Effectiveness of metaheuristic algorithms in optimization of CFDST short columns

Abstract

Concrete-Filled Double Steel Tubular (CFDST) columns, a subclass of composite members, are increasingly utilized in modern construction due to their superior structural performance, including higher strength-to-weight ratio, stiffness, fire resistance, and durability compared to traditional reinforced concrete (RCC) and steel structures. While the optimization of CFDST columns has been widely studied using statistical methods like Response Surface Methodology (RSM), this study focuses on applying advanced metaheuristic algorithms for optimization. A comparative evaluation of the Algorithm (GA), Particle Swarm Optimization (PSO), and Jaya Algorithm (JA) is conducted to optimize the axial load capacity of CFDST columns by fine-tuning key design parameters. The optimization objective is to maximize axial load capacity, which serves as the fitness function in the algorithms. An empirical equation representing the axial load capacity is used as the objective function.The comparative analysis demonstrates that metaheuristic algorithms are efficient and straightforward for optimizing CFDST columns. Among the three, the JA outperforms the others, achieving faster convergence and more consistent results across various population sizes and generations. PSO also exhibits faster convergence than GA, though JA remains the most robust and reliable. The study observes that the axial load capacity of the columns increases by 4.17%, 13.9%, 1.16%, and 10.81% for the considered series of specimens utilizing the same or fewer geometric values of key design components. These findings highlight the potential of metaheuristic algorithms in enhancing the design efficiency of CFDST columns for modern structural applications.