge for hospital administrators in providing better quality care at lower cost. 

Increasing demand against tighter resource constraints reinforces the institution’s 

focus on their mission: better outcomes for patients. These challenges, as well as 

MSK’s high volume of longitudinal care, create an opportunity to develop a 

fundamental platform for using quantitative methods to gain operational, 

strategic, and clinical insights. In 2006 MSK established the department of 

Strategic Planning and Innovation (STPI) based on the idea that better patient 

care means finding financially sustainable, patient-centered solutions. By working 

at the intersection of leadership, clinical staff, and physicians, STPI integrates 

OR/MS concepts within the hospital. Quantitative analysis, industrial engineering 

and design research teams collaborate to find innovative approaches to some of 

the hospital’s most difficult challenges, including improving process flow, 

modeling resource utilization, and introducing a volume forecasting system. MSK 

continues to push the frontiers of cancer care as it partners with IBM to use 

powerful machine learning technology to develop a comprehensive evidencebased 

medical intelligence engine. 

■ SC55 


Healthcare Applications 



Chair: Janny Leung, Professor, Chinese University of Hong Kong, 

System Eng. and Engineering Management, Shatin, New Territories, 

Hong Kong - PRC, jleung@se.cuhk.edu.hk 

1 - A Tracking Study in Medical Wards using RFID 

Ziye Zhou, The Chinese University of Hong Kong, Dept. of Syst. 

Eng. & Eng. Mgmt., Shatin, NT, Hong Kong, Hong Kong-PRC, 

zhouzy@se.cuhk.edu.hk, Chun-Hung Cheng, Dorbin Tobun Ng 

A large-scale tracking pilot study was conducted in two 40-bed Medical wards in 

a Hong Kong hospital since Fall 2011 for 6 months. This work is to illustrate how 

the tracking data were analyzed to support data-driven activity analysis & personto-person 

contact tracing in the hospital ward environment. 

2 - Evaluating Impacts of Staffing Policies on Patient Flows in an 

Emergency Department using Simulation 

Yong Hong Kuo, The Chinese University of Hong Kong, SEEM 

Dept,Chinese Uni. of Hong Kong, Shatin, New Territories, 

Hong Kong, NT, Hong Kong-PRC, yhkuo@se.cuhk.edu.hk, 

Benedetta Lupia, Omar Rado, Colin Graham, Janny Leung 

We studied impacts of staffing policies on patient flows in an Emergency 

Department in Hong Kong. We built a simulation model of the Emergency 

Department, which captures all relevant treatment processes, complicated arrival 

rates and adjustable staff deployment. To tackle a challenge of data 

incompleteness, we estimated required parameters in our model with the help of 

meta-heuristics. We will also present some insights and suggestions for 

operational improvements by using our model. 


124 

■ SC56 


Entrepreneurship and Innovation I 

Sponsor: Technology Management 


Chair: Sinan Erzurumlu, Assistant Professor, Babson College, 

231 Forest Park, Babson Park, MA, 02457, United States of America, 

serzurumlu@babson.edu 

1 - Optimal Bargaining Power in an Entrepreneurial Market 

Process 

Mohammad Keyhani, York University, 4700 Keele St., Schulich 

School of Business, Toronto, On, M3J 1P3, Canada, 

mkeyhani08@schulich.yorku.ca, Moren Lévesque 

We investigate the role of bargaining power in the economic returns to agents 

engaged in repeated cooperative games over time representing an entrepreneurial 

market process in line with the Austrian economics of entrepreneurship. We find 

that as expected, players with greater bargaining power accrue greater returns 

under various conditions, but under certain conditions closer to the Kirznerian 

school of thought the pattern is more complicated and not necessarily monotonic. 

2 - Are Innovative Companies also More Interactive? Evidence 

from Company Websites 

John Angelis, Rochester Institute of Technology, 105 Lomb 

Memorial Drive, Rochester, NY, United States of America, 

jangelis@saunders.rit.edu, Rajendran SriramachandraMurthy 

For innovative firms, interaction with customers may lead to improved 

innovation, but also may lower innovation efficiency. We assess firms ranked by 

innovation premium in Forbes Magazine’s list of Top 100 Innovative Small-Cap 

Firms. Each website is evaluated on interaction opportunities for customers based 

on such areas as educational opportunities, co-creation options, and social media 

opportunities. We find strong industry-related impacts on innovation types 

selected by firms. 

3 - Growth of New Business Ventures through Innovation 

Replicability 

Juliana Hsuan, Professor, Copenhagen Business School, 

Department of Operations Management, Solbjerg Plads 3, 

Frederiksberg, DK-2000, Denmark, jh.om@cbs.dk, 

Moren Lévesque 

We propose a formal model of firm growth through replication that considers the 

extent of the investment to adapt routines as replication unfolds and the portion 

of this investment that goes toward innovation in the routines. Two growth 

policies for new business ventures emerge: bull-imitation and bear-innovation. 

4 - Operations Design to Enhance ARPA-E Funding for 

Transformational Clean Technology Start-ups 

Sinan Erzurumlu, Assistant Professor, Babson College, 231 Forest 

Park, Babson Park, MA, 02457, United States of America, 

serzurumlu@babson.edu, Jane Davies, Nitin Joglekar 

This paper analyzes the role of operations design in securing funding for highly 

risky projects. We use data from 36 transformational innovative clean technology 

projects to examine to what extent the operational decisions can mitigate risk and 

enhance firm valuation during clean technology start-up. We find that operations 

design for risk reduction induces more funding, but design for market 

competitiveness shows significant negative correlation to level of funding. 

■ SC57 


Queueing Systems 


Chair: Philippe Chevalier, Professor, UCLouvain, core, 

34 Voie du Roman Pays, Louvain-la-Neuve, 1348, Belgium, 

philippe.chevalier@uclouvain.be 

1 - Interpolation Approximations for Many Single Queues in Series 

Kan Wu, Assistant Professor, Nanyang Technological University, 

50 Nanyang Ave., School of MAE, Singapore, 639798, Singapore, 

kan626@gmail.com, Leon McGinnis 

We propose a new approximation approach, based on observed properties of the 

behavior of tandem queues, which we call the intrinsic gap and intrinsic ratio. 

The approach exploits what we call the nearly-linear and heavy-traffic properties 

of the intrinsic ratio, which appear to hold in realistic production situations. 

Across a broad range of examined cases, this new approach outperforms earlier 

approximation methods and offers a way to analyze the dependence among 

servers.

2 - A Queueing System with Service Phases of Random Length 

and Vacations 

George Mytalas, Athens University of Economics and Business, 

76 Patission Str., Athens 104 34, Athens, Greece, mytalas@aueb.gr, 

Michael Zazanis 

We examine a queueing model with Poisson arrivals, vacations and service phases 

of random length with distributions with bounded or unbounded support. 

Standard techniques are used involving RouchÈ’s theorem and Wiener-Hopf 

factorization techniques. We obtain our results under minimal assumptions, 

namely the finiteness of the first moment of the service and vacation distributions 

together with the stability condition. 

3 - Real-time Delay Estimation in Multi-class Many-server 

Queueing Systems 

Zinan Yi, PhD Student, North Carolina State University, Room 304, 

1430 Collegeview Ave., Raleigh, NC, 27606-3133, United States of 

America, zyi@ncsu.edu, Yunan Liu, Rouba Ibrahim 

We develop real-time delay estimators for multi-class queueing networks with 

general service and patience distributions, time-varying arrival rates and staffing 

functions, skill-based routing and global first-come-first-served service discipline. 

To construct these estimators, we adopt the corresponding many-server heavytraffic 

fluid models. Computer simulation experiments verify that these estimators 

are effective. 

4 - Efficient Queuing Network Structures with Limited Flexibility 

Kuang Xu, Massachusetts Institute of Technology, 77 


kuangxu@mit.edu, John N. Tsitsiklis 

We study a multi-server model where n flexible servers are connected to rn 

queues through a fixed connectivity graph, with average degree d. We show that 

as n tends to infinity, both a diminishing delay and a large capacity region are 

jointly achievable, even under limited flexibility (d

SC60 

■ SC60 


Efficient Models and Algorithms for Air Traffic 

Flow Management 



Chair: Vikrant Vaze, Member Research Staff, Philips Research North 

America, 345 Scarborough Road, Briarcliff Manor, NY, 10510, 

United States of America, vikrant.vaze@philips.com 

1 - Equipage-based Resource Allocation in Air Traffic Flow 

Management 

Shervin AhmadBeygi, Lead Operations Research Analyst, Metron 

Aviation Inc., 45300 Catalina Ct, Dulles, VA, 20166, United States 

of America, shervin.ahmadbeygi@metronaviation.com 

Successful implementation of the Next Generation Air Transportation System 

(NextGen) requires a critical mass of aircraft to be equipped with a range of 

communication, navigation, and surveillance avionics. In this research we study 

different mechanisms to incorporate equipage information in Traffic Flow 

Management resource allocation algorithms. 

2 - A Decomposition Approach for the Stochastic ATFM Problem 

L. Corlli, University of Milano-Bicocca, Department Informatics 

Systems Communication, Milano, 20126, Italy, 

corolli@disco.unimib.it, Lewis Ntaimo, Guglielmo Lulli 

The stochastic Air traffic Flow Management (ATFM) problem is now attracting a 

lot of interest from the research community. In this talk, we present a two-stage 

stochastic integer programming formulation for the problem, which balance the 

trade-off between the complexity of the model and its representation of reality. To 

solve the problem, we present a decomposition approach to take advantage of the 

underlying structure of the problem. 

3 - Applying Air Traffic Flow Management Optimization Models to 

Railroad Train Scheduling 

Lingyun Meng, School of Traffic and Transportation, 8507D, No.8 

Teaching Building, No.3 ShangYuanCun, HaiDian District, China, 

lymeng@bjtu.edu.cn, Xuesong Zhou 

By analyzing a variety of air traffic flow management models, this paper proposes 

a cumulative flow count based optimization model for train scheduling problem 

under stochastic segment capacity breakdowns. The “capacity aggregation” 

technique is used to capture the uncertainty of capacity breakdown incidents. A 

Lagrangian relaxation based solution approach is presented to solve this model. 

Numerical experiments are conducted to demonstrate the performance of the 

proposed solution approach. 

4 - Application and Evaluation of Bibliometrics Methods for 

ATM Researches 

Hidenori Chida, University of Tokyo, Hongo 7-3-1, Bunkyo-ku, 

Tokyo, Japan, hidenori.chida@gmail.com, Katsuhiro Nishinari, 

Hiroko Nakamura 

Innovation in ATM (air traffic management) is expected to afford rapidly 

increasing air traffic more efficiently and safely. To implement the innovation, 

interdisciplinary cooperation is required in ATM researchers such as with civil 

engineering or navigation research of other transportation. Therefore, we applied 

and evaluated a bibliometrics method to get overviews of areas unknown to ATM 

researchers, and to support effective and strategic ATM development. 

■ SC61 


Airport Operations 



Chair: Mourad Boudia, Research Engineer, Amadeus, 485, 

Route du Pin Montard BP 69, Sophia Antipolis, 06902, France, 

mourad.boudia@amadeus.com 

1 - Pre-departure Sequence Planning 

Mourad Boudia, Research Engineer, Amadeus, 485, 

Route du Pin Montard BP 69, Sophia Antipolis, 06902, France, 

mourad.boudia@amadeus.com, Olivier Ratier, Baptiste Chatrain 

High performance departure sequence planning contributes to improve air traffic 

and aircraft operations at airports by maximizing usage of available runways. We 

present here an innovative approach based on linear programming and discuss its 

contribution and performances. 


126 

2 - Aircraft Stand Allocation with Associated Resource Scheduling 

Tor Fog Justesen, Industrial PhD Student, Copenhagen Airports 

A/S, Lufthavnsboulevarden 6, P.O. Box 74, Kastrup, 2770, 

Denmark, tor.justesen@cph.dk, Richard Lusby, Jesper Larsen, 

David Ryan 

When handling an aircraft turnaround different ground handling resources are 

required at different times. Each resource can be claimed by at most one 

turnaround at a time. In this project we develop a set partitioning model 

formulation of the aircraft stand allocation problem with associated resource 

scheduling and an optimization based solution algorithm in which knowledge 

about the structure of the problem is exploited to generate good feasible solutions. 

3 - Maximizing Resilience of Airport Runway and Taxiway 

Pavement Networks through Stochastic Programming 

Reza Faturechi, PhD Candidate, University of Maryland, 1173 

Glenn L. Martin Hall, College Park, MD, 20742, United States of 

America, reza@umd.edu, Xiaodong Zhang, Eyal Levenberg, 

Elise Miller-Hooks 

A two-stage binary integer, stochastic program and integer L-shaped method 

proposed for its solution are presented for assessing and maximizing the resilience 

of an airport given the possibility of future, randomly arising events affecting 

airport runway and taxiway pavement networks. Their application provides 

insights into optimal allocation of limited resources for response and preparedness 

actions that facilitate them. The approach is demonstrated on a case study 

involving Laguardia Airport. 

4 - Stand Allocation in Airports - New Solution Approaches 

and Results 

Rodrigo Acuna Agost, Operations Research-Innovation, Amadeus, 

485 Route du Pin Montard, Sophia Antipolis, 06902, France, 

rodrigo.acunaagost@amadeus.com, Mourad Boudia, Thierry 

Delahaye, Julien Guepet, Daniel Perez 

The Stand Allocation Problem consists in finding an assignment of aircraft 

turnarounds to parking positions. The usual objectives include maximizing the 

number of passenger assigned to contact stands, and minimizing the number of 

tows. In this talk we show new methods including MIP-based, CP-based, Column 

Generation, and metaheuristics approaches. The results of experiments show 

significant improvements on several KPIs compared to the current practice on real 

data from two European airports. 

■ SC62 


Automated Analysis and Design of Auctions 

Cluster: Auctions 


Chair: David Thompson, PhD Student, University of British Columbia, 

191-2366 Main Mall, Vancouver, BC, V6T 1Z4, Canada, 

davidrthompson@gmail.com 

1 - Automated Mechanism Design, with Applications to Auctions 

Yevgeniy Vorobeychik, Sandia National Labs, 7011 East Ave., 

Livermore, CA, 94550, United States of America, 

eug.vorobey@gmail.com 

We present general methods for automated mechanism design when both the 

design, and player strategy spaces are constrained. We offer Bayesian and worstcase 

(“robust”) formulations, and present a technique based on stochastic search. 

Finally, we demonstrate the effectiveness of our methods in a variety of auction 

settings. 

2 - Automated Mechanism Design using CAT Games 

Jinzhong Niu, Postdoctoral Research Associate, The City College of 

New York, 160 Convent Avenue, New York, NY, 10031, 

United States of America, jniu@ccny.cuny.edu, Simon Parsons 

In this paper, we present a genetic algorithm-based approach to automated design 

of complete auction mechanisms in the domain of CAT games. In a CAT game, 

multiple players, each providing his/her own design of a double-sided auction, 

compete against each other to attract traders for more profit and high transaction 

success rate. We show that this evolutionary method is effective in creating 

auction mechanisms in the CAT setting. 

3 - Revenue Optimization and the Generalized Second-price 

Auction 

David Thompson, PhD Student, University of British Columbia, 

191-2366 Main Mall, Vancouver, BC, V6T 1Z4, Canada, 

davidrthompson@gmail.com 

Search engines generate billions of dollars in advertising revenue using the 

generalized second-price auction (GSP). They have tried to further increase 

revenue using GSP variants including different kinds of reserve prices, and 

quality-score squashing. Our contributions are a comparison of different reserve 

price regimes (with and without squashing) and a study of how equilibrium 

selection affects the revenue generated by these variants.

4 - Profit-charging Market Makers with Bounded Loss, Vanishing 

Bid/Ask Spreads, and Unlimited Depth 

Tuomas Sandholm, Carnegie Mellon University, Computer Science 

Dept., Pittsburgh PA , United State of America, 

sandholm@cs.cmu.edu, Abe Othman 

It was an open question whether a market maker can simultaneously satisfy 

bounded loss, the ability to make a profit, a vanishing bid/ask spread, and 

unlimited market depth. We design market makers that do. We achieve this by 

introducing a practical framework that extends constant-utility cost functions 

with two functions that are added to the prices quoted to the trader: the liquidity 

function uses its proceeds to increase the liquidity, and the profit function is a 

“lockbox” of savings. 

■ SC63 


Joint Session Behavioral/HAS: Antecedents and 

Consequences of Worker and Patient Behaviors in 

Health Care Settings 

Sponsor: Behavioral Operations & Health Applications Society 


Chair: Anita Tucker, Associate Professor and Marvin Bower Fellow, 

Harvard University, 413 Morgan Hall, Soldiers Field, Boston, MA, 

02478, United States of America, atucker@hbs.edu 

Co-Chair: Reidar Hagtvedt, Assistant Professor of Management Science 

and Statistics, University of Alberta School of Business, 2-43 Business 

Building, Edmonton, AB, T6G2C7, Canada, 

reidar.hagtvedt@business.ualberta.ca 

1 - Speaking Up About Internal Supply Chain Problems: 

Experiments on Medication Administration 

Anita Tucker, Associate Professor and Marvin Bower Fellow, 

Harvard University, 413 Morgan Hall, Soldiers Field, Boston, MA, 

02478, United States of America, atucker@hbs.edu 

Employees who experience internal supply chain problems can provide valuable 

information about improvement opportunities. To understand the conditions 

under which employees are more likely to speak up about problems, we ran 

experiments using medication administration tasks for which we purposely 

created problems. Nurse spoke up under two conditions: (1) to obtain materials if 

there was no other way to work around the issue and (2) when their role identity 

was primed to include improvement. 

2 - Desensitization to Stimuli and Optimal Expenditures to Increase 

Hand-hygiene in Hospitals 

Reidar Hagtvedt, Assistant Professor of Management Science and 

Statistics, University of Alberta School of Business, 

2-43 Business Building, Edmonton, AB, T6G2C7, Canada, 

reidar.hagtvedt@business.ualberta.ca, Sarah Forgie 

Hand hygiene in hospitals is typically lower than recommended by the WHO. One 

strategy for change is to run educational campaigns, although empirical studies 

show health-care workers are aware of the importance of hand-hygiene. Such 

campaigns eventually recede into the background. We have purchased sensors for 

gel dispensers, in order to establish baseline usage and forgetting rates, which will 

then serve as input to a model to optimize resources used to increase handhygiene 

compliance. 

3 - Multi-server Systems with Adaptive Server Behavior 

Mohammad Delasay, Alberta School of Business, University of 

Alberta, Edmonton, AB, T6G2R6, Canada, delasays@ualberta.ca, 

Armann Ingolfsson, Bora Kolfal 

We investigate service rates in the Calgary EMS system and show that service 

rates decrease with the number of busy ambulances. Then, we model and analyze 

systems in which service rates change with “load” (the number of busy servers) 

and “overwork” (the number of users served in the current full busy period). We 

demonstrate how ignoring adaptive server behavior results in inconsistencies 

between the planned and realized performance. 

4 - Health Care Supply Chain for Behavioral Health: Evaluating 

Community Environment and Primary Care 

David Zepeda, Assistant Professor, University of Minnesota, 

Minneaplis, MN, United States of America, zepe0003@umn.edu, 

Kingshuk K. Sinha 

One in four people experience a behavioral health condition in their lifetime. 

Making treatment affordable and cost effective requires the integration of 

behavioral and physical care through primary care. Yet, the community 

environment can also influence behavioral health. Using clinic and community 

level data we evaluate the effects of the community environment (physical and 

social) and quality primary care (technology-enabled, evidence based, and 

affordable) on improving behavioral health. 


127 

■ SC66 


Data Mining and OR in Health Application 



Chair: Sung Won Han, Fellow, Hoffmann-La Roche Inc., 340 Kingsland 

St., Nutley, NJ, United States of America, sung_won.han@roche.com 

1 - A Comparison of MCUSUM-based and MEWMA-based 

Surveillance Methods in Non-homogeneous Population 



sung_won.han@roche.com 

Motivated by applications in healthcare surveillance, we studied the 

spatiotemporal surveillance problem of detecting the mean change of Poisson 

count data in non-homogeneous populations. We investigate several likelihood 

ratio-based approaches and EWMA-based methods and compare them under 

various scenarios. We discuss under which scenarios the EWMA-based methods 

are more robust than the likelihood ratio-based approaches. 

2 - Dynamic Spatiotemporal Warping for the Detection and 

Localization of Heart Attacks 



Myocardial infarction (MI), also known as heart attack, is the leading cause of 

death. This paper presents a novel warping approach to quantify the dissimilarity 

of disease-altered patterns in 3-lead VCG signals. The hypothesis testing shows 

there are significant space-time signal pattern variations between healthy controls 

(HC) and various MIs. The new approach yields an accuracy of 94.7% for 

separating MIs and HCs and an accuracy of 96.5% for anterior-related MIs and 

inferior-related MIs. 

3 - Predicting Sequential Health Events 

Benjamin Letham, Massachusetts Institute of Technology, 

Operations Research Center, Cambridge, MA, United States of 

America, bletham@mit.edu, David Madigan, Cynthia Rudin 

Using a database of sequences of health events, we wish to learn a predictor that 

can use a patient’s current medical history to predict future health events. We use 

ideas from supervised ranking to develop a framework for sequential predictions 

that allows for a variety of modeling choices. We apply our method to health 

event data from a clinical trial and demonstrate good performance. 

4 - Contact Network Based Risk Assessment to Prevent 

Potential Pandemics 

Shengpeng Jin, Teaching Asistant, University of Louisville, 782 

Frederick Stamm Ct. Apt #4, 4, Louisville, KY, 40217, United 

States of America, jinshengpeng@gmail.com, Suraj Alexander 

Early information is crucial for policy makers and public health officials. Current 

indicators of the spread of contagious outbreaks lag behind the actual spread of 

the epidemic, leaving no time for a planned response. Our focus, however, is on 

the effective control of the spread of contagious outbreaks in their early stages. In 

this paper, we use information from “central” individuals in “contact” networks 

for early “spatial - temporal” prediction of virulent contagious outbreaks. 

■ SC67 

SC67 


Data Fusion for Manufacturing System Performance 

Improvement II 



Chair: Ran Jin, Assistant Professor, Virginia Tech., 1145 Perry St., 

Blacksburg, VA, 24060, United States of America, jran5@vt.edu 

Co-Chair: Jian Liu, Assistant Professor, The University of Arizona, Rm 

221 ENGR Building, 1127 E. James E. Rogers Way, Tucson, AZ, 85721, 

United States of America, jianliu@sie.arizona.edu 

1 - Surface Variation Monitoring by Integrating Spatial Data with 

Cutting Force Models 

Hui Wang, University of Michigan, Ann Arbor, MI, 

United States of America, huiwz@umich.edu 

Variation control of surface shape becomes a key enabler of high-precision 

machining. This talk presents a method for efficiently measuring and monitoring 

variations of machined surfaces. Cutting force modeling is conducted and is 

correlated with surface spatial data to improve surface profile prediction. Using 

the prediction, a strategy under Bayesian framework is proposed to monitor 

surface variations without significantly increasing false alarms. A Ford case study 

demonstrates the method.

SC68 

2 - The Use of Spatial Statistical Analysis for Image Based 

Part Inspection 

Jaime Camelio, Associate Professor, Virginia Tech, 235 Durham 

Hall, Blacksburg, VA, United States of America, jcamelio@vt.edu, 

Lee Wells 

Traditional quality control tools are based upon extracting features and measuring 

discrete dimensional data. However, new data-sets that are highly heterogeneous 

(multiple sensor types measuring different phenomena) and data-rich (thousands 

or even millions of data points). This presentation will use a case study of highspeed 

micro-scale part inspection using digital images and 3D laser scanning data. 

A cross-sensor spatial inspecting technique will be presented and its performance 

analyzed. 

3 - Informative Sensor and Feature Selection via Hierarchical Non- 

Negative Garrote 


United States of America, kamip@umich.edu 

In order to effectively extract the information from a distributed sensing system, it 

is essential to develop a systematic method that not only can select important 

sensors/signals, but also can extract a low dimension of interpretable features 

from the high-dimensional vector of a selected signal. For this purpose, we 

develop a new hierarchical regularization approach called hierarchical nonnegative 

garrote. Performance of the method is evaluated through simulations 

and a case study. 

4 - Damage Identification for High-Speed Train Gearbox Shell 

Material Base on Acoustic Emission Technology 

Weidong Zhang, University of Science & Technology Beijing, 

30 Xueyuanroad, Haidian District, Beijing, 100083, China, 

zwd@ustb.edu.cn, Xiwen Zhang 

The failure of high-speed train gearbox shell material is very important for the 

safety of high-speed train.In this paper,we use acoustic emission technology to 

monitor the the degree of damage in high-speed train gearbox shell material.The 

research work includes two main aspects: the data pre-processing method for 

imbalanced dataset and SVM modification method for imbalanced dataset. Then 

we apply these methods in the damage identification of high-speed train gearbox 

shell material. 

5 - Engineering Driven Force Model Enhancement for the Laser 

Assisted Micro Milling (LAMM) Process 

Chia-Jung Chang, Georgia Institute of Technology, 765 Ferst Drive, 

Main Building, Office 217, Atlanta, GA, 30332-0205, 

cchang43@gatech.edu, Mukund Kumar, Roshan Vengazhiyil, 

Shreyes N. Melkote 

LAMM is capable of generating three dimensional microscale features in hard 

metals. An engineering-driven force model that captures the effect of thermal 

softening due to laser heating has been developed. We proposed a systematic way 

to estimate unknown model parameters with experimental force measurements. 

This model provides accurate force predictions over a range of production 

conditions in a timely manner which facilitates the LAMM process optimization 

to yield the maximal force reduction. 

■ SC68 


Finance: Theory and Empirics 


Chair: Yunchao Xu, National University of Singapore, Mochtar Riady 

Building, BIZ 1 8-69, 15 Kent Ridge Drive, Singapore, 119245, 

Singapore, bizxuy@nus.edu.sg 

1 - Asset Price Bubbles in Experimental Markets 

Vinod Cheriyan, Georgia Institute of Technology, 

755 Ferst Drive NW, Atlanta, GA, 30332, United States of America, 

vinod.cheriyan@gatech.edu, Anton Kleywegt, Federico Bonetto 

Asset price bubbles have been observed in experimental markets. However, in 

most of these studies the supply and demand of assets are specified exogenously. 

Also, their finite duration causes end of the horizon effects. In this work, we 

design an experimental market with endogenous supply and demand, and 

random duration. We show that asset price bubbles are possible in such markets. 

We present findings from our experiments. We also fit our model and other 

models of investor behavior to the data. 

2 - How to Make Optimal Instruments for GMM More Efficient 

Stephen Goldberger, University of North Carolina at Chapel Hill, 

Chapel Hill, NC, United States of America, sgoldber@live.unc.edu 

For instrumental variables estimation of time series models with martingale 

difference sequence errors the optimal instruments are known but unobserved. 

We use mixed frequency data in order to create improved feasible optimal 

instruments, which in turn leads to more efficient parameter estimation. 


128 

3 - Minimum Quadratic Benchmark Regret in Portfolio Optimization 

Yunchao Xu, National University of Singapore, Mochtar Riady 

Building, BIZ 1 8-69, 15 Kent Ridge Drive, Singapore, 119245, 

Singapore, bizxuy@nus.edu.sg, Zhichao Zheng, Chung Piaw Teo, 

Karthik Natarajan 

We propose a minimum quadratic benchmark regret model to decide optimal 

portfolio weights, which combines flavor of traditional regret and tracking 

models. We provide closed-form expressions of the optimal portfolio weights, 

with or without consideration of transaction volumes, when the risky asset 

returns follow a multivariate normal distribution. Our portfolios demonstrate 

superior out-of-sample performance compared to numerous portfolio strategies 

studied in the literature. 

■ SC69 


Portfolio Management in Frictional Markets 



Chair: Gerry Tsoukalas, University of Pennsylvania, The Wharton 

School, Philadelphia, PA, United States of America, 

gtsouk@wharton.upenn.edu 

Co-Chair: Dan Iancu, Assistant Professor, Graduate School of Business, 

Stanford University, Stanford, CA, 94305, United States of America, 

daniancu@stanford.edu 

1 - Fairness and Efficiency in Multiportfolio Optimization 

Nikos Trichakis, Assistant Professor, Harvard Business School, 

Boston, MA, United States of America, ntrichakis@hbs.edu, 

Dan Iancu 

When a manager is in charge of multiple accounts, the performance of each 

account depends on the strategies of others due to market impact. We propose a 

novel approach for jointly optimizing the trading activities of all accounts and also 

splitting the associated market impact costs. Our approach allows one to balance 

the conflicting objectives of maximizing aggregate gains and distributing them 

fairly. We perform numerical studies that suggest that our approach outperforms 

existing methods. 

2 - Optimal Dynamic Portfolio Construction with Transaction Costs 

John Birge, Professor, University of Chicago, Chicago, IL, 

United States of America, john.birge@chicagobooth.edu 

Challenges for practical portfolio optimization include both time-varying return 

distributions and the presence of transaction costs. We consider simultaneous 

portfolio optimization and parameter estimation in a GARCH model with 

proportional transaction costs using consistent estimates of the opportunity cost of 

sub-optimal positions. 

