Tunnel Fires, Smoke Control and Life Safety

By Ricky Carvel


Traditionally, fire strategies have relied on the common assumption that people and smoke do not mix well. In the event of a fire, we generally use compartmentation, protected routes, smoke management, etc., to try to keep escaping people away from the smoke, and keep smoke away from escaping people. If we can ensure this happens, regulators are generally satisfied that we have ensured life safety and that our fire strategy is acceptable.

What happens, though, when, due to the nature of the specific environment we are working in, we cannot ensure that the people and the smoke are kept separate? In those circumstances, regulators generally are satisfied that we do all that is possible to maximize the number of people who are kept away from smoke, and accept that small numbers of people may be exposed to smoke. Readers of a certain age and a certain taste in movies will recall the primary message of the second Star Trek movie, “The Wrath of Khan”: “The needs of the many outweigh the needs of the few.” The same logic is commonly applied in our fire strategies. Spock would be proud.

 However, those same readers may well recall that the message of the third Star Trek movie, “The Search for Spock,” was the exact opposite: “The needs of the few outweigh the needs of the many.” It wasn’t logical then and seems illogical now, when we are looking at smoke control, but sometimes, as will be shown, this could be the right thing to do.

A Train in a Tunnel

The scenario of interest here is that of a high-speed, inter-city passenger train in a long tunnel, forced to stop due to a fire. This tunnel is equipped with regular cross-passages to a place of safety, much like the Channel Tunnel or the Gotthard Base Tunnel. In our imagined scenario, these cross-passages are about 325 to 375 m apart. This tunnel, again much like the Channel Tunnel or Gotthard Base Tunnel, has a ventilation system able to blow smoke longitudinally along the tunnel in either direction.

Ideally, the train should stop so that the train doors line up perfectly with the cross-passage doors. If that happened, all the passengers could walk along the train to the exit beside the cross-passage and hop directly into the safety of the cross-passage. However, accidental fires are not ideal, so here we assume that this does not happen.

Indeed, for a worst case scenario, we assume that the train stops so that the fire lines up with one cross-passage door, rendering it useless for egress. Our passengers now have to evacuate away from the fire, possibly on the train to begin with, then in the tunnel for the remainder of their 375 m escape. This is shown schematically in Figure 1.

Figure 1 – Schematic of egress route, along train, then in tunnel.

For simplicity, we assume that the passengers are safe from smoke ingress while on the train, and only are affected by smoke, if smoke is present, once they exit the train and begin the second stage of their egress in the tunnel itself. Of course, the presence or absence of smoke on the egress route depends entirely on what we decide to do with the smoke control system.

Logic (and common practice) dictates that if the fire is near the rear end of the train, then the ventilation system should be used to direct the smoke away from the majority of the people; that is, blow from the front of the train toward the rear. If the fire is near the front, then the opposite ventilation direction should be adopted. If the fire is in the middle of the train, then either strategy could be adopted, because you have to save someone, don’t you?

That logic doesn’t necessarily apply here, because smoke control is not independent of fire behavior; fire behavior is itself dependent on air movement. If we blow air to clear smoke along a tunnel, we are also blowing air on the fire, which will change the fire dynamics and hence change the properties and quantity of the smoke itself.

Does this really matter, though? To find out, we carried out an analysis of all the factors at work in this scenario.

Putting the Jigsaw Together

The solution to this problem requires us to consider the following inter-connected questions—what is the effect of:

  • ventilation on fire size?
  • ventilation on smoke production?
  • ventilation on smoke toxicity?
  • ventilation on smoke stratification?
  • ventilation on smoke movement?
  • smoke on human behavior?
  • smoke on egress speed?
  • smoke and heat on life safety?

It is only by putting together all these pieces of the puzzle that we can see the overall picture and assess the suitability of the various possible fire scenarios. Further details of this study, including how each of the above puzzle pieces is addressed, some using CFD, some using hand calculations, can be found in the proceedings of the 2016 ISTSS conference [1], and the 2015 ISAVFT conference [2]; a larger paper is in preparation for submission to Fire Technology in due course.

For simplicity, this article considers only four scenarios:

  1. Fire in the middle of the train, with longitudinal ventilation
  2. Fire in the middle of the train, without longitudinal ventilation
  3. Fire near the rear of the train, with longitudinal ventilation
  4. Fire near the rear of the train, without longitudinal ventilation

It is acknowledged that “no flow” conditions hardly ever occur in tunnel environments, so the contrast here is between scenarios where the ventilation system is used to intentionally create longitudinal flow, versus scenarios where the ventilation system is used to minimize longitudinal flow.

