2015 Informs Annual Meeting

WD76

INFORMS Philadelphia – 2015

WD76 76-Room 204C, CC Simulation and Optimization Contributed Session Chair: Siyang Gao, Assistant Professor, City University of Hong Kong, Tat Chee Avenue, P6605, AC1, Kowloon, Hong Kong - PRC, siyangao@cityu.edu.hk 1 - A Simulation Based Traffic Control Policy for Hazardous Materials Transportation Sara Masoud, The University of Arizona, A214, 1300 E. Fort Lowell Rd,, Tucson, AZ, 85719, United States of America, saramasoud@email.arizona.edu, Sojung Kim, Young-jun Son A simulation-based traffic control policy for hazardous materials (hazmat) vehicles is devised to enhance public safety without sacrificing traffic efficiency, by considering two traffic control policies simultaneously: 1) Network Design which restricts hazmat vehicles from freeways; and 2) Dual Toll Pricing which levies toll on both hazmat vehicles and regular vehicles in tollways. The proposed approach is demonstrated via AnyLogicÆ ABS software with a real traffic data of San Antonio, Texas. 2 - General-Purpose Ranking and Selection Soonhui Lee, UNIST, UNIST-gil 50, Ulsan, Korea, Republic of, shlee@unist.ac.kr, Barry Nelson In this study we take a step toward general-purpose Ranking & Selection procedures that work for many types of performance measures and output distributions, including situations in which different simulated alternatives have entirely different output distribution families. To obtain the required PCS we exploit intense computation via bootstrapping, and establish the asymptotic PCS of these procedures. 3 - Convex Risk Measures: Efficient Computations via Monte Carlo Zhaolin Hu, Associate Professor, Tongji University, School of Economics and Management, Shanghai, China, huzhaolin@gmail.com In this paper, we study an important class of convex risk measures called utility- based shortfall risk (SR). We develop efficient Monte Carlo methods for estimation of SR, sensitivity analysis of SR, and optimization of SR. Numerical experiments are studied extensively, which further demonstrate the effectiveness of our methods. 4 - Territory Design under Uncertainty Peter Verderame, Air Products and Chemicals, Inc., 7201 Hamilton Boulevard, Allentown, PA, United States of America, verderpm@airproducts.com, Subhajit Ghoshal, Erdem Arslan, Pratik Misra Territory design looks to maximize the efficiency of a company’s resources through intelligent, objective placement and allocation of assets. Balancing workload across territories is a critical factor for successful deployment; however, projected workload is often uncertain which in turn greatly impacts design robustness. We developed a sophisticated optimization-and-simulation-based framework which explicitly considers the parametric or nonparametric uncertainty surrounding workload forecasts. 5 - Efficient Feasibility Determination with Multiple Performance Measure Constraints Siyang Gao, Assistant Professor, City University of Hong Kong, T at Chee Avenue, P6605, AC1, Kowloon, Hong Kong - PRC, siyangao@cityu.edu.hk, Weiwei Chen Feasibility determination has emerged as a widely applied problem in simulation optimization. It seeks to provide all the feasible designs from a finite set of design alternatives. In this paper, we consider the this problem in presence of multiple performance measure constraints. The optimal solution to maximize the probability of correct selection is derived under asymptotic approximation. The numerical testing shows that our approach can enhance the simulation efficiency significantly.

3 - Estimation of Servers Utilization in an Unreliable Queueing System with Stacked Servers Maboubeh Madadi, University of Arkansas, Fayetteville, AR, United States of America, mmadadi@uark.edu, Richard Cassady, Shengfan Zhang, Lisa Maillart We consider a queueing system comprised of a set of identical servers that are stored in a stack when not in use. In such cases, customers may find it more convenient to select the server that is on top of the stack. We develop a continuous-time Markov chain model to compute the cumulative, transient utilization and age of each server based on the number of servers in the system, the server’s initial position in the stack, the arrival rate and the service rate. 4 - Optimal Preventive Maintenance Planning in Anticipation of Imperfect Implementation Kai He, University of Pittsburgh, 1048 Benedum Hall 3700 O’Hara Street, Pittsburgh, PA, 15261, United States of America, kah167@pitt.edu, Oleg Prokopyev, Lisa Maillart Maintenance planners usually assume perfect implementation of their prescribed preventive maintenance (PM) policies. However, the maintenance workers often deviate from the intended PM policy resulting in imperfectly timed PM actions. We formulate cost rate-minimizing models to investigate the impact of such deviations, assuming that the actual PM time differs from the scheduled PM time in a probabilistic manner. We establish results for age replacement with and without minimal repair policies. Chair: Faranak Fathi Aghdam, The University of Arizona, 3125 E. Bellevue Street, Tucson, AZ, 85716, United States of America, faranakf@email.arizona.edu 1 - A Reliability Model for Multi-State Systems with Multi-State Components with Different Failures Carlos Solorio, Assistant Professor, CETYS Universidad, Calzada S/N, Mexicali, Mexico, carlos.solorio@cetys.mx A general model that evaluates the reliability of complex engineering systems that suffer soft failures due to common degradation of physical systems and catastrophic failures due to sudden shocks that provoke powerful stresses is presented. The general reliability model considers multi-state systems with multi- state components, where system reliability is evaluated based on the states of the components. Performance measures are presented that help us decide which system is better. 2 - Optimal CBM Policies under the Gamma Degradation Process David Han, University of Texas, One UTSA Circle, San Antonio, TX, United States of America, David.Han@utsa.edu CBM is an effective method to reduce unexpected failures as well as the O&M costs. This talk discusses the CBM policy with optimal inspection points under the gamma degradation process. A random effect parameter is used to account for population heterogeneities and its distribution is continuously updated at each inspection epoch. The observed degradation level along with the system age is utilized for making the optimal maintenance decision, and the structure of the optimal policy is examined. 3 - An Opportunistic Condition-Based Maintenance Policy with Two Degradation Thresholds Joeri Poppe, KU Leuven, Naamsestraat 69, Leuven, 3000, Belgium, joeri.poppe@kuleuven.be, Robert Boute, Marc Lambrecht Condition-based maintenance (CBM) is a maintenance strategy that makes uses of the actual condition of the component. We introduce CBM in combination with preventive and corrective maintenance. We propose a CBM policy that relies on two thresholds of the degradation level, which can initiate a maintenance action on the monitored component. We find that major reductions in both maintenance cost and system unavailability can be realised, compared to the established maintenance policies. 4 - Reliability Study of High-k Bi-layer Dielectrics with Non-weibullian Failure Time Distribution Faranak Fathi Aghdam, The University of Arizona, 3125 E. Bellevue Street, Tucson, AZ, 85716, United States of America, faranakf@email.arizona.edu, Haitao Liao As electronic devices get smaller, reliability issues pose new challenges due to unknown underlying physics of failure. This necessitates new reliability analysis approaches related to nano-scale devices. The time-dependent breakdown of dielectric films is one of the most important reliability issues. In this research, we study two new approaches for modeling the time to breakdown of high-k bi-layer dielectrics. WD74 74-Room 204A, CC Reliability III Contributed Session

477