Data-driven axial load-carrying capacity prediction of FRP-RC columns through random forest regression

Numerous past investigations have shown that the behavior of concrete columns reinforced with fiber-reinforced polymer (FRP) bars differs from that of traditional reinforced concrete columns, mainly due to the distinct mechanical properties of FRP bars. However, existing empirical models for predicting the axial load-carrying capacity (ALCC) of FRP-reinforced concrete (FRP-RC) columns often low accuracy. The aim of the present work is to introduce a novel model for accurately predicting the ALCC of concrete columns reinforced with FRP bars. To achieve this, random forest (RF) regression was applied on collected dataset of 377 concrete columns samples reinforced with FRP bars from previous literature works. A range of input variables was considered, including concrete type, column height, gross section area, compressive strength, area of FRP reinforcement bars, FRP reinforcement ratio, number of FRP bars, shape of tie bars, type of longitudinal and transverse FRP reinforcement, diameter of stirrups, diameter of main FRP bars, modulus of elasticity of FRP bars, their tensile strength, and stirrup spacing to estimate the ALCC of FRP-reinforced concrete columns. The suggested RF prediction showed a well correlation with dataset, achieving R² values of 0.99 and 0.97, respectively. Additionally, a comparative analysis of RF model against fifteen existing empirical formulas confirmed that the proposed machine learning model significantly improve the accuracy of ALCC predictions for FRP-reinforced concrete columns.