U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
REPORT |
This report is an archived publication and may contain dated technical, contact, and link information |
|
Publication Number: FHWA-HRT-13-091 Date: November 2014 |
Publication Number: FHWA-HRT-13-091 Date: November 2014 |
The strength of the LTPP classification rule set is that it more accurately differentiates between heavy and light vehicles with three to five axles, with particular emphasis on the ability to identify and remove passenger vehicles pulling trailers from the truck categories used in pavement design. Because most of the trailers pulled by passenger vehicles are very light, the resulting LTPP load spectra should be heavier for the truck class into which many State rule sets most commonly incorrectly classify these vehicles: Class 8. (That is, when light axles are incorrectly included in Class 8, the large number of these axles in the lightest axle weight bin of the load spectrum produces a normalized load spectrum that reflects a large fraction of very light axles. The LTPP spectrum will be based on a much lower percentage of very light axles and thus higher percentages of heavier axles.)
This expected change in Class 8 load spectra can be observed by comparing the Class 8 load spectra computed on the basis of the LTPP rule set and the Ohio rule set. The Ohio rule set classifies a large number of passenger vehicles pulling trailers as trucks. When the Ohio rules are applied to the Pennsylvania SPS-6 site, a significant number of the trailers pulled by passenger vehicles that shift between classes have tandem axles. This adds a lot of very light tandem axles to the calculation of Class 8 load spectra. The resulting normalized load spectra for the LTPP and Ohio rule sets are shown in figure 6.
Figure 6. Graph. Comparison of Class 8 normalized tandem load spectra
for the LTPP and Ohio rule sets.
The Class 8 load spectra for these two rule sets are based on essentially the same number of heavy axles. However, the spectra for the Ohio rule set contain a very large number of light axles that are not included in the LTPP spectra, so the total number of axles in the Ohio rule set load spectra is larger. Because of these extra axles, the denominator used to convert the actual load spectrum into the normalized version shown in figure 6 is larger for the Ohio rule set, and therefore the fraction of axles attributed to each of the heavier load ranges is smaller for the Ohio spectrum than for the LTPP spectrum. In contrast, the normalized load spectrum for the Ohio rule set has a considerably greater proportion of axles in the two lightest load ranges. Thus, the normalized Ohio spectrum is much lighter than the normalized LTPP spectrum.
If the light Ohio load spectra were multiplied by the much greater Class 8 vehicle volume produced by the Ohio rules, the total number of heavy axles would be the same as that computed with the LTPP rule set’s load spectra and the LTPP rule set’s Class 8 volume estimate.[8] However, if the larger Ohio Class 8 volume were multiplied by the heavier LTPP load spectra, the total load (cumulative over all load ranges) computed for this class of trucks would be much higher than that applied by actual truck traffic.
While the Class 8 example above is a good illustration of how the load spectrum for a specific vehicle class and axle type can change—given a different classification rule set—changes do not occur in just this one vehicle class or axle type. Changes occur simultaneously across multiple vehicle classes and differ depending on the vehicle classification rule set to which the LTPP rule set is compared. The combinations of these multiple changes are extremely difficult to predict, in large part because at each site, the vehicles transferring into or out of a given classification may be light or heavy (particularly in the heavier vehicle classes), and until a site has been examined carefully, analysts cannot know whether the group of vehicles changing classes makes up a large or small percentage of the vehicles in each class.
The variety of effects generated by using different State classification rule sets is illustrated in figure 7 and figure 8. Figure 7 shows the normalized single-axle load spectra for Class 8 vehicles for the Pennsylvania SPS-6 site under nine different classification rule sets. Figure 8 shows the normalized Class 8 tandem spectra for those same rule sets.
These two figures readily show which classification rule sets significantly change the nature of the vehicles (and thus the loading characteristics of the vehicles) identified as Class 8. The Florida AVC, Ohio, Wisconsin, and Washington systems all primarily rely on axle spacing to differentiate these vehicles, whereas the California WIM, LTPP, and Florida WIM rule sets all use weight to help with this differentiation. The rule sets produce very different outcomes. One interesting observation is that the California AVC system matches the California WIM system reasonably well, despite its inability to use weight data in that differentiation.
Figure 7. Graph. Normalized single-axle load spectra for Class 8 vehicles,
Pennsylvania SPS-6 site, given different classification rule sets.
Figure 8. Graph. Normalized tandem-axle load spectra for Class 8 vehicles,
Pennsylvania SPS-6 site, given different classification rule sets.
