Compressive response of concrete-filled double-skin steel elliptical tubular columns: A hybrid finite element and machine learning approach

This study examined the structural behavior of concrete-filled double-skin steel elliptical tubular columns (CFDSETC) under different loading scenarios. These columns’ considerable advantages in strength, longevity, and architectural adaptability make them a potential replacement for traditional structural components. Eight finite element models (FEM) that had previously been validated using experimental data were created using ABAQUS software. An expanded dataset with 330 additional FEM was created by meticulously modifying a number of material and design parameters, such as the area of the concrete portion, yield strength of the outer and inner steel tubes, concrete cubic compressive strength, thickness of the outer and inner steel tubes, and loading eccentricity. Additionally, five machine learning (ML) approaches that predicted the CFDSETC load-bearing capacity during eccentric and axial compression were trained using the dataset obtained from FEM. The results showed that the load carrying ability of the CFDSETC was enhanced by increasing the concrete’s strength and area, the outer and inner steel tubes’ yield strength and thickness, and lowering the load eccentricity. Out of all the models evaluated, the categorical boosting model achieved the highest prediction accuracy. The findings provide useful information on the design and optimization of CFDSETC for use in civil engineering structures.