Informs Annual Meeting 2017

SB17

INFORMS Houston – 2017

3 - Warehouse Storage Layout Optimization using Community Detection and Mathematical Modeling Husam Dauod, PhD Student, Binghamton University, 99 Main Street, Apt 5, Binghamton, NY, 13905, United States, hdauod1@binghamton.edu, Sung Hoon Chung, Sang Won Yoon This research presents a community detection approach to analyze and optimize warehouse storage layout in order picking environments. Storage layout is a key warehouse management function that greatly affects the overall performance. However, storage layout optimization is challenging because it requires the analysis of large amounts of orders data. To analyze this data efficiently, networks community detection is used to extract association relationships between items. These relationships are then used in a mixed integer programming model to determine optimal inventory locations. Results indicate that the proposed approach provide good results with short computational time. 4 - Dynamic Substitution Policy for Selling Multiple Products with Supply Uncertainty Chengzhang Li, Purdue University, 2367 Edison Dr, West Lafayette, IN, 47906, United States, li1392@purdue.edu, Qi Feng, J. George Shanthikumar We study a firm selling multiple substitutable products over a planning horizon of multiple periods. The firm faces replenishment cycles of fixed lengths with uncertain supply. In each period, when the random demands materialize, the firm may substitute one product with another in meeting the demands. Extending the notion of stochastic linearity, we show that the replenishment problem is concave via transformation. We design an efficient algorithm to decide the substitution policy that delivers close-to-optimal performance. We also show that restricting substitutions between products with adjacent characteristics can yield a benefit close to allowing full substitution among all products. 5 - Optimum Pricing and Inventory Control for Spare Parts in Existence of Secondary Markets Mustafa Hekimoglu, Assistant Professor, Isik University, Sile Campus, Sile, Istanbul, 34980, Turkey, mustafa.hekimoglu@isikun.edu.tr, Rommert Dekker, Erwin Van der Laan Secondary markets for spare parts are online platforms where Original Equipment Manufacturers (OEMs), distributors, customers can obtain spare parts with short lead times and for cheaper prices or they can dump their excess inventory. From an OEM’s perspective, pricing against secondary market while controlling inventory is critical as higher prices may lead customers to secondary markets or even early phase-outs of their capital products. In this study, optimum policy for single period problem of pricing and inventory control is developed under linear demand model assumption. Also, we developed insights for multi-period problem under discounter cost criterion. 340A Healthcare/Service Systems in Applied Probability Sponsored: Applied Probability Sponsored Session Chair: Song-Hee Kim, University of Southern California, Los Angeles, CA, 90089, United States, songheek@marshall.usc.edu 1 - Dynamic Decision Making in a Queueing System with Secondary Service Wanyi Chen, wanyic@live.unc.edu Motivated by operational practices in emergency departments, we consider a queueing model where each job is one of two types. Type 1 jobs need only a primary service given by a single server. Type 2 jobs need an additional secondary service. Secondary service is conducted by infinitely many servers. Primary servers cannot serve a new job until secondary service of a job is over. Jobs incur waiting costs and there is an option of starting primary and secondary services together with an extra cost. The decision is whether or not to use that option for each job given the probability that the job is type 1. We show the optimal policy is of threshold-type and numerically test the performance of heuristic methods. 2 - Improving Itinerary Completion at a Destination Healthcare Institution Jonathan Helm, Indiana University, Kelley School of Business, 1309 East Tenth Streeet, Bloomington, IN, 47405, United States, helmj@indiana.edu, Mark P.Van Oyen At a destination hospital, patients travel long distances to receive treatment. For these patients, itinerary completion - defined as the completion of their treatment segment before the weekend - is an important metric. Motivated by itinerary completion, we present a new methodology for modeling and optimizing flows in a network of outpatient clinic visits to minimize flow times of priority patients within a queueing network with hard deadlines for service completion. SB17

3 - A Real-time Risk Prediction and Discharge Optimization Framework for Inpatient Flow and Patient Outcomes Pengyi Shi, Purdue University, 403 W. State St, Krannert School of