Figure 9 shows the normalized Class 8 single-axle load spectra for the Kansas SPS-2 site under the different classification rule sets. This graph illustrates that the biases associated with classifying vehicles as trucks and cars have reasonably consistent effects on Class 8 load spectra across all of the TPF sites examined. (That is, the sites that have large increases in light vehicles in Class 8 all show large increases in the number of light single axles.) However, while the shape of these graphs changes predictably, the actual numerical changes in load spectra percentages are less predictable.
Figure 9. Graph. Normalized single-axle load spectra for Class 8 vehicles,
Kansas SPS-2 site, given different classification rule sets.
These figures also show that the effect of adding cars pulling trailers to Class 8 has a much more significant impact on the visual shape of the normalized single-axle load spectra than it does on that of the tandem axle spectra. This is likely because cars pulling trailers add at least two single axles per vehicle (and sometimes three or four single axles) to the load spectra for Class 8 when they are incorrectly shifted into this vehicle class. On the other hand, only a fraction of these vehicles have trailers with tandem axles, and they only have one tandem axle per vehicle. Therefore, the shift toward light axles may not be as strong in the tandem axle category when a State rule set places pickups pulling trailers in Class 8.
The patterns shown above follow a consistent bias: specific types of cars with trailers either are—or are not—included in Class 8. Those classification systems that include more passenger vehicles in Class 8 produce load spectra that have large numbers of light single and tandem axles. Those rule sets that are more restrictive have heavier load spectra.
In the larger truck categories, there is generally less consistency. Although the same bias concept does exist, in the case of larger trucks, there are two other important factors: 1) are the specific vehicle configurations identified by those biases present, and 2) are those vehicles light or heavy at a particular site?
In the Class 8 example, there is a reasonably large number of passenger vehicles pulling trailers, and all of those vehicles tend to be light. Therefore, the same basic trend occurs at all sites. The only question is how large those changes are at a given site. For heavier truck classes, neither of these factors is constant.
Once again the Class 13 and Class 10 examples illustrate the impact of those biases. The load spectra change in the same varied way that the estimated number of axles per truck changes for these classes, and for the same basic reasons.
Figure 10 shows how the heaviest portion of the normalized load spectra for quad axles changes for Class 10 under the different classification rule sets at the Maryland SPS-5 site. Figure 11 shows this same portion of the quad-axle load spectra for Class 13 at the Tennessee SPS-6 site. Figure 12 and figure 13 show these same two graphs, but for the New Mexico SPS-5 site. Because the vehicles are very different, and even more important, because the loads carried by those vehicles are very different, the effects of the classification shift differ.
Figure 10. Graph. Normalized quad-axle load spectra for Class 10 vehicles,
Maryland SPS-5 site, given different classification rule sets.
In figure 10, extreme peaks result at 54,000 lb for the LTPP and California WIM load spectra. These peaks occur because only one quad axle is observed in Class 10 under the LTPP and California WIM classification rule sets. That one axle falls in the 54,000-lb load range bin. Thus, this load spectrum has 100 percent of all quad axles in this load range. Just as important, no Class 10 quad axles are observed at this site when using the California AVC rule set. Conversely, the Washington, Ohio, and to a lesser extent, Wisconsin rule sets all observe a significant, heavy set of quad axles heavier than 72,000Â lb, while the two Florida rule sets observe a large percentage of quad axles in the 60,00-lb range.
When a classification rule set introduces large numbers of quad axles into this classification, the spike shown in the LTPP classification disappears. The specific axle that caused the LTPP rule set spike is still present in the load spectrum calculation, but it is now only one of many. Therefore, in this case, this particular load spectrum becomes heavier simply because a different classification rule set was applied. However, if that one axle had been very heavy, the alternative rule set could just as easily have made those load spectra lighter.
Figure 11 continues this example by now examining the quad distribution of Class 13 at the same Maryland site. In figure 11, we can see that the heavy quad axles observed in Class 10 in the Washington, Ohio, and Wisconsin rule sets in figure 10 now appear in Class 13 when the Florida and California classification rule sets are used. They are missing from both figures in the LTPP rule set, because these axles were part of trucks that were left unclassified by the LTPP rules. Also note that the large number of 60,000-lb axles observed in figure 10 (Class 10) for the Florida sites are not present in Class 13 for any of the classification rule sets. These axles were not exchanged between Class 10 and 13. Instead, these axles belong to trucks that are classified as Class 7 by the Washington, Ohio, and Wisconsin rule sets, and thus are not included in either figure 10 or figure 11. The lack of heavy axles in Class 13 for the Washington, Wisconsin, and Ohio rule sets means that for those rule sets, the Class 13 quad load spectrum is quite light for this Maryland site, while it is quite heavy for the California and Florida rule sets.
