Skip to main content
leaderboard1

Early warning leads to swift response in French road tunnel

With the help of Citilog’s early warning, authorities were able to close the Vuache Tunnel in France before a fire fully broke out, reports David Arminas
October 6, 2025 Read time: 6 mins
The risk of fire in a road tunnel is greater than many would think, says Piarc (© Arbel Rom | Dreamstime.com)

Tragedy was averted when in June a fire broke out in the Vuache Tunnel in the Haute-Savoie region of the French Alps. Automatic incident detection (AID) systems from Citilog played their part in quickly alerting officials about the incident which ensured a swift and effective response from emergency services.

The twin-bore 1.4km-long tunnel was opened in 1982. It was in the southbound tunnel that a heavy goods vehicle laden with plastic pellets caught fire. Because of Citilog’s early warning, officials were able to close the tunnel - one of two bores - before the fire fully broke out. This prevented other vehicles from entering and avoiding what could have been a far more serious situation (see box, response timeline).

ATMB (Autoroutes et Tunnel du Mont Blanc) is the company responsible for managing the French side of the Mont Blanc Tunnel, which runs between France and Italy, as well as the surrounding alpine highways. ATMB plays a key role in ensuring safe and efficient cross-border transportation. This recent positive outcome is the result of a remarkably fast and coordinated response by ATMB’s operations teams and emergency services - and, crucially, of the automatic incident detection system that triggered the first alert.

Citilog is a major supplier of video based automatic incident detection solutions for tunnels, bridges and highways, with more than 60,000 cameras installed world-wide.

 

Nothing missed

Citilog says that its solution turns cameras into AID sensors, creating a broad detection network where nothing is missed. No matter where an incident occurs, it will be detected instantly to reduce its impact on road users - and sometimes help save lives and prevent secondary accidents. Its broad incident detection automatically identifies wrong-way vehicles, pedestrians and debris, smoke and occlusion as well as lane changes, congestion and slow or stopped vehicles.

The tunnel runs through the Haute-Savoie region of the French Alps (image: ATMB)

Citilog’s AID is deployable on a server or directly within the cameras, ensuring it meets all customers’ needs. It is compatible with central servers for future scalability and is completely camera-agnostic, allowing one server to handle up to 100 streams. It is deployable on Axis cameras as a cost-effective alternative for distributed architectures.

The system’s video analytics algorithms are trained to identify vehicles and remain unaffected by shadows, reflected light, or moving vegetation. Citilog says that its combination of pixel analytics and deep learning algorithms enables a detection range of 300m - not achievable through deep learning alone.

Traffic through the dual-bore Vuache Tunnel, part of the A40, resumed the next day when officials reopened the other two-lane tunnel using one lane in each direction for vehicles. Meanwhile, as a safety precaution, ATMB reported that the transport of dangerous materials through the tunnel was forbidden until repairs to the damaged tunnel were completed and the restriction officially lifted.

 

Incidents will happen

Incidents will always occur. No system is going to be perfect, explains Grégoire Latourte, regional sales manager for Citilog. It is all about risk management. When an incident such as the recent one in the Vuache Tunnel happens, it reinforces the ongoing efforts of tunnel operators and owners to secure their infrastructure and invest in systems like AID.

In fact, the risks of a fire in a road tunnel are greater than many people would think, according to the World Road Association (Piarc). Since 1957, Piarc has had a road tunnel committee to address a range of issues related to the use of road tunnels, such as geometry, equipment and its maintenance, operation, safety and the environment.

 

 

Piarc says that vehicle fires give rise to particular concern simply because they are, actually, not very rare in road tunnels. The discussion on tunnel fires is often dominated by the extreme events which have occurred, such as in the 11.7km-long Mont Blanc Tunnel between France and Italy and in Austria’s 6.5km-long Tauern Tunnel. However, in reality, the majority of tunnel fires are relatively small events in comparison. Nevertheless, says Piarc, they may have the potential to develop into more serious events, depending on various influencing factors. This is because the vehicle fires are contained for the most part within a vehicle’s physical structure and are slow to grow outside the vehicle. Basically, vehicles catch fire more often than the public thinks; it is just that they catch fire on an open road.

