Estimating Cumulative Traffic Loads, Volume II:
Traffic Data Assessment and Axle Load Projection for The Sites With Acceptable Axle Weight Data, Final Report for Phase 2

CHAPTER 6. SUMMARY AND RECOMMENDATIONS

Summary

A principal objective of the LTPP program is to quantify the relationship between pavement performance and traffic loads. Consequently, traffic data collection and analysis, required to obtain traffic loads, is the key activity within the LTPP program.

In 1998, the FHWA sponsored a study to estimate traffic loads on LTPP sites. Phase 2, described in this report, included the assessment of the overall quality of traffic data for all 890 LTPP traffic sites, the distribution of summary traffic data reports to all participating agencies describing what traffic data are available for the sites in their jurisdictions, and soliciting their input regarding traffic projections and the projection of axle loads for all LTPP sites with adequate traffic data.

Axle load projections were developed for all in–service years for 558 LTPP traffic sites that had adequate traffic monitoring data in the IMS database. The axle load projections were expressed as annual axle load spectra for single, tandem, and triple axles, and were placed into IMS computed parameter tables.

The main traffic data assessment and traffic projection activities carried out during the course of the Phase 2 study were:

• Data assessment and traffic projection carried out for all individual LTPP traffic sites.
• Preparation of the LTPP Traffic Feedback and Resolution Packages for all participating agencies; these summarized traffic data assessment and traffic projection results.
• Review of LTPP Traffic Feedback and Resolution Packages by RCOs.
• Review of LTPP Traffic Feedback and Resolution Packages by participating agencies.
• Implementation of review comments received from participating agencies.

Because of the large variation in the quality of traffic data and the uncertainty associated with traffic load projections developed for individual sites, traffic projections were assigned projection confidence codes to characterize the level of confidence associated with projected traffic loads:

• A: Acceptable projection results. Cumulative ESAL estimates are probably within ±50 percent of the actual cumulative ESAL values.
• Q: Questionable projection results. Cumulative ESAL estimates are probably within ±100 percent of actual cumulative ESAL values.
• N: Not available projection results. Axle load estimates could not be provided at the time.

Of the 890 LTPP traffic sites, 194 (21.8 percent) were assigned the acceptable projection confidence code and 364 (40.9 percent) were assigned the questionable projection confidence code. For 269 sites (30.2 percent), only the projection of AADT truck volumes was provided because of the unavailability of adequate site–specific axle load data. Many of these 269 sites had site–specific axle load data in the database, but the data were considered inadequate to be used for the projection of axle loads. No traffic projections were carried out for 63 sites because of lack of traffic data. The projection results for all 890 sites are summarized in appendix A.

The projection results were provided for the LTPP sites with unique traffic identification numbers. The total number of all LTPP sections is larger than the number of the LTPP sites because one traffic site may provide traffic data for several sections.

No axle load projections could be developed for 332 LTPP sites because of inadequate or missing data. Nonetheless, the 332 sites, representing 37 percent of all LTPP sites, contain a wealth of information regarding pavement materials, environment, and pavement performance. Without the required traffic data, this information cannot be utilized for the development of load–related pavement performance models. In order to provide traffic projections for these sites, truck classification and axle load distributions must be estimated in lieu of missing site–specific data. Any such estimates must be done judiciously. However, no guidelines exist for the estimation of truck class and axle load distributions, and the knowledge of typical or characteristic distributions is widely dispersed.

To overcome the difficulty of estimating the missing traffic data, it was proposed to develop the LTPP PLG. This report contains a description of the purpose, design parameters, and functionality of the PLG, a blueprint for the development of the PLG, and two examples of using the PLG to obtain traffic load projections for LTPP sites without site–specific truck class or axle load data.

