R56-10

TS-5a R 56-1 AASHTO

Standard Practice for

Certification of Inertial Profiling Systems

AASHTO Designation: R 56-101

1. SCOPE

1.1 This practice describes a certification procedure for test equipment used to measure a longitudinal surface elevation profile of highways based on an inertial reference system that is mounted on a host vehicle. The minimum requirements stipulated herein are intended to focus on the need for accurate and repeatable profile measurements during construction. It is also applicable to certifying the performance of profiler systems utilized in network data collection.

1.2 This practice describes minimum performance requirements for inertial profiling systems to be used for quality control/quality assurance (QC/QA) of surface roughness on Owner-Agency paving projects where a profile-based roughness construction specification is applicable. The same approach should be utilized for network data collection systems; however, different performance criteria may be employed.

1.3 The filtering methods and threshold values recommended in this document were established to ensure adequate determination of the International Roughness Index (IRI) and the profile features that affect it. An Owner-Agency planning to use an index other than the IRI should filter the profile and set thresholds accordingly. In particular, replace the preprocessor filter in Step 3, Appendix X.1 with a filter that corresponds to the relevant waveband for the alternative index.

1.4 If any part of this practice is in conflict with referenced documents, such as ASTM Standards, this practice takes precedence for its purposes.

1.5 This practice does not purport to address all of the safety concerns, if any, associated with its use. It is the responsibility of the user of this practice to establish appropriate safety and health practices and determine the applicability of regulatory limitations related to and prior to its use.

2. REFERENCED DOCUMENTS

2.1 AASHTO Standards: M 328, Inertial Profiler R 57, Operating Inertial Profiling Systems R 40, Measuring Pavement Profile Using a Rod and Level

2.2 ASTM Standards: E 867, Standard Terminology Relating to Vehicle-Pavement Systems E 1926, Standard Practice for Computing International Roughness Index of Roads from

Longitudinal Profile Measurements

© 2010 by the American Association of State Highway and Transportation Officials.All rights reserved. Duplication is a violation of applicable law.

Copyright American Association of State Highway and Transportation Officials Provided by IHS under license with AASHTO Licensee=PennDOT/5958818001

Not for Resale, 02/10/2011 10:53:30 MSTNo reproduction or networking permitted without license from IHS

--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

TS-5a R 56-2 AASHTO

2.3 Other: Karamihas, S. M. Development of Cross Correlation for Objective Comparison of Profiles.

International Journal of Vehicle Design, Vol. 36, Nos. 2/3 (2004), pp. 173–193.

3. TERMINOLOGY

3.1 Definitions:

3.1.1 high-pass filtering—reduces the effect of long wavelengths that are associated with gradual elevation changes such as hills.

3.1.2 International Roughness Index (IRI)—a statistic used to determine the amount of roughness in a measured longitudinal profile. The IRI is computed from a single longitudinal profile using a quarter-car simulation as described in the paper “On the Calculation of International Roughness Index from Longitudinal Road Profile” (Sayers 1995). Computer programs to calculate the IRI statistic from a longitudinal profile are referenced in ASTM E 1926.

3.1.3 longitudinal profile—a two-dimensional slice of the roadway surface taken along the travel direction, such as along the wheel path. It represents the perpendicular deviations of the pavement surface from an established reference parallel to the horizontal.

3.1.4 report interval—the longitudinal distance between the outputs of a profile index value.

3.1.5 sample interval—the longitudinal distance between data capture points. The data include location, height, and accelerometer values. These data points are combined to create one profile data point. These points, in turn, may be combined to create a final value in the reported profile.

4. SIGNIFICANCE AND USE

4.1 This practice outlines standard procedures for certifying and verifying the operational accuracy and repeatability of inertial profiling equipment. It also provides guidance for qualifying the equipment operators.

5. EQUIPMENT

5.1 Minimum Requirements—The inertial profiling system must meet all requirements and specifications found in M 328.

5.2 Operating Parameters—The inertial profiler must be capable of providing relative elevation measurements that meet the following requirements:

5.2.1 Report Interval—The interval at which relative profile elevations are reported must be less than or equal to 2 in.

5.2.2 High-Pass Filter—The algorithm for high-pass filtering the profile data shall have an undistorted response (profile amplitude and location error of less than 5 percent) for all wavelengths shorter than 150 ft when operated between 20 and 70 mph for high-speed profilers and 15 and 25 mph for low-speed profilers. It shall also have a minimum of a 30 percent reduction in profile amplitude for wavelengths longer than 300 ft. The filter will have a minimum of a 70 percent reduction for wavelengths longer than 450 ft.




