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 1. INTRODUCTION

In 1998, the Federal Highway Administration (FHWA) sponsored a two–phase study to develop traffic load estimates for Long–Term Pavement Performance (LTPP) sites. This report describes the results of the Phase 2 study. The Phase 1 study resulted in the development of a methodology for estimating axle load spectra for all years the LTPP sites were in service.^[1] Phase 2 used this methodology to estimate axle loads for all LTPP sites that had acceptable site–specific axle weight data. In total, traffic load estimates were made for 558 LTPP traffic sites.

Objectives of Phase 2

The original goal of Phase 2 was to apply the methodology developed in Phase 1 to obtain annual axle load spectra for 500 LTPP sections. During the course of Phase 2, this goal was refined to encompass the following specific objectives:

• Review traffic data for all LTPP sites in terms of quality and quantity.
• Carry out initial truck volume projections for all LTPP sites and submit them for review by participating agencies.
• Carry out initial axle load spectra projections for all LTPP sites with monitoring site–specific axle loads and submit them for review by participating agencies.
• Modify the initial traffic projections according to the review comments provided by participating agencies.
• Develop computed parameter tables for storage of the projected axle load spectra and supplemental projection data in the LTPP Information Management System (IMS) and upload the traffic projections into the database.
• Develop a prototype Pavement Loading Guide (PLG).
• Develop reliability or variability indicators for the projected traffic loads.

Background

A principal objective of the LTPP program is to answer key questions about pavement design and rehabilitation characteristics that will help the States and Provinces achieve pavement performance that is both long lived and cost effective.^[2] For interstate and other major highways, this objective is directly related to the need to quantify the relationship between pavement performance (deterioration of pavement structure with time) and traffic loads. Consequently, traffic data collection and analysis, required to obtain traffic loads, is the key activity within the LTPP program.

Since the inception of the LTPP program, traffic data collection has been the responsibility of the participating highway agencies, while the storage and analysis of traffic data have been done by LTPP program. Over the course of the program, participating agencies received a series of guidelines on how traffic data should be collected and reported to the LTPP. Briefly, the guidelines recommended that truck volume data should be collected using continuously operating automatic vehicle classifiers (AVC), and that truck axle weights should be collected using weigh–in–motion (WIM) scales or other scales operating during specified time periods.^[3,4]

Traffic data collected in the field are sent by the participating agencies to LTPP Regional Coordination Offices (RCOs) in an electronic format as individual vehicle records for processing and storage. The traffic data processing includes quality assurance (QA) checks and factoring. Factoring is used to obtain annual traffic data, such as annual average daily truck volumes, from the data collected during a portion of the year only. After processing by RCOs, data are stored in the Central Traffic Database (CTDB), and selected aggregated traffic data are also stored in the IMS.^[5]

At some LTPP sites, traffic monitoring equipment was not installed, was not operational, or was not calibrated. Also, in spite of the best plans and intentions of the participating agencies, traffic data collection in the field has been prevented by such factors as equipment malfunction, power failure, inclement weather, lack of funding, and lack of personnel. As a result, the amount and quality of traffic data collected or measured in the field vary considerably from agency to agency, from site to site, and from year to year. Typically, the amount of measured traffic data (the number of trucks that have been classified and weighed) represents only a fraction of the traffic that a typical LTPP site carried during the course of the LTPP program, and an even a smaller fraction of the traffic that occurred over the entire time the pavements were in service.

To quantify the relationship between pavement performance and traffic loads, it is necessary to estimate the total amount of traffic loads that pavement sections carried since opening to traffic. The estimating process relies on: (a) traffic data provided by the participating agencies and (b) mathematical modeling procedures that utilize available traffic data to fill in gaps in data. This report describes the results of the modeling procedures applied to the 558 LTPP sites for which participating agencies supplied acceptable site–specific axle weight data.

Description of Traffic Variables and Terms

The following description of traffic variables and terms provides background information to facilitate understanding of this report.