3 - Estimation Risk in Dynamic Portfolio Optimization 

Alberto Martin-Utrera, Mr, University Carlos III of Madrid, 

C/ Madrid, 126-28903, GETAFE, Sp, 28903, 

amutrera@est-econ.uc3m.es, Francisco J. Nogales, Victor DeMiguel 

We consider an investment decision problem for an investor whose objective 

function is to maximize her lifetime utility in the presence of trading costs. We 

study the effects of estimation error of the estimated optimal trading strategy 

within the investor’s expected utility, and we propose several shrinkage methods 

to mitigate the effects of estimation error.

■ SC70 


Theoretic Approach and Optimal 

Coordination Policies 



Chair: Monica Johar, Assistant Professor, University of North Carolina 

at Charlotte, 9201 University City Blvd, Charlotte, NC, 

United States of America, msjohar@uncc.edu 

1 - Optimal Information System Security Investment – A Control 

Theoretic Approach 

Jing Zhou, Assistant Professor, University of North Carolina- 

Charlotte, 9201 University City Blvd, Friday 353C, Charlotte, NC, 

28223, United States of America, jzhou7@uncc.edu, Xun Li, 

Monica Johar 

Most of the IT security investment literature is based on single period analysis. In 

addition to the initial investment, a firm continues to exert effort to maintain the 

effectiveness of the IT infrastructure. We use a control theoretic framework to 

find the optimal IT security investment to minimize the expected total cost over a 

planning horizon. Our results can help firms decide how to invest in IT security 

given the type of the information system and the efficiency of the investment. 

2 - Optimal Coordination Policies in Distributed 

Software Development 

Vijay Mookerjee, University of Texas at Dallas, 800 W. Campbell 

Road, Richardson, TX, 75080, United States of America, 

vijaym@utdallas.edu 

We study the problem of designing optimal coordination policies in distributed 

software development (DSD). Coordination in DSD can be both within one site 

and across different sites. The latter type coordination involves communication 

across spatial boundaries (different locations) and possibly temporal boundaries 

(different time zones) and is the main challenge that DSD faces. We consider both 

symmetric and aymmetric sites with randomness. 

3 - Analyzing the Influence of Minimum Requirements on Coupon 

Selling in Groupon’s Daily Deal 

Gang Wang, University of Connecticut, 2100 Hillside Road, Storrs, 

CT, United States of America, gang.wang@business.uconn.edu, 

James Marsden, Xue Bai, Bill Ross 

Minimum requirements have been one of the key features in Groupon’s daily 

deal setting in which a certain number of coupons must be bought to activate the 

deal. Comparing two groups of Groupon deals, those with and those without 

minimum requirements, we isolate the influence of minimum requirements on 

social sharing, indicated by Facebook likes, and then analyze possible impacts on 

coupon sales and revenue for various retail categories. 

■ SC71 


Economics of Information Systems 



Chair: Mingfeng Lin, University of Arizona, 1130 E. Helen St, Tucson, 

AZ, 85721, United States of America, mingfeng@eller.arizona.edu 

1 - Optimal Choice for Market Price Information 

Chris Parker, Pennsylvania State University, Smeal College of 

Business, University Park, PA, 16802, United States of America, 

chris.parker@psu.edu, Kamalini Ramdas, Nicos Savva 

Information is only valuable if it is both new and actionable. We model a seller’s 

selection of a market for which to purchase price information. We discuss the 

results of the model in the context of farmers in rural India and use data to 

analyze whether actual decisions are in line with our model. 

2 - The Impact of IT on Economic Capacity 

Dawei Zhang, University of Calgary, Haskayne School of Business, 

Calgary, Canada, dzhang@ucalgary.ca, Barrie Nault 

We define economic capacity as the output level that maximizes short-run 

industry profit, and empirically explore the impact of IT on economic capacity and 

CU under a production function framework. 


129 

3 - Content Sharing in a Social Broadcasting Environment: 

Evidences from Twitter 

Huaxia Rui, University of Texas at Austin, Austin, TX, 

United States of America, huaxia@utexas.edu 

We collect a detailed dataset about the information-sharing activity on Twitter, 

called retweet, and estimate an econometric model regarding tie strength and 

retweeting behavior. The empirical results convincingly support our model and 

we find that after an author posts a median quality (as defined in the sample) 

tweet, the likelihood that a unidirectional follower will retweet is significantly 

higher than the likelihood that a bidirectional follower will. 

4 - Geography in Online Peer-to-peer Lending 

Mingfeng Lin, University of Arizona, 1130 E. Helen St., Tucson, 

AZ, 85721, United States of America, mingfeng@eller.arizona.edu, 

Siva Viswanathan 

We study how geography affects investors’ choice of loans to fund using data 

from Prosper.com, one of the largest online peer-to-peer lending websites. Using 

data from regular market conditions and from a natural experiment, we show 

that geography affects lender decisions, but this relationship is moderated by the 

nature of the loan request and the salience of geography in the marketplace. 

■ SC72 

SC72 


Resource Management 



Chair: Natarajan Gautam, Associate Professor, Texas A&M University, 

3131 TAMU, College Station, TX, 77843, United States of America, 

gautam@tamu.edu 

1 - Efficient Management of Demand Loads in Cloud Computing 

Environments 

Samyukta Sethuraman, Texas A&M University, 3131 TAMU, 

College Station, TX, 77843, United States of America, 

samyukta@tamu.edu, Natarajan Gautam, Lewis Ntaimo 

Energy efficiency can be achieved in cloud computing environments by cleverly 

assigning applications to servers, routing requests to appropriate servers, and 

speed-scaling of servers. We formulate and solve an optimization problem that 

minimizes expected energy consumption subject to satisfying QoS constraints. We 

develop a hybrid model that combines a deterministic optimization model and a 

stochastic fluid flow analysis in a unified manner. 

2 - A Method to Generate Image for SPC by Point Cloud Data 

Ketai He, Associate Professor, University of Science and 

Technology Beijing, No. 30, Xueyuan Rd., Haidian Dist., Beijing, 

100083, China, ketaih@vt.edu, Jaime Camelio, Fadel Megahed 

Image analysis is an important method of statistical process control in 

manufacturing. It is difficult to obtain suitable images for defect detection and 

analysis when products have irregular surfaces. Three-dimension surface scanner 

can get the surface information, but the data is difficult to analyze directly due to 

massive data points. We propose a method to generate image by point cloud data 

from surface scanner, which can be used in defect analysis. 

3 - Optimal Arrival and Service Rate Control of Multi-server Queues 

with Application to Could Computing 

Nelson Lee, Department of Statistics and Operations Research, 

University of North Carolina, Chapel Hill, NC, 27599, 

United States of America, leent@email.unc.edu, 

Vidyadhar Kulkarni 

We consider the problem of optimal control of a multi-server queue with 

controllable arrival and service rates. This study is motivated by its potential 

application to the design and control of data centers or cloud computing facilities. 

The cost structure includes customer holding cost, server operating cost, and 

system operating reward. We derive structural properties of the optimal control 

policies under both discounted cost and average cost criterions.

SC73 

■ SC73 


Computational Finance I 



Chair: Nan Chen, Assistant Professor, The Chinese University of Hong 

Kong, Shatin, N.T., Hong Kong, Hong Kong - PRC, 

nchen@se.cuhk.edu.hk 

1 - Coco, Bail-in and Tail Risk 

Nan Chen, Assistant Professor, The Chinese University of Hong 

Kong, Shatin, N.T., Hong Kong, Hong Kong-PRC, 

nchen@se.cuhk.edu.hk, Paul Glasserman, Behzad Nouri 

We develop a capital structure model to analyze the incentives created by 

contingent convertibles (Cocos) and bail-in debt, two variants of debt that 

converts to equity as a bank nears or reaches financial distress. Our model 

combines endogeous default, debt rollover, and jump risks. We find that Cocos 

generally have positive incentive effects when the conversion trigger is not set too 

low. 

2 - Optimal Deleveraging with Market Impact 

Jingnan Chen, University of Illinois at Urbana-Champaign, 

Urbana, IL, 61801, United States of America, jchen98@illinois.edu, 

Jiming Peng, Liming Feng, Yinyu Ye 

We consider an optimal liquidation problem where the objective is to meet 

specified leverage ratio at the minimal cost. Under linear price impact, the 

problem reduces to a generally non-convex quadratic program with quadratic and 

box constraints. We derive some analytical results regarding the optimal 

liquidation strategy. A Lagrangian method is proposed to solve the problem 

numerically. 

3 - Consistent Pricing of Options on Leveraged ETFs 

Martin Haugh, Columbia University IE&OR, 500 West 120th 

Street, Mudd 332, New York, NY, 10027, United States of America, 

mh2078@columbia.edu, Andrew Ahn, Ashish Jain 

We consider the problem of pricing options on leveraged ETFs in a manner that is 

consistent with the pricing of options on the underlying security. The difficulty is 

that tractable underlying price dynamics do not in general translate into tractable 

LETF dynamics. We overcome this problem by approximating the LETF dynamics 

so that approximate LETF option prices can be computed very quickly via 

standard transform methods. We justify our approximations via Monte-Carlo 

experiments. 

4 - Rare Event Simulation for Portfolio Default Losses 

Xiaowei Zhang, HKUST, Hong Kong, Hong Kong-PRC, 

xiaoweiz@ust.hk 

We use a self-exciting point process L(t) to model the default losses of a portfolio. 

Fixing the time horizon t, we develop an importance sampling (IS) simulation for 

estimating P(L(t)>x), i.e. the probability that the portfolio suffers a large default 

loss before time t. We will show that this IS scheme achieves a significant variance 

reduction compared to plain Monte Carlo simulation. 

Sunday, 4:30pm - 6:00pm 

■ SD01 

01- West 101- CC 

Joint Session Opt-Global/ENRE-Energy: Global 

Optimization in Energy 

Sponsor: Optimization/Global Optimization & Energy, Natural Res & 

the Environment/Energy 


Chair: Steffen Rebennack, Assistant Professor of Operations Research, 

Colorado School of Mines, 1500 Illinois St., Golden, CO, 80401, 

United States of America, srebenna@mines.edu 

Co-Chair: Timo Lohmann, PhD Student, Colorado School of Mines, 

816 15th Street, Golden, CO, 80401, United States of America, 

tlohmann@mines.edu 

1 - Decomposition Strategies for Stochastic Mixed-integer 

Nonlinear Programs with Energy Applications 

Paul I. Barton, Lammot du Pont Professor, Massachusetts Institute 

of Technology, 77 Massachusetts Avenue, 66-464, Cambridge, MA, 

02139, United States of America, pib@mit.edu 

We present a recently developed deterministic global optimization approach with 

running time that scales linearly with the number of scenarios considered. This 


130 

enables the solution of large-scale instances of problems that arise in the 

optimization of a variety of energy systems under uncertainty, such as the 

planning of natural gas production infrastructure, the design of flexible energy 

polygeneration systems, feedstock selection for oil refineries, etc. 

2 - The Optimal Power Flow Problem Model as a MILP 

(Mixed-Integer Linear Programming) 

Maria de Luján Latorre, Researcher, PSR, Praia de Botafogo 

228/1701-A Botafogo, Rio de Janeiro, RJ, 22250-145, Brazil, 

lujan@psr-inc.com, Sérgio Granville, Mario Veiga Pereira, 

Rafael Ferreira, André Dias 

The OPF problem determines the optimal operation of generation and 

transmission system representing the AC network constraints in an integrated 

form. In this work we apply McCormick inequalities and disjunctive constraints to 

linearize the original nonlinear and non-convex optimization problem. Mixed 

Integer Programming models are derivate both in Polar as in Cartesian 

coordinates systems. Results compare both formulations with the original OPF 

model. 

3 - Global and Local Approaches for Large-scale Securityconstrained 

Optimal Power Flow 

Dzung Phan, Research Staff Member, IBM T.J. Watson Research 

Center, 1101 Kitchawan Road, Yorktown Heights, NY, 10598, 

United States of America, phandu@us.ibm.com, 

Jayant Kalagnanam 

We present global and local methods for solving the security-constrained optimal 

power flow problem. We propose a class of global methods based on Lagrangian 

duality to solve the problem to optimality. For practical uses when dealing largescale 

instances, two decomposition algorithms based on Benders cut and ADMM 

are investigated. These schemes often generate solutions whose objective values 

are smaller than the conventional approach and are very close to the optimal 

points. 

4 - Optimization of Electrical Efficiency using Mixed 

AC/DC Distribution 

Stephen Frank, PhD Student, Electrical Engineering, Colorado 

School of Mines, 1500 Illinois St., Golden, CO, 80401, United 

States of America, stfrank@mymail.mines.edu, Steffen Rebennack, 

Pankaj (PK) Sen 

The majority of electrical loads and distributed generation sources in modern 

buildings require direct current (DC), but grid-level electricity generation, 

transmission, and distribution systems continue to use alternating current (AC). A 

significant amount of energy is lost due to AC/DC conversion. This presentation 

explores formulations and solution methods for the optimal design of mixed 

AC/DC distribution systems to minimize the losses due to AC/DC conversion. 

■ SD02 

02- West 102 A- CC 

2012 DAS Practice Award Presentations 



Chair: Gregory Parnell, United States Military Academy, West Point, NY, 

10996, United States of America, gregory.parnell@usma.edu 

1 - Catalyze Supports UK Ministry of Defence Training 

Rationalisation 

Carmen Carmona, Consultant, Catalyze, 1 The Old Diary, 

Bunstead Barns, Hursley, SO21 2LL, United Kingdom, 

CarmenCarmona@catalyze.co.uk, Patrick Warburton, 

Rex Mazonowicz 

Catalyze designed and implement a two level stakeholder socio-technical process 

aiming to consistently assess 24 possible options, using a mix of financial, socio 

and technical criteria. The process allowed the wider stakeholders and MOD 

scrutiny community to be convinced of the integrity of the decision consequences 

and engage MOD Experts from all three armed services. (Navy, Royal Air Force 

and Army) 

2 - AESA: Decision Analysis in Search for Safer Skies 

David Rios Insua, Professor on Statistics and Decision Sciences, 

Royal Academy of Sciences, Calle Valverde, 22, Madrid, Spain, 

david.rios@urjc.es, Pablo Hernandez-Coronado, Esperanza Herraiz, 

Jorge Valero, Veronica Elvira 

We describe the pioneering efforts of the Spanish Agency for Air Transportation 

Safety in using Decision Analysis within the State Safety Programme. We 

illustrate modeling issues in forecasting threats and their severities, identifying 

and communicating the most severe ones, allocating resources to mitigating 

them, and delivering safe sky policies in public-private partnership.

3 - Operations Analysis Workforce Development 

Capabilities Assessment 

William Nanry, Member, Group Technical Staff, Lockheed Martin 

Missiles and Fire Control, 1701 W. Marshall Drive, Grand Prairie, 

TX, 75265, United States of America, william.nanry@lmco.com, 

Steve Dininger 

To enhance operations analysis (OA), the Lockheed Martin OA Workforce 

Development Team assessed the value each OA capability / skill provides to 

anticipate needs, assess shortfalls, and articulate mission understanding to 

generate affordable and relevant solutions for decision-makers and end-users. A 

multi-attribute decision support exercise was conducted to identify those key 

critical OA capabilities / skills across Lockheed Martin. 

■ SD03 

03- West 102 B- CC 

Joint Session DAS/HAS: Decision Analytic Models 

in Healthcare 

Sponsor: Decision Analysis & Health Applications Society 


Chair: Turgay Ayer, Assistant Professor, Georgia Institute of Technology, 

765 Ferst Drive, Atlanta, GA, 30332, United States of America, 

ayer@isye.gatech.edu 

1 - Monotone Policies in Dynamic Medical Treatment Planning 

Archis Ghate, University of Washington, Industrial and Systems 

Engineering, Seattle, WA, United States of America, 

archis@uw.edu 

A central problem in medicine requires physicians to choose dose levels that 

achieve adequate disease control while limiting adverse effects over a treatment 

course with multiple decision points. We propose a stylized stochastic dynamic 

programming formulation of this problem and prove the existence of a monotone 

optimal policy in which dose levels increase with disease severity. This result is 

illustrated using an example from cancer radiotherapy. 

2 - Optimizing Radiation Delivery Schedules for Gliomas 

Kevin Leder, Assistant Professor, University of Minnesota, 111 

Church St., Minneapolis, MN, 55455, United States of America, 

Kevin.Leder@me.umn.edu, Ken Pitter, Eric Holland, 

Franziska Michor 

We construct a model to study the response of gliomas to radiation therapy. In 

particular, we consider a model considering two distinct cell populations, a slow 

growing radio-resistant stem cell population and a faster growing radio-sensitive 

progenitor cell population. Cells dynamically switch back and forth between these 

two states in response to radiation doses. In the context of this model we identify 

near optimal dosing schedules, and compare their performance to standard 

therapies. 

3 - Evaluating the Impact of Performance Goals on Emergency 

Medical Service Dispatching Decisions 

Laura McLay, Virginia Commonwealth University, Statistics & 

Operations Research, 1015 Floyd Ave., Box 843083, Richmond, 

VA, 23284, United States of America, lamclay@vcu.edu, 

Maria Mayorga 

Emergency medical services measure performance based on response time 

thresholds, i.e., the proportion of calls responded to within a fixed timeframe. The 

ultimate goal of emergency medical services is to save lives, which suggests that it 

might be worthwhile to consider using performance measures more closely tied to 

patient outcomes. This paper examines how different response time threshold 

performance goals impact patient survival rates in the context of emergency 

medical dispatch. 

4 - Improving Access to Community-based Chronic Care through 

Improved Capacity Allocation 

Tingting Jiang, Student, Northwestern University, 2145 Sheridan 

Road Room C210, Evanston, IL, 60208, United States of America, 

tingting-jiang@northwestern.edu, Seyed Iravani, Sarang Deo, 

Karen Smilowitz 

We study a model of community-based healthcare for chronic diseases. Patients 

periodically access the healthcare system, influencing their disease progression 

and health outcomes. The provider’s goal is to maximize community-wide health 

outcomes, subject to capacity constraint. We investigate how the provider can 

improve her objective through better operational decisions. 


131 

■ SD04 

SD04 

04- West 102 C- CC 

Scoring Rules, Probability Assessment, and Bias 



Chair: Robert Hammond, The University of Texas at Austin, 

1 University Station, Austin, TX, United States of America, 

robertkh@gmail.com 

Co-Chair: Eric Bickel, Assistant Professor, The University of Texas at 

Austin, 1 University Station, C2200, Austin, TX, 

United States of America, ebickel@mail.utexas.edu 

1 - Lossed in Translation: Adjusted Scoring Rules to Recover 

Subjective Beliefs from Loss-Averse Experts 

Asa Palley, The Fuqua School of Business, Duke University, 

100 Fuqua Drive, Box 90120, Durham, NC, 27708, 

United States of America, abp12@duke.edu, Theo Offerman 

We apply a prospect theory model of risk preferences to examine the incentives of 

an expert under an asymmetric quadratic scoring rule. We use these results to 

develop the L-adjusted scoring rule, a modified quadratic rule that corrects these 

distortions and allows for the recovery of an accurate estimate of probabilistic 

beliefs. Finally, we conduct an experiment that demonstrates the effectiveness of 

this modification in eliciting truthful beliefs and eliminating conservative baseline 

reports. 

2 - Gender and Performance in Risky Situations 

Susan Fisk, PhD Candidate, Stanford University, Sociology 

Deparment, Building 120, Room 031, Stanford, CA, 94305, 

United States of America, sfisk@stanford.edu, Ross Shachter 

We explore the role gender plays in risk-taking, using data from a large 

undergraduate decision analysis class in which the midterm uses a probabilistic 

scoring rule but the final exam does not. Controlling for academic major, we find 

that women take fewer risks than men, confirming established research on 

gender and risk-taking. In addition, we find that women perform worse on exams 

when their grade is determined by the scoring rule instead of conventional 

grading methods. 

3 - Strictly Proper Mechanisms with Cooperating Players 

Ross Shachter, Associate Professor, Stanford University, 

Department of Management Science and Eng, Huang Engineering 


shachter@stanford.edu, SangIn Chun 

Prediction markets provide an efficient means to assess uncertain quantities from 

forecasters. Although traditional and competitive strictly proper scoring rules 

have been shown to incentivize forecasters to provide accurate probabilistic 

forecasts acting alone, when forecasters with different beliefs can cooperate, these 

mechanisms instead discourage honest reports. Nonetheless, we show conditions 

under which the decision maker can confidently apply the assessed probabilities. 

4 - Scoring Rules and the Measurement of Uncertainty 

Eric Bickel, Assistant Professor, The University of Texas at Austin, 

1 University Station, C2200, Austin, TX, United States of America, 

ebickel@mail.utexas.edu 

Strictly proper scoring rules (SPSRs) continue to play an important role in 

probability assessment. While many such rules have been developed, relatively 

little guidance exists as to which rule is the most appropriate. All SPSRs reward 

calibration. In this presentation, we focus on how differing rules reward 

sharpness and argue that log scoring is the only rule that is consistent with 

entropy as a measure of uncertainty.

SD05 

■ SD05 

05- West 103 A- CC 

Process Engineering 


Chair: Raed Naebulharam, Research Assistant, University of Wisconsin- 

Milwaukee, 3200 North Cramer St, EMS 280, Milwaukee, WI, 53211, 

United States of America, naebulh2@uwm.edu 

1 - Refinery Operation Decisions with the Consideration of the 

Price Fluctuations 

Ruoran Chen, PhD Student, Department of Industrial Engineering, 

Tsinghua University, Building ZJ14#0429A, Beijing, 100084, 

China, crr11@mails.tsinghua.edu.cn, Simin Huang, Ruwen Qin 

Refineries face the price fluctuations in crude oil and products when making 

decisions in procurement and production. We first construct an LP model for a 

refinery to maximize the total profit. Then the model is extended to a large-scale 

multi-period stochastic dynamic programming to optimize the procurement, 

inventory and refinery decisions. A numerical method is developed to solve the 

problem. Numerical experiments are set up to demonstrate the importance of 

integrating these decisions. 

2 - Integrated Maintenance and Production Scheduling under 

Deteriorating Machine Conditions 

Maliheh Aramon Bajestani, PhD Candidate, University of Toronto, 

Mechanical & Industrial Engineering Dept, Unit 804, 141 

Davisville Ave., Toronto, ON, M4S 1G7, Canada, 

maramon@mie.utoronto.ca, J. Christopher Beck, Dragan Banjevic 

We address the problem of maintenance and production scheduling in a 

deteriorating multi-stage, multi-product system with random demand. Exploiting 

machine condition information, we formulate a Markov decision process model 

to determine the maintenance plan over the long term and then formulate an 

integer programming model to find the maintenance and the production schedule 

in the current period. Our results demonstrate that the planned schedule is very 

close to the actual schedule that is executed. 

3 - Hub and Chain: A Variance-based Method for Designing 

Process Flexibility 

Zhiguang Han, PhD Student, Nanyang Business School, Nanyang 

Technological University, 50 Nanyang Avenue, Singapore, 639798, 

Singapore, hanz0007@ntu.edu.sg, Shaoxiang Chen, Geoffrey Chua 

We study the process flexibility design issue. To our best knowledge, the demand 

variance has been generally ignored in the literature. We investigate the effect of 

demand variance and present a simple variance-based method to construct good 

flexible structures that link several small chains together in a hub. Our numerical 

tests show that our method performs better than existing methods such as 

chaining and constraint sampling, in both computational time and structure 

quality. 

4 - Bernoulli Serial Lines with Deteriorating Product Quality: 

Evaluation & System-theoretic Properties 

Raed Naebulharam, Research Assistant, University of Wisconsin- 

Milwaukee, 3200 North Cramer St., EMS 280, Milwaukee, WI, 

53211, United States of America, naebulh2@uwm.edu, 

Liang Zhang 

This paper assumes that the probability that each unfinished part is of good 

quality is a decreasing function of its residence time in the preceding buffer. Then, 

in serial production lines with machines having Bernoulli reliability model, 

closed-form formulas for performance evaluation in the two-machine line case 

were derived, and an aggregation-based procedure to approximate the 

performance measures in M > 2-machine lines was developed. Monotonicity 

properties were also studied. 

5 - Demand Forecasting in Revenue Management under Availability 

Constraints 

Shadi Sharif Azadeh, Polytechnique Montreal, 2900 Boulevard 

Edouard-Montpetit Montre, Montreal, QC, Canada, 

shadi.sharifazadeh@polymtl.ca, Gilles Savard 

We investigate the problem of demand forecasting in transportation. Seats are 

limited therefore, transportation companies continue to accept reservations in a 

fare class until the booking limit is reached. We propose an optimization model 

which takes availability constraints into account in order to have more accurate 

insight about demand behavior. 


132 

■ SD06 

06- West 103 B- CC 

Metamodeling Techniques for 

Simulation Optimization 



Chair: Sujin Kim, Assistant Professor, National University of Singapore, 

1 Engineering Drive 2, Singapore, 117576, Singapore, iseks@nus.edu.sg 

1 - A New Framework for Combining Global and Local Surrogates 

in Black-box Optimization 

Yibo Ji, National University of Singapore, Kent Ridge Crescent, 

Singapore, 119260 SK, Singapore, jiyibo@nus.edu.sg, Wendy Xu, 

Myun-Seok Cheon, Sujin Kim 

We propose a new framework for solving computationally expensive optimization 

problems, where neither their closed-forms nor derivatives are available. We first 

build a global metamodel to identify the promising area, and then conduct a local 

search to ensure local optimality. One of the key performance factors is how we 

balance the two procedures. The framework is designed to accommodate various 

types of global and local search methods. Numerical studies demonstrate the 

feasibility of our work. 

2 - Enhancing Stochastic Kriging Metamodels with 

Gradient Estimators 

Barry Nelson, Professor, Northwestern University, 2145 Sheridan 

Road, Department of IEMS, Evanston, IL, 60208-311, 

United States of America, nelsonb@northwestern.edu, Xi Chen, 

Bruce Ankenman 

Stochastic kriging is a metamodeling technique for effectively representing the 

mean response surface implied by a stochastic simulation. We show that 

incorporating gradient estimates into stochastic kriging tends to significantly 

improve surface prediction. We also evaluate the properties of infinitesimal 

perturbation analysis and likelihood ratio gradient estimators when incorporated 

into stochastic kriging. 

3 - Augmenting Simulation Metamodels with Direct 

Gradient Estimates 

Huashuai Qu, University of Maryland, College Park, MD, 

United States of America, huashuai@math.umd.edu, Michael Fu 

Traditional metamodel-based optimization methods assume experiment data 

collected consist of performance measurements only. However, in many settings 

found in stochastic simulation, direct gradient estimates are available. We 

investigate techniques that augment existing regression and stochastic kriging 

models to incorporate additional gradient information. The augmented models 

are shown to be compelling compared to existing models, in the sense of 

improved accuracy or reducing simulation cost. 

■ SD07 

07- West 104 A- CC 

Machine Learning I 



Chair: Cynthia Rudin, Assistant Professor, Massachusetts Institute of 


United States of America, rudin@mit.edu 

1 - Using Latent Semantic Analysis and Support Vector Machine in 

Multi-level Polarity Sentiment Analysis 

Yilu Zhou, Assistant Professor, George Washington University, 

2201 G St NW Suite 515, Washington, DC, 20052, 

United States of America, yzhou@gwu.edu 

We propose a multi-level polarity sentiment analysis framework that utilizes a 

Natural Language Processing technique: Latent Semantic Analysis (LSA) to 

reduce feature dimensionality and a multi-classifier SVM to detect sentiment 

level. We conducted a series of experiments to estimate the optimal kernel 

parameters and the dimension of features. The results shows that LSA can 

effectively reduce term feature dimensions and the optimal dimension and kernel 

parameters can boost the performance. 

2 - Machine Learning Methods from Statistical Physics 

Mohsen Bayati, Stanford University, Stanford, CA, 

United States of America, bayati@stanford.edu, Andrea Montanari 

We present results on approximate message passing (AMP) that is an iterative 

algorithm for high-dimensional data analysis and is inspired by statistical physics 

ideas.

3 - A Comparison of Criteria for the Smoothing Parameter 

Selection in Partial Spline with Change Points 



sung_won.han@roche.com, Hua Zhong 

In many applications such as bio-informatics, the estimation of the starting and 

ending points of drop-down in the longitudinal data is one of the key problems. 

We used a partial spline model based on the reproducing kernel Hilbert spacebased 

spline and multiple change points. Since the existing criteria for the 

smoothing parameter selection in partial spline with change points are not proper, 

we propose simple and better criteria to estimate change points. 

■ SD08 

08- West 104 B- CC 

Joint Session SPPSN/MAS: Humanitarian Logistics 

and Disaster Response I 

Sponsor: Public Programs, Service and Needs & Military 

Applications 


Chair: Jarrod Goentzel, Massachusetts Institute of Technology, 

Room E38-650, 77 Massachusetts Ave., Cambridge, MA, 02139, 

United States of America, goentzel@mit.edu 

1 - Airport Congestion During Relief Operations 

Mike Veatch, Gordon College, Department of Mathematics, 

255 Grapevine Rd, Wenham, MA, 01984, United States of 

America, Mike.Veatch@gordon.edu, Jarrod Goentzel 

Scheduling the airlift of relief supplies into a damaged or small airport during a 

crisis is a major challenge. Adapting to the local conditions and coordinating the 

many international donors and airlift providers is complex. Alternative scheduling 

methods are proposed and assessed using queueing models with data from the 

Haiti earthquake. 

