MEPDG Traffic Loading Defaults Derived From Traffic Pooled Fund Study

CHAPTER 11-SUMMARY OF FINDINGS AND RECOMMENDATIONS

SUMMARY OF FINDINGS

Traffic Loading Defaults Based on SPS TPF WIM Data

The focus of this study was to develop MEPDG traffic loading defaults based on the SPS TPF WIM data. The main traffic loading default is NALS. Other defaults that are based on WIM data are axle-per-truck coefficients, axle spacing, and wheelbase.

The researchers assessed available SPS TPF data and found them to be sufficient for developing representative site-specific values and alternate MEPDG axle loading defaults. The primary benefit of these new data is that they are of known acceptable data quality and sufficient quantity to develop representative NALS for each SPS TPF site. The limitation of the SPS TPF data is a limited study scope that includes only 26 WIM sites located in 22 States. Future expansion of the program would expand the applicability of the defaults and, thus, should be considered by FHWA LTPP Program or in conjunction with other programs, like the Long-Term Bridge Performance Program.

The following two tiers of NALS defaults were developed:

• Tier 1: Global defaults based on all applicable SPS TPF data. These defaults were based on averaging of NALS computed for the 26 SPS TPF sites by vehicle class and axle group type. One default NALS was computed for each vehicle class and axle group type combination, resulting in 25 unique default NALS. These defaults are fully compatible with the original MEPDG defaults in format and methodology used.
  
  
• Tier 2: Supplemental defaults based on subsets of SPS TPF data representing several specific loading conditions. These defaults are based on identifying groups of SPS TPF sites representing similar axle loading conditions and averaging of NALS for these sites. Multiple NALS were developed for each vehicle class and axle group type combination. NALS were identified by descriptive loading category. The most frequently observed NALS was labeled as "typical" for each vehicle class and axle group type. If a particular SPS TPF site had loading characteristics significantly different from any other site (usually much lighter or heavier), NALS for these sites were identified as special cases.

In addition to NALS, one set of default axle-per-truck coefficients was developed based on all SPS TPF sites. Representative axle spacing and wheelbase values were also computed.

MEPDG input files for all default NALS were developed in the formats compatible with DARWin-ME^TM and NCHRP 1-37A MEPDG software.

LTPP-PLUG

LTPP-PLUG was developed to facilitate the selection and use of axle loading defaults for MEPDG applications.⁽⁵⁾ The guide consists of two parts and a software application. The first part of the guide provides guidelines for selecting and using LTPP SPS TPF axle loading defaults within the for MEPDG and DARWin-ME^TM software. The purpose of the second part of the guide is to provide practical guidelines for generation of additional MEPDG traffic loading defaults that can be used by States and LTPP users to generate axle loading defaults based on their own WIM data or specific to their analysis purposes. The guide also contains an operator's manual that supports the use of the LTPP-PLUG interactive traffic loading library software application.

The LTPP-PLUG software application provides guidance in the selection of axle loading conditions (the traffic loading library) for pavement designs, and it produces input files for axle load data for use with the AASHTO DARWin-ME^TM software. It also allows States to add their own site-specific and default NALS to the PLUG database, which can be used to generate MEPDG input files and compute the new defaults. This software could be used to generate .alf or .xml files for any other NALS added to the LTPP-PLUG database tables.

Comparison of Default Values Based on SPS TPF Data with NCHPR 1-37A Defaults

NALS default values based on SPS TPF data were compared with NCHPR 1-37A defaults. The results of the comparison using tier 1 defaults indicated that significantly different MEPDG outcomes could be expected for some cases (JPCP slab cracking and total rutting of AC pavements). In addition, significantly different MEPDG outcomes are expected when different sets of tier 2 NALS defaults are used (light versus heavy). This conclusion highlights the importance of accurate measurement of the axle load spectra, along with the importance of the local knowledge of the expected axle loading conditions for pavement design and analysis.

The newly computed NALS defaults had fewer very light and fewer very heavy loads. This is most likely due to the fact that the new defaults were collected with more consistently calibrated WIM equipment compared to the dataset used for the development of the original NALS defaults under the NCHRP 1-37A project. The better calibration of the WIM scales used to develop the new defaults means that fewer very light loads (caused by under calibrated scales observing light loads) and fewer very heavy loads (caused by over calibrated scales observing heavy loads) are observed in the new default database.

Assuming that the new defaults are more accurate and representative of typical loading conditions, a conclusion could be made that pavement designs using the new defaults will be thinner than the designs using the original MEPDG defaults. However, from a practical perspective, the difference in the design thickness was significant only for a limited number of pavement scenarios tested.

WIM Data Selection Criteria for Generation of MEPDG Traffic Loading Defaults

WIM data selection criteria used in this study to develop the MEPDG traffic loading defaults addressed data availability, data quality, and data reasonableness. A number of statistical analyses were conducted to evaluate the reliability of computed NALS using different data quality and availability scenarios.

