Integration of scheduling and control for batch process based on generalized Benders decomposition approach with genetic algorithm

Abstract

Abstract Batch process plays an important role in various fields of industrial production, such as chemical engineering and pharmacy, given its characteristics such as small batch size, flexible production, and additional value of the product. Efforts to integrate the scheduling and process control for improving the benefits of batch process are recent. The integration of scheduling and process control is described by state equipment network which is closely related to the processing variables due to the feature of the network structure of the batch process where material splitting and mixing are allowed. The integrated formulation invokes logical disjunctions and operational dynamics which represents a typical mixed-logic dynamic optimization (MLDO) problem. To solve such a MLDO problem, we transform it into a mixed-integer nonlinear program (MINLP) using the Big M reformulation and the simultaneous collocation method. Then, the MINLP problems are solved through a generalized Benders decomposition (GBD) approach and genetic algorithm. The decomposed master problem is a scheduling problem with variable processing times, processing costs, and the Benders cut. Accordingly, the genetic algorithm is implemented to increase benefit. The primal problem comprises a set of separable dynamic optimization problems in the processing units. By collaboratively optimizing the process scheduling and dynamics, the proposed method substantially improves the overall economic performance of the batch production. At last, the feasibility and superiority of the proposed integration model and optimization algorithm can be determined by dealing with specific production instances.