A numerical study on smoke propagation in metro tunnel fires with different door opening scenarios

Numerical study was carried out to analyze the effect of door opening scenario on the smoke propagation in metro tunnel fires. The smoke back-layering length, critical velocity, and temperature distribution were studied under a total of 8 working conditions. Results show that the opening of emergency exit doors could result in a shorter smoke back-layering length in the train compartment. In such conditions, a smaller driving force was needed to approach the critical velocity. On the other hand, the high temperature area of the tunnel evacuation platform would change under different opening scenarios. The research result could be used as a deeper understanding of the smoke propagation of metro fire and a reference for the formulation of smoke control.


Introduction
When a fire occurs in a running train in a subway interval, normally the train will continue to the front station and organize the evacuation of passengers in the front station.But when the long interval subway tunnel is on fire and the train cannot continue to the front station, it can only stop in the interval tunnel, start the adjacent subway station ventilation exhaust smoke control, in this process a large amount of toxic smoke will first spread in the narrow train car.Followed by the consideration of personnel evacuation, train doors and the emergency ventilation near the source of the fire area are opened.Therefore, it is of engineering significance to consider the characteristics of the smoke entrainment and spread in the double narrow space formed by the train and the tunnel from the perspective of smoke control and personnel evacuation.
There are relatively more domestic studies on fires in the interior of stopped train cars.Numerous studies have investigated the impact of tunnel characteristics on smoke behavior, specifically focusing on the smoke back-layering length and critical velocity.The influence of tunnel slope (Du et al., 2018 [1]; Ko et al., 2009 [2]; Li et al., 2019 [3]), the influence of cross-sectional geometry (Li and Ingason, 2017 [4]; Weng et al., 2015 [5]), and the influence of bifurcation angle (Huang et al., 2020 [6]) in the smoke back-layering length and critical velocity were studied.In the context of vehicle blockage fire scenarios, researchers have focused on understanding the smoke behavior in underground tunnels with different train lengths and longitudinal ventilation velocities.Zhang et al. [7] conducted numerical simulations to study the smoke back-layering length.Hu et al. [8] investigated the critical velocity and transition velocity resulting from train fires in underground tunnels, establishing a relationship between dimensionless smoke back-layering length and dimensionless longitudinal ventilation velocity.Zhu et al. [9] studied the critical velocity and back-layering length by considering both blockage and slope.Tang et al. [10] performed tests to analyze the influence of vehicle obstructions on the back-layering length and critical velocity in longitudinally ventilated tunnels, and they developed a predictive formula for estimating these parameters.These studies collectively contribute to our understanding of smoke behavior in tunnels and provide insights for designing effective ventilation systems to control smoke movement and ensure the safety of occupants during fire incidents.This paper mainly analyzes the impact of mechanical ventilation on smoke control during fire conditions in stopping trains in the area and discusses it through indicators such as critical velocity, smoke back-layering length, and temperature distribution at personnel height.

Physical model
A Beijing subway tunnel is used as a prototype to build a numerical model of the subway interval.The dimension of this rectangular cross-section tunnel is 500m ×4.8m × 5.2m(length× width ×height).The subway train is 19m×2.8m×3.8m(length×width×height),and total length of the subway train is 120m.Subway train evacuation doors mainly include emergency exit doors and side doors.According to the actual measurement and relevant information, the train side doors are 1.5m × 1.8m(width×height) and the emergency exit doors are 1m×1.8m(width×height).

Grid independence
The size of the characteristic diameter of the fire source is determined according to Equation 2-1, and the calculation can reach the characteristic diameter of the fire source of 1.8m for the power of 5MW, for which four grid sizes of 0.16, 0.18, 0.2, 0.24 are selected for the verification of the grid size.The temperature distribution in the near fire source area (60m downstream of the fire source) under different grid sizes of the constructed physical model without mechanical ventilation conditions was compared, as shown in figure 1.According to the simulation results, the longitudinal temperature distribution in the tunnel does not differ much when the grid size is less than 0.18, so a grid size of 0.18m is considered. (1)