2 - Managing Bottlenecks in Port and Overland Transport Networks 

for Food Aid 

Mallory Soldner, Georgia Institute of Technology, Atlanta, GA, 

United States of America, msoldner@gatech.edu, Jarrod Goentzel, 

Ozlem Ergun 

Based on experience with WFP in East Africa, port bottlenecks can restrict the 

flow of life saving food aid. We introduce a canonical framework for building 

acyclic queuing networks to model port and overland transportation networks. 

Steady state equations and approximations are used to define cost and time 

bottlenecks. We classify how these bottlenecks shift in the network depending on 

parameters and decisions. Finally, we consider how adding capacity to the 

network impacts efficiency. 

3 - Modeling and Analyzing Supply Chain Viability 

Mariah Magagnotti, Graduate Research Assistant, Clemson 

University, 110 Freeman Hall, Clemson, SC, 29634, 

United States of America, mariahm@clemson.edu, Scott Mason 

The supply chain literature reveals many performance measures focused on 

localized costs and/or throughput. A holistic view is more appropriate when a 

supply chain is disrupted due to an emergency or disaster, as many of the key 

decisions to be made are interdependent. We present a capacitated, multicommodity 

flow model under two competing objectives and demonstrate how 

viability-based decision making can impact both performance measures of 

interest. 

4 - Scheduling and Routing for a Bus-based Evacuation with 

Constant Evacuee Arrival Rate 

Victor Pereira, PhD Student, Virginia Tech, Department of 

Industrial and Systems Eng., 250 Durham Hall, Blacksburg, VA, 

24061-0118, United States of America, vpereira@vt.edu, 

Douglas Bish 

In this presentation, a variant of the vehicle routing problem for bus-based 

evacuation planning is introduced. The problem assumes that evacuees arrive at a 

constant rate, and seeks a schedule of pick-ups and the routes for a fleet of buses 

that minimizes their waiting time at various pick-up locations. In addition, the 

effect of service level over the waiting time is explored and upper and lower 

bounds on the fleet size and an upper bound on the number of arc traversals per 

bus are proposed. 


133 

■ SD09 

SD09 

09- West 105 A- CC 

AHP/ANP Applications 



Chair: Birsen Karpak, Professor of Management, Youngstown State 

University, One University Plaza, WCBA 3303, Youngstown, OH, 

44555, United States of America, bkarpak@ysu.edu 

1 - Integrating Multicriteria Analysis and Geographic Information 

Systems: An Application of Spatial ANP 

Valentina Ferretti, PhD, Politecnico di Torino, Corso Castelfidardo 

30/A, Torino, 10138, Italy, valentina.ferretti@polito.it, 

Giulio Mondini 

Spatial Multicriteria Evaluations are rapidly gaining traction for planning and 

policy-making as useful tools to make decision processes effective and reliable. 

The contribution proposes a Spatial Multicriteria approach based on the 

integration of Geographic Information Systems and the Analytic Network Process 

for the development of a land suitability analysis in the context of undesirable 

facilities location problems. The integrated approach is able to cope with complex 

decision problems. 

2 - Evaluating the Performance of Brazilian Financial Managers 

with the AHP 

Luiz F. Autran M. Gomes, Professor, Ibmec/RJ, Av. Presidente 

Wilson, 118, Room 1110, Rio de Janeiro, 20030020, Brazil, 

autran@ibmecrj.br, Renata Andrade 

The key objective of this paper is to show, through a case study, how managers 

working for a financial organization in Brazil were evaluated in their performance 

by using the AHP method. Based on the success of the study the organization has 

decided that the AHP method should definitely be considered as the analytical 

tool to be utilized in their future performance evaluation processes. 

3 - Evaluation of Emergency Distribution Centers in Istanbul 

Ozay Ozaydin, Research Assistant, Dogus University, Zeamet S. 

No:21, Acibadem, Istanbul, 34722, Turkey, oozaydin@dogus.edu.tr, 

Ilker Topcu 

In disaster management, logistics play a major role and distribution centers are 

namely in the center of this role. As Istanbul has a highly dynamic environment, 

the previous studies on location of distribution centers can rapidly become 

obsolete and even infeasable. Using Geographical Information Systems (GIS) as 

main data input for an AHP model, this study not only evaluates the current 

situation but also proposes a dynamic system to make additional research and 

update process more manageable. 

4 - Cloud Computing Vendor Selection for the City of Pittsburgh 

Enrique Mu, MBA Program Director, Carlow University, 

3333 Fifth Avenue, Pittsburgh, PA, 15213, United States of 

America, ENRIQUEM@pitt.edu, Howard Stern 

Like many government agencies, the technology department at the City of 

Pittsburgh has had to deal with shrinking budgets and increased demands for 

state-of-the-art technology like cloud computing. The City employed the Analytic 

Hierarchy Process (AHP) methodology to select the most innovative and costeffective 

cloud computing solution provider. Lessons learned and issues with 

respect to this collaboration and public group decision-making in the context of 

new technology will be discussed.

SD10 

■ SD10 

10- West 105 B- CC 

Stochastic Programming with Discrete 

Random Variables 



Chair: Anh Ninh, Rutgers Center of Operations Research, 

Rutgers University, Piscataway, United States of America, 

ninhtuananh@gmail.com 

1 - Probabilistic Constrained Stochastic Programming with 

Independent Discrete Random Variables 

Kunikazu Yoda, PhD Candidate, Rutgers University, 640 

Bartholomew Rd, Piscataway, NJ, 08854-8003, United States of 

America, kyoda@rutcor.rutgers.edu, Andras Prekopa 

We consider the probabilistic constrained stochastic programming problem where 

the random right-hand side vector of independent components follows a discrete 

distribution with a finite support. The decision variables can be integers or real 

numbers. The problem is formulated by a mixed-integer linear programming with 

the concept of p-efficient points of a distribution. We present some techniques to 

improve the performance of the solution with MILP solver software and show 

numerical examples. 

2 - Numerical Solution of Large Scale Moment Problem in Case of 

Piecewise Higher Order Convex Objective 

Anh Ninh, Rutgers Center of Operations Research, 

Rutgers University, Piscataway, United States of America, 

ninhtuananh@gmail.com, Gabriela Alexe, Andras Prekopa 

We propose efficient numerical solution for the (continuous) power moment 

problem where the objective function is piecewise higher order convex. The 

solution goes through the solution of discrete moment problems with 

Vandermonde matrices in the equality constraints. Problems with up to 30 

moments can efficiently be solved but we can go higher in special cases. 

Numerical results will be reported. 

3 - Solution of the Univariate Discrete Moment Problem for New 

Classes of Functions 

Mariya Naumova, RUTCOR, Rutgers University, 640 Bartholomew 

Rd, Piscataway, NJ, 08854, United States of America, 

mnaumova@rci.rutgers.edu 

We characterize the dual feasible bases, in connection with univariate discrete 

moment problem for classes of objective function not dealt with until now. The 

new classes include, e.g., step functions with finite number of values. Applications 

will be mentioned to engineering design and finance. 

■ SD11 

11- West 105 C- CC 

Models in Robust Optimization 


Chair: Andreas Thorsen, PhD Candidate, Industrial Engineering and 

Operations Research, Pennsylvania State University, 310 Leonhard 


aht105@psu.edu 

1 - Robust Optimization and Uncertain Prices in an Oil Refinery 

Supply Chain 

Jens Bengtsson, School of Economics and Business, Norwegian 

University of Life Sciences, P.O. Box 5003, Aas, Norway, 

jens.bengtsson@umb.no, Patrik Flisberg, Mikael Ronnqvist 

Uncertain prices of crude oil and refinery products may seriously affect the 

profitability of the refinery value chain. In this project we analyze the oil refinery 

planning problem using robust optimization in order to handle uncertainties in 

prices and to determine robust plans. The uncertainties can be described using 

historical and/or forecasted data in a flexible model. 

2 - Distributionally Robust Convex Optimization 

Wolfram Wiesemann, Imperial College London, 180 Queen’s Gate, 

London, United Kingdom, wwiesema@imperial.ac.uk, 

Melvyn Sim, Daniel Kuhn 

Distributionally robust optimization studies stochastic programs whose uncertain 

parameters follow a distribution that is itself uncertain. The distribution is only 

known to belong to an ambiguity set defined in terms of statistical or structural 

properties, and the goal is to hedge against the worst-case distribution within the 

ambiguity set. In this talk we present several distributionally robust optimization 

problems that admit polynomial-time solvable reformulations or approximations. 


134 

3 - The Robust Model for City Logistic Terminal Location 

Afrooz Ansaripour, University of Oklahoma, School of Industrial 

Engineering, Norman, OK, 73072, United States of America, 

afrooz.ansaripour@ou.edu, Theodore Trafalis, 

Mostafa Badakhshian 

The concept of logistic terminals has been presented to help cities reduce the total 

number of urban trips and negative impacts of transportation. This paper aims to 

apply robust optimization in the city logistic terminal location problem. The main 

goal is to minimize the transportation volume between the terminals considering 

uncertainty. The indicators include the flow and distances between terminals and 

also the weight of each flow. Preliminary computational results will be presented. 

4 - Robust Optimization Models for Military Airlift under 

Demand Uncertainty 

Andreas Thorsen, PhD Candidate Industrial Engineering and 

Operations Research, Pennsylvania State University, 310 Leonhard 


aht105@psu.edu, Tao Yao 

The military airlift problem involves allocating cargo and passenger demands to 

aircraft and determining routes in order to achieve cost goals such as reducing 

overall cost and fuel consumption. Using robust optimization, we study the robust 

military airlift problem under demand uncertainty and present several mixedinteger 

program formulations including (static RO, two-stage RO, and multi-stage 

RO). We present results using data from the US military and examine the 

cost/robustness tradeoffs. 

5 - Estimating the Violation of the KKT Conditions 

Delphine Mico-Umutesi, University of Florida, 358 Little Hall, 

Gainesville, FL, 32611, United States of America, delmico@ufl.edu, 

William Hager 

Given x, not necessarily a local minimizer of an optimization problem, we 

propose a quantitative way of estimating how close x is to a solution of the KKT 

conditions. This estimate of the distance to a KKT point is the solution of a nonconvex 

optimization problem. We show how to solve this problem using an 

Active Set Algorithm (ASA) . 

■ SD12 

12- West 106 A- CC 

Advances in Integer Programming 



Chair: James Ostrowski, University of Tennessee, 1425 S Stadium Dr, 

Knoxville, TN, 37996, United States of America, jostrows@utk.edu 

1 - On Linear Programming Formulations of the TSP Polytope 

Sebastian Pokutta, Friedrich-Alexander-Universitat Erlangen- 

Nürnberg, Cauerstrasse 11, Erlangen, 91058, Germany, 

sebastian.pokutta@me.com, Serge Massar, Hans Raj Tiwary, 

Ronald de Wolf, Samuel Fiorini 

We solve a 20-year old problem posed by M. Yannakakis and prove that there 

exists no polynomial-size linear program (LP) whose associated polytope projects 

to the traveling salesman polytope, even if the LP is not required to be symmetric. 

Moreover, we prove that this holds also for the maximum cut problem and the 

stable set problem. These results follow from a new connection that we make 

between one-way quantum communication protocols and semidefinite 

programming reformulations of LPs. 

2 - Solution Approaches for the Cutting Stock Problem with 

Setup Cost 

Ali Ekici, University of Houston, Department of Industrial 


aekici@central.uh.edu, Azadeh Mobasher 

In this research, we study the Cutting Stock Problem with Setup Cost (CSP-S). In 

CSP-S, unlike traditional Cutting Stock Problem (CSP), we have separate cost 

factors for the number of setups and material usage, and the objective is to 

minimize total production cost including both material and setup costs while 

satisfying demand. We propose two local search algorithms and a column 

generation algorithm, and test these algorithms on a set of instances from the 

literature.

3 - Partial Objective Function Inequalities for the Multi-Item 

Capacitated Lot-Sizing Problem (MCLSP) 



esra.b@wichita.edu, J. Cole Smith, Joseph Hartman 

We derive valid bounds on the partial objective function of the MCLSP 

formulation via dynamic programming and integer programming techniques. We 

then develop algorithms for strengthening these valid inequalities by back-lifting 

binary and continuous variables. These inequalities are utilized in a cutting-plane 

strategy, in which we perturb the partial objective function coefficients for the 

MCLSP polytope. We present computational experiments with the proposed valid 

inequalities. 

4 - Optimizing the Layout of Proportional Symbol Maps 

Tallys Yunes, Department of Management Science, University of 

Miami, Coral Gables, FL, 33124-8237, United States of America, 

tallys@bus.miami.edu, Guilherme Kunigami, Pedro de Rezende, 

Cid de Souza 

Proportional symbol maps help cartographers and geo-scientists visualize geopositioned 

data, e.g. population and earthquakes. At specific locations, symbols 

are placed and scaled to make their areas proportional to the magnitude of the 

data. We present a novel IP model to draw opaque disks on a map while 

maximizing the visible border of all disks. We implement new decompositions and 

facets in a branch-and-cut algorithm, and are the first to optimally solve real 

instances of this problem. 

■ SD13 

13- West 106 B- CC 

Efficient Optimization Algorithms and Their 

Recent Applications 



Chair: Hongchao Zhang, Louisiana State Univeristy, Department of 

Mathematics, Baton Rouge, United States of America, 

hozhang@math.lsu.edu 

1 - Minimizing Rational Functions by Exact Jacobian SDP 

Relaxation 

Li Wang, University of Southern California, 3715 McClintock Ave., 

RM 236, Los Angeles, CA, 90089, United States of America, 

Wang40@usc.edu, Feng Guo 

We study the optimization problem of minimizing a rational function. We 

reformulate this problem as polynomial optimization by the technique of 

homogenization. These two problems are shown to be equivalent under some 

general conditions. The exact Jacobian SDP relaxation method proposed by Nie is 

used to solve the resulting polynomial optimization. 

2 - Mini-batch Stochastic Approximation Methods for Constrained 

Nonconvex Stochastic Programming 

Saeed Ghadimi, University of Florida, P.O. Box 210020, 

Gainesville, FL, 32611, United States of America, 

sghadimi@ufl.edu, Hongchao Zhang, Guanghui Lan 

We present a randomized mini-batch stochastic approximation method for solving 

constrained nonconvex stochastic programming problems. Complexity analysis of 

finding an approximate stationary point of this kind of problems is provided. 

Some computational results are also presented to show practical performance of 

the proposed algorithm. 

3 - A Quadratic C0 Interior Penalty Method for the Displacement 

Obstacle Problem of Clamped Plates 

Yi Zhang, Graduate Student, Louisiana State University, 103 

Lockett Hall, Department of Mathematics, LSU, Baton Rouge, LA, 

70803, United States of America, yzhang24@math.lsu.edu, 

Li-yeng Sung, Hongchao Zhang, Susanne Brenner 

The displacement obstacle problem of clamped Kirchhoff plates is an example of a 

fourth order variational inequality whose numerical analysis is more subtle than 

that of second order variational inequalities. In this talk we will introduce C0 

interior penalty methods for this problem. Both error estimates and numerical 

results will be discussed. This is joint work with Susanne Brenner, Li-yeng Sung 

and Hongchao Zhang. 


135 

■ SD14 

14- West 106 C- CC 

United States Presidential Election Forecasting: Who 

Will Win the White House in 2012? 

Cluster: Public Policy 


Chair: Sheldon Jacobson, Professor, University of Illinois, 201 N. 

Goodwin Avenue (MC258), Urbana, IL, 61801, United States of 

America, shj@illinois.edu 

1 - The Keys to the White House: Forecast for 2012 

Allan Lichtman, Distinguished Professor, Department of History, 

American University, Washington, DC, 20817, 

United States of America, lichtman@american.edu 

This paper will present the Keys to the White House and provide a forecast for the 

2012 presidential election. The Keys to the White House are a historically-based 

prediction system that retrospectively account for the popular-vote winners of 

every American presidential election from 1860 to 1980 and prospectively 

forecast well ahead of time the popular-vote winners of every presidential 

election from 1984 through 2012. 

2 - Modeling a Presidential Prediction Market 

Keith Chen, Yale School of Management, New Haven, CT, 

United States of America, Keith.chen@yale.edu, Edward Kaplan 

Using data from past Intrade US presidential prediction markets, we examined 

three alternative models for security prices, focusing on consistency with electoral 

college rules. We show that a simple diffusion model provides a good description 

of the overall distribution of electoral college votes, while an even simpler 

ranking model provides excellent predictions of the probability of winning the 

presidency. We derive implications for predicting the 2012 US presidential 

election. 

3 - The 2012 United States Presidential Election: And the 

Winner is? 

Sheldon Jacobson, Professor, University of Illinois, 201 N. 

Goodwin Avenue (MC258), Urbana, IL, 61801, United States of 

America, shj@illinois.edu, Edward Sewell, Steve Rigdon, 

Jason Sauppe 

Predict the number of electoral college votes for each candidatein the United 

States Presidential election can be a difficult task. This presentation discusses a 

Bayesian approach incorporating undecided voter effects into the analysis. A web 

site electionanalytics.cs.illinois.edu has been created for people to follow the 

resulting forecasts. 

■ SD15 

SD15 

15- West 202- CC 



1 - Lumina Decision Systems, Inc. - Introduction to Analytica, 

What it Does that Spreadsheets Can’t 

Surya Swamy, Consulting Decision Analyst, Lumina Decision 

Systems, 26010 Highland Way, Los Gatos, CA, 94117, 

United States of America, surya@lumina.com 

Experienced analysts prefer Analytica to spreadsheets because of its visual influence 

diagrams, Intelligent Arrays, fast Monte Carlo and scalability. Because it’s 

designed by expert modelers, it’s also an ideal tool for teaching students the art 

of effective modeling. View this software demo to see why users say that they 

can build, verify and analyze models in a quarter to half the time it takes with a 

spreadsheet. 

2 - Exploiting LINKS Simulations Web-based Resources for 

Maximum Teaching and Learning Impact 

Randall G. Chapman, Founder and President, LINKS Simulations, 

320 Forest Haven Drive, Winter Garden, FL, 34787, 

United States of America, chapman@LINKS-simulations.com 

How can an instructor teach effectively and efficiently with a large-scale, teambased, 

competitive supply chain management simulation? By exploiting the simulation’s 

web-based teaching and learning resources! This software demo 

explores web-based resources for supporting instructors and their students 

throughout LINKS simulation events.

SD16 

■ SD16 

16- West 207- CC 

Application of Kalman Filter in Traffic Management 

and Control 



Chair: Hans van Lint, Associate Professor, Delft University of 

Technology, Stevinweg 1, Delft, 2628CN, Netherlands, 

j.w.c.vanlint@tudelft.nl 

1 - Application of Kalman Filter in Traffic Management and Control 

Hans van Lint, Associate Professor, Delft University of Technology, 

Stevinweg 1, Delft, 2628CN, Netherlands, j.w.c.vanlint@tudelft.nl 

In many areas of traffic management and control the variables that are of most 

interest are often the ones that are most difficult to measure and estimate. Take 

for example vehicular density (veh/km) and space-mean speed (km/h). A reliable 

real-time estimate of these quantities is critically important for real-time control 

of traffic networks. However, neither can be straightforwardly deduced from 

available sensor data. Similarly elusive quantities are origin-destination flows. 

These depict the amount of vehicles per hour planning to go from one place to 

another at a certain moment. Unless we literally know the origin and destination 

of \textit{all} vehicles on a traffic network, advanced estimation techniques are 

required to extract OD patterns from whatever available data and prior 

knowledge we have available. One intuitive and highly effective method to solve 

these types of problems (i.e. estimating a quantity x when all we have are 

observations y and prior knowledge about the process) is the Kalman Filter, first 

proposed by Rudolf Kalman in 1960. In this TutORial we explain with many 

examples how this technique can be applied in the domain of traffic management 

and control to solve real-world problems. We’ll see that Kalman filtering is a 

powerful technique, that works surprisingly well in many cases, but there are also 

clear limitations that relate to the many assumptions underlying its application. 

■ SD17 

17- West 208 B- CC 

Panel Discussion: Design Thinking for Creating a 

Capstone Course in Management Consulting 

Sponsor: INFORM-ED 


Chair: Paul Szwed, Professor, U.S. Coast Guard Academy, Department 

of Management, 15 Mohegan Avenue, New London, CT, 06320, 

United States of America, Paul.S.Szwed@uscga.edu 

1 - Using Design Thinking to Create Meaningful 

Learning Experiences 

Moderator: Paul Szwed, Professor, U.S. Coast Guard Academy, 

Department of Management, 15 Mohegan Avenue, New London, 

CT, 06320, United States of America, Paul.S.Szwed@uscga.edu, 

Panelists: Laurel Goulet, Charles Coiro 

How do you go about developing a new course or improving an old one? One 

emerging method is to use concept and principles from design thinking. We had 

an award-winning capstone business consulting course that we knew was no 

longer getting us the desired results. Using design thinking concepts and 

principles, we conducted a “deep dive” on our course and developed an entirely 

new model. We will share the philosophies used and our experiences and lessons 

learned. Design thinking can be applied to a wide variety of courses or even 

modules in virtually any discipline. 

■ SD18 

18- West 208 A- CC 

Simulation for Transportation and Reliability 


Chair: Bex G. Thomas, Researcher, General Electric Global Research 

Center, 1 Research Center, Niskayuna, NY, 12309, 

United States of America, thomasb@ge.com 

1 - Incentives for Pedestrian Crossing Behavior with Vehicle 

Interactions under Unsignalized Conditions 

Hui Xi, The University of Arizona, 1127 E. James E, Rogers Way, 

Tucson, AZ, United States of America, huix@email.arizona.edu, 

Young-Jun Son 

Pedestrian crossing behavior under unsignalized conditions has been recognized 

as a main reason of pedestrian-vehicle crashes. This paper analyzes the incentives 

to prevent pedestrian jaywalking and their impact on pedestrian safety. By 


136 

considering heterogeneous behaviors of drivers and pedestrians as well as their 

interactions under various scenarios, we intend to identify the significant 

variables that help improve comfort and convenience as well as safety of 

pedestrian crossing. 

2 - Simulation of Container Truck-sharing Service (TSS): Reducing 

Empty Slots for Capacity Improvement 

Samsul Islam, University of Auckland, ISOM Department, 

12 Grafton Road, Auckland, New Zealand, 

misl086@aucklanduni.ac.nz, Tava Olsen 

Little attention has been paid by researchers to formulating strategies to resolve 

the issue of empty container-truck hauling, which decreases the system-wide 

truck transport capacity. The objective of this simulation paper is to perform and 

demonstrate scenario testing to compare three main strategies. These strategies 

include a dynamic truck sharing facility for an internet-based matching system to 

assign exporters to truckers to solve the empty-truck issue. 

3 - Modeling Life Cycle Performance of Asbestos Cement 

Water Mains 

Kent Kostuk, University of Saskatchewan, 57 Campus Drive, 

Saskatoon, SK, S7N 5A9, Canada, kent.kostuk@usask.ca, 

Gordon Sparks 

Municipalities in developed countries are facing a looming infrastructure crisis. 

Because of their visibility, roads and bridges are often the first infrastructure assets 

that come to mind. But buried infrastructure, such as water mains, are also 

decaying. To help quantify future costs and to identify ways to better manage and 

control these costs, a simulation model of an urban subdivision was developed. 

4 - Estimating Warranty Reserves for Photovoltaic Modules using 

Simulation Modeling and Analysis 

Bex G. Thomas, Researcher, General Electric Global Research 

Center, 1 Research Center, Niskayuna, NY, 12309, 

United States of America, thomasb@ge.com, Rajesh Tyagi 

GE Solar manufacturers and sources modules to be used within PV systems or 

resells it to end customers. The reliability and performance of these modules is 

guaranteed through product warranties and guarantees. We have developed a 

simulation based approach to estimate the warranty reserves over time that 

accounts for product specifications, manufacturing information, product 

reliability, product durability, warranty terms and associated costs. We highlight 

the advantages of our approach. 

■ SD19 

19- West 211 A- CC 


and Resource Allocation Applications in Disaster 

Relief 



Chair: Ashlea Bennett Milburn, Assistant Professor, University of 

Arkansas, 4207 Bell Engineering Center, Fayetteville, AR, United States 

of America, ashlea@uark.edu 

1 - Online Point of Distribution Location in Disaster Relief 

Ashlea Bennett Milburn, Assistant Professor, University of 

Arkansas, 4207 Bell Engineering Center, Fayetteville, AR, 

United States of America, ashlea@uark.edu, Chase Rainwater 

Information regarding affected populations, roads, buildings, and bridges becomes 

known on a rolling basis in the hours following a disaster. We develop online 

models that use real-time information regarding demand and infrastructure status 

to specify shelter locations and supply distribution points as a specific damage 

scenario is realized. Approximate solution approaches are employed to obtain 

good plans quickly. Computation results are presented for a New Madrid 

earthquake scenario. 

2 - Can Social Media Information Improve Disaster Relief 

Routing Plans? 

Emre Kirac, PhD Student, University of Arkansas, Department of 

Industrial Engineering, Fayetteville, AR, United States of America, 

ekirac@uark.edu, Clarence Wardell, Ashlea Bennett Milburn 

This research examines the impact of incorporating two distinct classes of 

information regarding the location and need of individuals impacted by a disaster 

when routing vehicles to deliver relief supplies: verified requests obtained 

through traditional means, and unverified requests obtained from social media. 

Competitive ratio analysis is performed for alternative decision approaches and 

computational results for a wide variety of disaster scenarios are presented.

3 - Developing a Dynamic Emergency Response Plan for a 

Post-earthquake Scenario 

Shuva Ghosh, Doctoral Candidate, Missouri University of Science 

and Technology, 223 Engineering Management Building, Rolla, 

MO, United States of America, sg2mb@mst.edu, Abhijit Gosavi 

An effective emergency response plan can play a vital role in minimizing risk to 

the people affected when an earthquake strikes. Emergency resources are 

generally stored in multiple responding centers. Emergency management 

personnel have to select an optimal combination of responding centers to respond 

effectively in a dynamically evolving situation. In this work, we integrate inputs 

from a so-called domino-effect analysis within a semi-Markov decision process to 

develop a response plan. 

■ SD20 

20- West 211 B- CC 

Optimization 



Chair: Yavuz Gunalay, Bahcesehir University, Faculty of Economics and 

Admin Science, Besiktas, Istanbul, 34353, Turkey, 

yavuz.gunalay@bahcesehir.edu.tr 

1 - Using Cross-entropy for Solving the Stochastic Shortest 

Path Problem 

Adji Cisse, Texas Tech University, Box 43061, Lubbock TX 79409, 

United States of America, gnouma.cisse@ttu.edu, Timothy Matis 

In this presentation, we show how Cross-Entropy methods may be used to find 

the stochastic shortest path through a networks. At the core of this approach is 

the use of truncated saddlepoint approximations for the distribution of path 

lengths. 

2 - Uncertainty and Risk Analysis in Federal Cost 

Estimating Agencies 

Tomeka Williams, Senior Operations Research, Optimum 

Performance Solutions, 2486 Glengyle Drive, Vienna, VA, 22181, 

United States of America, tomeka.williams@gmail.com 

Uncertainty and risk analysis is a process that is unique, tailorable, and often 

times misunderstood within Federal Departments and Agencies. In working in 

three different cost estimating organizations, each has had several initiatives to 

bridge that gap. This session is intended to share the initiatives of each of these 

agencies, and practical tools that have been employed to ensure that uncertainty 

and risk analysis is included in Federal cost estimating products. 

3 - Probabilistic Interpretation for the Elasticity of Carried Load 

Revenue from the Erlang Loss Model 

William Massey, Professor, Princeton University, Sherrerd Hall 

Princeton University, Princeton, NJ, 08540, 

United States of America, wmassey@princeton.edu 

Suppose that we know the demand function for some telephone service as a 

function of the phone rate. If we model the number of busy channels as an Erlang 

loss model, then we have a formula for the carried load revenue rate as a function 

of this price rate. The explicit formula for the elasticity of this carried load rate can 

be expressed in terms of the elasticity of the corresponding offered load rate. 

■ SD21 

21- West 212 A- CC 

Clique Relaxations in Networks 

Sponsor: Optimization/Networks 


Chair: Sergiy Butenko, Associate Professor, Texas A&M 

University,Industrial & Systems Engineering, 3131 TAMU, College 

Station, TX, 77843, United States of America, butenko@tamu.edu 

1 - New Developments in Finding Maximum 2-clubs in Graphs 

Foad Mahdavi Pajouh, Oklahoma State University, Industrial 

Engineering and Management, 322 Engineering North, Stillwater, 

Ok, 74075, United States of America, mahdavi@okstate.edu, 

Illya Hicks, Baski Balasundaram 

A k-club is an induced subgraph of diameter at most k which is a distance-based 

graph-theoretic generalization of a clique. The k-clubs can be used to model 

clusters in social networks, biological networks and internet graphs. This talk will 

outline the research challenges associated with k-club model because of its nonhereditary 

nature. New complexity results, theoretical results concerning the 

2-club polytope and preliminary numerical results will be presented. 


