Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WB02

2 - Robust Power Dispatch and Renewable Energy Management Ruiwei Jiang, University of Michigan, Ann Arbor, MI, 48109, United States, Hongyan Ma In this paper, we propose a robust power dispatch approach through incorporating a corrective re-dispatch and integrating active management of renewable energy, which co-optimizes the power pre-dispatch strategies and the admissible ranges of renewable outputs. Case studies on the modified IEEE systems display the effectiveness and scalability of this approach. 3 - Risk Averse Energy Storage Optimization for High Penetration of Wind Energy We propose a modified stochastic dual dynamic programming (SDDP) method aimed at optimizing energy storage for a set of batteries scattered across the energy grid. With the fine-grained time scale of battery storage, we also have to optimize over hundreds of time periods. We consider a hidden semi-Markov model (HSMM) that accurately reproduces the crossing-time behavior of the wind, which captures the amount of time that actual wind sample paths are above or below the forecast. We show that we can significantly decrease the risk of shortages when we consider our HSMM model coupled with the proposed modified SDDP method over the classical iid stochastic model coupled with a standard SDDP algorithm. 4 - Planning Transmission Storage and Generation for Renewable Energy and Carbon Policies Uncertainty Jing Peng, Johns Hopkins University, 3400 N. Charles Street, Baltimore, MD, 21218, United States, Qingyu Xu, Benjamin Field Hobbs Renewable energy policies and carbon policies are reshaping the power system by providing incentives to invest more in renewables and displace coal generation. However, the particular timing and implementations of these policies are uncertain to the power system planners. Due to rapidly dropping costs, the energy storage is becoming an increasingly competitive option. We propose an optimization model to plan energy storage, accounting for its substitution and complementary relations with transmission and generation, to hedge against policy uncertainties in the western interconnection of North America. Joint Session OPT/Practice Curated: Stochastic Assignment Problem Applications to Healthcare and Network Design Sponsored: Optimization/Optimization under Uncertainty Sponsored Session Chair: Onur Tavaslioglu, University of Pittsburgh, Pittsburgh, PA, United States 1 - Stochastic Operating Room Scheduling Under Emergency Arrivals with Integrated Block Assignments Onur Tavaslioglu, University of Pittsburgh, Houston, TX, 77025, United States, Oleg A. Prokopyev, Andrew J. Schaefer Operating room scheduling for elective procedures is challenging due to the nonlinear structure and stochasticity. In this paper, we model the operating room scheduling problem as a two-stage stochastic mixed integer program with nonlinear objective and chance constraints. We then present a value function reformulation, which converts the mentioned model into a pure binary program. We propose a dynamic programming based approach to calculate the value functions. We compare our approach to a cut generation approximation from the literature. The performance of our approach is better in computational time and it offers optimal solutions. 2 - Facility Protection and Network Design when the Effect of Protection is Uncertain Tanveer Hossain Bhuiyan, Mississippi State University, McCain Engineering Building, Room 321, Starkville, MS, 39759, United States, Hugh Medal We study a facility location and network design problem that involves protecting facilities subject to random disruptions where the protection is imperfect, multi- level, and the effect of disruption is imperfect. The goal of our study is to optimally allocate protection resources to the facilities, and construct links in the network to minimize the expected transportation cost. We model the problem as a two-stage stochastic programming with decision dependent uncertainty where the post-disruption capacity states of the facilities depends probabilistically on the resource allocation decision and the disruption intensity. We implement an L- shaped algorithm to solve the model. Juliana Nascimento, Princeton University, Sherrerd Hall, Operations Research and Fincl Engineering, Princeton, NJ, United States, Joseph Durante, Warren B. Powell n WB02 North Bldg 121B

3 - Chance-constrained Bin Packing Problem with an Application to Operating Room Scheduling Shanshan Wang, Beijing Institute of Technology, 5 South Street

Zhongguancun, Haidian District, Beijing, 100081, China Shanshan Wang, Northwestern University, Evanston, IL, United States, Sanjay Mehrotra, Jinlin Li

We develope a branch-and-cut solution scheme for chance constrained optimization of bin-packing problems with random technology matrices. The problem is to allocate items with random weigh to a set of bins with respect to bin size, so as to minimize the total open and allocation cost subjecting to the packing constraints with given probabilistic guarantees. We formulate the integer programs by assuming discrete distributions of random weight. We propose a branch-and-cut framework with several new classes of valid inequalities. Computational study on chance-constrained formulation of surgery assignment problem is conducted to verify the performance of our algorithms. 4 - Multi-mode Resource Constrained Project Scheduling with Optional Activities and Uncertain Network Structure Chrysanthos Gounaris, Carnegie Mellon University, Doherty Hall 3107, Dept of Chemical Engineering, Pittsburgh, PA, 15213, United States, Nikolaos Lappas, Hua Wang The classical Multi-mode Resource Constrained Project Scheduling Problem (MMRCPSP) is defined upon a fixed directed graph modeling the precedence of activities that have to be executed exactly once in one of their available modes. In practice, however, there are cases where network structure can be altered both by internal strategic decisions as well as uncertain external factors. To that end, we extend MMRCPSP to accommodate flexible network structures that involve recycles and optional activities via the use of propositional logic, and we present a set of new applicable models that are compared extensively across a comprehensive list of benchmark problems. n WB03 North Bldg 121C Risk-averse and Robust Optimization Sponsored: Optimization/Optimization under Uncertainty Sponsored Session Chair: Alexander Vinel, Auburn University, Auburn, AL, 36832, United States 1 - Hazardous Materials Routing and Network Design under Uncertainty Considering Risk Equity Nasrin Mohabbati Kalejahi, PhD Candidate, Auburn University, Auburn, AL, United States, Alexander Vinel Moving hazardous materials raises an inherent risk for public safety. One aspect that has received a significant attention in the literature recently relates to the observation that using a single route repeatedly can lead to an undesirable overload of hazmat risk on specific links of the network and risk inequity. In this work, we study hazmat route planning and network design with consideration of risk equity. We combined the concept of Risk Parity with modern risk functions and developed a two stage risk-reward-diversification framework for optimally selecting routes and distributing the exposure to the risk in a network. 2 - Distributionally Robust Optimization with Decision-dependent Ambiguity Set Nilay Noyan, Sabanci University, Faculty of Engineering and Natural Sciences, Orhanli/Tuzla Istanbul, 34956, Turkey, Miguel Lejeune, Gabor Rudolf We introduce a new class of distributionally robust optimization problems under decision-dependent ambiguity sets. In particular, as our ambiguity sets we consider balls centered on a decision-dependent probability distribution. The balls are based on a class of Earth Mover’s Distances that includes both the Total Variation Distance and the Wasserstein-1 metric. We also consider a special class of problems where decisions are binary and the inherent randomness is characterized by a set of binary vectors, and develop mixed-integer linear programming reformulations. 3 - Adjustable Robust Optimization Applied to a Planning Problem in the Wine Industry Jorge R. Vera, Catholic University of Chile, Dept of Ind Eng Campus San Joaquin, Vicuna Mackenna 4860, Santiago, Chile, Rodrigo Cofr Uncertainty is common in industry and particularly in agriculture, where climate and crop yield are some of the relevant sources. In this work we consider a harvest planning problem in the wine industry, where grapes are harvested and transported to wineries. However, variability in the fermentation process introduces uncertainty in the processing capacity of the winery. Grapes cannot be stored, so harvest planning must take this into consideration. In this work we present a 2 Stage Stochastic approach as well as an Adjustable Robust approach for harvest planning, considering also the potential effect in product quality. We show results of different solution schemes for a real industrial case.

440