Decomposition based algorithm for the design and scheduling of multipurpose batch plants

Abstract

The optimal design of plants incorporating multiple resources, whose main characteristic is their intrinsic flexibility, it is a complex task, since the design of the plant resources and scheduling should go together. Due to the nature and dimension of these problems, they often result into large Mixed Integer Linear Program formulations (MILP) that come associated with a high computational burden. In order to overcome this difficulty, a decomposition algorithm has been developed, which is described in this paper. It comprises two different levels (a master and a sub-problem) that interact through an iteration procedure that guarantees optimality. The high level consists in a time aggregation model, where the main equipment choices are defined. These choices once established, are transferred and used as input at the lower level, where the optimal scheduling coupled with resources capacities is calculated. The algorithm performance is analysed and improved with the addition of three classes of cuts, while retaining control of the run length of the sub-problem solution. Some examples illustrating the effectiveness of the proposed decomposition approach are presented.