A mathematical programming based model and algorithm for a two stage production process

Abstract

The research conducted in this thesis is concerned with the study of a two-stage production process existing at the Standard Register's (previously Burroughs Corporation) Paper Products Division plant in Rocky Mount, VA. The objective is to develop a mathematical programming based model and algorithm to allocate and sequence work orders in order to improve the plant's productivity and reduce the labor and material costs. The proposed algorithm is based on Lagrangian Relaxation and Benders' decomposition techniques, which exploit the inherent generalized assignment and travelling salesman problem substructures in the model. The algorithm is computationally intensive and generates (near) optimal solutions with a reasonable amount of effort. Heuristic algorithmic procedures for the generation of good solutions for large-sized problems is also proposed.