A priority-rule based method for scheduling in chemical batch production

We present a priority-rule based method for a scheduling problem arising in chemical batch production. The problem consists in scheduling a set of operations on the processing units of a multi-purpose chemical batch plant in such a way that the production makespan is minimized. An operation transforms one or several input products into one or several output products. In general, each operation can be executed on several alternative processing units. Certain intermediate products can be stocked in storage facilities of limited capacity, whereas chemically instable intermediates have to be consumed immediately. Moreover, a changeover may be necessary between the processing of different operations on the same processing unit. The problem is distinguished by the presence of both material-availability and storage-capacity constraints. Consequently, already the problem of finding a feasible schedule is NP-hard. That is why classical priority-rule based methods cannot be used. In the literature, a two-phase approach has been proposed, dealing with both kinds of inventory constraints separately. In the present paper we devise a new priority-rule based method where the two types of constraints are treated jointly. The main idea is to allow for partial schedules causing capacity overflows at storage facilities and to deduce temporary latest start times of operations from the time points at which the capacity overflows occur. Each time such a latest start time cannot be met, an unscheduling procedure is called, which introduces appropriate release dates for operations having caused the respective overflow. We report on the results of an experimental performance analysis comparing the new method to state-of-the-art algorithms from the literature. The analysis shows that in particular for large problem instances, our method is able to significantly improve upon the results of the previous algorithms.