2016 INFORMS Annual Meeting Program

TB35

INFORMS Nashville – 2016

2 - Decoupling Job Size And Server Slowdown In Modeling Redundancy

2 - An Approximate Dynamic Programming Approach To Online Capacity Planning For Rehabilitation Treatment Ingeborg A. Bikker, University of Twente, Enschede, Netherlands, i.a.bikker@utwente.nl Ingeborg A. Bikker, Sint Maartenskliniek, Nijmegen, Netherlands, i.a.bikker@utwente.nl, Martijn Mes, Richard J Boucherie We study an online capacity planning problem in which rehabilitation patients require a series of appointments with several disciplines, within a certain access time. In practice, appointments are typically planned in the first available time slots, leaving no space for urgent patients. In our research, we plan capacity for the appointment series of a patient at the moment of his/her arrival, in such a way that the total number of requests planned within their required access time is maximized. We formulate this problem as a Markov decision process, that takes into account predicted future arrivals. An approximate dynamic programming approach is used to obtain approximate solutions. 3 - Real-time Assignment Of Inpatients To Care Teams And Beds Aleida Braaksma, Massachusetts Institute of Technology, 100 Main Street, E62-389, Cambridge, MA, 02142, United States, braaksma@mit.edu, Elizabeth Ugarph, Rhodes Berube, Cecilia Zenteno, Walter O’Donnell, Retsef Levi The Department of Medicine at Massachusetts General Hospital has undergone a major care team and inpatient units redesign. In this work, we exploit the potential of the redesign to develop algorithms for real-time assignment of patients to care teams and to beds, aiming at shortening patient wait times, and decreasing the load of Medicine patients boarding in the Emergency Department. We use data-driven simulation to assess the effectiveness of the algorithms as well as to evaluate several other interventions aimed at optimizing patient flow. TB35 205A-MCC Managing Service Systems with Strategic Servers Sponsored: Manufacturing & Service Oper Mgmt, Service Operations Sponsored Session Chair: Philipp Afeche, University of Toronto, Rotman School of Management, Toronto, ON, M5S 3E6, Canada, afeche@rotman.utoronto.ca 1 - Product Support Forums: Customers As Partners In The Service Delivery Stouras Konstantinos, INSEAD, konstantinos.stouras@insead.edu, Serguei Netessine, Karan Girotra Online product support forums where customers can post complaints and questions, or report issues about a product or service of a firm abound. A large number of companies choose to crowdsource their product and service support back to their customers, employing a few dedicated service operators. We characterize the equilibrium behavior of such a novel business model for service and compare it with a call center model. 2 - Incentive Based Service System Design: Staffing And Compensation To Trade Off Speed And Quality Amy Ward, USC Marshall School of Business, Los Angeles, CA, United States, amyward@marshall.usc.edu, Dongyuan Zhan In many service systems, there is a trade-off between service speed and quality, and employees are paid based on both. We assume that the employees each selfishly choose their own service speed in order to maximize their own expected utility, which can have both a monetary and a non-monetary component. We show that a simple linear staffing and compensation policy is a first best solution in a large system limit. We further show the conditions under which a critically loaded, efficiency-driven, quality-driven, or mixed regime - in which there is simultaneous customer abandonment and server idling - emerges under a first- best linear policy. 3 - Managing Workplace Flexibility: The Case Of Agents With Task Preferences Vasiliki Kostami, London Business School, vkostami@london.edu, Rouba Ibrahim In many workplaces, employees are expected to excel in different skills as part of their job and are usually heterogeneous in their preferences to perform certain tasks. They might be willing to give up a grain of their salary to avoid working on the unlikable ones or prioritize the preferred ones. The manager, in turn, gains extra freedom in his decision to allocate tasks by charging his servers for this task discretion. We study how the choices change in equilibrium and we also derive the optimal flexibility fee under two innovative flexibility schemes and different heterogeneity scenarios.

