MILP-based Decomposition Method for the Optimal Scheduling of an Industrial Batch Plant

摘要

This paper considers a periodic scheduling problem from a fine chemicals plant that synthesizes active pharmaceutical ingredients (API). The objective is to select from the set of available equipment units those better suited for the production of a certain product. The trade-off involved is that, as the number of selected equipments increases the makespan decreases but so does the free capacity of the plant, making it harder to accept new orders. The problem is tackled with a Resource-Task Network (RTN) discrete-time formulation, and a decomposition method is used to overcome a potentially large problem size resulting from the multiple possibilities of distributing the equipment units over the processing tasks. The results have shown that the method is very efficient, since it can find good solutions in a few minutes of computational time.