Informs Annual Meeting 2017

MB53

INFORMS Houston – 2017

3 - Organizing for Disaster Management Across Many Districts Barrie Nault, University of Calgary, Calgary, AB, Canada, nault@ucalgary.ca, Hong Guo, Yipeng Liu We examine districts’ choice of whether to centralize in a setting with more than two districts. Districts face interoperability challenges and convex costs in deciding their investment in disaster management assets, where these assets can be shared across districts. We derive conditions for cost sharing that maximize social welfare where some, but not all, districts decide to centralize. 4 - Devising Management Response Strategies in a Competitive Market Mingwen Yang, The University of Texas at Dallas, Richardson, TX, 75080, United States, mxy131030@utdallas.edu, Vijay Mookerjee, Eric Zheng, Hongyu Chen We address a practical question of when a firm should respond to customer reviews in a competitive market, where consumers’ purchasing decisions are influenced by the reviews of the focal and the competing firms. Factoring in competition is essential for a firm to derive an optimal response strategy to effectively manage online WOM and subsequently influence sales. Our study extends the literature by 1) designing an optimal response strategy to online reviews, 2) prescribing the relationship between firm response and sales and 3) considering competition. We develop a Stochastic Differential Equation (SDE) model that captures the data generating processes and estimate the model using real data. 361E Energy Contributed Session Chair: Adam Ng, National University of Singapore, Singapore, isentsa@nus.edu.sg 1 - Welfare Effects of Pricing Schemes for Non-Convex Markets Jacob Mays, Northwestern University, 2145 Sheridan Road, Room C230, Evanston, IL, 60208, United States, jacob.mays@u.northwestern.edu A major goal of competitive electricity markets is encouraging optimal investment in generation capacity, demand response capabilities, and energy efficiency measures. For market-integrated buyers and sellers, these investments are guided by prices announced by system operators. However, the usefulness of these prices has been called into question due to the proliferation of zero-marginal-cost wind and solar resources, the influence of out-of-market supports, and the presence of non-convex technical constraints. This work investigates the influence of price formation decisions on welfare in electricity markets. 2 - Robust Design of Energy Consumption Subsidies Reform Hossein Mirzapour, University of Montreal (HEC School of Business), 920-6275 Northcrest, Montreal, QC, H3S.2N3, Canada, hossein.mirzapour@hec.ca, Amir Ardestani-Jaafari To raise the necessary public support for any energy subsidies reform, a prevalent recommendation is to redistribute its revenue among the most affected households/industries. However, the main challenge to design an effective redistribution mechanism is how to reach those needy entities, particularly in the presence of some uncertain side-effects such as the price inflation of non-energy goods. We study the problem of a social planner. The goal is to minimize the gap between the local subsidized price and its international level, while expecting a minimum purchasing power for the most affected entities. We show the value of a flexible distribution mechanism in such reforms. 3 - Energy and Reserve Co-optimization – Reserve Availability, Lost Opportunity and Uplift Compensation Cost Ivan Pavic, Research Associate, University of Zagreb, Faculty of Electrical Engineering and Computing, Unska 3, Zagreb, 10000, Croatia, ivan.pavic@fer.hr, Hrvoje Pandzic, Yury Dvorkin In a deregulated power systems, energy and reserve are often co-optimized to decrease costs payed by end-users. Generating companies can still encounter monetary losses caused by the provision of reserve. Currently, compensation of these losses is based on ex-post calculations which do not adequately factor real costs and create distortions in market prices. Proposed algorithm run energy and reserve co-optimization with an explicit consideration of two compensation mechanisms, i.e. lost opportunity and uplift payments. The problem is structured as a bi-level and converted to single level using strong duality equation. The case study shows that the proposed model increases market efficiency. MB52

4 - A Computable General Equilibrium Model for Power Interruption Contracts

Lakshmi Palaparambil Dinesh, University of Cincinnati, 601 McAlpin Avenue, Apt 5, Cincinnati, OH, 45220-1539, United States, lakshmi603@gmail.com, Dr. Kenneth Skinner, Dr. Uday Rao Energy demand response programs change the energy consumption behavior of consumers with changes in price and demand. Our supply demand integrated cost minimization model shows that demand response lowers overall costs for suppliers and promotes energy savings for customers. In addition to the cost savings for the supplier, we also look at the societal welfare using a Computable General Equilibrium (CGE) model that will aid the energy policy regulators. This model helps to understand the generation costs and activity levels at equilibrium, the tax rates needed to float demand response and the amount of capital that can be saved by demand response. 5 - Energy-economic Recovery Resilience with Input-output Linear Programming Models Adam Ng, National University of Singapore, 10 Kent Ridge Crescent, Singapore, 119260, Singapore, isentsa@nus.edu.sg, Peijun He, Bin Su We develop a novel input-output linear programming model to study the energy- economic resilience of an economy by analyzing the relationships between energy disruption, impacts on production and demands, and post-disruption recovery efforts. The model evaluates the minimum level of recovery investments required to restore production levels so that economic impacts are acceptable over the post-disruption duration. The optimization model is solved using a cutting plane method which involves computing a small sequence of mixed integer programs. A case study using China input-output data is performed to demonstrate the model’s ability to uncover critical inter-sectoral dependencies. 361F Advancements in Detecting and Modeling Traffic Flow Sponsored: TSL, Intelligent Transportation Systems (ITS) Sponsored Session Chair: Wei-Hua Lin, University of Arizona, Tucson, AZ, 100190, United States, whlin@email.arizona.edu 1 - Transportation Network Traffic Detector Error Estimation Han Yang, UCDavis, 1 Shield Ave, Davis, CA, 95616, United States, cnhyang@ucdavis.edu, Yudi Yang, Yueyue Fan One of the main challenges regarding traffic sensor data is how to assess its quality, as empirical works have shown inconsistency from a large portion of road sensors even they are not labelled as “broken” by major traffic data providers. This paper aims to address the problem of identifying the malfunctioned sensors and correcting systematic error using a network modelling based statistical approach. The measurement error is defined conditionally on the true flow which is unobservable. Network flow balance is assumed to be preserved among traffic flows. We provide two different estimators based on moment matching and maximum likelihood respectively to estimate the systematic error ratio. 2 - Multi-class Support Vector Machines Based Traffic Pattern Detection Sayyid Vaqar, King Fahd University of Petroleum and Minerals, P.O. Box 983, Dhahran, 312600, Saudi Arabia, savaqar@kfupm.edu.sa The traffic jam detection problem is formulated as a pattern recognition problem with scarce data utilizing Principal Component Analysis and Multiclass Support Vector Machines. The collected data is viewed as a snapshot of traffic flow condition on the road. For each time step this data will provide snapshots of driving conditions for the node of interest. With proper analysis these snapshots can be used in predicting traffic jam conditions on the freeways.The model is tested under assumption of different levels of vehicles equipped with Mobile Ad hoc network communication capability. The trained MSVMs are then used to predict traffic conditions for unlabeled test data. 3 - A Generalized Theory of Mixed Traffic Flow Based on Behavioral Assumptions Jia Li, Research Associate, University of Texas at Austin, 7630 Wood Hollow Dr, Apt 251, Austin, TX, 78731, United States, lijia.ust@gmail.com This research proposes a generalized theory of mixed traffic flow. Most existing mixed traffic flow models are descriptive in nature and lack sound behavioral interpretations. Towards filling this gap, a theory is developed from natural behavioral assumptions and explicitly addresses factors such as lane and information access. The stationary and dynamic behaviors of mixed traffic flow in several application scenarios, including mixed connected/automated and regular traffic, will be discussed. MB53

183