2016 INFORMS Annual Meeting Program

TB05

INFORMS Nashville – 2016

4 - Changes In Community State Types Reflect Major Shifts In The Human Microbiome Paul Brooks, Virginia Commonwealth Univ, jpbrooks@vcu.edu The human microbiome consists of the micro-organisms that reside in various body habitats. We present a re-analysis of five sequencing-based surveys of the vaginal microbiome with repeated measures. Our goal is to understand the expected persistence in a state, the rate of state changes, our ability to predict an upcoming change, and predictors of upcoming changes. Our results provide insight into microbiome dynamics. TB03 101C-MCC Marketing Strategies of Entertainment Products Invited: Entertainment Analytics Invited Session Chair: Yong Liu, University of Arizona, 1130 E Helen Street, Tucson, AZ, 85721, United States, yoliu@eller.arizona.edu 1 - Pre-launch Analysis Of Competitive Dynamics Natasha Zhang, University of Virginia, ynf8a@comm.virginia.edu, Fang Wu, Vithala R Rao The market structure literature has focused on mature products and post-launch competition. Yet many industries such as entertainment see frequent product introductions with short lifecycles; and need to identify dynamic market structure pre-launch to guide marketing mix decisions largely made pre-launch. However, such identification is almost impossible because of lack of sales data and dynamic competition resulting from entries, exits, and changes in preannounced release timing. The authors thus propose a dynamic market structure model to capture the evolution of competitive sets and brand positioning, calibrated on a film prediction market. 2 - The Choice And Effects Of Movie Previews In Video-on-Demand Services A key strategy that the Video-on-demand (VOD) providers use to help consumers choose among numerous programs is the provision of program reviews. In this paper we investigate when consumers choose to watch previews and how preview watching may influence movie choice. Our research context are the on- demand movies in the Internet Protocol TV (IPTV) systems from a large Chinese VOD service provider. We find that the movie attributes, household characteristics, and time of the week significantly affect consumer choice for previews. At the same time, the ratio of preview watching time to the length of the preview program has a significant impact on the subsequent movie choice. TB04 101D-MCC Developing Country Electricity Systems Sponsored: Energy, Natural Res & the Environment, Energy I Electricity Sponsored Session Chair: Valerie Thomas, Georgia Tech, 765 Ferst Drive, Atlanta, GA, 30332, United States, vthomas@isye.gatech.edu 1 - Electricity Development In Africa - A Multi-objective Optimization Approach Amelia Musselman, Georgia Institute of Technology, Atlanta, GA, United States, amusselman@gatech.edu, Dima Nazzal, Valerie Thomas Many people across Africa are without sufficient access to electricity. The unavailability and unreliability of energy resources in Africa contribute to developmental challenges in many areas including business, education, and healthcare. We develop a mixed integer optimization model for power generation and transmission system expansion planning in Africa. In addition to solving the least cost optimization, we consider optimal use of a fixed electrification budget when demand cannot fully be met. We test our model on the Democratic Republic of Congo, Uganda, Tanzania, Rwanda, and Burundi. 2 - Can Developing Countries Leapfrog The Centralized Electrification Paradigm? Todd Levin, Argonne National Laboratory, tlevin@anl.gov Due to the decreasing costs of decentralized technologies, centralized power systems are no longer a necessary condition of universal electricity access. Developing countries with less developed infrastructures may be able to adopt these new technologies more quickly. We determine the electricity consumption levels at which the costs of centralized and decentralized electrification are Yong Liu, University of Arizona, Tucson, AZ, United States, yoliu@eller.arizona.edu, Zhen Chen, Madhu Viswanathan, Hongtao Pan, Pei Huang

equivalent in three African regions. We then calculate capital costs necessary for distributed technologies to cost-effectively provide each of five tiers of energy access, as defined by the UN SE4All Framework. 3 - Analysis of a Wind-Hydro Storage System in Kenya Maureen Murage, Cornell University, Ithaca, NY, Contact: mwm88@cornell.edu Kenya has set a target of increasing the wind power capacity from 1% to nearly 15% of total generation capacity. The Lake Turkana Wind Power (LTWP) wind farm will be the largest installation, contributing to half of this total. In this presentation, we will discuss the reliability impact of LTWP project on the Kenyan Power system. The results indicate that the addition of wind power into the system only marginally improves the reliability of the power system, though other implementation strategies may improve this outcome. TB05 101E-MCC Optimal Power Flow Sponsored: Energy, Natural Res & the Environment, Energy I Electricity Sponsored Session Chair: Andy Sun, Georgia Institute of Technology, Atlanta, GA, United States, andy.sun@isye.gatech.edu 1 - Visualizing The Feasible Spaces Of Optimal Power Flow Problems And Their Convex Relaxations Daniel K Molzahn, Argonne National Laboratory, dan.molzahn@gmail.com Optimal power flow (OPF) problems can have non-convex feasible spaces. Visualizing these feasible spaces can aid in understanding these non-convexities. Work towards developing such visualizations uses the Numerical Polynomial Homotopy Continuation (NPHC) algorithm, which is guaranteed to find all power flow solutions. By discretizing the inequality constraints, repeated power flow solutions using NPHC enables calculation of the feasible spaces for small OPF problems. This presentation illustrates the feasible spaces of several challenging OPF problems and various convex relaxations. 2 - Global Optimization Techniques For The Optimal Power Flow Problem Burak Kocuk, Carnegie Mellon University, Pittsburgh, PA, United States, bkocuk@andrew.cmu.edu, Santanu S. Dey, Andy Sun In this study, we aim to solve the Optimal Power Flow problem using global optimization techniques. Our analysis starts with an equivalent SDP formulation of the problem with additional nonconvex minor constraints. We propose several convexification approaches to deal with these nonconvexities via cutting planes and envelopes. We make efficient use of bound tightening to stregthen the convex relaxations. We also develop an SOCP-based spatial branch-and-cut algorithm. Our approach is successful in proving small optimality gaps for challenging power systems instances from the literature. 3 - Convex Cuts For Optimal Power Flow Global optimality and feasibility guarantees are highly desirable outcomes for problems arising in application areas with critical infrastructures. These include energy, transportation, telecommunication and cyber-security systems. A global approach is necessary to prove that the underlying model is infeasible or that the provided solution is optimal. Unfortunately, off-the-shelf global optimization tools are unable to scale up to real-world size problems. This presentation will cover recent results on the generation of valid convex cuts for nonconvex quadratically constrained programs, focusing on key application problems such as the Optimal Power Flow. 4 - Convexification Of The Power System State Estimation Problem Yu Zhang, University of California, Berkeley, Berkeley, CA, United States, yuzhang49@berkeley.edu, Ramtin Madani, Javad Lavaei This presentation deals with the power flow (PF) and power system state estimation (PSSE) problems. The PF problem is cast as an optimization problem by adding a well-designed quadratic objective. It is shown that with a suitable set of measurements and under mild angle conditions, the semidefinite programming (SDP) relaxation can recover the true solution. Capitalizing on this result, a penalized SDP using the weighted least absolute value data fitting cost is tailored for the PSSE. The optimal SDP solution has a dominant rank-one component formed by lifting the true state. An upper bound on the estimation error is also derived, which depends on the noise power. Hassan Hijazi, Australian National University, Canberra, Australia, hassan.hijazi@anu.edu.au, Carleton Coffrin, Pascal Van Hentenryck

265