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 previous chapter described the kinds of changes (errors) that occur in the outcome of the vehicle classification process when different classification rule sets are applied to the basic axle weight and spacing data collected by WIM and AVC equipment, and the effect those changes have on the computation of normalized load spectra. When truck volumes are counted using one classification rule set and load spectra are computed using a different classification rule set, mismatches between the volumes reported and load spectra used create errors in the traffic load estimate used for pavement design. Whether those errors over- or under-estimate the load varies depending on which classification rule sets are used and what the vehicle characteristics are at each site.
This chapter explores the combined effects of all of the previously examined changes in classification. It computes traffic loads using truck volume data computed from the individual truck classification records at LTPP SPS TPF sites as processed using a variety of different classification rule sets and applied against the LTPP load spectra. These results are compared against the known traffic load at the SPS TPF sites. The errors computed are used to estimate the size and nature of traffic load errors that can be expected if load spectra from the SPS TPF sites are used at LTPP test sections that do not have valid, site-specific load spectra, and that use truck volume counts based on classification counts made using State-specific vehicle classification rule sets.
To understand the relative importance to pavement design of the wide variety of biases/ changes in estimated traffic volume that are caused by the use of alternative vehicle classification rule sets, it is necessary to simplify the traffic load estimate. This is most commonly done by selecting a set of coefficients or factors that allow conversion of a set of axles of different weights into a new, single value. These coefficients serve as weights of relative importance of individual axle weight measurements with respect to their effect on pavement deterioration.
In this approach, axle loads are divided into load spectra. Each weight bin within a load spectrum is assigned a weight (or impact factor) that is related to the damaging potential caused by a single application of an axle of that weight. The impact factors provide a measure of the relative importance of one load level against another with regard to potential pavement damage. (Note that these impact factors are not intended to be used in a direct computation of pavement damage because different pavement deterioration mechanisms are expected to have different sensitivity to various load levels.) [9]
Once a conversion method has been selected, it is possible to make direct comparisons of the combined effects of using different vehicle classification rule sets for the development of load spectra, and the determination of traffic volumes by vehicle classification that are used in the pavement design process.
The axle load spectra used in this study come from the LTPP SPS TPF data. They are computed as normalized load spectra for each type of axle (single, tandem, tridem, and quad) for each class of vehicle. To develop an estimate of the total annual loading for the analysis, the following procedure was applied:
Using this procedure, the estimates of total annual impact load for one classification rule set can be compared with that produced by another classification rule set.
A single total annual impact load value was then computed for each of the 18 TPF sites for which load spectra and vehicle classification data were available for this analysis. At most sites, data for a complete year were used to compute this single loading estimate. (In a few cases, the site had slightly less than 12 months of traffic data, but the available data were more than sufficient to estimate the effects of using different classification rule sets. These sites are treated as though there were 12 months of data present.)
Table12. Impact factors used to compute total loading estimate.
Single |
Tandem |
Tridem |
Quad |
||||
Weight Bin (lb) |
Factor |
Weight Bin (lb) |
Factor |
Weight Bin (lb) |
Factor |
Weight Bin (lb) |
Factor |
0–999 |
0 |
0–1,999 |
0 |
0–2,999 |
0 |
0–2,999 |
0 |
1,000–1,999 |
0 |
2,000–3,999 |
0 |
3,000–5,999 |
0 |
3,000–5,999 |
0 |
2,000–2,999 |
0 |
4,000–5,999 |
0 |
6,000–8,999 |
0 |
6,000–8,999 |
0 |
3,000–3,999 |
0 |
6,000–7,999 |
0 |
9,000–11,999 |
0 |
9,000–11,999 |
0 |
4,000–4,999 |
0 |
8,000–9,999 |
0 |
12,000–14,999 |
0 |
12,000–14,999 |
0 |
5,000–5,999 |
0 |
10,000–11,999 |
0 |
15,000–17,999 |
0.04 |
15,000–17,999 |
0 |
6,000–6,999 |
0 |
12,000–13,999 |
0.01 |
18,000–20,999 |
0.09 |
18,000–20,999 |
0.02 |
7,000–7,999 |
0 |
14,000–15,999 |
0.04 |
21,000–23,999 |
0.15 |
21,000–23,999 |
0.05 |
8,000–8,999 |
0.02 |
16,000–17,999 |
0.08 |
24,000–26,999 |
0.21 |
24,000–26,999 |
0.09 |
9,000–9,999 |
0.04 |
18,000–19,999 |
0.14 |
27,000–29,999 |
