A genetic approach to multi-product modular supply network redesign

For many industries, capacity reconfigurations need not be restricted solely to facility shutdown or new construction activities, but may incorporate modular reconfigurations of pre-existing capacities. Supply network models to date, attempting to deal with flexibly configured sites have restricted consideration to "all-or-nothing" production capability types and not with modularity per se. This study presents a first attempt at modeling the multi-product modular supply network reconfiguration problem. Sample cases are solved via a genetic algorithm (RAGA) designed specifically to deal with this modularity. The algorithm provides solutions within 2.5% of the optimal while requiring about one tenth of the time utilized by an industry standard package (LINDO).