137 

2 - Degree-bounded Vertex Partitions 

John Arellano, Rice University, 6100 Main St., Houston, TX, 

77005-182, United States of America, jda2@rice.edu, Illya Hicks, 

Benjamin McClosky 

This paper studies degree-bounded vertex partitions, derives analogues for wellknown 

results on the chromatic number, and presents two algorithms for 

constructing degree-bounded vertex partitions. The first algorithm minimizes the 

number of partition classes. The second algorithm minimizes a weighted sum of 

the partition classes, where the weight of a partition class depends on the level of 

adjacency among its vertices. 

3 - Algorithms for Computing Optimal Clique Relaxation Structures 

that are Hereditary 

Chitra Balasubramaniam, 401 Stasney STreet, Apt 605, 

College Station, TX, 77840, United States of America, 

bcvaidyanath@neo.tamu.edu, Sergiy Butenko, 

Baski Balasundaram, Svyatoslav Trukhanov 

Heredity is a fundamental property of cliques that is in the core of some of the 

most successful combinatorial algorithms for the maximum clique problem. We 

use this observation to develop exact algorithms for computing maximum weight 

hereditary clique relaxation structures, such as s-plex and s-defective clique. 

■ SD22 

SD22 

22- West 212 B- CC 

Multilevel/Multistage Programming Applications 



Chair: Soheil Hemmati, PhD Student, University of Florida, 

406 Weil Hall, Gainesville, FL, 32603, United States of America, 

soheilmn@ufl.edu 

1 - An Algorithm for Solving Inverse Integer Programs 

Ted Ralphs, Associate Professor, Lehigh University, 200 West 


ted@lehigh.edu, Aykut Bulut 

An inverse optimization problem is that of determining the smallest perturbation 

in a given objective function required to make a given solution optimal. In this 

talk, we introduce the inverse mixed integer linear programming problem, discuss 

its complexity, and propose a general algorithm for solving it. The relationship to 

bilevel programming will also be explored. 

2 - Selecting Land Preserves to Enable Species Migration 

Jorge Sefair, PhD Student, University of Florida, Department of 

Industrial & Systems Eng., Gainesville, FL, United States of 

America, j.sefair@ufl.edu, Miguel Acevedo, Robert Fletcher, 

. Cole Smith 

We consider a problem in which a species migrates randomly among a network of 

patches across some region. Development on these patches negatively influences 

the probability that an organism survives when it moves to a patch, and if a 

critical set of patches is selected for development, the survival of the species 

becomes threatened. Our research presents a three-stage optimization model for 

optimally protecting patches so that the worst-case impact of land development is 

mitigated. 

3 - A Parametric Integer Programming Algorithm for Bilevel Mixed 

Integer Programs 

Chris Ryan, University of Chicago, Chicago, IL, United States of 

America, chris.ryan@chicagobooth.edu, Matthias Koeppe, 

Maurice Queyranne 

We consider discrete bilevel optimization problems where the follower solves an 

integer program with a fixed number of variables. Using recent results in 

parametric integer programming, we present fixed-dimension polynomial time 

algorithms for pure and mixed integer bilevel problems. 

4 - Mixed-integer Bilevel Programming (MIBLP) Approach for 

Competitive Prioritized Set Covering Problem 

Soheil Hemmati, PhD Student, University of Florida, 406 Weil 

Hall, Gainesville, FL, 32603, United States of America, 

soheilmn@ufl.edu, J. Cole Smith 

In competitive set covering problem, which has several applications, including 

non-cooperative product introduction, two players aim to maximize their rewards 

by selecting sets of items to satisfy clauses. Each selected item incurs a cost to 

players and each clause is satisfied by the highest-priority selected item, yielding a 

reward for the player that introduced the item. We propose an MIBLP model with 

0-1 decision variables appear in both levels, and suggest an exact cutting-plane 

algorithm.

SD23 

■ SD23 

23- West 212 C- CC 

Panel Discussion: Managing a Career in OR 



Chair: Peter Bell, Professor, University of Western Ontario, Richard Ivey 

School of Business, London, ON, N6A 3K7, Canada, pbell@ivey.uwo.ca 

1 - CPMS Practitioners Offer Tips for Success 

Moderator: Peter Bell, Professor, University of Western Ontario, 

Richard Ivey School of Business, Londo, ON, N6A 3K7, Canada, 

pbell@ivey.uwo.ca, Panelists: Nicholas Nahas, Manoj K. Chari, 

Doug Samuleson, Vijay Mehrotra 

This panel will share why advanced analytics are needed today more than ever. 

Business leaders lack abilities to apply analytical tools and data structure to 

address their challenges. Despite years of talk about scorecards and metrics, most 

organizations still rely on gut feelings and experience for urgent critical matters. 

We can glean insights from data and guide them along innovative ways as 

analytical thinking can lead their organization into the future. 

■ SD24 

24- West 213 A- CC 

Empirical Research in Health Care Operation 



Chair: Lijie Song, MIT-Zaragoza International Logistics Program, 

Edificio N·yade, 5 C/ Bari 55, PLAZA, Zaragoza, 50197, Spain, 

lsong@zlc.edu.es 

1 - Physician Variation in Resource Utilization 

Jillian Berry, Doctoral Candidate, Harvard University, 

Soldiers Field, Boston, MA, 02478, United States of America, 

jberry@hbs.edu, Anita Tucker 

We study the variation in physician resource utilization in two academic 

Emergency Departments (ED) and how this variation is affect by resource 

availability. 

2 - Building Flexibility in US Healthcare Delivery through 

Medical Tourism 

Muer Yang, Assistant Professor, University of St. Thomas, 

1000 LaSalle Avenue, SCH 435, Minneapolis, MN, 55403, 

United States of America, yangmuer@stthomas.edu, 

Sameer Kumar 

Medical tourism is a large and growing phenomenon. In 2011 the medical 

tourism market was an estimated $20 billion, up from under $2 billion in 2005. 

The combination of cheap price and good medical care has caused an increasing 

number of Americans to choose medical tourism. We develop a mathematical 

model illustrated through discrete event simulation to show the impact of medical 

tourism on the cost and effectiveness of US healthcare delivery system. 

3 - Partially Flexible Operating Rooms 

Yann Ferrand, Assistant Professor, Clemson University, 

Department of Management, Sirrine Hall, Clemson, SC, 29634, 

United States of America, yferran@clemson.edu, 

Michael Magazine, Uday Rao 

When organizing the operating theatre and scheduling surgeries, hospitals face a 

trade-off between the need to be responsive to emergency cases and to conduct 

scheduled elective surgeries efficiently. We develop a simulation model to 

compare resource-allocation policies of complete flexibility, and of complete 

focus, and compare them to hybrid partial flexibility policies. We evaluate these 

policies on patient and provider outcome measures, including patient wait time 

and physician overtime. 

4 - Understanding Client Preferences for Preventive Care 

Lijie Song, MIT-Zaragoza International Logistics Program, 

Edificio Náyade, 5 C/ Bari 55, PLAZA, Zaragoza, 50197, Spain, 

lsong@zlc.edu.es, Vedat Verter, Beste Kucukyazici 

We studied a government-funded breast cancer screening program in Montreal, 

using stated preference discrete choice modeling as framework. This enables us to 

better understand the clients’ preference concerning facility attributes. 

Incorporating our findings in a simulation model, we are able to generate 

managerial insights on how to improve the participation to such programs. 


138 

■ SD25 

25- West 213 B- CC 

Chemotherapy Patient and Resource Management in 

Oncology Clinics 



Chair: Lewis Ntaimo, Associate Professor, Texas A&M University, 

3131, College Station, TX, 77843, United States of America, 

ntaimo@tamu.edu 

Co-Chair: Tanisha Cotton, Doctoral Candidate, Texas A&M University, 

3131 TAMU, College Station, TX, 77843, United States of America, 

tanbgreen05@neo.tamu.edu 

1 - Uncertainty in Chemotherapy Patient Scheduling 

Sara Shashaani, ISE PhD Student, Virginia Technical University, 

Blacksburg, VA, United States of America, sshashaa@vt.edu, 

Mark Lawley, Ayten Turkcan, Hong Wan 

Chemotherapy patient scheduling is a complex problem due to varying treatment 

lengths and nursing requirements, unpredictable same-day cancellations and 

unscheduled add-ons. Sensitivity analysis of uncertainties in the workflow is 

studied using a discrete-event simulation model to distinguish the stochastic 

elements that impose significant changes in the performance of the system when 

they encompass variability. Changes to the current deterministic scheduling 

model are proposed for flexibility. 

2 - Multi-criteria Nurse Assignment for Chemotherapy Patients in 

Oncology Clinics 

Ayten Turkcan, Northeastern University, Boston, MA, 

United States of America, A.Turkcan@neu.edu, Bohui Liang 

In oncology clinics, nurses with different skills are assigned to patients with 

different acuities to administer the chemotherapy. It is important to have a 

reasonable nurse workload for several reasons such as patient safety, quality of 

care, patient and staff satisfaction. We propose a multi-objective integer 

programming model with the objectives of balancing workload among nurses, 

and minimizing waiting time and overtime. We also propose a genetic algorithm 

to reduce the computation times. 

3 - Mean-risk Stochastic Optimization Approach for 

Chemotherapy Scheduling 

Tanisha Cotton, Doctoral Candidate, Texas A&M University, 3131 


tanbgreen05@neo.tamu.edu, Michelle M. Alvarado, 

Eduardo Perez, Lewis Ntaimo 

Oncology clinics face tight resource and scheduling constraints due to increased 

demand, cyclic chemotherapy treatment plans, and complex and uncertain 

duration of treatments. This talk presents the model and results of using a risk 

neutral and risk averse two-stage stochastic optimization approach to efficiently 

schedule chemotherapy nurses, patients, and chairs based on a real oncology 

clinic. 

4 - Modeling and Simulation of an Oncology Clinic using DEVS 

Michelle M. Alvarado, PhD Student, Texas A&M University, 3131 


michelle.mcgaha@gmail.com, Eduardo Perez, Tanisha Cotton, 

Lewis Ntaimo 

Scheduling of chemotherapy treatments is a difficult task due to the cyclic nature 

of the prescribed treatment regimen, variability of treatment lengths, and patient 

acuity levels. We present a discrete event system specification (DEVS) simulation 

model integrated with a stochastic mixed integer programming model (SMIP) to 

schedule and manage patients in an oncology clinic. Preliminary results are 

presented from the management perspective.

■ SD26 


Effective Decision-making in Pricing and 

Inventory Control 



Chair: Esma Gel, Arizona State University, Department of Industrial 

Engineering, Tempe, AZ, 85287, United States of America, 

esma.gel@asu.edu 

1 - A Comparison of Integrated Inventory/Outbound 

Dispatch Policies 

Sila Cetinkaya, Texas A&M University, College Station, TX, 

United States of America, sila@tamu.edu, Bo Wei 

We present an analytical comparison of alternative dispatch policies in the context 

of integrated inventory replenishment and outbound dispatch scheduling 

decisions. We prove that under the same expected replenishment cycle length and 

the same expected dispatch cycle length, quantity-based dispatching achieves the 

least inventory cost and waiting cost within one replenishment cycle, and, hence, 

it is superior to hybrid dispatching. 

2 - Leveraging Future Cost Reductions: A Dynamic Two-Sided 

Pricing Model 

Xiajun Amy Pan, University of Florida, United States of America, 

amypan@ufl.edu, Mei Lin 

Technology innovation engenders new products of higher qualities and reduces 

production costs. We study how a firm leverages future cost reductions to position 

its product line toward the high-end in both buyer-side only and two-sided 

markets. We find that future cost reductions raise the optimal price of the current 

device and trigger an intertemporal buyer-side demand shift forward. 

3 - Increasing Revenues from Parking Tickets 

Nichalin Summerfield, Adjunct lecturer, University of Arizona, 

1130 E. Helen Street, Mclnd 430, Tucson, United States of 

America, nichalin@email.arizona.edu, Morris Cohen, Moshe Dror 

Parking inspection is modeled as a revenue collecting Chinese Postman Problem. 

This approach maximizes expected revenues. We investigate online rules that 

allow the enforcement officers to adjust their routes in response to the observed 

parking permits’ times. Allowing an officer to selectively wait by legally parked 

cars till their permit expires could increase the expected revenues by 7%. 

4 - Price and Lead Time Quotation for Contract and Spot 

Customers 

Baykal Hafizoglu, Arizona State University, 699 S. Mill Avenue, 

Tempe, AZ, 85281, United States of America, baykal@asu.edu, 

Pinar Keskinocak, Esma Gel 

We consider dynamic price and lead time quotation for a make-to-order company 

with demand from contract customers as well as spot purchasers. Contract 

customers are offered a uniform price and lead time, whereas, spot purchasers are 

subject to dynamically quoted price and lead times. We discuss the potential of 

dynamic quotation for such environments, various properties of optimal control 

policies and the optimal mix of contract customers and spot purchasers. 

■ SD27 


Consumer Behavior Issues in Operations 



Chair: Omar Besbes, Columbia Business School, 3022 Broadway, New 

York, NY, 10027, United States of America, ob2105@columbia.edu 

Co-Chair: Ilan Lobel, New York University, Stern Business, 44 W 4th St, 

New York, NY, United States of America, ilobel@stern.nyu.edu 

1 - The Perils of Rationing under Consumer Regret and 

Misperception of Availability 

Yanchong Karen Zheng, Massachusetts Institute of Technology, 

Sloan School of Management, 100 Main Street, Cambridge, MA, 

United States of America, yanchong@mit.edu, Ozalp Ozer 

We study the impact of consumers’ regret and misperceived product availability 

on a firm’s inventory decision under a markdown policy. We show that both 

factors greatly reduce the need for rationing. They act as complements on the 

seller’s profit when consumers are modestly risk-averse and markdown is 

substantial. In contrast, they act as substitutes for highly risk-averse consumers. 

Ignoring these factors can subject the seller to the peril of over-rationing and lead 

to excessive profit losses. 


139 

2 - Strategic Stock-Outs with Social Learning 

Yiangos Papanastasiou, PhD Candidate, London Business School, 

Regent’s Park, London, NW14SA, United Kingdom, Nicos Savva, 

Nitin Bakshi 

We study a monopoly firm selling its product to customers holding uncertain, 

heterogeneous valuations. Limited availability induces customer self-selection, 

with high-valuation customers exerting effort to obtain the product first. We 

account for social learning and demonstrate that by inducing an early supply 

shortage, a firm can optimally control the information signal emitted from early 

reviews. Thereby, the firm can closely mimic dynamic pricing outcomes while 

charging a fixed price. 

3 - Dynamic Pricing Without Priors 

Ying Liu, New York University, 44 West 4th Street, New York, NY, 

United States of America, yliu2@stern.nyu.edu, Rene Caldentey, 

Ilan Lobel 

We consider a setting where a firm is selling a limited inventory of items over 

time to strategic customers, but it does not have a prior on the customers’ 

distributions of values. We construct policies that minimize the maximum regret 

the firm could face and present structural properties of such pricing policies. 

4 - Cheap Talk in Queues with Multiple Customer Classes 

Gad Allon, Northwestern University, 2001 Sheridan Rd, Evanston, 

IL, United States of America, g-allon@kellogg.northwestern.edu, 


We examine the problem of information communication by considering a model 

in which both the firm and the customers act strategically. In this model, the 

customers are heterogeneous both with regards to their waiting cost and the 

value they obtain from the service. The customer type is private information. We 

characterize the equilibrium language that emerges between the service provider 

and her customers. 

■ SD28 

SD28 


Models of Customer Behavior 



Chair: Xuanming Su, University of Pennsylvania, 


xuanming@wharton.upenn.edu 

1 - Consumer Bounded-Rationality: Evidence from the Click 

Stream Data 

Jun Li, University of Pennsylvania, The Wharton School, 

Philadelphia PA 19104, United States of America, 

lijun1@wharton.upenn.edu, Serguei Netessine 

Using click stream data of online search for air travels, we study how price 

volatility affects strategic purchase delays and search frequencies. 

2 - Capacity Investment under Bankruptcy Risk with Forward- 

Looking Consumers 

Robert Swinney, Associate Professor, Graduate School of Business, 

Stanford University, 655 Knight Way, Stanford, CA, 94305, 

United States of America, swinney@stanford.edu, Song Alex Yang, 

Erica Plambeck 

We consider the capacity investment decision of a firm that is in financial distress 

and may enter bankruptcy. Consumers value firm survival and experience a loss 

in utility if bankruptcy occurs. Thus, forward-looking consumers anticipate the 

risk of bankruptcy and may delay a purchase to see whether the firm survives. 

We analyze optimal capacity decisions in this setting and determine the impact of 

such strategic customer behavior on the firm’s profit, survival probability, and 

capacity level. 

3 - Social Comparisons in Service Cohorts 

Guillaume Roels, Assistant Professor, UCLA, 110 Westwood Plaza, 

Los Angeles, 90095, United States of America, 

guillaume.roels@anderson.ucla.edu, Xuanming Su 

We study the effect of social comparisons on individual performance in service 

cohorts such as arise in education, health care, and sports industry. We consider 

two kinds of social comparison depending on whether customers experience a 

disutility from under-performing or a utility gain from over-performing relative to 

their peers. We discuss how a service provider should manage her service to 

enhance or mitigate the effect of social comparison.

SD29 

4 - Pricing Restaurant Reservations 

Jaelynn Oh, University of Pennsylvania, 


jaelynn@wharton.upenn.edu, Xuanming Su 

Many restaurants offer reservations to delay sensitive customers and charge a fee 

for no-shows. Strategic customers make a reservation only when the expected 

utility of doing so is greater than that of walking in without a reservation. We 

study the restaurant’s pricing problem when serving both reservation holders and 

walk-in customers. We characterize equilibrium prices and profits and discuss 

model implications. 

■ SD29 


Staffing and Scheduling in Healthcare Operations 



Chair: Arvind Sainathan, Assistant Professor, Nanyang Business School, 

Nanyang Technological University, S3-B2A-03, 50 Nanyang Avenue, 

Singapore, Singapore, asainathan@ntu.edu.sg 

1 - Capacity Allocation between Appointments and Walk-ins with 

Endogenous Patient Choice 

Sarang Deo, Assistant Professor, Indian School of Business, 

Gachibowli, Hyderabad, 500032, India, Sarang_Deo@isb.edu, 

Aditya Jain 

We consider a healthcare system, wherein each patient decides the mode of 

arrival (appointment vs. walk-in) depending on the resulting waiting times 

(indirect vs. direct) as well as the severity of patient’s condition. The healthcare 

provider decides the allocation of capacity between walk-in and appointment 

slots. We analyze the equilibrium of this game and derive managerial insights 

about the impact of patient characteristics on capacity allocation decisions. 

2 - Multi-objective Robust Adaptive Scheduling for Hospital 

Staffing 

Kibaek Kim, Northwestern University, 2145 Sheridan Rd., C210, 


Kibaek.kim@u.northwestern.edu, Jonathan Turner, 

Sanjay Mehrotra 

For given mid-term (or long-term) schedules, it is necessary to make short-term 

adjustments, either by rescheduling staffs when shortages are anticipated or by 

canceling staffs when patient volume drops are anticipated. We develop a multiobjective 

robust adaptive scheduling model for the hospital staffing problem. 

3 - Hospital Inpatient Design: Analysis and Insights 

Mabel Chou, NUS Business School, Level 8, Biz1, Singapore, 

Singapore, mabelchou@nus.edu.sg 

We study the patient flow management in an inpatient department of a 

Singaporean hospital and discuss how to design the system to make the planning 

more cost effective. Our analysis generates a number of managerial insights for 

inpatient management. 

4 - Scheduling Patients by Improving Open Access (same day 

appointment) Policy in a Family Medicine Clinic 

Abraham Seidmann, Professor, Simon School of Business, 

University of Rochester, CS 3-333C, Rochester, NY, 14627, 

United States of America, avi.seidmann@simon.rochester.edu, 

Balaraman Rajan 

We present new empirical and analytical results of our research that focuses on 

improving the Open Access (same day appointment) models for scheduling 

patients in a family practice setup. We show the value of dynamically 

incorporating the effect of anticipated behavior of patients on the expected load of 

the clinic. 

5 - Nurse Absenteeism and Staffing Strategies for Hospital 

Inpatient Units 

Wen-Ya Wang, University of Minnesota, 111 Church Street S.E., 

Minneapolis, MN, United States of America, wenya@ie.umn.edu, 

Diwakar Gupta 

We develop models for determining the number of nurses to assign to inpatient 

units when nurse absentee rates are heterogeneous. The assumption of 

heterogeneous absentee rates is supported by a detailed analysis of data from 

multiple nursing units of two hospitals. We establish structural properties of an 

optimal nurse assignment strategy and propose heuristics that perform well in 

numerical experiments. 


140 

■ SD30 


Energy and Commodity Real Options 



Chair: Nicola Secomandi, Tepper School of Business, Carnegie Mellon 

University, Pittsburgh, PA, 15213, United States of America, 

ns7@andrew.cmu.edu 

Co-Chair: Selvaprabu Nadarajah, Tepper School of Business, Carnegie 

Mellon University, 5000Forbes Avenue, Pittsburgh, PA, 15213, United 

States of America, snadarajah@cmu.edu 

1 - Storage Operations and Investments in an Electrical System 

Santhosh Suresh, PhD, Ross School of Business, University of 

Michigan, 701 Tappan St., Ann Arbor, MI, 48109, United States of 

America, sssandy@umich.edu, Roman Kapuscinski, Owen Wu 

Storage technology is becoming increasingly prevalent in today’s electric grid. We 

discuss the optimal siting and sizing of storage units in the view of minimizing 

overall operating costs accounting for transmission losses unlike individual profit 

maximization setting in most previous literature. We show the degree of storage 

pooling to be related structurally with: the efficiency of storage and transmission, 

variance, correlation and mean in the demand distribution; using a three node 

model. 

2 - Valuing Renewable Energy Resources 

Michael Pavlin, Assistant Professor, Wilfrid Laurier University, 

75 University Avenue West, Waterloo, ON, Canada, 

Michael.Pavlin06@Rotman.Utoronto.Ca, John Birge 

Wind, solar, and other renewable energy resources such as tides and rain have 

intermittent availability that makes effective operation and valuation challenging. 

We consider the valuation of wind generation for electricity given the structure of 

this market and the risk associated with commitment to production. 

3 - Managing Wind-based Electricity Generation with Storage and 

Transmission Capacity 

Yangfang Zhou, PhD Candidate, Carnegie Mellon University, 

5000 Forbes Ave., Pittsburgh, PA, 15213 United States of America, 

yangfang@andrew.cmu.edu, Alan Scheller-Wolf, 

Nicola Secomandi, Stephen Smith 

We consider the problem of operating wind-based electricity generation with 

storage (for instance, a grid-level electricity battery) in a market. We demonstrate 

that mismanaging such a system can significantly reduce its value. We also show 

that storage can greatly increase the monetary value of the wind farm, and, while 

it typically increases the total energy sold to the market, in certain situations it 

may also-– paradoxically—decrease the total wind energy sold to the market. 

4 - Valuing Multiple Exercise Options using Term Structure Models 

and Approximate Dynamic Programming 

Selvaprabu Nadarajah, Tepper School of Business, Carnegie Mellon 

University, 5000 Forbes Avenue, Pittsburgh, PA, 15213, 

United States of America, snadarajah@cmu.edu, Francois Margot, 

Nicola Secomandi 

We develop computationally efficient least squares Monte Carlo (LSM) and 

approximate linear programming (ALP) methods for valuing multiple exercise, 

such as energy swing and storage, options, using term structure price models. Our 

theoretical and numerical investigation shows the superiority of an LSM version 

based on value, rather than continuation value, function approximation. We also 

structurally relate our LSM and ALP methods. This suggests further research on 

optimal ALP relaxations. 

■ SD31 


Control the Inventory Risk of Stockout 



Chair: Jim Shi, Georgia State University, College of Business, Atlanta, 

GA, United States of America, shigsu@gmail.com 

1 - Stock-out Risk Control of the Production-Inventory System with 

Compound Poisson Demands 

Jim Shi, Georgia State University, College of Business, Atlanta, GA, 

United States of America, shigsu@gmail.com 

We study a single-product continuous-time production-inventory system with a 

constant replenishment rate, compound Poisson demands and lost-sales. The 

objective is first to derive the expected discounted inventory cost until the 

stockout occurrence for any given initial inventory level so as to investigate the 

stockout risk.

2 - The Effects of Risk Aversion and Service Constraints on a Twoperiod 

Newsvendor Policy 

Tatyana Chernonog, Bar Ilan University, Ramat Gan, 52900, Israel, 

Tatyana.Chernonog@biu.ac.il, Matan Shnaiderman 

This paper presents a two-period newsvendor problem with postponed pricing 

strategy. Risk aversion and service constraints are taken into account. These 

constraints may affect the optimal inventory capacity as well as the optimal 

prices. The effects of the constraints on supply chain performance are compared. 

The current two-period model leads to complicated and non-trivial solutions and 

insights, compared with one-period models, found in the literature. Exact 

analytical formulas are given. 

3 - A Risk-averse Dynamic Inventory Model with Fluctuating 

Purchasing Costs 

Sungyong Choi, Assistant Professor, Yonsei University, 

1 Yonseidae-gil, Wonju, 220-710, Korea, Republic of, 

sungyongchoi@gmail.com 

In my work, dynamic consumption models are studied for finite and infinite time 

Markov Decision Process (MDP) problems with fluctuating purchasing costs. I use 

additive general and exponential utility functions with finite and infinite time 

horizons. In addition, we assume that purchasing costs follow a discrete-state 

Markov chain. Then, I consider parametric sensitivity analysis for the model 

parameters using super-modularity. 

■ SD32 

32- North 223- CC 

Topics in Manufacturing I 


Chair: Sharafali Moosa, Associate Professor of OM (Edn.), LKC School 

of Business, Singapore Managment University, 50 Stamford Road, 

Singapore, 178899, Singapore, sharafalim@smu.edu.sg 

1 - Wire Path Planning for Multi-axis Wire EDM 

Zhi Yang, Kimberly Clark, 1400 Holcomb Bridge Road, Roswell, 

GA, 30076, United States of America, zack.yang@kcc.com, 

Richard A. Wysk, Sanjay Joshi 

Wire path generation methods are introduced. Cutting wire trajectories are 

generated under each specified coordinate system based on tangent visibility 

results. In order to accomplish the lead-in/lead-out wire path generation 

automation, an algorithm is developed to find paths for different types of tangent 

visibility results. A cut off plane is introduced to generate the cut off wire path as 

the final step of machine operation. 

2 - 3D Layout Optimization with Flexible Objects 

Sima Maleki, PhD Candidate, University of Tennessee, 416 East 

Stadium Hall, Knoxvile, TN, 37996, United States of America, 

smaleki@utk.edu, Rapinder Sawhney 

This work addresses the multi-layer 3-dimensional packaging of flexible objects 

and containers with the aim of maximizing the space utilization. The flexible 

containers are filled with rigid and non-rigid objects. Then, the containers are 

packed in a bin with fixed size. The problem of filling up containers with 

heterogeneous objects and locating the containers in a bin is formulated as a 

Mixed Integer Linear Programming model. Results illustrate the effectiveness of 

the layout. 

3 - Modelling Geometric Variation of Compliant Sheet Metal 

Assembly using Statistical Modal Analysis 

Abhishek Das, University of Warwick, International Digital 

Laboratory, WMG, Coventry, CV4 7AL, United Kingdom, 

Abhishek.Das@warwick.ac.uk, Darek Ceglarek 

Efficient modelling and analysis of geometric variation is crucial for quality 

control of deformable assemblies in automotive and aerospace industries, 

especially for emerging technologies such as remote laser welding. Statistical 

Modal Analysis using discrete-cosine-transformation is presented through a case 

study. The approach models modes of deformation for developing surface-based 

SPC. 

4 - A Failure-prone System with Delivery Deadline and 

Outsourced Maintenance 

Sharafali Moosa, Associate Professor of OM (Edn.), LKC School of 

Business, Singapore Managment University, 50 Stamford Road, 

Singapore, 178899, Singapore, sharafalim@smu.edu.sg, 

Hakan Tarakci, Shailesh Kulkarni 

We consider a production-maintenance system which is subject to random 

breakdowns. The manufacturer has to offer a firm delivery deadline to her 

customer. She also has to decide on the maintenance decisions with a 

maintenance contractor. A regenerative stochastic process is identified and 

analysed to develop the cost function over the finite horizon. Numerical examples 

will illustrate the optimization problem. Some managerial insights with regard to 

coordination will also be provided. 


141 

■ SD33 

SD33 


Joint Sessions MSOM/ENRE: Modeling & 

Optimization for Electric Vehicles 


Operations & Energy, Natural Res & the Envi/ Environment 

and Sustainability 


Chair: Mustafa Dogru, Alcatel-Lucent Bell Labs, 600 Mountain Avenue, 

Murray Hill, NJ, 07974, United States of America, 

mustafa.dogru@alcatel-lucent.com 

1 - Optimal Location of Public Electric Vehicle 

Charging Infrastructure 

Ramteen Sioshansi, Assistant Professor, The Ohio State University, 

240 Baker Systems, 1971 Neil Avenue, Columbus, OH, 43215, 

United States of America, sioshansi.1@osu.edu, Xiaomin Xi, 

Vincenzo Marano 

We develop a simulation optimization based approach to optimize the location of 

public electric vehicle charging infrastructure. The model includes budget 

constraints and accounts for driving and charging behavior. We use a case study 

based on the central Ohio region to demonstrate the model. 

