An MILP framework for Short-Term Scheduling of Single-Stage Batch Plants with Limited Discrete Resources

Abstract Dealing with limited discrete resources in batch scheduling problems usually produce a sharp increase in model size and computational requirements. This work introduces a novel MILP formulation where all discrete resources including processing units are treated uniformly. Moreover, the ordering of batches at any resource item is handled by a common set of sequencing variables so as to achieve an important saving in 0–1 variables. Pre-ordering rules significantly reducing the problem size can be easily embedded in the MILP framework. In addition, discrete resources could even be sequentially assigned when real world resource-constrained scheduling problems are tackled. Two examples involving the scheduling of up to 29 batches in a single-stage batch plant under severe manpower restraints were successfully solved. Comparison with prior work shows a notable reduction in CPU time of at least two orders of magnitude.