Modeling the Shear Capacity of Externally Bonded Fiber Reinforced Polymer Strengthened Beams by Artificial Neural Network

The current guidelines, namely ACI 440.2R and fib TG-9.3, quantify the nominal shear strength of strengthened reinforced concrete (RC) beams by simply summing up the shear strength contribution of the three components: concrete (V_c), steel stirrups (V_s), and externally bonded fiber reinforced polymer (EB-FRP) (V_f). However, as reported in the literature, this assumption is inaccurate due to an adverse interaction between EB-FRP and steel stirrups. This research study adopts a machine learning method called Artificial Neural Network (ANN) to model EB-FRP shear strength contribution. The model is built by considering the possible interaction mentioned above. Considering the collective response of these three components is crucial in accurately predicting the shear behavior because the performance potential of EB-FRP strengthening depends on the material properties of existing members. To implement the ANN modeling method, a database of 191 test specimens reported by a total of 40 individual studies is gathered. For the assessment of the proposed ANN model, a sensitivity analysis is conducted. The proposed model prediction is then compared statistically with experimental results and with the predictions of the existing guidelines such as ACI 440.2R and fib TG-9.3. The results showed that the developed ANN model could predict the EB-FRP shear strength contribution with higher accuracy than the current guidelines. Moreover, a user-modified executable program is also developed in this study to readily execute the proposed ANN model by inputting the properties of the existing beam and EB-FRP.