Optimization Algorithms for a Car Resequencing Problem

Abstract

This thesis considers a car resequencing problem (CRP) in automotive paint shops, where a set of cars conveyed from the upstream shop to one of the multiple conveyors is retrieved sequentially before painting operation. The aim of CRP is to obtain a car retrieval sequence which minimizes the sequence-dependent changeover cost in the paint shop. The changeover cost is incurred when two consecutive cars do not share the same color. In this study, we propose exact and heuristic algorithms for CRP. First, we consider a mathematical formulation of CRP and propose a branch-and-cut algorithm. Second, we present an accelerated dynamic programming algorithm by using strong combinatorial lower bounds, which outperforms existing dynamic programming algorithm. We also present several heuristic algorithms based on the proposed accelerated dynamic programming algorithm that yield good solutions quickly for large-sized instances. Computational results show that the proposed algorithms are more efficient than existing approaches and also more applicable in practice.