Title: Design and Admission Control of a Stochastic Overflow Loss Network
Speaker: Canan Pehlivan
Date/Time: February 17, 2016 – 13:40
Place: FENS G029
Abstract: This study addresses the problem of rejections in a stochastic loss service network and proposes some methodologies to improve the service level in the network. From a strategic perspective, first a dynamic location and capacity planning problem is studied. We propose a multi-period, multi-facility, multi-service nonlinear optimization model in order to ensure a minimum desired customer acceptance rate in each service. Various linearization approaches are developed using a point-wise representation and structural properties of these linearization models are proved. The models are applied to healthcare network of Hauts-de-Seine in Paris, France. Second, from an operational perspective, we address the admission control problem in the overflow loss network. We consider a multi-server overflow loss queueing network where rejected customers are not lost but overflow to other stations. A station may receive several incoming overflow streams from other stations. Our objective is to find the optimum admission control policy for each incoming stream at each station. We adopt a research strategy that evolves from simpler networks to more complicated ones. First, we consider a 1-station multiple-arrival loss system whose results form a fundamental basis for more complicated networks. Second, we consider a 2-station overflow loss network where customers of one station can overflow to the other station and vice versa. For this system, we establish the existence of an optimal threshold admission policy. Last, we consider a general N-station overflow loss network (N>2) and propose a near-optimal local control policy and a linearized mathematical model that is proved to compute a tight upper-bound. We assess the performance of the proposed local control policy on several numerical studies from different application areas including emergency healthcare networks and telecommunication networks.
BIO: Canan Pehlivan is a visiting professor at the Faculty of Engineering and Natural Sciences, Industrial Engineering Program, Sabancı University. Prior to Sabancı University, she was an assistant professor at Department of Automation and Production in Ecole des Mines de Nantes, France from 2013 to 2014. She holds a Ph.D. in Operations Research from Ecole des Mines de Saint-Etienne, an M.S. degree in Industrial Engineering from Middle East Technical University and a B.S. degree in Industrial Engineering from Bilkent University. Her research interest includes analytical modelling, optimization and performance evaluation in stochastic systems and their applications mainly on healthcare and sustainable energy systems
Contact: Barış Balcıoğlu