Feature-Based Decomposition of Architectural Spaces: Outline of a Procedure and Research Challenges

Abstract

Space modeling is a key activity in building design that is supported by BIM authoring systems. However, decomposing spaces with specific properties, such as spaces with non-convex volumes, into disjoint partial spaces, or sub-spaces, is a manual and non-trivial task. This study identifies the need for automated 3d space decomposition methods. Informed by a review of related space and shape decomposition methods, the problem scope for feature-based space decomposition is defined. A procedure is outlined that addresses this problem. The procedure recursively decomposes a space into sub-spaces using cutting operations derived from recognized features. The decomposition of a (sub-)space ends if it meets decomposition goals or if no features are recognized. The procedure is divided into space evaluation, feature recognition, and space cutting phases. The recursive decomposition of a space may be represented as a tree where nodes correspond to iterations of the procedure and edges to sub-spaces. Research challenges related to architectural feature taxonomies, feature recognition, and cutting operations are identified.