The scenarios are ranked largely on the basis of fractional effecting dose (FED) calculations, using Purser’s methods [3], as detailed in the SFPE Handbook [3]. In each scenario, FED is calculated for the first escaping passenger and the last escaping passenger, accounting for the length of time spent in the tunnel and exposure to toxic gases and heat. It is acknowledged that the absolute values of FED calculated in this way may not be accurate, but that the relative values between scenarios will indicate the safer fire strategies.

The example under consideration is a typical modern, 18-car, high-speed train. When the fire is in the middle of the train and the passengers detrain from the foremost or rearmost carriages, all passengers in these scenarios enter the tunnel environment about 180 m away from the fire. Exposure to high temperature is therefore limited in both conditions.

It is clear from Figure 2, below, that the FED due to toxic exposure is considerably greater for those escaping downstream in the ventilated scenario, compared to the low-flow case. Note: FED data are not presented for upstream egress in the ventilated case, since no smoke or heat exposure occurs in this scenario. It is apparent that in the low-flow scenario, even though there is some smoke exposure, the FED values remain very low for passengers leaving the train in either direction.

Figure 2 – FED for first escaping passenger in each scenario, with fire in middle of train.

When the fire occurs near to the rear of the train, there are many more people upstream of the fire than there are downstream. However, the few who are downstream are close to the fire when they enter the tunnel and must spend more time in the tunnel to reach a place of safety. The FED data for the first egressing passenger in this scenario are shown in Figure 3, below.

Figure 3 – FED for first escaping passenger in each scenario, with fire near rear of train.

Again, it is apparent that the FED values for the egressing passengers in the low-flow scenarios are considerably lower than those for the forced-flow scenario. If the FED values are taken as realistic, it should be noted that the first egressing passenger in the downstream ventilated scenario attains a FED of above 1.0 before the cross-passage is attained. Such a value would usually be understood to entail incapacity or fatality.

This relates to the first egressing passenger. Things are considerably worse for the final passenger to detrain, who encounters a lethal FED within seconds of entering the tunnel. Conversely, with a low-flow strategy, the final egressing passenger in the downstream direction attains a FED of only 0.15 in the 12 minutes it takes to reach the safety of the cross-passage door.


These results seem extreme—and they are. The implication is that ventilating a train fire exposes all passengers downstream to potentially lethal conditions. If we aim to minimize the flow, all passengers may be exposed to smoke, but none of them experience anything close to lethal conditions. To save the few downstream, we need to expose the many to potentially toxic smoke. This is not a decision to be made lightly.

These conclusions come with a huge disclaimer: These results only apply to the very specific case studied, and rely on the assumptions inherent in the method (see references for details). These results cannot be generalized and are not transferrable to any other scenarios. Different scenarios will require further analysis.

These results do, however, include all the fire dynamics, smoke behavior and egress behavior that we could include in the study. Our study explicitly included the assumption that the underventilated fire produced a greater proportion of toxic gases than the ventilated fire did, yet the results show that toxic exposure is greater in the ventilated case. The reason for this is that the fire grows much faster in the ventilated scenario compared to the underventilated one, and this behavior completely dominates the results. The choice to blow fresh air at a fire is the choice to have a large, fast-growing fire. All studies of vehicle fires in tunnels show this.

Ventilation can be used to move smoke, but it also considerably changes fire behavior. Designers of fire strategies must always remember this: You cannot divorce smoke management from fire behavior, particularly in a tunnel. Careful consideration of both fire and smoke dynamics will help passengers “live long, and prosper.”

By Ricky Carvel with BRE Centre for Fire Safety Engineering, University of Edinburgh


[1] Winkler, M. & Carvel, R. “Ventilation and egress strategies for passenger train fires in tunnels,” 6th International Symposium on Tunnel Safety & Security, Montreal, Canada, March 2016. pp. 453–464.

 [2]Winkler, M. & Carvel, R. “The effect of longitudinal ventilation on tenability during egress from passenger trains in tunnels during fire emergencies,” Proceedings of the 16th International Symposium on Aerodynamics Ventilation and Fire in Tunnels, Seattle, USA, September 2015. pp. 211–226.

 [3]Purser, D.A. & McAllister, J.L. “Assessment of hazards to occupants from smoke, toxic gases and heat,” SFPE Handbook of Fire Protection Engineering, 5th edition (2016), pp. 2308–2428.

© SFPE® | All Rights Reserved
Privacy Policy