Performance of single stage representation genetic algorithms in scheduling flexible flow shops

This is a report on the performance of a set of techniques for the solution of the problem of scheduling flexible flow shops (FFS). Four variants of the single stage representation genetic algorithm (SSRGA) and four well known dispatching rules, first in first out (FIFO), shortest processing times (SPT), longest processing times (LPT), and shortest due dates (SDD), are considered as well as the shifting bottleneck procedure (SBP). These nine approaches are applied to randomly generated instances of the FFS problem for each of four different optimisation criteria: makespan, maximum tardiness, total weighted completion times and total weighted tardiness. Because of the rather wide gaps in the previous experimental work on this topic, the motivation behind this study is to find out about the performance of dispatching rules not considered in this framework, i.e. combined with the SSRGA, before, as well as on FFS problems with different optimisation criteria. Computational results as well as implementation details are included.