Axial Strength of Short Hot-Rolled Steel Equal Angle Members: Experimental Analysis, Numerical, and Machine Learning Modeling

Equal angle sections used as columns exhibit bending or buckling behavior under axial loads, influenced by their shape and potential for uneven load distribution. Their performance is significantly affected by factors such as slenderness ratio, boundary conditions, and alignment during construction. This paper presents findings from experiments aimed at determining the maximum load capacity of equal-angle sections used as column members. The experiments were conducted on 12 different hot-rolled steel equal-angle sections with both ends hinged. To validate the experimental results regarding ultimate load and failure mode, a nonlinear finite element model was developed in ABAQUS. The results of this parametric study were compared with the code standards IS 800: 2007 and BS 5950–1: 2000. Furthermore, machine learning (ML) techniques were employed to predict the ultimate loads of hot-rolled steel equal-angle sections. ML models, including artificial neural networks (ANN) and gradient boosting regression (GBR), were created to forecast the ultimate load based on finite element analysis results. Several statistical metrics were employed to assess the accuracy of these models in predicting the ultimate load. This study highlights the effectiveness of ML techniques, showing that ANN and GBR are the most reliable forecasting methods for analyzing the ultimate load of hot-rolled steel angle sections.