2 - Modeling and Optimization for Electric Vehicle 

Charging Infrastructure 

Mustafa Dogru, Alcatel-Lucent Bell Labs, 600 Mountain Avenue, 

Murray Hill, NJ, 07974, United States of America, 

mustafa.dogru@alcatel-lucent.com, Matthew Andrews, Yue Jin, 

John Hobby, Gabriel Tucci 

We have developed a modeling framework to quantify the load on the power grid 

due to electric vehicles (EVs) in a metropolitan area during the course of a day. 

Our analysis uses driving patterns as an input and utilizes the data from publicly 

available surveys conducted in the US. We use this information to optimize the 

placement of public charging stations. The analysis also allows us to determine 

the percentage of households that are candidates for adopting EVs given the 

current technology. 

3 - An Integer Programming Model for Generating Vehicle Fleet 

Purchase Recommendations 

Daniel Reich, Operations Research Analyst, Ford Motor Company, 

2101 Village Road, MD 2122, Dearborn, MI, 48124, United States 

of America, dreich8@ford.com, Sandy Winkler, Erica Klampfl 

Sustainability and environmental impact are areas of growing importance to 

many of Ford’s fleet customers. New green vehicle technologies have emerged, 

which present organizations with an opportunity to increase the fuel economy of 

their fleets. “Fleet Purchase Planner (FPP)” (patent pending) is a software system 

designed to assist Ford’s fleet customers in planning their purchases. This talk 

introduces FPP and the integer programming model we use to provide customized 

purchase recommendations. 

4 - Controlled Plug-in Vehicle Charging in High Wind 

Penetration Scenarios 

Jeremy Michalek, Associate Professor, Carnegie Mellon University, 

Scaife Hall 324, 5000 Forbes Ave., Pittsburgh, PA, 15213, 

United States of America, jmichalek@cmu.edu, Allison Weis, 

Paulina Jaramillo 

We construct a MILP model of optimal power plant capacity expansion, dispatch, 

and plug-in vehicle charging to estimate the benefits of electric vehicle demand 

management using variable rate charging (less costly than bidirectional V2G). We 

estimate the implications for system cost, integration of intermittent wind power, 

wind curtailment avoided, and peaking plant construction avoided.

SD34 

■ SD34 

34- North 224 B - CC 

The Architecture of Coordination 

Cluster: New Product Development 


Chair: Alan MacCormack, Harvard Business School, Soldiers Field, 

Boston, MA, 02163, United States of America, amaccormack@hbs.edu 

1 - Innovation Analytics: using Complexity Science and Big Data 

for Sequencing of Products 

Nitin Joglekar, Boston University, One Silber Way, Boston, MA, 

United States of America, joglekar@bu.edu, Edward Anderson 

Many projects are sequentially dependent on one another because their features 

and pricing influence market tastes, and vice-versa, thus creating a complex 

system. Firms have begun to use “Big Data” from social media and allied sources 

to inform underlying decisions. We explore an information scaling framework for 

navigating through this complexity. 

2 - Managing Distributed Product Development Projects: 

Integration Strategies for Language and Geography 

Edward Anderson, University of Texas McCombs School of 

Business, 1 University Station B6500, Austin, TX, 78712, United 

States of America, Edward.Anderson@mccombs.utexas.edu, 

Xiaoyue Jiang, Aravind Chandrasekaran 

We study two barriers to distributed product development work, (1) geographic 

distance and (2) language. Our empirical setting involves 55 projects in 20 focal 

firms that outsource complex innovation work to their suppliers. We investigate 

the effects of two integration mechanisms used to overcome these barriers, (1) 

colocating firm and supplier personnel and (2) giving a “supply chain integrator” 

personnel that manage the DPD projects control over key aspects of their projects. 

3 - Transparent Environments for Leaky Modules 

Jim Herbsleb, Professor, Carnegie Mellon University, 5000 Forbes 

Avenue, 5307 Wean Hall, Pittsburgh, PA, 15213, United States of 

America, jdh@cs.cmu.edu, Laura Dabbish, Colleen Stuart, 

Jason Tsay 

Design rules enable concurrent work when organizational structure mirrors the 

technical architecture. Perfect modularity, however, is difficult to achieve. 

Imperfect modularity requires ongoing coordination of inter-module 

dependencies. In this qualitative study, we examine how an emerging generation 

of workspaces, explicitly designed around transparency, expose details of ongoing 

work and support surprisingly rich inferences about people, activities, technical 

direction, and dependencies. 

4 - The Human Cost of Complexity: How System Architecture 

Affects Developer Productivity 

Alan MacCormack, Harvard Business School, 

Soldiers Field, Boston, MA, 02163, United States of America, 

amaccormack@hbs.edu, Dan Sturtevant 

Prior work shows that systems with greater complexity suffer greater quality 

problems, in terms of experiencing defects. Few studies however, examine the 

human costs of complexity - that is, the impact on the productivity of system 

developers. We explore this relationship in a commercial software system. We 

find that i) developer productivity is impacted by where in the system 

architecture they work, and ii) only skilled developers are allowed to work on the 

most complex parts of the system. 

■ SD35 


Bid Price Control in Practice 



Chair: Darius Walczak, PROS, 3100 Main Street, Suite 900, Houston, 

TX, 77002, United States of America, dwalczak@prospricing.com 

1 - Bid and List Price Trends in RM and Forward-Looking 

Consumers 

Ming Hu, Rotman School of Management, University of Toronto, 

Toronto, Canada, Ming.Hu@rotman.utoronto.ca, Oded Berman, 

Zhan Pang 

We characterize a probabilistic property of the bid price process driven by an 

optimal policy in a general RM framework: the bid price process always follows 

an upward trend. We explore a couple of applications of this structural property. 

For example, we show that the optimal list price process also follows an upward 

trend if the customers’ WTP does not decrease over time. We draw implications 

on forward-looking behavior of strategic consumers. 


142 

2 - Randomization Approaches for Network RM with Choice 

Behavior 

Sumit Kunnumkal, Assistant Professor, Indian School of Business, 

Gachibowli, Hyderabad, A.P, AP, India, sumit_kunnumkal@isb.edu 

We present new approximation methods for the network RM problem with 

customer choice behavior. The starting point for our methods is a dynamic 

program that allows randomization. An attractive feature of this dynamic 

program is that the size of its action space is linear in the number of itineraries. 

We present two approximation methods that build on this dynamic program and 

use ideas from the independent demands setting. 

3 - Estimating Mixtures of Willingness-To-Pay Distributions from 

Hotel Rate and Occupancy Data 

Chris Keller, Assistant Professor, East Carolina University, College 

of Business, 3136 Bate Building, Greenville, NC, 27858, 

United States of America, kellerc@ecu.edu 

This presentation uses daily rate and occupancy data for a mid-market hotel and 

it competitive set to estimate mixtures of distributions for customer willingnessto-pay. 

The presentation considers various numbers of mixture distributions and 

various types of distributions in order to maximize goodness-of-fit to observed 

rates and occupancies. The distributions are also collectively varied by day, week, 

and month. 

4 - Practical Aspects of Network Bid Price Control 

Darius Walczak, PROS, 3100 Main Street, Suite 900, Houston, TX, 

77002, United States of America, dwalczak@prospricing.com 

Many network carriers rely on bid price control in their day-to-day revenue 

management. At the same time they are also often task with providing estimates 

of expected revenue, load factor or other measures. We discuss challenges that 

arise in deriving those business metrics under bid price control in the network 

setting. 

■ SD36 


Pricing and Energy Cost of Cloud 

Computing Services 



Chair: Parijat Dube, IBM, TJ Watson Research Center, Hawthorne, NY, 


Co-Chair: Genady Ya. Grabarnik, St. John’s University, Queens, NY, 

United States of America, grabarng@stjohns.edu 

1 - Cost Comparison between Infrastructure Cloud Services 

Huan Liu, Accenture Technology Labs, 50 W San Fernando St., 

San Jose, CA, 95113, United States of America, 

huan.liu@accenture.com 

Infrastructure cloud service providers, such as Amazon EC2, differ widely from 

each other in terms of their price structure, which makes it very difficult to 

perform an apple-to-apple comparison. In this paper, we describe a methodology 

to dissect the price structure into its individual components, including CPU, 

memory, and hard disk capacity. We then use the price components to compare 

offerings from a few dominate cloud providers. 

2 - Selective Commitment and Selective Margin: Techniques to 

Minimize Cost in an IaaS Cloud 

Yu-Ju Hong, Purdue University, 2550 Yeager Rd Apt 12-1, 


yujuhong@purdue.edu, Mithuna Thottethodi, Jiachen Xue 

We aim to minimize compute cost for day-to-day operations in an Infrastructureas-a-Service 

(IaaS) cloud, while satisfying statistical response-time targets. First, a 

dynamic programming algorithm is used to compute load-dependent margins (a 

set of servers to handle unexpected load bursts). Second, we exploit the 

differences of price/commitment of on-demand and reserved virtual machine 

instances. Simulations with real Web server traces using Amazon EC2 cost model 

show 21% of average cost saving. 

3 - Online Dynamic Capacity Provisioning in Data Centers 

Minghong Lin, Caltech, 1200 E. California Blvd., Pasadena, 

United States of America, mhlin@caltech.edu, Adam Wierman, 

Eno Thereska, Lachlan Andrew, Zhenhua Liu 

Power consumption imposes a significant cost for data centers, yet much of that 

power is used to maintain excess service capacity during periods of low load. In 

this work, we study how to avoid such waste via an online dynamic capacity 

provisioning. We show that the optimal offline algorithm has a simple structure 

when viewed in reverse time, and develop a lazy online algorithm which is 3competitive. 

Additionally, we analyze the traditional receding horizon control and 

introduce a new variant.

4 - Distributed Resource Allocation in Large Scale Data 

Intensive Computing 

Cathy Xia, Professor, Ohio State University, 210 Baker, 

1971 Neil Ave., Columbus, OH, 43210, United States of America, 

xia.52@osu.edu, Kevin Liu, Ness Shroff 

In cloud computing, distributed management of resources for effective dataintensive 

in-network computing is critical. We consider the optimal resource 

allocation problem in this context using a network utility optimization 

framework. We provide second-order based distributed control algorithms with 

guaranteed quadratic convergence speed. 

5 - Energy Cost and Availability Considerations in Cloud 

Services Composition 

Parijat Dube, IBM, TJ Watson Research Center, Hawthorne, NY, 


Compute service provider has the flexibility to use its own private nodes or use 

nodes from public domain when composing services. However there are cost and 

availability concerns associated with both private nodes and public nodes. The 

goal is to construct compute services which satisfy end-to-end delay requirements 

of a job while minimizing cost and maintaining reliability. We propose simple 

heuristics and their empirical verification. 

■ SD37 


Alternative Fuels, Stations, and Vehicles I 

Sponsor: Location Analysis 


Chair: Michael Kuby, Professor, Arizona State University, Tempe, AZ, 

85287-5302, United States of America, mikekuby@asu.edu 

1 - Effects of Electric Vehicles Adoption Pathway on the 

US Energy System 

Dong Gu Choi, Georgia Institute of Technology, 765 Ferst Drive, 

NW, Room 336, Atlanta, GA, 30332, United States of America, 

doonggus@gatech.edu, Frank Kreikebaum, Deepakraj Divan, 

Valerie Thomas 

Adoption of electric vehicles (EVs) can affect the development of the electric 

power system, the integration of intermittent renewable energy, and the overall 

efficiency of the vehicle fleet, with implications for costs, environmental impacts, 

and policy implementation. Using a set of models spanning the light-duty vehicle 

fleet and the electric power system, we evaluate numerous scenarios for EV 

integration into the energy system in eastern U.S. and argue linkages among the 

considerations. 

2 - Shortest Routes with Refueling Stops 

Pitu Mirchandani, Arizona State University, 699 South Mill 

Avenue, Tempe, AZ, 85287, United States of America, 

pitu@asu.edu, Jonathan Adler, Guoliang Xue 

Unlike the problem of shortest paths and shortest path trees, when refueling stops 

are not required, and where well known algorithms exist for finding shortest path 

trees, the problem of finding shortest routes with refueling may require detouring 

for refueling, which may result in walks with cycles. A polynomial algorithm is 

presented for finding the set of shortest walks for all node pairs in the network. 

3 - The Vehicle Scheduling Problem with Electric Vehicles 

Jonathan Adler, Arizona State University, 699 S. Mill Ave., 

Tempe, AZ, 85281, United States of America, jonadler@asu.edu, 

Pitu Mirchandani 

The Vehicle Scheduling Problem (VSP) consists of assigning a fleet of vehicles to 

travel a certain set of trips with given start and end times. VSP changes when the 

fleet is composed of electric vehicles, since the vehicles can only travel a limited 

distance before needing to recharge or exchange batteries. In this talk we will 

define and formulate the problem of using electric vehicles in these scheduling 

settings, show the problem is NP-hard, and present a heuristic solution. 

4 - Distributed Joint Budget Allocation and Location Problem for 

Alternative-fuel Stations 

Ismail Capar, Assistant Professor, Texas A&M University, TAMU 

3367, College Station, TX, 77843, United States of America, 

capar@tamu.edu, In-Jae Jeong 

Government agencies (GAs) around the world are providing grants to service 

providers (SPs) for building infrastructure for alternative-fuel vehicles (AFVs). 

While GAs aim to increase the number of adopters; SPs want to maximize profit. 

These objectives may not be well aligned. We consider this joint funding 

allocation and AFV station location problem under partial information sharing 

where parties do not reveal the private information of individual objective 

functions and budget constraints. 


143 

5 - Hydrogen Charging Station Location Analysis with Scheduling 

and Routing Considerations 

Jamie Kang, Institute of Transportation Studies, University of 

California, Irvine, 4000 AIRB, Irvine, CA, United States of 

America, jekang@uci.edu, Will Recker 

A facility location problem with full-day scheduling and routing considerations is 

proposed for hydrogen refueling stations in the early adoption stage. The model 

uses the set covering model as a location strategy, and the Household Activity 

Pattern Problem as the scheduling and routing tool. The formulation isolates each 

vehicle’s routing problem from other vehicles as well as the set covering problem, 

and uses a modified column generation that finds a column with a negative 

reduced price. 

■ SD38 

SD38 


Public Services Applications 



Chair: Robert Saltzman, Professor, Decision Sciences, San Francisco 

State University, 1600 Holloway Avenue, San Francisco, CA, 94132, 

United States of America, saltzman@sfsu.edu 

1 - Planning for an Aging Fleet of Shuttle Vehicles with Simulation 

Robert Saltzman, Professor, Decision Sciences, San Francisco State 

University, 1600 Holloway Avenue, San Francisco, CA, 94132, 

United States of America, saltzman@sfsu.edu 

A university operates an aging fleet of shuttle vehicles that facilitate travel by its 

affiliates to, from and around campus. This paper presents an animated discreteevent 

simulation model that can be used by campus planners to make capital 

planning decisions about the fleet. Experiments were run to identify the best type 

of shuttle to replace existing vehicles and to quantify how different fleet 

configurations might accommodate various future increases in ridership. 

2 - Using Stochastic Processes and Behavioral Analysis to Support 

Search and Rescue Operations 

Johnathon Dulin, Concurrent Technologies Corporation, 771 

Fairdale Ct, Castle Rock, CO, 80104, United States of America, 

dulinj@ctc.com 

Optimizing search patterns has long been a focus in the OR community, and it is 

generally accomplished in two veins: probabilistic analysis to develop a fixed 

pattern or automated learning to direct progress. But is it possible that the two 

could meet in the middle? In this talk we describe an approach that does just this. 

By merging the behavioral elements of game theory with the policy optimization 

of MDPs, our combined approach finds smarter search paths by allowing real-time 

updates. 

3 - Police Control Room Emergency Response Behaviour Analysis 

Partha Datta, Associate Professor, IIM Calcutta, Diamond Harbour 

Road, Joka, Calcutta, WB, 700104, India, ppdatta@iimcal.ac.in 

The primary issue explored in this research is how demand distortions called the 

‘bullwhip’ effect, create inefficiencies for responding to emergency calls in police 

control rooms. In this paper, we demonstrate the existence of bullwhip effect in 

the emergency service sector through case study of Indian police control room. 

We explore the impact of call handler’s behavioral differences. 

4 - Evaluating an Automatic Data Sharing Tool for 

Municipal Governments 

Stacy Hobson, IBM T. J. Watson Research Center, 19 Skyline 

Drive, Hawthorne, NY, 10532, United States of America, 

stacypre@us.ibm.com 

Coordination of data from multiple departmental applications are critical for 

municipalities to provide services to its citizens, yet the prevalence of silos 

between applications requires significant amounts of manual data sharing. We 

present our work in creating the Shared Data Manager (SDM) – a tool to for 

loosely-coupled and customizable application integration and findings of our 

study on the use of the SDM as a means to improve data sharing and 

coordination within municipal governments.

SD39 

■ SD39 


Electronic and Mobile Retailing Services 



Chair: Howard Hao-Chun Chuang, PhD Candidate, Mays Business 

School at Texas A&M University, College Station, TX, 77843-4217, 

United States of America, hchuang@mays.tamu.edu 

1 - From E-commerce to M-commerce: Business Value of Mobile 

Sales Channel 

Yen-Chun Chou, Arizona State University, United States of 

America, Yen-Chun.Chou@asu.edu, Benjamin Shao 

M-commerce represents an emerging subset of e-commerce and is projected to 

reach 31 billion by 2016. While firms embrace mobile sales channels, one critical 

question is how they can reap values from the new channels. We aim to answer 

the question by considering prior e-commerce establishments, timing of mobile 

channel implementation, and a firm’s general business operations. Our study 

contributes to research and practice by exploring factors that influence mcommerce 

performances. 

2 - Visualization of Contents-oriented Services 

Chie-Hyeon Lim, Pohang University of Science and Technology, 

Engineering Building #4-316, Pohang, Korea, Republic of, 

arachon@postech.ac.kr, Kwang-Jae Kim 

A contents-oriented service is a type of service in which service contents have 

major roles in service provision. Examples include car-infotainment services and 

internet-based education services. This talk presents issues regarding the 

visualization of contents-oriented services. The conventional service blueprint and 

other existing tools for service visualization are reviewed for the visualization of 

contents-oriented services; the limitations and alternatives are also discussed. 

3 - Impact of Website Functionalities on e-Retailer Sales: An 

Econometric Analysis 

Guanyi Lu, PhD Candidate, Mays Business School at Texas A&M 

University, 320N, 4217 TAMU, College Station, TX, 77843-4217, 

United States of America, glu@mays.tamu.edu, Xiaosong Peng, 

Howard Hao-Chun Chuang, Gregory Heim 

Utilizing a three-year panel data set of top500 Internet retailers, we examine the 

effects of various web functionalities on retailer sales. The results suggest that 

only value-added web functionalities exhibit a direct and sizable effect on web 

sales; the effects of fundamental and customer-centered web functionalities 

appear to be non-substantial. 

4 - Operating Efficiency of Internet Retailers 


at Texas A&M University, College Station, TX, 77843-4217, United 

States of America, hchuang@mays.tamu.edu, Gregory Heim, 

Rogelio Oliva 

We empirically assess the operating efficiency of Internet retailers using stochastic 

non-smooth envelopment of data. In the first stage, we adopt concave 

nonparametric least squares (CNLS) to tackle estimation bias incurred by 

contextual variables. In the second stage, we decompose CNLS residuals and 

derive efficiency estimates. We conclude by discussing the methodological 

implications of contextual variables and the managerial implications of e-retail 

efficiency. 

■ SD40 


Ecosystem Conservation 

Sponsor: Energy, Natural Res & the Envi/ Environment and 

Sustainability 


Chair: Hayri Onal, University of Illinois at Urbana-Champaign, 

305 Mumford Hall, Urbana, IL, 61801, United States of America, 

h-onal@illinois.edu 

1 - Optimal Surveillance and Eradication of Invasive Species 

Robert Haight, USDA Forest Service, 1992 Folwell Ave., 

St. Paul, MN, 55108, United States of America, rhaight@fs.fed.us, 

John Kean, Rebecca Epanchin-Niell, Andrew Liebhold, 

Ludek Berec 

Cost-effective surveillance strategies are needed for efficient responses to 

biological invasions of terrestrial ecosystems. Surveillance strategies are applied in 

environments with continual invasion pressure, unknown number, size, and 

location of populations, and trade-offs between surveillance effort and 


144 

management costs. We develop a decision model that accounts for these features 

and determines the long term equilibrium sampling effort that minimizes total 

expected costs of new invasions. 

2 - A Modeling Framework for Life History-based 

Conservation Planning 

Sandor Toth, Assistant Professor, University of Washington, School 

of Environmental and Forest Scien, Seattle, WA, 98195, United 

States of America, toths@uw.edu, Eileen S. Burns, Robert Haight 

We present a new population protection function for reserve site selection models 

that determines whether minimum habitat requirements are present in a given 

reserve network. We embedding the function in an integer programming-based 

reserve selection model and apply it to a case study of Myotis bat conservation on 

Lopez Island, United States. Finally, we highlight the need in future reserve 

design models to capture intra- and inter-species competition for habitat 

resources. 

3 - Robust Network Design for Multispecies Conservation 

Bistra Dilkina, Cornell University, Ithaca, NY, 14853, United States 

of America, bistra@cs.cornell.edu, Ronan Le Bras, Yexiang Xue, 

Carla Gomes 

Ideally conservation plans should ensure the connectivity of habitat areas is 

robust to changes in the landscape. We introduce the Shortest Maximally 

Connected Network Design problem, which captures the need to both protect 

multiple edge-disjoint paths as well as minimize the species-specific resistance 

length of such paths, and provide solution approaches. Our models extend to 

multispecies conservation, providing a tool to systematically study tradeoffs 

between species, in addition to costs. 

4 - Designing a Conservation Reserve with Optimum 

Functional Connectivity 

Hayri Onal, University of Illinois at Urbana-Champaign, 

305 Mumford Hall, Urbana, IL, 61801, United States of America, 

h-onal@illinois.edu, Yicheng Wang 

Spatial contiguity of conservation reserves is crucial for effective functioning of 

the reserved areas and improving species survival chances. Just physical 

contiguity may not always be adequate, quality of the connecting areas is 

important also. We present a linear MIP model for determining an optimal 

functionally connected reserve while meeting other conservation criteria and an 

empirical application to a real data set. 

■ SD41 


Uncertainty in Energy 


Chair: Bruno Flach, RSM, IBM Research, Brazil, bflach@br.ibm.com 

1 - Payment Uncertainty Management in a Combined Market of 

Energy and Reserve 

Taìsa Felix, Universidade de Brasìlia, SHCGN 703 Bloco L apto 103, 

Brasìlia, BR, 70730712, Brazil, taisafelix@gmail.com 

As a rule, power system markets operate considering bid cost minimization 

(BCM) in clearing processes. This procedure has inconsistency because the total 

BCM is different from the load payment. This work introduce a combined market 

model of energy and reserve operating under Payment Minimization (PM), 

transforming the problem in a bi-level linear model, which can be solved using 

GAMSÆ. Expected payment is evaluated using Monte Carlo simulation, 

providing results to evaluate the PM energy policy. 

2 - Robust Unit Commitment Problem with Demand and 

Market Price Uncertainty 

Michele Samorani, Alberta School of Business, University of 

Alberta, Edmonton, AB, Canada, samorani@ualberta.ca, 

Manuel Laguna, Fabio Furini 

The unit commitment problem must be solved periodically by the grid operator in 

order to find the optimal energy production plan. We extend existing solution 

methods by taking into account the uncertainty of the demand and of the energy 

prices. We test the performance of robust optimization strategies on real data of 

energy demand. Our results suggest that significant savings can be achieved by 

adopting the minimax regret criterion. 

3 - An MIQP Approach to the Determination of Analogous 

Reservoirs 

Bruno Flach, RSM, IBM Research, Brazil, bflach@br.ibm.com 

Oil companies are constantly faced with decision under uncertainty problems 

related to the analysis of potential investments in target reservoirs. Our work 

focuses on the determination of analogous reservoirs as a way to estimate 

unknown properties and production forecatsts by formulating it as a mixed 

integer quadratic programming (MIQP) problem. Computational results on the 

application of different solution algorithms to a realistic large-scale problem will 

be discussed.

4 - An Optimal Solution to the Discrete Price-based Dynamic 

Economic Dispatch Problem 

Pawel Kalczynski, Professor of Inf. Syst. and Dec. Sci., California 

State University-Fullerton, Mihaylo College of Business and Econ., 

P.O. Box 6848, Fullerton, CA, 92831-6848, 

United States of America, PKalczynski@fullerton.edu 

A new approach to the price-based dynamic economic dispatch (PBDED) of fossilfuel 

electricity generators is presented; temperature is the decision variable. The 

objective is to maximize profit over a discrete time horizon when the fuel and 

electricity prices (or forecasts) are given. An optimal solution can be found using 

the dynamic-programming approach. The model allows for representing different 

types of fossil-fuel generators; it can used to compare bidding strategies and price 

forecasts. 

■ SD42 


Traffic and ITS 


Chair: Mostafa Badakhshian, PhD Student, Concordia University, 

2525 Boul Cavendish, Montreal, QC, H4B 2Y6, Canada, 

m.badakhshian@gmail.com 

1 - Priority Traffic Signal Control in an Actuated Control System 

Larry Head, Department Head, Systems and Industrial 

Engineering, University of Arizona, P.O. Box 210020, Tucson, AZ, 

85721, United States of America, larry@sie.arizona.edu, Jun Ding, 

Qing He, Wei Wu 

Traffic signals provide service for multiple modes of travelers including vehicles, 

trucks, transit, pedestrians and bicycles, and emergency vehicles. Historically, the 

treatment for each mode of traveler was addressed somewhat independently 

within the normal traffic signal operation (vehicles). This paper addresses a 

decision framework for prioritizing requests for service from multiple modes and 

accommodating the authorized requests within an integrated framework. 

2 - Monitoring Traffic Shock Waves on the Roads via Massive 

GPS Probes 

Young-Ji Byon, Assistant Professor, Khalifa University of Science 

Technology and Research, Al Saada St. (19th) and 4th Street, 

P.O. Box 127788, Abu Dhabi, 127788, United Arab Emirates, 

youngji.byon@kustar.ac.ae, Raja Jayaraman, Young Seon Jeong, 

Said Easa 

Until now, intelligent transportation system (ITS) applications have focused on 

speed, density and volume of the general traffic on particular sections of roads. 

A more direct method of traffic monitoring can be achieved by capturing 

“shockwaves” with GPS probes. Shockwaves generally travel backwards on the 

road and can create grid-locks at major intersections. By monitoring the 

movements of the shockwaves, real-time route guidance can be improved. 

3 - Travel Time Reliability for Extended Freeway Facilities 

Ali Hajbabaie, Postdoctoral Research Scholar, Institute for 

Transportation Research and Education-North Carolina State 

University, 909 Capability Drive, Suite 3600, Research Building IV, 

Raleigh, NC, 27606, United States of America, ahajbab@ncsu.edu, 

Behzad Aghdashi, Nagui Rouphail 

In this study, we introduce a methodology to incorporate reliability measures in 

travel time estimation of extended freeway facilities. The methodology is designed 

to account for non-recurring sources of congestion such as inclement weather 

events and traffic incidents. Travel time reliability method, several reliability 

performance measures, and numerical findings of a real-world case study will be 

explained. 

4 - An Analytical Approximation of the Joint Distribution of Queuelengths 

in an Urban Network 

Carter Wang, Massachusetts Institute of Technology, 235 Albany 

Street, #2021, Cambridge, MA, 02139, United States of America, 

carterw@mit.edu, Carolina Osorio 

We propose a methodology to approximate the joint distribution of queue lengths 

in an urban network. The main issues are dimensionality and modeling 

dependency analytically. We approximate network-wide joint distributions by 

combining a finite capacity queueing network model with an analytical 

aggregation-disaggregation technique. The approach is analytical and 

differentiable and can be embedded within optimization frameworks to address 

efficiently a variety of urban transportation problems. 


145 

■ SD43 


Joint Session RAS/CPMS: Panel Discussion 

Continued: Analytics and Intermodal 

Sponsor: Railway Applications & CPMS, The Practice Section 


Chair: Bruce Patty, Vice President, Veritec Solutions, 824 Miramar 

Terrace, Belmont, CA, 94002, United States of America, 

bpatty@veritecsolutions.com 

1 - Roundtable on Analytics and Intermodal 

Moderator: Bruce Patty, Vice President, Veritec Solutions, 824 

Miramar Terrace, Belmont, CA, 94002, United States of America, 

bpatty@veritecsolutions.com, Panelists: Michael Gorman, 

Steve Sashihara, Dharma Acharya 

This roundtable will take place during two sessions. During this second session, 

there will be three panelists from the OR community (Mike Gorman from the 

University of Dayton, Steve Sashihara from Princeton Consultants, and Dharma 

Acharya from CSX) who will share their thoughts about current and potential 

applications of analytics to the intermodal industry. After their opening 

presentations, there will be an opportunity for the audience to ask questions and 

participate in an open discussion. 

