Prediction of axial load-carrying capacity of GFRP-reinforced concrete columns through artificial neural networks

Abstract The present research aims to propose new models for predicting the axial load-carrying capacity of concrete columns reinforced with glass fiber reinforced polymer (GFRP) bars. Two different approaches i.e. Artificial Neural Networks (ANNs) and empirical modeling, were adopted for achieving the aim. A large database of 279 specimens of GFRP-reinforced concrete columns was developed from the literature. The proposed ANN model was calibrated for the different number of neurons in the hidden layers to achieve an optimized model. Furthermore, a preliminary evaluation of the previously proposed empirical models for predicting the axial capacity of GFRP-reinforced concrete columns was performed over the developed database to obtain a more general form of the model. The currently proposed ANN and empirical models presented a close agreement with the experimental database with R2 = 0.848 and R2 = 0.730, respectively. The comparative study of the predictions represents that the currently proposed models are more accurate than the previously proposed models for predicting the axial capacity of GFRP-reinforced concrete columns. Moreover, an extensive parametric study of 600 specimens was performed using the proposed empirical model to examine the effect of various material and geometric parameters of GFRP-reinforced concrete columns.