Three-Dimensional Shape Generation of Low-Energy Architectural Solutions using Pareto Genetic Algorithms

Abstract

This paper extends on a previous work on the application of a Generative Design System [GDS] to the evolution, in a computational environment, of three-dimensional architectural solutions that are energy-efficient and adapted to the climatic environment where they are located. The GDS combines a well-known building energy simulation software [DOE2.1E] with search procedures based on Genetic Algorithms and on Pareto optimization techniques, successfully allowing to tackle complex multi-objective problems. In the experiments described, architectural solutions based on a simplified layout were generated in response to two often-conflicting requirements: improving the use of daylighting in the space, while controlling the amount of energy loss through the building fabric. The choice of a cold climate like Chicago provided an adequate framework for studying the role of these opposing forces in architectural form generation. Analysis of results shows that building characteristics that originate successful solutions extend further than the building envelope. Issues of massing, aspect ratio, surface-to-volume ratio, orientation, and others, emerge from the analysis of solutions generated by the GDS, playing a significant role in dictating whether a given architectural form will prove adapted to its climatic and energy requirements. Results suggest that the questions raised by the exploration of form generation driven by environmental concerns are complex, deserving the pursuit of further experiments, in order to better understand the interaction of variables that the evolutionary process congregates.