Low-energy design: combining computer-based optimisation and human judgement

Abstract

Simply minimising the heat loss from a building will not necessarily lead to an exemplary low-energy design: overheating may occur, leading to a large amount of cooling energy being used, and the shape and form of the design may not fit with other sensitivities and elements of the design brief. This paper couples a population-based optimisation algorithm (a genetic algorithm) to a dynamic thermal model with the idea of identifying large numbers of distinctly different low-energy designs. These designs are then presented to the user in the form of a visual summary for judgement as to potential use. In order that sufficiently different designs are evolved, and the thermal model can be run over a complete year on an hourly grid, several adaptations to the genetic algorithm have had to be made. The approach is illustrated by the design of a community hall. An extensive range of design possibilities is identified which achieve low-energy status by greatly different means with some concentrating on reducing losses and others on maximising their use of causal gains, including solar gains.