New Guidelines & Experience in Europe for Early Fire Detection in Road Tunnels 2011 Symposium on Tunnels and ITS, Bergen/Norway, June 15/16t - 2011 Willy Schuldt Abstract Major fire incidents in road tunnels with a larger number of casualties led to a discussion and measures to increase the safety in road tunnels. Technical evaluation of the various incidents showed that the monitors typically installed for visibility monitoring and ventilation control gave early and reliable signals indicating the start of a fire. In most cases, fire start due to technical reasons on the vehicles, such as defective engines, turbocharger, tires, etc. Such defects typically lead first to the development of smoke before an actual fire starts (in many cases, there s even no fire, just smoke). In some European countries, namely in Switzerland and Germany, new guidelines for the construction and ventilation of road tunnels have been introduced, including recommendations for a suitable and reliable early fire warning detection system. SIGRIST-PHOTOMETER was the first company developing and manufacturing a new smoke/fire detector addressing the needs for this application. In the meantime, several tunnels in Switzerland and Germany have been equipped with this monitor. The longest experience data over more than three years have been collected in one of the world largest and busiest road tunnel, the Gotthard in Switzerland. This paper gives an overview about the new guidelines and experiences for early fire detection in road tunnels. Introduction The following table gives an overview about major fire incidents over the past 30 years 1)» 1978 Velsen (The Netherlands) 5 fatalities and 5 injured» 1979 Nihonzaka (Japan) 9 fatalities» 1982 Caldecott (USA) 7 fatalities and 2 injured» 1983 Pecorile (near Genova, Italy) 8 fatalities and 22 injured» 1989 Brenner (Austria) 2 fatalities and 5 injured» 1995 Pfänder (Austria) 3 fatalities due to car crash» 1996 Isola delle Femmine (Italy) 5 fatalities and 10 injured» 1999 Mont-Blanc (France-Italy) 39 fatalities and 25 injured» 1999 Tauern (Austria) 12 fatalities (7 due to car crash)» 2001 Gleinalm (Austria) 5 fatalities due to car crash» 2002 St. Gotthard (Switzerland) 11 fatalities» 2005 Frejus (France Italy) 2 fatalities and 6 injured» 2006 Via Mala (Switzerland) 11 fatalities and 5 injured The next table summarizes fire and smoke incidents, including false alarms (in brackets), in the Gotthard road tunnel between 1996 and 2007 2)» 1996: 2 (8)» 1997: 10 (15)» 1998: 8 (9)» 1999: 10» 2000: 8» 2001: 3» 2002: 7 (32)» 2003: 6» 2004: 13 (19)» 2005: 5 (24)» 2006: 9 (27)» 2007: 14 (31) Richtlinien für die Tunnelsicherheit - Symposium SIGRIST, 26. Oktober 2006 8 Bundesamt für Strass en ASTRA Page 1 of 6
Technical investigations of tunnel fires As mentioned in the abstract, most fires (except in case of an explosion, e.g. after a crash) start with smouldering fires due to technical problems: overheated engine, turbocharger, blocking brakes, tires, etc. Smouldering fires develop smoke and poisonous gases: people are normally killed because of the gases, not because of the fire! The visibility quickly decreases. Investigations after each incident showed that the instruments installed for visibility monitoring always gave the first clear sign. Video detection is not reliable as demonstrated several incidents (e.g. in the Frejus tunnel on 7.4.08, the smoke was not detected!) Fire cables only react once a real fire starts and there can be delays of several minutes as shown during a fire test in Runehamar, Norway in March 2007 3). Visibility monitoring Visibility is measured in units expressing the extinction of light over a certain distance, E/m (extinction per meter) or, more convenient, me/m (milli-extinction per meter). 1 me/m means the light intensity is reduced by a factor of 10 over a distance of 1 000 meters. For visibility measurement, two technical possibilities are known: 1. Transmission measurement over a distance of 6..20 m for direct monitoring of the light extinction 2. Scattered light measurement of the dust concentration in the tunnel, which is mainly caused due to the light extinction Visibility measures in normal operation the soot from exhaust fumes, wear debris from tires, dirt particles, etc. In case of fire, visibility measures the soot particles and other products produced by the fire. There are other components which can affect the measurement, mainly fog/water steam. This will falsify the reading and trigger a wrong alarm. Instruments based on scattered light can eliminate this effect by implementing heating elements. Systems working on transmission measuring principles cannot compensate and, therefore, are not considered for the use of early smoke/fire detection. In