Structural Exploration of Constraint-Based Scheduling Problems

Constraint-based scheduling (CBS) is a highly efficient and widely applied method in solving resourceconstrained project scheduling problems – including job-shop scheduling problems. However, many recent studies suggest that practical problems highly differ from benchmark instances used in theoretical scheduling research, generated intentionally to challenge solvers. Although industrial problems require a rich and large-size model, they are simple in the sense that they often have a loosely connected structure of easy and hard subproblems. In this paper, we present how we extended a general CBS framework by methods that exploit these structural properties. Our system was validated on industrial job-shop scheduling problem instances containing up to 2000 tasks, and succeeded in solving problem instances that previously remained unsolvable due to their sheer size.