National Academies Press: OpenBook

Enhancement of the Practice for Certification of Inertial Profiling Systems (2023)

Chapter: Chapter 3 - Analysis of Certification Data

« Previous: Chapter 2 - Review of Existing Practices
Page 24
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 24
Page 25
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 25
Page 26
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 26
Page 27
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 27
Page 28
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 28
Page 29
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 29
Page 30
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 30
Page 31
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 31
Page 32
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 32
Page 33
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 33
Page 34
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 34
Page 35
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 35
Page 36
Suggested Citation:"Chapter 3 - Analysis of Certification Data." National Academies of Sciences, Engineering, and Medicine. 2023. Enhancement of the Practice for Certification of Inertial Profiling Systems. Washington, DC: The National Academies Press. doi: 10.17226/27182.
×
Page 36

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

24 Analysis of Certification Data Data Processing Data used for analysis were obtained from the following test events: 1. LTPP Profiler Rodeos conducted in 2010 and 2015; 2. FHWA Profiler Rodeo conducted in 2015; 3. MnDOT certification performed in 2021; 4. Mississippi DOT (MDOT) certifications performed in 2018, 2019, and 2021; 5. Tennessee DOT (TDOT) certification data from 2019; and 6. Florida DOT (FDOT) certifications performed in 2017, 2019, 2020, and 2021. These data were organized in a hierarchical folder structure to facilitate batch analysis to obtain detailed cross-correlation analysis results for repeatability, accuracy, and IRI values. The cross-correlation algorithm is documented in Appendix X1 of AASHTO R 56-14. The folder structure includes test events, equipment information, test conditions, and sections as the first, second, third, and fourth folder levels. Specific codes were designed to track machine brands, models, DMI types, height sensor types, pavement types, surface texture types, roughness levels, data collection speed, and filters applied. The Command-Line Interpreter (CLI) version of the ProVAL software was used to perform the batch analysis for the datasets. The analysis types include no sampling interval adjustment (“Non-Adjusted”) and sampling interval adjustment (“Adjusted”) for the “comparison profile” of a comparison pair. The adjustment sweep analysis was made within ±0.1 percent of the original sample interval with an increment at 0.01 percent to obtain the optimum cross-correlation. The results were then merged and organized by a Microsoft Excel macro program for subsequent statistical analysis. Routine LTPP data were considered for incorporation. The LTPP pavement performance database (PPDB) documents the surface material for each section. For concrete pavements, the database documents whether the portland cement concrete surface was tined or diamond ground; however, the direction of tining is not contained within the database. Since the surface texture information was inexact, these data were subsequently excluded from analyses. On the other hand, the LTPP Profiler Rodeos conducted in 2010 and 2015 include detailed information on the surface texture, including direction of tining of the concrete surfaces; therefore, these data were included in the analyses. The SurPRO walking profiler was the reference device used with each of the datasets evaluated. Of the nearly 4,200 records for the accuracy analysis, approximately 50 percent have reference measurements performed the same day as the inertial profiler measurements. One set of records has a time difference with the reference data measured over 300 days before the inertial profiler tests. Another 80 records were collected within 1 week, and 388 records were collected within 2 weeks of the inertial profiler collection. The correlation between the cross-correlation and C H A P T E R   3

