Using MILP and CP for the Scheduling of Batch Chemical Processes

摘要

A hybrid framework that uses Mathematical and Constraint Programming for the scheduling of batch chemical processes is proposed. Mathematical programming is used for the high-level optimization decisions (number and type of tasks, and assignment of equipment units to tasks), and Constraint Programming is used for the low-level sequencing decisions. The original problem is decomposed into an MILP master problem and a CP subproblem. The master MILP is a relaxation of the original problem, and given a relaxed solution, the CP subproblem checks whether there is a feasible solution and generates integer cuts. The proposed framework is based on the hybrid algorithm of Maravelias and Grossmann, and can be used for different objective functions and different plant configurations. In this paper we present the simplifications and enhancements that allow us to use the proposed framework in a variety of problems, and report computational results.