Most conventional route guidance systems select the optimal route on the basis of the minimum expected route travel time. However, as a result of the inherent variation in driving conditions, the route travel time that drivers will experience may be longer or shorter than the expected mean travel time. Drivers are likely to have a lower tolerance for experiencing travel times that exceed their expectation, and as a result, drivers may want to select routes that provide them with smaller travel time variation. While this is clearly desirable, it is not clear whether the paths providing the minimum average travel time are also the optimal paths for other levels of travel time reliability. This paper examines the impact that routing strategy has on route selection. Specifically, the proportion of paths that change when drivers alter their route selection criteria from selecting the minimum travel path computed on the basis of the average travel time to that computed on the basis of say the 95th percentile travel time, is computed. The analysis, conducted using simulation results from a network of a portion of Seattle, Washington, indicates that if a routing system selects routes on the basis of the 95th percentile travel time, then approximately 15% of all routes would be different than the minimum mean travel time routes. The results indicate that there may exist opportunities for improving the reliability of trip travel times by changing the route selection strategy to directly consider travel time variability. (A*)
Samenvatting