2015 Informs Annual Meeting

WC03

INFORMS Philadelphia – 2015

WC02 02-Room 302, Marriott Scheduling III Contributed Session Chair: Pravin Tambe, Dr., RCOEM, 49, Parate Nagar, Near H.B.Estate, Sonegaon, Nagpur, MS, 440025, India, tambepp@gmail.com 1 - Hybrid Flow Shop Batch Scheduling Problem with a Bi-Criteria Objective Rasaratnam Logendran, Professor, Oregon State University, School of Mech, Indust, and Mfgr. Engr, 204 Rogers Hall, Corvallis, OR, 97331-6001, United States of America, logendrr@engr.orst.edu, Omid Shahvari We address a batch scheduling problem in hybrid flow shops, wherein one or more stages have unrelated-parallel machines. The objective is to minimize the weighted sum of total weighted completion time and total weighted tardiness. Job release times and machine availability times are assumed to be dynamic. The performance of search algorithms, based on tabu search, is evaluated by developing a mixed-integer linear programming model in order to find the best algorithm, if any, for this problem. 2 - Flowshop Batch Processing Problem with Different Batches on Machines Nasser Salmasi, Sharif University of Technology, Department of Industrial Engineering,, Sharif University of Technology, Tehran, Te, 11365, Iran, nsalmasi@sharif.edu, Hossein Nick Zinat Matin, Mohsen Varmazyar We approach the multi-stage flowshop batch processing problem with minimization of makespan. Each batch on each machine has both the maximum number of jobs in each batch and the batch capacity limitations at the same time. The size of batches on machines can be different. The jobs can be assigned to different batches on different stages. We propose a general mathematical model and a metaheuristic algorithm based on particle swarm optimization (PSO) to solve the problem. 3 - Adjustable Robust Optimization for Handling Uncertainty in Process Scheduling Nikolaos Lappas, Graduate Student, Carnegie Mellon University, DH3122, 5000 Forbes Avenue, Pittsburgh, PA, 15212, United States of America, nlappas@cmu.edu, Chrysanthos Gounaris We develop an Adjustable Robust Optimization (ARO) framework to address uncertainty in Process Scheduling. Unlike RO, which results in a here-and-now solution, ARO results in a policy that is a function of parameter realizations. We derive the ARO counterpart, propose suitable decision-dependent uncertainty sets, and describe a cutting-plane-based solution approach. ARO results in considerably less conservative solutions, and unlike RO, can address also instances with zero-wait tasks. 4 - Robust Scheduling of Parallel Machines under Uncertainty Selçuk Güren, Abdullah Gul University, Dept. of Industrial Engineering, Kayseri, Turkey, selcuk.goren@agu.edu.tr, Seyma Bekli We consider stochastic scheduling of a set of jobs subject to processing time variability on identical parallel machines subject to random breakdowns. We develop a scenario-based integer-programming model that can handle small-size problems without machine breakdowns. We then propose a heuristic algorithm that can also handle large problems with machine breakdowns. Our computational experiments indicate that the performance of the proposed algorithms is promising. 5 - A Simulated Annealing Approach for Scheduling Jobs on Identical Parallel Machines Pravin Tambe, Dr., RCOEM, 49, Parate Nagar, Near H.B.Estate, Sonegaon, Nagpur, MS, 440025, India, tambepp@gmail.com, Makarand Kulkarni This paper presents a scheduling problem on identical parallel machines. A hybrid approach using simulated annealing combined with backward-forward heuristic is used for scheduling. The objective is to minimize the total penalty cost of jobs allocation on all the machines. Both earliness and tardiness penalties are considered. The computational experiment results for different problem sizes have been presented.

WC03 03-Room 303, Marriott Inventory Management for Supply Chains I Contributed Session Chair: Alireza Sheikhzadeh, University of Arkansas, 4207 Bell Engineering Center, Fayetteville, Ar, 72701, United States of America, asheikhz@uark.edu 1 - Multi-Period Dynamic Inventory Classification: Models and Applications Dorothy Liu Yang, University of Missouri - St. Louis, 240 JCPN, One University Blvd, Saint Louis, MO, 63121, United States of America, liuyang@umsl.edu, Haitao Li This talk presents a multi-period inventory classification model that aims to help company optimize profitability when facing nonstationary, stochastic demand. The benefits of multi-period, dynamic inventory policies over single-period, static inventory policies are examined. We offer managerial insights about how various purchasing and inventory parameters/settings impact the optimal dynamic policy and performance measures. 2 - Service-level Estimation in Inventory System Simulations with Model Uncertainty Canan Gunes Corlu, Assistant Professor, Boston University, We consider the simulation of a stochastic inventory system in the presence of input model uncertainty — the input distributions are unknown and there is only a limited amount of historical input data available for input distribution estimation. We develop an algorithm that estimates the mean service level of the inventory system accounting for the model uncertainty without making any assumptions on the form of the input model(s) that can represent the data on hand. 3 - Service Level and Contract Design with Shelf Space Dependent Demand Peeyush Mehta, Indian Institute of Management Calcutta, NF3/4, IIM Caclcutta, Kolkata, 700104, India, pmehta@iimcal.ac.in, R K Amit We consider supply chain coordination problem when demand is influenced by the amount of inventory displayed on the shelves. First, we determine an optimal service level for shelf space dependent demand. Next, we design individually rational contracts that coordinate the supply chain when the retailer faces inventory level dependent demand. 4 - Incorporating Order Crossover Information in Service-oriented Base Stock Policy Decisions Alan Pritchard, University of Maryland, Robert H. Smith Scholl of Business, Van Munching Hall, College Park, MD, 20742, United States of America, apritchard@rhsmith.umd.edu, Dean Chatfield Order crossover, when orders arrive in a different sequence than the sequence in which they were placed, is a prevalent issue for modern supply chains. We utilize a hybrid discrete/continuous simulation model to investigate multiple methods of including order crossover information in the safety stock decision for an (R,S) inventory system with a service level target. Issues related to effective lead times, the normal approximation, and the appropriate protection period are investigated. 5 - Segmentation Methods for Large-scale Multi-echelon Repairable Parts Provisioning Systems Alireza Sheikhzadeh, University of Arkansas, 4207 Bell Engineering Center, Fayetteville, Ar, 72701, United States of America, asheikhz@uark.edu, Manuel Rossetti Large-scale manufactured systems are an extremely crucial part of the infrastructure and worldwide competitiveness. Innovative approaches will significantly reduce the size and complexity of problems through the use of segmentation methods. In this research, grouping SKUs based on a no-backorder stocking policy (NBS) considerably reduced the size of problem with the lowest penalty cost. This procedure represents a unique segmentation model that has not been considered in the literature before. 808 Commonwealth Avenue, Boston, MA, 02215, United States of America, canan@bu.edu, Alp Akcay

427