Management, Kran 472, West Lafayette, IN, 47907, United States, shi178@purdue.edu, Jonathan Helm, Jivan Romain Deglise-Hawkinson, Julian Pan

Hospital discharge processes are critical to quality and access. Inpatient congestion can be mitigated by early discharge, which, however, carries risks of adverse events for individual patients. In collaboration with a healthcare analytics company, we develop a real time risk prediction model for readmission risk on each day of a patient’s hospital stay. We integrate this prediction with a discharge optimization framework that balances the risk of early discharge for individual patients with the risks of inpatient congestion. In this talk, we present our methodology and discuss our joint implementation efforts at a partner hospital. 4 - Analysis of Hospital Networks via Time-Varying Fluid Models with Blocking Noa Zychlinski, Technion - Israel Institution of Technology, Haifa, 3200003, Israel, noazy@tx.technion.ac.il, Avishai Mandelbaum, Petar Momcilovic, Izack Cohen This research focuses on modeling, developing and analyzing time-varying fluid networks with blocking. These models are motivated by two applications. The first is patient flow analysis between hospitals and geriatric institutions in order to improve their joint operation. The second application includes analysis of tandem flow lines with blocking and finite waiting rooms before the first station, and between stations. The finite capacity of the first station leads to customer loss, and therefore, requires reflection analysis. We demonstrate that our models yield operational insights on network performance. 340B Stochastic Networks and Queueing in Applied Probability IV Sponsored: Applied Probability Sponsored Session Chair: Maria Vlasiou, Eindhoven University of Technology, P.O. Box 513, Eindhoven, 5600MB, Netherlands, m.vlasiou@tue.nl Co-Chair: Bert Zwart, Eindhoven, 5629RD, Netherlands, Bert.Zwart@cwi.nl 1 - Effect of Presorting on the Delay in Multiclass Queues with a Global FCFS Service Discipline Willem Mélange, Ghent University, Ghent, Belgium, Willem.Melange@UGent.be, Joris Walraevens, Dieter Claeys, Herwig Bruneel We study the effect of presorting on the delay of a random customer in a continuous-time queueing system with two types of customers and dedicated servers. In this system, all arriving customers are accommodated in a single FCFS queue. All customers, except for the first P are served in a FCFS order, regardless of their types. The first P customers however can overtake customers of the other type in order to be served. The motivation for this work comes from traffic. Our SB18 Junqi Hu, Georgia Institute of Technology, Atlanta, GA, 30332, United States, hujunqi1210@gatech.edu, Sigrun Andradottir, Hayriye Ayhan Consider a Markovian queueing network with two stations in tandem,flexible servers, finite intermediate buffer, and infinite supply of jobs in front of the first station. We assume that the servers always work in teams and allow the team service rates to be general. Our objective is to maximize the long-run average throughput by dynamically assigning servers to teams and teams to stations. We first establish sufficient criteria for eliminating inferior teams, and then we show how to find the optimal policy among the remaining teams. Finally, we investigate special cases with constraints on team formation and with structured team service rates. 3 - Strategic Customers in Systems with Batch Arrivals: On How to Protect Your Party Olga Bountali, Southern Methodist University, 4210 Fairmount Street, Apartment 3063, Dallas, TX, 75219, United States, obountali@smu.edu, Apostolos Burnetas, Lerzan Ormeci Customers in batch-arrival systems that are served individually vacillate upon the “join-balk” dilemma: should each member decide under the objective of maximizing his revenue or should they all aim to maximize the batch welfare? We consider a Markovian queue and address the question above under two rules with regards to the decision framework: the 0-1 case, where the whole batch decides to join or balk, and the partial case, where only a fraction of the customers joins. We analyze and compare the equilibrium behavior in both cases and the corresponding implications on the social welfare. results can be used to determine optimal lengths of filter lanes. 2 - Dynamic Control of Service Systems with Teams

40