2015 Informs Annual Meeting

WC74

INFORMS Philadelphia – 2015

WC74 74-Room 204A, CC Reliability II Contributed Session

simulating hitting time of Brownian Meander to a fixed boundary. Numerical experiments show that our method performs 10-15x faster than current methods. 2 - Modeling and Simulation of Automotive Assembly Line Ahad Ali, Associate Professor, Lawrence Technological University,

21000 West Ten Mile Road, Canton, MI, 48188, United States of America, aali@ltu.edu, Don Reimer

Chair: Yin Shu, University of Houston, E206 Engineering Bldg.2, Houston, TX, 77204, United States of America, yinshulx@gmail.com 1 - Detecting Entropy Increase in Categorical Data using Maximum Entropy Distribution Devashish Das, Research Assistant, University of Wisconsin, Madis, 124 North Breeze Terrace Apt. E, Madison, WI, 53726, United States of America, ddas3@wisc.edu, Shiyu Zhou In work, we propose a statistical monitoring method to detect the increase of entropy in categorical data. First, we propose a distribution estimation method to approximate the probability distribution of the observed categorical data. Then we use this procedure to estimate the non-parametric, maximum entropy distribution of an observed data sample and use it for statistical monitoring. 2 - A Conditioned-Based Maintenance Policy for a Two-Component System Ameneh Forouzandeh Shahraki, NDSU, 1220, 10th St. N, #101, Fargo, ND, United States of America, ameneh.forouzandehsh@ndsu.edu, Om Yadav This paper proposes an optimal conditioned-based maintenance policy for a two- unit deteriorating system with economic dependency. Each unit is monitored by remaining useful life based-inspection policy and is maintained by corrective, perfect or imperfect maintenance actions.We propose a general maintenance decision framework to select optimal maintenance actions and to optimize the grouping of maintenance actions for both components at the same time. 3 - Optimal Decision Making for Systems with Mulifunctional Components Yiwen Xu, University of Arizona, RM111, 1127 E. James E. Rogers Way, Tucson, AZ, United States of America, yiwen.xu6@gmail.com, Haitao Liao We studied systems with multifunctional components. The goal is to make an optimal decision to maximize the system reliability when a failure occurs. Properties and numerical studies are included. 4 - Non-Gaussian Ornstein-Uhlenbeck Processes in Degradation-based Reliability Analysis Yin Shu, University of Houston, E206 Engineering Bldg.2, Houston, TX, 77204, United States of America, yinshulx@gmail.com, Qianmei Feng, Hao Liu, Edward Kao We use non-Gaussian Ornstein-Uhlenbeck (OU) processes to model the evolution of degradation with random jumps. The superiority of our models stems from their flexibility in modeling stylized features of degradation data series such as nonlinearity, jumps fluctuation, asymmetry, and heavy tails. Based on Fokker- Planck equations, we derive explicit results for reliability characteristics represented by Levy measures. Our models are applicable for analyzing a great deal of degradation phenomenon. 5 - A Flexible Random Effects Zero-inflated Model for the Study of Copper Hillocks Growth Guilin LI, National University of Singapore, #07-17 Blk E1, Engineering Drive 2, NUS, Singapore, 117576, Singapore, guilin_li@u.nus.edu, Szu Hui Ng, Daniel Chua, Royston Tan An experiment was conducted to measure the metal layer shorts in integrated circuits caused by copper hillocks to improve the process. Two features of the collected short counts are identified: excessive zeros and multi-level variations. A zero-inflated model with flexible random effects was proposed. Statistical inference procedures are also developed.