Class 13 quads are not always made lighter by the use of rule sets like Washington’s. Figure 12 shows that at the Tennessee SPS 6 site, the addition of the multi-axle Class 13 vehicles, which are not classified by the LTPP rules, creates a much heavier quad load spectrum than is observed in the LTPP spectrum, despite the fact that many of the LTPP Class 13 vehicles are reclassified as Class 10s by the Washington rules. Similarly, at this site, the Florida WIM rules have produced a much heavier load spectrum than most of the other rule sets. At this site, the Florida WIM rule set has observed trucks with more than 300 more quad axles than the Florida AVC rule set, and those additional axles generate a much heavier load spectrum. Conversely, the Ohio rule set classifies three-quarters of the trucks that LTPP designates as Class 13 as Class 10 trucks. Many of the heavy axles moved to Class 10 along with those trucks, and the resulting quad load spectrum is therefore much lighter than reported using the LTPP rules.
Figure 11. Graph. Normalized quad-axle load spectra for Class 13 vehicles,
Maryland SPS-5 site, given different classification rule sets.
Figure 12. Graph. Normalized quad-axle load spectra for Class 13 vehicles,
Tennessee SPS-6 site, given different classification rule sets.
These same changes are not present at the New Mexico SPS-5 site, even though the classification rule sets shift the same types of vehicles. Because the number of vehicles of each type and the weights they carry are different, the impacts of the rule set are different.
In figure 13, unlike figure 12, there are very few loaded axles. In fact, the LTPP rule set counts very few axles at all. Therefore, the differences in load spectra are essentially the normalized load spectra of the vehicles shifting into Class 10. In the Florida AVC rule set, only 10 quad axles are included in Class 10. Two of them are moderately heavy, explaining the peak loading observed. None of the other rule sets shift vehicles with heavy quad axles into Class 10, even though several rule sets (such as Washington’s) result in counting more than 100 quad axles on Class 10 trucks.
Figure 13. Graph. Normalized quad axle load spectra for class 10 vehicles,
New Mexico SPS-5 site, given different classification rule sets.
In figure 14, the major differences among heavy axle loads come from the vehicles that are not classified by the LTPP rules. In the Washington and Florida WIM rule sets, many of these vehicles are identified as Class 13, and the heavy axles of these vehicles are incorporated into this normalized load spectrum. In the classification rule sets that do not identify these vehicles as Class 13, these heavy quad axles remain in Class 15 and therefore do not appear in this graph. They also are not included in the LTPP load estimate, regardless of the classification rule set used to estimate the truck volumes used in the pavement design.
Figure 14. Graph. Normalized quad axle load spectra for Class 13 vehicles,
New Mexico SPS-5 site, given different classification rule sets.
In summary, there is no simple answer to the question, “How does truck volume by class and the associated load spectra for those vehicles vary if a classification rule set other than the LTPP rule set is used?”
When compared with the LTPP classification rule set, different State classification rule sets shift vehicles of different characteristics from one FHWA-defined class to another. Thus, without specifically examining the State classification system used to collect the traffic volume (by class) count at a specific site, it is not possible to predict which truck classes will gain or lose volume if a State’s classification rule set is used instead of the LTPP rule set.
However, even understanding the differences in how two classification rule sets are designed does not allow accurate prediction of the magnitude of truck volume changes, nor how those volume changes would affect the load spectra that apply to those trucks. This is because traffic characteristics tend to vary enough from site to site (even across multiple sites within a single State) that the percent of vehicles that change FHWA classes when different classification rule sets are applied also varies considerably from site to site. For heavy truck classes, although these variations are particularly influenced by State-specific truck size and weight laws, other factors, such as the presence or absence of large numbers of recreational vehicles, also significantly affect the size of observed changes caused by the application of any given pair of classification rule sets.
One approach for solving this problem in the future would be to develop files containing per-vehicle records for the vehicle classification data, similar to the ones currently used in W-cards for the weight data. In this way, vehicle classification data collected using one algorithm or rule set could be reprocessed if a different algorithm or rule set is desired (for example, to align vehicle classification and weight data collected using two different classification rule sets).
The wide variety of changes observed in truck volumes given the application of any specific State classification rule set are further reflected in the axle weight spectra produced from those truck volumes.
When these differences in load spectra are combined with the volume changes observed in individual truck classes, it appears that significant differences in predicted pavement loading for many vehicle classes may occur. The size of these differences is explored in the next chapter of this report. Whether these differences are sufficient to affect pavement design outcomes is discussed in Part II of this report.
8 The Ohio rule set would also estimate that a very large number of very light axles would "load" this pavement section. These axles are not present in the LTPP load estimate because the LTPP rule set places these vehicles in Classes 2 and 3, which are not incorporated into the load estimation process.