Kristen Gardner, Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh, PA, 15213, United States, ksgardne@cs.cmu.edu, Mor Harchol-Balter, Alan Scheller-Wolf Redundancy is an increasingly common technique for reducing response time in multi-server queueing systems. In a system with redundancy, jobs replicate themselves and wait in multiple queues; a job is complete as soon as any one copy completes service. Much of the existing theoretical work on analyzing systems with redundancy makes the unrealistic assumption that a job’s service times are independent across different servers. This assumption can lead to results that are at odds with implementation results. We introduce a new, more realistic model of redundancy and design a new dispatching policy that is both analytically tractable within our model and has provably excellent performance. 3 - Gated Queues With Impatience Customers George Mytalas, NJIT, Newark,NJ, NJ, United States, mytalas@aueb.gr We analyze an M/M/1 queueing model with gated service discipline. In this discipline there is a waiting room and a service queue. Each time the service queue becomes empty all customers in the waiting room are instantaneously put in the service queue. Customers in waiting room can be impatient after waiting a random amount (or a certain) of time and abandon the system. We derive the joint distribution of the number of customers in waiting room and service queue, and obtain various quality of service measures. 4 - Appointment Systems Under Service Level Constraints Rui Chen, University of Toronto, 105 St George St, Toronto, ON, M5S 3E6, Canada, rui.chen@rotman.utoronto.ca, Rowan Wang, Zhenzhen Yan, Saif Benjaafar, Oualid Jouini We consider a new model of appointment scheduling where customers are given the earliest possible appointment times under the service level constraint that the expected waiting time of each individual customer cannot exceed a given threshold. We apply the theory of majorization to analytically characterize the structure of the optimal appointment schedule. We show that, the optimal inter- appointment times increase with the order of arrivals. We prove that, when customer service times follow an exponential distribution, our system converges asymptotically to the D/M/1 queueing system as the number of arrivals approaches infinity. We also extend our analysis to systems with multiple servers. 5 - Approximations For Heavily-loaded G/ Gi/n+ Gi Queues Yao Yu, North Carolina State University, 4335-3 Avent Ferry Road, Raleigh, NC, 27606, United States, yyu15@ncsu.edu, Yunan Liu, Ward Whitt Motivated by applications to service systems, we develop convenient engineering approximation formulas for the steady-state performance of heavily-loaded G/GI/n+GI multi-server queues. Based on established Gaussian many-server heavy-traffic limits in the efficiency-driven regime, however, the approximations also apply to systems in the quality-and-efficiency-driven regime where traffic intensity is close to 1 from above. Simulation experiments show that the proposed approximations are effective for large-scale queueing systems for a significant range of the traffic intensity and the abandonment rate. Scheduling and Workload Assignment in Healthcare Sponsored: Manufacturing & Service Oper Mgmt, Healthcare Operations Sponsored Session Chair: Retsef Levi, MIT, 100 Main Street, Building E62-562, Cambridge, MA, 02142, United States, retsef@mit.edu Co-Chair: Cecilia Zenteno, Massachusetts General Hospital, 55 Fruit Street, White 400, Boston, MA, 02114, United States, azentenolangle@mgh.harvard.edu 1 - Integrated Scheduling And Capacity Planning With Considerations For Patients’ Length-of-stay Nan Liu, Columbia University, New York, NY, United States, nl2320@columbia.edu, Van-Anh Truong, Xinshang Wang, Brett Anderson Motivated by the shortcoming of current hospital scheduling and capacity planning methods which often model different units in isolation, we introduce the first dynamic multi-day scheduling model that integrates information about capacity usage at more than one location in a hospital. In particular, we analyze the first dynamic model that accounts for patients’ length-of-stay and downstream census in scheduling decisions. Through numerical experiments on real data, we show that there is substantial value in making integrated scheduling decisions. In contrast, localized decision rules that only focus on a single location of a hospital can result in up to a three-fold increase in total expenses. TB34 204-MCC

275