Classification of Highway Vehicles

LTPP uses the vehicle classification schema recommended by the FHWA Traffic Monitoring Guide and shown in table 1.^[6]

Trucks are defined as highway vehicles having dual tires on one or more axles. Buses are defined as highway vehicles with two axles and six tires, or three or more axles, manufactured to carry passengers. Trucks and buses are called commercial vehicles. Because the proportion of buses in the traffic flow is usually small, traffic composition is expressed in the form of a truck percentage that also includes buses. Similarly, the term truck volume is assumed to include both trucks and buses.

Truck class distribution is the distribution of commercial vehicles into the vehicle classes defined in table 1.

Table 1. FHWA commercial vehicle classification schema.

Classification of LTPP Sites by Highway Type

LTPP sites are classified according to the functional class of the highway on which they are located. At present, the LTPP sites are classified into six rural highway functional classes and five urban functional classes, listed below:

• Rural Principal Arterial–Interstate.
• Rural Principal Arterial–Other.
• Rural Minor Arterial.
• Rural Major Collector.
• Rural Minor Collector.
• Rural Local Collector.
• Urban Principal Arterial–Interstate.
• Urban Principal Arterial–Other Freeways or Expressways.
• Urban Other Principal Arterial.
• Urban Minor Arterial.
• Urban Collector.

Axle Loads and Axle Load Spectra

Equivalent single axle load (ESAL) is a quantity that is related to pavement damage caused by a standard axle load of 80 kilonewtons (kN) (18,000 poundforce (lbf)) carried by a single axle with dual tires. Because LTPP uses pounds (lb) as the unit for measurement and storage of axle weights, this unit is the primary unit used to report traffic projection results in this report.

Truck Factor (TF) is the number of ESALs per truck.

Axle load spectrum is defined as a frequency distribution of axle weights, of a given axle type, into weight ranges. Axle types are classified by the spacing between consecutive axles. Axles that are far apart (usually more than 2.44 meters (m) (8 feet (ft)) are called single axles. Two axles close together are called tandem axles; three axles spaced close together are triple axles; and four axles closely spaced are quadruple axles. Axle load spectrum is also referred to as axle load distribution.

Normalized axle load spectrum provides proportions of total axle loads that occur within designated load ranges. For example, the portion of 0.10 in the load range of 5,448 to 5,902 kilograms (kg) (12,000 to 12,999 pounds (lb)) for tandem axles means that 10 percent of all tandem axles are in the load range of 5,448 to 5,902 kg (12,000 to 12,999 lb). The normalized spectra (rather than actual spectra) are used to facilitate comparison of axle load spectra obtained for different truck volumes or sample sizes.

Axle–per–class coefficients provide the number of single, tandem, and triple axles for each vehicle class. Because the LTPP IMS database does not contain non–zero quadruple axle counts, no axle–per–class coefficients for the quadruple axles are discussed in this report. An example of axle–per–class coefficients for vehicle Class 9 (5–axle single trailer trucks) is presented in table 2.

Table 2. Example of possible axle–per–class coefficients for Class 9 vehicles.

Types of LTPP Traffic Data

The three main categories of LTPP traffic data are historical, monitoring, and supporting.

Historical data were estimated by the participating agencies for the LTPP sections before the use of monitoring equipment. Typically, historical data include annual average daily traffic (AADT) volumes, AADT truck volumes, TFs, and annual ESALs.

The years from the time the site was open to traffic to the time the site was included in the LTPP program are referred to as historical years. The years from the time a site was assumed by the LTPP program to present are referred to as monitoring years.

Monitoring data are data that have been submitted by the participating agencies for the years since the LTPP experiment began. There are two types of traffic monitoring data: measured and estimated. Measured monitoring data were obtained by field measurements using AVCs and WIM scales. Measured monitoring data are commonly referred to as monitoring data and typically include measured axle load data. Estimated monitoring data are data estimated by the participating agencies for monitoring years without measured monitoring data. Estimated monitoring data typically include annual truck volumes, TFs, and annual ESAL estimates.

Supporting data include site–specific characteristics such as site location, highway number, and pavement type.

Site–Specific, Site–Related, Regional, and Generic Data

Site–specific data are traffic data collected at or near the LTPP site using equipment that measures actual traffic that crosses the site. Site–specific data include truck class and axle load distributions.

Site–related data are traffic data collected on the same highway as the LTPP site but some distance from the site so that the collected data may be influenced by the presence of a major truck traffic generator (e.g., an intersection).

