Prediction of Axial Capacity of Concrete Filled Steel Tubes Using Gene Expression Programming

Area of Science:

• Structural Engineering
• Materials Science
• Computational Mechanics

Background:

• Infrastructure project safety and economy rely on accurate material simulation and design.
• Concrete-filled hollow steel sections (CFSS) offer composite action for axially loaded members.
• Existing models may not fully capture the complex behavior of CFSS.

Purpose of the Study:

• To develop a mathematical expression for the ultimate compressive strength (P_u) of CFSS using gene expression programming (GEP).
• To identify key input parameters influencing CFSS performance.
• To validate the developed GEP model against experimental data.

Main Methods:

• Utilized 149 data points from literature for CFSS ultimate compressive strength.
• Incorporated ten input parameters: D, wall thickness, f_c', E_c, f_y, E_s, L, ζ, D/t ratio, and L/D ratio.
• Employed GEP to formulate predictive mathematical expressions for P_u.

Main Results:

• The optimal GEP Model T3 (N_c=100, H_s=8, N_g=3) demonstrated high accuracy.
• Model T3 achieved an overall R-squared of 0.99, with low RMSE and MAE values in training and testing phases.
• Regression slope analysis confirmed Model T3's superior performance (R=0.99).

Conclusions:

• GEP provides a robust method for modeling CFSS ultimate compressive strength.
• Model T3 offers a reliable predictive tool for structural engineers.
• Parametric analysis indicates P_u increases linearly with outer diameter (D), wall thickness (t), and steel yield strength (f_y).