Scheduling batch processing: Genetic Algorithms Versus Mathematical Programming

Genetic algorithms have during the recent years gained popularity also in the domain of chemical engineering, among other applications for scheduling purposes. A genetic algorithm builds new sequences by combining and mutating previous sequences of genes, i.e. chromosomes, into a new set of chromosomes. In this new set, only the fittest survive, and the procedure is repeated. As a schedule in chemical batch plant can be seen as a sequence of starting points for the batches, the methodology of genetic algorithms can be applied also to batch scheduling. In this work, the genetic algorithm approach is used combined with a Discrete Event Simulation (DES) approach. Here the genetic algorithm determines the order of the batches, whereas the DES-approaches insert the batches in the actual schedule. Using this two-stage optimisation and some tuning of the DES procedures, schedules with similar objective function values as with mathematical optimisation can be achieved, but usually much faster, which is essential for operational scheduling systems.