Heuristic approaches for scheduling jobs in large-scale flexible job shops

In this paper, we discuss scheduling problems for flexible job shops that contain identical and unrelated parallel machines. The performance measure is the total weighted tardiness (TWT). The main contribution is an efficient iterative local search approach for flexible job shops with TWT measure using the disjunctive graph representation. Escaping from local optima is ensured by using the simulated annealing acceptance criterion. The assessment of moves is based on a dynamic topological ordering of the disjunctive graph. We hybridize the shifting bottleneck heuristic (SBH) with the proposed local search approach and a variable neighborhood search (VNS) approach. In addition, list scheduling techniques for a variety of due date-oriented dispatching rules are discussed. The proposed heuristics are compared by computational experiments for problem instances available in the literature and a set of new large-size problem instances. The local search scheme is able to determine high-quality solutions within a short amount of computing time. If the processing flexibility increases, i.e. the number of parallel machines, the improvement of the advanced techniques compared to list scheduling techniques decreases. In case of identical parallel machines, the SBH-type algorithms outperform the local search scheme with respect to TWT. However, the SBH requires more computing time. Scheduling problems for flexible job shops with TWT objective are considered.We propose an efficient local search (LS) approach.The shifting bottleneck heuristic is hybridized with LS and VNS schemes.Computational experiments demonstrate that the heuristics perform well.