Trip Travel-Time Reliability: Issues and Proposed Solutions

Abstract

The estimation of path or trip travel-time reliability is critical to any advanced traveler information system. The state-of-practice procedures for estimating path travel-time reliability assume that travel times follow a normal distribution and that segment travel times are independent (i.e., trip variance is a summation of segment variances). The present study analyzes Automatic Vehicle Identification (AVI) data from San Antonio, Texas, and simulated data to demonstrate through goodness-of-fit tests that a log-normal travel-time distribution is valid only under steady-state conditions, whereas a normal distribution is not valid. In the present article, the authors propose five methods for the estimation of path travel-time variance from its component segment travel-time variances. The analysis demonstrates that computing the trip travel-time coefficient of variation as the conditional expectation over all realizations of roadway segments provides estimates within 70% of trip travel-time variance for both uncongested and congested conditions.