U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
2023664000
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
This report is an archived publication and may contain dated technical, contact, and link information 

Publication Number: FHWAHRT05079
Date: May 2006 

Optimization of Traffic Data Collection for Specific Pavement Design ApplicationsAlternative Descriptions of FiguresPDF Version (545 Kb)
PDF files can be viewed with the Acrobat® Reader® Figure 1. Diagram. Schematic of the sensitivity of distress predictions to load spectra input. The figure shows a diagram that conceptualizes how the various lowpercentile input in the load spectra is translated by pavement design guide analysis into a range in pavement lives by distress type. Beginning at the left, the diagram shows a narrow vertical load bin, which is marked off in three sections. The bottom twothirds of the rectangle is designated as truth from 100 percent sample. The top section, which is just less than half of the rectangle, is labeled range from sampling scenario. It is derived by mean plus Z sub a S equals a. The bottom of the rectangle, which is slightly more than half, is labeled as frequency. It is derived by mean minus Z sub a S equals a. In the middle of the diagram is a process square labeled PDG pointing to the right. The right side of the diagram shows the pavement life expectancy predictions obtained by using the PDG software. Figure 2. Bar chart. LongTerm Pavement Performance Program sites with weighinmotion data available for periods longer than 359 days per year. This bar chart that shows the number of LongTerm Pavement Performance Program sites with 359 days of weighinmotion data or more on the vertical axis versus the number of years over which data are available on the horizontal axis. The number of sites range from 46 sites with 1 year of data to 6 sites with 2 years of data to 1 site with 5 years of data. Figure 3. Bar chart. LongTerm Pavement Performance Program sites with weighinmotion data available for periods longer than 299 days per year. This chart shows the number of LongTerm Pavement Performance Program sites with 299 days of weighinmotion data or more on the vertical axis versus the number of years over which data are available on the horizontal axis. The number of sites range from 76 sites with 1 year of data to 45 sites with 2 years of data to 2 sites with 6 years of data. Figure 4. Scatter chart. Flexible pavement site selection by average annual daily truck traffic and structural number. This scatter plot shows the combination of average annual daily truck traffic on the vertical axis versus structural number on the horizontal scale for the flexible LongTerm Pavement Performance Program sites selected. The data points range from 50 to 2,000 trucks per day with corresponding structural numbers ranging from 1 point 5 to 10. Figure 5. Scatter chart. Rigid pavement site selection by average annual daily truck traffic and slab thickness. This scatter plot that shows the combination of average annual daily truck traffic versus slab thickness for the rigid LongTerm Pavement Performance Program sites selected. The data points range from 100 to 3,800 trucks per day with corresponding slab thickness values ranging from 150 to 280 millimeters (6 to 11 inches). Figure 6. Clustering tree diagram. Annual distributions of tandemaxle loads, Washington State LongTerm Pavement Performance Program sites. This clustering tree diagram shows the level of similarity between extendedcoverage weighinmotion LongTerm Pavement Performance Program sites in Washington State in terms of frequency distribution of tandemaxle loads. The degree of similarity is indicated by a low number of the Euclidean distance. Table 16 provides the Euclidean distances represented in figure 8. Figure 7. Graph. Tandemaxle load distributions for the cluster of Washington State LongTerm Pavement Performance Program site 6048. This graph shows an example of the tandemaxle load distribution similarities for the cluster of Washington State site 6048. Axle loads, represented on the horizontal axis, range from 15 to 230 kilonewtons (3,300 to 51,700 pounds force). The corresponding frequencies on the vertical axis range from 0 to point 12. It is characterized by a dominance of lightly loaded tandem axles in the 40 to 70kilonewton (9,000 to 15,700pounds force) range. Figure 8. Graph. Tandemaxle load distributions for the cluster of Washington State LongTerm Pavement Performance Program site 1007. This graph is an example of the tandemaxle load distribution similarities for the cluster of Washington State site 1007. Axle loads are represented on the horizontal axis and range from 15 to 230 kilonewtons (3,300 to 51,700 pounds force), while the corresponding frequencies on the vertical axis range from 0 to point 14. It is characterized by a dominance of heavily loaded tandem axles in the 115 to 145kilonewton (25,900 to 32,600pounds force) range. Figure 9. Graph. Example of National Cooperative Highway Research Program 137A design guide output, site 181028. This graph is an example predicted rutting on the vertical axis and pavement age in months on the horizontal axis. Scenarios 10, 20, 30, 40, and 44 are shown on the graph. In addition, failure was established at 10 millimeters (0 point 4 