A unified solution framework for flexible job shop scheduling problems with multiple resource constraints

This paper examines flexible job shop scheduling problems with multiple resource constraints. A unified solution framework is presented for modelling various types of non-renewable, renewable and cumulative resources, such as limited capacity machine buffers, tools, utilities and work in progress buffers. We propose a Constraint Programming (CP) model and a CP-based Adaptive Large Neighbourhood Search (ALNS-CP) algorithm. The ALNS-CP uses long-term memory structures to store information about the assignment to machines of both individual operations and pairs of operations, as encountered in high-quality and diverse solutions during the search process. This information is used to create additional constraints for the CP solver, which guide the search towards promising regions of the solution space. Numerous experiments are conducted on well-known benchmark sets to assess the performance of ALNS-CP against the current state-of-the-art. Additional experiments are conducted on new instances of various sizes to study the impact of different resource types on the makespan. The computational results show that the proposed solution framework is highly competitive, while it was able to produce 39 new best solutions on well-known problem instances of the literature.