Data-driven investigation of compressive strength of FRP-confined concrete columns using a unified model based on RSM considering interactions between parameters

In this study, a data-driven model was developed to predict compressive strength of concrete columns after confining with Fiber-Reinforced Polymer (FRP). Through application of Response Surface Method (RSM), interactions of input parameters are investigated in addition to their individual influence. Moreover, RSM is utilized in conjunction with unification to enhance model's accuracy. Furthermore, sensitivity analysis is employed together with RSM, to determine the effect size of statistically significant parameters. It is revealed that the real behavior of FRP confined concrete cannot be predicted by single-parameter studies. Therefore, interaction terms should be included in the model to yield a more accurate prediction. Moreover, the model shows the highest sensitivity to FRP failure strength which this sensitivity decreases as corner radius increases. Therefore, regarding rounded-corner sections, high improvements are achievable even by utilizing low strength FRPs. It is also shown that higher improvement may be limited due to premature debonding which elastic modulus mismatch between different phases is supposed to be responsible for it.