Informs Annual Meeting 2017

WC83

INFORMS Houston – 2017

4 - Data-driven Design and Market Participation of Energy Systems Alexander W. Dowling, University of Notre Dame, Notre Dame, IN, 46556, United States, adowling@nd.edu, Farshud Sorourifar, Jose A. Renteria, Victor M. Zavala We propose a multiscale optimization framework for economic assessment of energy systems (e.g., batteries, industrial demand response, large-scale solar) from historic electricity market prices. Our framework considers both energy and ancillary service transactions in day-ahead and real-time markets. Energy system physics are incorporated as constraints. We analyze over 1 trillion prices from the California electricity markets from 2015 and find strong incentives for fast dynamic flexibility. We show how co-optimizing design decisions (e.g., storage sizing, location) and market participation can reduce payoff periods from over 100 to less than 2 years for utility-scale batteries in CA. 382B Robust Optimization Applications in Queueing and Data Centers Sponsored: Optimization, Optimization Under Uncertainty Sponsored Session Chair: Chaithanya Bandi, c-bandi@kellogg.northwestern.edu 1 - A Robust Queueing Network Analyzer for a Series of Single-server Queues Wei You, Columbia University, 100 Haven Ave, Apt. 18G, New York, NY, 10032, United States, wy2225@columbia.edu, Ward Whitt Whitt and You (2016) developed a new Robust Queueing (RQ) algorithm to expose the impact of dependence on the mean steady-state workload of a single server queue. For the G/GI/1 model, that algorithm requires the index of dispersion for counts (IDC) and the rate of the arrival process. We show how that algorithm can be applied recursively to queues in series by (i) providing systematic ways to calculate or estimate the IDC for the external arrival process and (ii) developing an approximation for the IDC of a departure process. Extensive simulation experiments show that the algorithm is fast and accurate. 2 - Delay Performance of Scheduling Algorithms for Data Center Networks and Input Queued Switches Siva Theja Maguluri, Georgia Institute of Technology, Atlanta, GA, United States, siva.theja@gatech.edu Today’s era of cloud computing is powered by massive data centers hosting servers that are connected by high speed networks. It is therefore desirable to design scheduling algorithms for data packets that have low computational complexity and result in small average packet delays. We consider the scheduling problem in an input-queued switch, which is a good abstraction for a data center network. We present low complexity scheduling algorithms that have optimal queue length (equivalently, delay) behavior in the heavy traffic regime. We also present bounds on the queue length in light traffic. These results are obtained using drift based arguments. 3 - Designing Optimal Priority Policies in Multi Class Queues via Robust Queueing Theory Chaithanya Bandi, 2001 Sheridan Road, Suite 566, Evanston, IL, 60208, United States, chaitu1287@gmail.com We consider the problem of dynamic resource provisioning and use tools from robust queueing theory and multi-stage optimization to present tractable solutions. WC82