inches), which is represented on the graph. The closely clustered scenarios 20, 30, and 40 are predicted to reach failure at approximately 312 months. Scenario 10 accumulates rutting at a slower rate and reaches failure at an estimated 336 months, while the predicted rutting for scenario 44 accumulates at a significantly lower rate, taking 516 months to reach failure. Figure 10. Graph. Summary of mean in life predictions, site 181028 in Indiana, confidence 50 percent. This graph shows the mean estimated life of LongTerm Pavement Performance Program section 181028 under each scenario. The estimated mean life for scenario 10, used as a reference, was estimated to be 30 years. The estimated lives of all scenarios, excluding scenarios 44, 45, 46, and 47, are relatively close to 30 years. Scenarios 44, 45, 46, and 47 are significantly higher, ranging from 45 to 50 years. Figure 11. Graph. Summary of the range in predictions, site 181028 in Indiana, confidence 75 percent. This graph shows the seventyfifth percentile range of estimated life of LongTerm Pavement Performance Program section 181028 under each scenario. The estimated mean life for scenario 10, used as a reference, was estimated to be 30 years. The estimated lives of all scenarios, excluding scenarios 44, 45, 46, and 47, are relatively close to 30 years. Scenarios 44, 45, 46, and 47 are significantly higher, ranging from 45 to 50 years. In addition, confidence intervals are shown where the confidence interval of scenario 47 is the largest, at approximately 12 years. Figure 12. Graph. Summary of the range in predictions, site 181028 in Indiana, confidence 85 percent. Graph. This graph shows the eightyfifth percentile range of the estimated life of LongTerm Pavement Performance Program section 181028 under each scenario. The estimated mean life for scenario 10, used as a reference, was estimated to be 30 years. The estimated life of all scenarios, excluding scenarios 44, 45, 46, and 47, are relatively close to 30 years. Scenarios 44, 45, 46, and 47 are significantly higher, ranging from 45 to 50 years. In addition, confidence intervals are shown where the confidence interval of scenario 47 is the largest at approximately 15 years. Figure 13. Graph. Summary of the range in predictions, site 181028 in Indiana, confidence 95 percent. This graph shows the ninetyfifth percentile range of the estimated life of LongTerm Pavement Performance Program section 181028 under each scenario. The estimated mean life for scenario 10, used as a reference, was estimated to be 30 years. The estimated life of all scenarios, excluding scenarios 44, 45, 46, and 47, are relatively close to 30 years. Scenarios 44, 45, 46, and 47 are significantly higher, ranging from 45 to 50 years. In addition, confidence intervals are shown where the confidence interval of scenario 47 is the largest at approximately 24 years. The confidence interval of scenario 43 is around 11 years. Figure 14. Graph. Summary of the range in predictions, site 181028 in Indiana, confidence 99 point 9 percent. This graph shows the ninetynine point nine percentile range of the estimated life of LongTerm Pavement Performance Program section 181028 under each scenario. The estimated mean life for scenario 10, used as a reference, was estimated to be 30 years. The estimated life of all scenarios, excluding scenarios 44, 45, 46, and 47, are relatively close to 30 years. Scenarios 44, 45, 46, and 47 are significantly higher, ranging from 45 to 50 years. In addition, confidence intervals are shown where the confidence interval of scenario 47 is the largest at approximately 48 years. The confidence interval of scenario 43 is around 17 years. Scenario 23 exhibits a 12year confidence interval. Figure 15. Bar chart. Pavement life prediction comparison between actual annual average annual daily truck traffic growth rate and 4 percent annual average annual daily truck traffic growth rate, flexible pavement sites. This bar chart compares predicted pavement life for the flexible pavement sites. The chart includes predicted life using a 4 percent annual growth rate as well as predicted life using the actual annual growth rate. Values used in this figure can be found in table 26 (4 percent annual growth rate) and table 28 (actual annual growth rate. Figure 16. Bar chart. Pavement life prediction comparison between actual annual average annual daily truck traffic growth rate and 4 percent annual average annual daily truck traffic growth rate, rigid pavement sites. This bar chart compares predicted pavement life for the rigid pavement sites. The chart includes predicted life using a 4 percent annual growth rate as well as predicted life using the actual annual growth rate. Values used in this figure can be found in table 27 (4 percent annual growth rate) and table 29 (actual annual growth rate). Figure 17. Diagram. Components of the percent difference between pavement life predictions for scenario X and those for scenario 10. In the center of the diagram is a bold vertical line that illustrates the life estimate error band. The length of the line is designated as A, the life prediction using the lowest percentile for all traffic input under discontinuous Scenario X. The distance from the baseline scenario 10 to the bottom of the line, B, is designated life prediction