2016 INFORMS Annual Meeting Program

WC10

INFORMS Nashville – 2016

WC10 103C-MCC Designing Energy and Water Supply Chains for Prosperity Sponsored: Energy, Natural Res & the Environment, Energy Sponsored Session Chair: Yao Zhao, 1 Washington Street, Newark, NJ, 07102, United States, yaozhao@andromeda.rutgers.edu 1 - Distressed Selling By Farmers: Policy Recommendations Shivam Gupta, University of Texas Dallas, sxg104920@utdallas.edu, Milind Dawande, Ganesh Janakiraman, Ashutosh Sarkar In many developing countries, farmers sell a significant portion of their produce at prices much lower than the guaranteed support price offered by the government. We build a dynamic programming model to analyze this practice and illustrate how it can serve as a useful decision making tool for policy recommendations. 2 - Designing Hydro Supply Chains For Water, Food, Energy And Flood Nexus Kwon Gi Mun, Assistant Professor, Fairleigh Dickinson University, Teaneck, NJ, United States, kgmun@fdu.edu, Raza Ali Rafique, Yao Zhao The interconnected issues of water, food, energy and flood are among the most formidable challenges faced by developing countries. We apply SCM principles to water resource development and provide the end-to-end and dynamic perspectives needed in the expansion of hydropower network, and also identify the unique features and economies of hydropower systems and construct an integrated location optimization model to capture the conflicts of these issues, to explore the synergy among different sectors, and to maximize the overall benefit. With the real-life situation of Pakistan, we provide solutions that outperform common practices in all aspects of energy, irrigation, and flood control. 3 - Agricultural Support Prices In Developing Economies: Operational Analysis And Its Use In Policy-making Harish Guda, University of Texas Dallas, hxg131530@utdallas.edu, Tharanga Kumudini Rajapakshe, Milind Dawande, Ganesh Janakiraman The Guaranteed Support Price (GSP) scheme has been adopted in several developing economies. Through this scheme, the government promises to procure a crop from farmers at a guaranteed (and attractive) price announced ahead of the selling season, and then distributes the procured amount to the underprivileged population. The goal of this scheme is twofold: (a) as a supply- side incentive, to ensure high output from farmers, and (b) as a demand-side provisioning tool, to subsidize the consumption of the poor. In this talk, I present our work on the operational decisions of the farmers and the government under the GSP scheme, its impact on social welfare, and the use of our analysis to policy-makers. 4 - Coordinating And Sharing Demand-side Energy Resources – A Conceptual Design Wei Qi, Lawrence Berkeley National Laboratory, WQi@lbl.gov, Bo Shen, Hongcai Zhang, Zuo-Jun Max Shen We present a coordination scheme for shared use of demand-side energy resources (e.g. distributed generation, electric vehicles, etc.). Multiple users form a sharing community within which trading electricity improves economic efficiency. We develop a cost splitting scheme to ensure the participation of the aggregator and the users. Decision Making Under Multistage Uncertainty Sponsored: Optimization, Linear and Conic Optimization Sponsored Session Chair: Kartikey Sharma, Northwestern University, 2145 Sheridan Rd, Evanston, IL, 60208, United States, kartikeysharma2014@u.northwestern.edu Co-Chair: Omid Nohadani, Northwestern University, Northwestern University, Evanston, IL, 60208, United States, nohadani@northwestern.edu 1 - Adjustable Robust Optimization Via Fourier-motzkin Elimination Jianzhe Zhen, Tilburg University, Tilburg, Netherlands, J.Zhen@tilburguniversity.edu, Melvyn Sim, Dick den Hertog We demonstrate how adjustable robust optimization (ARO) problems with fixed recourse can be casted as static robust optimization problems via Fourier-Motzkin WC11 104A-MCC

elimination (FME). Through the lens of FME, we characterize the structures of the optimal decision rules for a broader class of ARO problems. A scheme based on a blending of classical FME and a simple Linear Programming technique, that can efficiently remove redundant constraints, is used to reformulate ARO problems. This generic reformulation technique, contrasts with the classical approximation scheme via linear decision rules, enables us to solve adjustable optimization problems to optimality. 2 - Distributionally Robust Inventory Control When Demand Is A Martingale Linwei Xin, U of Illinois at Urbana-Champaign, lxin@illinois.edu, David Goldberg Independence of random demands across different periods is typically assumed in multi-period inventory models. In this talk, we consider a distributionally robust model in which the sequence of demands must take the form of a martingale with given mean and support. We explicitly compute the optimal policy and value, and shed light on the interplay between the optimal policy and worst-case martingale. We also compare to the analogous setting in which demand is independent across periods, and identify interesting differences between these two models. 3 - Robust Optimization With Decision Dependent Uncertainty Sets Kartikey Sharma, Northwestern University, kartikeysharma2014@u.northwestern.edu, Omid Nohadani Robust optimization is increasingly used to solve multistage optimization problems. In most such problems, the uncertainty sets are fixed. However in many cases, these sets can be influenced by decision variables. We present a two- stage robust optimization approach in which future uncertainty sets can be affected by the decisions made in the first stage. We illustrate the advantages of this model on a shortest path problem with uncertain arc lengths. 4 - Adaptive Probabilistic Satisficing Models Zhi Chen, National University of Singapore, National University of Singapore, Singapore, Singapore, chenzhi@u.nus.edu, Melvyn Sim In this paper, we study adaptive probabilistic satisficing models that can be used for multi-stage decision making. We introduce the finite adaptability into probabilistic satisficing models to overcome the difficulties of incorporating recourse decisions as arbitrary functions of unfolded uncertain parameters. For two-stage problems, we show that the complete adaptability is exact to the finite adaptability, under a mild monotone condition. We propose an iterative scheme for increasing the level of probabilistic satisficing. We discuss extensions of our results for multi-stage problems. Our computational studies present that the probabilistic satisficing solutions can be competitive. WC12 104B-MCC Recent Advances in Decision Diagrams for Optimization Sponsored: Optimization, Integer and Discrete Optimization Sponsored Session Chair: Andre Augusto Cire, University of Toronto Scarborough, Toronto, ON, Canada, acire@utsc.utoronto.ca 1 - A Generic Approach To Solving Sequencing Problems With Time-dependent Setup Times Tailoring dedicated solution approaches to solve scheduling and routing problems is often complex and time consuming. This work presents a flexible framework based on Constraint Programming, Mixed Integer Programming and Decision Diagrams to solve hard scheduling problems including the Time-Dependent (TD) TSP, TD-SOP and TD-TSP with Time Windows. The proposed method is sufficiently generic to render it applicable to a variety of related sequencing problems. Moreover, experiments indicate significant performance improvements over pure MIP or CP approaches. 2 - Decision Diagram Bounds For Integer Programming Models Christian Tjandraatmadja, Carnegie Mellon University, Pittsburgh, PA, United States, ctjandra@andrew.cmu.edu, Willem-Jan Van Hoeve Decision diagrams are capable of generating strong bounds in practice for discrete optimization problems when a certain structure is present. However, generalizing decision diagram techniques to integer programming can be challenging due to the lack of clear structure to exploit. We propose a framework to generate decision diagram bounds that aims to overcome this issue. We start with a set of base constraints that are well suited for decision diagrams and progressively incorporate further constraints via strengthening and Lagrangian relaxation. We discuss computational experiments on a set of binary optimization problems. Joris Kinable, Carnegie Mellon University, Pittsburgh, PA, United States, jkinable@cs.cmu.edu, Andre Augusto Cire, Willem-Jan Van Hoeve

430