Genetic algorithm for batch sequencing

Abstract

[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT REQUEST OF AUTHOR.] Batch sequencing is an elementary function in production control in a manufacturing facility due to increased customer expectations and fierce global as well as domestic competition. In the light of increasing global competition, companies are increasingly required to deliver made-to-order products in shorter production time and on schedule. As a result, production planning and control is under constant pressure to shorten lead times and meet due dates. However, in this research study we do not follow an earliest due date approach. We weigh all batches equally and try to find an optimal sequence with least changeover time. Most production planning and control problems are computationally inflexible in terms of an optimal solution and as a result different dispatching rules are used. Yet due to the current advances in computer and information technology, it is virtually viable and cost effective to implement an optimization based sequencing strategy that utilizes comprehensive information in order to improve shop performance. Optimal batch sequencing is attributed to cost reduction and profit maximization. As sequencing relates to operational cost and performance it directly affects an organization's survivability. Batch sequencing can be approached by Travelling Salesman Problem (TSP) framework. In this case, batch insertion sequence represents cities. The cost/distance matrix represents the changeover times. The TSP framework may consist of precedence constraints, reflecting that certain batch must be inserted before others. TSP has a starting node (location) which is a home location. In this thesis, we do not account for precedence constraints in order to get a least cost sequence. In this research, we develop a genetic algorithm (GA) that generates a least cost sequence. The results obtained by GA are validated by mathematical model branch & bound algorithm.