Flexible Job-shop Scheduling for Semiconductor Manufacturing with Hybrid Answer Set Programming (Application Paper)

The complex production processes in modern semiconductor manufacturing involve hundreds of operations on the route of a production lot, so that the period from lot release to completion can stretch over several months. Moreover, high-tech machines performing each of the operations are heterogeneous, may operate on individual wafers, lots or batches of lots in several stages, and require product-specific setups as well as dedicated maintenance procedures. This industrial setting is in sharp contrast to classical job-shop scheduling scenarios, where the production processes and machines are way less diverse and the primary focus is on solving methods for highly combinatorial yet abstract scheduling problems. In this work, we tackle the scheduling of realistic semiconductor manufacturing processes and model their elaborate requirements in hybrid Answer Set Programming, taking advantage of difference logic to incorporate machine processing, setup as well as maintenance times. While existing approaches schedule semiconductor manufacturing processes only locally, by applying greedy heuristics or isolatedly optimizing the allocation of particular machine groups, we study the prospects and limitations of scheduling at large scale.