Prediction of Ultimate Load Capacity of Concrete-Filled Steel Tubes with Circular Sections under Axial Load by Using Predictive Analytics Methods

Steel concrete composite hollow section columns with circular cross-sections, i.e. concrete-filled steel tubes (CFST), have been a popular choice for column design due to their superior structural performance both under static gravity loads as well as their contribution in the lateral force resistance systems of both braced and unbraced framed structures. This is mainly due to the ideal combination of the ductility and stiffness characteristics inherent in steel and concrete, respectively, resulting in structural members and systems with outstanding performance particularly in regions highly exposed to earthquakes. Numerous studies are available that analyzed the behavior of CFSTs under axial compressive loadings experimentally. However, experimental studies require long times, high effort, and cost. So, this study evaluates predictive models used to predict the ultimate load capacity of CFST with circular sections under axial compressive load. We identified 1673 experimental data from the existing literature for the analysis. Four prevalent Machine Learning techniques such as Support Vector Machines (SVM), Artificial Neural Networks (ANN), Random Forest (RF) and Multiple linear regression (MLR) were used and then the performances of these algorithms were compared. The prediction models' results have been found in a match with the experimental results in this study. ANN outperforms the other algorithms with the highest coefficient of determination (R2) and the smallest root mean square error (RMSE).