Travelling in tunnels is an everyday activity for millions of people around the world. Prior to the Mont Blanc tunnel fire in 1999, many of these people probably assumed that sufficient fire safety measures were installed in modern tunnels to cope with fires as and when they occur. The deaths of 39 people in the Mont Blanc tunnel brought about the realisation that a fire in a tunnel can lead to a disaster. The public, particularly in Europe, began to question the safety of tunnels that they passed through every day. The issue was highlighted again two months later when a vehicle collision in the Tauern Tunnel, Austria, lead to the deaths of 12 people. The conception that train tunnels are somehow safer from fire than road tunnels was shattered the following year when a funicular train fire near Kaprun in Austria was responsible for the deaths of over 150 people. Despite efforts to raise levels of fire safety in European tunnels since these events, another fire in an Alpine tunnel occurred in 2001; a head on collision involving two heavy goods vehicles (HGV) in the St. Gotthard tunnel lead to a fire which resulted in 11 fatalities and the collapse of 250m of the tunnel lining.
Surprisingly, the recent spate of tunnel fires does not seem to have rekindled interest in carrying out experimental tunnel fire research. Perhaps there is a feeling that sufficient experiments have already been carried out to understand fire and smoke behaviour in tunnels, it is just that the fire protection capabilities of existing tunnels (some of which are several decades old) may not be up to modern standards.
Inquiry
The first proper tunnel fire test series was carried out in an abandoned rail tunnel at Ofenegg in Switzerland in 1965à. Twelve tray fire tests were carried out to investigate the influence of different ventilation systems (natural, semi-transverse and longitudinal) on fire development and smoke movement. Some sprinkler configurations were also tested. For the first time, people had a reasonably clear idea of what a fire in a tunnel might be like. Over the next three decades a number of fire test series were carried out in tunnels. In the early/mid 1990s two extensive fire test series were carried out; the EUREKA EU499 “Firetun” project in Europe (the majority of fire tests were carried out in an abandoned single-lane tunnel in the north of Norway)á and the Memorial Tunnel Fire Ventilation Test Program (MTFVTP) in the US (all tests carried out in an abandoned twin-lane tunnel in Virginia)³.
The MTFVTP series was funded by the US Federal Highway Administration and the Commonwealth of Massachusetts for the Boston Central Artery Tunnel project4. The project cost approximately US$40M and involved a series of 98 diesel tray fire tests. It was largely concerned with investigating the capabilities of natural, semi-transverse, fully-transverse and longitudinal ventilation in controlling smoke from a range of fires, nominally 10, 20, 50 and 100MW in severity. Some sprinkler/deluge systems were also tested. These tests greatly increased scientific knowledge of smoke behaviour and control in tunnel fire situations, but did not produce much data on the behaviour of the sort of fire loads that are commonly found in tunnels, i.e. vehicles.
The EUREKA project, on the other hand, concerned itself with fire testing realistic tunnel-fire loads, including cars, train carriages, subway cars and a heavy goods vehicle (HGV). Fire tests involving wooden cribs and heptane trays were also carried out. This project cost approximately US$10M and was funded by a number of different organisations from Austria, Finland, France, Germany, Italy, Norway, Sweden, Switzerland and the United Kingdom. Although the project was proposed, organised and administered by the German collaborators, the experiments were carried out by a team of engineers and scientists from each of the collaborating countries. In all, 21 fire tests were carried out in the tunnel at Hammerfest in Norway, while other tests were carried out in labs and other tunnels across Europe to supplement the Hammerfest data. The objectives of the project were to gain understanding of (i) fire phenomena, (ii) escape, rescue and fire-fighting possibilities, (iii) the effect of the surrounding structure on the fire, (iv) reusing the structure (damage done, time required for redevelopment etc.). (v) accumulation of theory, and (vi) formation, distribution and precipitation of contaminants.
Since the fire in the Mont Blanc tunnel occurred, a number of other tunnel fire tests have been performed. However, there is a slightly worrying trend emerging.
Sophistication?
