Architectural design and geometry optimization

Abstract

Architecture has a long history, and humans have tried to build complex and aesthetic architectural forms since the days of the earliest civilizations. Such complex shapes satisfy functional requirements (e.g. forts, temples) as well as aesthetic ones. The design and construction cycle of such complex structures is non-trivial and often, many sub-tasks are completed by tedious manual work. Furthermore, the design obtained by such efforts may not be optimal from the point of view of constructions costs, risks, or even functionality. With the rapid development of computer science and technology, various architectural modeling, analysis and optimization techniques have been proposed in recent times to improve the level of design automation and to accelerate the design and construction processes. In this thesis, we develop geometric algorithms to focus on three sub-problems in the pipeline. First, for the problem in very early stages of architectural design, i.e., conceptual design, we develop an approach called shape inspired architectural design. The architect specifies some simple shape requirements, by inputting a few binary images that resemble the shape of the building from different viewpoints. These are used to generate a conceptual design that is guided by various aesthetic and structural requirements. To solve this problem, we develop a modified cuckoo search algorithm. Furthermore, a novel iterative mesh smoothing algorithm that yields a smooth surface while preserving sharp creases and roof structures is also developed. Second, we explore an interactive technique in planning the space within given architecture model with user-specified 3d room elements. The architect therefore specifies the ideal requirements on the key functional rooms by using instances of some simple primitive shapes. The size and location of each instance is a parameter in a global optimization search process. By coupling this optimization together with the shape inspired design algorithm, we can construct good initial conceptual designs. Third, we tackle a problem in rationalization of the building structure. The shape of a building is defined by its facade, or skin. Building skins are fabricated with a large set of tiles, or facade panels. The panels are attached to the building structure by means of a structural frame, typically made up of a mesh of steel links connected at joints. Given an arbitrary structural mesh, we introduce a methodology called joint based architectural surface remeshing to reduce the fabrication cost. The rationalization is achieved via the use of a new k-set clustering of the joints. A heuristic is also developed to minimize the number of clusters in a given mesh by remeshing it via a sequence of small topology preserving perturbations. All of the above have been coded into software programs and tested on realistic or real-world design cases to illustrate the effectiveness of our techniques.

Date of Award	2019
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology