著者

野畑 剛史 本間 裕大 今井 公太郎

出版者

日本建築学会

雑誌

日本建築学会計画系論文集 (ISSN:13404210)

巻号頁・発行日

vol.84, no.766, pp.2545-2552, 2019 (Released:2019-12-30)

参考文献数

16

---

In this paper, we propose a new morphological analysis method to evaluate architectural space. Specifically, we focus on the inner convex spaces as a partial inner space and enumerate all possible patterns. Since people can recognize each other in a convex space, they share five senses such as seeing each face, talking without being distracted by any obstacles. It means that a convex space is appropriate as one unit in the architectural space. Recently, some real architectural buildings have complicated and amorphous shapes with seamless special connections. In such buildings, the convex space could be a useful tool to comprehend the spatial composition. Therefore, we regard the building as concave polygons and enumerate all possible patterns of maximal inner convex space. The maximal convex spaces are enumerated in the following procedure. First, the possible candidates for the endpoints of the maximal convex spaces are enumerated as much as possible. The candidate points can be obtained by rotating lines around each reentrant angle of the objective concave polygon. Next, by confirming visibility, half-plane, and tangent line conditions, we create an adjacency matrix for candidate points. Finally, based on this adjacency matrix, we enumerate all possible maximal cliques, which correspond the maximal convex spaces for the analysis. Derived inner convex spaces can be applied to various architectural planning issues. For example, we can clarify the characteristics of inner spaces such as area, circularity. These results indicate the diversity of usage in objective architectural space. Furthermore, the distributions of inner spaces correspond to the openness of space. If space is overlapped by various convex space, it is for public usage and vice versa. We also analyze the connectivity of each inner convex space by connecting centroids by a minimum spanning tree (MST) based on the similarity. This structure would be the backbone of architectural space. Since we enumerated all possible patterns of inner spaces, it enables us to find the optimal inner spaces subject to various conditions. The optimal inner space which maximizes area, circularity, and so on, helps to understand the abilities of buildings. We believe that these new analyses expand the potential of quantitative researches for architectural planning. This study can be regarded as a critical extension of Isovist theory which has been used in numerous earlier studies because a scanning vector which constructs the Isovist is a proper subset of some maximal convex spaces. All of the above computational procedure can be completed within a realistic time, and we believe that this study proposes a new possibility of morphological analysis for architectural space. Future prospects include developing methods applicable to more complex buildings, extending the method to three dimensions, and establishing useful search methods.