Batch scheduling on parallel machines with dynamic job arrivals and incompatible job families

Abstract

We study the scheduling problem of minimising weighted completion times on parallel identical batching machines with dynamic job arrivals and incompatible job families. Each job is associated with a family, weight (priority), release time, and size. Batching machines can process simultaneously up to a specified total size of the jobs of a particular family. The scheduling problem can be represented by . We present a mathematical model and heuristic algorithms employing different local search procedures individually and sequentially under a variable neighbourhood search scheme. We have shown that among local searches, repositioning the batches instead of jobs yields better results. The best-performing heuristic algorithm is capable of generating solutions within 0.6% of the best overall heuristic solution for each instance in a reasonable amount of time. When this heuristic is compared against the mathematical model, solutions that are 3.7% above optimal on average in the 15-job problem instances are possible.

Keywords:

• scheduling
• heuristics
• batch scheduling
• mathematical modelling
• variable neighbourhood search