Conclusions

• Traffic data collection in the field is essential for obtaining reliable traffic load data but not sufficient to obtain traffic loads for all years the pavement was in service because it is not possible to collect past traffic data or to collect traffic data all the time. To obtain pavement loads for all years the LTPP sites were in service, it is necessary to use the combination of traffic data (both historical and monitoring) and mathematical modeling referred to as traffic projection.
• The projection procedure used to estimate traffic loads reflects the quantity and quality of available historical and monitoring data. However, the projection procedure cannot fully overcome limitations caused by the presence of questionable data in the database and the use of data that has been factored to annual values. Recommendations to overcome the limitations caused by data quality concerns and factoring of data to annual values are presented in the next section.
• Considering that only about 23 percent of all GPS sites, and about 10 percent of all SPS sites, have acceptable axle load estimates, greater attention should be paid to the quality of traffic data. About 37 percent of all sites (332 sites) had insufficient or missing site–specific truck class or axle weight distribution data to carry out axle load projections. To obtain traffic load estimates for those 37 percent, replacement traffic data will need to be used.
• The LTPP traffic database provides good representation of traffic flows on major highways. Of the 558 sites with traffic projections assigned the acceptable or questionable projection confidence codes, 471 sites (84.4 percent) are located on rural or urban interstates and principal arterial highways. The minimum annual average daily truck volume on LTPP sites ranged from 30 trucks per day for a site located on rural minor collector highway to 6310 trucks per day on a site located on an urban interstate.
• The mean annual growth in truck volumes between 1994 and 1998 ranged from 6.5 percent for urban freeways and expressways to 3.0 percent for rural minor arterial highways. The mean annual growth in truck volumes on rural interstates was 4.6 percent.
• Since 1993, there has been a steady decline in the amount of monitoring axle load data available for traffic projection. The number of annual axle load spectra that could be used for the projection declined from 231 sections in 1993 to 150 sections in 1998. In other words, in 1998, of the 890 sites, only 150 (16.8 percent) had axle load spectra that could be used for projection of axle loads that yielded acceptable or questionable confidence codes. Only 51 sites in 1998 (5.7 percent of all sites) had axle loads yielding acceptable axle load projections. The decline in the amount of acceptable traffic data highlights the need for a renewed data–collection effort. It also highlights the importance of traffic modeling to extend limited sampling data and to compensate for the lack of data, and the need for the proposed LTPP PLG.

Recommendations

This section describes recommended future analytical, modeling, and traffic data–management activities needed to improve traffic load estimates for LTPP sites. There is an obvious need to collect more traffic data and, particularly, to collect high–quality traffic data; however, the activities recommended in this section are concerned only with traffic data analysis and management and not with data collection. The recommendations are divided into:

• Short–term activities required to carry out traffic projection, initiated in Phase 2, to completion.
• Long–term activities that go beyond the scope and methodology of Phase 2 work and are needed for better utilization of the existing traffic data.

Short–Term Activities

Responding to Participating Agencies

At the conclusion of Phase 2, approximately 60 percent of the participating agencies had not yet completed the review of the initial traffic projections. When the agencies' reviews of the initial traffic projections are completed, the reviews should be clarified and discussed with the participating agencies, and utilized to develop reviewed projections.

Developing the LTPP PLG

The LTPP PLG is essential for the estimation of pavement loads for the LTPP sites that do not have site–specific traffic monitoring data. Of 890 LTPP sites, 332 are without traffic load projections. To provide traffic load projections for these sites, axle load spectra and/or truck classification volumes must be estimated in lieu of missing data. The proposed PLG will provide a knowledge base, guidelines, and computational software to facilitate the estimation of traffic loads for the LTPP sites without site–specific traffic data. It will also be a very useful product emerging from the LTPP program for estimating traffic loads for general pavement design purposes. For these reasons, it is recommended to proceed with the development of the proposed PLG on a priority basis.

Completing Traffic Projection for all LTPP Sites

After the development of the LTPP PLG, the initial traffic projection for the remaining 332 LTPP sites without traffic load projections should be carried out, sent to the participating agencies for review, and included in the database. The completion of the initial traffic projections for all LTPP sites will enable pavement performance analyses that require the knowledge of traffic loads for the additional 334 LTPP sites. Potentially, the number of additional LTPP sections that could be used in the load–dependent performance modeling could be even greater because more than one LTPP section could use traffic data from a single LTPP traffic site.

Traffic projections carried out in Phase 2 were done for the period from the time the pavement was open to traffic to 1998. The projections should be updated to incorporate additional traffic data when the data become available.

Long–Term Activities

The recommended long–term activities include:

• Development of an action plan.
• QA of traffic data.
• Use of monthly traffic data.
• Regional traffic modeling.

