Flexible Job-Shop Scheduling with Batching for Semiconductor Manufacturing

Scheduling decisions in the diffusion and cleaning area of a semiconductor manufacturing facility have an important impact on the overall performance of a plant. Consequently, we want to optimize those decisions while taking real-world constraints into account. An important property of machines in this work area is their batching capability: They can perform multiple operations at the same time. We want to take account of this in our algorithm. We need to schedule a given set of jobs. For each of them, a fixed sequence of operations must be performed. This sequence is called the route of the job. Operations can only be performed on qualified machines and their processing durations depend on the selected machine. A capacity limit constrains the number of jobs that can be processed per batch. Each operation is assigned to a family and only operations of the same family can be combined in the same batch. For each job, we are given a ready date and a due date. Those constraints describe a flexible job-shop scheduling problem with batching. We aim to minimize total weighted tardiness. For the described problem, we present a simulated annealing algorithm that is based on an extended evaluation of disjunctive graphs. In our proposed approach, batching decisions are taken dynamically during graph traversal.