Regional data are traffic data collected on highways of the same functional class and located in the same region as the LTPP site. The site of the regional data and the LTPP site should be the subject of the same truck size and weight regulations.

Generic data are traffic data that represent typical traffic conditions, for example typical truck class distribution on rural interstates.

Projected Traffic Data

Traffic projection is a mathematical modeling process used to estimate traffic loads from samples of monitoring traffic data and other information. In the context of this study, the objective of traffic projection was to obtain axle load distributions for all years the LTPP sites were in service.

Projected traffic data are data that have been obtained by the projection process involving factoring or expanding sampled traffic data to obtain traffic data for an entire period. If the period used to factor or expand the data is a year, the result is annual estimates. If the period is the entire period the pavement was open to traffic, the result is cumulative estimates. In this report, the period for which traffic estimates were carried out is from the date the pavement was open to traffic to the end of 1998.

LTPP Traffic Data Structure

LTPP traffic data reside in two locations–the CTDB and the IMS.

The CTDB stores traffic data in five levels. Levels 1 through 4 store only measured monitoring data, whereas level 5 stores historical and supporting data. Level 1 features annual axle load spectra for all vehicle classes combined. Level 2 data contain annual axle load spectra for individual vehicle classes (FHWA Classes 4 through 13). Level 3 data feature daily axle load spectra for individual vehicle classes. Up to 365 tables (1 for each day of the year) may appear in level 3 for each of the 10 vehicle classes and for each monitoring year. Level 4 contains raw data submitted by the participating agencies. Level 5 contains supporting data.

The IMS contains level 1 data of the CTDB, including the monitoring axle load spectra for all vehicle classes combined, as well as annual ESALs.

Traffic Data Sources

Traffic data used in this report were obtained from the following IMS tables (first quarter of 2000, Level E release):

• TRF_BASIC _INFO (F01)–Basic information about site characteristics (sheet 1).
• TRF_EST_ANL_TOT_GPS_LN (F02)–Estimate of annual totals (volume and ESAL) in study lane when traffic monitoring equipment was not in service.
• TRF_MONITOR_BASIC INFO (F00)–Summary information concerning data collection and traffic characteristics (volume and ESAL) on a yearly basis.
• TRF_MONITOR_AXLE_DISTRIB (F04)–Annual axle load distribution by weight range and axle group from monitoring data, all vehicle classes combined.
• TRF_MONITOR_VEHICLE_DIST (F05)–Annual vehicle type distribution by FHWA vehicle class from monitoring data.
• TRF_MONITOR_AXLE_SUMMARY (F06)–Annual number of axles in each axle group from monitoring data.

In addition, CTDB level 2 data were utilized to investigate axle load spectra for individual vehicle classes.

The Challenge of Estimating Traffic Loads for All In–Service Years

Pavement damage caused by traffic loads is cumulative. Consequently, to quantify the relationship between pavement performance and traffic loads, all traffic loads imposed on the pavement during its service life must be taken into account. Estimating traffic loads for all the years the pavement was in service requires knowledge of both historical and monitoring traffic data.

The quantity and quality of available historical and monitoring data vary considerably between the LTPP sites and also between years for the individual LTPP sites. Historical data do not contain any truck class and axle load distribution data, and are unavailable for some sites for all or some of the historical years. Monitoring data are in the form of samples of uneven duration and quality taken during the monitoring years. To obtain axle loads for all years the pavement was in service (i.e., for both historical and monitoring years), appropriate mathematical modeling procedures must be used that fully utilize the available fragmented historical and monitoring data.

In addition, because emerging mechanistically based pavement performance models (such as the 2002 Pavement Design Guide^[7]) require knowledge of axle loads, axle loads must be estimated in terms of axle load spectra for all in–service years.

The estimation of cumulative axle loads for LTPP sites is done in two steps involving annual estimates and cumulative estimates.

