2015 Informs Annual Meeting

WC23

INFORMS Philadelphia – 2015

WC21 21-Franklin 11, Marriott Healthcare Capacity Planning Models Sponsor: Health Applications Sponsored Session

expected delays, would she drive to the service facility? A common assumption in queueing models is that there is no time gap between observing the queue and joining it. We challenge it by allowing the queue to evolve while the customer is on her way. We show that as opposed to intuition, customers who balk when the queue is mid-congested departs towards it when it is highly congested. 2 - From Product Form Queues to Queue Decomposition: The State-dependent Mn/gn/1 Example Opher Baron, University of Toronto, 105 St. George St, Toronto, ON, Canada, opher.baron@rotman.utoronto.ca, Hossein Abouee Mehrizi In the analysis of Product Form compatible queueing systems we decomposed them into subsystems that can be analyzed independently of each other. Queueing decomposition (QD) was used to approximate solutions for other queueing networks. We formulate QD and discuss its implementation for exact analysis. We demonstrate QD for many systems focusing on Mn/Gn/1 where arrivals and service times are state dependent and service rates can change at arrivals and departures is analyzed. Vahid Sarhangian, University of Toronto, 105 St. George Street, Toronto, ON, Canada, vahid.sarhangian11@rotman.utoronto.ca, Philipp Afeche The literature on the economics of queues predominantly focuses on the queue- joining decisions of customers and ignores subsequent abandonment decisions. Such abandonment behavior is particularly important in priority queues, which are quite prevalent in practice. We study the equilibrium joining and abandonment behavior of utility-maximizing customers in the context of an observable two-class priority queue and identify important pricing implications. WC23 23-Franklin 13, Marriott Control of Queues Sponsor: Applied Probability Sponsored Session Chair: Douglas Down, McMaster University, 1280 Main Street West, Hamilton, Canada, downd@mcmaster.ca 1 - ATM Replenishment Scheduling Yu Zhang, UNC Chapel Hill, B04 Hanes Hall, UNC Campus, Chapel Hill, NC, 27599, United States of America, yuzhang@live.unc.edu, Vidyadhar Kulkarni We develop an ATM replenishment policy for a bank that operates multiple ATMs with an aim to minimize the cost of stock-outs and replenishments, taking into account the economies of scale involved in replenishing multiple ATMs simultaneously. We present the structure of the optimal strategy that minimizes the long run cost per unit time and study a heuristic policy which is easy to implement. 2 - An Emergency Department Resource Allocation Model for Patients of Deteriorating Health Mark Lewis, Professor, Cornell University, Rhodes Hall, Ithaca, NY, 14853, United States of America, mark.lewis@cornell.edu, Douglas Down, Carri Chan We consider the allocation of medical service providers (MSPs) when patients health continues to deteriorate while waiting for service. The decision maker must balance the need to see more severely injured patients with the need to not allow those that are injured to continue to deteriorate. Conditions are provided when to prioritize each patient class. 3 - Robust Performance and Optimization of a Series Queue Michael Veatch, Gordon College, 255 Grapevine Rd, Dept. of Mathematics, Wenham, MA, 01984, United States of America, Mike.Veatch@gordon.edu Robust optimization (RO) is conservative, but when Central Limit Theorem uncertainty sets are used in an RO queue, the result is optimistic. We explain why, and propose a correction based on a diffusion model of a queue with general arrival and service times. For a series queue, the method is tractable for performance analysis and for optimal allocation of server capacity. 4 - Service Rate Control of an On/Off Server Douglas Down, McMaster University, 1280 Main Street West, Hamilton, Canada, downd@mcmaster.ca, Guang Mo, Vincent Maccio We study a server that may be turned on and off, where there is a delay to turn on the server. In addition, the service rate of the server can be chosen from a finite set. Using a cost function that involves server usage and holding costs, we discuss the following questions. Under what conditions should the server never be turned off? If it is advantageous to turn off the server, when should it be turned off? When the server is on, what speed should be chosen? 3 - Rational Abandonment from Priority Queues: Equilibrium Strategy and Pricing Implications

