2016 INFORMS Annual Meeting Program

POSTER SESSION

INFORMS Nashville – 2016

Demand-driven Movie Scheduling In A Multiplex Julia Charlotte Krake, Research Associate, University of Hamburg - Institute for Operations Research, Von-Melle-Park 5, Hamburg, 20146, Germany, julia.krake@uni-hamburg.de A cinema chain is confronted with the weekly problem of creating a movie schedule. Decisions need to be taken about the set of movies, the assignment to screens and the start times. The proposed model improves this decision-making process with possible objectives like the increase of total attendance or revenue. Unittracker: Enhancing Conventional Generation Modeling Resolution In The Regional Energy Deployment System (ReEDS) Model Venkat Krishnan, Engineer, National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, CO, 80401, United States, Venkat.Krishnan@nrel.gov, Jonathan Ho, Kelly Eurek The Regional Energy Deployment System (ReEDS) model is the National Renewable Energy Laboratory’s flagship planning model for projecting the long- term build-out and operation of the U.S. electric power generation and transmission system. In this poster, we explore the effects of model resolution on solution quality and tractability. Specifically, for each of 134 load balancing areas in ReEDS, we increase the resolution of conventional plant thermal efficiencies and examine the consequent impacts on the planning results from ReEDS. Network Simplex Based Algorithm For The Minimum Cost Flow Problem With Linear Interdependencies Adam Rumpf, Illinois Institute of Technology, 10 West 32nd Street, Room 208, Chicago, IL, 60616, United States, arumpf@hawk.iit.edu We consider a generalization of the minimum cost network flow problem in which the flows through certain arcs are bounded by linear functions of the flows through other arcs. This formulation can be used to model interdependent infrastructure systems, for example a subway system whose components require delivery of electrical power from a separate system. We characterize the basis of this problem as a spanning forest plus some supplementary structures, and use these to develop an efficient solution algorithm based on the well-known network simplex algorithm. This is joint work with H. Kaul. Modelling Competitive Equilibrium Prices For Energy And Balancing Capacity In Electricity Markets Involving Non-convexities Andre Ortner, Researcher, Technical University of Vienna, In economic analyses of markets often the dual variables of market clearing equations derived from the optimal solution of cost-benefit optimization models are interpreted as efficient market prices. Whereas in convex (linear) problems the validity of this approach is undisputed it cannot be generalized to problem formulations containing non-convexities. The withholding of spinning reserves in electricity markets are a good example of such cases as costs in these markets are essentially driven by indivisibilities. In this paper we present a novel modelling approach designed to find equilibria in binary games to derive equilibrium prices of self-committed electricity market models. Multi-period Matching Under Relaxed Stability Gusshausstraße 25-29, Vienna, 1040, Austria, ortner@eeg.tuwien.ac.at, Daniel Huppmann Organizations sometimes face challenges in producing assignments of staff to jobs that are good (in the sense that preferences are met), and stable (which promotes high morale and less turnover in staff). In some situations, assignments have to be made more than once, allowing organizations to be more flexible in making assignments and negotiate with workers more efficiently. We extend the stable matching problem to a multi-period setting and consider stability as measured over all periods. We consider notions of relaxed stability to improve the quality of matching by analyzing stability over the entire horizon of the assignment. Stochastic Optimal Power Flow With Forecast Errors And Failures In Communication Basel Alnajjab, Lehigh University, 19 Memorial Drive, West, Bethlehem, PA, 18015, United States, bra212@lehigh.edu, Alberto J Lamadrid, Rick S Blum, Shalinee Kishore, Lawrence V Snyder The role of communication networks in supporting the operation of power grids will become increasingly more critical as we continue to integrate renewable energy sources into power grids, We present a stochastic optimal power flow formulation in which we account for errors in forecasting future load and renewable generation while also considering random failures in the communication network employed to communicate the realized values of the quantities for which we have forecasts. The communication network is also assumed to be employed for the control of loads and generators in the power system. We present results comparing different topologies for the communication network. Zihao Li, Georgia Institute of Technology, 765 Ferst Drive, School of ISyE, Atlanta, GA, 30332-0225, United States, zli66@gatech.edu, Özlem Ergun, Julie L Swann

