COMBINING SENSOR TRAFFIC AND SIMULATION DATA TO MEASURE URBAN ROAD NETWORK RELIABILITY

Travel time is likely to be dissimilar for same spatial trajectories. This underlines the need to think about travel time interms of frequency, and not just of magnitude. A measure of the dispersion of the travel time distribution is Travel Time Reliability(TTR). Its valuation has been receiving considerable attention in recent years, in freight and especially in passenger transport.This study addresses this need by proposing a methodology for estimating travel time reliability of an extended traffic network atdifferent aggregation level, by using historical radar detector data and a real-time traffic simulation model, incorporating the analysisand the comparison of various TTR measures. The basic variables used are travel times and traffic flows of each link which belongsto the traffic network. Travel times are estimated through an assignment traffic simulation model and the application of a periodicupdate with movable horizon called rolling horizon, in order to reproduce the dynamic interaction between the level of congestionand users route choices of a congested road network, that is significant non-linear. The statistical measures used to quantify traveltime reliability are Standard Deviation, Travel Time Window, Buffer Index, Planning Time and Buffer Time. The describedmethodology is applied to the urban area of Catania (Italy). Model is calibrated and tested for different network’s sizes at variousaggregation levels. The findings of this study should be of interest to transportation planners, traffic engineers, and transit agencies inthe form of recommendations for implementing travel time reliability in an extended transport network.