Scheduling hybrid flow shops with time windows

Abstract

Hybrid flow shops can be encountered in various industrial settings. In this paper we develop methods for scheduling hybrid flow shops with hard time windows. Specifically, we study a two-stage hybrid flow shop scheduling problem with time windows to minimize the total weighted completion times. Each stage consists of one or more identical parallel machines, and each job visits two processing stages in series. Finding a feasible schedule with hard time windows is a challenging task in this setting, because it is NP-complete in the strong sense even for a single machine in a single stage. We propose two matheuristics to find an initial feasible solution by local branching. We also develop two schedule improvement procedures, one based on stage-by-stage decomposition, and one using adapted local branching. The performance of our methods is validated via extensive computational experiments.