Étude de la dangerosité d'une courbe en section courante sur une autoroute

The primarily objective of this document is to analyze the dangerousness of freeways through an exploratory approach based on accident data. To achieve its purpose, three main stages are proposed: the literature review on analysis and modeling of freeway geometric elements, the formation and characterization of linked databases to accidents on freeways, and finally a phase analysis of the dangerousness of a specific geometric element. The research is distinguished by the size of the sample used in the analysis that corresponds to all freeway accidents in the province of Quebec in a range of five years (beginning of the year 2000 to the end of 2004). Before today's technology, this type of study required a relatively tedious process limiting the previous analysis and results. The improvements in the accident databases and operating systems of the Ministère des Transports du Québec in the recent years greatly facilitate this kind of study. To meet this new need, it is necessary now to update the methodology for processing and analyzing accidents. This is what the report tries to achieve. The literature review is an important part of this report. This review focuses specifically on the analysis and modeling of freeway geometric elements, because it is a major part in achieving the objectives of reducing accidents today. Given the continuous improvement of network and significant investment in safety studies in recent years, it is increasingly difficult to reduce accidents. The most promising current method is to carry out safety audits during all engineering stages of road projects (concept, pre-feasibility, preliminary and detailed design) to make sure that the final design is the safest. These audits however can't be done without support of tools to assess the dangerousness of road's elements. To highlight the different models of literature explaining the accidents and better predict behaviour in accidents on the road thus represents a step that can not be neglected. Models have been found and analyzed in detail for a comprehensive breakdown of geometric elements and sets of elements, except for curves and straight segments on mainline. These two last items may seem trivial, but most of the total accidents on freeways occur on them. They were therefore selected for further deepening. Previously, it is essential to understand the general trends of data. Thus, a reconstruction of the three major databases is performed: geographic data, traffic data and accident data. These various databases are then merged to characterize the fundamental dimensions of accidents: exposure time and space. In this research, it was decided to limit the analysis to these two dimensions.