2016 INFORMS Annual Meeting Program

WC33

INFORMS Nashville – 2016

4 - Cardinality Constrainted CDS Portfolio Optimization Dexiang Wu, Stockholm University, Stockholm Business School, Stockholm, 106 91, Sweden, dexiang.wu@sbs.su.se, Desheng Dash Wu We study the Credit Default Swap (CDS) market from the portfolio perspective. CDS-based portfolios are constructed through incorporating solvency and cardinality constraints for the purpose of decentralization. Portfolio size is controlled exactly and therefore mixed integer quadratic and linear programs that consider different risk measurements are studied. We found that moderate size can generally obtain better performance in terms of portfolio Sharpe ratio and other metrics. Moreover, due to the specific structure of CDS correlation matrix, a cluster simplification process is applied to speed up the computation. WC33 203B-MCC Production and Scheduling III Contributed Session Chair: Steffen T. Klosterhalfen, BASF SE, Ludwigshafen, Germany, steffenklosterhalfen@googlemail.com 1 - Fitting Clearing Fun Actions Using Generalized Regression Methods Reha Uzsoy, North Carolina State University, Dept. of Industrial & Systems Engg, 300 Daniels Hall Camps Box 7906, Raleigh, NC, 27695-7906, United States, ruzsoy@ncsu.edu, Karthick Gopalswamy Clearing functions are widely used in production planning models to capture the nonlinear relationship between workload and output. Traditionally least squares methods have been used to fir the data, which weight the errors of both signs equally and assume errors to be iid normally distributed. In this work we relax the assumption of normality and provide a generalized regression approach to fit the data taking into consideration that the data is not balanced. Computational experiments evaluate the performance of the proposed methods for fitting clearing functions. 2 - A Two-stage Stochastic Programming Model For Lot-sizing And Scheduling Under Uncertainty Lot-sizing and scheduling is one of the medium-term production planning problems in manufacturing. In this study, demand uncertainty has been considered and a robust production plan has been proposed with a two-stage stochastic programming framework. A case study proves that uncertainty has a significant impact on production planning. 3 - Managing Product Transitions In Semiconductor Wafer Fabrication Facilities Reha Uzsoy, North Carolina State University, Dept. of Industrial & Systems Engg, 300 Daniels Hall Camps Box 7906, Raleigh, NC, 27695-7906, United States, ruzsoy@ncsu.edu, Atchyuta B. Manda, Sukgon Kim, Karl Kempf We consider the problem of managing the introduction of new products into a wafer fabrication facility when the new product is subject to higher levels of process uncertainty than the current one. We propose a model for the impact of the new product on the cycle time of the fab using queueing concepts, and illustrate the behavior of the model with computational experiments. 4 - Scheduling With Batching Decisions And Energy Constraints For Steelmaking Continuous Casting Production Wenjie Xu, Northeastern University, NO. 3-11, Wenhua Road, Heping, Shenyang, 110819, China, xuwenjie.neu@outlook.com, Lixin Tang We study a novel scheduling problem with batching decisions of steelmaking continuous- casting (SCC) production. The energy constraints in this problem represent the conversion process from the Linz-Donawitz process gas (LDG) to electricity. The problem uses the minimum makespan as the scheduling objective and the minimum total electricity cost as the energy objective. A multi-objective optimization framework which incorporates an improved epsilon-constraint method is proposed to solve the problem. Preliminary results show the effectiveness of the multi-objective optimization framework and demonstrate the tradeoffs between minimum makespan and energy cost. Zhengyang Hu, Research Assistant, Iowa State University, 100 Enrollment Services Center Ames, Ames, IA, 50011, United States, zhengya@iastate.edu, Guiping Hu

5 - Integrated Production And Safety Stock Planning In The Process Industry Steffen T. Klosterhalfen, BASF SE, Ludwigshafen, Germany, steffenklosterhalfen@googlemail.com, Stefan Minner, Dariush Tavaghof Gigloo We develop and apply a new approach for integrated production lot-sizing and safety stock planning in the process industry where high demand uncertainty and large production campaigns are the rule. Our approach is based on mixed-integer linear programming and benchmarked with common sequential lot-sizing and safety stock planning frameworks characterized by different levels of sophistication in optimization methodology and parameter updating. 204-MCC Joint Session HAS/MSOM-HC: Advances in Healthcare Operations Sponsored: Manufacturing & Service Oper Mgmt, Healthcare Operations Sponsored Session Chair: Van-Anh Truong, Columbia University, 500 120th Street, New York, NY, 10027, United States, vt2196@columbia.edu 1 - A Pomdp For Reducing Readmissions Through Inpatient Outpatient Joint Control Xiang Liu, University of Michigan, liuxiang@umich.edu, Jonathan Helm, Mariel Sofia Lavieri, Ted Skolarus Hospital readmissions affect hundreds of thousands of patients, placing a tremendous burden on the healthcare system. We develop a two-stage POMDP that spans the inpatient stay and the post-discharge outpatient monitoring to reduce readmission. By learning and reducing readmission risk in the inpatient stage, and monitoring and intervening patient’s health condition in the second stage, our model jointly optimizes both discharge and post-discharge decisions to reduce readmissions. 2 - A Fluid Model For An Overloaded Bipartite Queueing System With Scoring Based Priority Rules Yichuan Ding, Assistant Professor, University of British Columbia, We consider an overloaded bipartite queueing system (BQS) with multi-type customers and service providers. A service provider assigns each customer a score based on customer type, waiting time, and server type. Service is always provided to the customer with the highest score. We approximate the behavior of such a system using a fluid limit process, which has two important features: (1) the routing flow rates at a transient state coincide with the maximal flow in a parameterized network; (2) the routing flow rates in the steady state coincide with the minimal-cost maximal-flow in a capacitated network. We illustrate the application of our machinery via an example of public housing assignment. 3 - Routing Shared Vehicles With Matching Constraints For Medical Home Care Delivery Miao Yu, University of Michigan, 1205 Beal Avenue, Ann Arbor, MI, 48109, United States, miaoyu@umich.edu, Viswanath Nagarajan, Siqian Shen In this paper, we study a vehicle routing problem variant for medical home care delivery. A health care provider assigns multiple vehicles to transport homecare devices and/or medical staff to patients’ home locations given that each patient can only be served by a subset of vehicles. We construct an integer-programming model solved by column generation, to minimize the total traveling distance of all the vehicles. We also propose an approximation algorithm that yields fast assignment, and conduct out-of-sample simulation to numerically evaluate the performance of the proposed methods. 4 - A Periodic Little’s Law And Its Application To Emergency Department Data Xiaopei Zhang, Columbia University, 1 Morningside Drive, Apt. 1710, New York, NY, 10025, United States, xz2363@columbia.edu, Ward Whitt We establish a new periodic discrete-time version of Little’s law and apply it to explain the remarkable fit of a data-driven stochastic process model, which is periodic over a week, of the emergency department occupancy over time in the Israeli Rambam hospital. WC34 2053 Main Mall, Vancouver, BC, V6T1Z2, Canada, daniel.ding@sauder.ubc.ca, Mahesh Nagarajan, S. Thomas McCormick

438