Effect of Door Openings on Train Fire Scenarios within a Subway Depot

It is challenging to control smoke in the case of a train fire occurred in the subway depot with a complex internal structure. In this paper, the effect of door opening state on smoke behavior characteristics induced by subway depot fire scenario was investigated. A series of numerical simulation were conducted by Fire Dynamics Simulator (FDS) software. Three key parameters were analyzed corresponding to the temperature, visibility and CO concentration for evaluating smoke propagation respectively. Results show that temperature distribution and CO concentration at 2 m height inside the train with train door opening state are lower than those with the train door closed. However, visibility at 2 m height inside the train shows the opposite trend. The results can provide reference in the ventilation system design and emergency evacuation scheme for the subway depot.


Introduction
With the increasing tension of urban land, coupled with the large area of the subway depot, developing subway depot with over-track buildings has become a new tendency.It is characterized by large space, wide area, complex building structure and limited opening area.Once the fire, enough oxygen in a large space will allow the fire to develop rapidly.In the subway depot, there are not only operators, maintenance personnel, but also many trains parked.Improper control can result in significant economic loss and pose a danger to people inside the depot as well as a challenge to fire rescue.
In view of the uncertainty and hazards of fire occurrence and the characteristics of subway depots with over-track buildings, it is necessary to study the smoke dispersion characteristics of fires in the subway depot in order to set up appropriate smoke prevention and exhaust strategies.
Rickard [1] analyzed the smoke characteristics of subway train tunnel fires by full-scale tests, showed the risk in the process of evacuation during the rapid development stage.Zhou [2] investigated the influence of tunnel slopes on the temperature distribution characteristics of a subway.Lee [3] conducted a full-scale experiment on a subway train and derived the maximum heat release rate with all four doors of the carriage open.A series of model tests were carried out in a single-tunnel double-track subway tunnel model to study the temperature variation characteristics of the vault and the smoke diffusion law in the tunnel [4].Wang [5] conducted 1:10 and 1:15 scale train model tests to obtain the smoke concentration distributions for mobile and fixed fire sources.And the relationship between air velocity and smoke distribution was analyzed.Li [6] conducted a 1:10 reduced-scale experiment to study the change pattern of the temperature field in the roof of the tunnel during a fire.And a prediction model was established.The results of the 1:50 reduced-scale experiments conducted showed that the temperature of the bottom layer in hybrid ventilation was much lower than that in natural ventilation [7].
At present, there is still a lack of studies that consider the effect of door opening state on fire development when a subway car fire occurs in the subway depot.This study aims to study the smoke behavior characteristics when a train fire occurs in the subway depot by numerical simulations.

Fire scenario
The scenario is set that a fire occurs in the middle of the train during the process of entering the depot.The specific cases are set in table 1.

Grid system
It is known that the size of the grid has a great influence on the calculation results of numerical simulation.In fact, when the size of the grid is fine to a certain extent, the results of the simulation with the further reduction of the grid and the improvement of the effect is very limited, so it is necessary to carry out the grid size sensitivity analysis [8].The grid size is screened to ensure that the divided computational grid does not affect the final simulation results while saving computational resources.The ratio of the characteristic diameter of the combustion fire source D* to the computational grid size is used as a criterion for grid size selection.Where the D* is calculated by the equation (1).
In this paper, six different size grid systems are set up to simulate the same fire scenario.Figure 2 shows that the distribution of the temperature in the vertical direction at a distance of 30 m from the fire source at 500 s.It can be seen that the distribution curve of temperature changes from rough to relatively smooth as the grid size is continuously reduced.Therefore, the grid size of 0.5 m  0.5 m  0.25 m is chosen to be used in simulation.With a grid size of 0.25 m,D * / =8.592, which is within the ratio range of 4 to 16 recommended by the FDS user's guide [9].