Analysis of Certification Data 25   time between the reference and profiler measurements is very low (approximately 3 percent), indi- cating that the time difference between collection of reference profile data and inertial profiler data should not impact the results. However, the records with reference data collected on a dif- ferent day from the inertial profiler collection were from asphalt pavements, and would likely have been different if on jointed concrete pavement sections. Seven vendors were represented within the data used. Sensor footprints included line lasers, spot lasers, wide spot lasers, and Laser Cracking Measuring System (LCMS). Table 4 provides the number of datasets available by vendor, sensor footprint, pavement type, and surface texture. A few profilers and sensor types were not definitively identifiable from the data pro- vided and were therefore designated as “unknown.” In the analyses where sensor type was con- sidered a factor, the unknown records were excluded. Speeds of data collection ranged from 25 to 60 mph. Test events include asphalt and concrete pavements, but not every event includes all surface types. For example, the FDOT and MDOT certification practices do not include concrete pave- ments. Asphalt pavements included dense-graded and open-graded surfaces. The open-graded surfaces included both negative and positive texture types. Concrete pavement sections included diamond-ground, longitudinally tined, and transversely tined surfaces. Figure 9 provides a histogram summarizing the IRI values from the dataset. Figure 10 provides a histogram of cross-correlation values within the dataset. These two figures demonstrate that the dataset contains a broad range of data for use in the analysis. As noted in Table 4, the sensor type was unknown for some of the profile data collected. Table 5 provides a comparison of the basic statistics for cross-correlation, IRI, and percent difference in IRI by height sensor type. The data suggest that the unknown height sensors appear to most closely match the distributions for the line lasers. Analyses conducted to evaluate the impact of height sensor type excluded the data with an unknown sensor type. Other analyses incorporated these data. Vendor Sensor Footprint Asphalt Pavement Concrete Pavement Dense- Graded Open- Graded Diamond Ground Long. Tining Trans. Tining Vendor 1 Line Laser 64 0 0 0 0 Spot Laser 28 8 0 0 16 Unknown 36 16 8 8 8 Vendor 2 Spot Laser 6 4 2 2 2 Vendor 3 Line Laser 12 0 0 0 0 Vendor 4 Unknown 6 4 2 2 2 Vendor 5 Line Laser 14 4 2 2 2 Spot Laser 16 8 0 0 24 Wide Spot 6 4 2 2 2 Unknown 128 128 0 0 0 Vendor 6 LCMS 6 4 2 2 2 Unknown 6 4 2 2 2 Vendor 7 Line Laser 6 4 4 4 4 Unknown 2 0 0 0 0 Vendor 8 Line Laser 24 0 0 0 0 Unknown 20 8 4 4 4 Unknown Line Laser 8 0 0 0 0 Spot Laser 8 0 0 0 0 Unknown 48 0 0 0 0 Table 4. Datasets available by profile vendor, sensor footprint, and pavement.

26 Enhancement of the Practice for Certification of Inertial Profiling Systems N o. o f O bs er va tio ns Figure 9. Histogram of IRI values. N o. o f O bs er va tio ns Cross Correlation, % Figure 10. Histogram of cross-correlation values.

Analysis of Certification Data 27   Analysis The following analyses were performed to identify recommended revisions to AASHTO R 56-14: • Comparison of adjusted and non-adjusted data; • Impact of other elements on cross-correlation including: – Pavement type, – Surface texture, and – Height sensor. • Impact of roughness on cross-correlation; • Cross-correlation versus difference in IRI; and • Analysis of reference device data. The results from these analyses are documented in the following sections. Comparison of Adjusted Data and Non-Adjusted Data AASHTO R 56-14 requires the DMI error to be less than 0.15 percent (Section 8.4.3) to pass the DMI test. It may be possible to adjust the sample interval to maximize the cross-correlation within the 0.15 percent DMI error. The impact of adjusting the sampling interval was evalu- ated, allowing for a maximum adjustment of up to 0.1 percent, which is less than the allowed DMI error. This analysis considers the impact of the “adjusted” sample interval compared with “non-adjusted.” A paired t-test was used to compare the resulting mean cross-correlations between the two datasets of adjusted and non-adjusted data. A paired t-test compares the difference in two vari- ables for the same subject. In this case, the cross-correlation was first computed for each pair of runs without adjusting the sampling interval and then a second time allowing for adjustment in the sampling interval. Table 6 and Table 7 show the results for comparing the values represent- ing accuracy and repeatability, respectively. The results demonstrate an improvement in the Sensor Type Mean Standard Deviation Cross-Correlation Line 95% 5.0% Spot 92% 6.0% Unknown 95% 4.6% Percent Difference in IRI Line 2.0% 2.8% Spot 2.2% 2.3% Unknown 1.9% 2.2% IRI, in./mile Line 87 29 Spot 103 53 Unknown 74 24 Table 5. Basic statistics for dataset by sensor type. Cross-Correlation, % IRI, in./mile Non-Adjusted Adjusted Non-Adjusted Adjusted Mean 88.3 90.3 83.21 83.19 Variance 101 74 1368 1364 Observations 4131 4131 4131 4131 t Stat -25.032 7.196 p-value 0 0 Table 6. Comparison of accuracy results.

28 Enhancement of the Practice for Certification of Inertial Profiling Systems cross-correlation for accuracy with a minimal change in the IRI values on average. Additionally, the variance of the cross-correlation for accuracy is reduced when allowing the adjustment to the sampling interval. This analysis approach has minimal impact on the repeatability results, as may be observed from Table 7. In addition to the paired t-test, a review was conducted by surface texture and laser type. Table 8 shows the mean difference in cross-correlation for each surface texture to compare the accuracy cross-correlations and repeatability cross-correlations. Table 9 provides a comparison of the difference in cross-correlation between non-adjusted and adjusted data by sensor type. Consistent with the results from Table 6 and Table 7, the results indicate that the accuracy impacted by the adjustment may be significant while the repeatability is not greatly impacted. Impact of Data Items on Cross-Correlation In the remaining analyses, the adjusted data were used, and the relationships between cross- correlation and various data items were reviewed. Pavement Type The first element considered was the pavement type. Data were available from both asphalt and concrete pavement sections. Table 10 provides a comparison of the cross-correlation, absolute value of percent difference IRI, and IRI by pavement type, and also provides the results of a t-test comparing the statistics between pavement types. A t-test is used to compare two distri- butions with a p-value representing the probability that the distributions are not the same and Cross-Correlation, % IRI, in./mile Non-Adjusted Adjusted Non-Adjusted Adjusted Mean 94.5 94.6 83.7 83.7 Variance 49.3 46.5 1240 1239 Observations 17403 17403 17403 17403 t Stat -21.226 8.100 p-value 0 0 Table 7. Comparison of repeatability results. Mean Cross-Correlation Difference–Accuracy, % Mean Cross-Correlation Difference–Repeatability, % Dense-Graded 1.84 0.19 Open-Graded 2.63 0.16 Longitudinally Textured 0.94 0.13 Transversely Tined 1.17 0.15 Table 8. Comparison of the mean difference in cross-correlation value between adjusted and non-adjusted data by surface texture. Mean Cross-Correlation Difference–Accuracy, % Mean Cross-Correlation Difference–Repeatability, % Line Laser 1.19 0.12 Spot Laser 2.57 0.28 Table 9. Comparison of the mean difference in cross-correlation value between adjusted and non-adjusted data by sensor type.

Analysis of Certification Data 29   identified as a statistically significant difference. Table 10 demonstrates that the cross-correlation achieved between the two pavement types is very similar, and suggests that a similar result may be achieved on average regardless of pavement type. Table 10 indicates that the cross-correlation observed on asphalt pavements is not the same as that observed on concrete pavements. The comparison between the IRI values is observed to be statistically significant; however, the mean and standard deviation values suggest that the difference in the sections is not important. The percent difference in IRI is not statistically significant. Surface Texture Table 11 provides a comparison of the cross-correlation, the absolute value of the percent difference in IRI, and IRI by texture type for asphalt pavements. In addition to these statistics, the table provides the results of the t-test comparing the distributions. In each case, the test indicates that the distributions of data observed on dense-graded pavements are statistically different from those obtained on open-graded pavements. Table 12 provides a similar comparison for the concrete pavements. These textures were grouped based on longitudinal orientation (tining and diamond grinding) and transverse orientation Cross-Correlation, % Absolute Value of IRI Difference, % IRI, in./mile Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Asphalt 94 6.7 2.3 5.3 83 38 Concrete 93 8.9 2.2 3.3 86 28 p-value 0 0.74 0 Statistically Significant? Yes No Yes Cross-Correlation, % Absolute Value of IRI Difference, % IRI, in./mile Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Dense 95 9.8 1.6 1.3 87 9.3 Open 91 9.5 3.7 1.9 73 8.5 p-value 0 0 0 Statistically Significant? Yes Yes Yes Cross-Correlation, % Absolute Value of IRI Difference, % IRI, in./mile Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Longitudinal 94 8.7 2.3 3.1 76 11 Transverse 93 9.1 2.2 3.6 94 35 p-value 0 0.09 0 Statistically Significant? Yes No Yes Table 10. Comparison by pavement type. Table 11. Comparison by surface texture for asphalt pavements. Table 12. Comparison by surface texture for concrete pavements.

30 Enhancement of the Practice for Certification of Inertial Profiling Systems (tining only). These comparisons indicate that the cross-correlation and IRI were statistically different between the two groups, while the percent difference in IRI was not. Height Sensor Table 13 provides a comparison of cross-correlation and percent difference in IRI by height sensor type. These distributions were observed to be statistically different. The line lasers were observed to collect lower values of IRI at a higher cross-correlation and lower percent differ- ence in IRI across the pavement and textures contained in the dataset. Table 14 provides a comparison of the mean and standard deviation for the cross-correlation, percent difference in IRI, and the IRI by surface texture and sensor type. While Table 13 indicates that the differences in these statistics are significant across all surface textures, Table 14 illustrates that these differences are not consistent between surface types. The open-graded asphalt pave- ments and longitudinally oriented texture on concrete pavements demonstrate a much larger difference than that observed on dense-graded asphalt pavements and transversely oriented texture on concrete pavements. The lowest cross-correlations and largest differences in IRI were observed on the longitudinal texture of concrete pavements, suggesting that spot lasers may not be appropriate for use in collecting data on these types of pavements. The data for the dense-graded and open-graded textures were further subdivided based on whether the data reflect the accuracy or the repeatability of the equipment. These values are presented in Table 15 and Table 16. These tables demonstrate that the cross-correlations are lower for accuracy than those for repeatability. The difference observed between the two sensor types for the open-graded textured surface is larger for the repeatability of the equipment than Cross-Correlation, % Absolute Value of IRI Difference, % IRI, in./mile Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Line 95 5.6 2.0 2.9 87 29 Spot 90 10.2 2.6 3.1 102 52 p-value 0 0 0 Statistically Significant? Yes Yes Yes Surface Texture Sensor Type Cross-Correlation, % Absolute Value of IRI Difference, % IRI, in./mile Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Dense Line 97 3.3 1.3 1.8 94 35 Spot 93 6.7 1.9 2.2 108 69 Open Line 90 7.0 3.6 3.6 73 10 Spot 87 8.4 4.2 3.7 83 17 Longitudinal Line 95 6.0 2.5 4.4 83 11 Spot 72 14.9 5.2 4.6 78 14 Transverse Line 94 5.5 1.6 1.5 73 10 Spot 91 10.7 2.4 2.9 108 40 Table 13. Comparison by sensor type. Table 14. Comparison by texture and sensor type.

Analysis of Certification Data 31   for the accuracy of the equipment. This difference suggests that, on average, the spot laser foot- prints are collecting data as expected; however, more variability is associated with the spot sensor footprint than the line sensor footprint. There are not sufficient data to perform a similar com- parison for the concrete pavement textures. Impact of Roughness on Cross-Correlation In considering the various aspects of AASHTO R 56-14, the lower end of the IRI was consid- ered in order to identify whether there is a level of IRI that makes achieving repeatability imprac- tical. Figure 11 provides a graph showing the relationship between cross-correlation and IRI. The minimum IRI observed on the test sections included in the analysis was 35 in./mile. The graph shows that minimum levels of cross-correlation are lower at lower levels of IRI. Surface Texture Sensor Type Cross-Correlation, % Absolute Value of IRI Difference, % Mean Standard Deviation Mean Standard Deviation Dense Line 95 4.3 2.2 2.4 Spot 91 7.5 2.4 2.2 Open Line 85 7.8 5.4 4.5 Spot 83 9.5 4.5 4.0 Surface Texture Sensor Type Cross-Correlation, % Absolute Value of IRI Difference, % Mean Standard Deviation Mean Standard Deviation Dense Line 98 2.1 1.0 1.3 Spot 94 6.2 1.7 2.1 Open Line 92 5.6 2.9 3.0 Spot 87 7.9 4.1 3.7 Table 15. Comparison by texture and sensor type for accuracy. Table 16. Comparison by texture and sensor type for repeatability. Figure 11. Comparison of cross-correlation and IRI.

