2.1.2. Ventilation Conditions

2.1.2. Ventilation Conditions The tunnel adopts a full-jet longitudinal ventilation system. To focus on the influence of ventilation velocity conditions on ASET, only the velocity value was considered, and other factors affecting smoke diffusion were ignored. Due to the large range of possible ventilation velocities in actual situations of tunnel fire, extensive selections of velocity conditions would make the total workload too large. Therefore, the ventilation velocity was simply divided into several modes, and the typical velocity values of each mode were set according to the ventilation velocity studies mentioned above. The velocity modes include "turn off mechanical ventilation," "low-velocity," "medium-velocity," "high-velocity The tunnel adopts a full-jet longitudinal ventilation system. To focus on the influence of ventilation velocity conditions on ASET, only the velocity value was considered, and other factors affecting smoke diffusion were ignored. Due to the large range of possible ventilation velocities in actual situations of tunnel fire, extensive selections of velocity conditions would make the total workload too large. Therefore, the ventilation velocity was simply divided into several modes, and the typical velocity values of each mode were set according to the ventilation velocity studies mentioned above. The velocity modes include "turn off mechanical ventilation," "low-velocity," "medium-velocity," "high-velocity (close to the critical velocity vc)," and "high-velocity (satisfying the critical velocity vc)". Typical velocity values are shown in Table 1, which are the average air velocities in the tunnel after the stable operations of jet fans.

(close to the critical velocity vc)," and "high-velocity (satisfying the critical velocity vc)". **Table 1.** Different velocity modes and values in tunnel fire.


Turn off mechanical ventilation 0.0

### Low-velocity 1.0 2.1.3. Evacuation Conditions

Medium-velocity 2.0 High-velocity (close to vc) 4.5 High-velocity (satisfying vc) 6.0 2.1.3. Evacuation Conditions The tunnel adopts a longitudinal evacuation method with a bottom passage linked The tunnel adopts a longitudinal evacuation method with a bottom passage linked by evacuation slides. Evacuation slides are installed from the middle position (Y = 500 m) to both sides of the tunnel equidistantly. According to the engineering experience range of slide spacing in tunnels, six schemes of different slide spacings from 30 m to 80 m were proposed, as shown in Table 2. The priority was to satisfy the evacuation spacings in the middle of the tunnel. Due to the high evacuation capability at the entrance and exit of the tunnel, the spacing between the slides on the edge and the entrance (exit) is slightly larger.

by evacuation slides. Evacuation slides are installed from the middle position (Y = 500 m) to both sides of the tunnel equidistantly. According to the engineering experience range


**Table 2.** Schemes of tunnel slide spacing. 40 m 23

of slide spacing in tunnels, six schemes of different slide spacings from 30 m to 80 m were proposed, as shown in Table 2. The priority was to satisfy the evacuation spacings in the middle of the tunnel. Due to the high evacuation capability at the entrance and exit of the tunnel, the spacing between the slides on the edge and the entrance (exit) is slightly larger.

> **Slides Spacing (D) Number of Slides**  30 m 31

*Fire* **2022**, *5*, x FOR PEER REVIEW 4 of 17

**Table 2.** Schemes of tunnel slide spacing.

### *2.2. Modeling and Validation* 2.2.1. Physical and Mathematical Model

### 2.2.1. Physical and Mathematical Model A full-scale 3D model of the tunnel was constructed by PyroSim software and sim-

A full-scale 3D model of the tunnel was constructed by PyroSim software and simplified vehicle entities in the whole tunnel were built to correspond to reality situations, as shown in Figure 2. The calculation kernel of PyroSim is based on FDS (version 6.7.5), developed by the National Institute of Standards and Technology (NIST). FDS numerically solves a form of the Navier–Stokes equations appropriate for low speeds (Ma < 0.3), thermally driven flow with an emphasis on smoke and heat transport from fires [22]. plified vehicle entities in the whole tunnel were built to correspond to reality situations, as shown in Figure 2. The calculation kernel of PyroSim is based on FDS (version 6.7.5), developed by the National Institute of Standards and Technology (NIST). FDS numerically solves a form of the Navier–Stokes equations appropriate for low speeds (Ma < 0.3), thermally driven flow with an emphasis on smoke and heat transport from fires [22].

**Figure 2.** Schematic diagram of the full-scale 3D physical model of tunnel. **Figure 2.** Schematic diagram of the full-scale 3D physical model of tunnel.

