Large-Scale Scheduling with Routing, Batching and Release Dates for Wafer Fabs using Tabu Search

Semiconductor fabricating facilities (wafer fabs) aim for manageable, heuristics that can reproducibly solve their typical large-scale dynamic complex job shop scheduling problems. Hence, the main contribution of this work is a deterministic tabu search algorithm that sequences, routes and batches the process operations of a continuous reentrant material flow using a large-scale mix of single, flexible (parallel) and batch machines. Due to continuous production, release dates and due dates are considered to minimize job tardiness. Using known large-scale benchmark instances, sixteen neighborhood structures are validated to find a practical compromise between solution quality and computational effort. The result is a novel combination of neighborhoods for simultaneous sequencing, routing and batching that finds good schedules for problem instances of industrial size and complexity in an acceptable time.