Informs Annual Meeting 2017

SB78B

INFORMS Houston – 2017

SB77

We will develop a decision-support tool Multi-stage and Multi-timescale robust Co-Optimization Planning (MMCOP) that augments existing power utility capabilities to support collaborative planning, analysis, and implementation of emerging variable and distributed power systems and help effectively mitigate risks and uncertainties in both short-term operation and long-term policy/technology changes. This tool will assist power market participants, utilities and regulatory agencies in analyzing economic, reliability, and sustainability issues when considering options for planning new and upgraded transmission facilities to accommodate existing and emerging generation sources.

372F 11:00 - 11:45 GAMS/ 11:45 - 12:30 Optimization Direct, Inc. Invited: Vendor Tutorial Invited Session 1 - GAMS - An Introduction Steven P. Dirkse, GAMS.Development Corporation, DC, WA, United States, sdirkse@gams.com We’ll introduce the key concepts of the GAMS language (e.g. sets, data, variables, equations) as we show how to build an optimization-based decision support application. Along the way, we’ll show how GAMS supports an easy growth path to larger and more sophisticated models and provides access to the most powerful large-scale solver packages. We will also look at some of the data management tools included in the GAMS system and show how to analyze and debug large problems using the various tools available within GAMS. Finally, we’ll show how our latest release supports integrating GAMS with object-oriented programming languages like C#, Python, and Java. 2 - DSx and CPLEX: Overview of Datascience Experience for Advanced Modeling and Optimization and LatestDevelopments/Results in CPLEX and ODh+CPLEX Alkis Vazacopoulos, Optimization Direct, Inc., 202 Parkway, Harrington Park, NJ, 07640, United States, alkis@optimizationdirect.com Organizations are increasingly hiring Data Scientists with Open Source skills. Leverage the capabilities of Data Science Experience DSx to work with Open Source tools like R, Python, Spark, as well as integration with CPLEX and Weather Company data. Come learn how DSx integrates with Open Source tools to enable clients to get the best of both worlds (Open Source programming and the Modeler GUI for those who prefer not to code). Furthermore we will review the latest developments/results in CPLEX Optimization Studio and the new ODh+CPLEX. 381A Addressing Variability and Uncertainty in Power Systems Operations with Alternative Approaches Sponsored: Energy, Natural Res & the Environment Electricity Sponsored Session Chair: Alberto J. Lamadrid, Lehigh University, Bethlehem, PA, 18015-3120, United States, ajl259@cornell.edu 1 - The Role of Stochastic Optimization in the Electric Power Sector Kory W. Hedman, Arizona State University, P.O. Box 875706, GWC206 School Of ECEE, Tempe, AZ, 85287-5706, United States, kwh@myuw.net The electric power sector is seeing an increase in stochastic resources, in the form of non-dispatchable or semi-dispatchable renewable resources as well as distributed energy resources. This talk will begin with a discussion on existing industry practices to manage uncertainty and recently proposed changes within the electric power sector. The presentation will cover how stochastic optimization techniques are already being used to help manage uncertainties and the presentation will cover how recent advanced in two-stage stochastic programming can further assist the electric power sector in the management of stochastic resources. 2 - Evaluating Electric Power Systems Miguel Ortega-Vazquez, Electric Power Research Institute, 3420 Hillview Avenue, Palo Alto, CA, 94304, United States, mortegavazquez@epri.com, Erik Ela This presentation will provide an update on research evaluating electric power system operating reserve methods and their comparison with advanced optimization methods for improved reliability and economic efficiency. 3 - Dynamic Electric Vehicle Travel Management in Coupled Power and Transportation Networks Hao Zhu, University of Illinois Urbana-Champaign, 306 N.Wright Street, 4056 ECE Building, Urbana, IL, 61801, United States, haozhu@illinois.edu This talk presents a framework to integrate the variable loads due to the changing of electrical vehicles into the operations of power systems. 4 - Multi-stage and Multi-timescale Robust Co-optimization Planning for Reliable and Sustainable Power Systems Lei Wu, Clarkson University, 8 Clarkson Avenue, P.O. Box 5720, Potsdam, NY, 13676, United States, lwu@clarkson.edu, Bo Zeng SB78

SB78B

380B Bayesian Approach Contributed Session Chair: Thomas G. Yeung, IMT-Atlantique, Nantes, France

1 - Valuing Demand Sample Information Prior to its Collection Adam Jason Fleischhacker, Associate Professor of Operations Management, University of Delaware, 222 Alfred Lerner Hall, Newark, DE, 19716, United States, ajf@udel.edu, Pak-Wing Fok, Mokshay Madiman We present both closed-form and computational approaches to valuing sample information prior to its collection. In particular, we look at the value of collecting demand data and its relationship to improving forecasts. Closed form approaches extend the existing literature on preposterior analysis and computational approaches leverage the probabilistic programming language Stan. 2 - Tuning a Bayesian Framework for B2b Pricing Fang Liang, SAS.Institute, Inc., 100 SAS.Campus Drive, Cary, NC, 27513, United States, fang.liang@sas.com, Maarten Oosten As in many other B2B markets, rail cargo companies negotiate prices with customers in the form of special agreements that are renewed on a regular basis. We propose a framework for optimizing the agreement prices that includes assessment of existing agreements, estimating the relationship between prices and performance of the agreements, leveraging this relationship to optimize agreement under various scenarios, and utilizing the scenario analysis to support negotiation. This approach needs to respond to changes in the business environment by learning quickly from the latest relevant observations. Bayesian approach is applied to accomplish this and the sensitivity of this approach is tuned. 3 - Evidential Reasoning Rule for Probabilistic Inference & Machine Learning with Ambiguous Data Dong-Ling Xu, Professor, Manchester University, Manchester combination, what assumptions are behind each rule & where the extensions are. We show how to obtain evidence from data (big or small) with ambiguity so that the inference using the ER rule constitutes a maximum likelihood probabilistic inference process, & how to use it in machine learning to build transparent models with maximum likelihood predictions. 4- Representing Markov Process as Dynamic Copula Bayesian Networks Thomas G.Yeung, Associate Professor, IMT-Atlantique, 4 rue Alfred Kastler BP.20722, La Chantrerie, Nantes, 44307, France, thomas.yeung@imt-atlantique.fr, Alex Kosgodagan-Dalle Torre In multivariate/high-dimensional statistics, recent attractive approaches include copula-based graphical models such as pair-copula Bayesian networks. Their attractiveness is largely due to the flexibility of copulae to capture dependence. However, very little attention has been given for these models to fit within a full probabilistic framework and for which inference could be performed efficiently. We prove that any k-th order Markov process can be represented as a dynamic pair copula-based Bayesian network. We also show the requirements in order to perform analytical conditioning. We motivate this work through a bridge degradation application and an example focused on Brownian motion. Business School, F37 MBS.East, Manchester, M13 9SS, United Kingdom, L.Xu@Manchester.ac.uk, jian-bo yang The Evidential Reasoning (ER) rule extends Bayes rule for inference with ambiguous data, unknown prior & data with various degrees of reliability. We explain its relationship with Bayes rule & Dempster rule for evidence

63