• Chukka, Naga Dheeraj Kumar Reddy
• Kumar, Rakesh
• Isleem, Haytham F.
• Oyebisi, Solomon
• Tang, Qiong
• Bahrami, Alireza

Abstract

Local buckling of steel and excessive spalling of concrete have necessitated the need for the evaluation of reinforced concrete columns subjected to axial compression loading. Thus, this study investigates the behaviour of concrete filled steel tube (CFST) columns and reinforced concrete filled steel tube (RCFST) columns RCFST columns under axial compression using finite element modelling (FEM) and machine learning (ML) techniques. To achieve this aim, a total of 85 columns from existing studies were analysed using FEM and simulation. The ultimate load of the generated datasets was predicted using various ML techniques. The findings showed that the columns' compressive strength, ductility, and toughness were improved by reducing transverse reinforcement spacing, increasing the number of reinforcing bars, and increasing the thickness and yield strength of the outer steel tube. Under axial compression loading, the FEM analysis provided an accurate assessment of the structural performance of reinforced concrete filled steel tube columns. Compared to other ML approaches, gradient boosting exhibited the best performance metrics with R2 and RMSE of 99.925% and 0.00708 and 99.863% and 0.00717 in training and testing phases, respectively to predict the column's ultimate load. Overall, gradient boosting can be applied in the ultimate load prediction of CFST and RCFST columns under axial compression, conserving resources, time, and cost to investigate ultimate load of columns through laboratory testing.

Topics

• phase
• simulation
• strength
• steel
• yield strength
• ductility
• machine learning