Database of flanged reinforced concrete squat walls and its utilization based on machine learning

Abstract

Flanged reinforced concrete (RC) squat walls are widely used in conventional buildings, owing to their capability of meeting certain architectural demands and providing significant lateral strength and stiffness in both principal directions. Despite their popularity and advantages, understanding of these walls is still limited due to the complexities brought by the non-rectangular cross-section and the relatively small shear span ratio. To facilitate future investigations through, a database containing comprehensive information about 182 flanged RC squat walls is presented in this paper. The criteria for selecting data and relevant sources are first introduced. Then items in the database are categorized into five groups, with histograms to assist understanding. The characteristics of these items are also discussed, and suggestions for future research are proposed. Furthermore, potential uses of this database are suggested, and the peak shear strength of these walls is explored based on machine learning as an example of utilization of this database. Compared with the mainstream building codes, this data-based approach produces much better predictions. Researchers are also encouraged to explore the use of this database through other data-driven approaches.