Freight Performance Measure Approaches for Bottlenecks, Arterials, and Linking Volumes to Congestion Report

Chapter 1. Introduction

1.1 Purpose of This Guide

The U.S. freight system serves the world’s largest economy. The highway portion of the freight system comprises 4 million miles of paved public roads. It also involves not only transportation agencies, but also private trucking firms and shippers. Because of the strong ties to private industry and the health of the general economy, freight performance becomes a unique functional performance area. Understanding freight performance and matching solutions to performance problems is critical to improving the movement of goods in the Nation.

The Federal Highway Administration (FHWA) has developed this Guide with two purposes in mind: 1) to provide best practices and approaches on several key areas of freight performance measurement; and 2) develop practical guidance in analyzing truck freight bottlenecks to State departments of transportation (DOT) and metropolitan planning organizations (MPO).

This Guide will to assist analysts with identifying and analyzing truck performance in general and truck freight bottlenecks specifically. This Guide specifies step-by-step procedures for analysts to follow, including data assembly and manipulation. Where applicable, limitations of the data and procedures are identified.

A Technical Working Group (TWG) formed by FHWA and comprised of State and local agency personnel with a stake in analyzing freight bottlenecks, reviewed the development of this Guide. They were instrumental in ensuring that the methodology developed is comprehensive and useful to practitioners.

1.2 Guide Organization

The structure of this Guide is as follows:

• Chapter 2—Summarizes of previous work on truck freight performance and bottleneck analysis is presented.
• Chapter 3—Covers the issue of matching traffic volumes to congestion data.
• Chapter 4—Presents data and methods for measuring congestion on signalized arterials.
• Chapter 5—Incorporates material from the previous chapters—along with new material—into a comprehensive methodology for analyzing truck freight bottlenecks.

1.3 Summary of Past Work

1.3.1 Background

A series of case studies provided documentation of the existing methodologies and practices related to truck freight bottleneck analysis, including the measurement of truck travel times on signalized arterials and the development of truck volumes to use in conjunction with travel-time data. The documentation used a matrix approach with a common template with a number of criteria:

• Reasons for conducting the analysis.
• Stakeholders/users.
• Data and models used.
• Methodology.
• Identification of bottlenecks.
• Raw data processing.
• Performance and cost measures.
• Recommended treatments.
• Scalability.
• Follow-up activities.
• Areas for methodological improvements.

There were 15 case studies that were undertaken:

1. Texas Freight Mobility Plan.
2. North Carolina I-95 Economic Impact Study.
3. Houston/Galveston Intermodal Connector Analysis.
4. American Transportation Research Institute (ATRI)/Federal Highway Administration (FHWA) Congestion and Bottleneck Analysis.
5. Georgia Department of Transportation (GDOT Freight) and Logistics Plan (Bottleneck Analysis Component).
6. Atlanta Regional Commission (ARC) Strategic Truck Route Master Plan.
7. Southern California Association of Governments (SCAG) Regional Goods Movement Project.
8. Texas Top 100 Congested Roadways.
9. 2012 Urban Mobility Report.
10. 2011 Congested Corridors Report.
11. Northwest Arkansas Congestion Analysis.
12. Twin Cities Arterial Mobility Performance.
13. Freight Chokepoint Analysis Tool (FCAT).
14. Assessment of Multi-modal Freight Bottlenecks for the Upper Midwest Region.
15. Bottleneck Performance in the I-95 Corridor.

1.3.2 Case Study Findings

The results of the case study analyses are presented in appendix A using a common template. A summary of the findings follows:

