Norbert Trautmann, University of Bern
Abstract:
We consider batch production of chemicals on a multi-stage, multi-product plant. The problem discussed in this talk is to compute a feasible schedule with minimum makespan such that some given primary requirements for the final products are fulfilled. The constraints include variable batch sizes, finite intermediate storage, perishable intermediates, alternative non-identical processing units, and cleanings between process changes and before idle times.
This problem decomposes into a batching problem and a scheduling problem. Most of the approaches discussed in the literature address both problems simultaneously. In contrast, we propose to solve the batching and the scheduling problem sequentially. We formulate the batching problem as a mixed-integer linear program, and we present a multi-pass procedure for the scheduling problem. Eventually, we report on the results of an experimental performance analysis.