Finite element and soft computing modeling of axial load carrying capacity of concrete filled composite short columns

Composite columns have higher strength and ductility performance, and recently they have been in use in the engineering applications. However, the filled tubular columns need paying more attention. This study aims to develop a finite element analysis (FEA) model and to present new formulations generated by gene expression programming (GEP) and artificial neural network (ANN) techniques for the axial load carrying capacity (Nu) of circular concrete filled steel tubular (CFST) short columns. For this purpose, 314 comprehensive experimental data samples presented in the previous studies were examined to prepare a data set. The prediction parameters were selected as outer diameter of column (D), wall thickness (t), length of column (L), compressive strength of concrete (fc), and yield strength of steel (fy). Although there have been some code equations and empirical relations for predicting Nu in the literature, a genetic algorithm-based explicit formulation is not available and the required attention to the ANN and FEA modeling on these structural composite members has been limited. In this regards, firstly, the failure modes and load-displacement behaviour of the columns were studied by using FEA, and then the axial capacity prediction model obtained by means of the FEA model as well as the GEP and ANN techniques proposed in this study were compared with available ones presented in the current design codes (ACI/AS, AISC, AIJ, EC4, DL/T, CISC) and some existing empirical models proposed by the researchers. The prediction performance of all models was also evaluated by the statistical parameters.