1. INTRODUCTION

The Department currently uses the 1993 AASHTO Design Guide for the design of pavement structures. The methodology used in that document is based on the concepts of the Present Serviceability Index and the Equivalent Single Axle Load (ESAL). ESAL calculations are dependent on the types and thicknesses of the pavement layers.

Research has shown that pavement designs based on the prediction of types of distress in addition to ride are useful in practice. These predictions require the use of mechanistic based distress prediction models which, in turn, require detailed axle load and vehicle classification data rather than a single value for the total number of ESALs.

For more than ten years, the States, through the National Cooperative Highway Research Program (NCHRP) of the Transportation Research Board (TRB) have sponsored the development of the Mechanistic-Empirical Pavement Design Guide(MEPDG) and successive versions of its implementing software.This effort presents a significant departure from what has been standard practice in pavement design for more than 30 years, especially in the consideration of traffic inputs to the design process. Adoption of the mechanistic-empirical approach requires that users discontinue the use of ESALs in favor of frequency distributions (spectra) of axle load weight groups, where each axle load is applied incrementally to the pavement structure.

As before, collection of traffic data for the MEPDG process includes truck weight and vehicle classification information as well as traffic volumes. Each of these forms of traffic data is now routinely acquired using automated means – weigh-in-motion (WIM) devices for truck weights, automatic vehicle classification (AVC) equipment for vehicle classification, and automatic traffic counters for traffic volumes. The means of collection of these data do not significantly affect the requirements of the MEPDG software.

  1. BACKGROUND

When initiated, this study used as its basis the traffic data input requirements for Version 1.1 of the MEPDG software. The data elements required include:

  • Base-year truck traffic volume
  • Vehicle (truck) operational speed
  • Truck traffic directional and lane distribution factors
  • Truck type and axle load distribution factors
  • Axle and wheelbase configurations
  • Tire characteristics and inflation pressure
  • Truck lateral distribution factor
  • Truck growth factors

As originally designed, these data are presented to the pavement designers at one of three levels. Level 1 data are site-specific. The data are collected either at the site or at a nearby location on the same route that has similar traffic characteristics. Level 2 data are taken from sites within a group of routes that have been shown to exhibit similar traffic characteristics within the State. Level 3 data are default values that were produced using data contained in the Long Term Pavement Performance (LTPP) program national database. However, in recent months the software designers removed the Level 2 option from the traffic data.

The following sections present detailed information about the traffic data requirements of the new guide, the process followed for deriving Florida’s input values, and the resulting recommended values.

  1. DEVELOPMENT OF FLORIDA-SPECIFIC INPUT VALUES FOR THE MEPDG TRAFFIC COMPONENT

The MEPDG implementing softwaredata includes the following data inputs to the software traffic module:

  • Base Year Traffic Information

Design Life

Opening Date

Initial Two Way Annual Average Daily Truck Traffic (AADTT)

Number of Lanes in Design Direction

Percent Trucks in Design Direction

Percent Trucks in Design Lane

Operational Speed

  • Traffic Volume Adjustment Factors

Monthly Adjustment Factors

Vehicle Class Distribution

Hourly Distribution

Traffic Growth Factors

  • Axle Load Distribution Factors
  • General Traffic Inputs

Lateral Traffic Wander

Number of Axles per Truck

Axle Configuration

Wheelbase

Each of these types of data may be provided for the design section as Level 1or 3, depending on the level of information available for the specific site. All traffic values are provided with national defaults with the exception of annual average daily truck traffic. Eachtraffic data type is discussed in the following sections.

In considering these data requirements, it is important to understand four underlying assumptions used in development of the traffic module during the performance of the work for the MEPDG:

  1. “The normalized axle load distributions by axle group type for each vehicle class remain constant from year to year unless there are political or economical (sic) changes that have an effect on the maximum axle or gross vehicle loads. However, the vehicle class distributions can change from year to year.
  2. The normalized axle load distribution by axle type and vehicle class does not change throughout the time of day or over the week (weekday versus weekend and night versus day). However, the vehicle class or truck distributions can change over time of day or day of week.
  3. The normalized axle load distribution for each axle type and vehicle class does not change from site-to-site within a specific region.(The term ‘region’ is not defined in the MEPDG software).
  4. The truck classifications for pavement design discussed previously provide a better description or grouping of roadways with similar truck traffic characteristics than the more traditional functional characteristics.”

These assumptions are based on the normalization of axle load distributions by vehicle class and by month for a number of LTPP WIM sites. Normalization was performed on these LTPP data sets to provide a common basis for comparison among sites with different numbers of vehicles. The normalization reduced the data to proportions rather than absolute numbers.However, the implementing MEPDG software does not allow any of the distributions to change over time. To achieve this effect it is necessary to estimate those parameters at the beginning and end of the design period and then use a nominal intermediate value as the traffic input value.

