An integrated simulation–optimization method for flexible assembly job shop scheduling with lot streaming and finite transport resources

Abstract

Lot streaming (LS) in manufacturing enhances production efficiency but often leads to increased job waiting times and prolonged makespan when combined with assembly job shop scheduling. The practical constraint of finite transport resources (TRs) is prevalent in practice but is frequently overlooked or simplified. This study addresses the Flexible Assembly Job Shop Scheduling Problem incorporating LS and finite TRs (FAJSSP-LS-TRs), where jobs are divided into equal-sized sublots, each requiring specific TRs with limited capacity for movement between machines and assembly stations. To efficiently explore the complex interplay between time-sensitive machining, transportation, and assembly activities within a network of heterogeneous TRs, we introduce an integrated simulation–optimization approach to minimize the makespan, using a self-repair genetic algorithm to optimize LS strategy, job operation allocation, machine processing sequence, and TRs designation. A case study in a furniture manufacturing workshop yields several pivotal insights: Firstly, practical constraints like setup time and finite TRs significantly affect the benefits of job splitting for makespan reduction. Secondly, compared to a single lot sizing strategy, a mixed lot sizing (LS-Mix) strategy leads to makespan reductions across varying-sized problem instances. Thirdly, the LS-Mix strategy outperforms the approach of merely increasing the number of TRs. Finally, managerial insights are derived.