--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

TS-5a R 56-3 AASHTO

5.3 The profiler software must also be able to calculate and report the IRI (in inches per mile and appropriate SI units) from the corresponding measured true profile and permit the operator to:

Automatically trigger the start of data collection at the designated location; Provide the measured true profiles in electronic text files following the formats prescribed by

R 57, in addition to any binary data file storage provided. These profile data are necessary to evaluate profiler accuracy and repeatability and to verify the height and distance measurements as described herein.

6. EQUIPMENT CALIBRATION VERIFICATION

6.1 General—This section refers to two important calibration and verification procedures: the block test and the bounce test. It is highly recommended that operators perform these procedures and calibrate their distance measurement instrument (DMI) immediately prior to certification and on a regular basis during normal operation as recommended by the manufacturer. Operators should adhere to the procedures specified in R 57 for calibrating and checking their equipment. The certifying agency shall provide adequate opportunity and facilities to perform these procedures, but it is the responsibility of the profiler operator to provide the needed equipment (e.g., calibration blocks, etc.).

6.1.1 Calibration Verification Log—Maintain a log that is to be kept with the inertial profiler to provide a verification of calibration history. The results of the routine bounce tests and verification runs should also be included in this log. The log should contain a record any repairs, replacement of components, and changes in native software versions. If the log is electronic, a backup copy should be kept in a secure location.

7. OPERATOR CERTIFICATION

7.1 Operators of inertial profilers used for QA testing of pavement ride quality must pass a proficiency test and be certified to operate an inertial profiler. Applicants for certification will be tested on the following:

Owner-Agency ride smoothness specification, Operating inertial profilers, Collection of profile data, and Evaluating the quality of the data collected and the IRI value calculated.

Applicants for certification shall undergo both written and practical examinations. They must pass both portions of the test to be certified. Prior to taking the proficiency test, applicants should have completed a profile training course such as NHI Course 131100 and must have undergone training on the use of the specific inertial profiler they will be operating in the field. Applicants must know how to perform vertical and horizontal calibrations and collect profile data with the inertial profiler. Applicants must bring the inertial profiler with them at the time of testing. Upon passing the proficiency test, successful applicants will be given documentation that will verify that they are certified to operate the inertial profiler for QA testing on paving projects. The documentation shall identify the specific type or brand of inertial profiler the operator is certified to operate. The initial duration and requirements for renewal of this certification shall be set by the Owner-Agency. Renewal should be required at least every 3 years.

8. EQUIPMENT CERTIFICATION

8.1 Certification frequency shall be as specified by the Owner-Agency. The inertial profiler must successfully perform and pass certification tests to establish compliance with the minimum




--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

TS-5a R 56-4 AASHTO

requirements for accuracy and repeatability set forth in this procedure. An inertial profiler shall be recertified after any major component repairs or replacements as identified in R 57.

8.2 Dynamic Certification Testing—Certification tests shall be conducted at a site approved by the Owner-Agency, and involve test sections, reference profiles, and devices used to establish reference profiles as described in the following:

8.2.1 Test Sections—Perform dynamic certification testing on sections over a range of roughness. The reference profiles on the smooth section shall have an average IRI within the range of 30 to 75 inch per mile while the corresponding reference profiles on the medium-smooth section shall have an IRI within the range of 95 to 135 inch per mile. If pavements with considerable distress are to be measured, as in network data collection, a third, medium-rough, site should be selected with roughness up to 200 inch per mile. The surface macro-texture of the test sections should reflect common Owner-Agency specifications for the pavement surfaces to be evaluated. It is imperative that all the surface types on which the profiler will be expected to collect data be included in this process since current height sensors remain sensitive to some surface features. Each test section will be at least 528 ft in length, with proper lead-in distance and a safe stopping distance available. For complete evaluation, the test section length should approach four times the length of the longest wavelengths being considered. Test sections should not include significant grade or grade change. Also, significant horizontal curvature or super-elevation should be avoided. If any lateral “g” forces or vehicle roll are sensed during testing at the highest speed employed, the course has too much curvature or single wheel-path elevation changes.

