Informs Annual Meeting 2017

WB83

INFORMS Houston – 2017

WB80

processes under this type of uncertainty. We then discuss the implementation of a stochastic dual dynamic programming (SDDP) algorithm to compute an optimal solution for this problem. 4 - Robust Affine Control over Finite Time Horizon: The Stochastic Case Georgios Kotsalis, Georgia Intitute of Technology, Atlanta, GA, United States, gkotsalis3@gatech.edu, Guanghui Lan, Arkadi Nemirovski We provide a computationally tractable design procedure of affine policies for the constrained multistage robust optimization problem as it pertains to linear regime-switching systems that are subject to quadratic constraints while being affected by uncertain initial state and disturbances. We derive our results under the general assumptions of Markovian evolution of the stochastically switching signal and provided that the initial state and external disturbances lie within some nominal range expressed as the intersection of ellipsoids centered at the origin. 382C Optimization, Stochastic Contributed Session Chair: Xiaoyi Ma, Tongji University, Shanghai, China, 18917288185@163.com 1 - Co-optimizing the Interplan Between Micro and Macro Grids Luckny Zephyr, Postodoctoral Associate, Cornell University, 308 E Court Street, Apt 3, Ithaca, NY, 14850, United States, lz395@cornell.edu The increase in renewable resource penetration in bulk power systems introduces challenges in operations, due to the inability to confidently predict resource availability. The result of inaccurate forecasting is a reduction in reliability, with volatility in prices, line flows and voltages on the power transmission network. However, the ability of microgrids to be operated in both interconnected and islanded modes shows potential to leverage these smaller systems to manage disturbances in the bulk network, contributing to the integration of renewables in the power system. We will present our in-progress work to co-optimize the interaction between the main network and microgrids. 2 - Development of Variable Speed Limit System under Uncertain Driver Behaviour Nadia Moshahedi, University of Calgary, Calgary, AB, T3A2E1, Canada, nadia.moshahedi2@ucalgary.ca, Lina Kattan variable Speed Limit (VSL) is an Intelligent Transportation System solution that enables dynamic changing of speed limit in an attempt to improve mobility. Performance of these systems is highly influenced by fluctuation in traffic flow, which is caused by factors such as driving behaviour. To develop a proper VSL system uncertainty must be quantified as a part of optimization framework. In this work, we develop a Model Predictive Control VSL under uncertain driver behaviour. A stochastic MPC VSL system that minimizes total system travel time while constraining the change in speed limit over space and time is developed and tested in VISSIM simulation software using a rolling horizon approach. 3 - Scenario Group Approach for Risk Averse Mixed Integer Multi Stage Stochastic Programming Problems with Dynamic Mean CVAR Ali Irfan Mahmutogullari, Bilkent University, Ankara, Turkey, a.mahmutogullari@bilkent.edu.tr, Ozlem Cavus, Selim Akturk We propose a scenario tree decomposition approach, namely scenario grouping, to obtain different lower and upper bounds for risk-averse mixed-integer multi- stage stochastic programming problems with dynamic mean-CVaR objective. The method requires no assumption such as convexity, linearity, and complete recourse. For the case where the first-stage decision variables are binary, we also propose an exact solution algorithm by using the obtained bounds iteratively. We conduct computational experiments to observe the quality of the obtained bounds and the efficiency of the proposed algorithm. 4 - Safe Approximations of Chance Constrained Programs Xiaoyi Ma, Tongji University, Shanghai, China, 18917288185@163.com, Zhaolin Hu Chance constrained programs (CCP) are important models in stochastic optimization. In this paper, we seek to construct some approximations of the chance constraints. We show that the approximation problems are sometimes more tractable than the original CCPs. Our approach provides a new idea of balancing the conservativeness and the tractability of the CCPs. WB83

381C Joint Session PSOR/ENRE: Models for Energy Policy 2 Sponsored: Energy, Natural Res & the Environment, Energy Sponsored Session Chair: Erin Baker, University of Massachusetts-Amherst, Amherst, MA, 01003, United States, edbaker@ecs.umass.edu Co-Chair: Zana Cranmer, Bentley University, Waltham, MA, United States, zanacranmer@gmail.com 1 - Analysis of Competitive Energy Prices when Transportation is Capacitated Ian Yihang Zhu, University of Toronto, 144A, 60 Harbord Street, Toronto, ON, M5S3L1, Canada, i.zhu@mail.utoronto.ca, Timothy Chan, Michael Pavlin We study the effect that infrastructure and transportation constraints have on prices in a competitive energy market. In particular, we note the potential that these constraints have in creating market segmentation and localized markets in an otherwise perfectly competitive economy. We apply our model to the gasoline market, where we use inverse optimization to study the movement of prices across different states. 2 - Demand Response Aggregation with Satisficing Consumers in Electricity Markets Daniel E. Olivares, Assistant Professor, Pontificia Universidad Catolica de Chile, Vicuna Mackenna 4860, Macul, Santiago, Chile, dolivaresq@ing.puc.cl, Florian Salah, Rodrigo Henriquez, George Wenzel, Matias Negrete-Pincetic This talk discusses the impact of consumers’ behaviour on the portfolio design of a Demand Response (DR) aggregator. We develop an optimization model to decide the optimal portfolio of DR contracts for an aggregator participating in the electricity market, where consumers’ decisions are modeled using elements of satisfying theory. The model determines the premiums to be offered to consumers in order to obtain a DR portfolio that maximizes the aggregator’s operating surplus while satisfying the aspiration levels of participating consumers. Several simulations are performed to obtain insights on the value of the DR resource, and the importance of parameters used to model the consumers’ behavior. 382A Multistage Risk-Averse and Robust Stochastic Programming Sponsored: Optimization, Optimization Under Uncertainty Sponsored Session Chair: Lewis Ntaimo, Texas A&M University, College Station, TX, 77843, United States, ntaimo@tamu.edu 1 - Decomposition for Risk-averse Multistage Stochastic Programs with Expected Conditional Risk Measures Lewis Ntaimo, Texas A&M.University, 3131 TAMU, College Station, TX, 77843, United States, ntaimo@tamu.edu, Maryam Khatami, Bernardo Pagnoncelli We present a study of decomposition algorithms for risk-averse multistage stochastic programs (MSLPs) with expected conditional risk measures (ECRMs). The ECRMs have the advantage of being nested and time consistent. We study ECRMs in the context of both quantile and deviation mean-risk measures. 2 - Stochastic Decomposition for Risk-averse Multistage Stochastic Linear Programs Prasad Parab, Texas A&M.University, 1901 Holleman Drive W, #403, College Station, TX, 77840, United States, prasaddparab@tamu.edu, Lewis Ntaimo, Bernardo Kulnig Pagnoncelli Risk-averse multistage stochastic linear programs (MSLPs) can be hard to solve not only because of their large-scale nature, but also due to the risk measure being used. In this talk, we present a study of stochastic decomposition for MSLPs with quantile and deviation risk measures. 3 - The Application of Stochastic Optimization in Chemical Engineering Process Hanxi Bao, University of Florida, baohanxi@ufl.edu, Zhiqiang Zhou, Guanghui Lan, Zhaohui Tong Chemical engineering problems often involve uncertainty such as randomness in materials, reactions, and operations. In this work we establish a multistage stochastic programming model for the optimization and control of chemical WB81

503