The Mont Blanc Tunnel recently reopened, three years after the tragic fire of 24th March 1999 in which 39 lives were lost. The technical and organisational objectives defined in the renovation programme, approved in December 1999, required two years of intensive effort and an investment of over US$273M. Fire tests were carried out in January 2002, as part of the final verification of the new safety equipment.

Testing of equipment, focusing on safety-relevant components, encompassed several levels including factory tests, individual testing of each component, the verification of the integration of the tunnel’s supervision system and global functionality tests. As part of the latter, great attention was devoted to the new ventilation system. The final verification was based on the combined use of cold smoke generators and real fires. Despite their complexity and cost, only real fire tests can provide a genuine verification of the safety level achieved. The new semi-transversal ventilation system, provided by Howden (Paris) and illustrated in figure 1, distributes fresh air along the tunnel through nozzles located every 10m in the lower part of the profile. The exhaust system can’t be used for fresh-air supply and its use during normal operation is limited to particular situations.

Key elements in case of fire are:

  • Reduction to 30% of the fresh-air supply;

  • Extraction of at least 150 m3/s smoke over 600m (6 dampers);

  • Control of the longitudinal air flux in the tunnel by means of 38 pairs of jet fans;

  • Improved shelter ventilation system, supplied through the fresh-air ducts.

The improvement of the smoke-extraction system, required a number of heavy interventions:

  • Smoke-extraction nozzles every 100m (previously 300m), equipped with dampers;

  • Installation of 3 large centrifugal smoke-extraction fans on each tunnel side;

In order to reduce the pressure difference between the exhaust duct and the tunnel, four axial fans were distributed along the smoke-extraction duct and in caverns under the road plane; The smoke-extraction capability exceeds 150 m3/s for every fire location.

Fire tests

The key objectives of the fire tests were to verify the fire ventilation system, particularly smoke-extraction and mastering of the longitudinal velocity and shelter ventilation; and to verify the capability of the fire-fighting rescue teams and their equipment in very low visibility due to smoke destratification.

Several levels of testing were defined:

  • Cold-smoke tests;

  • Low-intensity fires (1.3-1.4MW);

  • High-intensity fires (10MW);

  • Fires with destratified smoke (ca. 3MW).

Cold-smoke tests are comparatively easy and inexpensive. They allow a realistic assessment of the aerodynamic behavior of the system without involving the complexities of real fires. Moreover, a quick cold-smoke test carried out before a fire test allows for a last verification of all systems and procedures.

The choice of the fire intensity is a compromise between the necessity of testing under real conditions and constraints of protecting the equipment and the structure. A heat-release rate of 1.3-1.4MW is representative for small tunnel fires involving cars, which develop typically around 2.5–5MW. The higher heat-release rate of 10MW is representative of high-intensity fires: a “typical” van is expected to develop about 15MW, a bus about 20MW and an HGV carrying combustible goods reaches typically 20-30MW. Significantly higher values, of up to 100-120MW, can be observed over short time intervals or in cases of dangerous goods (which are not allowed in the Mont Blanc tunnel). The tests carried out using destratified smoke have verified the ability of the intervention teams in difficult conditions, with decreasing visibility.

The local air velocity at fire onset is a particularly important parameter for defining a fire scenario, particularly in the case of longitudinal or semi-transversal ventilations. The initial test conditions were subsequently varied, in order to achieve a reasonably wide velocity range, between 1-2m/s to 4-5m/s. The main effects on longitudinal velocity in the tunnel are:

  • Atmospheric pressure difference between the two entrances, dependent on the conditions;

  • Traffic;

  • Ventilation conditions;

  • Fire location;

The injection of fresh air generates a linear increase of the velocity towards both entrances of up to about 7m/s. It was decided to operate with an initial fresh-air rate of about 75% (100% corresponds to about 82.5 m3/s for each of the 8 fresh-air segments), which corresponds to the conditions appropriate for medium-high traffic volumes. The initial effect of traffic in case of fire, characterised by a strong transient for about 5-10 minutes, could not be simulated. In order to simulate the influence of variable meteorological pressure differences it was decided to temporarily install two large axial fans in the tunnel, which will equip the portal air-extraction facility at the French entrance. Additionally, a number of jet fans were used manually for the same purpose. Usually the full system capability and automation was utilised, however part of the testing was carried out with manual control of the ventilation system. This allowed an assessment of the benefits resulting from complex automation.

