Scheduling of serial multiproduct batch processes

Abstract

In this work, we address the short-term scheduling of a serial multiproduct plant or flowshop using Mixed Integer Linear Programming. We develop seventeen new flowshop models by changing the definition of binary variables, coupling two contrasting binary variables, addition of new cuts etc. with the objective of minimizing makespan. We compare the models that fall under one of slot-, sequence-, event-based and the new coupling approach in the continuous time domain representation of short-term scheduling. A simple and systematic empirical analysis of models is presented, which uses model characteristics as weights. Contrary to the conventional notion, we realize that the efficiency of the model does not depend only on the number of binary variables but also on the types of constraints that govern the model. We also develop a few Linear Programming based heuristics for larger flowshop problems and attempt reformulating a resource constrained flowshop.