Buckling behavior analysis of prestressed CFRP-reinforced steel columns via FEM and ANN

Prestressed (PS) carbon fiber reinforced polymer (CFRP)-reinforced steel columns are novel multiparameter systems exhibiting complex nonlinear buckling behavior. In this study, this behavior was investigated with the finite element method (FEM) and an artificial neural network (ANN). First, FEM models of the columns under axial and eccentric compression were built. The numerical and experimental force–displacement curves, failure modes, and CFRP stress–displacement curves were in good agreement. Moreover, the influencing rules of 9 key parameters (i.e., CFRP initial prestressing force, supporting length, eccentricity, steel yield strength, slenderness, CFRP elastic modulus, initial imperfection and boundary conditions) on the buckling capacity and reinforcing efficiency of the reinforced columns were determined. Afterward, 312 datasets from the validated finite element model covering 8 input parameters were generated via the ANSYS parametric design language (APDL). Finally, as ANNs can manage highly complex and computationally intensive nonlinear problems, a practical ANN tool was developed to predict the buckling capacity of PS CFRP-reinforced steel columns.