The eleven tables and other traffic data that are needed by the MEPDG software are illustrated in Figure 1 and described in the following sections.

Figure 1. Traffic Input Panel from MEPDG 1.1 Software.

Prior to providing the eleven tables, several items are required for the base year in the general project information. These are described in the following section.

Base Year Traffic Information

Design Life – The number of years that the pavement is expected to provide anacceptable level of service.

Opening Date – This is the date that the pavement designer intends that the project will be opened to traffic. It is entered directly on the screen.

AADTT – The two-way annual average daily truck traffic.

Number of Lanes in Design Direction – The number of through lanes in the design direction. The default is two lanes.

Percent Trucks in Design Direction – The percent trucks in the design direction (used to account for uneven distribution). The software provides a default value of 50%.

Percent Trucks in Design Lane– Allocates the greater number of trucks to a specific lane, usually the outermost lane. The software gives a default value of 95%.

Operational Speed –This value is the average speed for trucks. The default value in the software is 60 mph.

With the exception of AADTT and Percent Trucks in Design Direction, these base year values were not directly addressed in this study.

Traffic Volume Adjustment Factors

Each of the following data categories was addressed in this study.

The traffic volume adjustment factors are used to allocate the truck traffic to categories based on monthly adjustment, vehicle class, hourly distribution, and traffic growth considerations.

Monthly Adjustment Factors–This is actually a table of months of the year by vehicle class that shows the monthly variation in the volume of each truck type by month throughout the year. The default table shows a value of 1.0 in each cell of the table, indicating no monthly variation for any vehicle class.The twelve monthly adjustment factors for each vehicle class must sum to 12.0. Default values based on the LTPP national database are provided.

Vehicle Class Distribution – The vehicle class distribution allocates the total traffic among the truck vehicle classes. Each vehicle class receives a percentage value. The total of the values must be 100. Default values based on the LTPP national database are provided.

Hourly Distribution – The hourly distribution allocates a percentage of total truck traffic to each hour of the day. The percentages must sum to 100. Default values based on the LTPP national database are provided.

Traffic Growth Factors–Truck traffic growth can be none, linear, or compound. The rate of growth can be selected by the pavement designer. The default growth is compounded at four percent per annum.

Axle Load Distribution Factors – The axle load distribution factors are provided in four concatenated tables. Each of the four axle group configurations (single, tandem, tridem, and quad) has a table that provides the percentage of axle weights for each vehicle class that falls within each weight range. The weight ranges for each axle groupconfiguration are:

Single - 3,000 lb to 40,000 lb at 1,000 lb intervals

Tandem - 6,000lb to 80,000 lb at 2,000 lb intervals

Tridem and Quad– 12,000 lb to 102,000 lb at 3,000 lb intervals

The percentages for each vehicle class and axle group type must sum to 100. Default values based on the LTPP national database are provided.

General Traffic Inputs

Lateral Traffic Wander–These data fall within three types. The first is Mean Wheel Location, which is the distance from the outer edge of the rightmost wheel to the left edge of the right marking for the outermost lane of the pavement. The default Mean Wheel Location is 18 inches. The second is Traffic Wander Standard Deviation, which is the standard deviation of a sample of wheel location measurements. The default Traffic Wander Standard Deviation is 10 inches. The third data type is the width of the design lane.

Number of Axles per Truck – The average occurrence of each axle group configuration (single, tandem, tridem, and quad) for each vehicle class is provided in a table. Default values based on the LTPP national database are provided. The national default values for this parameter are shown in a later section.

Axle Configuration – This includes a number of individual data items to be provided, including:

  • Average axle width, measured from the outer edge of the outmost tire on one side to the outer edge of the outmost tire on the other side. The default width is 8.5 feet.
  • Dual tire spacing, measured from the center of one tire of a dual pair to the center of the other tire of that pair. The default spacing is 12 inches.
  • Tire pressure for both single and dual tires. The default pressure for both is 120 psi.
  • Axle spacing for tandem, tridem, and quad axle groups. The default spacings are 51.6 inches, 49.2 inches, and 49.2 inches, respectively.

Wheelbase –The spacing between the steering and next axle of a tractor or heavy single unit truck. The values are given for short, medium, and long with a percentage of the applicable vehicles that fall in the short, medium, and long categories. The default proportions are 33%, 33%, and 34%, respectively. The default spacings are 12 feet, 15 feet, and 18 feet.

