A Proposed Model for Axial Strength Estimation of Non-compact and Slender Square CFT Columns

The current international codes divide composite columns into three classes based on the width-to-thickness ratio of the steel tube: compact, non-compact, and slender. The present paper studies the specification of AISC 360-10 to determine the axial strength of concrete-filled steel tube (CFT) members. Three efficient approaches have been developed based on the computational intelligence technique, using a comprehensive database (experimental and numerical) of non-compact and slender square CFT columns. To achieve a wide variety in geometric and material properties of composite columns, the numerical models are simulated using OpenSEES finite element analysis package. The proposed models were created using the Levenberg-Marquardt artificial neural network, group method of data-handling approach, and gene expression programming, based on the mechanical (yield stress of tube and compressive strength of concrete) and geometrical (column length and dimensions of tube) properties of the CFT members. Comparison of the results of developed models and experimental specimens indicates superior performance and their acceptable accuracy in the determination of the axial strength of non-compact and slender CFT columns.