Owing to the limitations of actual fire experiments, research on large-scale fires, such as those in tunnels, has primarily relied on numerical simulations. Scaled models are used to ensure computational efficiency and reproducibility. However, scaling can introduce physical distortions, necessitating a careful review of the reliability of analysis results. In this study, tunnel fire simulations were performed using PyroSim, and smoke spread and error characteristics were analyzed according to varying scales, starting from a scale of 1:1. Our analysis targeted a bus fire scenario that occurred 200 m from a 1,000 m tunnel. Scales of 1:1, 1:2, 1:5, 1:10, and 1:20 were analyzed. Based on smoke arrival times, error rates were compared across various scales using regression estimation (RE), maximum accuracy (MAE), and root mean square error (RMSE). Walking-speed conditions were applied to examine the impact of varying scales on evacuation time. The results revealed differences in smoke flow and error rates depending on scale, with large-scale models tending to provide more conservative estimates of evacuation time.
Im et al. (Thu,) studied this question.