During the EUREKA test series, two car fire tests were carried out. A car fire test was also carried out in a tunnel in Sweden in 19975. Aside from these three tests, no fully instrumented (i.e. including measurement of the heat release rate of the fire) car fire tests have been carried out. That is not to say that no more cars have been burned in tunnels under test conditions, several have, but the trend in recent car fire tests is not to investigate how cars burn in tunnels, but rather to investigate the capabilities of certain tunnel fire protection products, such as deluge systems or tunnel lining systems6. The worrying trend is that these tests are not research driven, but are commercially driven.
Another slightly worrying trend is the use of computer modelling in tunnel fire safety. Today’s computer fire models (many of the leading ones are based on the principles of Computational Fluid Dynamics or CFD) are highly complex pieces of software able to model fire and smoke behaviour in very complex tunnel configurations. They are also able to produce some very pretty pictures to convey the meaning of their numerical calculations to a wide audience. However, it is important to realise that the results produced by a computer model are highly dependent on the assumptions and knowledge of the model operator. However good a model’s results may look, it is important to question the assumptions behind them. Having said all that, CFD and other fire models are excellent tools in the hands of a well informed operator.
Today’s fire models are based on today’s understanding of fire phenomena. Although some advances in our understanding of fire phenomena have been made using computer fire models (for example, it was CFD that first identified the “trench effect” following the Kings Cross Station fire in 19877), there is only a limited amount of new things that can be discovered using a model.
More inquiry
Research into tunnel fire phenomena is still ongoing. The Swedish National Testing and Research Institute (SP)8 is continuing to perform experimental fire tests in lab scale, medium scale and full scale tunnels.
Some recent research has approached existing experimental tunnel fire data from a different perspective and attempted to squeeze as much information out of it as possible. It has occasionally been observed that fires in tunnels seem to be more severe than similar fires in the open air, but this effect has not been adequately investigated until recently. By comparing heat release rate (HRR) data (taken by many to be a good measure of the severity of a fire) from vehicle, wooden crib and fuel tray fires in tunnels with HRR data from similar fires carried out in the open air, it has been discovered that the tunnel itself can enhance the HRR of a fire by up to four times. Further investigation has shown that the degree of enhancement varies with the ratio of fire width to tunnel width in a cubic manner (see graphic):
Where _ is the degree of enhancement, WF is the width of the fire object and WT is the tunnel width. The relationship appears to hold for all vehicle, wooden crib and tray fires up to approximately half the width of the tunnel9 (see figure 1). It has not yet been adequately investigated for fires larger than half the width of the tunnel and applies only to naturally ventilated tunnels.
In tunnels with longitudinal ventilation systems, the ventilation tends to have a further effect on the HRR of a fire. By comparing HRR data from tunnel fire experiments in longitudinally ventilated tunnels with HRR data from similar experiments in naturally ventilated tunnels, the influence of longitudinal ventilation has been investigated. It has been found that different types of fire behave in different ways when subject to longitudinal ventilation; the HRR of HGV fires can be significantly enhanced, even by low ventilation rates, whereas the HRR of small tray fires (less than half the width of the tunnel) may be reduced under the same ventilation conditions10. There is no simple equation to describe the complex relationships between ventilation velocity and HRR enhancement, but the relationships can be approximated by the graphs in figure 2.
Into the unknown
Although much progress has been made in recent decades regarding tunnel fire phenomena, there are still many unanswered questions including:
In November 2000, a supposedly “fireproof” train in a tunnel in the Austrian Alps caught fire and lead to the deaths of 155 people. While many factors contributed to the disaster, one of them was thinking that a vehicle can be fireproof.
If we ever decide that we have solved the problem of tunnel fires we will be caught off guard when the next disaster occurs.
Related Files
Degree of enhancement equation
Fig 1 – the relationship between fire enhancement and tunnel fire dimensions. Some experimental data points are overlaid
Fig 2 – the relationship between HRR enhancement and ventilation velocity† for (a) a HGV fire and (b) a small/ medium sized pool fire