This paper provides methods to create a valid representation of a automotive assembly using modeling and simulation. Various performance analysis are presented with statistical validation. The use of the mean steady state 90% confidence interval was used to measure job per hour for the simulation model to

the current system. DOE and RSM will be used and analyzed. 3 - Promoting Loose Coupling in Simulation Models: The Service Broker Approach

Sunil Kothari, Researcher, Hewlett-Packard, 1501 Page Mill Road, Bldg 2U, Palo Alto, CA, 94304, United States of America, sunil.kothari@hp.com, Jun Zeng, Gary Dispoto, Francisco Oblea Many simulation models are not designed for reuse since the development of reusable models can take significant investment in time and effort since they have to be abstracted over all possible use cases. We illustrate the service broker architecture in the context of industrial printing domain. The service broker architecture ensures that the resources are loosely coupled to demand. We highlight two case studies to show our approach. 4 - Maximal On-time Delivery by Soft-pegging in Wafer Fab under Consideration of Hierarchical Mes Joon Young Lee, ASU ME, 850 S. McAllister Ave., Tempe, AZ, 85287, United States of America, joon.lee@asu.edu Under given hierarchical production planning and control of the MES, assignment policy (soft-pegging) is considered in order to maximize customer service level in semiconductor wafer fabrication processes. A wafer fab is modeled and several input scenarios were experimented to see the effects of the policy. On-time delivery is maximized in equilibrium state at a diversification point by proper assignment policy. 5 - Knowledge Evolution in a Dynamic R&d Team from the Perspective of Task Performance Ni Xia, Mr., School of Management, Huazhong University of Science and Technology, No.1037, LuoYu Road, Hongshan District, Wuhan, 430074, China, xntx128390@163.com We leverage learning by doing to establish a model of knowledge evolution in a R&D team,and simulate the effects of strength of team dynamics,task load,task complexity.Simulation results indicate that dynamic strength has greater damage to group knowledge than individual knowledge.It not only destroys group knowledge,but also benefit knowledge recovery.We also find low task load benefits knowledge restore but harms knowledge increase, while task complexity cannot take significant effects. Engineering, North Carolina State University, 111 Lampe Dr., Daniels 443, Raleigh, NC, 27695, United States of America, zwang23@ncsu.edu 1 - Price and Quantity Competition in Mixed Market with a Common Retailer Jian Liu, Dr., Hohai University, Focheng West Road No.8, Business School, Hohai University, Nanjing, JS, 211100, China, liujane1124@126.com, Sun Li, Huimin Wang This paper investigates the public supplier and the private supplier’s equilibrium pricing and quantity decisions in mixed market. Through comparing these equilibrium results under different game, we find that under Cournot competition, the whole market is in prison’s dilemma. Under Bertrand competition, the market equilibrium exists when the product differentiation is lower, where the public supplier is leader and the private one is follower. 2 - Real-time Demand Forecasting in a 3-stage Fast-fashion Retail Supply Chain Sanchoy Das, Professor, New Jersey Institute of Technology, University Heights, Newark, NJ, 07102, United States of America, das@njit.edu, Jingran Zhang Demand forecasting is a critical issue as a premise of supply positioning in a fast fashion supply chain. This paper is focused on a fast fashion retailer that sells a single product in a finite selling season, with demand uncertainty and stage switching. Products are sold in three sequential stages, regular, clearance and outlet. We derive the demand forecast by a modified exponential-weighted moving average. WC77 77-Room 300, CC Supply Chain Competition I Contributed Session Chair: Ziteng Wang, Department of Industrial and Systems

WC76 76-Room 204C, CC Simulation III Contributed Session

Chair: Ni Xia, School of Management, Huazhong University of Science and Technology, No.1037, LuoYu Road, Hongshan District, Wuhan, 430074, China, xntx128390@163.com 1 - Exact Efficient Simulation of Stochastic Differential Equations Apaar Sadhwani, PhD Student, Stanford University, 212L, Huang Engineering Center, 475 Via Ortega, Stanford, Ca, 94305, United States of America, apaars@stanford.edu, Kay Giesecke We present practical techniques to improve the efficiency of exact simulation of SDEs. Our proposed acceptance-rejection algorithm is an order of magnitude faster than existing methods and alleviates the problem of sample paths hitting unreachable boundaries. To achieve this speedup, we develop a novel method for

452