from a continuous scenario or from mean traffic input for discontinuous Scenario X. Figure 18. Bar chart. Statistics of error component A in life predictions (percent), flexible pavement sites with average annual daily truck traffic less than or equal to 800 trucks per day per lane. This bar chart of the mean and standard deviation of error component A in life predictions for flexible pavement sites with average annual daily truck traffic less than or equal to 800 trucks per day per lane. For scenarios 11, 12, 20, 21, 22, and 23, the mean values range from 0 to minus 10 percent, while the standard deviation values range from 5 to 20 percent. The mean and standard deviation for scenarios 30 and 31 are approximately 5 and 55 percent, respectively. Scenarios 40, 41, 42, and 43 yield mean values around 10 percent and standard deviations around 50 percent. The remaining scenarios (44, 45, 46, and 47) have mean values that range from minus 40 to minus 45 percent, with corresponding standard deviations ranging from 25 to 30 percent. Figure 19. Bar chart. Statistics of error component A in life predictions (percent), flexible pavement sites with average annual daily truck traffic greater than 800 trucks per day per lane. This bar chart shows the mean and standard deviation of error component A in life predictions for flexible pavement sites with average annual daily truck traffic greater than 800 trucks per day per lane. Scenarios 11, 12, 20, 21, 22, and 23 have a mean values range from plus 1 to minus 15 percent. The standard deviation values range from 5 to 10 percent. The mean and standard deviation for scenarios 30 and 31 are approximately minus 15 and plus 5 percent, respectively. Scenarios 40, 41, 42, and 43 yield mean values around minus 5 percent and standard deviations around 10 percent. The remaining scenarios (44, 45, 46, and 47) have mean values that range from 15 to 25 percent, with corresponding standard deviations ranging from 40 to 45 percent. Figure 20. Bar chart. Statistics of error component A in life predictions (percent), rigid pavement sites with average annual daily truck traffic less than or equal to 1,200 trucks per day per lane. This bar chart shows the mean and standard deviation of error component A in life predictions for rigid pavement sites with average annual daily truck traffic less than or equal to 1,200 trucks per day per lane. For scenarios 11, 12, 20, 21, 22, and 23, the mean values range from 5 to 10 percent, while the standard deviation values range from 5 to 15 percent. The mean and standard deviation for scenarios 30 and 31 are approximately 15 and 45 percent, respectively. Scenarios 40, 41, 42, and 43 yield mean values that range from minus 5 to plus 45 percent and standard deviations that range from 55 to 70 percent. The remaining scenarios (44, 45, 46, and 47) have mean values that are approximately minus 20 percent with corresponding standard deviations of 15 to 20 percent. Figure 21. Bar chart. Statistics of error component A in life predictions (percent), rigid pavement sites with average annual daily truck traffic greater than 1,200 trucks per day per lane. This bar chart shows the mean and standard deviation of error component A in life predictions for rigid pavement sites with average annual daily truck traffic greater than 1,200 trucks per day per lane. For scenarios 11, 12, 20, 21, 22, and 23, the mean values range from minus 5 to plus 5 percent, while the standard deviation values range from 5 to 25 percent. The mean and standard deviation for scenarios 30 and 31 are approximately minus 10 and plus 25 percent, respectively. Scenarios 40, 41, 42, and 43 yield mean values that range from minus 5 to plus 5 percent and standard deviations that are around 25 percent. The remaining scenarios (44, 45, 46, and 47) have mean values that are approximately 25 to 30 percent, with corresponding standard deviations of 45 to 50 percent. Figure 22. Bar chart. Estimated range in National Cooperative Highway Research Program 137A design guide pavement life prediction errors from mean traffic input. This bar chart shows the range in estimated pavement design guide error by scenario type and confidence level under mean traffic output. In other words, this is equal to 1 minus the probability of exceeding this error. This bar chart is threedimensional. Scenarios are on the x axis, range in mean error is on the y axis, and reliability level is on the z axis. For a 75 percent reliability, errors range from 5 point 3 percent for scenario 11 to 32 point 5 percent for scenario 47. For 85 percent reliability, the errors range from 8 point 3 percent to 50 point 4 percent, respectively. For 95 percent reliability, the errors range from 13 point 4 percent to 82 percent, respectively. Figure 23. Bar chart. Estimated range in National Cooperative Highway Research Program 137A design guide pavement life prediction errors from lowpercentile traffic input. This bar chart is threedimensional. Scenarios are on the x axis, range in mean error is on the y axis, and reliability level is on the z axis. For 75 percent reliability, errors range from 21 percent to 83 point 8 percent for scenario 11 to 32 point 5 percent for scenario 47. For 85 percent reliability, the errors range from 27 point 4 percent to 139 point 2 percent, respectively. For 95 percent reliability, they range from 41 point 6 percent to 206 point 8 percent, respectively. Figure 24. Cluster tree chart. Clusters of LTPP sites by annual tandemaxle load distribution, Washington. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Washington State for frequency distribution of tandemaxle loads. The degree of similarity is indicated by a low number of the Euclidean distance. There are 17 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 15. Figure 25. Cluster tree chart. Clusters of LTPP sites by annual tandemaxle load distribution, Vermont. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Vermont for frequency distribution of tandemaxle loads. The degree of similarity is indicated by a low number of the Euclidean distance. There are 5 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 0095. Figure 26. Cluster tree chart. Clusters of LTPP sites by annual tandemaxle load distribution, Mississippi. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Mississippi for frequency distribution of tandemaxle loads. The degree of similarity is indicated by a low number of the Euclidean distance. There are 22 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 15. Figure 27. Cluster tree chart. Clusters of LTPP sites by annual tandemaxle load distribution, Minnesota. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Minnesota for frequency distribution of tandemaxle loads. The degree of similarity is indicated by a low number of the Euclidean distance. There are 18 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 23. Figure 28. Cluster tree chart. Clusters of LTPP sites by annual tandemaxle load distribution, Michigan. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Michigan for frequency distribution of tandemaxle loads. The degree of similarity is indicated by a low number of the Euclidean distance. There are 11 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 035. Figure 29. Cluster tree chart. Clusters of LTPP sites by annual tandemaxle load distribution, Indiana. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Indiana for frequency distribution of tandemaxle loads. The degree of similarity is indicated by a low number of the Euclidean distance. There are 14 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 11. Figure 30. Cluster tree chart. Clusters of LTPP sites by annual tandemaxle load distribution, Connecticut. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Connecticut for frequency distribution of tandemaxle loads. The degree of similarity is indicated by a low number of the Euclidean distance. There are 4 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 01. Figure 31. Cluster tree chart. Clusters of LTPP sites by annual average truck class distribution, Washington State. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Washington State for truck class distribution. The degree of similarity is indicated by a low number of the Euclidean distance. There are 17 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 78. Figure 32. Cluster tree chart. Clusters of LTPP sites by annual average truck class distribution, Vermont. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Vermont for truck class distribution. The degree of similarity is indicated by a low number of the Euclidean distance. There are 5 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 18. Figure 33. Cluster tree chart. Clusters of LTPP sites by annual average truck class distribution, Mississippi. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Mississippi for truck class distribution. The degree of similarity is indicated by a low number of the Euclidean distance. There are 22 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 2 point 9. Figure 34. Cluster tree chart. Clusters of LTPP sites by annual average truck class distribution, Minnesota. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Minnesota for truck class distribution. The degree of similarity is indicated by a low number of the Euclidean distance. There are 18 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 1 point 5. Figure 35. Cluster tree chart. Clusters of LTPP sites by annual average truck class distribution, Michigan. This clustering tree that establishes the level of similarity between extendedcoverage WIM LTPP sites in Michigan for truck class distribution. The degree of similarity is indicated by a low number of the Euclidean distance. There are 11 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 1 point 1. Figure 36. Cluster tree chart. Clusters of LTPP sites by annual average truck class distribution, Indiana. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Indiana for truck class distribution. The degree of similarity is indicated by a low number of the Euclidean distance. There are 14 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 1 point 6. Figure 37. Cluster tree chart. Clusters of LTPP sites by annual average truck class distribution, Connecticut. This clustering tree chart establishes the level of similarity between extendedcoverage WIM LTPP sites in Connecticut for truck class distribution. The degree of similarity is indicated by a low number of the Euclidean distance. There are 4 LTPP sites shown in the clustering tree. The Euclidean values range from 0 to 0 point 6. FHWAHRT05079
