Design and development of effective decomposition methods for scheduling complex workshops
Abstract
This work addresses the problem of scheduling job shops with different types of workcenters, reentrant product flows, sequence-dependent setups and due-date-related performance measures, motivated by semiconductor manufacturing. In the scheduling literature, these types of problems are mostly approached by myopic dispatching rules, which may lead to poor solutions. We propose a Decomposition Procedure (DP), which decomposes the job shop into single-machine workcenters, schedules them in order of their criticality and integrates their schedules to construct a schedule for the whole facility. A tabu search is embedded into the decomposition framework to reoptimize partial schedules to generate high-quality solutions in reasonable computation time. We compare the performance of this procedure with those of several dispatching rules and heuristics using randomly generated test problems. We show that the DP outperforms the dispatching rules and heuristics in terms of solution quality. To reduce the computational burden of a DID, a careful study of its elements is needed. We observe that computation time can be decreased and solution quality can be enhanced with control structures that exploit job routings in industrial settings. We also analyze the performance of DPs in shops with well-defined bottlenecks and both setup and non-setup machines, which are closer to real-life scenarios. We show that DPs perform better in such environments. This is encouraging, as DPs become viable alternatives in real-life scenarios. Our results yield interesting insights into the strengths and limitations of DMs.
Degree
Ph.D.
Advisors
Uzsoy, Purdue University.
Subject Area
Industrial engineering|Systems design
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.