  1. DEVELOPMENT OF SITE-SPECIFIC AXLE LOAD DISTRIBUTION FACTORS

This portion of the data development effort was performed during Phase I of the study. As indicated previously, the axle load distribution factors required by the MEPDG are provided in four concatenated tables. Each of the four axle group configurations (single, tandem, tridem, and quad) has a table that provides the percentage of axle weights for each vehicle class that fall within each weight range. The weight ranges are different for each axle group configuration and are:

Single - 3,000 lb to 40,000 lb at 1,000 lb intervals

Tandem - 6,000 lb to 80,000 lb at 2,000 lb intervals

Tridem and Quad – 12,000 lb to 102,000 lb at 3,000 lb intervals

The percentages for each vehicle class and axle group type must sum to 100. Default values based on the LTPP national database are provided. Figure 2 shows the axle load distribution factors panel from the MEPDG version 1.1 software.

Figure 2. Axle Load Distribution Factors Panel from MEPDG 1.1 Software.

The first part of the axle load distribution factor analysis addressed the factors for individual continuously operated WIM sites, thus providing site-specific axle load distribution factors.

Archived Florida WIM data were analyzed to determine the most appropriate approach to determining axle load distribution factors. Examination of the archives showed that more than 50 million individual truck weight records were available for the years 2001 through 2009. Further consideration led to the conclusion that the best analysis period was 2005 through 2009 based on knowledge that the data for those years had been more closely monitored for quality. This period also lends itself to annual review and update and a planned implementation of a five year averaging methodology. This analysis period provided 35 million individual truck records.

SAS© software was used to analyze the archived WIM data by month, year, day of the week, and vehicle type. Close examination of the WIM data files revealed that there were many cases of missing data. It was therefore important to analyze the data in a manner that minimized possible sources of bias. Specifically, the basic unit of analysis was defined as each day of the week within each month of data. That is, values were computed that represented an average for each Monday, Tuesday, etc., within eachmonth. Further aggregation was then performed using averaging across the days of the week.

Specific attention was given to the handling of:

  1. Multiple lanes of data
  2. Multiple days of data
  3. Multiple years of data

The issue of multiple lanes of data is associated with the fact that the axle load distribution factors are to be applied to a design lane.In those cases where there are two directions of traffic flow data collected, it was decided that the lane with the highest loading total would be used for computation of the site-specific axle load distribution. Analysis of the data showed that this was usually the outermost lane, except occasionally otherwise in urban areas.Total daily loads were calculated for each lane at each site for the analysis period. The design lane for each WIM site was then identified and flagged. Figures 3 and 4 illustrate the differences in the directional axle load distribution factors for Site 9904 on I-75.

Figure 3. Axle load distribution factors for I-75 northbound traffic.

Figure 4. Axle load distribution factors for I-75 southbound traffic.

Figures 3 and 4 show mean values of the annual axle load spectra for the tandem axle group of Class 9 vehicles for the years 2005 through 2009. The horizontal axes show the weight bin numbers. Increasing bin numbers indicate increasing weight. The vertical axes indicate the proportion of all observations falling in the indicated weight bin. The red line shows the mean values for all of the years taken together. The line labeled “mepdg” shows the national default values. The most important difference between Figures 3 and 4 is the peaking characteristic. Figure 4 shows a higher proportion of loads at the higher weight categories. This lane will therefore produce higher total loads when the MEPDG software is run.

As previously stated, variation in loading among days of the week was addressed by processing each day of the week in each year separately. Intermediate computations produced day-specific axle load distribution factors. These were combined to produce axle load distribution factors for the year. In this way, if unequal amounts of data were available for different days of the week, it would not bias the results.

The issue of acceptable sample sizewas addressed to determine the appropriate method for computing the axle load distribution factors by axle group, vehicle type, and month of the year. It is (almost) universally accepted that approximately 30 samples are sufficient to produce reasonable statistical estimates. In this application, there were multiple cases when the available data were sparse, while in other cases the data were huge. In this study, judgments were made to deal with both extreme cases. The following rules of thumb for sample sizes seemed to be reasonable for estimating axle load factors:

  1. Compute monthly axle load factors when there are at least 100 vehicles per month and 1,200 vehicles per year.
  2. Compute annual load factors when there are at least 100 vehicles per year, but fewer than 1,200 vehicles per year.
  3. Do not compute site specific annual load factors when there are fewer than 100 vehicles per year.

In application, these criteria mean that, when monthly factors are provided, each vehicle type will have a different set of axle load factors for each of the twelve months of the year. When annual load factors are provided, the same set of axle load factors for each vehicle type is repeated for each month. When neither monthly nor annual factors are provided, default values must be used.

In keeping with these criteria, axle load spectra type tables were generated that were flagged to determine the appropriate level of aggregation as illustrated in Figures 5 and 6.Figure 5 shows that monthly axle load distribution factors should be computed for Vehicle Type 9 at this site. This table summarized for each vehicle type in each month of the year the daily, monthly, and annual number of samples. In each monthly case, the monthly total is greater than 100, so monthly values can be used.Likewise, Figure 6 shows that annual axle load distribution factors should be computed for Vehicle Type 13 at this site. In each annual case, the monthly total is less than 100 but the annual total exceeds 100.