Decomposition heuristics for complex job shop scheduling
Abstract
We address the problem of scheduling complex job shops characterized by well-defined bottleneck machines, various types of machines and due date related performance measures. The motivating application is semiconductor wafer fabrication. The main theme of this thesis is to develop effective decomposition methods for scheduling problems encountered in semiconductor wafer fabrication (FAB), which we model as complex job shop. To the best of our knowledge, no researchers in the past have solved complex job shop scheduling problems with similar complexity to our problems. The research in job shop with distinct bottleneck machines shows that the Shifting Bottleneck (SB) procedure is well suited for solving shops with unbalanced loading among machines. The research of a flow shop problem involving a batch processing machine and a unit-processing machine with setup requirement explores heterogeneous decomposition of a complex job shop. The results shows the subproblem of such decomposition can be solved efficiently and hence have great potential in future research. The research of scheduling a batch processing machine with dynamic job arrival provides an important building block for developing decomposition method of complex job shop. Finally, we design a decomposition heuristics using both temporal and entity based decomposition for solving the scheduling problems associated with a FAB. Computational experiments show that our Rolling Horizon (RH) algorithm outperforms a number of dispatching rules that are widely used in practice. The results demonstrate that our decomposition heuristics has great potential to be applied in practice.
Degree
Ph.D.
Advisors
Uzsoy, Purdue University.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.