An empirical model is presented which relates incidents with congestion and travel time prediction. Incident and travel time data from the Attica Tollway (AT) in Athens, Greece have been studied according to the framework and the following conclusions are drawn: 1) There is significant time-of-day variability in the effect of incidents on delay; 2) There is significant route-to-route variability in the effect of incidents on delay. Delay characteristics, especially maximum delay observed for the route, is strongly correlated (with correlation coefficient of 0.68) with the incident-delay correlation; 3) Breakdowns and accidents have largest impact on delay and thus need to be managed with high priority, while other incidents have small impact; and 4) The real-time travel time prediction, such as what is displayed on a CMS, could be improved by bias-reduction using the knowledge of an accident that has just happened. These results imply that it is critical for incident management plan and real-time travel time prediction to be based on empirical study using site-specific data on incidents and congestion. The proposed empirical modeling framework can facilitate development of such incident management plan and improve real-time travel time prediction reflecting incident information.
Samenvatting