Note 1—Coarse surface textures, such as those found on open-graded surface mixes, chip-sealed, tined (transverse or longitudinal), or grooved pavement, for example, are very challenging to inertial profilers with dot, single point height sensors. Coarse texture is likely to adversely affect the repeatability and accuracy of the profile and the calculated index. Adherence to the requirements of this standard on smooth-textured surfaces does not guarantee acceptable performance of inertial profilers on coarse-textured surfaces. Certification testing on surfaces with the same texture type expected in the final application is required.

8.2.2 Reference Profiles—The reference profile for each test section shall be measured in accordance with R 40 except that the maximum horizontal measurement interval and reporting interval shall not be greater than 2.75 in. and the minimum vertical resolution shall not exceed 0.01 in. Other Owner-Agency approved survey techniques that produce third order or better survey accuracy may be used. Devices that measure and integrate differential elevations, such as the Dipstick® and Walking Profiler, may be used to establish the reference profiles using multiple staggered runs as required to meet the interval requirement. However, the measurements from these devices must be checked with the rod and level at distances along the test profile trace that are multiples of the reporting interval for these devices. The rod and level measurement locations shall be no more than 100 ft apart. Reference profile measurements shall be made on the designated profile trace of each test section as well as on the lead-in to the section. The measured lead-in distance shall be at least equal to the longest wavelength of interest (150 ft for the profiles used to collect IRI). The total lead-in should be at least two times the longest wavelength of interest, with four times being very desirable.

8.2.3 Ten repeat runs of the candidate inertial profiler shall be made on the designated profile trace of each test section in the prescribed direction of measurement. For high-speed profilers, make five runs at the maximum desired certification speed and five runs at the minimum certification speed. Data collection shall be automatically triggered at the starting location of each run of the section, and reported so that a longitudinal position of “zero” occurs at the starting location. An




--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

TS-5a R 56-5 AASHTO

automatically detected mark at the end of the section may be used to verify the DMI repeatability and accuracy.

8.3 Test Data—Profile data shall be reported in a format specified by the Owner-Agency.

8.3.1 During the certification tests, the same trace(s) is measured for all runs on a given test section. There will be twice as much data collected and analyzed when dual-path inertial profilers are tested. To facilitate the analysis of the data, the files from the tests described herein shall be named according to the following convention:

The first three characters of the file name are reserved for identifying the profiler tested. This identification will be established by the testing agency and given to the operator of the profiler on or before the day of testing.

The fourth character shall be “S” for runs made on the smooth section or “M” for runs on the medium-smooth section and “R” for runs made on the medium-rough section.

The fifth character shall be “L” or “H” for low- or high-speed runs. “G” is used for governed runs on lightweight profilers.

The sixth character shall designate the trace or wheel path tested. For dual-sensor profilers, the letter “B” shall be used to indicate that both wheel paths were profiled in the same run. For single-sensor profilers, the designation for the test trace will be given by the testing agency to the operator of the profiler on or before the day of testing.

The seventh and eighth characters shall designate the run number (01 to 10). The file extension shall be the one appropriate for the data type being transmitted and will be

specified by the agency.

8.3.1.1 The performance of the profiler is evaluated by analyzing the test data using cross correlation as described in the following sections to establish the repeatability and accuracy for the application of the device. The profiles are filtered to focus the evaluation on the profile features critical to the index being subsequently applied to the collected data. The available free software ProVAL can perform the calculations specified below. The method employed is described in the “Development of Cross Correlation for Objective Comparison of Profiles” by S. M. Karamihas, published in the International Journal of Vehicle Design, Vol. 36, Nos. 2/3 (2004), pp. 173–193.

8.3.1.2 Equipment Repeatability—Evaluate repeatability using cross correlation of the filtered output as described in Section 8.3.1.4. Calculate the repeatability agreement score of each trace. For single-sensor set profilers, one score for each path will be determined: one for the path in the smooth section, one for the medium-smooth section, and one for the medium-rough section as defined in Section 8.2.1. For dual-path profilers, twice as many scores will be determined, two for each test section. When the IRI is applied to the profile, the IRI filter should be used. On each trace, cross correlate each of the ten profiles to each of the remaining nine. (Since all of the profiles have the same reporting interval, 45 comparisons will be needed rather than 90 because the process has reciprocity.) The repeatability agreement score for each trace is the average of all 45 values. For IRI, experience has shown that on pavements with IRI values less than 150 inch per mile, an agreement score of 0.92 or greater is required on all traces to provide IRI values within 5 percent with a 95 percent confidence level. A lower agreement score may be acceptable for other applications such as network-level collection given the higher distress levels encountered. If an alternate index is applied, a filter appropriate to that index should be applied and suitable thresholds established based on independent analysis.

