Optimum Carbon Fiber Reinforced Polymer (CFRP) Design for Flexural Strengthening of Cantilever Concrete Walls Using Artificial Neural Networks

Abstract

This study introduces a hybrid framework combining an Artificial Neural Network (ANN) with the Jaya optimization algorithm to predict the minimum Carbon Fiber Reinforced Polymer (CFRP) area required for flexural strengthening of reinforced concrete (RC) cantilever walls. A multilayer perceptron (MLP) network was trained on 500 Jaya-optimized design scenarios incorporating twelve design variables, including geometry, loads, and material properties. The ANN achieved high predictive accuracy, with R-values near 1.0 across training, validation, and testing phases. Five independent test cases yielded an average error of 3.69%, and 10-fold cross-validation confirmed model robustness (R = 0.9996). A global perturbation-based sensitivity analysis was also conducted to quantify the influence of each input parameter, highlighting wall length, moment demand, and concrete strength as the most significant features. This integrated ANN-Jaya model enables rapid, code-compliant CFRP design in accordance with ACI 318 and ACI 440.2R-17, minimizing material usage and ensuring economic and sustainable retrofitting. The proposed approach offers a practical, data-driven alternative to traditional iterative methods, suitable for application in modern performance-based structural engineering.