Scheduling Unrelated Parallel Machines with Attribute-Dependent Setup Times: A Case Study

We study a practical production planning problem that was encountered by an industrial partner. Mathematically, the problem can be described as an unrelated parallel machine scheduling problem with deadlines and sequence-dependent setup times. The setup times depend on certain job attributes (e.g. material, color, size) of the successive products. For all products, dated demands are given that must be met with the means of one or multiple production orders that can be assigned to a set of eligible machines. We describe and evaluate different approaches to solve the scheduling problem. Our final algorithm is a combination of an iterated greedy construction heuristic together with a constraint program to locally improve the solution found by the heuristic. For most of our test instances, our algorithm achieves more than 20% reduction of setup times compared to the original solutions used by our industrial partner.