■ SD44 


Information in Supply Chain 


SD44 

Chair: Eirini Spiliotopoulou, PhD Candidate in Supply Chain 

Management, MIT-Zaragoza International Logistics Program, 

Zaragoza Logistics Center, Zaragoza, 50197, Spain, 

espiliotopoulou@zlc.edu.es 

1 - Optimal Replenishment Policies for Deteriorating Items Based 

on the Forecasting Method of Grey Model 

Shoufeng Ji, Professor, Northeastern University, No.11, Lane 3, 

WenHua Road, HePing District, Liaoning, Shenyang, 110004, 

China, sfji@mail.neu.edu.cn, Haifei Yu 

The collected and forecasting profit model was briefly introduced focus on a series 

of conditions such as limited inventory, stochastic replenishment periods, 

uncertain of whether lending stock and service level constraint. Then determinate 

the optimal replenishment quantity and optimal policy and utilize MATLB realtime 

simulation to analyze how lending stock rate and stochastic replenishment 

periods effluence the profit .Finally it is proved that model is validated to the 

policy. 

2 - Mechanisms to Induce Buyer Forecasting: Do Suppliers Always 

Benefit from Better Forecasting? 

Thunyarat (Bam) Amornpetchkul, University of Michigan, 

701 Tappan St., Ann Arbor, MI, 48109, United States of America, 

thunyara@umich.edu, Ozge Sahin, Izak Duenyas 

We consider a supplier’s contract offerings to a buyer who may obtain more 

precise or more accurate demand forecasts closer to the selling season. We 

investigate optimal contract types both when the supplier is certain and uncertain 

of the buyer’s forecasting capability. We propose new mechanisms which 

effectively induce the buyer to obtain better forecasts and always yield a higher 

profit to the supplier, compared to conventional contract types considered in the 

literature. 

3 - The Value of Real-time Advance Demand Information in Lotsizing 

Decision Making 

Qiannong Gu, Sam Houston State University, 1419 Eden Meadows 

Dr, Spring, TX, 77386, United States of America, 

qian_nong@yahoo.com, Basheer Khumawala, John Visich 

Advance demand information captured by a real time information tracking 

technology offers an opportunity for the manufacturer in a supply chain to make 

more efficient lot-sizing decisions. This research studies the trade-off between lot 

setup costs and inventory holding costs in a lot-sizing problem by using the realtime 

ADI.

SD45 

4 - Information Reliability in Supply Chains: The Case of 

Multiple Retailers 

Eirini Spiliotopoulou, PhD Candidate in Supply Chain 

Management, MIT-Zaragoza International Logistics Program, 

Zaragoza Logistics Center, Zaragoza, 50197, Spain, 

espiliotopoulou@zlc.edu.es, Mustafa Cagri Gurbuz, 

Karen Donohue 

Multiple retailers consider to form a pooling coalition and delegate inventory 

management to a benevolent central planner. Each retailer faces uncertain 

demand and has private information about it due to his proximity to the market. 

We focus on whether reliable demand information sharing occurs between 

retailers and the central planner and on the impact of a) the size of the pooling 

coalition and b) the market uncertainty on the level of trust and trustworthiness 

of supply chain parties. 

5 - Information Sharing in Supply Chain of Durable Goods 

Neda Khanjari, PhD Student, Northwestern University, 2145 

Sheridan Road, Evanston, IL, 60202, United States of America, 

neda@u.northwestern.edu, Seyed Iravani, Hyo duk Shin 

We study a supply chain of a durable good product consisted of a retailer with 

advance demand information and a manufacturer. The retailer and the 

manufacturer play a pricing game. We study how the policy of the retailer to 

share his information with the manufacturer depends on durability of the product 

and other parameters of the model. We find conditions under which the retailer 

shares information with the manufacturer and study the impact of information 

sharing on the supply chain profits. 

■ SD45 


Panel Discussion: WORMS: Advice on Promotion to 

Full from Associate Professor 

Sponsor: Women in OR/MS 


Chair: Mary E. Kurz, Associate Professor, Clemson University, Industrial 

Engineering, 110 Freeman Hall, Clemson, SC, 29634, United States of 

America, MKURZ@clemson.edu 

1 - WORMS: Advice on Promotion to Full from Associate 

Professor Panel 

Moderator: Mary E. Kurz, Associate Professor, Clemson University, 

Industrial Engineering, 110 Freeman Hall, Clemson, SC, 29634, 

United States of America, MKURZ@clemson.edu, Panelists: 

Aleda Roth, Susan Sanchez, Beril Toktay, Julie Higle 

In this panel, we will discuss the experiences, observations and receive advice 

from several successful female faculty housed in different types of departments 

and institutions, focusing on the transition from associate to full professor. 

■ SD46 


Joint Session ORG/TMS: Organizing for Innovation in 

the Digitized World 

Sponsor: Organization Science & Technology Management 


Chair: Youngjin Yoo, Temple University, 1810 N. 13th Street, 

Speakman Hall, Philadelphia, PA, 19122, United States of America, 

yxy23yoo@gmail.com 

1 - Digital Science and Knowledge Boundaries 

Deborah Dougherty, Rotgers University, 94 Rockafeller Rd, Janice 

H. Levin Rm 26, Piscataway, NJ, 08854, United States of America, 

doughert@business.rutgers.edu 

Drug discovery has many unpredictable interdependencies. Science digitalization 

affords ways to measure, analyze, and model chemical compounds, diseases, and 

biology.We build on epistemic cultures to develop theory for integrating digital 

sciences. Digitalization creates essential knowledge and new knowledge 

boundaries around central activities of innovation. We explain how innovation 

activities can be transformed to integrate digital science into drug discovery. 


146 

2 - Reconfiguring Boundary Relations: Robotic Innovations in 

Pharmacy Work 

Eivor Oborn, Professor, School of Management, Royal Holloway 

University of London, Egham London, United Kingdom, 

eivor.oborn@rhul.ac.uk, Wanda Orlikowski, JoAnne Yates, 

Michael Barrett 

We explore the influence of robotic innovations on the boundary dynamics of 

three different occupational groups and extend conceptually Pickering’s tuning 

approach to examine the entanglement of mechanical elements and digital 

inscriptions. The robot’s hybrid and digital materiality over time reconfigured 

boundary relations among the three occupational groups with important and 

contradictory consequences for the pharmacy workers’ skills, jurisdictions, status 

and visibility. 

3 - Catalyzing Collaboration Amongst Strangers: Field 

Experimental Evidence 

Karim Lakhani, Assistant Professor, Harvard Business School, 

Boston, MA, United States of America, k@hbs.edu, Patrick Gaule, 

Christoph Riedl, Kevin Boudreau, Anita Woolley 

We used a field experiment to investigate the emergence of collaboration in 52 

groups (5 members) of software developers. We test theoretical predictions from 

different disciplines – regarding the factors leading to the emergence of 

collaboration. We find that monetary rewards powerfully shape the effort that 

workers exert in their work. However, we also find substantial evidence for the 

existence of peer effects, whereby individuals exert more effort when other team 

members work harder. 

4 - Digital Innovation and the Division of Innovative Labor: Digital 

Controls in the Automotive Industry 

Jaegul Lee, Wayne State University, Detroit, MI, 48202, 

United States of America, jaegul.lee@wayne.edu 

Much of organizational scholarship holds that, accompanying a shift toward 

modular structures, suppliers tend to engage in invention around the components 

they supply. In this study, we find that in the wake of a major shift in the digital 

controls technology, suppliers engage in relatively less component innovation in 

comparison with their OEMs. By analyzing the evolution of automotive emission 

control systems, we explain this shift by offering two views about the interfirm 

division of labor. 

■ SD47 

47- North 230- CC 

Traffic Control Schemes and Applications II 




Chair: Qing He, IBM Watson Research Center, 1101 Kitchawan Rd, 


qhe@us.ibm.com 

1 - Energy-efficient Traffic Management: A Microscopic 

Simulation-based Approach 

Carolina Osorio, Assistant Professor, Massachusetts Institute of 

Technology, Room 1-232, 77 Massachusetts Avenue, Cambridge, 

MA, 02139, United States of America, osorioc@mit.edu, 

Kanchana Nanduri 

We use detailed traffic and fuel models to address a fixed-time signal control 

problem for the Swiss city of Lausanne. We have developed an optimization 

framework with good short-term performance using a metamodel that combines 

information from the simulator and an analytical queuing model. The method is 

able to identify signal plans with improved travel times and energy consumption 

even under tight computational budgets. 

2 - Optimal Speed Trajectory for Fuel Consumption Reduction at 

Signalized Intersection 

Xiaozheng He, University of Minnesota, 500 Pillsbury Dr. SE, 

Minneapolis, MN, 55414, United States of America, 

hexxx069@umn.edu, Henry Liu 

In this study, we formulate optimal speed trajectories at signalized intersection as 

solutions to a multi-phase optimal control problem, which considers the impact 

from vehicle queues due to traffic lights. By applying the pseudospectral method, 

this multi-phase optimal control problem is reformulated into a nonlinear 

programming problem which can be solved effectively. We demonstrate the 

effectiveness of the proposed model by an example.

3 - A Bi-level Formulation for the Combined Dynamic Equilibrium 

Based Traffic Signal Control 

Kien Doan, Purdue University, 550 Stadium Mall Drive, West 

Lafayette, IN, 47906, United States of America, dtkien@gmail.com, 

Satish Ukkusuri 

This paper formulates a combined dynamic signal control optimization (SCO) 

with dynamic user equilibrium (DUE) using a bi-level formulation. A path-based 

CTM is embedded as the traffic flow model to capture the different O-D flows at 

all cells and links and there is no holding-back problem in the entire network. A 

specialized solution technique will be developed for this problem and rigorously 

studied to solve general networks. We use a heuristic algorithm based on the 

projection method to solve the lower level (DUE) problem. A mixed integer 

programming (MIP) and other extensions are proposed to solve the upper level 

signal optimization. Extensive numerical results will illustrate the benefits of 

using this model and algorithm. 

4 - Traffic Control Agency Planning and Signal Optimization for 

Special Planned Events 

Qing He, IBM Watson Research Center, 1101 Kitchawan Rd, 


qhe@us.ibm.com, Arun Hampapur, Xuan Liu 

Large scale special planned events, attracting both high volume of pedestrians and 

vehicles, result in significant nonrecurrent congestion as well as queue spillover. 

In this research, we proposed a mixed intersection control model utilizing both 

manual control from traffic control agency (TCA) and automatic signal control. 

Different with traffic signals, TCA can effectively balance queues and increase 

throughput, and prevent intersection capacity drop caused by large volume of 

pedestrians. 

■ SD48 


Logistics System Planning 

Sponsor: Transportation Science & Logistics/ Freight 

Transportation & Logistics 


Chair: Alan Erera, Professor, Georgia Institute of Technology, 765 Ferst 

Dr NW, Atlanta, GA, 30339, United States of America 

1 - A Pickup and Delivery Problem using Crossdocks and 

Truckload Lane Rates 

Kathleen Lindsey, Georgia Institute of Technology, Atlanta, GA, 

United States of America, kate.abercrombie@gatech.edu, 

Alan Erera, Martin Savelsbergh 

A path-based integer programming model is presented for a shipper pickup and 

delivery planning problem to move freight from origins to destinations. Each 

shipment is moved either direct via a less-than-truckload (LTL) carrier or 

consolidated with other shipments and moved by truckload routes. When using a 

truckload carrier, the shipper takes advantage of contracted lane rates that are 

significantly less than comparable LTL prices for a full truckload. For larger 

instances that cannot be directly solved using commercial solvers, an IP-based 

local search approach is developed. 

2 - A Network Flow Methodology to Estimate Empty Trips in Freight 

Transportation Models 

Satish Ukkusuri, Associate Professor, Purdue University, 550 

Stadium Mall Drive, G175F, West Lafayette, IN, 47906, United 

States of America, sukkusur@purdue.edu, Rodrigo Mesa-Arango 

This research presents a novel methodology to estimate the number of empty 

trips useful for regional freight planning. This methodology is based on a time 

expanded network flow approach that captures behavioral and operational characteristics 

of the carriers. The model is formulated as a linear optimization problem 

that minimizes the system costs associated to truck trips. The model is validated 

with a commercial vehicle survey from Colombia. 

3 - Sugarcane Harvest Logistics in Brazil 

Kamal Lamsal, University of Iowa, 108 Pappajohn Business Bldg, 

S210, Iowa City, IA, 52242, United States of America, 

kamal-lamsal@uiowa.edu, Barrett W. Thomas, Philip Jones 

In the face of rising world sugar prices, the logistics of sugar mills in the Brazil, 

the world’s largest sugar producer, are of great interest. In this presentation, we 

present a rolling horizon approach to coordinating sugarcane harvest operations 

for Brazilian sugar mills. After each random event, we update an integer program 

to account for both new information and operations already underway. The 

solutions from our dynamic approach are compared to a perfect information 

bound. 


147 

4 - Customer Order Rationing in Shipment Consolidation 

Benhür Satir, Assistant Professor, Cankaya University, 

Yenimahalle, Ankara, 06810, Turkey, benhur@cankaya.edu.tr, 

James Bookbinder, Fatih Safa Erenay 

A delivery company faces two classes of demand for dispatching a single package 

type, viz. standard and express; the latter is more profitable. Shipments are made 

by capacitated vehicles. The optimal customer rationing policy for shipment 

consolidation is modeled and solved as a continuous-time Markov Decision 

Process. 

■ SD49 

SD49 


Public Transit: Network Modeling 

Sponsor: Transportation Science & Logistics 


Chair: Wei Fan, Associate Professor, The University of Texas at Tyler, 

Department of Civil Engineering, 3900 University Blvd, Tyler, TX, 

75799, United States of America, wfan@uttyler.edu 

1 - Spatial and Temporal Tradeoffs in Transit Stop Location 

Problems 

Sha Mamun, Graduate Student, University of Conneticut, 261 

Glenbrook Road, Unit 2037, Storrs, CT, 06269, United States of 

America, msm08014@engr.uconn.edu, Nicholas Lownes 

A transit network design problem is presented to optimize transit service 

efficiency by adjusting service frequency and transit stop locations. A bi-level 

mathematical programming model is constructed to allow tradeoffs between the 

spatial design (stop location) and the temporal operations (frequency setting). The 

stop locations are optimized by minimizing the cost components in the lower 

level, while the upper level optimized the transit line frequencies by maximizing 

network connectivity. 

2 - Trade-offs in Bus Frequency Allocation: System Level 

Optimization and Application to Large Scale Network 

Omer I. Verbas, Northwestern University, Evanston, IL, 60208, 

United States of America, ismailverbas2012@u.northwestern.edu, 

Hani S. Mahmassani 

We present a system-level model for frequency allocation to bus routes. The 

formulation solves for optimal frequencies that vary throughout the course of the 

day on individual routes and associated service patterns. Objectives include 

ridership maximization, waiting time minimization and cost minimization, subject 

to fleet and crowding considerations. Detailed demand elasticity levels are used in 

an application to the Chicago area network. 

3 - Equitable Redesign of the Existing Public Transit Route 

Networks: A Bi-level Optimization Approach 

Wei Fan, Associate Professor, The University of Texas at Tyler, 

Department of Civil Engineering, 3900 University Blvd, Tyler, TX, 

75799, United States of America, wfan@uttyler.edu 

A bi-level optimization model is formulated, which can explicitly account for the 

spatial equity, for solving the public transit route network redesign problem 

(PTNRP). A genetic algorithm based solution approach is developed to solving this 

PTNRP bi-level optimization model. Network experiments are conducted. 

Numerical results and related characteristics are also described in detail. 

4 - A Simulation-based Transit Assignment Model 

Mark Hickman, Associate Professor, University of Arizona, 1209 E. 

Second Street, Tucson, AZ, 85721-0072, United States of America, 

mhickman@email.arizona.edu, Alireza Khani 

In the context of a scheduled fixed-route transit service, we explore a more 

realistic passenger assignment using an integrated traffic and transit simulation 

tool to generate transit vehicle trajectories. Using these trajectories, we assign 

transit passengers to the network using either a trip-based shortest path or a tripbased 

hyperpath assignment. The simulation model and the associated transit 

assignment methods are illustrated on a case study from Sacramento, California.

SD50 

■ SD50 


Military Search and Surveillance 



Chair: Michael Hirsch, Raytheon Company, Orlando, FL, 

United States of America, mjh8787@ufl.edu 

1 - Hunting Drug Smugglers: An Optimal Search Problem in 

Continuous Time and Space 

Jesse Pietz, Naval Postgraduate School, 1411 Cunningham Road, 

Monterey, CA, United States of America, japietz@nps.edu, 

Johannes Royset 

We consider search for moving targets in continuous time and space, and 

formulate a mixed-integer nonlinear program whose solution yields an optimal 

search plan. We highlight features of this highly challenging problem that lead to 

specialized algorithms and discuss solutions for application to drug smuggling 

interdiction operations. 

2 - Optimal Search for a Random Moving Intruder 

David Casbeer, Air Force Research Lab., AFRL/RBCA, B146 R300, 

WPAFB, OH, 45433, United States of America, 

David.Casbeer@wpafb.af.mil, Meir Pachter, K. Krishnamoorthy, 

Phillip Chandler 

Using reachability arguments, the optimal control of an agent searching for an 

intruder on a graph is considered. The agent is only able to perceive partial and 

delayed information about the intruder. Sufficient conditions for guaranteed 

isolation of the intruder, before he reaches his goal, and the corresponding 

optimal agent control policy, are provided. 

3 - A Joint Surveillance and Patrol Problem for Law Enforcement 

Belleh Fontem, PhD Candidate, The University of Alabama, 300 

Alston Hall, 361 Stadium Drive, Tuscaloosa, AL, 35487, United 

States of America, bafontem@crimson.ua.edu, Sharif Melouk, 

Burcu Keskin 

We investigate the problem of using unmanned aerial systems (UASs) as visual 

aids to ground law enforcement agents while assigning dangerous incidents to 

these agents in order to maximize the cumulative harm averted to society. We 

introduce a very tight formulation of the resulting team orienteering problem and 

and draw insights from theoretical analyses as well as from numerical 

experiments. 

■ SD51 


Joint Session MAS/SPPSN:Defense Operations and 

Homeland Security II 

Sponsor: Military Applications & Public Programs, Service and 

Needs 


Chair: Anna Doro-on, Sr. Consultant/Project Engineer, Water, Tunnel & 


91101, United States of America, anna@annaforhomelandsecurity.com 

1 - Terror Queues and the Duration of Terror Plots 

Edward Kaplan, William N. and Marie A. Beach Professor of 

Management Sciences, Yale School of Management, 135 Prospect 

Street, New Haven, CT, 06520, United States of America, 

edward.kaplan@yale.edu 

Starting with a review of US terrorism-related indictments, lower and upper 

bounds for the initiation date of 30 distinct Jihadi plots were identified in addition 

to the date of arrest or an attempted/actual terror act. Estimated mean duration 

equals 9 months, while 95% of all plots are estimated to fall between 1 and 25 

months. These estimates suggest that in the United States, on average 

approximately three ongoing Jihadi terror plots have been active at any point in 

time since 9/11/2001. 

2 - Intelligent-IEDs Application and Risk Acceptability Optimization 

in a Successful Conduct of War 

Anna Doro-on, Sr. Consultant/Project Engineer, Water, Tunnel & 



anna@annaforhomelandsecurity.com 

The Improvised Explosive Devices (IEDs) were used by terrorists against United 

States and Allies’ military forces and responsible of deaths in Iraq and 

Afghanistan. This paper presents the application of intelligent-IEDs for the 


148 

effectiveness of military operations and defense against the enemy, to decrease 

military casualties, protect infrastructures, minimize risk of terrorism, and 

maximize risk acceptability. 

3 - Finding Fakes: Counterfeit Product Detection at United States 

Border Crossings 

Gregory DeYong, Assistant Professor, University of Michigan-Flint, 

2145 Riverfront Center, 303 E. Kearsley Street, Flint, MI, 48502, 

United States of America, gdeyong@umflint.edu 

The problem of counterfeit products continues to plague manufacturers. Using 

data from the U.S. Customs and Border Patrol, we analyze the effectiveness of 

border crossing points using data envelopment analysis and apply game theory to 

suggest interdiction strategies. 

4 - Optimal Placement of Multiple Types of Detectors under a Small 

Vessel Attack Threat to Port Security 

Xiaofeng Nie, Assistant Professor, Nanyang Technological 

University, 50 Nanyang Avenue, Singapore, 639798, Singapore, 

xiaofengnie@ntu.edu.sg, Xihong Yan 

Because of the strategic importance, ports are potential terrorist targets. We focus 

on the threat scenario where the terrorists would use small vessels as waterborne 

improvised explosive devices to attack maritime targets. We consider how to 

locate multiple types of detectors such that the expected damage cost is 

minimized. The model is formulated as a nonlinear binary integer program and 

some exact and heuristic algorithms are proposed. 

■ SD52 


Optimization in Smart Grid 



Chair: Andrew Liu, Assistant Professor, Purdue University, 315 N. Grant 

Street, West Lafayette, IN, 47907, United States of America, 

andrewliu@purdue.edu 

1 - Decision Making in the Operation of Smart Distribution Systems 

Mingguo Hong, Midwest ISO, 5407 Alvamar Place, Carmel, IN, 

46033, United States of America, mingguoh@hotmail.com 

The Smart Grid operation involves both the eletrical transmission and distribution 

systems where infrastructure and operational strategy changes should take place 

in parallel. This talk focuses on the operational decisions of the Smart Grid at the 

distribution system level to ensure greater grid reliability and economic efficiency. 

The talk also looks at a potential market design for the distribution system 

involving electrical utility and their customers. 

2 - Quantify Effects of Home Energy Management System under 

Dynamic Electricity Pricing 

Jingjie Xiao, PhD Candidate, Purdue University, 315 N. Grant 

Street, West Lafayette, IN, 47907, United States of America, 

jingjie.xiao@gmail.com, Andrew Liu, Joseph F. Pekny 

A key hurdle for implementing real-time pricing of electricity is the lack of 

consumers’ responses. Solutions to overcome the hurdle include energy 

management systems that can automatically optimize household appliance usage 

and PHEV charging via two-way communication with the utilities (or the grid). 

This talk will show how the optimization can be done and how such systems can 

be integrated with electricity markets’ operation under various uncertainties 

through approximate dynamic programming. 

3 - SMART-ISO: A Stochastic, Multiscale Model of the 

PJM Power Grid 

Warren Powell, Princeton University, 230 Sherrerd Hall, Princeton, 

NJ, 08544, United States of America, powell@princeton.edu, 

Boris Defourny, Hugo Simao 

We will describe the development of a stochastic, multiscale, dynamic model of 

the PJM power grid. SMART-ISO includes a full model of the PJM power grid, a 

robust day-ahead model for stochastic unit commitment, hour-ahead modeling 

for planning natural gas simulation, and economic dispatch solved at five minute 

increments for accurate modeling of ramp rates of natural gas units and variations 

in renewables. We report on the status of this multi-year development effort. 

4 - Dynamic Stability of Electricity Markets 

Victor Zavala, Argonne National Laboratory, Mathematics & 

Computer Science Division, Chicago, IL, 60439, 

United States of America, vzavala@mcs.anl.gov 

We present new insights on the dynamic stability of markets. We discuss how 

stability issues arise from incomplete gaming and physical constraints and propose 

strategies to mitigate these.

■ SD53 


Optimization Models for Demand-Response 



Chair: Miguel F. Anjos, Canada Research Chair in Discrete Nonlinear 

Optimization in Engineering, Ecole Polytechnique de Montreal, C.P. 

6079, succ. Centre-ville, Montreal, QC, H3C 3A7, Canada, 

miguel-f.anjos@polymtl.ca 

1 - Real-time Demand Response Model 

Antonio Conejo, Professor, Universidad Castilla-La Mancha, 

Campus Universitario, Ciudad Real, 13071, Spain, 

Antonio.Conejo@uclm.es, Luis Baringo, Juan Miguel Morales 

This presentation describes an optimization model to adjust the hourly load level 

of a given consumer in response to electricity prices. The objective of the model is 

to maximize the consumer utility. Price uncertainty is modeled through robust 

optimization techniques. The model materializes into a simple LP that can be 

integrated in the Energy Management System of a household or a small business. 

A simple bi-directional communication device enables the implementation of the 

proposed model. 

2 - Demand Response in the California ISO market 

Guillermo Bautista Alderete, Manager of Market Validation and 

Quality Analysis, California ISO, 250 Outcropping Way, 

Folsom, CA, 95630, United States of America, 

bautista.guillermo@gmail.com 

This presentation explains the initiatives and mechanisms in place to facilitate the 

participation of retail demand response programs in the California ISO wholesale 

market, including the implementation of a new demand response product known 

as the proxy demand resource. In conjunction with the ISO’s existing 

participating load product, the proxy demand resource product will add to the 

demand response capability available to market participants. 

3 - Optimization for Time-of-use Control in Electricity Demand 

Side Management 

David F. Rogers, Associate Professor of Operations and Business 

Analytics, University of Cincinnati, 2925 Campus Green Dr, 531 

Carl H. Lindner Hall, Cincinnati, OH, 45221-0130, 

United States of America, David.Rogers@UC.edu, George G. Polak 

Time-of-use pricing of electricity in demand side management requires 

determining the time periods of a day that should be grouped for identical pricing. 

Integer & constraint programming models to find clusters of contiguous periods 

potentially with a “wrap-around” feature for which time periods at both the 

beginning and end of the day may be in the same cluster are examined. A realworld 

data set consisting of kwh usage for 93 buildings was employed for testing 

the models. 

4 - A System Architecture for Autonomous Demand Side Load 

Management in Smart Buildings 

Miguel F. Anjos, Canada Research Chair in Discrete Nonlinear 

Optimization in Engineering, Ecole Polytechnique de Montreal, 

C.P. 6079, succ. Centre-ville, Montreal, QC, H3C 3A7, Canada, 

miguel-f.anjos@polymtl.ca, Giuseppe T Costanzo, Gilles Savard, 

Guchuan Zhu 

We present a new system architecture for demand-side load management. The 

system is composed of modules for admission control, load balancing, and 

demand/response management that operate using online operation control, 

optimal scheduling, and dynamic pricing respectively. It can integrate different 

energy sources and handle autonomous systems with heterogeneous dynamics in 

multiple time-scales. Simulation results confirm the viability and efficiency of the 

proposed framework. 


149 

ENRE Poster Session 

ENRE POSTER SESSION 

North Building Foyer 

Cluster: ENRE Poster 


Chair: Juan Pablo Vielma, University of Pittsburgh, 1043 

Benedum Hall, 3700 O’Hare Street, Pittsburgh, PA, 15261, 


1 - Competitive Markov Decision Process Model for 

Energy-water-climate Change Nexus 

Ivan Savedra, Research Assistant, University of Wisconsin- 

Milwaukee, 3200 North Cramer Street, Milwauke WI 53211, 

United States of America, saavedr9@uwm.edu,Vishnuteja Nanduri 

Drought-like conditions brought on by climate change are causing water shortages 

leading to power disruptions in many regions in the world. It is noted by 

the DOE that U.S. electricity industry uses more water for electricity production 

than the entire agricultural and horticultural industries combined. Energy-waterclimate 

change nexus is a critical issue that needs to be addressed. However, 

there are no existing quantitative models that examine this nexus. Our paper 

addresses this challenge. 

2 - Investigating through the Effects of Renewable Energy 

Technologies Exploitation on Economic Growth 

Seyed Hadi Nourbakhsh, University of Tehran, No 370, 

66 Alley, Dardash St.,Narmak, Tehran, Iran, 

hadi.noorbakhsh@gmail.com, Setareh Ariafar, Hessam Kazari 

In this research we tried to survey the effect of utilizing renewable energy technologies 

in electricity generation on economic growth by benefiting from a system 

dynamics approach. At first, we designed a causal model with considering 

main exogenous energy systems variables and economic variables. Then we executed 

proposed model for the case of Iran. 

3 - Modeling Investment Planning in Thermal Power Generation in 

India using Optimal Control Methods 

Sudhakar Achath, Professor, Amrita School of Business, 

Amrita Vishwa Vidyapeetham, Coimbatore, TN, 641112, India, 

s.m.achath@gmail.com 

Using linear quadratic control with learning, this study formulates an optimal 

plan for investments in plant capacity, R&D expenditure and imported technology 

for the Indian thermal power sector. The modelling system has twelve linear 

equations and the control problem formulated in state-space is solved as a 

dynamic programming problem using Lagrange multipliers and a backward solution 

method, following Chow (1975). The problem was initially written in BASIC 

but later resolved in open-source R. 

4 - Multi-objective Retrospective Optimization using 

Zigzag Search 

Honggang Wang, Rutgers University, 212 CoRE Building, 

Busch Campus, Rutgers, Piscataway, NJ, 08854, 

United States of America, honggang.w@rutgers.edu 

We propose a new retrospective optimization (RO) method with a gradient-based 

zigzag search approach for multi-objective simulation optimization (MOSO)problems. 

