Genetic algorithms for multiobjective scheduling of combined batch/continuous process plants

Systems in the process industry commonly incorporate both batch and continuous processes. These processes must be scheduled to satisfy product specifications, requirements from downstream processes and physical plant constraints. In doing so, the need to maximise various production objectives within a highly constrained environment can present an extremely difficult problem. This paper demonstrates the difficulties of attempting to schedule the combination of discrete tasks of varying cycle times with continuous elements. Two implementations of a heuristic genetic algorithm (GA) approach are demonstrated on a processing problem that has similar characteristics to a sugar mill. The implementations include problem-specific representations, single and multiobjective approaches to handle the four problem costs, and various uses of penalty functions to avoid constraint violations. In addition, highly tailored problem-specific operators allow the GA to match its behaviour to the critical elements in the problem definition, specifically those relating to a highly constrained shared storage facility. The results and implications of using such techniques for this type of problem are presented and discussed