Abstract

Several metrics of reliability have been recommended for use. These metrics may need to be calculated at relatively fine levels of aggregation of travel time data when used for advanced management strategies and transportation system modeling and analysis. This study investigates the variation of travel time distributions by time of day at fine temporal aggregation levels, the sensitivity of various reliability metrics to these variations, the effect of the aggregation level choices on the calculated metrics, and the amount of data required to estimate stable values of the reliability metrics. The investigation is made for a facility that has general purpose lanes and high-occupancy-toll lanes. The results show that the parameters of travel time distributions vary during the peak period, reflecting the effects of traffic congestion, traffic flow dynamics, and how much nonrecurrent factors, such as incidents, contribute to the unreliability of travel time on the investigated facility. Trends of variations are observed when reliability metrics are used to assess reliability as the congestion level changes during the peak period. The results also show that examining combinations of time-variant static distribution parameters and reliability metrics of the general purpose and high-occupancy-toll lanes can provide valuable information that cannot be obtained when the analysis is performed at higher aggregation levels. The study results suggest using at least one year of data to obtain stable values of reliability metrics.