Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TA45

4 - Detection and Attribution of Climate Change Using Offline Model Simulations for Streamflow in the Columbia River Basin Mingzhou Jin, University of Tennessee-Knoxville, 525D John D. Tickle Engineering Building, Industrial and Systems Engineering, Knoxville, TN, 37996, United States, Whitney Forbes, Jiafu Mao Detection and attribution analysis is the processes of statistically detecting a change in a particular climate variable or affected variables. Variables studied here are annual and seasonal in the Columbia River Basin for 1950 - 2008. The availability of a daily naturalized streamflow dataset and a new ensemble of semi- factorial land surface model simulations with less biased precipitation make the streamflow study for the basin more accurate. For forcings, the effects of climate change and variability, carbon dioxide concentration, nitrogen deposition, and land use and land cover change are used. Changes of the center of timing and seasonality were detected and attributed to climate changes. Managing Uncertainty in Electric Power Networks Sponsored: Energy, Natural Res & the Environment/Electricity Sponsored Session Chair: Johanna L. Mathieu, University of Michigan, Ann Arbor, MI, 48109, United States 1 - Data-driven Distributionally Robust Stochastic Optimal Power Flow Tyler Summers, University of Texas Dallas, Dallas, TX, United States We propose a data-based method to solve a multi-stage stochastic optimal power flow (OPF) problem based on limited information about forecast error distributions. Instead of assuming the uncertainties follow prescribed probability distributions, we consider ambiguity sets of distributions centered around a finite training dataset. By utilizing the Wasserstein metric to quantify differences between the empirical data-based distribution and the real unknown data- generating distribution, we formulate a multi-stage distributionally robust OPF problem to compute control policies that are robust to both forecast errors and sampling errors inherent in the dataset. 2 - Statistical Ranking for Flexible Robust Unit Commitment Lindsay Anderson, Cornell University, 316 Riley-Robb Hall, Ithaca, NY, 14853-5701, United States, Amandeep Gupta With increasing scale, complexity, and variability in modern power systems, there is a benefit to combining statistical tools with analytical framework of power system methods to provide much needed flexibility to system operators. This work describes a statistical ranking methodology that allows for adaptive robust stochastic unit commitment using a modular and customizable structure. The method and its applications are illustrated via case studies performed on IEEE-30 bus and 118-bus test systems and compared to other established approaches. The ranking model is further implemented in a multi-objective framework to analyze performance with competing worst-case scenarios. 3 - Operating Under Uncertainty in Smart Communities Kyri Baker Evolving energy systems are introducing heightened levels of uncertainty and stress on the electric grid. Increasing renewable generation, new market structures, and the proliferation of connected devices are transforming the operation of traditional energy systems, from the bulk grid to the operation of individual building components. To alleviate these issues, practical operational strategies which directly incorporate uncertainty need to be developed. This talk will introduce chance constrained and joint chance constrained optimization methods that aim to ensure operational guarantees in distribution networks and grid interactive buildings in future smart communities. 4 - Distributionally Robust Chance Constrained Optimal Power Flow Assuming Log-concave Distributions Bowen Li, University of Michigan, Ann Arbor, MI, United States, Ruiwei Jiang, Johanna Mathieu Chance constrained (CC) optimization is widely used to solve the optimal power flow (OPF) problem under uncertainty but its solution can be affected by an inaccurate estimate of the underlying distribution. To obtain a robust solution, distributionally robust (DR) techniques are used by considering a set of distributions sharing the common properties rather than a single distribution. Here, we develop a new DRCC OPF formulation that incorporates moment, support, and log-concavity information. We then derive approximations based on second-order cone programs and evaluate the results on IEEE test cases. 5 - Learning Demand Response Elasticity in Chance-constrained Optimal Power Flow Yury Dvorkin, New York University, Metrotech Center, Fay Street, Brooklyn, NY, 11201, United States The flexibility of demand response (DR) resources can be leveraged to accommodate the stochasticity of some distributed energy resources. This presentation will describe an approach to continuously learn time-variant price n TA44 North Bldg 227C

elasticity of DR resources and integrate this updated knowledge in the chance- constrained optimal power flow framework. This integration is shown to adequately remunerate DR resources and efficiently co-optimize the dispatch of DR and conventional generation resources.

