2016 INFORMS Annual Meeting Program

MC11

INFORMS Nashville – 2016

MC10 103C-MCC Food, Energy, Water and Extreme Events Sponsored: Energy, Natural Res & the Environment, Energy II Other Sponsored Session Chair: Sauleh Ahmad Siddiqui, Johns Hopkins University, 3400 N. Charles St., Latrobe Hall 205, Baltimore, MD, 21218, United States, siddiqui@jhu.edu Co-Chair: Craig Bakkar, Johns Hopkins University, 3400 N. Charles St., Baltimore, MD, 21218, United States, cbakker2@jhu.edu 1 - Mixed Complementarity Modelling In Food Systems Sauleh Ahmad Siddiqui, Johns Hopkins University, 3400 N. Charles St., Latrobe Hall 205, Baltimore, MD, 21218, United States, siddiqui@jhu.edu, Craig Bakkar We are developing a new food systems model investigate the impacts of climate change. This model is designed to build upon past climate-food research by modelling the entire food supply chain, connecting more fully with energy and water sector models, and capturing shock propagation. Our model uses a microeconomic Mixed Complementarity Problem (MCP) formulation and a new MCP decomposition method that both reduces wall clock time to solution and increases the size of solvable problems. 2 - Post-disaster Distribution Systems Restoration Yushi Tan, University of Washington, 185 Stevens Way, Seattle, WA, 98195, United States, ystan@uw.edu, Feng Qiu We investigate the problem of repairing a distribution network after a natural disaster. Such a post-disaster restoration is identified as an NP-hard variant of a job scheduling problem, in which crew are dispatched to repair the damaged lines in a way that minimizes the total harm caused by the outages. We implement a time-indexed ILP formulation with valid inequalities as a benchmark. Two practical methods are also proposed to solve the problem: a conversion algorithm and a linear-relaxation-based list scheduling algorithm. Worst case bounds are analyzed for both algorithms. Numerical results validate the effectiveness of the proposed methods. 3 - Nonlinear Optimization Of Water Supply Network Pumping Plans In A Deregulated Electricity Market Maxime Fender, Optimization Consultant, Artelys, 2001 Boulevard Robert-Bourassa, Suite 1700, Montreal, QC, H3A 2A6, Canada, maxime.fender@artelys.com Within the context of deregulated electricity markets, electro-intensive companies are encouraged to optimize their electricity consumption patterns. This talk presents a study led by Artelys showing how a water distribution network operator, facing varying electricity wholesale market prices, can minimize its water pumping costs. This MINLP (nonlinear pressure loss constraints in pipes / OnOff pumps binary variables) has been solved using an iterative process composed of a linear relaxation and a continuous nonlinear optimization solved respectively by FICO Xpress-Optimizer and Artelys Knitro.

4 - Balancing Diagnostic And Resolution Efforts In A Nonprofit Service Delivery Organization Priyank Arora, Georgia Institute of Technology, 800 W Peachtree St NW, Atlanta, GA, 30308, United States, Priyank.Arora@scheller.gatech.edu, Morvarid Rahmani, Karthik Ramachandran This paper studies service design of a nonprofit organization (NPO) that aims to maximize overall utility delivered to its clients. We examine how the NPO should balance levels of effort between diagnostic and resolution stages of its service delivery, in the presence of client heterogeneity and resource constraint. Our analytical model is based on secondary data collected from an NPO working towards empowerment of victims of domestic violence. MC09 103B-MCC Nonlinear Optimization Problems for Power Systems Invited: Energy Systems Management Invited Session Chair: Javad Lavaei, University of California, Berkeley, 4121 Etcheverry, Berkeley, CA, 94720, United States, lavaei@berkeley.edu 1 - Power System State Estimation With A Limited Number Of Measurements Ramtin Madani, University of California, Berkeley, Berkeley, CA, United States, ramtin.madani@berkeley.edu, Javad Lavaei, Ross Baldick This work is concerned with the power system state estimation (PSSE) problem, which aims to find the unknown operating point of a power network based on a given set of measurements. We develop a set of convex programs with the property that they all solve the non-convex PSSE problem in the case of noiseless measurements if the voltage angles are relatively small. This result is then extended to a general PSSE problem with noisy measurements, and an upper bound on the estimation error is derived. The objective function of each convex program has two terms to account for the non-convexity of the power flow equations and estimate the noise levels. The proposed technique is demonstrated on a 9000-bus network. 2 - Stochastic Unit Commitment With Topology Control Recourse For Renewables Integration Jiaying Shi, University of California, Berkeley, Berkeley, CA, 94720, United States, shijy07@berkeley.edu, Shmuel S Oren We propose a two-stage stochastic unit commitment formulation with topology control recourse decisions for power systems with renewables integration. We investigate applying progressive hedging algorithm to solve this problem. Preliminary test results on both IEEE 118 system and central European system show that such capability of controlling the system configuration actively through switching transmission lines can help improve the efficiency of unit commitment. 3 - A Strong Semidefinite Programming Relaxation Of The Unit Commitment Problem Javad Lavaei, University of California-Berkeley, Berkeley, CA, United States, lavaei@berkeley.edu, Morteza Ashraphijuo, Salar Fattahi, Alper Atamturk The unit commitment (UC) problem aims to find an optimal schedule of generating units subject to the demand and operating constraints for an electricity grid. We develop a strengthened semidefinite program (SDP) based on first deriving certain valid quadratic constraints and then relaxing them to linear matrix inequalities. These valid inequalities are obtained by the multiplication of the linear constraints of the UC problem. The performance of the proposed convex relaxation is evaluated on several hard instances of the UC problem. By solving a single convex problem, globally optimal integer solutions are obtained in most of the experiments that we have conducted. 4 - Optimal Distributed Control Of Power Systems Salar Fattahi, University of California, Berkeley, Berkeley, CA, United States, fattahi@berkeley.edu, Abdulrahman Kalbat, Javad Lavaei This talk is concerned with the optimal distributed control of power systems under input disturbance and measurement noise. This optimal control problem is highly nonlinear and NP-hard. In this work, we design an efficient computational method that transforms the optimal centralized controller to a near-optimal distributed controller. We also study how the connectivity of its underlying communication network affects the optimal performance of the stochastic power system under control. As a case study, the proposed technique is used to design a distributed primary frequency controller for the IEEE 39-Bus New England test System.

MC11 104A-MCC

Network Vulnerability and Criticality Sponsored: Optimization, Network Optimization Sponsored Session

Chair: Chrysafis Vogiatzis, Assistant Professor, North Dakota State University, CIE 202K, P.O. Box 6050, Fargo, ND, 58108, United States, chrysafis.vogiatzis@ndsu.edu 1 - Critical Arcs Detection In Influence Networks Colin P. Gillen, University of Pittsburgh, Pittsburgh, PA, 15261, United States, cpg12@pitt.edu, Alexander Veremyev, Oleg A Prokopyev, Eduardo Pasiliao The influence maximization (MAXINF) problem chooses an optimal set of seed nodes to maximize the propagation of influence (cascading behavior) in a network. Given a set of seed nodes and the linear threshold model, our work proposes to determine which edges - e.g. relationships in a social network - are most critical to the influence propagation process. NP-completeness of the problem is proved. Naïve time-dependent and time-independent mixed-integer programming (MIP) models are stated. An improved MIP-based exact algorithm and a heuristic are proposed, and computational results presented.

185