AI-Augmented Analytical–Numerical Model for Cyclic Behavior of Basalt- FRP Strengthened RC Beams: MATLAB-Based Theoretical Method.

Abstract

This study proposes a hybrid analytical–AI modeling framework for predicting the cyclic flexural behavior of reinforced-concrete (RC) beams strengthened with externally bonded Basalt Fiber-Reinforced Polymer (BFRP) sheets. The approach combines a MATLAB-based analytical–numerical solver with an artificial-intelligence surrogate predictor to enable accurate and rapid assessment of strength, ductility, and energy dissipation under repeated loading.

A fiber-section analytical model was formulated to simulate the nonlinear hysteresis response, incorporating concrete degradation, steel kinematic hardening, and bond-controlled BFRP strain limits. Displacement-controlled cyclic simulations were executed to generate a comprehensive dataset spanning BFRP thicknesses (0.5–2.0 mm) and bonded-length ratios (0.3–0.9). A deep-learning surrogate network, trained on 300 analytical simulations, achieved a high predictive accuracy (R² = 0.964, RMSE = 98.4 kN), confirming strong correlation with both analytical and published experimental data.

Parametric results revealed that increasing effective FRP strain up to 0.016–0.018 enhances peak load by ≈ 20 %, while extending the bonded length to Lₐ/L ≈ 0.8 improves ductility and equivalent damping (ξ_eq ≈ 5 %). Beyond these limits, performance gains plateau, defining an optimal retrofit range.

The proposed framework establishes a non-experimental digital-twin pathway for performance-based FRP retrofit design, offering a computationally efficient alternative to finite-element or laboratory testing. The model’s scalability supports future integration into AI-driven code calibration and design-chart development for cyclic and seismic applications.