Scheduling approaches for the batch process industries

Abstract

The elements of scheduling, a decision methodology which is appropriate whenever competing activities must be assigned resources over time so as to achieve desired performance objectives, are outlined. The key difficulty of such problems, namely their characteristically high combinatorial complexity, is reviewed and the range of approximate as well as exact methodologies which have been advanced for their solution are highlighted. Batch process scheduling problems are shown to be highly structured and resource constrained and thus not readily amenable to the full range of generic solution approaches. Model-based optimization approaches, validation using detailed simulation models, are proposed as the most appropriate decision support technologies for this domain. The optimization models use simplified product recipes, time discretization constructions, linear material and other resource balance constraints, and application specific economic objective functions. The core solution methodology consists of mixed integer and linear programming methods, specialized to exploit model structure. The role of combined continuous-discrete simulation models is to allow adjustment of the scheduling information obtained from deterministic, linear models to account for the detailed recipes, fine grained representation of time, nonlinear state dependent effects, and stochastic operational and order information, which are characteristic of plant operations. It is proposed that the model-based optimization and simulation approaches are inherently complementary and offer the best prospects for the routine creation and validation of robust and economical operational schedules for the batch environment. Work in progress to integrate these methodologies is noted and some open issues discussed.