Development of an Action Plan

The action plan should identify all issues facing LTPP traffic data collection and analysis and recommend specific actions for their resolution. Several components of such an action plan are recommended here.

Quality Assurance of Traffic Data

The previous traffic data QA process has resulted in traffic data that cannot be used without reservations. As documented in chapter 3, about 50 percent of all annual axle load spectra stored in the IMS database were judged unacceptable for estimating traffic loads.

The scope and effectiveness of the original LTPP traffic data QA process faced several limitations:

• The previous LTPP traffic data QA reports were carried out for periods equal to, or less than, 1 year. Consequently, it was not possible to assess long–term trends in historical and monitoring traffic data, such as annual differences in monitoring truck volumes (e.g., to compare AADT truck volumes for two consecutive years) and axle load spectra.
• Comparisons between historical and monitoring traffic data were not carried out, and no long–term assessment of trends in traffic data (spanning the entire time the pavement sections were in service) was performed.
• When the original traffic data QA work started, monitoring traffic data was available for only a few years, and often the amount of data was insufficient to establish trends and to obtain an understanding of what type of data are expected on a particular site. This paucity of data has made it difficult to screen data and reject suspect data confidently.
• Vehicles that could not be properly classified by the monitoring equipment (usually referred to as Class 14 vehicles) were typically not included in the traffic data QA process.

These barriers have been removed through: the development of the computerized procedure to display and summarize long–term trends in traffic data developed under this project; better understanding of trends and variation in traffic data; and the availability of additional traffic data. The anticipated development of the LTPP PLG will also contribute to the body of knowledge that can be used to assess and verify traffic data.

In summary, there is a fundamental need to carry out a basic traffic data QA process to better use traffic data that have been collected in the field. The traffic data QA process is necessary to provide reliable traffic load estimates. The collection of traffic data in the field is very expensive considering the cost of purchasing, installing, and maintaining AVC or WIM system equipment, and data collection and processing. The QA process will help agencies obtain a return on this investment and enhance the LTPP data.

The following activities may be required to carry out the basic traffic data QA process and to remove nonsensical data from the database:

• Reprocessing of some of the raw traffic data (for example, for agencies that have reported a large percentage of Class 14 vehicles).
• Improvements in the software used to produce LTPP traffic data QA reports so the graphs displaying traffic data are more user–friendly and understandable to representatives of participating agencies.
• Preparation of new LTPP traffic data QA reports (or re–examination of the old reports) and identification of erroneous data.
• Purging erroneous data.

Use of Monthly Traffic Data

If an annual axle load spectrum were judged to be unacceptable, all axle loads constituting the annual spectrum were rejected and not used in the projection. Yet annual axle loads may consist of an amalgam of both valid and questionable axle load data collected at different times of the year. Rejecting the entire annual spectrum from the projection process could have serious consequences for sites for which only a few monitoring annual axle load spectra are available.

Using shorter time periods, such as a month, enables a better utilization of available traffic data because axle load spectra are accepted or rejected as monthly, not annual, chunks. Consequently, a portion of the annual axle load data that would have been rejected as part of the annual data could be utilized as monthly data. Monthly axle load spectra are not available and need to be calculated from the raw binary files. Because of the large computational requirements, the monthly axle load spectra can be calculated only through CTDB data managers.

The use of annual data for the projection of traffic loads has been appropriate as the starting point. However, to overcome the disadvantages of using annual data, it is recommended to develop software to calculate monthly traffic data and, in particular, monthly axle load spectra, and to use the monthly axle load spectra for the projection of axle loads. The availability of monthly data will also provide a measure of seasonal variation in traffic loads.

Regional Traffic Modeling

Regional traffic modeling may be the best and most cost effective way to extend limited LTPP data and supplement them with data that are not part of the LTPP database. This situation specifically applies to agencies such as Florida and Texas that have a large corporate traffic database containing extensive WIM–type data but relatively little LTPP data. The feasibility of this approach (utilizing both LTPP and non–LTPP data for LTPP traffic projection) should be investigated using a pilot study in one or two agencies.

The proposed PLG, with its database management, display, and comparison features, will also facilitate the utilization of regional traffic data in lieu of missing LTPP data.