Modeling and optimization of key parameters for temperature control by fire sealing in tunnel fires using response surface methodology

Fires in road tunnels have emerged as a critical public safety threat worldwide. Among various fire suppression strategies, the sealing method, which physically cuts off oxygen supply, has proven to be a key tactic for controlling tunnel fire spread. However, a systematic quantification of the coupled effects of multiple parameters during dynamic sealing processes remains largely unexplored. To bridge this gap, this study introduces an integrated framework combining Response Surface Methodology and Fire Dynamics Simulator to quantify the effects of enclosed space length (A), sealing time (B), and sealing ratio (C) on maximum ceiling temperature (Tmax) in tunnel fires. A highly accurate predictive model (R2 = 0.9978) was developed based on Box–Behnken-designed numerical simulations. Results identify C as the dominant factor (31.1% contribution) and exhibit a significant quadratic nonlinear relationship, exhibiting a critical threshold near 75% where combustion transitions from ventilation-controlled to fuel-controlled, drastically reducing Tmax via oxygen depletion. While A moderately influences heat accumulation through smoke recirculation (5.3% contribution) and B shows limited individual significance, their interactions reveal delayed sealing risks in long tunnels. Sensitivity and Monte Carlo analyses confirm the model's robustness. Under a ±15% variation in Heat Release Rate, the predicted Tmax remains within a 95% confidence interval of 588.1–622.3 °C, demonstrating the model's stability against input uncertainty. This work provides a quantitative foundation for advancing tunnel fire sealing strategies from empirical practice to optimized design.