• In general, the degree of sophistication in the methods, and the ability to provide high-resolution and accurate results, depends on the quality and nature of the data that are used, regardless of the analytic method. Older studies relied on “planning-level” data—typically traffic volumes, truck percentages, and information about physical capacity. The primary performance metric of interest—travel-time or a variant of it—was developed by the use of models. Recent studies have used primarily vehicle probe travel-time data from private vendors. These data offer a great step forward in that they are actual measurements rather than modeled estimates.
• Global positioning system readings produce data from on-board or personal devices, but are heavily processed by the vendors so that any origin and destination (O/D) data are lost. This is true of the vendors that provide the data as speeds assigned to a link, usually defined by the Traffic Message Channel (TMC) standard. It also is more advantageous to have specific freight truck data rather than general traffic data, as several studies have confirmed both route and speed differences between cars (including taxis and limousines) and freight trucks. The American Transportation Research Institute (ATRI) has access to unprocessed truck data and thus has been able to develop O/D information from their data.
• Methodologies for producing performance measures from data are very similar, but would benefit from consistent/standardized processing procedures. Developing high-level performance measures from low-level data requires multiple processing steps, and there are usually multiple ways to perform each step. Default values also are often necessary, and these can vary depending on the methodology. The result is that different values can result from processing the same basic data.
• Past studies have taken the facility-based view of truck impacts but truckers and shippers are usually more concerned with the performance of the entire trip. The reason that facility performance is so prevalent is that this is the level at which the data are available; true O/D data that would define entire trips are very rare. One study explored synthesizing trip travel-time data from facility-specific data as an option. In general, incorporating the user’s perspective into freight performance is a key step to advance the state of the practice. However, both perspectives provide for a comprehensive freight performance management program, especially since the majority of improvements that can be effected by transportation agencies are facility focused.
• The current methodologies do not “drill down” to identify the causes of congestion. Model-based methods have considered only recurring (physical bottleneck-related) congestion. Methods based on continuously collected travel-time data capture all potential sources of congestion, but the methods to date have not decomposed congestion into its sources.
• Bottlenecks related to the physical geometry of the roadway are the most prevalent type in these studies. While these types of bottlenecks tend to be the most severe, other types such as those created by operating policies and restrictions have not been covered. Again, we suspect that this is an artifact of the facility-based data that are used in the studies.
• Only a few studies have translated truck congestion into general economic impacts. Data on the commodities carried by trucks traversing bottlenecks is the main limitation.
• Freight bottleneck methodologies start with simple methods to “scan” for potential locations. The scans can be based on:
  • Vehicle Probe Data—Looking for links with a significant amount of slow vehicle speeds, especially if the links are in sequence (indicates queuing).
  • Inventory Data (e.g., Highway Performance Monitoring System (HPMS))—Highway sections with high volume-to-capacity ratios.
  • Anecdotal Information—Often the best information is to ask State and local engineers and planners who are usually aware of problem locations.
• Since most methodologies reviewed consider geometric (recurring) bottlenecks, physical characteristics such as interchanges, bridge crossings, and lane drops are matched against the data scans. This step is necessary to identify other causes of “links with slow speeds” such as long-term work zones or bad weather, and to give the bottleneck an identity.
• Performance measures used in the studies are the usual suite of travel-time-based measures, made truck-specific. Delay is a common metric because it can be valued.
• Arterial truck performance is very important for access to port/transfer facilities, but this has not been widely studied, perhaps because more serious truck bottlenecks occur on freeways. Data limitations are another possible reason. A few studies have used vehicle probe data in largely the same way as for studying freeway bottlenecks. While vehicle probe data should provide the necessary content, there are technical concerns (e.g., stopped delay at signals appears to be a problem).
• Even the more recent studies, which use vehicle probe data to develop detailed travel-time measures, rely on planning-level estimates of truck volumes. Short count-produced truck volumes may not be reflective of actual truck volumes on an individual facility. Further, matching the network georeferencing used for vehicle probe data with the georeferencing used by transportation agencies is a huge technical obstacle.

1.3.3 Suggestions for Final Methodology Based on Technical Working Group

1.3.3.1 Discussion

• There was much support by the TWG favoring the gathering of anecdotal data at several levels, including the initial identification of potential bottleneck locations and truck stakeholders’ perceptions of the significance/severity of impact. It was suggested that the methodology provide guidance on how agencies develop anecdotal information (who to contact, what questions to ask, how to use the information, etc.). However, anecdotal information is not a substitute for data in performance analysis. For example, it is not possible to determine the percent of trucks on roadway by interviewing several shippers or analyzing a small number of supply chains.
• The issue of impacted users at freight bottlenecks was discussed. From the perspective of a supply chain, delay at a bottleneck may only account for a small portion of the total trip, and thus would not be viewed as a significant problem. In addition, if the bottleneck is related to peak weekday periods or times of inclement weather, scheduling would reduce the impact to the supply chain. The methodology should at least acknowledge this fact and offer guidance on how to decompose the different types of users affected by freight bottlenecks. The Research Plan recommended this as one of the topics for further discussion.
• The TWG suggested that part of the methodology should be identifying the causes of congestion at bottlenecks. As mentioned in the presentation, it is important for analysts to define the nature of the bottleneck, especially when reviewing system-wide scans based on travel-time data. However, no original research on this topic was conducted, as there is other recent material that can be appropriated (e.g., Strategic Highway Research Program 2 (SHRP 2) Projects L02 and L03). Additionally, National Cooperative Highway Research Program (NCHRP) Project 8-98 (Guide for Identifying, Classifying, Evaluating, and Mitigating Truck Freight Bottlenecks), which is being conducted simultaneously, will be exploring the congestion-by-source issue in depth.
• The methodology should be scalable to both rural and urban settings. A major issue in rural issues is the nature of the current generation of vehicle probe data: the roadway sections that vendors use for reporting travel times can be excessively long in rural areas (more than 20 miles) and long sections can mask bottlenecks. This underscores the need to have good anecdotal data on bottleneck locations, but some form of supplemental travel-time data may be required in these cases.
• An ongoing freight performance measurement program should house the methodology. One way to approach this is to consider freight bottleneck identification and analysis as a second or “drill-down” step in a system-wide freight performance measurement system. System-wide monitoring produces high-level statistics on how the system is performing. The freight bottleneck methodology is a tier lower than this as it searches for specific locations that are congestion trouble spots. How to design and maintain a freight performance measurement system is beyond the scope of this work, but recommendations on now it can be integrated with the freight bottleneck methodology will be made.
• As part of the bigger picture, an agency-wide performance management program should incorporate the freight bottleneck methodology. Here, the methodology would provide input to procedures such as agency target setting and tradeoff analyses with other functional areas such as safety, pavement, and bridges. How to conduct tradeoff analyses with other project and program types also is beyond the scope of this project, but we will ensure that the freight bottleneck methodology’s usefulness is discussed.
• Recent work on bottleneck analysis has emphasized the use of vehicle probe data from private vendors as the primary data source. The reason for this that it has extremely wide coverage and is cost-effective. However, as of this writing, these data may not provide accurate performance information on signalized arterials. To a very large extent, the methodology is neutral as to how the data were collected; all that is required are travel times and volumes. Regardless of the technology, data from any source gets transformed into travel times and volumes before the analysis is started. So while the preprocessing may be different, the actual analysis stays the same.

previous | next