various electricity generation and water saving scenarios, including no-hydro and no-nuclear plant and the results, are presented. Modifications to proposed different scenarios have been applied and discussed to meet the practical and reliability constraints. 2 - Aerial Autonomous Drones Optimal Allocation for Incident Detection Houssam Ghandour, Student, Texas A&M.University, 1100 Hensel Drive, College Station, TX, 77840, United States, hmg06@tamu.edu, Luca Quadrifoglio Traffic incidents contribute to 25% of the congestion in the US. Incidents are detected through inefficient methods which heavily depend on operators. Early detection would reduce the impact through faster response. This paper aims to utilize aerial autonomous drones as incident detectors. The drones fly through optimized paths in which they discover incident locations and send footage to the operators who benefit from the fast process. The main objective of the formulated algorithm is to minimize the total number of drones in the network along with optimizing the locations of the charging stations. Moreover, the algorithm integrates the average incident detection time as an input to the model. 3 - Multi-period Stochastic Optimization of a Second Generation Bioethanol Supply Chain Atif Osmani, Assistant Professor, Minnesota State University Moorhead, 1104 7th Avenue South, 211F Hagen Hall, Moorhead, MN, 56563, United States, atif.osmani@mnstate.edu, Jun Zhang, Vinod Lall, Rachel Axness This work proposes a multi-objective optimization model to design a sustainable multi-period second generation bioethanol supply chain under multiple uncertainties. The objective is to simultaneously maximize the economic, environmental, and social performance. The epsilon-constraint method is used to trade-off among the competing objectives and to achieve desired levels of sustainability. A solution approach involving sequential application of a Sample Average Approximation method and Benders decomposition is utilized to solve the optimization model efficiently and effectively. 4 - A Multiple Traveling Salesman Problem Model for Unmanned Aerial Vehicles Routing Ashkan Mirzaee, Graduate Research Assistant, University of Missouri, 1133 Ashland Rd. Apt 1106, Columbia, MO, 65201, United States, amirzaee@mail.missouri.edu, Mohamed Awwad Unmanned Aerial Vehicles (UAV) are gaining momentum in the field of commercial last-mile logistics. UAVs, commonly referred to as drones, can provide flexibility and efficiency for delivery of goods and services through aerial delivery. Finding the shortest delivery route between distribution center, docking stations and customers within the battery life window will considerably extend the drones’ service level. This study considers Multiple Traveling Salesman Problem (mTSP) for UAVs’ routing in the presence of a time window to find optimal set of routes for drones to traverse in a network of vertices that represent a distribution center, docking stations and customers. 5 - Application of Linear Programming in Optimizing the Procurement and Movement of Coal for an Indian Coal-fired Power Generating Company Subrata Mitra, Professor, Indian Institute of Management Calcutta, Diamond Harbour Road, Joka, Kolkata, 700104, India, subrata@iimcal.ac.in, Balram Avittathur In this paper, an application of linear programming (LP) in optimizing the procurement and movement of coal for an Indian coal-fired thermal power generating company is presented. Results show that there is immense potential not only for significant cost savings but also for reduced logistics between different coal source-power plant pairs. The issue of greenhouse gas (GHG) emissions from coal-fired power plants has also been addressed. The trade-off between the optimal total cost and GHG emission targets has been explored. Finally, recommendations and concluding remarks are presented. 6 - Warmstart of Interior Point Methods for Second Order Cone Optimization via Rounding over Optimal Jordan Frames Sertalp Bilal Cay, PhD Candidate, Lehigh University, 200 West Packer Avenue, Department of Industrial and Systems Engineer, Bethlehem, PA, 18015-1582, United States, sertalpbilal@gmail.com, Imre Polik, Tamas Terlaky Interior point methods (IPM) are the most popular approaches to solve Second Order Cone Optimization (SOCO) problems. In this talk, we present a novel warm-start method for IPMs to reduce the number of IPM steps to solve SOCO problems that appear in a Branch and Conic Cut (BCC) tree when solving Mixed Integer Second Order Cone Optimization (MISOCO) problems. Our method exploits the optimal Jordan frame of a subproblem and provides a primal-dual initial point for the self-dual embedding model by solving two auxiliary linear optimization problems. Numerical results on test problems in the CBLIB library show on average around 61% reduction of the IPM iterations for a variety of MISOCO problems.”

WC83

382C Optimization, Linear Programming Contributed Session

Chair: Sertalp Bilal Cay, Lehigh University, 200 West Packer Avenue, Department of Industrial and Systems Engineer, Bethlehem, PA, 18015-1582, United States, sertalpbilal@gmail.com 1 - Water with drawal and Consumption Reduction for Electrical Energy Generation Systems Mohammad Hasan Balali, University of Wisconsin-Milwaukee, Milwaukee, WI, 53211, United States, mbalali@uwm.edu, Narjes Nouri There is an increasing concern over shrinking water resources. Water use in the energy sector primarily occurs in electricity generation. A linear model is developed to minimize water consumption while considering several limitations and restrictions. California has planned to shut down some of its hydro and nuclear plants due to environmental concerns. Studies have been performed for

531