In the context of MOSO, a stochastic zigzag search algorithm is used in the 

RO framework. Numerical results of applying this new algorithm to MOSO problems 

will be provided. 

5 - Optimization for Annual Unloading Arrangement and Inventory 

Management in LNG Receiving Terminals 

Hidetaka Shinozaki, Tokyo Gas Co.,Ltd., 5-20, Kaigan 1-Chome, 

Minato-Ku, Tokyo, 1058527, Japan, shinozaki@tokyo-gas.co.jp, 

Shinnosuke Kimura, Hiroshi Kashio, Yohei Nakai, Mariko Ito, 

Shirou Morimitsu, Kazuaki Nagai 

We receive over 300 LNG vessels annually. The operators not only allocate the 

vessels to 3 terminals but also manage the inventory of 52 LNG tanks considering 

many constraints related to stocks, calorific values and so on. Besides, we have to 

reschedule the plan whenever some changes by shipping delay or demand fluctuations 

occur. We optimized annual unloading arrangement and inventory 

management in LNG receiving terminals. It enables to reduce the manpower and 

the operating cost. 

6 - Optimizing Batteries in Electricity Markets 

Eugene Zak, Viridity Energy, 2370 130 Ave., NE, #104, Bellevue, 

WA, United States of America, ezak@viridityenergy.com, 

Ricardo Rios-Zalapa 

Battery can operate as a generator, load, or storage. The problem is how to integrate 

the battery into the grid given its life cycle characteristics, functional flexibility, 

and electricity market conditions. We propose a model to optimize battery 

use for energy and regulation. The model takes advantage of a price difference 

between distinct time periods and alternative products and services. As a result 

the model delivers a charge-discharge strategy and its impact on the battery life.

ENRE POSTER SESSION 

7 - Payment Uncertainty Management in a Combined Market of 

Energy and Reserve 

Taísa Felix, Instituto Federal de Brasília, Quadra SHCGN 703 

Bloco L, Apartamento 103, Brasilia, DF, 70730-712, Brazil, 

taisa.felix@ifb.edu.br, Pablo Cuervo 

As a rule, power system markets operate considering bid cost minimization 

(BCM) in clearing processes. This procedure has inconsistency because the total 

BCM is different from the load payment. This work introduce a combined market 

model of energy and reserve operating under Payment Minimization (PM), 

transforming the problem in a bi-level linear model, which can be solved using 

GAMS®. Expected payment is evaluated using Monte Carlo simulation, providing 

results to evaluate the PM energy policy. 

8 - Transmission Expansion Planning using a Linearized 

AC Model 

Hui Zhang, Arizona State University, 5th floor, Room 519, 

551 E. Tyler Mall, Tempe, AZ, 85281, United States of America, 

dahui911@hotmail.com, Vijay Vittal, Gerald Heydt 

This paper presents a MILP approach for solving transmission expansion planning 

(TEP) problems using a linearized AC model. Compared to existing DCbased 

TEP models, the proposed model includes reactive power, off-nominal bus 

voltage magnitudes as well as network losses. A practical approach to include the 

N – 1 criterion is also discussed. The simulation results show that the proposed 

TEP model has better accuracy and is applicable to solve TEP problems for large 

power systems. 

■ SD54 


Feature Articles from Recent Issues of INFORMS 

Transactions on Education (ITE) and Discussion of 

the ITE Review & Publication Process 

Cluster: INFORMS Journals Cluster 


Chair: James Cochran, Bank of Ruston Endowed Research Professor, 

Louisiana Tech University, College of Business, Ruston, 71272318 2, 

United States of America, jcochran@latech.edu 

Co-Chair: Matthew Bailey, Associate Professor, Bucknell University, 

School of Management, Lewisburg, PA, 17837, 

United States of America, matt.bailey@bucknell.edu 

Co-Chair: Vijay Mehrotra, University of San Francisco, School of 

Management, 2130 Fulton Street, San Francisco, CA, 

United States of America, vmehrotra@usfca.edu 

Co-Chair: Armann Ingolfsson, Dr., University of Alberta, School of 

Business, University of Alberta, Edmonton, AB, T6G 2R6, Canada, 

armann.ingolfsson@business.ualberta.ca 

Co-Chair: Jill Hardin Wilson, Northwestern University IEMS, 2145 

Sheridan Rd, Room C210, Evanston, IL, 60208, 

United States of America, jill.wilson@northwestern.edu 

1 - The Effectiveness of using a Web-based Applet to Teach 

Concepts of Linear Programming 

Christine Kydd, Professor, University of Delaware, Department of 

Business Administration, Lerner College of Business, Newark, DE, 

19716, United States of America, chriskyd@udel.edu 

This paper presents a web-based java applet that was used by instructors and 

students to graphically illustrate/learn fundamental concepts of LP models. It 

describes the results of a study which compares student performance of those 

who did use the applet vs. those who did not. Results show that the students who 

used the applet to learn about LP concepts performed significantly better than 

those who did not. Implications for using such techniques in the classroom are 

discussed. 

2 - Active Learning Exercises for a Service Operations 

Management Course 

Mark Davis, Professor, Bentley University, 175 Forest St., 

Waltham, MA, 02452, United States of America, 

mdavis@bentley.edu, Ravi Behara 

In today’s hi-tech environment, students are familiar with the Internet where 

they actively participate in the generation and consumption of information. Thus, 

if business schools want to stimulate student interest, they have to actively 

engage them in the learning process. This presentation introduces exercises, some 

of which are part of the course structure, that incorporate active learning in the 

classroom, and which align well with topics that are in a service operations 

management course. 


150 

3 - Case Article: Flight Delays at RegionEx 

Susan Martonosi, Harvey Mudd College, Department of 

Mathematics, Claremont, CA, United States of America, 

martonosi@hmc.edu, Amr Farahat 

In this case about two fictitious airlines, RegionEx, a small regional airline, is a 

contracted regional carrier for MDA, a major US airline. RegionEx is at risk of 

losing its contract due to an apparently poor flight delay record, and the Flight 

Operations Manager must analyze the flight delay records and explain RegionEx’s 

seemingly poor performance. Undergraduate and MBA students use basic data 

analysis techniques to discover important paradoxes in the flight delay data. 

4 - KEY Electronics - Sourcing and Warehouse Analysis 

Tim Kraft, University of Virginia, Darden School of Business, 


kraftt@darden.virginia.edu, Yenho Chung, Feryal Erhun 

KEY Electronics is a consumer electronics retailer that generated $1.4 billion in 

total revenue in 2008. KEY has a limited retail presence in Mexico, which it 

would like to expand. As part of its growth strategy, KEY must (1) improve its 

current sourcing of products for Mexico and (2) revamp its existing warehouse 

operations. KEY introduces students to a practical inventory modeling scenario 

with real-life data, where solutions are not solely based on minimum cost but also 

qualitative factors. 

■ SD55 


Operations Research and†Management Science in 




Chair: Siddhartha SenGupta, Principal Scientist, Tata Consultancy 

Services Ltd, Gateway Park, 2nd flr, MIDC Road 13, MIDC, Andheri 

East, Mumbai, Ma, 400093, India, Siddhartha.SenGupta@tcs.com 

1 - Rationing Policy and Inventory Optimization Models for a 

Three-Stage Divergent Supply Chain 

Chandrasekharan Rajendran, Professor, IIT Madras, Chennai, 

600 036, India, craj@iitm.ac.in, Kurian John, Brijesh Paul 

This work considers a static three-stage divergent supply chain model consisting 

of retailers, two distributors and one manufacturer. A distributor serves only a 

subset of retailers for whom she/he is the sole supplier. The unsatisfied demands 

at all the installations are backordered. Assuming the base-stock policy, the 

objective is to determine the best installation inventory order policy parameters 

such as the base-stock levels and inventory rationing quantities, with the 

objective of minimizing the total supply chain cost (TSCC) consisting of holding 

costs and shortage costs at all installations in the supply chain over a finite 

planning horizon. An integer programming model to determine the optimal 

inventory control-policy parameters is proposed in this work. Computational 

complexity limits the applicability of the exact solution method to large problem 

sizes. Therefore, in the current study we also propose a hybrid heuristic technique 

based on Genetic Algorithm (GA) and Particle Swarm optimization (PSO) 

algorithm to determine the best base-stock levels. A heuristic rationing rule 

determines the shipment (rationing) quantity in the case of shortage at a higher 

installation, while meeting the customer demands. The performance of the hybrid 

heuristic against the stand-alone GA and PSO is ascertained by testing the 

algorithms under various supply chain settings. 

2 - A Rough Set-decision Tree Based Approach for Design of 

Optimal Process Tolerances 

J Maiti, Associate Professor, Indian Institute of Technology 

Kharagpur, Industrial Engineering and Management, Kharagpur, 

India, jmaiti@iem.iitkgp.ernet.in, Sandipan Karmakar 

Tolerance design provides a leeway to counteract the inherent manufacturing 

uncertainties with desired product performance. It is important to design 

tolerances on process elements based on product dimensions. In this study, the 

authors propose a novel approach using rough set theory-decision tree to find out 

the best tolerance values on process elements. A example of an assembly of sheet 

metals for automotive body assembly is discussed to show the effectiveness of the 

proposed method.

3 - Business Models and Processes for 3PL Providers 

Siddhartha SenGupta, Principal Scientist, Tata Consultancy 

Services Ltd, Gateway Park, 2nd flr, MIDC Road 13, MIDC, 

Andheri East, Mumbai, Ma, 400093, India, 

Siddhartha.SenGupta@tcs.com 

Economically Services are becoming increasingly more important and it would be 

useful to develop appropriate Business Models and Business Processes for service 

value chains. Here we try to “Learn from the smokestacks” - from the well 

developed Models and Processes of Manufacturing. In particular, we find that the 

IHIP characteristics of Services make them similar to manufacturing models 

where the decoupling point is away from the customer processes, e.g. in Make- 

To-Order. We adapt and extend the relevant methods of these models to 3PL 

services, especially to shipping. 

4 - Food Security for Emerging Economies – An OR Perspective 

G Raghuram, Professor, Indian Institute of Management 

Ahmedabad, Wing 15, IIMA Campus, IIMA Campus, Ahmedabad, 

380015, India, graghu@iimahd.ernet.in, Ankan Pal 

Food security is provided by the Food Corporation of India which procured 50 

million tons of food grains in 2011-12 with a spend of over Rs 80,000 crores. 

Towards achieving this, the primary activity is managing logistics, including the 

type of storage (bulk/bagged, covered/open, own/outsourced), allocation from 

procurement to storage to demand points, mode choice, and quality 

management. This paper provides a perspective on OR methods which can help 

achieve food security while optimizing costs. 

■ SD56 


Entrepreneurship and Innovation II 

Sponsor: Technology Management 


Chair: Sinan Erzurumlu, Assistant Professor, Babson College, 

231 Forest Park, Babson Park, MA, 02457, United States of America, 

serzurumlu@babson.edu 

1 - Strategies for Profiting from Technologies with Weak 

Patent Rights 

Leonardo Santiago, Federal University of Minas Gerais, Escola de 

Engenharia da UFMG, Av. Antonio Carlos, 6627. Pampulha, 

Belo Horizonte, MG, 31270-901, Brazil, lsantiago@ufmg.br, 

Henrique Rocha 

We investigate a set of strategies that companies adopt in order to profit from 

innovations. The strategies depend not only on inherent features of the 

innovation at stake but also on the incentives potential competitors have to enter 

in the target market. Our main focus is on technologies characterized by weak 

patent rights. 

2 - Management of Innovation in the Product and 

Technology Markets 

Sinan Erzurumlu, Assistant Professor, Babson College, 231 Forest 

Park, Babson Park, MA, 02457, United States of America, 

serzurumlu@babson.edu, Karthik Ramachandran 

Firms in many technology intensive industries increasingly license innovations to 

potential competitors besides using them in their own products. Licensing not 

only creates an additional profit stream, but could also enhance the value of the 

innovator’s own product. However, it also makes the licensee a potent competitor 

in the product market. In this paper, we develop a model to understand if, when 

and how a firm should share its innovation with a rival. 

3 - Value of Reverse Factoring in Multi-stage Supply Chains 

Fehmi Tanrisever, Eindhoven University of Technology, Den 

Dolech 2, Eindhoven, 5612, Netherlands, f.tanrisever@tue.nl, 

Hande Cetinay, Matthew Reindorp 

We develop a mathematical model for integration, analysis, and optimization of 

operational and financial processes within a supply chain. Specifically, we 

consider commercial transactions of a large corporate customer with a small- or 

medium-sized supplier. We show how application of reverse factoring – an 

increasingly popular product within the broad field of supply chain finance – 

influences the operational and financial decisions of these firms. 


151 

4 - A Dynamic Model of Knowledge Creation from Exploration and 

Exploitation in the High-tech Venture 

Jennifer Bailey, Georgia Institute of Technology, 800 W. Peachtree 

Road, NW, Atlanta, GA, United States of America, 

Jennifer.Bailey@mgt.gatech.edu, Cheryl Gaimon 

We develop a dynamic model of exploration and exploitation for a high-tech 

venture. The uncertain nature of these learning activities is captured in terms of 

the mean and variance of the innovation outcomes. We consider the impact of 

the venture’s absorptive capacity as well as the lagged effect of exploration on 

innovation outcomes. We determine when the typical explore-exploit versus an 

atypical exploit-explore sequential strategy is optimal. 

■ SD57 


Applications of Queueing Models 


SD57 

Chair: Jun Luo, Hong Kong University of Science and Technology, Rm 

5567, Academic Building, Clear Water Bay, Hong Kong, Hong Kong - 

PRC, jluolawren@ust.hk 

1 - Modeling Sequential Zone Picking Systems 

Jelmer van der Gaast, Erasmus University, Burgemeester Oudlaan 

50, Rotterdam, Netherlands, jgaast@rsm.nl, René De Koster, 

Ivo Adan, Jacques Resing 

We develop an analytical model for analyzing sequential zone picking systems. 

The system is approximated by a multi-class product-form closed queueing 

network with jump-over blocking. An iterative algorithm using mean value 

analysis (MVA) is used to evaluate the performance statistics of the order picking 

system. The accuracy of the algorithm is compared with this results of a discrete 

event simulation. 

2 - A Simple, Practical Prioritization Scheme to Minimize Cycle 

Time and Maximize Throughput 

Shuping Zhang, University of Tennessee, Knoxville, 

201 Stokely Management Center, Knoxville, TN, 37996, 

United States of America, szhang12@utk.edu 

The MRO process is used to recondition equipment in the railroad, off-shore 

drilling, aircraft, and shipping industries. We model the back-shop as a multi-class 

queueing network with a ConWIP execution system and introduce a new priority 

scheme to maximize system performance. Simulation results show that this 

priority scheme increases system performance by lowering average cycle times 

without dramatically impacting total throughput. 

3 - Optimal Staffing for Ticket Queues 

Li Xiao, Business School, National University of Singapore, Biz 2 

Building, Basement B2-03, NUS, singapore, Singapore, 

a0075548@nus.edu.sg, Susan Xu, Hanqin Zhang 

This talk is concerned with ticket queues systems, where each arriving customer 

gets in the system with a numbered ticket and observes the number of working 

servers. The arriving customers abandon the system with some probability 

determined by the number of working servers. The system cost includes customer 

abandonment and server operations costs. Based on the ticket number 

information, we address the question how to staff such that the expected longrun 

average cost is minimized. 

4 - Staffing and Control of an Instant-messaging Based Customer 

Service Center 

Jun Luo, Hong Kong University of Science and Technology, 

Rm 5567, Academic Building, Clear Water Bay, Hong Kong, 

Hong Kong-PRC, jluolawren@ust.hk, Jiheng Zhang 

A new kind of customer service centers, in which agents communicate with 

customers via instant-messaging (IM) on the Internet, is prevalent nowadays. A 

distinctive feature of IM-based service centers is that one agent can serve multiple 

customers in parallel. We model it as a server pool consisting of many limited 

processor sharing servers. The underlying stochastic processes can be 

characterized by a fluid approximation, which provides insights into the optimal 

staffing and control policies.

SD58 

■ SD58 


Asymptotics and Optimization of Queueing Systems 



Chair: Hengqing Ye, Associate Professor, Hong Kong Polytechnic 

University, Kowloon, Hong Kong - PRC, lgtyehq@inet.polyu.edu.hk 

1 - Excursion-based Universal Approximations for the Erlang-A 

Queue in Steady-state 

Itai Gurvich, Northwestern University-Kellogg School of 


i-gurvich@kellogg.northwestern.edu, Junfei Huang, 

Avishai Mandelbaum 

We re-visit the Erlang-A queue and propose a universal diffusion approximation 

(rather than limit) that applies simultaneously to all of the existing many-server 

heavy-traffic regimes: QED, ED, QD and NDS. The approximation yields accurate 

estimates for steady-state metrics. In our proofs, we do not use the steady-state 

distribution of the Erlang-A queue. Rather, we relate the excursions of the 

underlying Birth-and-Death process to properly defined excursions of the 

universal diffusion. 

2 - Optimal Queue-size Scaling in Switched Networks 

Neil Walton, University of Amsterdam, Science Park 904, 

Amsterdam, 1098 XH, Netherlands, n.s.walton@uva.nl, 

Yuan Zhong, Devavrat Shah 

We consider a switched network with constraints on which queues may be served 

simultaneously; such networks have been used to effectively model input-queued 

switches. The scheduling policy for such a network specifies which queues to 

serve at any point in time, based on the current state or past history of the 

system. We provide a new class of online scheduling policies that achieve optimal 

average queue-size scaling for a class of switched networks including inputqueued 

switches. 

3 - Diffusion Limit of Fair Resource Control—Stationarity and 

Interchange of Limits 

Hengqing Ye, Associate Professor, Hong Kong Polytechnic 

University, Kowloon, Hong Kong-PRC, 

lgtyehq@inet.polyu.edu.hk, David Yao 

Consider a network in which each job requires the concurrent occupancy of a 

subset of servers, and the service capacity is shared among job classes following 

the proportional fair allocation. We demonstrate that the usual traffic condition of 

the diffusion limit is necessary and sufficient for both the diffusion limit and 

prelimit networks to have stationary distributions, and justify the diffusion 

approximation of the stationary distribution and p-th moment under a bounded 

workload condition. 

4 - Skill Based FCFS Service Systems 

Gideon Weiss, University of Haifa, Mount Carmel 31905, Haifa, 

Israel, gweiss@stat.haifa.ac.il, Ivo Adan 

We consider systems with several types of customers and several types of servers, 

where compatibility of servers to customers is given by a bipartite graph. This is 

motivated by call centers with skill based routing, and by organ transplant 

applications. A common approach is provided by infinite first come first served 

matching of two infinite multi-type random sequences. We obtain explicit results 

for this model, and approximations for many server queues with abandonments. 

■ SD59 


Stochastic Network Algorithms 



Chair: Devavrat Shah, Massachusetts Institute of Technology, 32-D670, 

77 Massachusetts Ave, Cambridge, MA, 02139, 

United States of America, devavrat@mit.edu 

Chair: Yuan Zhong, Massachusetts Institute of Technology, 

235 Albany Street, Cambridge, MA, 02139, United States of America, 

zhyu4118@mit.edu 

1 - Dynamic Scheduling in Queueing Networks with Heavy-tailed 

Traffic: Insights through Fluid Models 

Mihalis Markakis, Massachusetts Institute of Technology, 

77 Massachusetts Avenue, 32D-666, Cambridge, MA, 02139, 

United States of America, mihalis@mit.edu, John N. Tsitsiklis, 

Eytan Modiano 

In this talk we present the first steps towards a systematic methodology for 

evaluating the performance of dynamic scheduling policies for switched queueing 


152 

networks under heavy-tailed traffic. We establish a connection between delay 

instability of queues and the evolution of the fluid model from certain initial 

conditions. This enables us to identify the delay unstable queues by simulating 

the fluid model. Moreover, it allows us to reach a surprising conclusion for special 

networks of interest. 

2 - A Note on Queue-size Scaling for Input-queued Switches 

Yuan Zhong, Massachusetts Institute of Technology, 235 Albany 

Street, Cambridge, MA, 02139, United States of America, 

zhyu4118@mit.edu, Devavrat Shah, John N. Tsitsiklis 

We consider queue-size scaling in an input-queued switch. In the main result of 

the paper, we provide a new class of scheduling policies under which the longrun 

average total queue size scales as power 2.5 of the port number n, in a certain 

limiting regime of interest. This is a substantial improvement upon prior works, 

where the same quantity scales as power 3 of n. Our result also advances toward 

a conjecture about optimal queue-size scaling in input-queued switches. 

3 - Smoothed Online Convex Optimization 

Adam Wierman, Assistant Professor, Caltech, 1200 E. California 

Blvd., Pasadena, CA, 91125, United States of America, 

adamw@caltech.edu, Lachlan Andrew, Minghong Lin, 

Alan Roytman 

Out work on sustainable data center design during the last few years has 

repeatedly led us to a class of problems we term “smoothed online convex 

optimization (SOCO)” problems. SOCO is a variant of the class of “online convex 

optimization (OCO)” problems, and is strongly related to the class of “metrical 

task systems”, each of which have been studied largely independently. In this talk 

I will discuss some recent progress toward unifying the results and algorithms 

from these communities. 

4 - The UFOS Algorithm for Network Resource Allocation 

Yang Yang, Graduate School of Business, Stanford University, 

W146A, Knight Management Center, 655 Knight Way, 

Stanford, CA, 94305, United States of America, 

yang.yang1025@stanford.edu, J. Michael Harrison, 

Chinmoy Mandayam 

The UFOS algorithm (utilization first, output second) puts primary emphasis on 

greedy maximization of total resource utilization, and secondary emphasis on 

greedy maximization of total outflow. Focusing on connection-level models of 

internet congestion control, we compare the delay-related performance of UFOS 

with that of proportionally fair (PF) resource allocation. In our simulation studies, 

UFOS reduces overall average delay by about 30% relative to PF. 

■ SD60 


Air Transportation and High-Speed Rail: Competition 

and Cooperation 



Chair: Regina Clewlow, PhD Candidate, Massachusetts Institute of 

Technology, 77 Massachusetts Avenue, Cambridge, MA, 02139, 

United States of America, rclewlow@alum.mit.edu 

1 - The Impact of Climate Policy on U.S. Aviation and 

High-speed Rail 

Regina Clewlow, PhD Candidate, Massachusetts Institute of 

Technology, 77 Massachusetts Avenue, Cambridge, MA, 02139, 

United States of America, rclewlow@alum.mit.edu, 

Joseph Sussman 

In this study, we examine the energy and CO2 emission impacts of aviation and 

potential high-speed rail in major short-haul intercity corridors in the United 

States. We couple a simulation model of high-speed rail and aviation demand 

with a general equilibrium model that examines the impact of climate policy on 

the U.S. on a regional scale. We will present the modeling methodology, as well as 

the results of economic and emissions analyses for several policy scenarios. 

2 - Long-term, System-wide Impact Analysis of High-speed Rail in 

the Midwest Corridor 

Jeffrey Peters, Purdue University-Nextrans, 3000 Kent Ave., 


peters83@purdue.edu, En-Pei Han, Daniel DeLaurentis, 

Srinivas Peeta 

This study describes a methodology for forecasting system-wide ridership and the 

corresponding user and community impacts of a high-speed rail mode under the 

existing transportation infrastructure. Experiments with various social, economic, 

technological, and political trends over the next few decades are conducted using 

the proposed Chicago-Hub network to analyze the potential viability in the longterm. 

Both the methodology and the impacts of the proposed system will be 

discussed.

3 - HSR: Paying Our Way Forward 

Nicolas Norboge, Assistant Transportation Researcher, Texas 

Transportation Institute, 3135 TAMU, Room 334, College Station, 

TX, 77840, United States of America, n-norboge@ttimail.tamu.edu 

With renewed interest in intercity passenger rail brings the emergence of fresh 

efforts to develop comprehensive system plans. However, many plans fail to 

explain clearly how such capital-intensive projects will be paid for. This research 

provides a list of viable funding and financing strategies transportation 

professionals and policymakers to consider for paying for new HSR networks. 

■ SD61 


AAS Dissertation Prize / Anna Valicek Medal 



Chair: Laurie Garrow, Associate Professor, Georgia Institute of 

Technology, School of Civil Engineering, 790 Atlantic Drive, Atlanta, 

GA, 30332, United States of America, laurie.garrow@ce.gatech.edu 

1 - Building Reliable Air-travel Infrastructure using Empirical Data 

and Stochastic Models 

Mazhar Arikan, University of Kansas, School of Business, 

Lawrence, KS, United States of America, mazhar@ku.edu, 

Milind Sohoni, Vinayak Deshpande 

Flight delays have been a growing issue and they have reached an all time high in 

recent years. We develop stochastic models, using empirical data, to analyze the 

propagation of delays through air-transportation networks. We then analyze 

robustness measures for airline networks. Our analysis enables us to make policy 

recommendations regarding managing bottleneck resources in the air-travel 

infrastructure, which if addressed, could lead to a significant improvement in airtravel 

reliability. 

2 - The Recoverable Robust Tail Assignment Problem 

Stephen Maher, University of New South Wales, School of 

Mathematics and Statistics, Sydney, 2052, Australia, 

stephen.maher@unsw.edu.au, Gary Froyland, Cheng-Lung Wu 

Disruptions affect airline operations, resulting in increased operational costs. 

Robust planning attempts to reduce the impact of these disruptions. We develop a 

less conservative robust plan using the concept of recoverable robustness for the 

tail assignment problem. Simultaneously solving the tail assignment planning and 

recovery problems to integrate recovery decisions into the original plan. Resulting 

in a tail assignment that is recoverable with a minimal number of operational 

changes. 

3 - Methods for Improving Robustness and Recovery in 

Aviation Planning 

Marcial Lapp, US Airways, Phoenix AZ, United States of America, 

mlapp@gmail.com 

Flight disruptions can have major effects on scheduled aircraft maintenance 

events. Once aircraft rotations are disrupted, plans must be updated to comply 

with regulations for aircraft maintenance. We address this issue by providing preand 

post-disruption strategies to stay compliant with regulations through our 

maintenance planning and recovery framework. We present results that illustrate 

the effectiveness of our optimization strategy and its impact on an airline’s shortterm 

maintenance planning process. 

■ SD62 

62- Borein A- Hyatt 

Combinatorial Markets 

Cluster: Auctions 


Chair: Ben Lubin, Professor, Boston University, 595 Commonwealth 

Avenue, Boston, MA, 02215, United States of America, blubin@bu.edu 

1 - Algorithms for Finding Bayes-Nash Equilibria in 

Core-selecting Auctions 

Mark Schneider, University of Connecticut, Uconn School of 

Business, 81 Cheney Drive, Storrs, CT, United States of America, 

mschneider@business.uconn.edu, Robert Day 

Determining Bayes-Nash equilibria in combinatorial auctions is quite difficult, and 

to date, the analytical approach to finding an exact equilibrium has been 

successful only in very limited settings. In this work, we explore the alternative 

possibility of approximating equilibrium behavior using numerical methods, with 

a focus on the class of core-selecting auctions. 


153 

2 - New Payment Rules in Core-selecting Auctions 

Marissa Beck, Stanford University, 579 Serra Mall, Stanford, CA, 

94305, United States of America, mrbeck@stanford.edu, 

Robert Day 

The set of core-selecting auctions is large and previous literature has focused 

mostly on minimum-revenue core-selecting auctions. We explore alternative 

“core” pricing rules based on linear item prices and LP duality, but enforcing 

stricter competitive equilibrium properties. A primal-dual cutting-planes approach 

is used to strike a balance between the (sometimes infeasible) concept of item 

prices and the (always obtainable but less intuitive) concept of personalized 

package prices in the core. 

3 - Coordination in Combinatorial Auctions with Unknown Bundles 

Zhen Hao, Technical University Munich, Boltzmannstr 3, 

Garching, Germany, hao@in.tum.de, Martin Bichler 

One of the main strategic challenges in combinatorial auctions is the coordination 

problem of bidders when they do not know the bundles, which are of interest to 

other bidders. We propose a combinatorial auction format, in which we leverage 

information on losing coalitions, which is collected by the auctioneer. We analyze 

how bidders respond to different levels of information feedback and overall 

efficiency compared to the combinatorial clock auction. 

4 - Payment Rules through Discriminant-based Classifers 

Ben Lubin, Professor, Boston University, 595 Commonwealth 

Avenue, Boston, MA, 02215, United States of America, 

blubin@bu.edu, Felix Fischer, Pichayut Jirapinyo, John Lai, 

Paul Deutting, David Parkes 

In mechanism design one typically seeks socially optimal mechanisms constrained 

by incentive compatibility. By replacing this constraint with the goal of 

minimizing expected ex post regret, we are able to adapt statistical machine 

learning techniques to the design of payment rules in settings with multidimensional 

types. Specifically, we train a support vector machine with a special 

discriminant function, such that it implicitly establishes a payment rule with 

desirable properties. 