8.3.1.3 Equipment Accuracy—Evaluate accuracy using cross correlation of the appropriate filtered output as described in Section 8.3.1.4. On each trace, cross correlate each of the ten profiles to the reference profile. The accuracy agreement score for each trace is the average of the ten individual




--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

TS-5a R 56-6 AASHTO

cross correlation values. Based on the same rationale as in Section 8.3.1.2, a score of 0.90 or greater for applications using IRI should be required for passing for new construction quality control. A different value may be acceptable in applications utilizing other indexes.

8.3.1.4 Cross Correlation—Cross correlate profiles as described in Appendix X1. When cross correlating two profiles, the following processing steps are required:

If the reference profile contains grade information, it should first be removed by application of a high-pass filter set at least three times the longest wavelength of interest.

Apply the filter associated with the application to both traces before cross correlating them. For profiles used to determine the IRI, the IRI filter should be applied.

When comparing a profile from a candidate device to a reference profile, interpolate the candidate profile to the recording interval of the reference profile. (It is important that this step is performed after the filter is applied.)

Cross correlate the two profiles several times by shifting one profile over every possible offset up to 3 ft in either direction. When comparing a profile from a candidate device to the reference device, shift the candidate profile.

The cross correlation of the two profiles is the maximum (best) value found over the 6-ft range.

8.3.2 Verification of Computed Ride Statistics (IRI)—The test equipment software must be capable of computing and reporting the IRI of each profile trace tested. The performance of the calculation software is verified by comparing the calculated values with the values determined by a program of known accuracy. An example of such a program is ProVal. Each IRI value should be comparable to the value from the reference program with an error not greater than 2 percent. Other ride indexes may be verified in a similar fashion.

8.4 Distance Measurement Index Test—Test the accuracy of the DMI on one test section.

8.4.1 Distance Measurement Index Test Section—Provide a section for DMI testing. The test section will be at least 1000 ft in length, with proper lead-in distance and a safe stopping distance available. This test section may incorporate the test sections that are used for accuracy and repeatability testing. Clearly mark the starting and ending points of the test section. Measure the distance between the starting and ending points with a temperature-compensated steel tape, pulled taut but still following the pavement contour.

Note 2—The same runs may be used for verification of DMI accuracy as are used for testing accuracy and repeatability of the collected profile. The ideal test of DMI accuracy would be performed over a known but undisclosed length of pavement.

8.4.2 At least three auto-triggered runs at the lowest and highest test speeds of the candidate inertial profiler shall be made on the designated length of pavement in the prescribed direction of measurement. At the end of each run, record the reading from the profiler’s DMI. For high-speed profilers, this results in at least six values. Collection speed governed devices should make at least five runs.

8.4.3 Distance Measurement Index Accuracy—Compute the absolute difference between the DMI readings and the known distance of the path tested for each run. The average of the absolute differences for both the high-speed and low-speed runs, if applicable, must be less than 0.15 percent to pass the test.




--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

TS-5a R 56-7 AASHTO

8.5 Test Results—The results of the certification tests shall be documented by the testing agency. The distribution of the results of the certification shall be determined by the testing agency. Results of certification shall include the following information:

Identification of the profiler tested (i.e., make, model serial number, software version, Owner, etc.);

Date of last certification; Operator of the profiler; Name of the individual from the testing agency who conducted the test; Date of test; Number of paths the profiler can measure in the same run; Filter type, name of the filter program, and the applicable program version number used to

evaluate the profiler accuracy; Overall determination from the test: Pass or Fail; Known longitudinal distance of the DMI test section; and Average absolute difference between the DMI readings and the known distance, expressed in

distance unit and as a percentage of the known longitudinal distance.

The following information is to be provided for each trace: Overall repeatability score; Overall accuracy score; The average percent difference of the IRIs computed from the profiles and those from the

reference software.