In order to highlight the influences of ventilation velocity on smoke diffusion and ASET distributions, parameters such as installation position and the start sequence of the fans are ignored. The airflow of each ventilation condition was provided by the air-supply surface of constant flow outside the tunnel entrance, and the feasibility refers to the experimental studies of Tang et al. [23] and Chow et al. [24]. ASET is defined as the time interval between the time of ignition and the time passengers are estimated to be incapacitated [25], and it is also the tenability limit calculated depending on the hazard evaluation of the smoke environment. Carbon monoxide (FEDCO), cumulative heat (FEDheat), the tem-In order to highlight the influences of ventilation velocity on smoke diffusion and ASET distributions, parameters such as installation position and the start sequence of the fans are ignored. The airflow of each ventilation condition was provided by the air-supply surface of constant flow outside the tunnel entrance, and the feasibility refers to the experimental studies of Tang et al. [23] and Chow et al. [24]. ASET is defined as the time interval between the time of ignition and the time passengers are estimated to be incapacitated [25], and it is also the tenability limit calculated depending on the hazard evaluation of the smoke environment. Carbon monoxide (FEDCO), cumulative heat (FEDheat), the temperature of convective heat (Temp), and evacuation visibility (Vis) were used as indexes to calculate ASET in this study, as shown in Table 3.

perature of convective heat (Temp), and evacuation visibility (Vis) were used as indexes to calculate ASET in this study, as shown in Table 3. We used a visibility threshold of 3 m because of the dense slides in the bottom evac-We used a visibility threshold of 3 m because of the dense slides in the bottom evacuation mode. Most of the passengers could quickly reach the vicinity of the evacuation slide before the visibility drops significantly, and then they can keep evacuating until they would turn back than rather enter the smoke.

uation mode. Most of the passengers could quickly reach the vicinity of the evacuation slide before the visibility drops significantly, and then they can keep evacuating until they would turn back than rather enter the smoke. The sensors and slices were set up to obtain the CO concentration, temperature, air velocity, and visibility data of the evacuation space for monitoring ambient air velocity The sensors and slices were set up to obtain the CO concentration, temperature, air velocity, and visibility data of the evacuation space for monitoring ambient air velocity and recording data for the above indexes. The 2 m height of installed sensors was based on the range of evacuating activity, considering taller individuals, and leaving a safety margin to ensure that smoke hazards would not be underestimated. The sensors and slices for simulation are shown in Table 4.

and recording data for the above indexes. The 2 m height of installed sensors was based on the range of evacuating activity, considering taller individuals, and leaving a safety


**Table 3.** Calculation index of available safe escape time (ASET).

**Table 4.** Sensors and slice settings.


The initial parameters of model boundary conditions were set as follows:


The computational domain of the model was made up of 15 adjacent linked meshes and covers the entire tunnel. To shorten the simulation time, multi-core parallel computing of a single CPU was carried out. The grid and simulation settings are shown in Table 5.

**Table 5.** Grid and simulation settings.


Near-wall eddy viscosity 0.6

Traffic Modes

process.

Pathfinder software, which has been commonly used in the field of fire evacuation [29,30], was used to construct the evacuation model in this study and was used in steering mode to consider the passenger collisions [31]. According to the traffic flow data that will be mentioned in Section 2.2.2, assuming that all blocked vehicles are randomly distributed and the distance between vehicles is 1 m, a total of 310 vehicles and 1870 passengers could possibly be trapped in the tunnel. The average shoulder width of passengers was set to 0.5 m, in line with the relevant fire safety standard. Evacuation slides on the side walls of the tunnel were simplified as doors to the outside. In addition, passengers could flexibly choose to evacuate from the evacuation slide or tunnel entrance according to the distance or flow without crossing the fire source. The partial evacuation model is shown in Figure 3. The white closed lines denote vehicle outlines, the green points denote passengers, and the short green line below denotes an evacuation slide. mode to consider the passenger collisions [31]. According to the traffic flow data that will be mentioned in Section 2.2.2, assuming that all blocked vehicles are randomly distributed and the distance between vehicles is 1 m, a total of 310 vehicles and 1870 passengers could possibly be trapped in the tunnel. The average shoulder width of passengers was set to 0.5 m, in line with the relevant fire safety standard. Evacuation slides on the side walls of the tunnel were simplified as doors to the outside. In addition, passengers could flexibly choose to evacuate from the evacuation slide or tunnel entrance according to the distance or flow without crossing the fire source. The partial evacuation model is shown in Figure 3. The white closed lines denote vehicle outlines, the green points denote passengers, and the short green line below denotes an evacuation slide.