The following conservative minimum data availability criteria were identified to remove any potential DOW and monthly bias from computation of RANALS for individual WIM sites:

• Availability of at least 1 of each 7 DOW per month to avoid potential DOW bias. Days do not need to be consecutive.
  
  
• Availability of at least 1 of each 12 calendar months to avoid potential seasonal bias. Months do not need to be consecutive or from the same calendar year.

When more than 7 DOW per month or more than 12 calendar months of acceptable quality data are available, all available data should be used to compute site-specific RANALS.

The following acceptable data quality criteria were identified for this study:

• Data should be from a WIM site conforming to the ASTM E1318-02 specification for type I WIM system accuracy and performance requirements specified in the LTPP Program's internal field operations guide for SPS WIM sites, as evidenced by documented WIM installation reports, and annual or semiannual validation/calibration reports.^(6,12)
  
  
• Selected data must pass a quality review using the LTPP QC procedures for the SPS TPF study and be identified as Level E data in the LTPP SDR database.

In addition to data availability and data quality, data reasonableness criteria were applied to address cases when data may be valid but atypical. A series of tests were developed to aid in the identification of atypical data.

WIM System Accuracy

The effect of WIM system accuracy on axle weight measurements, NALS estimates, and the associated MEPDG outcomes was also investigated. The MEPDG analysis outcomes indicated that drift in WIM system calibration that leads to over 5 percent bias in mean error between true and WIM-measured axle weight could lead to significant differences in MEPDG design outcomes. Therefore, it is important to keep WIM systems properly calibrated and schedule calibration visits when a consistent drift in axle weight measurement close to 5 percent of the reference dataset (data collected right after previous calibration visit) is observed.

Current SPS TPF WIM performance requirements result in a level of error in WIM data that is insignificant for MEPDG applications, provided that excessive bias values are mitigated in a timely manner through WIM calibration and maintenance activities.

New Summary Statistics to Characterize NALS and to Quantify NALS Reliability

Two new statistical parameters were developed in this study to simplify the description and comparisons of different NALS. Because NALS required for MEPDG analyses include thousands of numbers (considering all load bins, all vehicle classes, and all axle types), comparing or evaluating different NALS is not straightforward. Therefore, the ability to summarize these complex sets of numbers into a single statistic that represents the whole NALS is very helpful. That summary statistic can be used to compare different NALS, characterize the relative size of the loading condition associated with a site or when comparing the likely effects of using different NALS or quantifying NALS accuracy.

Distributions like NALS can be summarized in many ways. For the success of this study, it was critical to define a summary statistic that would account for the following:

• The frequency of load applications, expressed as percentages of the total axle load applications at each load level.
  
  
• The impact of different axle load levels on the MEPDG-based pavement performance predictions, accounted for through MEPDG-derived weighting factors.

The following parameters were developed and used in this study:

• RPPIF: A summary statistic for the comparison of NALS. This statistic was also used to help group or cluster NALS that represent similar loading characteristics. This statistic converts NALS to a single value, considering both the frequency of load applications and the effect of different load magnitudes on pavement performance.
  
  
• PWLE: A parameter for quantifying errors associated with using several NALS to compute a single group NALS. The PWLE statistic can also be used to assess the reliability of that summary NALS. PWLE is a single value that provides a representative estimate of the accuracy of a summary NALS, as well as the range of expected errors associated with using that summary. To compute the PWLE statistic, it is necessary to identify the statistical CL at which the reliability of the NALS is to be computed. PWLE takes into consideration both the relative percentage of the traffic at each load level and the impact of each load level on the MEPDG-based pavement performance predictions.

These parameters are applicable for studies involving NALS comparison and grouping and for the evaluation of NALS accuracy. Values for weight factors, W_ij, developed in this study were based on generic pavement designs and globally calibrated MEPDG models. These values could be replaced with values developed using different or locally calibrated MEPDG models, or based on different pavement designs that are more applicable to studies being performed, or for local agency use.

RECOMMENDATIONS FOR USING TRAFFIC LOADING DEFAULTS

These recommendations address the selection of default or surrogate NALS based on the data collected at SPS TPF sites if no loading data are available to develop reliable NALS estimates.

Knowledge of Local Traffic Loading Conditions

Before selecting the NALS defaults, it is recommended that every effort is made to understand the expected traffic loading pattern at the site for which the NALS selections are being made. This should include an analysis of site location and likely truck traffic characteristics observed at the site, including the following:

• Dominant heavy vehicle classes observed at the site.
  
  
• Percentage of through trucks versus local delivery trucks.
  
  
• Dominant commodities being carried by trucks.
  
  
• Truck sizes and weight laws of different States (which can result in different truck body/configurations for the same FHWA truck class being utilized within different states).
  
  
• Presence and effectiveness of weight enforcement activities.
  
  
• Number of roadway lanes.

This information should be used to establish descriptive traffic loading conditions for the dominant heavy vehicle classes observed at the site. An example of axle loading categories by loading condition is provided in table 24.

