Performance evaluation of building designs with BIM-based spatial patterns

Abstract For design performance evaluation, this study proposes a novel method based on the theory of Pattern Language by Christopher Alexander. The proposed method evaluates the performance of designed spaces with three components: standard patterns, BIM model, and evaluation logic. The standard patterns are defined as optimal object configurations that can maximize the spatial performance, and quantitative evaluation is done by comparing the patterns extracted from the BIM model to the standard patterns. The evaluation logic includes 1) algorithm for extracting patterns from the BIM model, and 2) algorithm for computing performance scores with evaluation equations and weights. A prototype evaluation system was implemented to verify the method through a case study for apartment kitchen spaces. The result of the case study shows that the performance scores of the kitchen spaces computed by the prototype system were similar to those of human architects. The proposed method is expected to enable design practitioners to evaluate designed spaces and improve their performance by reflecting the corrections from the evaluation of original designs before the actual construction of the spaces.