Unveiling the combined thermal and high strain rate effects on compressive behavior of steel fiber-reinforced concrete: A novel predictive approach

Steel Fiber Reinforced Concrete (SFRC) is widely recognized for its exceptional performance under extreme conditions, such as high temperatures and high strain rates, due to its enhanced fracture resistance and energy absorption. However, the combined effects of these extreme conditions on SFRC’s mechanical behaviour remain insufficiently explored. This study presents a novel framework for predicting the compressive strength of SFRC subjected to temperatures ranging from 200°C to 1200°C and strain rates from 10⁻⁵/s to 10²/s using advanced machine learning (ML) approaches: Gene Expression Programming (GEP), Multi-Expression Programming (MEP), and XGBoost. A dataset comprising 307 experimental results from published studies was used, with 70 % allocated for training and 30 % for testing and validation. The GEP model demonstrated superior performance with R-values of 0.964, 0.968, and 0.960 for training, validation, and testing, respectively. The MEP and XGBoost models provided reasonable accuracy but underperformed compared to GEP. Global Sensitivity Analysis (GSA) identified temperature and strain rate as the most significant parameters influencing compressive strength, while heating rate had minimal impact. Notably, the study developed a simplified empirical equation through GEP, enabling efficient and accurate strength estimation. This research addresses critical gaps by integrating advanced ML models to predict SFRC behaviour under extreme conditions, offering a reliable and cost-effective alternative to experimental testing. The findings provide valuable insights for optimizing SFRC in critical infrastructure applications, enhancing safety and resilience against fire and blast scenarios. This study’s framework sets a foundation for future work in performance-driven material design.