Numerical simulation of working conditions
The current numerical simulation working conditions are shown in table 1.According to GB50157-2013, the velocity of the interval is controlled within the range of 2-11m/s.Therefore, the working conditions under different evacuation door opening scenarios are studied corresponding to the simulation of four ventilation velocity, and the air supply direction is to meet the running direction of the train.It should be noted that the return length of the simulated interval train model is a comprehensive consideration of the smoke propagation inside and outside the compartment, while the location of the fire source is used as the starting point for the return flow of smoke upstream.Based on the simulation results, depicted in figure 2, it can be observed that when the mechanical air supply velocity is set at 2m/s, the smoke begins to flow in the opposite direction of the fire source at 150s.At 240s, the smoke spreads towards the 6th train side door, which is located upstream of the fire source.Eventually, at 300s, the smoke flows back precisely to the emergency exit door.Notably, due to the presence of the emergency exit door, there is a sudden expansion in incoming velocity, which prevents further smoke backflow.However, the smoke within the car gradually descends until it fills the entire compartment.Based on the data presented in figure 3, when the mechanical air supply velocity is set at 3m/s, it is observed that at 120s, the air directed from above the fire source effectively prevents the smoke from spreading upstream.However, as the power of the fire source gradually increases, the smoke begins to flow back at 240s.At 360s, the smoke spreads towards the 5th car side door, which is upstream of the fire source.After this point, the smoke essentially ceases to flow back due to the incoming velocity of the air.Subsequently, at 600s, the smoke expands to the 6th car side door, also located upstream of the fire source.Based on the simulation results illustrated in figure 5, when the mechanical ventilation velocity is set at 2m/s, it is observed that the smoke begins to spread in the upstream direction of the fire source at 75s.By 120s, the smoke reaches the fourth train side door, which is located upstream of the fire source.After reaching the side of the emergency exit door at 180s, the smoke no longer continues to spread upstream.Consequently, the smoke remains confined within the length of the train and does not extend further.

Figure. 5. Smoke spread at 2m/s ventilation velocity in the open door scenario of condition 2
The mechanical air supply was carried out under the condition that the train side doors were all opened immediately after the fire, and the air velocity was 3m/s, as shown in figure 6.According to the simulation results, the smoke started to flow back at 90s, and the smoke spread to the emergency exit door position of the train at 213s, after which it did not continue to spread upward.

Figure. 6. Smoke spread at 3m/s ventilation velocity in the open door scenario of condition 2
Continue to increase the air supply velocity to 4m/s for the fire condition under the door opening scenario of condition 2, the smoke started to flow back at 140s, as shown in figure 7. The smoke spread to the emergency exit of the train at 257s, after which the smoke did not continue to spread upstream, and the train car was rapidly filled with smoke, which was still controlled within the length of the train.Based on the data presented in figure 8, in Case 2 where the air supply velocity for the door opening scenario was increased to 6m/s, several observations can be made.Firstly, the smoke does not initiate backflow until 245s, which is a delay compared to the previous scenario.Additionally, after flowing back to the emergency exit door of the train at 359s, the smoke no longer continues to spread upstream.It is worth noting that throughout the simulation, the smoke remains within the length of the train.This simulation result suggests that when comparing the train side door and the emergency exit door, the emergency exit door exhibits a more pronounced inhibitory effect on smoke backflow.However, when only the train side door is open, increasing the velocity shows that the time required for smoke backflow and spreading beyond the length of the train becomes longer.This indicates that the opening of the train side door has a limited effect in countering the buoyancy force of the smoke, but it still influences the overall dynamics due to the inertial force of the airflow.

Effect of mechanical ventilation on temperature distribution under different evacuation scenarios
According to the simulation results of Case 1, the increase of velocity can make the upstream return length of the fire source gradually decrease, and the temperature stratification in the vertical height of the compartment gradually becomes less obvious.After the velocity reaches 4m/s, the downstream of the fire source is filled with high temperature smoke as the smoke no longer spreads upstream of the fire source.At the same time, as the velocity increases, the temperature inside the compartment decreases.According to the temperature slices at this location, as shown in figure9, it can be seen that for the door opening scenario in which the side doors and emergency exit doors of the train are all open in Case 1, the smoke overflow from the car to the outside of the car is mainly through the train side doors near the end of the train.From the side door of the train to the rear end of the train, the overflow of high-temperature smoke from the side door of the train gradually increases, and the temperature of the evacuation platform also increases significantly as shown in figure 10.At the same time, according to the simulation results, it can be seen that as the velocity increases, the area of the evacuation platform at the rear of the train where the temperature suddenly rises increases, and the amount of smoke overflowing from the side doors of the train along the longitudinal length from the fire source to the rear of the train gradually increases.This phenomenon becomes more and more obvious with the increase of time.This is mainly because with the increase in velocity, the inertial force of the incoming velocity makes a large number of smoke quickly flow to the rear of the train, and the velocity of smoke overflow out of the door is less than the longitudinal spreading velocity of smoke by the role of incoming wind velocity.Therefore, with the increase of wind velocity, the longitudinal difference of temperature in the evacuation platform becomes larger.For Case 2, the train side doors are all opened for longitudinal ventilation, and the temperature inside the train car gradually decreases as the velocity increases.It can be seen that although the smoke backflow length is always maintained in this way with 60m on the train, that is, the end side of the train.But according to the temperature distribution, as shown in figure 11, it can be found that the upstream and downstream temperature distribution of the fire source shows an asymmetric distribution.Moreover, with the increase of velocity, the high temperature area at the top of the car upstream of the fire source was shortened and did not reach the upstream train end.This indicates that although the incoming velocity is large enough to completely block the backflow of smoke, the smoke temperature gradually decreases.The temperature stratification on the vertical height downstream of the fire source is no longer obvious, and the upper and lower layers of smoke and air are fully mixed, which is more significant when the velocity is 6m/s.For Case 2, the train side doors are all open, the high temperature area of the evacuation platform is also mainly concentrated in the rear of the train, and the high temperature area in the central area gradually decreases, as shown in figure 12.The phenomenon is the same as Case 1, so we will not repeat it here.