n TA45 North Bldg 228A

Sustainable Energy and Environmental Policy Sponsored: Energy, Natural Res & the Environment/Electricity Sponsored Session Chair: Makoto Tanaka, National Graduate Institute for Policy Studies (GRIPS), 7-22-1 Roppongi, Minato-ku, Tokyo, 106-8677, Japan 1 - Regulatory Jurisdiction and Policy Coordination: A Bi-level Modeling Approach for Performance-based Policy Makoto Tanaka, National Graduate Institute for Policy Studies (GRIPS), 7-22-1 Roppongi, Minato-ku, Tokyo, 106-8677, Japan, Yihsu Chen, Afzal Siddiqui This study discusses important aspects of policy modeling based on a leader- follower game of policymakers. We specifically investigate non-cooperation between policymakers and the jurisdictional scope of regulation via bi-level programming. Performance-based environmental policy under the Clean Power Plan (CPP) in the U.S. is chosen for our analysis. We argue that integration of policymakers is welfare enhancing. Somewhat counterintuitively, full coordination among policymakers renders performance-based environmental policy redundant. We also find that distinct state-by-state regulation yields higher social welfare than broader regional regulation. 2 - Market Power in Policy Mix: Cap-and-trade and Renewable Portfolio Standards Mari Ito, Tokyo University of Science, Chiba, Japan, Ryuta Takashima Policies for reducing greenhouse gas emissions, e.g., emissions permits trading as cap-and-trade (C&T), and renewable energy policy as renewable portfolio standards (RPS), have been introduced in various countries. In this work, we examine market equilibria under C&T and RPS mix in bi-level optimization framework. For the lower level, generation of outputs of renewable and non- renewable sectors and electricity prices are decided by maximizing sectors’ profits. For the upper level, the policy maker chooses the rate of emission cap and the RPS requirement in an attempt to maximize the social welfare. Our results indicate that their policy mix is the best scheme for increasing social welfare. 3 - Multiagent Simulation with Reinforcement Learning Agents for Integrative Evaluation of Renewable Energy Policies Masaaki Suzuki, Tokyo University of Science, 2641 Yamazaki, Noda-shi, Chiba, Japan, Mari Ito, Ryuta Takashima Even as governments combat greenhouse emissions through a range of initiatives, scholarship has yet to clarify how renewable energy policy, energy market structure, and number of energy producers impact social welfare. We model a deregulated market for electricity as a blind single-price call auction and construct a multi-agent system with reinforcement learning that facilitates more realistic market evaluations and observation of equilibrium processes. We validate our simulation by comparing its results with the results from theoretical analysis in a simplified market. 4 - Investment in Power Generation and Transmission: The Effect of Capacity Size and Expansion Kazuya Ito, National Graduate Institute for Policy Studies (GRIPS), Japan, Makoto Tanaka, Ryuta Takashima The penetration of renewable energy has induced a decrease in capacity factors and a decommissioning for existing generations, and a reduction in new investments. This problem implies a fear for the shortage of capacity in the electricity market. Thus, in order to meet the capacity in the market, policymakers implement various policies for capacity procurement. In this work, we analyze investments in power generation and transmission by means of real options theory. The ISO decides the investment timing by maximizing social welfare whereas a power generator invests by maximizing the own profit. Especially, we examine the effect of capacity procurement for the ISO on the investment decisions.

5 - Real Options in Renewable Portfolio Standards Makoto Goto, Hokkaido University, Sapporo, Japan, Ryuta Takashima

In this paper, we examine a market equilibrium under uncertainty in RPS by means of real options analysis. More concretely, we analyze an investment timing for renewable producer. After that, we derive optimal RPS target. We have found results about the effect of uncertainty on market equilibrium and optimal RPS target. For fixed RPS target, high RPS target accelerates investment and decreases greenhouse gas. For the optimal RPS target, high RPS target disturbs innovation due to decrease in RE’s revenue from REC market, and increases greenhouse gas due to investment delay. This is a new finding in this area.

259