2016 INFORMS Annual Meeting Program

TA17

INFORMS Nashville – 2016

2 - Sampling Based Optimization Algorithms For Power Systems Application Harsha Gangammanavar, Clemson University, harsha@clemson.edu

4 - Power System Planning In Fragile States: A Case Study Of South Sudan Evangelia Spyrou, Johns Hopkins University, Baltimore, MD, United States, elina.spirou@gmail.com, Morgan Bazilian, Debabrata Chattopadhyay, Benjamin Field Hobbs In countries suffering from fragility, conflict and violence, power system planning and investment is essential for development and economic growth. However, the sector has to contend with deep uncertainty that may impact on an already vulnerable power system. We propose the application of a multi-stage stochastic program that explicitly considers probability of conflict and its consequences on power system infrastructure. Results for the power system in South Sudan are provided and discussed. TA18 106A-MCC Finance, Portfolio I Contributed Session Chair: Markku Kallio, Professor, Aalto Univertsity, Runeberginkatu 22-24, Helsinki, FIN-00200, Finland, markku.kallio@aalto.fi 1 - Trade Space Exploration Tools And Methods With Applications To Capital Investments And Portfolio Management Decisions For Optimality Simon Miller, Applied Research Laboratory, The Pennsylvania State University, 411 Waupelani Drive, D-221, State College, PA, 16801, United States, swm154@psu.edu, Christopher M. Farrell, Michael A. Yukish, Gary M Stump Faced with constrained portfolio conditions, senior leaders must often make strategic choices and fiscal trades with implications on capabilities, capacities, and system attributes to maximize value, manage risk, and satisfy stakeholder requirements. Researchers have developed robust tools and methods to explore large scale, complex, multi-objective problems for portfolio analyses, where data visualization techniques and optimization algorithms are simultaneously applied to support decision processes in a binary combinatorial space. The tools and methods may be applied to a wide range of financial management and resource allocation problems to provide flexibility and options. 2 - Forth Order Stochastic Dominance Efficiency Test And An Empirical Evaluation Nasim Dehghan Hardoroudi, PhD Candidate, Aalto University School of Business, Runeberginkatu 22-24, Chydenia (4th floor), Helsinki, 00100, Finland, nasim.dehghan.hardoroudi@aalto.fi Stochastic dominance is an important tool in aiding decisions under uncertainty when the decision maker’s utility function is unknown. In this study, we propose a novel forth order stochastic dominance (FOSD) efficiency test. We derive the necessary and sufficient conditions for such test, which is based on nonlinear convex optimization problem. For comparison, we provide numerical illustrations for second and third order stochastic dominance (SSD, TSD) as well as decreasing absolute risk aversion stochastic dominance (DSD) besides the FOSD test, using stock market data of the US. The market index as benchmark is found inefficient Markku Kallio, Professor, Aalto Univertsity, Runeberginkatu 22-24, Helsinki, FIN-00200, Finland, markku.kallio@aalto.fi We consider third order stochastic dominance (TSD), decreasing absolute risk aversion (DARA) stochastic dominance (DSD) as well as stochastic dominance (ESD) based on the family negative exponential utility functions. These concepts are of interest because the respective classes of utility functions convey observed properties of individual preferences. Using the efficiency concept introduced by Post, we derive necessary and sufficient tests for efficiency under the three types of stochastic dominance. Our DSD efficiency test is new, it relies on our argument for the TSD test, and it circumvents shortcomings in recent literature. and dominated under the all types of stochastic dominance. 3 - Some Tests For Stochastic Dominance Efficiency

We present sampling-based approaches for addressing a class of stochastic optimization problems arising in power systems with significant renewable penetration, including economic dispatch and distributed storage control. These approaches provide a distribution-free alternative to methods based on Benders decomposition. This allows them to directly operate with external state-of-the-art simulators accessible to power systems operators. We will demonstrate their advantages in two-stage and multistage setups through computational experiments on real-scale power systems. 3 - Robust Strategic Bidding In Day Ahead Electricity Markets Bruno Fanzeres, Georgia Institute of Technology, 765 Ferst Drive NW, Atlanta, GA, United States, santosbruno85@gmail.com, Shabbir Ahmed, Alexandre Street The standard approach to devise bidding strategies in day-ahead electricity markets assumes available a joint probability distribution that drives the probabilistic nature of rival players’ behavior. Nevertheless, construct such probabilistic description is a challenging task due to its complex nature. In this talk, robust optimization techniques are adapted to the bidding strategy problem to characterize the uncertainty on rival players’ bids. A Column-and-Constraint Generation algorithm is constructed to solve the bidding problem. An illustrative example is presented to highlight the applicability of the proposed model as well as to provide intuition behind the algorithm. 4 - Modeling Power Markets With Multi-stage Stochastic Nash Equilibrium Joaquim Dias Garcia, PSR-Inc., Praia de Botafogo, 228, Botafogo, Rio de Janiero, Brazil, joaquimgarcia@psr-inc.com The modeling of modern power markets requires the representation of the following main features: (i) a stochastic dynamic decision process, with uncertainties related to renewable production and fuel costs, and (ii) a game- theoretic framework that represents the strategic behaviour of multiple agents., These features can be in theory represented as a stochastic dynamic programming recursion, where we have a Nash equilibrium for multiple agents. This work presents an iterative process to solve the above problem for realistic power systems. The proposed algorithm consists of a fixed point algorithm, in which, each step is solved via stochastic dual dynamic programming method. Stochastic Programming for Long Term Planning Sponsored: Optimization, Optimization Under Uncertainty Sponsored Session Chair: Anderson Rodrigo de Queiroz, NCSU, n, 1, NC, 12, United States, arqueiroz@ncsu.edu Co-Chair: Joseph F DeCarolis, North Carolina State University, Raleigh, NC, United States, jdecarolis@ncsu.edu 1 - The Value Of Stochastic Programming For Energy Systems Planning Anderson Rodrigo de Queiroz, North Carolina State University, Raleigh, NC, United States, arqueiroz@ncsu.edu, Joseph F DeCarolis Energy system models should reflect the reality that planners must make decisions prior to the realization of future uncertainties. Multi-stage stochastic programs, which embed uncertainty in the decision process, optimize over future possibilities to yield a near-term decision strategy. We use the expected value of perfect information and the value of the stochastic solution as metrics to quantify the value of such strategies for long-term capacity expansion of energy systems. 2 - Stochastic Optimization Of Design Under Heuristic Operation In Mixed Integer Programs Alexander Zolan, The University of Texas at Austin, alex.zolan@utexas.edu, David Morton, Alexandra M Newman We present a framework for optimizing system design in the face of a restricted class of policies governing system operation, which aim to model realistic operation for stochastic integer programs with a long operating time horizon. This leads to a natural decomposition of the problem yielding upper and lower bounds, which we can compute quickly. We illustrate application of these ideas using a model that seeks to design and operate a microgrid to support a forward- operating base under load and photovoltaic (PV) uncertainty, as well as other examples from the literature. TA17 105B-MCC

238