8.6 The report will also label each test result with a Pass or Fail depending on whether the given test value meets or fails to meet the prescribed criterion. The profiler must pass all tests to be certified. A decal or other approved marking shall be placed on the profiler as evidence of certification. This decal shall show the expiration date (month and year) of the certification.

9. KEYWORDS

9.1 Inertial profiler; IRI; International Roughness Index; pavement ride quality; profiler certification.

10. REFERENCES

10.1 NHI Training Course 131100. Pavement Smoothness: Use of Inertial Profiler Measurements for Construction Quality Control. March 2002.

10.2 Sayers, M. W. On the Calculation of International Roughness Index from Longitudinal Road Profile. In Transportation Research Record 1501. Transportation Research Board, National Research Council, Washington, DC, 1995, pp. 1–12.

10.3 Sayers, M. W., T. D. Gillespie, and W. D. O. Paterson. Guidelines for Conducting and Calibrating Road Roughness Measurements. The World Bank Technical Paper, Number 46. The World Bank, Washington, DC, 1996.

10.4 Sayers, M. W., and S. M. Karamihas. The Little Book of Profiling. University of Michigan Transportation Institute, Ann Arbor, Michigan, 1998.




--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

TS-5a R 56-8 AASHTO

APPENDIX

(Nonmandatory Information)

X1. CROSS CORRELATION

X1.1. Cross correlate any pair of profiles using the following steps. When cross correlation is performed for repeatability analysis, either profile may be selected as the “reference.”

Step 1: Identify a reference profile (Q). It will also be considered the location reference. As such, the start of the section of interest shall occur at a longitudinal location of “zero,” with at least 150 ft of profile upstream of the section start and 5 ft of profile after the section end. The profile will have a recording interval Δx.

Step 2: Identify a segment of the correlated (or candidate) profile (p). Using auto-triggering, the profiler should place the section starting point at a longitudinal location of “zero,” with at least 150 ft of profile upstream of the section start.

Step 3: Preprocess the reference profile and the candidate profile by applying the IRI filter to them. This includes the 9.84-in. moving average, the conversion to slope, and the quarter- car simulation.2

Step 4: Crop the preprocessed reference profile to include only the section of interest. For a section length of Lq, a number of samples within the section of interest is Nq + (Lq/Δx + 1). Offset the cropped profile to have a mean of zero. Calculate the standard deviation of the preprocessed, cropped, and shifted reference profile (q), and the result is σq.

Step 5: Rewrite the preprocessed candidate profile (p) to the recording interval of the reference profile (q). Calculate the value at each location using linear interpolation of the surrounding points. Offset the resulting profile to have a mean of zero. The preprocessed, interpolated profile (p) will be cross correlated to the reference profile.

Step 6: Cross correlate the reference profile (q) to the candidate profile (p) several times, over a range of longitudinal offsets from a lead of 3 ft to a lag of 3 ft. For a sample interval of Δx, cross correlation will be performed 2M + 1 times, where M = int(3/Δx), where Δx is in feet.3 This produces a correlation function, ρm, where m is an integer subscript that ranges from –M to M:

( )

( )21ˆ

1 ˆmax ,

qN

m i m miqp

p q+=

ρ =σ σ

∑

Each increment of the subscript m represents a unique longitudinal offset of the candidate profile of mΔx. At each step (i.e., each increment of the subscript m), the candidate profile (p) must be conditioned by removing the mean to produce a vertically shifted profile ( )p̂ . Note also that the standard deviation of the candidate profile ( )p̂σ must be

recalculated at each step.

Step 7: Search the function ρm for its maximum cross correlation value ρmax.




--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

TS-5a R 56-9 AASHTO

Step 8: Calculate the adjustment factor for overall roughness as follows: f = min(σp, σq)/max(σp, σq)

Step 9: Calculate the “agreement score” for the two profiles: f•ρmax.

1 Formerly PP 49. First published as a full standard in 2010. 2 Application of the IRI filter is described in “On the Calculation of International Roughness Index from Longitudinal Road Profile” (Sayers 1995). Do NOT rectify the signal (i.e., do not take the absolute value of every point in the signal). 3 The “int()” function truncates the argument at the decimal point.




--`,,`,,`,````,,```,`,`,,`,,``,,-`-`,,`,,`,`,,`---

sunil.dhoundeyal

Line

Documents

R56-10