Combining Constraint Programming and Temporal Decomposition Approaches - Scheduling of an Industrial Formulation Plant

摘要

This contribution deals with the development of a Constraint Programming (CP) model and solution strategy for a two-stage industrial formulation plant with parallel production units for crop protection chemicals. Optimal scheduling of this plant is difficult: a high number of units and operations have to be scheduled while at the same time a high degree of coupling between the operations is present due to the need for synchronizing charging and discharging operations. In the investigated problem setting the formulation lines produce several intermediates that are filled into a variety of types of final containers by filling stations. Formulation lines and filling stations each consist of parallel, non-identical sets of equipment units. Buffer tanks are used to decouple the two stages, to increase the capacity utilization of the overall plant. The CP model developed in this work solves small instances of the scheduling problem monolithically. To deal with large instances a decomposition algorithm is developed. The overall set of batches is divided into subsets which are scheduled iteratively. The algorithm is designed in a moving horizon fashion, in order to counteract the disadvantages of the limited lookahead that order-based decomposition approaches typically suffer from. The results show that the complex scheduling problem can be solved within acceptable solution times and that the proposed moving horizon strategy (MHS) yields additional benefits in terms of solution quality.