A special effort should be made to identify loading conditions for class 9 vehicles. For the majority of the LTPP sites (and the majority of U.S. primary roads), class 9 is the dominant heavy vehicle type. Class 5 vehicles are frequently dominant but not heavy enough to make a significant contribution to total traffic loading; thus, they may be excluded from determination of the dominant heavy vehicle classes, unless local knowledge exists of heavier-than-usual class 5 vehicles at the site.

In some instances, historical WIM data or recently collected portable WIM data may be available for the site. These data may not be accurate enough to compute NALS for the site but may be useful in establishing a descriptive loading condition, including evaluation of loading conditions by direction and for identifying loading condition for the design lane.

Default Selection

Once descriptive loading categories are determined for the dominant vehicle classes expected at the site, tier 2 default NALS corresponding to the identified loading categories could be selected from the database (see appendix D). If no local knowledge exists, either tier 1 NALS defaults or tier 2 NALS defaults representing typical conditions could be used. For a more conservative analysis, tier 2 NALS defaults representing heavy conditions could be used. Chapter 10 provides decision tree to aid in the default selection process.

If a State's truck sizes and weight laws allow loads exceeding Federal regulations 23 U.S.C. 127 and 23 CFR 658, and the site has a low percentage of interstate traffic, there is a higher likelihood of heavier-than-typical NALS, especially if there is local knowledge of heavy commodities being transported by certain vehicle classes using road facilities that include the site of interest.^(19,20)

MEPDG Testing of NALS Alternatives

Not all pavement designs have the same sensitivity to NALS; therefore, analysis of MEPDG outcomes from different NALS inputs may be beneficial to determine if using alternative NALS will result in differences in MEPDG outcomes that have practical significance. In this study, for example, MEPDG analysis considering different single axle NALS did not result in outcomes that carry practical significance for any vehicle classes, except class 11, while for tandem axles results were significantly different for all classes, except classes 5 (due to low weight) and 12 (due to low volume). NALS for tridem and quad axles may lead to significantly different outcomes if these axles carry a significant percentage of the total load for the site.

Applicability and Limitations of LTPP SPS TPF Defaults

The new defaults are based on high-quality WIM data from 26 sites primarily located on RI and ROPA roadways. These are the road types for which State agencies are likely to use the MEPDG design method. The sites were located throughout the U.S. and provide reasonable estimates of expected axle loading condition for these types of roads, in the absence of site-specific loading information.

However, based on the limited number of sites, it was not possible to determine if these defaults would be applicable to roadways in other functional categories or in jurisdictions that utilize types of trucks or serve industrial or agricultural facilities significantly different from the ones observed in the LTPP SPS TPF study. For example, only two SPS TPF sites were located on roads classified as "urban." Usability and applicability of the defaults developed in this study should be determined at the State level by comparing these defaults with the available site-specific data. Special care should be taken to assure that the data collected by the State are based on data selection criteria similar to the ones described in this report to assure data accuracy and data completeness for generation of representative NALS.

It is highly recommended that some local knowledge of loading conditions be applied when judging the applicability of the SPS TPF defaults, especially for roads that are likely to carry a large percentage of local truck traffic.

Use of LTPP SPS TPF Defaults with MEPDG or DARWin-ME^TM Software

The SPS TPF default NALS were developed to follow the formats established for the NCHRP 1-37A MEPDG and DARWin-ME^TM software. Defaults for axle spacing and APC coefficients could be uploaded to these software products as direct user inputs, while NALS could be imported as *.alf or *.xml files. This could be accomplished using the LTPP-PLUG software.

FUTURE RESEARCH RECOMMENDATIONS

Future research recommendations include the following:

• Axle loading defaults developed in this study are based on the data from 26 LTPP SPS TPF sites. It is recommended that as more sources (additional WIM sites) of the research-quality WIM data are identified, these data should be included in development of the updated NALS defaults.
  
  
• NALS defaults developed in this study were based on clustering of NALS representing axle loading conditions that are likely to produce significantly different MEPDG outcomes based on globally calibrated models. As MEPDG pavement performance prediction models and design criteria evolve, these clusters may change. Therefore, the defaults should be reviewed and updated when major revisions are made to the MEPDG models.
  
  
• NALS defaults were developed for single, tandem, tridem, and quad axle groups. However, heavy trucks that have axle groups with five or more axles could be found in some States. It is recommended that the effect of the axle loads and the frequency of occurrence of such trucks on pavement design should be evaluated to determine if the MEPDG should consider these trucks (with five or more axle group types) in routine pavement design.
  
  
• All NALS defaults were developed based on data collected using the LTPP vehicle classification scheme. In summer 2012, additional rules were introduced to that scheme, resulting in changes how vehicle classes 7, 10, and 13 are identified. It is recommended that NALS developed for these vehicle classes are revised once the new data become available.
  
  
• Since truck traffic characteristics, pavement design criteria, and pavement design philosophy (considering pavement thickness, frequency of maintenance/rehabilitation activities, and pavement design reliability) differ from State to State, it is recommended that States evaluate the applicability of the defaults developed in this study to their local conditions through comparison of the default and State-specific NALS, as well as MEPDG outcomes using these NALS.