Bond strength prediction of externally bonded reinforcement on groove method (EBROG) using MARS-POA

Abstract

The Externally Bonded Reinforcement on Grooves (EBROG) method represents an advancement in externally bonded reinforcement (EBR) techniques, specifically addressing the challenge of premature debonding often encountered in conventional Fiber Reinforced Polymer (FRP) applications directly bonded to concrete. This article introduces a novel and straightforward mathematical equation for predicting bond strength in the EBROG method using soft computing techniques for the first time. The study delves into the combined potential of the Multivariate Adaptive Regression Spline (MARS) model and the Pelican Optimization Algorithm (POA) for bond strength prediction in this method. The input parameters include FRP width, FRP thickness, elasticity modulus of FRP, concrete strength, groove length, groove width, and groove depth, while the output is EBROG bond strength. The study demonstrates exceptional accuracy with R2 values of 0.9629 for training and 0.9623 for testing, highlighting the model’s precision. The proposed bond strength prediction equation for the EBROG method undergoes validation against existing models, encompassing thirteen equations for EBR and a recent one specific to EBROG. Statistical metrics confirm the accuracy and reliability of the proposed equation. Notably, FRP stiffness emerges as the parameter with the highest relative importance, while groove width exhibits the lowest impact on bond strength.