One of the goals of the cold-smoke tests was to demonstrate how the ventilation system would cope with a very large smoke spread in the tunnel and with unstratified smoke, reproducing the conditions encountered when materials (that are characterised by large smoke production and low energy release) burn. The tunnel was filled with smoke over a length of 1100 –1300m. During the filling process the system’s automatic reaction was inhibited on purpose, and activated only after the desired spread was achieved. This test was carried out at three different locations, characterized by low, medium and high initial longitudinal air velocity. The atmospheric pressure difference during the test was negligible (Table 2).

Even under these extremely pessimistic conditions (the length of the smoke-invaded zone and the missing stratification) the results showed that all systems worked properly. The cold smoke could be perfectly confined between two smoke-extraction openings. This was required, as expected, because of the initial propagation length, roughly 15-20 minutes. Roughly two minutes after activation of the emergency procedure, the longitudinal velocity could be inverted and the direction of smoke propagation was observed a few minutes later. Further tests demonstrated the ability of the rescue and firefighting teams to operate under extremely difficult visibility conditions.

All the fire tests were carried out under controlled conditions, using shallow metallic fuel pools with a circular footprint and burning surface of 1.13m2. The fuel was domestic oil, which produces large amounts of dense black smoke. The fire tests were carried out at low (1 pool) and medium-high intensity (6 pools). The unitary heat-release rate was estimated at about 1.4MW before testing, based on a fuel consumption rate of 0.034kg/s and a calorific power of 41MJ/kg. The observed heat-release rate, based on fuel amount and total combustion time, was significantly higher than expected for the larger fires, and slightly lower for the smaller ones. During the higher-intensity fires the typical temporal development of the heat-release rate observed in medium-intensity tunnel fires was taken into account, as the fuel pools were ignited in pairs at intervals of 2 minutes:

  • to ignition of 2 fuel pools (HRR 3MW)

  • to +2 min ignition of further 2 fuel pools (HRR 7MW)

  • to +4 min ignition of last 2 fuel pools (HRR 10MW)

The peak heat-release rate was thus achieved roughly 5 minutes after ignition. Fire duration was about 25-30 mins. The system was operated to simulate an automatic response, based on the reaction of the event-detection system (video images):

  • to +1 min fire presumption

  • to +2 min confirmation of fire alarm and start of smoke extraction

Protection of the tunnel equipment and structure was one of the main concerns during the tests, to reduce disassembling, cleaning and verifying the functionality afterwards. For this reason it was decided to carry out all the fire tests at one location, PM 8780, about 2800m from the Italian entrance, between shelters 28 and 29 (about 3/4 of the tunnel length). The longitudinal air velocity at this location is mostly between 1-5m/s towards the Italian entrance. Representative conditions for other positions along the tunnel were obtained by varying the initial conditions, as discussed above. Clear indications for the required protection level of infrastructure was derived from the tests carried out early in 2000 (Brousse et al. 2001). These tests were ordered by the French instruction judge for a better understanding of the dynamics of the fire on 24 March 1999.

Delicate equipment (lamps etc.) were removed along a length of 500m, everything else was protected (in particular the cabling). The main protection consisted of isolating panels, with a simple layer over 400m, double over 40m and triple over 20m. An additional 50m on each side was protected by glass wool. The instrumentation was not very heavy, as the testing was not purely scientific. Attention was devoted to the monitoring of cable temperatures under the protection panels with thermocouples to avoid damage. The measurements taken were mainly: temperature using K-type thermocouples, air velocity using bidirectional Mac Caffrey anemometers, some opacimeters, paramagnetic analysers for oxygen concentration and video equipment.

Low-intensity fires

These fires test the performance of the ventilation in case of low-intensity fires. They differ from larger fires because of the weaker smoke production, but also because of the lower stability level of thermal and optical stratification. The tests were carried out at a heat-release rate of about 1.4MW (1 fuel pool). The results were excellent and from both a qualitative and quantitative point of view, very similar to those for the higher heat-release rate, described below. Based on these results the thermal protection of the tunnel equipment was reinforced.