Conclusion
6.This paper investigates the effect of different door-opening scenarios and longitudinal ventilation air velocity on smoke control in the case of a fire in the interior of a stopped subway train.The major conclusions are given as follows.
(1) From the point of view of critical velocity, for the evacuation scenario where the emergency exit doors and single side doors are all open at the same time, there is a critical velocity, which is 4～ 6m/s, making the smoke no longer to spread upstream of the fire source.For the fire conditions in the car, the air volume of mechanical can be increased by air supply and smoke exhaust to both ends of the subway station, when the emergency exit doors and single side doors are opened as much as possible, the smoke can be controlled from spreading upstream of the fire source by lower longitudinal air velocity.
(2) In terms of the back-layering distance and time, when the train single side doors and emergency exit doors at head and end are opened simultaneously, the increase in mechanical ventilation air velocity can effectively reduce the smoke back-layering length.When only the side doors of the train are open, although the moment when the backflow starts are delayed with the increase of mechanical velocity and the smoke backflow velocity is reduced, the increased ventilation velocity cannot effectively control the back-layering distance and can only control the smoke back-layering length within the train length.

Figure. 2 .
Figure. 2. Smoke spread at 2m/s ventilation velocity in the open door scenario of condition 1

Figure 3 .
Figure 3. Smoke spread at 3m/s ventilation velocity in the open door scenario of condition 1According to the findings presented in figure4, when the mechanical air supply is set at 4m/s and all train side doors and emergency exit doors are open, it can be observed that the smoke no longer flows back.This indicates that the inertial force generated by the incoming air is strong enough to prevent the smoke from flowing back above the fire source.Furthermore, the critical velocity at which this occurs is determined to be 4m/s when all train side doors and emergency exit doors are open.

Figure 4 .( 2 )
Figure 4. Smoke spread at 4m/s ventilation velocity in the open door scenario of condition 1 (2) The train side doors are all open and the emergency exit doors are closed.By comparing the smoke spread effects of the two door openings with different velocities, we can further analyze how the incoming velocity enters the train through the side doors or emergency exit doors and stops the smoke from spreading upstream.Based on the simulation results illustrated in figure5, when the mechanical ventilation velocity is set at 2m/s, it is observed that the smoke begins to spread in the upstream direction of the fire source at 75s.By 120s, the smoke reaches the fourth train side door, which is located upstream of the fire source.After reaching the side of the emergency exit door at 180s, the smoke no longer continues to spread upstream.Consequently, the smoke remains confined within the length of the train and does not extend further.

Figure 7 .
Figure 7. Smoke spread at 4m/s ventilation velocity in the open door scenario of condition 2

Figure 8 .
Figure 8. Smoke spread at 6m/s ventilation velocity in the open door scenario of condition 2

Figure 9 .
Figure 9. Temperature cloud of the tunnel centerline for Case 1

Figure 10 .
Figure 10.Temperature cloud of evacuation platform centerline for working condition 1

Figure 11 .
Figure 11.Temperature cloud of the tunnel centerline for Case 2

Figure 12 .
Figure 12.Temperature cloud of evacuation platform centerline for working condition 2

Table 1 .
Simulation of working conditions when starting mechanical ventilation.
4. Analysis of numerical simulation results41.Effect of ventilation on smoke control(1) All side and emergency exit doors of the train are open The mechanical air supply of 2m/s, 3m/s, 4m/s and 6m/s was carried out for the door opening scenarios with all side and emergency exit doors open, respectively, to analyze the smoke spread and control under different velocity conditions.