0.28 |
27,000–29,999 |
0.14 |
10,000–10,999 |
0.08 |
20,000–21,999 |
0.22 |
30,000–32,999 |
0.35 |
30,000–32,999 |
0.2 |
11,000–11,999 |
0.12 |
22,000–23,999 |
0.3 |
33,000–35,999 |
0.43 |
33,000–35,999 |
0.27 |
12,000–12,999 |
0.18 |
24,000–25,999 |
0.4 |
36,000–38,999 |
0.53 |
36,000–38,999 |
0.34 |
13,000–13,999 |
0.24 |
26,000–27,999 |
0.51 |
39,000–41,999 |
0.64 |
39,000–41,999 |
0.42 |
14,000–14,999 |
0.31 |
28,000–29,999 |
0.62 |
42,000–44,999 |
0.76 |
42,000–44,999 |
0.52 |
15,000–15,999 |
0.4 |
30,000–31,999 |
0.75 |
45,000–47,999 |
0.92 |
45,000–47,999 |
0.62 |
16,000–16,999 |
0.49 |
32,000–33,999 |
0.89 |
48,000–50,999 |
1.1 |
48,000–50,999 |
0.73 |
17,000–17,999 |
0.59 |
34,000–35,999 |
1.04 |
51,000–53,999 |
1.32 |
51,000–53,999 |
0.85 |
18,000–18,999 |
0.71 |
36,000–37,999 |
1.21 |
54,000–56,999 |
1.58 |
54,000–56,999 |
0.99 |
19,000–19,999 |
0.85 |
38,000–39,999 |
1.4 |
57,000–59,999 |
1.9 |
57,000–59,999 |
1.14 |
20,000–20,999 |
1.01 |
40,000–41,999 |
1.63 |
60,000–62,999 |
2.27 |
60,000–62,999 |
1.3 |
21,000–21,999 |
1.19 |
42,000–43,999 |
1.9 |
63,000–65,999 |
2.71 |
63,000–65,999 |
1.47 |
22,000–22,999 |
1.41 |
44,000–45,999 |
2.23 |
66,000–68,999 |
3.22 |
66,000–68,999 |
1.66 |
23,000–23,999 |
1.67 |
46,000–47,999 |
2.63 |
69,000–71,999 |
3.82 |
69,000–71,999 |
1.87 |
24,000–24,999 |
1.99 |
48,000–49,999 |
3.13 |
72,000–74,999 |
4.51 |
72,000–74,999 |
2.1 |
25,000–25,999 |
2.38 |
50,000–51,999 |
3.74 |
75,000–77,999 |
5.3 |
75,000–77,999 |
2.35 |
26,000–26,999 |
2.85 |
52,000–53,999 |
4.49 |
78,000–80,999 |
6.2 |
78,000–80,999 |
2.63 |
27,000–27,999 |
3.43 |
54,000–55,999 |
5.42 |
81,000–83,999 |
7.22 |
81,000–83,999 |
2.93 |
28,000–28,999 |
4.12 |
56,000–57,999 |
6.56 |
84,000–86,999 |
8.37 |
84,000–86,999 |
3.26 |
29,000–29,999 |
4.96 |
58,000–59,999 |
7.95 |
87,000–89,999 |
9.66 |
87,000–89,999 |
3.62 |
30,000–30,999 |
5.97 |
60,000–61,999 |
9.64 |
90,000–92,999 |
11.09 |
90,000–92,999 |
4.02 |
31,000–31,999 |
7.18 |
62,000–63,999 |
11.67 |
93,000–95,999 |
12.68 |
93,000–95,999 |
4.46 |
32,000–32,999 |
8.62 |
64,000–65,999 |
14.11 |
96,000–98,999 |
14.44 |
96,000–98,999 |
4.94 |
33,000–33,999 |
10.33 |
66,000–67,999 |
17 |
99,000–101,999 |
16.37 |
99,000–101,999 |
5.47 |
34,000–34,999 |
12.35 |
68,000–69,999 |
20.43 |
102,000–104,999 |
18.48 |
102,000–104,999 |
6.06 |
35,000–35,999 |
14.72 |
70,000–71,999 |
24.47 |
105,000–107,999 |
20.78 |
105,000–107,999 |
6.71 |
36,000–36,999 |
17.48 |
72,000–73,999 |
29.19 |
108,000–110,999 |
23.28 |
108,000–110,999 |
7.42 |
37,000–37,999 |
20.7 |
74,000–75,999 |
34.68 |
111,000–113,999 |
25.98 |
111,000–113,999 |
8.2 |
38,000–38,999 |
24.41 |
76,000–77,999 |
41.04 |
114,000–116,999 |
28.9 |
114,000–116,999 |
9.06 |
≥39,000 |
28.7 |
≥78,000 |
48.37 |
≥117,000 |
32.03 |
≥117,000 |
10.01 |
Load spectra and vehicle volumes by classification were computed for each of the available TPF sites, using each of the different classification rule sets tested. A single statistic representing total annual impact traffic load estimate was then computed for each TPF site for the following:
These outputs were then compared. The value computed using the LTPP load spectra and LTPP volumes is considered “ground truth” for this analysis.
The difference in the total annul traffic impact loading estimates that result from using different classification rule sets for count versus weight data collection is expressed as the ratio of the computed traffic load divided by the LTPP ground truth value (i.e., [Total Annual Impact Load with LTPP Load Spectra and State-Specific Class Volumes] divided by [Total Annual Impact Load with LTPP Load Spectra and LTPP Class Volumes]).
This statistic is referred to as the “Class Ratio” in this report, and this term is used throughout the remainder of this report to describe the results of the analysis of the effects of using traffic volumes collected with State-specific classification rule sets.
A Class Ratio greater than 1 indicates an increase in estimated traffic load when compared with what the LTPP system alone would have computed at these TPF sites. A ratio less than 1 indicates a decrease in estimated traffic load. At each TPF site, a mean and standard deviation of the Class Ratio values was computed across all of the tested classification rule sets. The mean and standard deviation are taken as the reasonable range of the “errors” associated with using different classification rule sets for developing load spectra and traffic volume by classification estimates.
9 The specific method chosen to compute any given set of impact factors will affect the results of such a study. Only one set of impact factors is presented in this report. The choice of a different set of impact factors based on different damage criteria will produce slightly different numerical results than those presented in this report. The project team is confident that the choice of any commonly agreed upon set of impact factors in place of the one chosen will not significantly alter the basic conclusions of this report.