Reducing Response Categories In Multinomial Regression Brad Price, West Virginia University, College of Business and Economics, PO Box 6025, Morgantown, WV, 26506, United States, brad.price@mail.wvu.edu, Adam Rothman, Charles Geyer In this work we propose penalized likelihood estimators to reduce the number of response categories in multinomial regression. Typically the multinomial model is made simpler by trying to reduce the number of covariates in the model. We instead approach it by combining response categories in situations where a set of covariates does a poor job of differentiating between these categories. An ADMM algorithm is proposed, and convergence properties presented. Tuning parameter selection is also addressed. Reliability And Economic Criteria To Determine Management Policies Of Wind Energy Systems With Storage Cristina M Azcarate, Universidad Publica de Navarra, Dep of Statistics & OR, Campus Arrosadia, Pamplona, Navarra, 31006, Spain, cazcarate@unavarra.es, Fermin Mallor, Pedro Mateo Electrical energy storage systems integrated into a renewable generation system reduce the effects of forecasting errors and enable the determination of management policies. In this work, we propose a management strategy for a wind energy system with storage capacity that integrates tactical and operational decisions in a single stochastic mathematical model. The mathematical model includes economical and reliability criteria, and an updated probabilistic wind speed forecast. A simulation model inspired on a real wind-hydrogen energy system is built to assess the performance of this strategy. Electric Transmission Expansion Considering Property Value Reduction On Routing Juan Andrade, University of Texas at Austin, 4405 Avenue A, Apt 11, Austin, TX, 78751, United States, jandraderam@utexas.edu, Ross Baldick The development of utility scale renewable generation requires new transmission infrastructure, whose proximity to urban areas produces social opposition. This opposition can be quantified as a social cost produced in property value reduction by transmission proximity to population. It is presented a MILP formulation that minimizes costs for generation, and investment and social impact for transmission considering an electrical and a routing networks. An implementation that uses geographical information was developed, and tested with typical IEEE test systems. Detection Of Copyrightable Images From Social Media Feeds Manoj Pooleery, Scopio LLC, 175 Varick St, New York, NY, United States, manoj@scopio.io, Binu Josephi, Jinjin Qin Finding ``original” images-those that can be potentially protected by copyright laws-from social media channels is a challenging problem. The images may contain objectionable/unusable content (spam) and are often modified by users by addition of text, change of texture and color. This paper presents a framework for detection of original images by first filtering out spam, then identifying presence of user generated text and finally augmenting the decision making process by manual curation & verification. Empirical results obtained from Twitter & Instagram feeds suggest that an automated technique for identification of original images with minimal manual intervention can be developed. Application Of Data-driven Analytics To Optimal Decisions Meng-Chen Hsieh, Assistant Professor I, Rider University, 2083 Lawrenceville Rd, SWG 313, Lawrence Township, NJ, 08648, United States, mehsieh@rider.edu, Jeffrey Simonoff, Clifford Hurvich, Avi H Giloni In operation research and management science problems, a traditional approach in deriving optimal decision rules under uncertainty has been to optimize an univariate target quantity while ignoring the presence of auxiliary variables. These auxiliary variables, if used wisely, can provide valuable information on their association with the target variable and thus substantially reduce the target variable’s prediction uncertainty. This work provides guidelines on applying statistical learning to leverage the association between target and auxiliary variables thereby enhancing the efficiency of optimal decisions. Modeling In-Process Machining Data as Spatial Point Clouds vs. Time Series: Research Challenges and Opportunities Mohammed Shafae, PhD Candidate, Virginia Tech, Blacksburg, VA, United States, shafae1@vt.edu, Lee Wells, Jaime Camelio Traditional approaches for analyzing machining process data revolve around representing them as time-series. What tends to be missing is the relationship between the time-series and the part physical dimensions. This research discusses the concept of a novel representation of machining data as spatial point clouds and the corresponding practical advantages.

179