Examining Travel Time Variability Using AVI Data

Abstract

Knowledge of travel time variability is valuable for improving the reliability of traffic information services and increasing the accuracy of travel time predictions. To identify the source of travel time variability, information on the travel time distribution properties is needed. Most investigations of travel time distributions rely on data from probe vehicles and consequently have limited sample size. Based on extensive Automatic vehicle identification (AVI) data collected from the CityLink Tollway in Melbourne, a comprehensive investigation of travel time distributions was conducted in terms of various time windows. Given the number of factors affecting travel time variability and their interaction effects, multiple regression with two-way interaction terms was used to quantify the contribution of the various sources to the variability in travel time. The application of the methodology to two groups of data, namely travel times in morning peak and afternoon peak, demonstrates that they have distinctive sources of variability. Morning peak travel times vary mostly because of demand related factors, while 25% variability of travel times in afternoon peak is related to capacity related factors.