Estimating freeway route travel time reliability from data on component links and associated cost implications
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
2017, 2017 The Institute of Urban Sciences. Typically, data on travel time (speed) and traffic volume are collected for relatively short homogenous links of a roadway (e.g. every to 2 miles or 0.83.2 km) and not for particular routes from origins to destinations. To accurately estimate a travellers trip, there is a need to build information on an entire route from data collected at the link level. Because it can be resource-intensive to define specific routes for the entire network of interest, route reliability performance measures are sometimes estimated using a vehicle-miles of travel (VMT)-weighted average of performance indices from links. In performance measurement, travel time index (TTI) is the ratio of average travel time and free-flow travel time while planning time index (PTI) is the ratio of 95th percentile travel time and free-flow travel time. TTI is considered a mobility measure while PTI reflects system reliability. Using indices based on free-flow (uncongested) travel time, this study finds that the VMT-weighting of link indices to get the corresponding route index gives reasonable results for the travel time index (TTI). However, this approach overestimates the planning time index (PTI). This highlights the statistical insufficiency of estimating route PTI values from link PTI values weighted by VMT. It points to the subsequent need to incorporate factors for dispersion and skewness characteristics of travel time distributions on constituent links. This paper introduces a promising method to estimate freeway route reliability considering these dispersion and skewness properties. For statewide Virginia Interstates considered here, economic analyses based on the value of travel time demonstrate that incorrect travel time reliability estimates result in upwards of 200% higher user costs. Researchers anticipate that the information documented in this paper will be useful for those conducting performance monitoring activities to estimate route reliability with limited resources when only link data are available.
