9 November 2023
Route bidirectionnelle en zone peu dense
Route bidirectionnelle - Pixabay
Cerema has carried out a study into accidentology on two-way roads, which account for a large number of road accidents. The Sécubidi project was carried out as part of a call for projects issued by the French Road Safety Delegation to improve knowledge of accidents on these roads outside built-up areas.

Two-way interurban roads make up a network of over 400,000 km managed by both the state and local authorities. This network performs multiple functions and has diverse characteristics in terms of infrastructure provision. It carries a significant proportion of journeys made for a variety of reasons, with a mix of different users such as passenger vehicles, heavy goods vehicles, vulnerable road users and agricultural machinery.

In terms of accidents, this network accounts for 89% of deaths on roads outside built-up areas (1,915 people killed per year), or 56% of all road deaths. What's more, although scattered along the routes, it appears that one quarter of this network accounts for two-thirds of the fatalities. While the BAAC file (compendium of road accidents in France) can provide some informations on the accident rate, there is a lack of more in-depth knowledge that would enable a diagnosis to be made that is better suited to the heterogeneity of this network.

As part of a call for projects issued by the Road Safety Delegation to improve knowledge of accidentology and usage on two-way roads outside built-up areas, Cerema proposed the Sécubidi project.

One of the main objectives of the project was to provide detailed knowledge of accidentology on two-way roads outside built-up areas, with particular emphasis on infrastructure analysis, in order to :

  • Better understand in order to act, in particular to estimate and prioritise safety issues,
  • Identify the main accident mechanisms in order to make a diagnosis and estimate the role of infrastructure and users,
  • Assess the possibility of proposing a specific analysis for each type of network.


Methodology and main results

Analyses were carried out on 2 types of incoming data:

  • The BAAC 2013/2017 files for the "stakes study" deliverable, making it possible to better estimate and prioritise safety issues,
  • The 2015 fatal accident database from the FLAM project, in order to identify the main accident mechanisms and factors, and taking into account the heterogeneity of the network
  • studied.

In mainland France, between 2013 and 2017, 3,400 people were killed each year, 2,400 of them outside built-up areas. Of these 2,400 people killed outside built-up areas, 1,900 (80%) were killed on two-way roads. Two-way roads are very serious: 80% of fatalities outside built-up areas, for 58% of bodily injury accidents outside built-up areas.

Moreover, the accident rate has not declined over the period observed: in 5 years, the number of bodily injury accidents on two-way roads outside built-up areas has risen by 25% (5 points more than on non-two-way roads), while the total number of bodily injury accidents in France has risen slightly by 3%.

The main issue is accidents outside intersections, and without pedestrians. The number of fatalities (1,900) per year on two-way roads outside built-up areas breaks down as follows:

  • 1,600 killed in non-intersection accidents with no pedestrians (84%),
  • 230 killed in accidents at intersections without pedestrians (12%),
  • 90 fatalities in accidents with pedestrians (4%).


Study of lane departures

In terms of primary safety, an analysis of accident mechanisms showed that 55% of drivers involved in accidents on two-way roads outside built-up areas had left the carriageway. These lane departures took place on bends in 54% of cases. The majority of lane departures were to the left (71% compared with 29% to the right), whether the driver was initially on a straight section (73% of lane departures were to the left) or on a bend (70% of lane departures were to the left). 

These lane departures break down as follows: 

  • 37% of drivers with a lack of guidance, generally progressive (falling asleep, dozing off, various inattentiveness), etc...,
  • 63% of drivers with loss of control of the vehicle (loss of dynamic control due to inadequate speed in relation to adhesion constraints, loss of control following avoidance or collision with an animal).

With regard to lane departures linked to a lack of route guidance (excluding those involving discomfort), it appears that this problem is present on all networks, although the issue is more acute on the national network.

Furthermore, it appears that the majority of drivers: 

  • Were driving in a straight line (55% of cases, and 180 drivers concerned),
  • Twice as often on right-hand bends as on left-hand bends,
  • made an initial shift to the left (88% of them, and 255 drivers concerned).

The 255 accidents involving first swerving to the left represent more than 15% of all fatal accidents on two-way roads outside built-up areas.


