Abstract : This paper presents the results of a research project called IRCAD, which aimed at developing a global system to warn drivers in real time, when their behaviour is not adapted to infrastructure characteristics. The warning system is based on the comparison between the speed of a vehicle before a curve and a safety speed. This safety speed is defined as the maximum speed value before crash all along the curve (safety speed profile). These thresholds values of speed are calculated in real time depending both on constant parameters like infrastructure geometry (radius of curvature, cross and longitudinal slopes) and changing parameters like skid resistance, which is evolving with the weather conditions (wind, rain). This study is divided into three parts. In a first step, a water-depth model was developed to predict the water film thickness in the curve taking into account the road geometry and the rainfall intensity. On a second step, the skid resistance decrease due to the water film is evaluated. Then, the maximum speed is calculated by considering these corrected friction values. In a third step, experimentation is realized on two sites located on French secondary roads. To conclude, this system is proved to be very efficient and useful considering the fact that the models use data easy to obtain for road managers and that the warning sign only starts when the situation presents a real risk.
