Combining Heuristics and Integer Programming for Optimizing Job Shop Scheduling Models

In this thesis, we review and synthesize a number of algorithmic approaches for optimizing the job-shop scheduling problem (JSSP). The complexity of the JSSP leads to large and difficult combinatorial optimization models. In Chapter 1, we review first an analytical method based on Mixed Integer Programming (MIP). Second, we review an assumption-based heuristic method. In Chapter 2, the JSSP is solved by the Branch and Bound algorithm for MIP using the CPLEXTM Callable Library. Experiments with the Branch and Bound search parameters are discussed. In Chapter 3, we review first Benders' Decomposition for MIP and the JSSP. Then, we discuss an approach combining Benders' Decomposition and the assumption-based heuristic. Experimental results with that approach are discussed. In Chapter 4, we discuss areas of future research and computational experimentation with the MIP and heuristic method for the JSSP. Thesis Supervisor: Jeremy F. Shapiro Title: Professor of Operations Research and Management