Chair: Mike Carter, University of Toronto, Mechanical & Industrial Engineering, 5 King’s College Rd., Toronto, ON, M5S 3G8, Canada, mike.carter@utoronto.ca 1 - Influence of Surge Capacity Protocols on Hospital Bed Capacity Planning Carolyn Busby, PhD Candidate, University of Toronto, Mechanical & Industrial Engineering, 5 King’s College Rd., Toronto, ON, M5S 3G8, Canada, carolyn.busby@mail.utoronto.ca, Mike Carter Management of finite hospital resources changes as hospitals near capacity. As such, we need to consider these “surge protocols” in order to accurately model bed capacity needs. Preliminary work is presented on a generalize discrete event simulation that models inpatient bed usage by medical and surgical patients, as well as off-servicing rules and surge protocols (ambulance consideration, early discharge, altered admission rates, cancellation of elective procedures etc.) 2 - Tools to Support Managing Access in the Veterans Health Administration Renata Konrad, Worcester Polytechnic Institute, Worcester, MA, United States of America, rkonrad@wpi.edu, Sharon Johnson, Bengisu Tulu Creating access to health services requires balancing supply and demand. A capacity cushion must be maintained, representing extra supply to absorb variability or the result is increased wait time. We describe a tool that links analysis of clinic data with actionable strategies to reduce variability in supply and demand in primary and specialty care clinics in the Veterans Health Administration. The tool is piloted in four clinics and has the potential to improve decisions surrounding access. 3 - Designing Offload Zones to Reduce Offload Delay Peter Vanberkel, Professor, Dalhousie University, P.O. Box 15000, Halifax, Canada, Peter.VanBerkel@Dal.Ca, Richard Boucherie, Corine Laan, Alix Carter We examine the offload zone - an area where multiple patients who arrive by ambulance can wait allowing the ambulance crew to return to service immediately. Although a reduction in offload delay was anticipate, it was observed that the offload zone is often at capacity. In this study we investigate why this is the case and use a continuous time Markov chain to evaluate interventions to prevent offload zone congestion. 4 - Analyzing Long-term Care Transition Data with a Multi-state Semi-markov Model Hambisa Keno, PhD Candidate, Purdue University, School of Industrial Engineering, West Lafayette, IN, United States of America, hkeno@purdue.edu, Nan Kong, Steven Landry, Mark Ward, Wanzhu Tu, Chris Callahan Capacity reconfiguration between nursing homes and home-and-community- based settings is a challenging decision for long-term care delivery. A good indicator to the capacity requirement in these facilities is length of stay. Semi- Markov models have been used to characterize patient flows in a single care facility. We extend these models to the context of multi-facility care networks. Further, we embed higher-order Markov chains to assess the impact of care pathway on the model fitting. Learning and Queues Sponsor: Applied Probability Sponsored Session Chair: Ricky Roet-Green, University of Toronto, 37 zola gate, Thornhill, L4J9A7, Canada, rgricky@gmail.com Co-Chair: Michael Kim, University of Toronto, 5 King’s College Road, Toronto, Canada, mikekim@mie.utoronto.ca Co-Chair: Andrew Lim, National University of Singapore/Department of Decision Sciences, Mochtar Riady Building, BIZ1 08-69, 15 Kent Ridge Drive, Singapore, Singapore, andrewlim@nus.edu.sg 1 - The Armchair Decision: to Depart Towards the Queue or Not Ricky Roet-Green, University of Toronto, 37 Zola Gate, Thornhill, L4J9A7, Canada, rgricky@gmail.com, Refael Hassin Consider a GPS user that inspects the traffic from his armchair at home. Given the WC22 22-Franklin 12, Marriott

433