Integrated behavioural analysis of FRP-confined circular columns using FEM and machine learning

Abstract

This study investigates the structural behaviour of double-skin columns, introducing novel double-skin double filled tubular (DSDFT) columns, which utilise double steel tubes and concrete to enhance the load-carrying capacity and ductility beyond conventional double-skin hollow tubular (DSHT) columns, employing a combination of finite element model (FEM) and machine learning (ML) techniques. A total of 48 columns (DSHT+DSDFT) were created to examine the impact of various parameters, such as double steel tube configurations, thickness of fibre-reinforced polymer (FRP) layer, type of FRP material, and steel tube diameter, on the load-carrying capacity and ductility of the columns. The results were validated against the experimental findings to ensure their accuracy. Key findings highlight the advantages of the DSDFT configuration. Compared to the DSHT columns, the DSDFT columns exhibited remarkable 19.54 % to 101.21 % increases in the load-carrying capacity, demonstrating improved ductility and load-bearing capabilities. Thicker FRP layers enhanced the load-carrying capacity up to 15 %, however at the expense of the reduced axial strain. It was also observed that glass FRP wrapping displayed 25 % superior ultimate axial strain than aramid FRP wrapping. Four different ML models were assessed to predict the axial load-carrying capacity of the columns, with long short-term memory (LSTM) and bidirectional LSTM models emerging as superior choices indicating exceptional predictive capabilities. This interdisciplinary approach offers valuable insights into designing and optimising confined column systems. It sheds light on both double-tube and single-tube configurations, propelling advancements in structural engineering practices for new constructions and retrofitting. Further, it lays out a blueprint for maximising the performance of the confined columns under the axial compression.

Keywords

• Confined concrete
• Double steel tube
• Finite element model
• FRP reinforcement
• Strength enhancement
• Machine learning