Sur le plan de la sécurité secondaire, les accidents (hors inter) sans tiers avec heurt d’obstacle aggravant représentent 29% de l’ensemble des accidents du réseau bidirectionnel hors agglomération, soit 497 cas. Dans 1 cas sur 2, l’obstacle heurté est un arbre. La distance des obstacles par rapport au bord de la chaussée a été estimée, et il ressort que 60% étaient situés à moins de 2 m, et 83% étaient situés à moins de 4 m. Par ailleurs, il ressort que les accidents contre obstacles (hors inter, sans tiers) ont majoritairement lieu en courbe (53% des cas). 

Les facteurs humains sont très fortement majoritaires dans les accidents mortels sur réseau bidirectionnels hors agglomération. La pratique d’une vitesse inadaptée ou excessive (38%) ainsi que l’alcoolémie (31%) représentent les principaux facteurs. 


The main factors linked to Infrastructure or traffic conditions are : 


  • Poor visibility, mainly caused by fixed masks (7%) linked to the environment (4%) and the profile or layout of the road (2%),
  • Poor legibility of the infrastructure, which prevents users from adapting their behaviour: poor legibility of a curve (4%) or an intersection (2%),
  • Inadequacy of the infrastructure to dynamic constraints: grip problems on wet roads (7%), poor road condition in 2% of accidents,
  • Recovery or avoidance possibilities limited by insufficient shoulder width (8%) or by the presence of an obstacle on the shoulder (3%),
  • Collision with an aggravating fixed obstacle on the shoulder was present in 35% of accidents.

Contribution of the project to the proposals for action

With regard to infrastructure, the project highlighted several points: 

  • Installing audible warning devices, mainly at the centre of the carriageway, to limit lane departures due to guidance problems (apart from discomfort). At least 328 fatal accidents occurred in this configuration in 2015, i.e. 19% of all accidents studied. The phenomenon is much more prevalent on the centre of the carriageway (255 accidents: 15%) than on the edge (39 accidents: 2%).
  • Initiate and continue measures to deal with side obstacles. Priority can be given to obstacles located within 2 metres of the edge of the carriageway, a distance that covers 60% of cases of collisions with aggravating obstacles in accidents (excluding inter-vehicle accidents) without third parties, i.e. at least 270 cases of collisions out of the 1,685 accidents studied. In relation to the issue of accidents occurring on bends, obstacles on bends must be considered very carefully, since 54% of accidents involving obstacles take place on bends.
  • Carry out grip measures to limit loss-of-control accidents in wet weather, an accident factor encountered 121 times out of the 1,685 accidents studied. Routes with curves could be treated as a priority. If the level of skid resistance proves to be poor, it may be possible to propose lowering the speed limit while waiting for the surface course to be repaired.
  • Limit the presence of obstructions to visibility, particularly fixed obstructions, which were a factor in 115 of the 1,685 accidents studied.
  • Pay particular attention to the legibility of the road, especially curves, which were a factor in 60 of the 1,685 accidents studied. Tight bends after a long straight stretch represent a challenge that seems to be more pronounced for motorbikes. To this end, the ALERTINFRA method and the principles of signposting bends should be applied more systematically in practice.
  • Check that sections with authorised overtaking offer good visibility conditions. Overtaking accidents account for 13% of all accidents on the main network and 11% on the secondary network. It would be worthwhile limiting the number of overtaking opportunities upstream of junctions with a high rate of left-turns, which presupposes an improvement in the current level of traffic knowledge.
  • Step up the policy of creating cycle routes or facilities on the secondary network, which accounts for 61% of accidents involving cyclists, i.e. 43 cases.

As far as users are concerned, the project has highlighted the importance of: 

  • Check speeds, blood alcohol levels (particularly at night and off the main network) and drug use,
  • Carrying out information/communication campaigns on the risks associated with fatigue and inattention, which are more prevalent on the main road network.
  • Reinforce communication on the need to share the road in order to take into account the presence of vulnerable road users, particularly at junctions, taking into account the increase in cycling.


Detailed results of the Sécubidi project are available on CeremaDoc: