Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WA43

2 - Coordinating Decisions in Decentralized Perishable-products Supply Chains via Contracts Sandra Transchel, Kuehne Logistics University, Grosser Grasbrook 17, Hamburg, 20457, Germany We consider a two-stage supply chain consisting of a vendor and buyer selling a perishable product, more specifically, a product with a finite life time. We investigate different contracts and their impact on both the vendor’s and the buyer’s inventory decision in a decentralized decision-making environment. Beside the wholesale price, the contracts specify the minimum residual shelf life the buyer is willing to accept as well as a service level agreement and penalty payments. We compare the decentralized solution with the inventory policy of a centralized decision-maker and show the impact on the supply chains profitability and waste performance. 3 - Head Against the Wall: The Connection between Concussions and Overconfidence Dominik Piehlmaier, Dissertator in Consumer Behavior, University of Wisconsin-Madison, Madison, WI, 53703, United States An estimated average of 3.8 million concussions occur every year within the US. However, only approximately half of all cases are reported (Harmon et al., 2013). This study sheds light on the influence of overconfidence on the propensity to experience a concussion and on the likelihood to report it. Data drawn from survey responses from around 1,000 athletes were used to analyze these aspects. A set of Bayesian regressions and Bayesian structural models, using an efficient Hamiltonian Monte Carlo sampler, was fitted to analyze the data. The results suggest that athletes who are overconfident in their ability to detect a concussion, are more likely to experience one but are less likely to report it. 4 - Influence Modeling: Mathematical Programming Representations of Persuasion under Either Risk or Uncertainty William Nicholas Caballero, Air Force Institute of Technology, 6465 Hemingway Road, Dayton, OH, 45424, United States, Brian J. Lunday Persuasion is a fundamental element of human interaction applied to both individuals and populations. Although the study of persuasion has historically been dominated by qualitative models, this research advances its quantitative characterization and future use. This research complements the qualitative psychological literature with respect to the processing of persuasive messages by developing an influence campaign design framework. We adapt the classic Decision Analysis problem to a bilevel mathematical program, wherein a persuader has the opportunity to affect the environment prior to the decisionmaker’s choice. Thereby, we define a new class of problems for modeling persuasion. 5 - When Payoffs Look Like Probabilities: Separating Form and Content in Risky Choice Johannes Müller-Trede, IESE Business School, Barcelona, Spain, Shlomi Sher, Craig R. McKenzie Prospect theory assumes a value function that is concave for gains and convex for losses, and an inverse S-shaped probability weighting function. But in typical studies, form and content are confounded: Probabilities are represented on a bounded scale, whereas representations of gains (losses) are unbounded above (below). To unconfound form and content, we ran studies employing a probability-like representation of outcomes and an outcome-like representation of probability. We show that interchanging numerical representations can interchange the resulting psychophysical functions. Traditional models may reflect subjective reactions to numerical form rather than substantive content. n WA42 North Bldg 227A Simulation and Optimization Contributed Session Chair: Harri K. Ehtamo, Aalto University, P.O. Box 1100, FIN-02015 HUT, Espoo, 02150, Finland 1 - Modeling Systematic Technology Adoption with Heterogeneous Interacting Agents Huayi Chen, Nanjing University of of Aeronautics and Astronautics, 29 Yudao street, Nanjing, 210016, China Traditional systematic technology adoption models often assume one representative agent. However, researchers have been arguing that even a well calibrated representative agent cannot represent heterogeneous agents due to several reasons, of which one, is that there could be no trade. This paper intends to build a systematic optimization model of technology adoption with heterogeneous interacting agents. Agents are cost minimizing entities and the interaction among them could be the trade in goods, new advanced technology,

and further, in carbon emission credits. 2 - An Analysis of Overlapping Appointments in Outpatient Oncology Centers Melissa Marquez, Binghamton University, New York, NY, 10033, United States This research proposes a simulation-based optimization approach to minimize the total cost by considering appointment overlapping for chemotherapy scheduling in outpatient oncology centers (OOC). The model considers waiting time, chair idle time, and total clinic overtime, uses on-line Best Fit bin packing heuristic, and optimizes pre-treatment and infusion duration. The model considers uncertainties such as patient no-shows, patient tardiness, and variability in scheduled infusion duration and processing times. The appointment overlapping with efficient scheduling increases infusion chair utilization, and reduces total cost in OOC with high no-show rates. 3 - A Surrogate-based Tabu Search Heuristic to Optimise the People Flows in a Timetable Hendrik Vermuyten, KU Leuven, Warmoesberg 26, Brussel, 2800, Belgium, Jeroen Belien, Liesje De Boeck, Tony Wauters In this work, we address the problem of minimizing people flows that are the result of timetabling decisions. We assume that each event has already been assigned to a certain timeslot and we only take the decision of assigning events to rooms into account. The crowd dynamics are modelled by Menge, which is an open source microscopic pedestrian simulator, and tabu search is used to iteratively find improved solutions. To speed up the computations, a surrogate model is used that approximates the real objective function values of candidate solutions and is computationally much less costly. 4 - Optimizing People Flow and Safety in Buildings Harri Ehtamo, Professor, Aalto University, P.O. Box 11100, 00076 Aalto, Espoo, 02150, Finland, Juha-Matti Kuusinen, Janne Sorsa, Marja-Liisa Siikonen, Henri Hakonen We discuss some key advances in people flow and building safety modeling based on several years of collaborative research carried out by KONE Corporation, one of the global leaders in the elevator and escalator industry, and Systems Analysis Laboratory of the AALTO University. The talk is partly based on our OR/MS Today, April 2017, article on the subject. n WA43 North Bldg 227B Stochastic Optimization I Contributed Session Chair: Cheng-Yu Rao, Shanghai St., Taitung County 950, New Taipei City, 95048, Taiwan 1 - Stochastic Integer Programming Formulation for Process Discovery Georges Spyrides, Student, PUC-Rio, Rua Marques de Sao Vicente 225 - Gávea, Rio de Janeiro, 22451900, Brazil, Marcus V. Poggi Process discovery algorithms try to build meaningful Petri-nets from a log of events. Bergethum (2008) proposes that candidates for Petri-net places are related as basic solutions on an ILP formulation. Van Der Werf et al (2009) proposed a general method called ILP Miner. This method generates fully replayable Petri- nets, but with a downside of producing hard to understand process models in practice. This work aims to address the simplicity problem by interpreting the frequency of a sequence of events as the probability of materialization in the future. We propose a novel two stage stochastic formulation with chance constraints, which produces models not so sensitive to outlier process variants. 2 - A Long-term Analysis of Renewable-dominated Regions: A Stochastic Co-optimization of Expansion Planning and Dynamic Probabilistic Reserve Alessandro Soares, Engineer, PSR, Praia de Botafogo 228, Rio de Janeiro, Brazil As consequence of the variability in the renewable generation, system reserve requirement must be increased to maintain security and thermal units may need to operate outside their efficient operation point. Since thermal flexibility and reserve may result in substantial costs to the system, expansion planning models need to consider these costs and co-optimize them along with the expansion. We present a Multiscale Stochastic Optimization approach to optimize dynamic probabilistic reserves (DPR) and system flexibility costs due to high renewable penetration. A real-case study of the Chilean system will be analyzed to illustrate the methodology.

425