Over 25% of congestion on freeways is caused by incidents (accidents, vehicle breakdowns and spilled loads). These incidents occur randomly in time and space, and can reduce the capacity of the freeway considerably. Because of the uncertain nature of incidents, they cause many problems to road-users. Over the past years many Incident Management programs have been implemented. Most of these programs aim to reduce the duration of incidents, and consist of measures to improve response and clearance procedures, or the deployment of emergency response units. The number of incidents and their distribution over a road network have to be known in order to make efficient use of Incident Management programs. Decisions to change road geometry or to implement traffic management systems can also be based upon data about the number and distribution of incidents. These measures aim to reduce the number of incidents, especially on road segments where the impact of incidents in terms of congestion is great. This report describes the results of a study of the influence of several roadway- and traffic characteristics on the number of incidents occurring on a road section. The freeways around the city of Utrecht were used as a case-study. Only car-accidents and vehicle breakdowns were studied, as there was not enough data on truck-accidents or spilled loads. The available data included traffic-volumes, the percentage of trucks, road geometry (such as horizontal curvature, the presence of weaving sections, on- and off ramps) and the number of lanes. A log-linear regression model was used to estimate the influence of these factors. The results of the model can be used in two ways: (1) to predict the number of accidents on a section with certain road- and traffic characteristics (future situations), to optimize the application of Incident Management programs, or (2) to indicate under which circumstances the probability that an incident will occur is high, and to try to avoid these circumstances in the future. The results indicate that traffic characteristics, such as traffic volumes and percentage of trucks, are the most important factors influencing the number of incidents. Traffic data are, however, only collected on through lanes, and not on parallel lanes and ramps. Separate estimations for trough lanes and for parallel lanes and ramps have been carried out. The results are clearly better when traffic volumes are included. The presence of weaving sections and the number of lanes also have a significant effect. The location of emergency telephones on the shoulder of the freeway is important in case of vehicle breakdowns: most vehicles can make it to these telephones. The costs of incidents are also taken into account, when decisions about the implementation of Incident Management systems have to be made. These costs were studied. They consist of the costs of emergency response units, damage and injuries, and congestion caused by the incidents. The delay of people and goods can, in turn, upset production processes. The total costs are high and it can be recommended to apply Incident Management in the area around Utrecht or to adapt road geometry or traffic characteristics. (A)
Samenvatting