Reducing the number of different nodes in space frame structures through clustering and optimization. (1st June 2023)

• Record Type: Journal Article
• Authors: Liu, Yuanpeng; Lee, Ting-Uei; Koronaki, Antiopi; Pietroni, Nico; Xie, Yi Min
• Is Part Of: Engineering structures. Volume 284(2023)
• Publication Date: 2023-06-01
• Subjects: Architectural geometry -- Clustering -- Optimization -- Space frame structure -- Free-form surface -- Node design; Structural engineering -- Periodicals; Structural analysis (Engineering) -- Periodicals; Construction, Technique de la -- Périodiques; Génie parasismique -- Périodiques; Pression du vent -- Périodiques; Earthquake engineering; Structural engineering; Wind-pressure; Periodicals; 624.105
• DOI: 10.1016/j.engstruct.2023.116016
• Languages: English
• ISSNs: 0141-0296

Abstract

Abstract: Space frame structures are increasingly adopted in contemporary free-form architectural designs due to their elegant appearance and excellent structural performance. However, a space frame structure in a doubly-curved form typically comprises nodes of different shapes. This often requires extensive node customization, hence incurring high manufacturing costs. In this study, we propose a new clustering–optimization framework to reduce the number of different nodes in space frame structures. In clustering, nodes are divided into different groups, with similar shapes grouped together, using an enhanced k -means clustering technique. In optimization, nodes within the same group are transformed towards congruence while closely approximating the target surface. Together, by interleaving clustering and optimization, our method can minimize the node shape variety under a user-defined error threshold. The effectiveness of the method is validated through a variety of numerical examples. The potential practical application of our method is demonstrated by re-designing a complex, free-form architectural project. Highlights: A new method is proposed to reduce the node shape variety in space frame structures. The core of the method is based on interleaving clustering and optimization. Congruent nodes can be achieved by equalizing the corresponding angles. The method is validated through various numerical examples of free-form geometries. The practical relevance of the method is demonstrated by re-designing a real project.