SCHEDULE OPTIMIZATION WITH SIMULTANEOUS LOT SIZING IN CHEMICAL PROCESS PLANTS

Operational problems are of increasing importance as the process industries turn to the batch manufacture of high-value specialty products. A method is presented to minimize stockout for multiunit plants where customer order may constitute nonintegral batch multiples. Product- and unit-dependent processing times and batch sizes, sequence-dependent changeover times, and product-to-unit assignment constraints are modeled for nonidentical parallel processes. Arbitrary initial conditions are also modeled. The algorithm determines the number of production runs of each product, run length (lot sizing), the assignment of run to units, and run sequence on each unit. Scheduling with simultaneous lot sizing has not previously been addressed in the literature. The evolutionary method presented here obtains near-optimal (and often globally optimal) solutions on industrial-sized test problems. These results are significant since near-optimal scheduling of many industrial process plants now becomes possible, although problems of this complexity are intractable using exact methods.