The Potential of Evolutionary Methods in Architectural Design

Abstract

In this paper we examine the potential of combining 2D shape packing algorithms and evolutionary methods in the design process. We investigate the ways such algorithms can be used in architectural design and how they may influence it. In the first part of this paper we introduce the theoretical framework of packing algorithms and genetic algorithms as well as the traditional design process and the nature of design problems. In the second part of the paper we introduce a software prototype that tests these algorithms in two contexts: the preliminary design of a shading façade pattern and the design of commercial housing layouts. The aim for both experiments was to generate optimal configurations based on user-defined criteria without resorting to exhaustive search. Several lessons were learned that point to the potential of evolutionary methods in architecture as well as the limitations of such methods. We conclude the paper with recommendations for further developing this research project.