Approximately ten to fifteen years ago many countries introduced tunnel safety management systems and started paying attention to tunnel safety in a more structured way. PIARC report 2007R07 recommended an integrated approach to tunnel safety. As a result of this, many countries have gained experience with the application of various tools for tunnel safety management. Experience with tunnel incidents and methods for incident evaluation and risk analysis have led to developments in organisation and management and to improvements in the systems in use. In this report several contributing countries share information on lessons learned from incidents and developments in safety management and risk analysis and conclusions are drawn on topics of general interest.

As documented in chapter 2 of this report, the collection of data on significant incidents in road tunnels is carried out in a systematic manner in many countries. In this chapter the process of data collection is described. It has been noted that in practise, collecting all necessary data for a good evaluation leading to improving safety procedures or to produce incident statistics that can be used in risk analysis, is not easy and can be very time consuming. There can be a conflict in available and required resources for data collection. It is therefore important that the required data is clearly identified and that it is determined which data shall be collected by each party involved and at what time (immediately after incident or at a later stage). To keep parties involved motivated to contribute, the purpose of data collection has to be made clear to them and feedback on lessons learned and improved procedures and systems should be provided.

Results of data collection can be found in the other chapters of this report.

In chapter 3 collision rates for several countries are presented. In general, collision rates represent average values for a certain set of tunnels investigated. Figures presented range from 2 to 12 collisions per 10⁸ vehicle-kilometres. It has been determined in several studies that collision rates are influenced by many parameters. In the report a list of the most influencing parameters is given., e.g. tunnel length, traffic volume, horizontal alignment, lane width, tunnel cross section, quality of tunnel lighting, traffic composition, driving speed, and - last but not least - national driving habits and the technical standard of the vehicles in use. All these influencing factors make it difficult to compare collision rates on a statistical level. Despite some recent studies on the topic, there is still a lack of knowledge on the interaction of the various parameters influencing collision rates in road tunnels. The report provides a comprehensive list of all relevant data concerning collisions, which should be collected in order to improve the quality of information.

In chapter 4, new information on fires rates has been compiled, based on tunnel fire statistics from 12 countries around the world. It seems that an "average tunnel" has a fire rate in the order of magnitude 5 - 15 fires per 10⁹ vehicle km. However, the scatter of the rates from tunnel to tunnel may be very significant as a number of factors may influence the recorded fire rates, for instance: tunnel design, location of the tunnel, geometry of the road, monitoring, technical standard of the vehicles, traffic regulation, speed limits, driving culture etc.. The fire rates should be used with care, and the assessment of the applicability and the modification of the basic rates required for an application for a given tunnel should be done by experts with experience in tunnel safety. When the above conditions are fulfilled, the fire rates can be applied in analysis to inform the development of requirements for safety systems for tunnels which take appropriate account of the fire risk.

Whereas chapter 3 and 4 are addressing incident data evaluation for tunnels in general, chapter 5 focusses on real incidents at the level of an individual tunnel. It provides a realistic idea of incidents happening in road tunnels as well as examples of conclusions and improvements that can be identified for specific incidents and tunnel systems. These findings can also be used in risk assessments for other (new) tunnels and as illustrative material in training and education.

Practical experience with risk assessments is described in chapter 6. Several countries shared their experiences with risk assessment in the past decade. From these contributions it can be noted that there are some differences between countries in methodology they use and even within one country often several methods are used, depending on the purpose of the investigation. Typical applications for risk assessments are:

• to demonstrate that the tunnel is safe enough
• as a decision making tool to compare various alternatives or to decide on risk
• to determine performance and/or reliability requirements of safety systems
• to provide insight into the residual risk
• for the classification of the tunnel with respect to transport of dangerous goods

In all countries risk analysis is applied as an additional tool complementary to prescriptive guidelines and regulations. Experiences in the past decade and input from real incident data and statistics have been used to improve risk analysis methods. Vice versa, risk analysis methods are used to improve prescriptive guidelines. Furthermore, the experience with risk analysis methods in the tunnel design and decision making process have also led to improvements of the risk analysis methods themselves and have instigated additional research with regard to collection and evaluation of data on real incidents and fires.

Based on the information in this report we can conclude that use of an integrated approach to tunnel safety is becoming practised more and more in several countries. Experience with this approach and lessons from real incident data are used for further improvements in tunnel safety. This report can be used as a reference to further improve the integrated approach to tunnel safety, especially the requirements and key aspects to data collection and risk assessment in this approach.