■ SD63 

SD63 


Joint Session Behavioral Oper./ MSOM: Empirical 

Studies of Behavioral Factors 

Sponsor: Behavioral Operations & Manufacturing & Service Oper 

Mgmt 


Chair: Bradley Staats, Assistant Professor, University of North Carolina- 

Chapel Hill, McColl Building, CB 3490, Chapel Hill, NC, 27599, 

United States of America, Bradley_Staats@kenan-flagler.unc.edu 

1 - Learning from My Success and from Others’ Failure 

Bradley Staats, Assistant Professor, University of North Carolina- 

Chapel Hill, McColl Building, CB 3490, Chapel Hill, NC, 27599, 

United States of America, Bradley_Staats@kenan-flagler.unc.edu, 

Diwas KC, Francesca Gino 

Drawing on attribution theory, we investigate how individuals learn from failure 

and success. We use 10 years of data from 71 surgeons using a new cardiac 

surgery technology. We find that individuals learn more from their own successes 

than from their own failures, while they learn more from the failures of others 

than they do from others’ successes. We also find that individuals’ prior successes 

and others’ failures help individuals overcome their inability to learn from their 

own failures. 

2 - A Hard Take on a Soft Asset? A Test of an Operational Approach 

to Human Resource Management 



Valery Yakubovich 

We analyze empirically agent behavior at a virtual call center in which work is 

allocated according to performance-based hierarchy. We use a controlled 

experiment to disentangle the effect of this system on agent performance. 

3 - Waiting Patiently: Abandonment from Emergency 

Department Queues 

Robert Batt, The Wharton School, University of Pennsylvania, 

3730 Walnut St., Huntsman Hall, Suite 500, Philadelphia, PA, 

19104, United States of America, batt@wharton.upenn.edu, 

Marcelo Olivares, Christian Terwiesch 

We present an empirical study of patient abandonment behavior from a hospital 

emergency department queue. We show that patients respond differently to 

queue length and expected wait time despite the fact that these two measures are 

proportional to each other. We also consider the impact of providing wait time 

estimates to patients and under what conditions this can lead to reduced social 

welfare.

SD66 

4 - Innovation and Learning in the Use of Components 

Kamalini Ramdas, Professor, London Business School, Regent’s 

Park, London, NW1 4SA, United Kingdom, kramdas@london.edu, 

Khaled Saleh, Steven Stern, Haiyan Liu 

We examine learning in the context of hip replacement surgery using a dataset 

from a major US hospital. We find that the learning associated with individual 

surgeons’ experience is driven by their experience at the micro level of specific 

device variants, with implications for how surgery is managed and taught, and for 

innovation in medical devices. 

■ SD66 


Mining Time Series Data 



Chair: Onur Seref, Assistant Professor, Virginia Tech, 1007 Pamplin 

Hall, Blacksburg, VA, 24061, United States of America, seref@vt.edu 

1 - Causal Inference for Time-series Analysis: Simultaneous 

Denoising and Feature Selection 

Mohammad H. Poursaeidi, University of Houston, E 209 

Engineering Building, Houston, TX, United States of America, 

mhpoursaeidi@uh.edu, Erhun Kundakcioglu 

We present a new hyperplane based approach that performs regression and 

feature selection while simultaneously removing noise from time-series data. 

Mixed integer nonlinear programming formulations are developed for linear 

regression. These formulations are also extended via kernel trick to perform 

nonlinear regression. The problem is proven to be NP-hard and a heuristic 

algorithm is developed for large data sets. 

2 - Supervised Time Series Pattern Discovery through 

Local Importance 

Mustafa Baydogan, Graduate Research Assistant, Arizona State 

University, 699 S. Mill Avenue, Tempe, AZ, 85281, United States of 

America, mbaydoga@asu.edu, George Runger, Eugene Tuv 

In time series classification, nearest neighbor (NN) classifiers lack the aspect of 

interpretability since they are based on the similarity of the whole time series 

although temporal relations within the time series are important. We present an 

exploratory approach that finds the regions of the time series that have 

potentially representative patterns used for classification based on a local 

importance measure. Our algorithm provides comparable and interpretable 

results than competitive methods. 

3 - Online Piecewise Linear Segmentation and Temporal Pattern 

Extraction of Time Series Data 

Art Chaovalitwongse, University of Washington, 

3900 Stevens Way, Seattle, WA, United States of America, 

artchao@u.washington.edu, Shouyi Wang 

Piecewise linear segmentation (PLS) is popular to monitor and summarize time 

series patterns online. This talk presents a robust and efficient online 

segmentation algorithm that can be easily applied to various time series using a 

data-driven decomposition strategy and a closed-form online updating formula. A 

set of skeleton-point-based features are proposed to characterize time series 

temporal patterns. Two real-world applications demonstrate the effectiveness of 

the proposed approaches. 

4 - Information-theoretic Feature Selection with Discrete 

k-Median Clustering 

Onur Seref, Assistant Professor, Virginia Tech, 1007 Pamplin Hall, 

Blacksburg, VA, 24061, United States of America, seref@vt.edu, 

Elan Borenstein, Art Chaovalitwongse, Ya-Ju Fan 

We propose a novel information-theoretic feature selection method using discrete 

k-median (DKM) clustering, where the center for each cluster is represented as a 

set of samples. Normalized mutual information (NMI) scores between the output 

of DKM and the clusters formed by the projection of center samples are used for 

feature selection. Average of NMI scores are used for determining the number of 

clusters. We present results on simluated data and real-life neural time series data. 


154 

■ SD67 


Joint Session QSR/ENRE: Variability and Uncertainty 

in Renewable Power Systems 

Sponsor: Quality, Statistics and Reliability & Energy, Natural Res & 

the Environment/Forestry 


Chair: Eunshin Byon, Assistant Professor, University of Michigan, 

1205 Beal Ave., Ann Arbor, MI, 48109-2117, United States of America, 

ebyon@umich.edu 

1 - On-site Renewable Generation: Portfolio Selection and 

Contracts under Uncertainty 

Diego Klabjan, Associate Professor, Northwestern University, 

2145 Sheridan Road, IEMS, Evanston, IL, 60208, United States of 

America, d-klabjan@northwestern.edu, Sangho Shim, Jorge 

Arinez, Stephan Biller 

Investment decisions about on-site renewable generation for industrial sites are 

driven by net present value which inhibits many uncertainties. We present an 

optimization model for portfolio selection and an options contract. The 

uncertainties are driven by prices of renewable energy credits, on-site usage and 

output of sources, and future grid rates. 

2 - Grid Integration and R&D policy 

Erin Baker, University of Massachusetts, Amherst, MA, 

United States of America, edbaker@ecs.umass.edu, 

Noubara Djimadoumbaye 

In previous work we have analyzed the optimal energy technology R&D portfolio 

in the face of climate change, considering solar PV, CCS, and nuclear. Here we 

consider how the results are impacted by two very different assumptions 

regarding the integration of solar PV onto the grid. We apply a mutli-model, 

decision making under uncertainty framework. 

3 - A GIS, Optimization and Simulation Framework for Optimal PV 

Size and Location in Campus Environments 

Amirreza M. Khaleghi, The University of Arizona, 1127 E. James 

E. Rogers Way P.O. Box 21, Tucson, AZ, 85721-0020, United States 

of America, amirreza@email.arizona.edu, Ye Zhang, Ferenc 

Szidarovszky, Guzin Bayraksan, Young-Jun Son, Sadik Kucuksari, 

Maryam Hamidi 

A framework integrating GIS, optimization, and simulation modules is proposed 

to obtain the optimal placement and size for photovoltaic (PV) panels in a campus 

area. The GIS module finds the suitable rooftops and their panel capacity. The 

optimization module is then used to maximize the long term profit of PV 

installations. Voltage profile of the electricity distribution network is investigated 

in the simulation module. The proposed framework has been successfully 

demonstrated for a real case. 

4 - Assessing Extreme Loads on Wind Turbines by using 

Aerodynamic Simulation 

Eunshin Byon, Assistant Professor, University of Michigan, 1205 

Beal Ave., Ann Arbor, MI, 48109-2117, United States of America, 

ebyon@umich.edu, Youngjun Choe 

We develop a new methodology to predict an extreme load on a wind turbine by 

employing high-fidelity simulation. Due to a small probability of extreme events, 

crude Monte Carlo simulation requires extensive runs, which is computationally 

prohibitive. We devise an efficient sampling method to guide the simulation to 

generate the extreme events more frequently with minimal efforts. The proposed 

method is applied to an aerodynamic simulation.

■ SD68 


Finance: Financial Engineering 


Chair: Xiangwei Wan, Assistant Professor, Shanghai Jiao Tong 

University, 535 Hua Fa Zhen Road, Shanghai, China, 

xwwan@sjtu.edu.cn 

1 - A Neuro-wavelet Method for the Forecasting of Financial 

Time Series 

Luis Ortega, PhD Candidate, Financial Engineering, Stevens 

Institute of Technology, Castle Point on Hudson, Hoboken, NJ, 

07030, United States of America, lortega@stevens.edu 

We propose a wavelet neural network model (neuro-wavelet) for the short-term 

forecast of stock returns from high-frequency financial data. The proposed hybrid 

model combines the inherent capability of wavelets and artificial neural networks 

to capture nonstationary and nonlinear attributes embedded in financial time 

series. Reasonable forecasting accuracy for one, three and five step-ahead 

horizons was achieved by the proposed Jordan neuro-wavelet model when 

compared against five other models. 

2 - Analysis of Non-linear Behavior - A Sensitivity-based Approach 

Xiangwei Wan, Assistant Professor, Shanghai Jiao Tong University, 

535 Hua Fa Zhen Road, Shanghai, China, xwwan@sjtu.edu.cn 

The standard dynamic programming fails to work in behavioral analysis due to 

the distortion in performance probability. In this paper, we apply a sensitivitybased 

approach to solve the portfolio management problem in an environment 

with probability distortion and obtain a closed-form optimal policy. We expect 

that this approach is generally applicable to optimization in other non-linear 

behavioral analysis. 

3 - Game Russian Option with the Finite Maturity 

Atsuo Suzuki, Meijo University, Nijigaoka 4-3-3, Kani, Japan, 

atsuo@urban.meijo-u.ac.jp, Katsushige Sawaki 

We consider Game Russian options with the finite maturity. Game Russian option 

is a contract that the seller and the buyer have the rights to cancel and to exercise 

it at any time, respectively. We discuss the pricing model of Game Russian options 

when the stock pays dividends continuously. We show that the pricing model can 

be formulated as a coupled optimal stopping problem which is analyzed as 

Dynkin game. 

4 - Estimation of Stochastic Volatility Models for Commodities with 

Few Options 

Gonzalo Cortazar, Pontificia Universidad Catolica de Chile, 

Vicuna Mackenna 4860, Santiago, Chile, gcortaza@ing.puc.cl, 

Federico Alonso 

Stochastic volatility models are required for commodity prices. To obtain good 

parameter estimates, prices for non-linear derivatives with different maturities are 

necessary, but some commodities trade only futures and short-term options. We 

propose a three-factor multicommodity model with stochastic volatility and show 

how to use long-term options written on one commodity to help estimate the 

dynamics of another. The model is implemented for Brent and WTI commodities 

with excellent results. 

■ SD69 


Optimization and Financial Engineering 



Chair: Martin Haugh, Columbia, martin.b.haugh@gmail.com 

1 - Hope, Fear and Aspirations 

Xuedong He, Assistant Professor, Columbia University, 316 S. W. 

Mudd Building, 500 W. 120th Street, New York, NY, 10027, United 

States of America, xh2140@columbia.edu 

We propose a rank-dependent portfolio choice model in continuous time that 

captures the role in decision making of three emotions: hope, fear and 

aspirations. We solve the portfolio choice problem both thoroughly and 

analytically, and these solutions motivate us to introduce a fear index, a hope 

index and a lottery-likeness index to quantify the impacts of three emotions, 

respectively, on investment behavior. 


155 

2 - Approximate Methods for Dynamic Portfolio Choice with 

Learning 

Poomyos Wimonkittiwat, University of California-Berkeley, 

Berkeley, CA, United States of America, poomyos@berkeley.edu, 

Andrew Lim 

We develop approximate methods for dynamic portfolio choice problems with 

hidden states, where the transition probabilities may also need to be learned. We 

derive semi-explicit expressions for the optimal policy which can be implemented 

using tools from MCMC. 

3 - Martingale Dual Methods for Zero-Sum Games 

Chun Wang, Columbia University, 500 West 120th Street, 

Mudd 323, New York, NY, 10027, United States of America, 

cw2519@columbia.edu, Martin Haugh 

Martingale duality based on information relaxations has recently been developed 

for zero-sum optimal stopping games. In this talk, we consider generalizing this 

duality for more general zero-sum dynamic games. 

4 - Portfolio Selection with Multiple Spectral Risk Constraints 

Carlos Abad, Columbia University IEOR, 500 W 120th St., New 

York, NY, 10027, United States of America, ca2446@columbia.edu, 

Garud Iyengar 

We propose an iterative algorithm to efficiently solve the portfolio selection 

problem with multiple spectral risk constraints (CVaR is a special case of the 

spectral risk function). In each step, the algorithm solves a very simple separable 

convex quadratic program. We report numerical results that show that our 

proposed algorithm is very efficient, and is at least two orders of magnitude faster 

than the state of the art general purpose solver for all practical instances. 

■ SD70 

SD70 


Social Media and Social Broadcasting Network 



Chair: Huaxia Rui, University of Texas at Austin, Austin, TX, 

United States of America, huaxia@utexas.edu 

1 - Modeling Multicategory Online Video Demand in Social Media 

Qian Tang, University of Texas at Austin, Austin, TX, 78703, 

United States of America, qian.tang@phd.mccombs.utexas.com, 

Liangfei Qiu, Andrew Whinston 

The popularity of social media has greatly increased the demand for online 

videos. The demand for online videos may be influencd by online WOM, social 

influence, the provider’s reputation, and competition. Our paper disentangles the 

influence of these factors. We model a viewer’s decision as first choosing 

categories and then videos within the categories. The hierarchical model allows 

videos within a category to have correlated demand. 

2 - The Value of Being Social: How Bloggers Attract Followers 

Hailiang Chen, City University of Hong Kong, Department of 

Information Systems, Hong Kong, Hong Kong-PRC, 

chenhl.2007@gmail.com, Yu Hu, Prabuddha De 

In a blog community, for a blogger to build readership and gain popularity, 

ultimately, the key should be to write good quality content. However, from a 

marketing perspective, how much additional value does being social with her 

readers bring about to the blogger? This study intends to quantify the value of 

being social in the blogging context and to examine how bloggers should manage 

their conversations with readers. 

3 - Information Exchange in Prediction Markets: How Social 

Networks Promote Forecast Efficiency 

Liangfei Qiu, University of Texas at Austin, 2225 Speedway, 

Austin, TX, 78712, United States of America, 

qiuliangfei@gmail.com, Andrew Whinston, Huaxia Rui 

This paper studies the effects of information exchange and social networks on the 

performance of prediction markets. We provide a game-theoretic framework to 

resolve the question: Can social networks and information exchange promote the 

forecast efficiency in prediction markets? Our study shows that social network is 

not a panacea in terms of improving forecast efficiency. The abuse of social 

networks could be detrimental to forecast performance.

SD71 

■ SD71 


Management and Economics of IS 



Chair: Dennis Kundisch, Professor, University of Paderborn, 

Warburgerstrasse 100, Paderborn, Germany, 

dennis.kundisch@wiwi.uni-paderborn.de 

1 - IT Investments and the Structure of Production 

Fengmei Gong, University of Calgary, Haskayne School of 

Business, Calgary, Canada, fgong@ucalgary.ca, Barrie Nault 

We examine the potential for determining how investments in information 

technology (IT) from an industry and its value chain partners affects the inputoutput 

structure of production. 

2 - Valuation of Interdependent IT Projects – A Real Option 

Approach Considering Unhedgeable Risks 

Steffen Zimmermann, Assistant Professor, University of Innsbruck, 

Universitaetsstr. 15, Innsbruck, 6020, Austria, 

Steffen.Zimmermann@uibk.ac.at, Marcel Mueller, Sebastian Stöckl 

To value interdependent IT projects, financial option pricing models like the 

Black-Scholes model (BSM) are frequently suggested. However, one major 

concern about applying these models for real options, is the assumption that all 

risks need to be hedged by a replicating portfolio. Thus, we present an 

enhancement of the BSM where we consider hedgeable and unhedgeable risks by 

two seperate stochastic processes. In doing so, we find out that the BSM 

undervalues real options systematically. 

3 - Risk-aware SLA Design for Enterprise Information Systems 

Dirk Neumann, Professor, Albert-Ludwigs Universitat Freiburg, 

Platz der Alten Synagoge 1, Freiburg, 79085, Germany, 

dirk.neumann@is.uni-freiburg.de, Markus Hedwig, 

Simon Malkowski 

Since effective IT systems are necessary for business practices, growing operation 

costs necessitate the search for efficient delivery models. Cloud Computing 

provides a powerful alternative to traditional IT operation. Firms still have 

reservations about migrating to the Cloud due to the lack of binding SLA 

offerings. Service providers hesitate to offer binding SLAs as assessing economic 

risks is challenging. We present a model for SLA design for clouds to estimate the 

induced operational risk. 

4 - Does It Pay Off to Bid Aggressively? An Empirical Study 

Philipp Herrmann, University of Paderborn, Warburger Strafle 100, 

Paderborn, Germany, Philipp.Herrmann@wiwi.uni-paderborn.de, 

Mohammad Rahman, Dennis Kundisch 

We empirically investigate the payoff of aggressive bidding in an electronic 

auction. We use a unique and very rich dataset containing actual market 

transaction data for approximately 7,000 pay-per-bid auctions. Our research 

design allows us to separate the effect of an aggressive bidding strategy to signal 

one’s own valuation from the effect of bidding aggressively due to impatience. We 

find a strong and significantly negative effect of aggressive bidding on the 

likelihood to win an auction. 

■ SD72 


Advances in Optimization Modeling Languages and 

Systems I 

Cluster: ICS-Advances in Optimization Modeling Languages and 

Systems 


Chair: Robert Fourer, President, AMPL Optimization, Inc, 2145 

Sheridan Road, Evanston, IL, United States of America, 4er@ampl.com 

1 - A Generic Benders Decomposition Algorithm for the AIMMS 

Modeling Language 

Marcel Hunting, Paragon Decision Technology, Schipholweg 1, 

Haarlem, 2034 LS, Netherlands, Marcel.Hunting@aimms.com 

We present an implementation of Benders Decomposition, in the AIMMS 

modeling language, that is easy to use on a broad class of problems. The user only 

needs to specify which variables belong to the master problem and which to the 

subproblems. Benders’ cuts are generated automatically. Several cut selection 

criteria are available. The Benders Decomposition algorithm is implemented as a 

white box algorithm. 


156 

2 - Automated Conversion of Common Optimization Problem 

Structures to Mixed-integer Linear Programs 

Jared Erickson, Northwestern University, 2145 Sheridan Road, 

Room C210, Evanston, IL, 60208, United States of America, 

JaredErickson2012@u.northwestern.edu, Robert Fourer 

Optimization problems conveniently represented through the use of disjunctive 

linear constraints or polynomials in binary variables are commonly recast as 

MILPs for solution. We present algorithms that detect these types of problems and 

transform them to MILPs automatically. Our implementation uses the AMPL 

modeling language and works with various solvers. The transformations allow for 

a wider variety of solvers to be used and can lead to reductions in solution time. 

3 - ZIMPL 3.3: An Open Source Modelling Language 

Gerald Gamrath, Research Assistant, Zuse-Institut Berlin, Takustr 

7, Berlin, 14199, Germany, gamrath@zib.de, Thorsten Koch 

ZIMPL is a freely available open source algebraic modelling language for 

describing linear and non-linear mixed integer programs. It is very portable and 

can be used either standalone, embedded into an application, or linked to a 

solver. One feature of ZIMPL is the use of rational arithmetic for all computations. 

We will give an overview of the latest release and in particular to the SCIP 

integration, which interfaces with the exact solver version of SCIP and also with 

the MINLP solver. 

4 - APMonitor Modeling Langauge for Mixed-integer Differential 

Algebraic Systems 

John Hedengren, Assistant Professor, Brigham Young University, 

350 CB, Provo, UT, 84602, United States of America, 

john_hedengren@byu.edu 

The APMonitor Modeling Language (APM) is an optimization platform for 

differential and algebraic equations (DAEs) and is coupled with large-scale solvers 

for data reconciliation, dynamic optimization, and nonlinear predictive control. 

Applications include computational biology, unmanned aerial systems, chemical 

process control, smart grid optimization, and oil and gas upstream monitoring 

systems. 

■ SD73 


Stochastic Optimization in Finance 



Chair: Tim Siu-Tang Leung, Columbia University, New York, NY, 

United States of America, leung@ieor.columbia.edu 

1 - Utility Maximization in Hidden Regime Switching Markets 

Agostino Capponi, Assistant Professor, Purdue University, 

315 North Grant Street, West Lafayette, IN, 47906, 

United States of America, capponi@purdue.edu 

We consider a regime switching market, where investors can trade a stock, a riskfree 

bond, and a defaultable bond, whose price is only observed with noise. The 

dynamics of the instruments are driven by hidden Markov chains. We provide the 

HJB equations for the partially observed stochastic control problem, and 

demonstrate how to solve them numerically. We then illustrate the dependence 

of the strategies on the level of noise in market, risk aversion of the investor, and 

default risk. 

2 - Portfolio Optimization under Convex Incentive Schemes 

Maxim Bichuch, Princeton University, Dept. of Operations 

Research and Finance, Sherrerd Hall, Charlton Street, Princeton, 

NJ, 08544, United States of America, mbichuch@princeton.edu, 

Stephan Sturm 

We consider a utility maximization problem of terminal wealth from the point of 

view of a portfolio manager paid by convex incentives. Even though the 

manager’s utility function is concave, the result is a non-concave optimization 

problem that does not fit into the classical portfolio optimization theory. Using 

duality theory, we prove existence and uniqueness of the optimal wealth in 

general (incomplete) markets. 

3 - Optimal Incentives for Delegated Portfolio Selection 

Stephan Sturm, Worcester Polytechnic Institute, c/o Princeton 

University, 116, Sherrerd Hall, Princeton, NJ, 08544, 

United States of America, ssturm@princeton.edu, Maxim Bichuch 

We study the problem of an investor who lets a fund manager manage his wealth. 

The latter is paid by an incentive scheme based on the performance of the fund. 

Manager and investor have different risk aversions; the manager may invest in a 

financial market to form a portfolio optimal for his expected utility whereas the 

investor is free to choose the incentives – taking only into account that the 

manager is paid enough to accept the managing contract.

4 - Private Valuation of Claims using Stochastic Programming 

Alan King, IBM Research, 1101 Kitchawan Road, Yorktown 

Heights, NY, 10598, United States of America, kingaj@us.ibm.com, 

Olga Streltchenko 

Valuing contingent claims can be formulated as a stochastic programming 

problem. We explore the advantages of the SP model in modeling realistic aspects 

of investing, including: calibration constraints, existing positions, spreads, and risk 

preferences. Capturing these realistic features of investing and hedging leads to a 

paradigm in which investors can construct and calibrate a private valuation 

measure for the evaluation of trades and the pricing of illiquid positions. 

Monday, 8:00am - 9:30am 

■ MA01 

01- West 101- CC 

Recent Advances in Nonlinear Optimization 

Sponsor: Optimization/Global Optimization 


Chair: Evrim Dalkiran, Assistant Professor, Wayne State University, 

4815 4th St Manufacturing Eng., Detroit, Mi, 48202, 

United States of America, evrimd@wayne.edu 

1 - Disjunctive Cutting Planes for Linear 

Complementarity Constraints 

Trang Nguyen, University of Florida, 303 Weil Hall, Gainesville, FL, 

32611, United States of America, trang@ufl.edu, 

Jean-Philippe Richard, Mohit Tawarmalani 

We discuss the problem of generating strong cutting planes for linear programs 

with linear complementarity constraints (LPCCs). In particular, we exploit 

complementarity constraints to derive cuts from the optimal simplex tableaux of 

the LP relaxation of the problem. We also introduce the notions of split cut and 

split closure for LPCC problems with bounded variables. We compare these 

notions with their integer programming counterpart and present preliminary 

numerical results. 

2 - Complexity of Deciding Convexity 

Amir Ali Ahmadi, Postdoctoral Associate, Massachusetts Institute 

of Technology, 195 Binney St., Cambridge, MA, 02142, 

United States of America, a_a_a@mit.edu 

We show that deciding convexity of polynomials of degree as low as four is 

strongly NP-hard. This solves a problem that appeared as one of seven open 

problems in complexity theory for numerical optimization in 1992. Joint work 

with A. Olshevsky, P.A. Parrilo, and J.N. Tsitsiklis. 

3 - An Algorithm for the Projection a Point on the Intersection of 

Two Nonnegative Hyperplanes and the Nonnegative Orthant 

Claudio Santiago, Postdoc Researcher, Lawrence Livermore 

National Laboratory, 7000 East Ave., Livermore, CA, 94550, 

United States of America, pratas@gmail.com, Nelson Maculan, 

Maria Helena Jardim 

In this work, we present a quadratic algorithm for the projection of a point on the 

intersection of two hyperplanes with nonnegative coefficients and the 

nonnegative orthant. Interior point methods are the most efficient algorithm in 

the literature to solve this problem. While efficient in practice, their efficiency is 

highly dependent on a series of parameters (especially for nonlinear problems). 

We propose a new method based on the KKT optimality conditions. 

■ MA02 

02- West 102 A- CC 

Issues and Advances in Real Options 



Chair: Reidar Bratvold, Professor, University of Stavanger, 

Department of Petroleum Engineering, Stavanger, 4036, Norway, 

reidar.bratvold@uis.no 

Chair: Joe HahnAssociate Professor, Pepperdine University, 

Malibu Campus, Malibu, CA, United States of America, 

Joe.Hahn@pepperdine.edu 

1 - Recent Developments in Valuation to Support Real Asset 

Decision-Making: A CO2 Storage Example 

David Laughton, Principal/Adjunct Professor, David Laughton 

Consulting Ltd./University of Alberta School of Business, 


157 

11006-125 St., Edmonton, AB, T5M 0M1, Canada, 

laughton.david@davidlaughtonconsulting.ca 

Value is an important metric in many large public corporations shaping the 

process of designing and choosing real assets. Separating and adopting the two 

aspects of so-called “real options” valuation (derivative asset valuation and 

decision tree analysis) can produce a technically superior form of valuation that 

conforms with the organisational structures and work flows in these corporations. 

This increases the net benefit of adopting the proposed changes and makes it less 

costly to do so. 

2 - A Decision Analysis Approach to Valuing Real Options 

Ross Shachter, Associate Professor, Stanford University, 

Department of Management Science and Eng, Huang Engineering 


shachter@stanford.edu, Stephen Derby 

We extend the standard finite-state multiple-period model of investment in a real 

option by including return to scale (RTS), the derivative of the buying price with 

respect to the scale of investment. RTS is either constant or decreasing, depending 

on whether all of the investment risk can be hedged, as assumed in financial 

option pricing theory. RTS increases with wealth and trading frequency. We 

provide bounds and approximations for the investment value without imposing 

constant RTS. 

3 - Risk-Sensitive Real Options Decisions 

H. Dharma Kwon, University of Illinois at Urbana-Champaign, 

1206 South Sixth Street, Champaign, IL, 61820, 

United States of America, dhkwon@illinois.edu, Michael Lim 

We consider a class of stopping problems under uncertainty with non-linear 

utility functions. We focus on stationary stopping problems and consider 

stationary stopping policies. We formulate credible stationary stopping policies, 

from which the decision-maker will have no incentive to deviate. 

4 - The Choice of Stochastic Process in Real Option Valuation 

Luiz Ozorio, Professor, IBMEC Business School, Av. Afranio de 

Melo Franco 149/403, Rio de Janeiro, RJ, 22430-060, Brazil, 

lmozorio@ibmecrj.br, Luiz Eduardo Brandao, Adrian Pizzinga, 

Carlos Bastian-Pinto 

A main issue in valuation modeling is the choice of the stochastic process that 

describes the asset price performance. Particularly, in investment projects that 

show a high level of managerial flexibility in conditions of uncertainty the 

assumption of a specific process can have an impact on the project value and the 

investment rule. This work discusses the choice of stochastic process in real 

options valuation and useful tests and theoretical considerations to give support 

to this task. 

■ MA03 

MA03 

03- West 102 B- CC 

Joint Session: DAS/Emergency Mgmt: Security and 

Behavioral Game Theory 

Sponsor: Decision Analysis & Applications in Emergency 

Management and Terrorism Security 


Chair: Richard John, Associate Professor of Psychology, University of 

Southern California, Los Angeles, CA, United States of America, 

richardj@usc.edu 

Co-Chair: Heather Rosoff, University of Southern California, 

Los Angeles, CA, United States of America, rosoff@usc.edu 

1 - Security Games: Challenges in Modeling Human 

Adversarial Behavior 

Milind Tambe, University of Southern California, Los Angeles, CA, 

United States of America, tambe@usc.edu, Rong Yang, James Pita 

Stackelberg security games have received significant attentions in recent years 

given their deployment for real-world security problems. Many of these realworld 

systems have adopted the standard game-theoretic assumption that the 

adversaries are perfectly rational; this assumption may not hold when facing 

human adversary with bounded rationality. We provide an overview of our 

research efforts to address human bounded rationality in the context of security 

games.

Sunday

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?