32 Enhancement of the Practice for Certification of Inertial Profiling Systems The relationship observed in Figure 11 does not indicate a strong relationship. While the data suggest that a minimum value may be needed, it does not provide an indication of what that minimum value should be. Cross-Correlation vs. Difference in IRI This analysis investigated the relationship between the cross-correlation and percent difference in IRI. These analyses examined the impacts of multiple factors, focusing on the impact of site factors on these relationships. A quantile linear regression was used for the analysis to evaluate the line associated with the 95th percentile. In other words, the line represented in these analyses is such that the percent difference has a 95 percent probability of being the same or less than the value provided by the line. Given the differences observed for various factors, the analyses started with a review of the surface texture. Table 17 shows the analysis results considering surface texture on con- crete pavements. The factor is considered statistically significant when the probability is less than 0.05. Table 17 indicates that the transverse and longitudinal orientations of texture on concrete pavements are statistically significant. The coefficients indicate that the percent difference in IRI is lower on transversely tined pavements than on concrete pavements with longitudinal texture. Concrete pavement sections were further examined incorporating the sensor footprint shown in Table 18. The results indicate that the spot laser is not significantly different from the line laser for concrete pavements. The lack of statistical significance does not mean that the sensor footprint does not impact the level of cross-correlation achieved. Instead, it indicates that the relationship between cross-correlation and the percent difference in IRI is the same between the two sensor footprints considered. The next analysis considered the impact of roughness. The sensor type was removed since it was identified as not statistically significant. These results are presented in Table 19. Value Standard Error t value Probability (>|t|) Intercept 101.320 1.769 57.260 0.000 Transverse -43.422 2.635 -16.480 0.000 Correlation -1.013 0.018 -56.152 0.000 Correlation—Transverse 0.438 0.027 16.203 0.000 Value Standard Error t value Probability (>|t|) Intercept 102.821 1.825 56.351 0.000 Transverse -44.016 3.408 -12.914 0.000 Spot -0.723 3.480 -0.208 0.835 Correlation -1.028 0.019 -55.305 0.000 Correlation—Spot -0.061 0.040 -1.530 0.126 Correlation—Transverse 0.443 0.035 12.803 0.000 Correlation—Transverse—Spot 0.068 0.005 12.469 0.000 Table 17. Results of correlation vs. percent difference in IRI, accounting for surface texture on concrete pavements. Table 18. Results of correlation vs. percent difference in IRI, accounting for surface texture and sensor footprint for concrete pavements.

