A Biased-Randomized Discrete-Event Heuristic for the Hybrid Flow Shop Problem with Batching and Multiple Paths

Abstract

Based on a real-life use-case, this paper discusses a manufacturing scenario where different jobs be processed by a series of machines. Depending on its type, each job must follow a pre-defined route in the hybrid flow shop, where the aggregation of jobs in batches might be required at several points of a route. This process can be modeled as a hybrid flow shop problem with several additional but realistic restrictions. The objective is to find a good permutation of jobs (solution) that minimizes the makespan. Discrete-event simulation can be used to obtain the makespan value associated with any given permutation. However, to obtain high-quality solutions to the problem, simulation needs to be combined with an optimization component, e.g., a discrete-event heuristic. The proposed approach can find solutions that significantly outperform those provided by employing simulation only and can easily be extended to a simheuristic to account for random processing times.