Europe, the following levels of visibility are typical in road tunnels:» Normal Traffic < 5 me/m» Heavy Traffic ~ 5 me/m» Traffic Jam ~ 7 me/m» Closing of the tunnel 12 me/m» Fire > 15 me/m Experiences from incidents demonstrated that the visibility values increase quite quickly over 15 me/m; however, this information was not used for smoke/fire detection. In the worst case, more fresh air was pumped into the tunnel by the ventilation system, based on the assumption that the emission is too high. Consequences: the fire gets more oxygen and the smoke is distributed in the tunnel! European Guidelines for Visibility (Smoke) Detection Up to now, only a few regulations and guidelines are implemented. Most of them are very general guidelines giving some indication about the minimum required safety standards in a road tunnel, such as the guideline 2004/54/EU and the RABT 2003 guidelines in Germany. The clearest recommendations have been set by the ASTRA 2007 (Federal Road Authority, Switzerland), giving concrete guidelines for fire detection in road tunnels. Other European countries just started to look into this topic and may eventually follow and adopt existing guidelines. The two existing guidelines from Germany and Switzerland list the following: - RABT 2006, Germany 4) : Visibility has to be measured in distances of maximum 300 meters. To support fire detection, distances for visibility monitoring can be reduced to 150 meters. - ASTRA 2007, Switzerland 5) : The automatic system must be able to localize a fire within a maximum distance of 100 meters in less than 60 seconds. The automatic fire detection system shouldn t trigger more than one faulty alarm for every 2 Km per year. Page 2 of 6
Tunnel construction and ventilation control New tunnels in many cases are now equipped with an intermediate ceiling and fire dampers, as shown in the pictures below: The intermediate ceiling consists of a fresh air section and an exhaust section with the fire dampers. The control of the ventilation system has been modified as well. During normal operation, a uniform distributed vertical suction of the exhaust air is maintained. The intensity is adjusted by the level of visibility, measured by the corresponding visibility monitors. In case of a fire, the ventilation system works in a different way: first, the suction of the exhaust air is increased as long as the object is still moving. Once the object comes to a stop, the position is localized, the fire dampers open and the flow direction of the ventilation is changed in such a way that air is directed towards the place of the smoke/fire and escapes through the intermediate ceiling (see pictures below): In this way, the affected area is limited to a minimum, allowing people in the car to leave the vehicle and escape from the tunnel via the emergency exits. Such systems have consequences on the density of the installed monitors. As the requested detection time should be less than 60 seconds, the installation density will depend on the flow velocity. Examples: a tunnel with separate traffic tubes for each direction and a flow velocity of 5m/s requests a distance of 250m between each monitor. For a tunnel with two-way traffic and a flow velocity of 2m/s, the distance is 100m. This makes it very clear, that the standard visibility monitors installed for ventilation control are too expensive and too sophisticated to be used as smoke/fire detectors. However, it is also clear, that commercially available smoke detectors as used for buildings are not suitable because they are not designed for the tough environmental conditions as present in road tunnels and they re not equipped with the necessary communication options, e.g. industrial Profibus. New concept for a smoke/fire detector Based on the experience and guidelines set, a new concept of a smoke/fire detector was necessary. It should fulfil the following requirements: Fast detection of smoke/fire incidents, low rate of faulty alarms, unsusceptible towards fog, pollution, spider webs, etc., minimum maintenance requirement, preferably possible to be done by the operator, and favourable cost of ownership. Page 3 of 6
