Identifying the Locations and Causes of Unreliable Truck Travel Times

Abstract: Truck travel time reliability (TTTR) is one of the most important performance measures for assessing freight movement on the interstates. State Departments of Transportation are required by the FHWA to establish TTTR targets for all interstates, and to report the performance and progress toward the achievement of the targets. Transportation agencies are increasingly use TTTR in various planning and performance measurement processes. However, a recent review of current practices found that most agency studies were focused on individual hotspots, and there is a lack of systematic, data-driven approaches to identify the locations and causes of unreliable truck travel times at network level.

This study filled the research gap by developing a systematic method to identify the location, time, and causes of unreliable truck travel times. A variety of data sources were used, including truck travel times from the National Performance Management Research Data Set (NPMRDS) and datasets for traffic and operation factors, roadway geometric factors, incidents, and weather. Using data collected for all interstate highways in Virginia, this study conducted statistical and machine learning analyses of TTTR metrics to develop criteria for defining and measuring unreliable truck travel times. For each studied road segment, the temporal and spatial variations in TTTR metrics were analyzed to reveal the changes of reliability metrics with factors such as the time of day, roadway geometry, traffic volume, incident, weather, and work zones. The impacts of relevant factors on TTTR were evaluated, and potential correlations among TTTR, truck safety index and truck bottlenecks were analyzed.

The systematic and data-driven approach developed in this study represents a significant step toward a more comprehensive understanding of truck travel time reliability. The findings from the analyses can help determine whether the TTTR of certain road segments or corridors are getting worse (or better) statistically over the years and help identify potential improvement locations. Other important considerations in the analyses include the determination of whether segments are (un)reliable during all analysis periods or only a subset of these, and the methodologies to handle potential outliers and incorporate truck-involved crash data in addition to general crash data. The findings could help transportation agencies to better identify the locations and causes of unreliable truck travel times, which could lead to more targeted treatment and better prioritization of traffic improvement projects.