This test series, carried out with an initial air speed of 4-4.4m/s, also proved the superiority of the automated system response compared to that of manual control of the ventilation equipment, even if the latter was carried out by one of the experts responsible for the development of the control system. This was especially true when controlling longitudinal air velocity.

High-intensity fires

The main test series, with a heat-release rate of about 8-10MW (6 fuel pools), was in January. The tests were carried out with a longitudinal velocity of 2 and 4.5m/s, directed towards the southern (Italian) entrance. The test conditions are presented in Table 2. On the southern side, along the initial air velocity direction, the following observations were made:

  • Initially, during the 2 minutes preceding the activation of the smoke-extraction system, rapid smoke propagation in southern direction with reduced visibility in the lower part of the tunnel;

  • starting about 4 minutes after ignition, return of the smoke front, due to the combined effect of the reduced longitudinal velocity at the fire location and the powerful smoke extraction;

  • rapid improvement of the visibility conditions;

  • stabilisation of the smoke front about 150m south of the fire site, corresponding to the second smoke extraction opening.

The observations on the northern side were:

  • No smoke propagation during the 4-5 minutes following ignition, due to the initial air velocity and smoke extraction;

  • later on, slow propagation of a perfectly stratified smoke layer (backlayering);

  • stabilisation of the smoke front about 150m north of the fire site, corresponding to the second smoke extraction opening.

In the final state, achieved after a few minutes, the smoke was perfectly confined to a distance of 300m, delimited by 4 of the 6 active smoke-extraction openings. This configuration was very stable until extinction. The stratification and visibility levels, initially slightly perturbed during the propagation in southern direction driven by the initial velocity, was finally perfect on both sides of the fire, up to approximately 3m from the floor. Based on fire duration and fuel consumption, a higher average heat-release rate than expected could be determined, about 10-11MW, corresponding to about 1.5-1.7MW/m2 fuel.

The shelters ventilation system performed perfectly and no smoke intrusion was observed. This was in verified in the shelters closest to the fire site (28 and 29) where the protection doors stayed open during most of the test time.

Destratified smoke

This somewhat a-typical test was conceived for verifying the capacity of the firefighting teams to operate in dense smoke under difficult conditions and the equipment, in particular Citilog, Ereca and Plettac’s thermosensitive video equipment. The smoke was produced by means of two fuel pools, with a heat-release rate of 2.8MW (estimate before testing) or 3.2MW (estimate after testing). A virtually complete optical destratification was achieved using two small portable fans, activated close to the fire. Additionally, the two dampers closest to the fire were closed. In this way the tunnel was filled with destratified smoke over a length of 300m, between shelters 28 and 29.

The results were entirely satisfactory. The new firefighting vehicles could travel safely in the smoke zone using the thermosensitive video equipment, even if at reduced speed. All obstacles could be localised safely and the fire was clearly visible. There was no smoke penetration into the open shelters even with vehicle movement.

Results and conclusions

The fire tests proved the functionality of the equipment, in particular the new ventilation system, which entirely satisfied ambitious programmatic goals. They also confirmed excellent levels of training and preparation of the rescue teams, and their equipment.

From the point of view of fire ventilation the main findings can be summarised as follows: After the unavoidable initial smoke propagation, the longitudinal velocity was mastered in a rapid and reliable manner. In most cases values below 0.5m/s were achieved within 5 minutes. The combined effect of smoke extraction and reduction of the longitudinal velocity allowed confinement of the smoke propagation to 300m even at the highest heat-release rates tested; The conditions for survival (visibility, temperature, radiation level and concentration of toxic substances) were slightly deteriorated in the initial phase, but rapidly improved and became excellent over most of the time, even at short distance from the fire; Smoke-penetration in the shelters was entirely prevented, even with open doors. This was the most difficult test for the “new” Mont Blanc tunnel.

The new structure, after three years of refurbishment and renovation, demonstrated to be one of the safest tunnels currently in operation.

Related Files
Figure 2: Approximate observed smoke-front location
Figure1: The “new” Mont Blanc Tunnel