An evolutionary strategy enhanced with a local search technique for the space allocation problem in architecture, Part 2: Validation and performance tests

The first part of this paper proposed a hybrid evolutionary technique which helps architects to generate sets of floor plans in the early design stage. The algorithm is an enhanced Evolutionary Strategy (ES) with a Stochastic Hill Climbing (SHC) technique.In this second part, the validity and performance of the technique is examined. Four tests are conducted. Three validation tests are used to determine the ability of the algorithm to replicate a floor plan made by an architect; the ability to produce a set of different floor plan designs by comparing the results with a well-known enumerated floor plan solution; its capability to work with less constrained problems, and to control the variation in the floor plan form with different compactness evaluators’ weights. The last test determines the evolving behavior, the breakdown of the evaluators’ performance for the objective function, and the robustness of the algorithm.The methodologies used in the tests and the problem specifications are presented. Results and observations on the evaluators’ weights and their importance and the limitations of the technique are analyzed and discussed. Finally, the conclusion to this part of the paper is made.

► A new hybrid evolutionary computation methodology and mathematical model are presented.
► The algorithm is tested in its validation and performance.
► The proposed technique generates floor plans to be used in the early architectural design stage.