2016 INFORMS Annual Meeting Program

WC39

INFORMS Nashville – 2016

4 - Education Engineering [SIC] Kingsley Anthony Reeves, University of South Florida, Tampa, FL, Contact: reeves@usf.edu The craft of engineering has been applied to numerous industries, especially the manufacturing-oriented industries such as automotive, pharmaceuticals, and petroleum. In more recent history we have seen the application of engineering approaches to the service sector; examples include financial services and health- care. A notable exception (and area of great need) is the application of engineer- ing to education. While there is currently significant momentum in the growth of engineering education programs across the nation, this paper calls for the cre- ation of education engineering as a discipline and explores the opportunities that exist particularly for industrial engineering. 5 - Operational Performance Of Retail Stores Andreas Holzapfel, Catholic University of Eichstaett-ingolstadt, Ingolstadt, Germany, andreas.holzapfel@ku.de, Heinrich Kuhn2, Michael Sternbeck We study which factors drive instore operations efficiency. For this purpose we develop explanatory models that quantify the impact of operational circum- stances on working hours required in the stores and financial performance. The results are valuable input for store and logistics planning as well as for staff assignment planning. WC39 207A-MCC Joint Session APS/MSOM: Service Systems in Applied Probability II Sponsored: Applied Probability/MSOM Sponsored Session Chair: Jing Dong, Northwestern University, Evanston, Evanston, IL, United States, jing.dong@northwestern.edu 1 - When The Past Does Not Predict The Future: Delay Announcements With Customer Priorities Rouba Ibrahim, University College London, rouba.ibrahim@ucl.ac.uk, Mor Armony, Achal Bassamboo Motivated by the problem of making delay announcements, we study the accuracy of announcements based on the history of delays in a system with multiple customer classes and a priority service discipline. We present ways of exploiting this historical information to design new and improved announcements. 2 - Managing Overstaying Electric Vehicles In Park-and-charge Facilities Ragavendran Gopalakrishnan, Research Scientist, Xerox Research Centre India, Bangalore, India, Ragavendran.Gopalakrishnan@xerox.com, Arpita Biswas, Partha Dutta With the increase in adoption of Electric Vehicles (EVs), ensuring proper utilization of the charging infrastructure is a key emerging challenge. Overstaying by EVs after charging is complete can be discouraged by imposing penalties, but the upfront uncertainties in parking and charging durations render higher penalties riskier, which might turn prospective users away, leading to decreased utilization (and revenue). We develop a framework that integrates models for realistic user behavior into queueing dynamics to locate the optimal penalty from the points of view of utilization and revenue, and discover a surprising alignment between the “green” objective and the “commercial” objective. 3 - The Superposition-traffic Game Harsha Honnappa, Purdue, 315 N. Grant St., West Lafayette, IN, 47906, United States, honnappa@purdue.edu, Ashish R. Hota, Shreyas Sundaram Motivated by ride-sharing and online-platform systems, we consider a model of a single-server service system where a finite number of traffic sources compete for service. The presence of a large, but finite, number of traffic sources is assumed to have positive network effects, but also causes an increase in congestion. The goal of the sources is to choose the traffic rate, trading-off these two effects. We present an analysis of the generalized Nash equilibrium (GNE), and discuss implications on pricing and mechanism design for such service systems. 4 - Finite-size Effects In Critically Dimensioned Emergency Departments Britt Mathijsen, Eindhoven University of Technology, PO Box 513,

system exhibits quality-and-efficiency-driven (QED) type behavior as it grows large, and approximate its performance through a fixed-point method. Building upon the asymptotic results, we ultimately propose a dimensioning scheme for the number of nurses and beds necessary to ensure good quality of care in both stationary and time-varying environments.

WC40 207B-MCC Queueing Systems Sponsored: Applied Probability Sponsored Session

Chair: Guodong Pang, Penn State University, University Park, University Park, PA, 16801, United States, gup3@psu.edu 1 - Parameter Uncertainty In Naor’s Model John Hasenbein, University of Texas-Austin, jhas@mail.utexas.edu, Ying Chen We examine the classical Naor’s model when the arrival rate is not known with certainty by either the system controller or the customers. Rather, only the arrival rate distribution is known. We analyze the system in the observable and unobservable queue length regimes from the point of view of individuals, a social optimizer, and a revenue maximizing firm. 2 - Strong Approximations For General Time Varying Queues Jamol Pender, Cornell University, jamol.pender@gmail.com We present a novel methodology for approximating the queue length distributions of non-Markovian and time varying queueing systems. The first step is to approximate the general distributions with phase type distributions and second step is to use strong approximations to construct fluid and diffusion limits for the phase type queueing process. We show that our approximations are quite accurate in a variety of parameter settings. 3 - Pull-based Load Distribution Among Heterogeneous Parallel Servers: The Case Of Multiple Routers Aleksandr Stolyar, Lehigh University, sasha.stolyar@gmail.com We consider a heterogeneous service system, consisting of several large server pools and multiple ‘routers’. Each router receives equal fraction of the customer arrival flow, and assigns each customer to a server immediately upon arrival. The asymptotic regime is considered such that the total arrival rate and pool sizes scale to infinity simultaneously, while the system load remains subcritical. We introduce a ‘multi-router’ version of the ‘pull-based’ routing scheme and prove that, under this scheme and certain assumptions, both waiting times and blocking probabilities asymptotically vanish. 4 - The Method Of Chaining For Many Server Queues Yuhang Zhou, Penn State University, YXZ197@psu.edu, Guodong Pang We discuss how the method of chaining can be applied to prove two-parameter process limits for many server queues. It provides useful maximal inequalities for two-parameter processes. The method is universal for models with general i.i.d. and dependent service times, and with general time-varying service times (e.g., arrival dependent services or entering-service-time dependent services). WC41 207C-MCC Real Options Sponsored: Financial Services Sponsored Session Chair: Kuno Huisman, Tilburg University, Best, Netherlands, kuno.huisman@gmail.com 1 - Predatory Pricing Under Uncertainty: Revisiting The Deep Pocket Argument Maria Lavrutich, Tilburg University, mlavrutich@gmail.com In this paper we develop a stochastic model of predatory pricing. When profits evolve stochastically, a negative demand shock can lead to bankruptcy for firms, that cannot immediately raise external capital. This creates incentives for market incumbents to use predatory pricing strategies in order to keep new players out of the industry. We show that firms may use a large cash reserve as a war chest to initiate a price war that could drive the opponent out of the market. Because of uncertainty the new player may wish to take a chance and enter based on the probability of success. Therefore, the realized market structure may vary for different sample paths of the stochastic process.

MetaForum 4.086, Eindhoven, 5600MB, Netherlands, b.w.j.mathijsen@tue.nl, Johan van Leeuwaarden, Foekje Sloothaak

Motivated by the desire to determine staffing levels in an emergency department, we study a queueing model in which patients alternate between being in need of direct care from a nurse and being stable, while the total number of patients present in the ED is limited. We identify a two-fold scaling policy for which the

440