1. Annual estimates–The sampled monitoring data are expanded or factored to obtain annual monitoring data. The shortest duration of a traffic sample that can be used to estimate annual monitoring data (e.g., annual axle load spectra) in the IMS database is 24 consecutive hours. The longest duration of a traffic sample is 365 days during the year. The quality and reliability of annual monitoring data depend on the quantity and quality of traffic samples and on the procedures used to expand the samples to obtain annual traffic data estimates. The procedure used to expand or project sampled data to obtain annual data is outlined in reference 5. The estimated annual data are stored in the CTDB and IMS databases.
2. Cumulative estimates–The annual monitoring data, available for some of the monitoring years, are combined with historical data and are projected for all the years the pavement was in service. Cumulative traffic estimates for any specified period expressed in number of years may be obtained by summation of the projected annual traffic data for these years.

Traffic data collection is essential to estimate traffic loads reliably. However, traffic data collection alone is not sufficient to obtain cumulative traffic loads because it is not possible to collect past data or to collect traffic data 100 percent of the time. To obtain pavement loads for all years the LTPP sections were in service, it is necessary to use a combination of traffic data (historical and monitoring) and traffic modeling procedures (traffic projection).

Results of Phase 1

The Phase 1 study encompassed preliminary assessment of the quantity and quality of LTPP traffic data.^[1] Because of the large differences in the quantity and quality of traffic data available for the LTPP sites, the LTPP sites were divided into five projection categories based on available data:

• Category 1 was intended for LTPP sites that have sufficient truck class and axle weight distribution data to enable the projection of annual and monthly variation in traffic loads.
• Category 2 was intended for LTPP sites with both truck class and axle weight distribution data; however, compared to Category 1, the amount and quality of data is insufficient for projection of monthly variation in traffic loads.
• Category 3 represents sites with adequate truck class distribution data, but without site–specific axle weight data.
• Category 4 represents LTPP sites with truck volume data but without site–specific truck class and axle weight distribution data.
• Category 5 represents LTPP sites without traffic data or with unacceptable traffic data.

Traffic projection procedures for estimation of axle loads for all in–service years were developed for each projection category. Regardless of the projection category, the basic procedure for projecting axle load spectra for the LTPP was as follows:

1. All available annual historical and monitoring data were used to establish a model predicting annual truck volumes for all years the section was in service.
2. A base annual axle load spectrum, representing a typical annual axle load spectra, was established.
3. For the years with missing annual monitoring axle load spectra, the missing spectra were obtained by multiplying the base annual spectrum by a factor related to annual truck volumes.

The methodology for projecting axle load spectra was evaluated and demonstrated using case studies for specific LTPP sites. Altogether, 12 case studies were conducted (3 for each projection category except 5, where no data can be acquired).^[1]

The main conclusions and recommendations of the Phase 1 study included the following:

• Proceed with Phase 2 study to develop and make available projected axle load spectra for selected LTPP sites using the projection methodology developed in Phase 1.
• Involving the participating agencies in the traffic projection process is crucial; many data problems cannot be resolved without input from local agencies.
• The projection of axle load spectra for LTPP sites without site–specific data (Categories 3 and 4) must be done judiciously and must be supported by a reference database source summarizing characteristic truck class and axle load distribution data. For this reason, the development of an LTPP PLG was proposed.
• Traffic projection and modeling is a highly cost effective and necessary process required to extend limited LTPP sampling traffic data and to compensate for the lack of measured traffic data in the past.

Report Overview

This report describes the findings and results obtained in Phase 2 of the FHWA study on the development of traffic loads for all years the LTPP sites were in service. It is organized into six chapters, including this one. Chapter 2 contains an outline of the projection procedure developed in Phase 1 and provides a detailed description of the process used to assess quality of traffic data, develop traffic projections in cooperation with the participating agencies, and assign projection confidence codes. It also includes the description of pilot studies that were used to develop procedures to involve participating agencies in the review of the initial projections and to facilitate the involvement of the regional LTPP data collection offices in the traffic data assessment and projection process. Chapter 3 summarizes the results of traffic data assessment and traffic load projection work. Chapter 4 describes the development of computed parameter tables used to store the projected traffic data in the IMS database. Chapter 5 describes the purpose, design parameters, and functionality of the proposed PLG. Finally, chapter 6 summarizes the study results and gives recommendations for future traffic analysis work.

This report also includes one appendix, appendix A, which tabulates traffic data assessment and projection results for all individual LTPP sites.