Evolving design genes in space layout planning problems

The paper describes the application of a genetic engineering based extension to genetic algorithms to the layout planning problem. We study the gene evolution which takes place when an algorithm of this type is running and demonstrate that in many cases it e ectively leads to the partial decomposition of the layout problem by grouping some activities together and optimally placing these groups during the rst stage of the computation. At a second stage it optimally places activities within these groups. We show that the algorithm nds the solution faster than standard evolutionary methods and that evolved genes represent design features that can be re-used later in a range of similar problems.