Production scheduling in assembly/disassembly systems

Abstract

This thesis considers scheduling problems concerned with manufacturing systems including assembly and disassembly features where each produced product is the final product in the assembly system but it is a component for the associated subsequent lines in the disassembly system. In the assembly system, several types of components are produced first at pre-assembly stage of feeding machines and then assembled into final products(jobs) at an assembly stage. Each assembly operation cannot start its processing until all of the associated components are prepared from the precedent pre-assembly stage. In the disassembly system, a disassembly operation for splitting a product into several components is performed and then subsequent parallel machines work with their respective components. The first problem to be considered in this thesis is to minimize the total weighted completion time of jobs at the pre-assembly stage. The problem is shown to be NP-complete in the strong sense even when there are only two feeding machines. In the analysis, two lower solution bounds are derived and tested in a proposed branch-and-bound scheme. Some solution properties are also characterized, based upon which three heuristic algorithms are derived. The effectiveness of the heuristics is tested with numerical problems and a worst-case analysis is also done. Secondly, three problems are considered to minimize makespan in a two-stage(pre-assembly stage and final assembly stage) assembly system, a disassembly system, and a combined assembly/ disassembly system, respectively. In the analysis, several solution properties for dominating schedules are characterized, based on which it is proved that the first two problems are equivalent. These results and Johnson*s rule are used to develop a heuristic algorithm. For the last combined system, two meta -heuristic algorithms, taboo search and simulated annealing, are suggested. The last problem is to minimize total job completion time in a two-...