Verification, Refinement, and Applicability of Long-Term Pavement Performance Vehicle Classification Rules

Chapter 5. Sensitivity of Pavement Design Models to Differences in Selected Vehicle Classification Rule sets

Analysis Approach

A sensitivity analysis of the MEPDG and American Association of State Highway and Transportation Officials (AASHTO) 1993 pavement design models was used as a means for evaluating the potential errors in pavement design life predictions that will occur as a result of combining axle load spectra collected using the LTPP classification rule set with vehicle classification and truck volume data collected using non-LTPP vehicle classification rule sets. The impact of these errors was quantified in terms of differences in pavement performance predictions for a set of typical pavement designs.

Because it is practically impossible to test MEPDG sensitivity to all possible combinations of site-specific vehicle streams and vehicle classification rule sets, a decision was made to identify and test a set of the worst cases that were identified based on the maximum differences in traffic load observed between LTPP and other vehicle classification rule sets. The logic supporting this approach is that if MEPDG outcomes are not sensitive to these extreme differences, then all other differences can be considered insignificant. On the other hand, if significant differences are observed, more detailed site-specific analyses may be required.

Traffic Scenarios

Because truck volumes and weights vary significantly on different types of roads, two extreme types of traffic loading conditions were tested. The following loading conditions are representative of road functional classes primarily covered in the LTPP program:

• High traffic loading condition (high truck volumes and high percentages of heavy trucks) characteristic of rural interstate (RI) roads.
• Low traffic loading conditions (low truck volumes and low percentages of heavy trucks) characteristic of non-interstate rural (“other”) principal arterial (ROPA) roads.

Traffic data from the Arkansas SPS-2 site, collected using the LTPP classification rule set, were used to generate base scenario traffic inputs for the high traffic loading condition. This site was selected to represent the loading patterns found on a typical interstate highway, with a very high percentage of Class 9 vehicles (70 to 90 percent) and very thick pavement structure. The input data consisted of annual average daily truck traffic (AADTT), normalized vehicle class distribution (VCD), normalized axle load spectra (NALS), and axle per class numbers. This site represents the highest overall cumulative axle loading among all SPS TPF sites and is representative of RIs with high truck volume. Because all vehicle classification rule sets accurately classify Class 9 vehicles, these types of roads are not expected to be affected greatly by the shifts in vehicle classification rule sets.

Traffic data from the Arizona SPS-1 site, collected using the LTPP vehicle classification rule set, were used to generate all base scenario traffic inputs for the low traffic loading conditions. This site has low truck volume and a low percentage of heavy trucks, and represents low overall cumulative axle loadings characteristic of many ROPA roadways. These roads also tend to have lower percentages of Class 9 vehicles in their truck traffic than interstate roads. Consequently, the effect of classification rule set changes is expected to be more pronounced for these roads. However, the normalized load spectra for Class 9 trucks for this particular site shows a percentage of heavy or overloaded trucks higher than typically observed.

For each of these base scenarios, three alternative traffic load estimates were developed by applying different vehicle classification systems to the base traffic data collected for the two base scenarios. This led to use of six traffic scenarios for pavement design sensitivity. The following rules were used to select the vehicle classification rule sets for the sensitivity analysis:

• Use LTPP vehicle classification rule set as a base case scenario.
• Use State classification rule set that produces the most significant increase in traffic load (when combined with load spectra collected using LTPP rule set) due to classification.
• Use State classification rule set that produces the most significant decrease in traffic load (when combined with load spectra collected using LTPP rule set) due to classification.

The results of the application of these rules are shown in table 14.

Table 14. Classification rule sets used to obtain AADTT and vehicle classification data for sensitivity analyses.

LTPP = Long-Term Pavement Performance

SPS = Special Pavement Studies

WIM = Weight in Motion

AVC = Automated Vehicle Classification

Volume and Classification Inputs

The AADTT and normalized VCD inputs used in the MEPDG analysis resulting from application of these vehicle classification rule sets are summarized in table 15 and table 16, respectively. These values were computed based on SPS TPF WIM data samples obtained and analyzed during Task 1 of the project. It is interesting to note that the same vehicle classification rule set—Washington WIM—resulted in very different percentile differences in AADTT values between the example RI and ROPA sites (2.5 percent and 35.4 percent, respectively), reinforcing the findings presented in Part I of this report that site-specific traffic composition plays a critical role in determining expected differences owing to application of different classification rule sets.

Table 15. AADTT Values based on different vehicle classification algorithms/rule sets.

LTPP = Long-Term Pavement Performance

AADTT = Annual Average Daily Truck Traffic

SPS = Specific Pavement Studies

AVC = Automatic Vehicle Classification

WIM = Weight in Motion

Table 16. Percentile VCD based on different vehicle classification algorithms/rule sets.

LTPP = Long-Term Pavement Performance

SPS = Specific Pavement Studies

AVC = Automatic Vehicle Classification

WIM = Weight in Motion

Table 17 shows AADTT by class computed using three vehicle classification rule sets. This table is useful for understanding which truck classes gained additional counts as a result of the application of various classification rule sets. For example, it shows an increase in heavy trucks (Classes 6 through 13) of 44 vehicles, or 59.4 percent, using the Washington rule set instead of the LTPP rule set for the Arizona SPS 1 site. Vehicle classes that gained the largest number of trucks are Classes 8 and 11.

Table 17. Actual VCD based on different vehicle classification algorithms/rule sets.

LTPP = Long-Term Pavement Performance

SPS = Specific Pavement Studies

AVC = Automatic Vehicle Classification

WIM = Weight in Motion

Loading Spectra Inputs

The Arkansas SPS-2 spectra were used to assess MEPDG sensitivity for the high traffic loading condition, and the Arizona SPS-1 spectra were used to assess MEPDG sensitivity for the low traffic loading condition. The same NALS and numbers of axles per class, developed using the LTPP rule set, were used with all three vehicle classification scenarios identified for each site. Appendix B contains the load spectra tables used.

Representative Pavement Structures

The sensitivity tests were applied against 16 base pavement structures. These were created by using two types of roadways, two pavement types, and four different climatic zones. AADTT, vehicle classification, and axle load spectra based on the LTPP classification rule set for Arizona SPS-1 and Arkansas SPS-2 sites were used to develop base pavement structures for high and low traffic loading conditions, respectively. Pavement structures for the base conditions were designed for 15 years of service life for flexible pavements and 20 years for rigid pavements. The same truck traffic growth (4 percent, based on the value provided in the MEPDG software) was assumed over the design period, and no errors in truck volume projection were considered in this analysis to isolate the error associated with application of different vehicle classification rule sets. The predicted pavement service life was established based on the condition when at least one of the pavement distress values or International Roughness Index (IRI) reached its terminal value using the 90-percent reliability design option.

Table 18 shows a summary of the final pavement designs.

Table 18. Summary of pavement and climate scenarios for sensitivity analysis.

AC = Asphalt Concrete

PCC = Portland Cement Concrete