A novel gene expression programming model for predicting ultimate axial capacity of FRP-reinforced concrete columns: sensitivity analysis and comparative evaluation

Fiber-reinforced polymer (FRP) bars have been recognized as a viable alternative to conventional steel bars in construction, particularly in severe environments susceptible to corrosion. Prior works indicate that the accurate prediction of the ultimate load carrying capacity of FRP-reinforced concrete columns by considering influential design parameters has been a challenging problem. In this paper, Gene Expression Programming (GEP) technique is used to develop a new model for predicting the ultimate axial capacity of concrete columns reinforced with FRP bars. A comprehensive and reliable collection of 155 data points from literature was collected. The model is developed using five main parameters that primarily control the ultimate axial capacity of concrete columns, which are, concrete compressive strength, column gross cross-sectional area, columns slenderness ratio, the ultimate tensile strength of FRP bars, and area of longitudinal FRP reinforcement. Based on the developed GEP model, the influence of each variable on the column capacity is evaluated by performing sensitivity analysis and parametric study. Moreover, the prediction performance of the new model is compared to the available design equations and models in the literature. The statistical assessments revealed that the proposed model predictions are in good agreement with experimental test results and outperforms current models in the literature. The statistical assessments revealed that the proposed model achieved high accuracy, with R2 values of 0.944 and 0.957 and RMSE values of 505 and 368 for the training and validation datasets, respectively. Furthermore, the model outperforms existing design equations, demonstrating superior predictive capability and stability across all analyzed specimens.