Data‐driven assessment and design of axially loaded <scp>FRP</scp> ‐reinforced concrete columns

Abstract Fiber‐reinforced polymer (FRP) bars are increasingly used in construction due to their high strength‐to‐weight ratio and resistance to corrosion. Despite these advantages, existing design codes for FRP‐reinforced concrete columns (FRP‐RCC) lack comprehensive equations that adequately account for factors such as slenderness, eccentricity, and the contribution of FRP bars to axial capacity. This study addresses these gaps by proposing an artificial neural network (ANN) model to accurately predict the axial capacity of short, slender, concentric, and eccentric FRP‐RCCs. A comprehensive database of 490 column samples from the literature was utilized to analyze the factors influencing FRP‐RCC behavior, develop the ANN model, and validate its performance. The database also enabled a detailed evaluation of current design codes, existing ANN models, and other design equations, comparing their predictive capabilities to the proposed ANN. The results demonstrate the effectiveness of the proposed ANN model, achieving an R 2 value of 0.93 during training and testing and 0.97 during verification with an independent dataset. These findings underscore the model's robustness and practical applicability in engineering design. Furthermore, comparative analysis revealed that the proposed ANN consistently outperforms existing design equations and available ANN models, highlighting its superior predictive accuracy and potential as a reliable tool for designing FRP‐reinforced concrete structures.