Module library development via graph mining

Jianpeng Cao, Daniel Hall
ETH Zurich, Switzerland

DOI: 10.35490/EC3.2021.151
Abstract: The use of prefabricated modules can benefit the construction industry with the economy of scales and production efficiency. However, the existing approach to develop module libraries is project-based, lacking the potential to reuse and manage in future projects. By taking the repeatability and manufacturability into account, this paper proposes a graph-based framework to identify possible modules automatically from multiple projects by frequent pattern mining. The results show that the repeated patterns share a degree of standardization and can be considered as module candidates. Finally, the framework is implemented as add-ons in the BIM environment to support module lifecycle management.
Keywords: module library; frequent pattern mining; BIM
Pages: 285 - 292

Presentation Video: