Nonlinear Finite Element and Machine Learning-Based Prediction of Circular CFRP-Confined Reinforced and Plain Concrete Columns under Axial Compression

This study presents a comprehensive nonlinear finite element study on the behavior of circular CFRP-confined reinforced and plain concrete columns under concentric static loading. A total of 65 test models, which consist of a group of FRP-confined plain concrete, concrete with longitudinal and circular hoop reinforcements, and spiral hoop-reinforced concrete, were designed for this study. The accuracy of the proposed finite element model was verified by comparing it with the existing experimental test results. The impact of unconfined concrete strength, hoop reinforcement ratio, thickness of FRP, and spiral hoop spacing on the confinement effectiveness, load-carrying capacity, and ductility behavior of circular FRP-confined concrete columns was demonstrated. The parametric study revealed that increasing unconfined concrete strength enhances the axial load capacity of FRP-confined concrete columns, but low-strength confined concrete achieves a higher strength enhancement ratio than high-grade concrete. From the study, it’s also found that the confinement effectiveness of hoop reinforcement is mainly dependent on the confining FRP thickness. Based on existing experimental test results and FEA results presented in this paper, a new predictive analytical model for determining circular FRP-confined concrete columns' peak stress with the associated strain was also developed. Furthermore, machine learning (ML) models for predicting the peak axial load and ultimate strain at the tensile rupture of FRP were developed using four different machine learning techniques. The predictive performance of the proposed machine learning models was evaluated by six performance metrics indices of statistical parameters.