Hybrid genetic algorithms with immunisation to optimise networks of retail outlets

Spatial interaction models are used to simulate interactions between a set of demands in a given area and a set of facilities, products and services that meet these demands. When coupled with some optimisation scheme, these models can be used to plan networks of outlets that provide maximum sales, profit or service results. Software developed by GMAP Ltd. combines a spatial interaction model with a heuristic optimisation technique, called the idealised representation plan (IRP), to find profitable networks of car dealerships; this software is used by Ford of Britain to help in planning their sales delivery system. This paper discusses the use of genetic algorithm (GA) techniques in place of the IRP, to optimise the configuration of networks of Ford dealerships in the UK. In particular, results show that, for this real business application, incorporation of immunisation and hybridisation techniques improves the GA's effectiveness and efficiency.