Predicting axial load capacity of CFST columns using machine learning

Abstract

Owing to their economic and structural advantages, concrete-filled steel tubular (CFST) columns have been implemented in diverse structural applications, especially in high-rise buildings, suspension bridges, and subway stations. However, there is no agreement between international standards regarding the ultimate compressive strength of CFST columns subjected to concentric axial force or combination of bending moment and axial force, especially for slender sections and high-strength materials. Considering such limitations, a comprehensive dataset on rectangular and circular (built up and hot rolled) CFST sections with different slenderness ratios was collected. The compiled experimental results were compared with corresponding values calculated by pertinent design code provisions, including AISC 360-16 and Eurocode 4. The accuracy of design codes in predicting the axial capacity of CFST sections under concentric and eccentric loading was assessed accordingly. The distribution of the experimental-to-predicted data confirmed that the prediction error was dependent on the slenderness ratio of columns. However, the effects of other parameters, including the mechanical properties of materials and eccentricity level on the model error, were negligible. In the second phase of the study, a surrogate Machine-Learning (ML) model was developed to estimate the axial capacity of circular and rectangular CFST columns under centric or eccentric loading condition. The accuracy of the proposed ML predictive model was appraised using several statistical metrics. The novel informational model attained superior accuracy and could be used to simplify generative design in future computational intelligence structural design platforms.

Prediction of Torque Capacity in Circular Concrete-Filled Double-Skin Tubular Members under Pure Torsion via Machine Learning and Shapley Additive Explanations Interpretation

“…Significant attention in various engineering disciplines due to their capacity to comprehensively analyze large datasets [10], identifying complex patterns and relationships that may not be evident through conventional methods [11], has been gained by ML in recent years. Several recent studies have explored the application of ML techniques in predicting the strength of concrete-filled steel tubular (CFST) columns under various loading conditions.…”

“…Their findings suggested superior accuracy and broader applicability compared to existing design standards. Similarly, a machine learning model capable of estimating the axial capacity of both circular and rectangular CFST columns under various loading conditions, offering a high level of accuracy and serving as a viable alternative to empirical and theoretical formulations, was developed by Faridmehr and Nehdi [11].…”

Application of ensemble model in capacity prediction of the CCFST columns under axial and eccentric loading

“…The development and application of machine learning techniques provide new insight to solve this problem 20 . It is foreseen that using machine learning to predict component performance will not only provide a reference for actual design but also save significant resources by making full use of completed experimental data and reducing the need for further testing 21 , 22 . Moreover, machine learning is based on patterns between large amounts of experimental data and is much less dependent on the users themselves.…”