Analysis of Certification Data 33   The equation for relating the cross-correlation to the percent difference in IRI is: • •DIRI 109.4 40.0 TT 0.12(IRI) 1.1 CC 0.4 TT CC 0.001 IRI CC= - - - + +` ` ` `j j j j (1) where DIRI = percent difference in IRI values TT = 1 for transversely textured pavement surfaces and 0 for a longitudinal orientation IRI = International Roughness Index in in./mile CC = cross-correlation in percent Table 20 was developed to provide an example of the use of this equation. The results indi- cate that there is some minor improvement in the percent difference in IRI at higher levels of roughness. Additionally, a lower percent difference in IRI may be observed at the same level of cross-correlation on transversely textured pavements over longitudinally textured concrete pavements. Equation 1 indicates that the 94 percent cross-correlation results in an approximately 5 percent difference in IRI on smooth concrete pavements with a longitudinally oriented surface texture. With increasing levels of roughness, the percent difference in IRI is expected to be lower at this same level of cross-correlation. The negative values in Table 20 are a result of the math- ematical relationship developed; in reality, it is expected that the 100 percent cross-correlation would result in a 0 percent difference in IRI. The analysis for asphalt pavements compared open-graded surface textures with dense-graded surface textures. The results from this analysis are provided in Table 21. The table indicates that the percent difference in IRI is lower for similar levels of cross-correlation on dense-graded pavements than on open-graded pavements. The next analysis incorporated the surface texture. Table 22 provides the results of this analysis. These trendlines are illustrated in Figure 12. Figure 12 shows that the line lasers are expected to Value Standard Error t value Probability (>|t|) Intercept 109.382 2.621 41.727 0.000 Transverse -40.034 2.624 -15.256 0.000 IRI -0.115 0.022 -5.098 0.000 Correlation -1.098 0.027 -40.886 0.000 Correlation—Transverse 0.403 0.027 15.004 0.000 Correlation—IRI 0.001 0.0002 5.175 0.000 Table 19. Results of correlation vs. percent difference in IRI, accounting for surface texture and roughness on concrete pavements. Cross- Correlation, % Absolute Value of Percent Difference in IRI, % Transverse Texture Longitudinal Texture IRI = 50 in./mile IRI = 100 in./mile IRI = 50 in./mile IRI = 100 in./mile 100 -1.6 -2.6 -1.6 -2.6 95 1.6 0.4 3.6 2.4 92 3.6 2.2 6.8 5.4 90 4.9 3.4 8.9 7.4 85 8.2 6.4 14.2 12.4 80 11.4 9.4 19.4 17.4 75 14.6 12.4 24.6 22.4 70 17.9 15.4 29.9 27.4 Table 20. Percent difference in IRI vs. cross-correlation for concrete pavements.

34 Enhancement of the Practice for Certification of Inertial Profiling Systems have a larger percent difference in IRI than the spot lasers for the same cross-correlation, and also shows that the dense-graded surface texture measured with a line sensor footprint generally has higher cross-correlation values than the other surface textures and sensor footprints. The final analysis incorporated roughness with the results, as presented in Table 23. The equation for asphalt pavements from Table 23 is: • • • DIRI 58.8 9.1 OG 17.9 SP 0.05 IRI 0.6 CC 0.2 CC SP 0.09 CC OG 0.0005 CC IRI = + - + - + - -` ` ` ` ` ` ` j j j j j j j (2) Value Standard Error t value Probability (>|t|) Intercept 56.985 1.606 35.488 0.000 Open 14.886 1.792 8.307 0.000 Spot -14.050 1.831 -7.673 0.000 Correlation -0.562 0.0164 -34.324 0.000 Correlation—Spot 0.144 0.0188 7.647 0.000 Correlation—Open -0.153 0.018 -8.339 0.000 Correlation—Open—Spot -0.012 0.002 -5.339 0.000 Figure 12. Percent difference in IRI vs. cross-correlation for asphalt pavements considering surface texture and sensor footprint. Table 22. Results of correlation vs. percent difference in IRI, accounting for texture type and sensor footprint on asphalt pavements. Value Standard Error t value Probability (>|t|) Intercept 52.735 0.827 63.744 0.000 Open 15.828 1.323 11.967 0.000 Correlation -0.519 0.009 -60.732 0.000 Correlation—Open -0.163 0.014 -11.851 0.000 Table 21. Results of correlation vs. percent difference in IRI, accounting for texture type on asphalt pavements.

