Scheduling of a multi-product batch process in the chemical industry

Abstract

We present an example of a mixed-integer linear programming (MILP) model for the scheduling of a multi-product batch process occurring in the chemical industry. The batch process considered is organized in several stages. Various final products are produced out of a single feedstock by a number of chemical processes. The major scheduling objective is to minimize the makespan, i.e., to complete the required production operations within the shortest possible time. The complexity of the scheduling problem is determined by such factors as variable batch sizes, shared intermediates, flexible proportions of output goods, blending processes, sequence and usage dependent cleaning operations, finite intermediate storage, cyclical material flows, and no-wait production for certain types of products. Due to the fact that computational times are prohibitive for problems of realistic size, we developed various LP-based heuristics. The heuristics proposed are applied to relaxations of the original multi-period MILP model. Thus, computational results are obtained a magnitude faster. Furthermore, near-optimal solutions are made possible for larger problems within reasonable computational time. In order to evaluate the applicability of the heuristics, a number of numerical experiments were performed.