A multi-indicator failure early warning model for Nano-SiO₂/polyethylene fiber modified concrete subjected to different temperatures

To enhance the comprehensive performance of concrete under high temperatures and rapid cooling conditions, and to establish a reliable fracture early warning method, concrete was modified using nano-silica and polyethylene fibers. Concrete specimens with different nano-silica contents (1%, 2%, and 3%) and polyethylene fiber contents (0.1%, 0.3%, and 0.5%) were treated at temperatures of 200 ℃, 400 ℃, and 600 ℃. After rapid cooling treatment, uniaxial compression tests were conducted, and acoustic emission (AE) and digital image correlation (DIC) techniques were synchronously employed for real-time monitoring. The AE b -value, AE parameters variances, and strain field differentiation rate were utilized to develop the early warning model. The results indicate that a combination of 3% nano-silica and 0.1% polyethylene fibers exhibits the optimal synergistic effect. After thermal treatment to 600 ℃, this mixture achieves a residual strength retention rate of 60%, with its elastic modulus and compressive strength larger than those of all other mixtures. A critical transition point is identified, where the elastic energy dissipation ratio shifts from a decreasing to an increasing trend. Concurrently, the AE energy variance reaches an abnormal peak, the AE b -value drops sharply, and the strain field differentiation rate increases suddenly. These indicators can collectively serve as early warning signals for concrete failure. The proposed multi-indicator model integrates AE energy variance, elastic energy dissipation ratio, AE hit rate, and strain field differentiation rate. It yields a stable, unified early warning interval across different temperature conditions, with the stress ratio at early warning ranging from 90.3% to 97.5% and the early warning coefficient ranging from 6.59% to 9.61%. Compared to single-indicator models, the proposed method narrows the warning interval by 0.9%. By effectively avoiding the “false instability” interval inherent in single-indicator model, the proposed model significantly enhances the accuracy and robustness of fracture early warning.