Temperature distribution
In the following analysis, y = 60 m is used to represent the position of the longitudinal section of the center of the fire source.In figure 3, the temperature distribution under the roof at y = 60 m at different moments of Case 1 and Case 2 is compared.It can be clearly seen that at 100 s and 200 s after ignition, the roof temperature distribution directly above the center of the fire are similar.This shows that the doors opening state of the burning carriage has little influence on the temperature distribution under the roof before the HRR of the fire reaches its maximum value.At 400 s and 600 s after ignition, the maximum roof temperature in the case of opening all the doors of the burning train is significantly larger than that in the case of opening the doors of the burning carriage.With the diffusion of smoke, the temperature distribution is basically similar.This demonstrates that opening all doors of the fire train can make the maximum temperature of the center of the fire rise, but it has little effect on the area far from the fire source location.
It can be clearly seen from figure 4 that the temperature at 2 m height inside the train on both sides of the fire source shows a symmetric distribution phenomenon within the 600 s time frame of the fire.At 100 s and 200 s after ignition, the temperature distribution of the roof directly above the center of the fire source are similar, which indicates that the opening and closing of the doors of the train on fire has little effect on the temperature distribution at 2 m height inside the train before the development of the fire source reaches its maximum value.At 400 s and 600 s after ignition, the maximum temperature at 2 m height inside the train with only the doors of the train car open is significantly higher than that of the train with all doors open.As the smoke spreads inside the train, Both the local and average temperatures inside the train when only the doors of the burning train are opened is significantly higher.
It can be seen from figure 5 that at 200 s after ignition, the HRR of the fire source reached its maximum value.During the range of 100 s to 200 s, the opening and closing of the doors of the train on fire had little effect on the temperature rise at 2 m height in the center of the fire source.From 200 s to 600 s, the temperature at 2 m height was lower with closed fire train doors.This indicates that the smoke and heat accumulation in the train had an effect on the temperature distribution of the fire center.

Visibility distribution
It can be clearly seen from figure 6 that the visibility distribution at 2 m height inside the train is similar for both Case 1 and Case 2 at 200 s after ignition.This indicates that the opening and closing of the train door has little effect on the visibility distribution at 2 m height inside the train before the HRR of the fire source reaches its maximum.After 300 s, the visibility distribution at 2 m height in the train basically stabilized, and the visibility at 2 m height in the train with only the doors of the burning carriage opened is significantly lower than that at 2 m height in the train with all the doors of the train opened.When the doors of the train at the far end of the fire source are closed, the smoke accumulation phenomenon is formed at this time.For this reason, lower visibility is observed in Case 2.

CO concentration
Figure 7 shows the comparison of CO concentration distribution at 2 m height.It is clear that the CO concentration distribution at 2 m height in the train shows an increasing phenomenon with the time.
The CO concentration on both sides of the fire source basically shows a symmetric distribution phenomenon in the range of 600 s after ignition.At 100 s after ignition, the CO concentration distribution at 2 m height is basically the same for both Case 1 and Case 2. This indicates that the door opening state has little influence on the CO concentration distribution at 2 m height in the train before the HRR of the fire source reaches its maximum.At 200 s after ignition, the CO concentration at 2 m height in the train with only the doors of the fire train carriage open is slightly higher.Finally, it was obvious that the CO concentration at 2 m height of the train with the train doors closed was significantly higher.It can be concluded that opening the train door is beneficial to the smoke diffusion in the train at 400 s and 600 s after ignition.And the CO concentration in the train can be effectively reduced.

Conclusion
In this paper, numerical simulation by FDS was conducted to investigate fire smoke spread characteristics of the subway depot.Major conclusions can be drawn as follows: (1) When the fire occurred in the train, the maximum value of the roof temperature in the case of opening the train doors is significantly greater than that in the case of closing the train doors.With the diffusion of smoke under the ceiling, the temperature distribution at 2 m height inside the train is significantly higher under train door-closed condition.It was shown that the smoke and heat could accumulate under train door-closed condition, resulting in a rise in the local and average temperature.
(2) During 600 s after fire ignition, the result of visibility in the depot area showed that there is a positive effect on the diffusion of smoke into the reservoir area when the all doors are opened.
(3) The CO concentration at 2 m height inside the train with the train doors closed is significantly higher than that when the train doors are opened at 300 s after ignition.
(4) When the doors of a burning carriage are closed, the temperature, visibility and CO concentration are at a more dangerous level at a height of 2 m inside the train.Therefore, when a fire occurs in the process of the train entering the depot, the diffusion of smoke to the subway depot will be delayed in a short time while the train doors are closed.

Figure 1 .
Figure 1.Model of the subway depot.Figure 2. Vertical temperature distribution at 30m from the fire source under different grid sizes.

Figure 2 .
Figure 1.Model of the subway depot.Figure 2. Vertical temperature distribution at 30m from the fire source under different grid sizes.

Figure 3 .
Figure 3.Comparison of the temperature distribution under the roof at different moments.

Figure 4 .
Figure 4. Comparison of temperature distribution at 2 m height inside the train at different moments

Figure 5 .
Figure 5. Temperature at 2 m height in the center of the fire source.

Figure 6 .
Figure 6.Visibility distribution at y = 60 m at different moments.

Figure 7 .
Figure 7.The CO concentration distribution at 2 m height at different moments.

Table 1 .
Cases of the door opening and closing of a fired train.