Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TD50

2 - How Sampling and Averaging Historical Solar and Wind Data Distorts Generation Expansion Planning Cynthia D. Bothwell, Johns Hopkins University, Baltimore, MD, 21286, United States, Benjamin Field Hobbs Industry practice includes developing typical load shapes to forecast future system requirements. As weather-dependent wind and solar generation increase, the prediction of their contribution to the system has also relied on averaging techniques that disregard correlations of load and renewable output. Using long- term optimized expansion planning including hourly dispatch, this work shows the pitfalls of data averaging ignoring dependencies and inadequate sample lengths, as quantified by cost increases and system reliability distortions. Methods to minimize distortions are introduced. 3 - Robust Optimal Control of Wind Farm with PHEVs Participating in Energy Market Xiaojie Wang, University of Florida, Gainesville, FL, United States, Yongpei Guan Previous studies suggest that plug-in hybrid electric vehicles (PHEVs), serving as electricity storage units, offer a perfect supplement to intermittent renewable energy. In this study, we manage a profit driven wind station with PHEVs as storage devices, where we can either sell the wind energy directly to the market or use it to charge the PHEV batteries and storage the electricity for future sales. We consider an innovative model and provide a robust solution considering price and renewable generation uncertainties. Our numerical results show the advantages of our proposed model. n TD50 North Bldg 231A Mixed-Integer Nonlinear Programming in Power Systems Sponsored: Optimization/Integer and Discrete Optimization Sponsored Session Chair: Burak Kocuk, Sabanci University, Turkey 1 - Lazy Cut Generation for Semidefinite Programming Hassan Lionel Hijazi, Los Alamos National Laboratory, P.O. Box 1663, MS B284, Los Alamos, NM, 87545, United States In this talk, we will present the theory and the implementation of efficient cut generation algorithms that capture the strength of semidefinite programming relaxations and their application in ACOPF. 2 - The Value of Optimization Based Bound Tightening for Power Network Optimization Carleton Coffrin, Los Alamos National Laboratory, Los Alamos National Laboratory, los Alamos, NM, United States Convexification is a fundamental technique in (mixed-integer) nonlinear optimization and many convex relaxations are parametrized by variable bounds, i.e., the tighter the bounds, the stronger the relaxations. This work investigates how optimization based bound tightening can improve convex relaxations for power network optimization. In particular, this work shows that the Quadratic Convex relaxation of AC power flow, enhanced by bound tightening, results in very small variable domains in the seminal Optimal Power Flow problem. Leveraging these improved bounds, experimental results demonstrate root-node optimality gaps that are less than one percent in the majority of test cases. 3 - Strengthening Optimal Power Flow Relaxations with Reference Bus Constraints and Bounds Tightening Michael Lee Bynum, Sandia National Laboratories, Albuquerque, NM, 87123, United States, Anya Castillo, Jean-Paul Watson, Carl Laird Math programming is a powerful tool frequently used to achieve reliable and efficient operation of the electric grid. We present recent advances toward including high fidelity, nonconvex transmission models in these optimization formulations. In particular, we show that existing relaxations of the alternating current optimal power flow (ACOPF) problem can be further tightened when we combine the McCormick relaxation with reference bus constraints and an appropriate bounds tightening procedure. With these strategies, we are able to reduce the optimality gap to less than 0.1% on all but 5 NESTA test cases with up to 300 buses by performing optimality based bounds tightening alone.

n TD51 North Bldg 231B Scheduling Contributed Session Chair: William Ellegood, Sam Houston State University, Huntsville, TX, United States 1 - On Submodular Search and Machine Scheduling Thomas Lidbetter, Assistant Professor, Rutgers Business School, 1 Washington Park, Newark, NJ, 07102, United States, Robbert Fokkink, László Vogh We explore the connections between search theory and machine scheduling by studying an ordering problem proposed by Pisaruk (1992). A target is hidden in one of many locations, which must be searched in some order. The cost of searching an initial segment A of a given ordering is f(A) and the probability the target is hidden in A is g(A), where f is a non-decreasing submodular function and g is a non-decreasing supermodular function. We want the ordering that finds the target in minimal expected cost. The problem is NP-hard and we present a new 2-approximation algorithm that generalizes well-known results in scheduling theory and establishes new ones. We also consider a game theoretic version of the problem. 2 - Simultaneous Scheduling of Machines and Vehicles with Alternative Machines in Flexible Manufacturing Systems Dalila B. M. M Fontes, INESC TEC, Rua Dr Roberto Frias, Porto, 4200-465, Portugal, Mahdi Homayouni This work proposes a mathematical programming model for jointly scheduling production and transport in Flexible Manufacturing Systems considering alternative processing routes. Although production scheduling and transport scheduling have been vastly researched, most of the works address them independently. In addition, the few that consider their simultaneous scheduling assume job routes as an input, i.e., machine-job allocation is previously determined; however, in FMSs this is an important source of flexibility that should not be ignored. Results are reported for small-sized instances. 3 - Multi-period Lot-sizing with Supplier Selection: Complexity and Algorithms Guohua Wan, Professor, Shanghai Jiao Tong University, 1954 Huashan Road, Antai College of Economics & Management, Shanghai, 200030, China, Meichun Lin, Woonghee Tim Huh We consider a multi-period lot-sizing problem with multiple products and multiple suppliers. The objective is to determine the order quantities for each product from each supplier in each period to minimize the total cost over the planning horizon, including supplier-dependent fixed costs, variable purchase costs and product-dependent holding costs. We show the intractability of the problem in general and study three polynomial-time solvable cases of the problem. For these three cases, we analyze the structural properties, and propose dynamic program solution algorithms. We also develop a heuristic algorithm for the general case and computationally show that it works well. 4 - Utility Based Priority Scheduling Rules in Dynamic Job Shops with Customer Balking William Ellegood, Sam Houston State University, Huntsville, TX, 77341, United States, Kevin D. Sweeney, Stanislaus Solomon, Mitch Millstein This research examines a particular class of job shops where individual jobs have the option to forgo using the job shop (known as balking) due to expected poor levels of service as a result of system congestion. Using a dynamic job shop with heterogeneous jobs, we use the expected utility of a job to predict the balking behavior of potential job shop users. We further use the metric of expected utility to derive and test new priority dispatching rules for job shops that face the possibility of customer balking. These new rules are then compared to other priority scheduling rules based on processing time. 5 - Scheduling And Sequencing To A Two-stage Stochastic Service System With Multiple Servers Yang Zhan, Shanghai Jiao Tong University, 1954 Huashan Road, Antai College of Economics and Management, Shanghai, 200030, China, Siqian Shen, Guohua Wan In this paper, motivated by the application in the surgical suite, we consider the scheduling and sequencing problem in a two-stage stochastic service system with multiple servers on each stage. The processing times of jobs on both stages are stochastic and there is no buffer between stages. Our goal to optimize the usage of servers, the assignment of jobs-to-server and the sequence of jobs with the objective of minimizing the total of opening cost and overtime cost. We investigate a stochastic optimization framework and a decomposition algorithm to address the problem. Finally, we apply the model and algorithm to surgical scheduling.

364