Detailed Analysis of Travel Time Reliability Performance Measures from Empirical Data

Abstract

Research on measuring travel time reliability has increased heavily in recent years. A host of measures have been proposed and researched as well as adopted by management agencies. For this study, 13 reliability measures are calculated for 983 freeway segments from 15 minute space mean speed data. Statistical tests are performed to determine how consistently the measures rank the segments at different times of day and how well correlated the measures are to the average. Analysis of the change in measures over the day as well as each measure’s relationship with the average is discussed. Temporal sampling by time of day has a large effect on the travel time reliability measures calculated. This means effects like directional demand peaks can affect direct comparisons across segment types. Ideal comparisons of reliability measures should include all 24 hours of the day, and time of day analysis can identify time periods where management strategies can have the most effect on reliability. No single measure was identified as ideal, though the semi-standard deviation performed well on most tests. It also reports values in reference to the free flow travel time, eliminating the false negative issue where the standard deviation may be very low while the average travel time is high. It is recommended that full distributions be compared where appropriate. With increasing exposure to these distributions and careful explanations as to what they represent, decision makers can effectively prioritize traffic management and geometric improvements.