*3.2. Simulated Conditions*

*3.2. Simulated Conditions* Two groups of simulation scenarios are designed, as summarized in Table 1. The first group establishes the model by changing the number of cross passages (CP) from one to four. The interval of cross passages is 50 m, with the cross-sectional area of 12 m<sup>2</sup> and a width × height of 4 m × 3 m. The velocity of longitudinal ventilation generated in an accident tunnel is 2 m/s; the airflow velocity at both ends of the service tunnel is 1.3 m/s. The other group has five conditions, with different ventilation velocities in the service tunnel, Two groups of simulation scenarios are designed, as summarized in Table 1. The first group establishes the model by changing the number of cross passages (CP) from one to four. The interval of cross passages is 50 m, with the cross-sectional area of 12 m<sup>2</sup> and a width × height of 4 m × 3 m. The velocity of longitudinal ventilation generated in an accident tunnel is 2 m/s; the airflow velocity at both ends of the service tunnel is 1.3 m/s. The other group has five conditions, with different ventilation velocities in the service tunnel, set as: 0.70, 0.75, 0.85, 1.00 and 1.30 m/s. Three cross passages remain open. Other parameter conditions are consistent with the first group.

set as: 0.70, 0.75, 0.85, 1.00 and 1.30 m/s. Three cross passages remain open. Other param-



**(m)**

**(m) (m/s) (m/s)** a01 310 300 -- -- -- 2 1.30 a02 310 300 250 -- -- 2 1.30 a03 310 300 250 200 -- 2 1.30 a04 310 300 250 200 150 2 1.30 b01 310 300 250 200 -- 2 0.70 The fire source is located at 310 m inside the accident tunnel, with a heat release rate of 20 MW, set to reach the maximum value within 1 s, then maintained during the simulation for up to 500 s. Air flow in the service tunnel enters the accident tunnel through cross passages, to raise the longitudinal ventilation velocity to critical velocity. As smoke movement would reach a quasi-steady state after 300 s, the average values from 450 s to 500 s is considered as quasi-steady state data for investigation.

**(m)**

### b02 310 300 250 200 -- 2 0.75 *3.3. Grid Sensitivity Analysis*

*3.3. Grid Sensitivity Analysis*

b03 310 300 250 200 -- 2 0.85 b04 310 300 250 200 -- 2 1.00 b05/a03 310 300 250 200 -- 2 1.30 As the gird size is crucial for the reliability of numerical simulation, we chose the value range *D*∗/16 to *D*∗/4, which has been widely used for assessing grid resolution for sensitivity analysis. Here, *D*∗ can be calculated by [26]:

$$D^\* = \left(\frac{Q}{\rho\_{\infty}c\_p T\_{\infty}\sqrt{\mathcal{g}}}\right)^{2/5} \tag{8}$$

(8)

As the gird size is crucial for the reliability of numerical simulation, we chose the value range \* *D* 16 to \* *D* 4 , which has been widely used for assessing grid resolution for sensitivity analysis. Here, \* *D* can be calculated by [26]: A grid size of between 0.2 m and 0.8 m is considered suitable for simulation. In order to obtain a reasonable and accurate mesh size, we took five grid sizes of 0.40 m, 0.50 m, 0.60 m, 0.70 m and 0.80 m, for the numerical simulation stability analysis. The

2 5

*p*

*Q*

*c T g*

A grid size of between 0.2 m and 0.8 m is considered suitable for simulation. In order to obtain a reasonable and accurate mesh size, we took five grid sizes of 0.40 m, 0.50 m,

\*

=

*D*

longitudinal ventilation velocity in the tunnel is 3.0 m/s and the fire load is 20 MW. As shown in Figure 2, when the mesh size is less than 0.5 m, the simulation results obtained stability and repeatability, while the CPU time of 0.4 m took 23 h more than 0.5 m (computer configuration: i7-7700, 16 GB RAM). In order to save storage space and computing time, we employed 0.5 m as the mesh size of the numerical simulation model. ventilation velocity in the tunnel is 3.0 m/s and the fire load is 20 MW. As shown in Figure 2, when the mesh size is less than 0.5 m, the simulation results obtained stability and repeatability, while the CPU time of 0.4 m took 23 h more than 0.5 m (computer configuration: i7-7700, 16 GB RAM). In order to save storage space and computing time, we employed 0.5 m as the mesh size of the numerical simulation model.

0.60 m, 0.70 m and 0.80 m, for the numerical simulation stability analysis. The longitudinal

*Fire* **2022**, *5*, x FOR PEER REVIEW 5 of 10

**Figure 2.** Temperature distribution with different grid sizes. **Figure 2.** Temperature distribution with different grid sizes.

### **4. Result and Discussion 4. Result and Discussion**