Analysis of Certification Data 35   Value Standard Error t value Probability (>|t|) Intercept 58.794 2.190 26.848 0.000 Open 9.065 1.499 6.046 0.000 Spot -17.876 1.568 -11.189 0.000 IRI 0.050 0.022 2.288 0.022 Correlation -0.584 0.023 -25.853 0.000 Correlation—Spot 0.179 0.017 10.733 0.000 Correlation—Open -0.093 0.015 -6.052 0.000 Correlation—IRI -0.0005 0.0002 -2.100 0.036 Table 23. Results of correlation vs. percent difference in IRI, accounting for texture type, sensor footprint, and roughness on asphalt pavements. Cross- Correlation, % Absolute Value of Percent Difference in IRI, % Dense-Graded Open-Graded Spot Line Spot Line 100 0.4 0.4 0.2 0.2 95 2.6 3.4 2.8 3.7 92 3.9 5.3 4.4 5.8 90 4.7 6.5 5.4 7.2 85 6.9 9.5 8.0 10.7 80 9.0 12.6 10.6 14.2 75 11.2 15.6 13.3 17.7 70 13.3 18.7 15.9 21.2 Table 24. Percent difference in IRI vs. cross-correlation for asphalt pavements for IRI of 50 in./mile. Cross- Correlation, % Percent Difference in IRI, % Dense-Graded Open-Graded Spot Line Spot Line 100 0.4 0.4 0.2 0.2 95 2.7 3.6 2.9 3.8 92 4.1 5.5 4.6 6.0 90 5.0 6.7 5.7 7.4 85 7.2 9.9 8.4 11.1 80 9.5 13.1 11.1 14.7 75 11.8 16.2 13.9 18.3 70 14.1 19.4 16.6 22.0 Table 25. Percent difference in IRI vs. cross-correlation for asphalt pavements for IRI of 100 in./mile. where DIRI = percent difference in IRI values OG = 1 for an open-graded surface texture and 0 for a dense-graded surface texture SP = 1 for a spot sensor footprint and 0 for a line laser footprint IRI = International Roughness Index in in./mile CC = cross-correlation in percent Table 24 and Table 25 were developed to provide an example of the use of this equation. The two tables represent two levels of roughness demonstrating that for asphalt pavements, higher levels of roughness result in a higher expected percent difference in IRI for the same level of cross- correlation. These tables also show that the line sensor footprint may result in a higher percent

36 Enhancement of the Practice for Certification of Inertial Profiling Systems difference in IRI for the same level of cross-correlation. Similarly, an open-graded texture may also expect to result in a higher percent difference in IRI for the same level of cross-correlation. Equation 2 also indicates that for an open-graded section with a medium level of roughness measured with a line laser, an approximate 5 percent difference in IRI may be achieved at a 94 percent cross-correlation. Reference Device Analysis The last round of analyses conducted considered the reference device. Figure 13 illustrates how the cross-correlation varies by IRI for each surface texture represented in the dataset. There is not a consistent factor associated with the low cross-correlation values. These were collected from multiple events by multiple agencies, including the LTPP Rodeo from 2010 and two of the state DOTs. The average values by surface type are provided in Table 26. The average values suggest that there could be a relationship between the cross-correlation and the pavement type; however, this is confounded by the relationship between the cross-correlation and the IRI of the section. These values in combination with the figures above suggest that a roughness below 50 in./mile may be problematic for collecting reference data. Figure 13. Reference device cross-correlation by IRI. Pavement Surface Texture IRI, in./mile Cross-Correlation, % Asphalt Dense 69 95 Open 73 97 Concrete Longitudinal 76 97 Transverse 100 97 Table 26. Average IRI and cross-correlation from reference device by surface type.

Next: Chapter 4 - Proposed Revisions »
Enhancement of the Practice for Certification of Inertial Profiling Systems Get This Book
×
 Enhancement of the Practice for Certification of Inertial Profiling Systems
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Inertial profilers are used by state departments of transportation and others to produce an accurate and repeatable measure of the longitudinal pavement profile, which can be analyzed to produce various smoothness statistics such as the International Roughness Index.

NCHRP Research Report 1057: Enhancement of the Practice for Certification of Inertial Profiling Systems, from TRB's National Cooperative Highway Research Program, proposes revisions to AASHTO R 56-14 to enhance the practice for certification of inertial profiling systems.

READ FREE ONLINE

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!