Axial compressive capacity of concrete‐filled steel tube columns: Experimental investigation, regression modeling, and code evaluation

Abstract Concrete‐filled steel tube (CFST) columns have gained popularity due to their ability to sustain a high axial load and composite action in structural applications; however, their axial strength is difficult to predict This study presents a combined experimental and data‐driven study on the axial compressive capacity of CFST columns. Under concentric axial load, CFST specimens were experimentally tested. These specimens consisted of three concrete types and three configurations of steel tubes in order to assess load‐deformation behavior and ultimate strength. In an effort to enhance strength prediction, an extensive database containing 862 experimentally tested CFST columns from 96 published studies was constructed. From this database, five multivariable regression models were created to predict the ultimate axial load (): linear, nonlinear, pure quadratic, interaction, and full quadratic (FQ). The models were evaluated using the coefficient of determination ( R 2 ), the root mean square error, and the mean absolute error. Of all the models tested, the most accurate predictor was the FQ regression model, which produced R 2 values of over 0.986 for . In order to examine the accuracy in predicting the axial strength of the models, the optimal regression model's predictions were compared to the Eurocode 4 and American Institute of Steel Construction (AISC) 360‐10 design codes. From the findings, it can be observed that the regression‐based approach significantly outperforms existing code formulations, proving to be a useful approach in the design and assessment of CFST columns.