Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WA01

4 - Routing for Post-disaster Needs Assessment to Improve Information Accuracy and Precision Duygu Pamukcu, Middle East Technical University, Middle East Technical Uni. Ind, Ankara, 06400, Turkey, Melih Celik, Burcu Balcik Reliable information in the post-disaster needs assessment depends on how much time is spent for statistical sampling to collect information and how many different beneficiary groups visited. Estimated information is an input for relief distribution, so accuracy and precision of the estimation directly effect the efficiency of relief operations. The scarce resource, namely time, restricts visits to affected sites. Since information collection and travel time decisions are subject to the same total time constraint, there exists an inherent trade-off between them. Exact and heuristic methods are presented to analyze this tradeoff and support sample size and site selection decisions. 5 - Multi-objective Robust Model of Stochastic and Possibilistic Programming for Mitigation Planning and Emergency Response Luis Yáñez, Universidad de Chile, Santiago, Chile, Cristian Eduardo Cortes, Pablo Andres Rey We propose a multi-objective robust model of stochastic and possibilistic programming for facility location, inventories allocation and distribution in the context of mitigation planning and emergency response. Our model seeks to optimize the risk associated with scenario planning and minimal total cost, including a social component, deprivation costs, faults of organization and convergence of materials. In addition, protection with respect to uncertain parameters is included in the model, as epistemic uncertainty given by the use of imprecise parameters and scarcely systematized information. Healthcare Operations under Uncertainty Sponsored: Optimization/Optimization under Uncertainty Sponsored Session Chair: Yiling Zhang, University of Michigan, Ann Arbor, MI, 48105, United States 1 - Appointment Scheduling under Time Dependent Patient No-Show Behavior Zhenzhen Yan, Nanyang Technological University, 21 Nanyang Link, Singapore 637371, Qingxia Kong, Shan Li, Nan Liu, Chung-Piaw Teo This paper studies how to schedule medical appointments with time-dependent patient no-show behavior and random service times. We observe a significant time-of-day effect on patient show-up probabilities from two clinic datasets. We deploy a distributionally robust approach to model the schedule problem. To tackle the case when patient no-shows are endogenous on the schedule, we construct a set of dual prices to guide the search for a good schedule and use the technique iteratively to obtain a near-optimal schedule. Our computational studies reveal a significant reduction in total expected cost by taking into account the time-of-day variation in show-up probabilities. 2 - Hedging the Overtime Riskiness in Online Surgery Assignments Minglong Zhou, National University of Singapore, Singapore, Melvyn Sim We study an advanced scheduling problem where patients are assigned to different operating theatre slots. We propose an extended riskiness index as the risk measure. We show such riskiness index shares a representation in terms of a family of measures of risk. We further propose the lexicographical riskiness index, which defines a unifying framework to capture multi-priority or piece-wise preferences. We minimize the risk of failing to meet some targets in a forward looking framework with a non-anticipative policy to incorporate future uncertainty. Simulation results show that our approach controls the risk of overtime while not compromising overall number of surgeries compared to a Benchmark. 3 - Nurse Staffing under Uncertain Demand and Absenteeism Minseok Ryu, University of Michigan, 1911 McIntyre St, Ann Arbor, MI, 48105, United States, Ruiwei Jiang This paper describes a data-driven approach for nurse staffing decision under uncertain demand and absenteeism. We propose a distributionally robust nurse staffing (DRNS) model with both exogenous (stemming from demand uncertainty) and endogenous uncertainty (stemming from nurse absenteeism). Wednesday, 8:00AM - 9:30AM n WA01 North Bldg 121A Joint Session OPT/Practice Curated:

We provide a separation approach to solve the DRNS model with general nurse pool structures. Also, we identify several classes of nurse pool structures that often arise in practice and show how the DRNS model in each of these structures can be reformulated as a monolithic mixed-integer linear program that facilitates off-the-shelf commercial software. Also, we propose an optimal nurse pool design model. 4 - Integer Programming Approaches for Appointment Scheduling with Random No-shows and Service Durations Yiling Zhang, University of Michigan, Ann Arbor, MI, 48105, United States, Ruiwei Jiang, Siqian Shen We consider a single-server scheduling problem given a fixed sequence of appointment arrivals with random no-shows and service durations, of which only the support and first moments are known. We formulate a class of distributionally robust (DR) optimization models that incorporate the worst-case expectation/conditional Value-at-Risk (CVaR) penalty cost of appointment waiting, server idleness, and overtime as the objective or constraints. We obtain exact mixed-integer nonlinear programming reformulations and derive valid inequalities to strengthen the reformulations. Convex hulls are derived for the least and the most conservative supports of no-shows. Various instances are tested. n WA02 North Bldg 121B Joint Session OPT/MIF: Multistage Mean-Risk Stochastic Programming Sponsored: Optimization/Optimization under Uncertainty Sponsored Session Chair: Prasad Parab, Texas A&M University, College Station, TX, 77840, United States 1 - Column Generation Algorithm for Risk-averse Two-stage Stochastic Programs Saravanan Venkatachalam, Wayne State University, Detroit, MI, 48377, United States, Sujeevraja Sanjeevi In the airlines operations, re-timing of flight legs is a tactic used in the schedules to minimize the cost of delay propagation. In this talk, we propose a two-stage mean-risk stochastic programming approach to determine the re-timing of flights and minimize the expected delay propagations due to disruptions on the day of operations. The mean-risk models pose computational challenges. In this talk, we present decomposition algorithms and computational results. 2 - Decomposition for Risk-averse Multistage Stochastic Programs with Expected Conditional Risk Measures Lewis Ntaimo, Professor, Texas A&M University, 3131 TAMU, College Station, TX, 77843, United States, Maryam Khatami, Bernardo Pagnoncelli We formulate risk-averse multistage stochastic programs (MSLPs) with expected conditional risk measures (ECRMs) in the context of both quantile and deviation mean-risk measures. We study decomposition algorithms for this class of problems. Using MSLPs with ECRM, we model and solve hydrothermal scheduling and lifetime portfolio selection problems. We investigate the impact of risk-aversion on decision making for these applications. 3 - Stochastic Decomposition for Mean-risk Multistage Stochastic Linear Programs Prasad Parab, Texas A&M University, College Station, TX, 77840, United States, Lewis Ntaimo, Bernardo Kulnig Pagnoncelli Mean-risk multistage stochastic linear programs (MR-SLPs) are difficult to solve due to the incorporation of risk measures in the objective function and their large-scale nature. In this talk, we present stochastic decomposition for MR-SLPs with deviation and quantile risk measures and report on its application to hydrothermal scheduling problems. 4 - Multi-stage Risk-averse Stochastic Mixed-integer Programming for Mixed-model Assembly Line Sequencing Ge Guo, Iowa State University, 4701 Todd Dr. Unit 208, Ames, IA, 50014, United States, Sarah M. Ryan The existing optimization formulations for mixed-model assembly line sequencing do not consider some major real-world uncertainty factors such as timely part delivery and material quality. We present a multistage stochastic mixed-integer formulation to make sequencing decisions with optimal on-time performance considering uncertainties of part delivery and material quality. A risk-averse model is further developed to guarantee customers’ satisfaction with on-time performance. Computational studies are performed with the Progressive Hedging algorithm and a lower bounding approach in real-time resequencing.

410