The sensor developed by SIGRIST-PHOTOMETER AG fully complies with all these requests. The sensor offers the following features:» Measuring principle: scattered light» No moving parts» Response time: T90 in 5 second» Integrated temperature sensor» Fog elimination with heating elements (option)» Signal output via relays or Profibus DP» Installation on the wall, ceiling or in the intermediate ceiling of the fresh air channel or even integrated into the fire damper system The FireGuard response to smoke is as follows: After approx. 30 to 60 seconds, depending on the distance of the visibility monitor from the fire, a limit of 30 me/m has been exceeded. At the same time it could be demonstrated, that approx. 30 m/em is a realistic threshold for a smoke alarm (this value is also mentioned in the ASTRA guidelines). The alarm levels have to be established individually for every tunnel as ventilation and pollution levels can differ significantly. ST [me/m] Brandversuch 26.05.06 160 140 120 100 80 60 40 20 0 00:00 00:30 01:00 01:31 02:01 02:31 03:01 03:32 04:02 04:32 05:02 05:33 06:03 06:33 Zeit [Min.] The FireGuard has been thoroughly tested in a number of field tests in Switzerland, Germany, and Norway. Practical experiences 800 Fire Test FireGuard at 62.5m FireGuard at 125m VisGuard at 125m 14 The Norwegian Road Authority conducted a fire test in Norway (Runehamar) to compare the reaction and sensitivity of fire detection systems based on optical cables and scattered light 3). The test demonstrated that both systems detect fire quickly and reliably. However, scattered light instruments detected the start of the car fire 2½ minutes earlier because there was only smoke in the beginning: Vi si bility [P L A = m 600 400 200 12 Te m pe rat ur e 10 S en so r C 8 ab le 6 0 16:00 16:01 16:02 16:03 16:04 08.03.2007 dat Direct Car fire (realistic incident): FireGuard triggered alarm after approx. 40 sec. The sensor cable only reacted after more than 3 minutes! This is because in the beginning there was only smoke (in 90% this is the case!) Page 4 of 6
Since September 2007, the Gotthard main tunnel (16.9Km long) and the tunnels before and after the main tunnel have been equipped with a total of 210 FireGuard. The sensors are installed every 100 meters in the intermediate ceiling and the whole ventilation and alarm concept was modified as explained earlier. The operator made it very clear that the detection of smoke and fire in the tunnel must be automatic, including all initial actions to safe valuable time and increase safety. Since the installation of the system, two major incidents occurred. In both cases, the new system proved to fulfil all requirements as expected as demonstrated below 6). 1. Incident: Defective tire on a truck Truck with a defective tire generating slight smoke. Video showed no reaction (no detection of smoke). Truck was moving about 1 Km into the tunnel. Time interval of the pictures = 32 s. slight smoke visible smoke clearly visible smoke truck stopped System reaction: Evaluation of the alarm sequence, determination of the truck speed, localisation of the truck when it stopped. 07:10:56 Beginning of an increased distributed suction of the ventilation 07:12:15 Change to a concentrated suction at the place where the truck stopped 2. Incident: Defective turbocharger on a truck Page 5 of 6
Truck with a defective turbocharger, developing strong smoke. Truck was moving South, about 2.1 Km into the tunnel. Air velocity direction North, 2.2 m/sec. Max. area covered with smoke: 2.5 Km System reaction: Evaluation of the alarm sequence, determination of the truck speed, localisation of the truck when it stopped. 20:32:31 Beginning of an increased distributed suction of the ventilation 20:35:51 Change of the flow direction towards the place where the truck stopped (S = 1.8 m/sec, N = 3.2 m/sec) Total time required to remove the smoke = 30 minutes Summary Visibility monitoring is now used not only for the ventilation control, but also for early fire detection. It allows detecting smoke and smouldering fires in an early stage. Real fires are typically detected 2 to 3 minutes ahead of fire sensor cables. For the installation, new requirements concerning the instrument density and mounting place have to be considered. Alarm levels for fire detection must be defined individually for each tunnel. In the past years, more than 1 000 FireGuard sensors have been installed in different European countries. References 1) Alain Jeannerett, ASTRA Switzerland: Guidelines for tunnel safety, Tunnel symposium Sigrist, Hergiswil/Switzerland, October 26, 2006 2) Urs Grässlin, Lombardi Engineers, Switzerland: Measuring technique for ventilation control and fire detection in the Gotthard tunnel, Tunnel symposium Sigrist, Hergiswil/Switzerland, October 26, 2006 3) Aralt, T.T., Nilsen, A.R.: Automatic fire detection in road traffic tunnels, Tunnelling and Underground Space Technology (2008), doi:10.1016/j.tust.2008.04.001 4) Guidelines for the equipment and operation of road tunnels, RABT: Edition 2006 5) Swiss government, Federal Road Authority, ASTRA: Guideline Fire detection in road tunnels, Edition 2007, V 2.00 6) Urs Grässlin, Lombardi Engineers, Switzerland: Experiences from the Gotthard tunnel by using visibility, smoke and fire detection, Tunnel symposium Sigrist, Brunnen/Switzerland, October 29, 2008 Page 6 of 6