 

Fire can escalate rapidly

Piarc affirms that the confined space of a tunnel provides an environment in which untenable conditions may develop rapidly in the case of a fire. Numerous real fire tests have been performed in the context of various national and international research programmes to confirm assumptions on fire sizes and fire behaviour. In these tests, the focus again has been on large-scale fires with high heat release rates.

Given all this, road tunnel operators and owners have been well aware for many, many years about the importance of an AID system and its ability to rapidly detect events, allowing emergency teams to react quickly and avoid the most serious of repercussions - loss of life.

“However, they may not all be aware of the leap in performance and reliability that has been achieved over the past 10 years, thanks in particular to the use of deep neural networks,” says Latourte. “This minimises the inconvenience caused by false alarms and enriches the information fed back to operators. For example, a system can provide vehicle class information or generate new types of alarm.”

The twin-bore 1.4km-long tunnel was opened in 1982 (image: ATMB)

All this paves the way for the deployment of AID systems on open roads under challenging environmental conditions where performance levels now meet operators’ expectations.

“I can't speak on behalf of ATMB, but what is certain is that they have demonstrated operational excellence in managing this incident and minimising its impact,” says Latourte. “First, in terms of the human toll, since there were no casualties. But also in material terms, since the fire caused damage that was kept to a minimum by the rapid and appropriate response of the emergency teams. From Citilog’s perspective, I can only commend ATMB's scrupulous and regular maintenance for their AID system to ensure optimal working order at all times.”

 

Beyond usual maintenance

The maintenance aspect of maintaining AID systems cannot be over-emphasised, continues Latourte. “Beyond the usual maintenance, ATMB organises detection tests every year, enabling them to validate the system's ability to perfectly detect any type of event. All this pays off when an incident such as a fire breaks out in a tunnel.”

As with all technologies, it will move forward and rapidly. “I don't think the race to higher levels of performance is over,” says Latourte. “We'll continue to work to exploit the potential offered by new developments in the field of AI applied to computer vision. It would not be surprising, either, if the reliability and enrichment of information were to be achieved through the fusion of different complementary technologies or the use of multiple information sources, starting with those made available by the vehicle itself, through V2X [Vehicle to Everything] technology.”

While there is no universal rule about when an AID provider is brought into a new tunnel project, Latourte says that Citilog is regularly consulted by engineering firms during structural design phases. “This is a virtuous circle. I would always recommend it, as it enables us to pass on the benefit of our experience accumulated over almost 30 years of AID projects around the world. It will give an AID system every opportunity to excel once deployed in operational conditions.”

-------------------------------------------------------------------------------------------

Vuache Tunnel fire: response timeline

 

• At 10:33 the incident detection system identifies a stopped vehicle; the ATMB control centre immediately activates tunnel closure procedures

• At 10:38 the fire breaks out on the truck, but the tunnel is already closed; firefighters intervene swiftly, no other vehicles are involved and the driver is safely evacuated

• By 5:30am the next day, safe traffic flow is restored through the unaffected tube

For more information on companies in this article

Related Content

  • Citilog ITS helps avert tragedy in Vuache Tunnel
    June 25, 2025
    Automatic incident detection systems allowed ATMB to close tunnel
  • Tolling changes are making electronic music
    October 3, 2025
    Hal Worrall and Douglas Chastain offer some perspectives on the evolution of electronic toll collection and how ownership shapes priorities in the industry
  • How the City of Augusta mastered traffic with Network Optix
    September 26, 2025
    The annual Masters golf tournament doesn’t only produce stumbling blocks for the players. Adam Hill talks to Marc Faubert of Network Optix and John Ussery of Augusta’s Traffic Engineering division about managing the annual influx of sports fans and their cars
  • Citilog’s early detection prevents tragedy in Alpine tunnel
    August 26, 2025

    Citilog is showcasing the real-world power of its automatic incident detection (AID) technology by highlighting how it operated in a real-world situation when the system helped prevent a potential tragedy in France’s Vuache Tunnel.

    At 10:33 on 11 June 2024, Citilog’s AI-powered system identified a stopped vehicle and immediately triggered tunnel closure procedures. By 10:38, when a fire broke out on the truck, the tunnel was already sealed, preventing other vehicles from entering what could have been a catastrophic situation.

catfish1