Pathfinder software, which has been commonly used in the field of fire evacuation [29,30], was used to construct the evacuation model in this study and was used in steering

**Figure 3.** Partial view of evacuation model. **Figure 3.** Partial view of evacuation model.

*Fire* **2022**, *5*, x FOR PEER REVIEW 6 of 17

Time step system default value without limiting time step

## 2.2.2. Operation Modes and Fire Scenarios 2.2.2. Operation Modes and Fire Scenarios

Traffic Modes

As an extremely unfavorable situation that may occur, the condition of traffic congestion in the entire tunnel was assumed in this study. In this case, the number of vehicles and the number of people to be evacuated are the largest. The traffic composition of the tunnel, as shown in Table 6, was based on the vehicle-type distribution data of the roads near the tunnel, which were obtained on the Macau DSAT traffic video monitoring plat-As an extremely unfavorable situation that may occur, the condition of traffic congestion in the entire tunnel was assumed in this study. In this case, the number of vehicles and the number of people to be evacuated are the largest. The traffic composition of the tunnel, as shown in Table 6, was based on the vehicle-type distribution data of the roads near the tunnel, which were obtained on the Macau DSAT traffic video monitoring platform (http://www.dsat.gov.mo, accessed on 15 August 2021), and the average passenger number of each vehicle was assumed from experience data.

form (http://www.dsat.gov.mo, accessed on 15 August 2021), and the average passenger **Table 6.** Traffic composition of tunnel.

MPV 20 6


### Small Bus 3 20 Evacuation Modes

Bus 8 45 HGV 14 2 Evacuation Modes The evacuation process can be divided into two parts, the ready process and the travel process, which could be reflected in the composition of RSET. RSET is defined as the calculated time period required for an individual passenger to travel from their location at the time of ignition to a safe refuge or place of safety [25], which consists of the detect time, alarm time, pre-travel activity time, and travel time, as shown in Figure 4. For simplicity, the ready time (RT) is considered to be the time period required for the ready process.

The evacuation process can be divided into two parts, the ready process and the travel process, which could be reflected in the composition of RSET. RSET is defined as the calculated time period required for an individual passenger to travel from their location at the time of ignition to a safe refuge or place of safety [25], which consists of the detect time, alarm time, pre-travel activity time, and travel time, as shown in Figure 4. For simplicity, the ready time (RT) is considered to be the time period required for the ready In the design stage, it is difficult to determine a clear safety margin, so the basic requirement ASET ≥ RSET was used as the safe evacuation condition, which was also adopted in the research of Song et al. [32], and Zhang et al. [33]. During the ready process, passengers close to the fire will start traveling immediately after the ignition, and RT is zero. However, passengers far from the fire need to wait for the arriving crowd or the ringing alarm to begin to move; similar to a domino effect, their RT goes up because the moving crowd or diffusing smoke will prompt others to realize the risk. Considering the prompt effect of smoke diffusion and crowd movement on evacuation, the RT was set according to the distance from the fire source, as shown in Table 7.

**Figure 4.** The relationship between the available safe escape time (ASET) and the required safe escape time (RSET) in evacuation process. **Figure 4.** The relationship between the available safe escape time (ASET) and the required safe escape time (RSET) in evacuation process.

**Table 7.** RT of evacuation in different distance from the fire source.


moving crowd or diffusing smoke will prompt others to realize the risk. Considering the prompt effect of smoke diffusion and crowd movement on evacuation, the RT was set according to the distance from the fire source, as shown in Table 7. **Table 7.** RT of evacuation in different distance from the fire source. **Distance from the Fire Source S(m) RT (s)**  S ≤ 20 0 20 < S ≤ 60 30 60 < S ≤ 120 60 The maximum walking speed of passengers may undergo complex changes influenced by the smoke environment and psychological fluctuation during the travel process, which is not the focus of this article. To simplify this, referring to the study of Shen et al. [20], Seike et al. [34], and the average walk speed given by PIARC [35], the maximum values of walking speed were set as 1.0 m/s in the horizontal part and 0.8 m/s in the slope part, which represent a low level in adverse situations and make sure that the RSET calculations would not be underestimated. The evacuation capacity of each slide was set as 24 person/min, referencing the experimental and simulation research data from Xie et al. [36], Zhang et al. [37], and Cao [38]. There is no upper limit for the evacuation capacity of the tunnel's entrance and exit.
