Quantification of Smoothness Index Differences Related To Long-Term Pavement Performance Equipment Type

Chapter 3: Profiler Comparison Studies

INTRODUCTION

Studies have been conducted at regular intervals to compare the output from the four LTPP profilers. For each study, several test sections were laid out, and reference profile measurements along each wheelpath were obtained using Dipstick. Thereafter, profile measurements were performed on the test sections by the profilers. The primary method for checking if the profilers were functioning accurately was to compare the IRI values computed from Dipstick data with the values computed from the data obtained from the profilers. The repeatability of the profilers was evaluated by analyzing the standard deviations of the IRI, which were computed using the IRI values obtained from repeat measurements on a test section. In the profiler comparison studies that have been performed since 1998, in addition to comparing the IRI values, profiles obtained by the profilers were compared to evaluate profiler repeatability and reproducibility. The details of these comparison tests are presented later in this chapter.

Whenever FHWA purchased new profilers, the profilers underwent rigorous testing to ensure that they met the requirements that were specified in the contract documents. After each regional contractor took delivery of a new profiler, a comparison of the new profiler and the old profiler was performed in each region before collecting data with the new profiler. The purpose of these verification tests was to compare the output from the old and the new profilers. Details about these verification tests are presented later in this chapter.

In this research project, a literature review was also performed to gather information on other profiler comparison studies that have been performed in the past. The results of this literature review are presented later in this chapter.

LTPP PROFILER COMPARISON STUDIES

Overview

The following six LTPP profiler comparison studies have been held since the start of the LTPP program:

• 1990: Austin, TX.
• 1991: Ann Arbor, MI.
• 1992: Ames, IA.
• 1998: Urbana, IL.
• 2000: College Station, TX.
• 2003: Minnesota Road Research Project (Mn/ROAD) of the Minnesota Department of Transportation (Mn/DOT).

This section presents a brief description of the following activities related to an LTPP profiler comparison: (1) purpose of comparison test, (2) selection of test sections, (3) collection of reference elevation measurements, (4) profiler data collection, (5) computation of IRI values, and (6) analysis of data from LTPP comparison studies.

Purpose of Comparison Test

The purpose of performing a comparison test of the four LTPP profilers is to ensure that the profilers are collecting accurate, repeatable, and reproducible data. Currently, the following analyses are performed on the data collected during a comparison test:⁽⁶⁾

• Evaluation of the static accuracy of the profiler height sensors: Performed before data collection, this test checks the static accuracy of the height sensors using a set of blocks to determine whether the readings are within a specified tolerance.
• Evaluation of the results from the bounce test: The bounce test determines whether the height-sensor readings and accelerometer readings are canceling each other. The results of this test for each of the four profilers are compared to determine whether all four profilers are providing similar results.
• Evaluation of the accuracy of the DMI: Performed before data collection, this test determines whether the DMI meets specified bias and precision criteria. A 300-m- (984 ft-) long section is laid out to perform this test.
• Evaluation of the repeatability of the IRI values obtained by the profilers and a comparison of the IRI values obtained by profilers with those obtained using Dipstick: The IRI values obtained from the repeat runs of a profiler on a test section are used to evaluate the precision (repeatability) of a profiler. The IRI values are also used to evaluate the bias of a profiler with respect to Dipstick. For comparison studies that have been performed since 1998, the following precision and bias criteria have been evaluated: (1) determination of whether the precision of the IRI along a wheelpath (obtained by computing the standard deviations of the IRI for the repeat profiler runs) is less than 0.04 m/km (2.5 inches/mi), and (2) determination of whether the difference in IRI for a wheelpath between the average profiler IRI (which is calculated by averaging IRI obtained from the five profiler runs) and the Dipstick IRI is within ±0.16 m/km (±10 inches/mi).
• Evaluation of the repeatability of the profile data: The point-to-point repeatability for each profiler along each wheelpath is computed to evaluate the repeatability of the profile data.
• Comparison of the profile data obtained by the four profilers: One representative run for each profiler is selected for a test section and overlaid profile plots for each wheelpath at each test section are prepared to compare the data collected by the four profilers.

For comparison tests performed before 1998, an evaluation of the profile data was not performed; the analysis of the data was confined to IRI values.

Selection of Test Sections

The current LTPP procedures for performing a profiler comparison indicate that five test sections, which satisfy the following criteria, should be selected:⁽⁶⁾

• Section 1: Smooth Asphalt: Asphalt concrete (AC) pavement with an average IRI for the two wheelpaths of less than 1.6 m/km (101 inches/mi).
• Section 2: Rough Asphalt: AC pavement with an average IRI for the two wheelpaths of greater than 2.2 m/km (139 inches/mi).
• Section 3: Smooth Concrete: Jointed portland cement concrete (PCC) pavement with an average IRI for the two wheelpaths of less than 1.6 m/km (101 inches/mi).
• Section 4: Rough Concrete: Jointed PCC pavement with an average IRI for the two wheelpaths of greater than 2.2 m/km (139 inches/mi).
• Section 5: Chip-sealed section.

The comparison study performed in 1990 used six test sections, the studies in 1991 and 1992 used eight test sections, the study in 1998 used four test sections, and the studies in 2000 and 2003 used five test sections.

The following guidelines are specified for selecting test sections:⁽⁶⁾

• Each test section should be 152.4 m (500 ft) in length, with the beginning and end marked.
• The test sections should be located on flat, tangential sections that have sufficient length at each end to allow for acceleration to a constant speed before the section and safe deceleration past its end.
• The speed limit of the roadway containing the test sections should be at least 80 kilometers per hour (km/h) (50 miles per hour (mi/h)).
• The test sections should have a marked outside lane-edge stripe that can be used as a lateral reference when profiling the test section.
• The AC sections should not be concrete sections that have been overlaid with asphalt.
• Where possible, test sections should be located within a centralized locale, with short travel distances between each test section to reduce travel time.

Collection of Reference Elevation Measurements

Dipstick has been used to collect reference elevation data for all profiler comparison studies, except for the study in 2000. In the 2000 study, the reference profile measurements were obtained using an Australian Road Research Board (ARRB) walking profiler. In the 1992 comparison, rod-and-level measurements were obtained in addition to Dipstick measurements.

Dipstick measurements are performed along both wheelpaths at all test sections. The current procedures for performing Dipstick measurements are outlined in the LTPP Manual for Profile Measurements and Processing.⁽⁶⁾ As described in this document, Dipstick data collection from a test section is performed according to the following sequence:

1. Start data collection from the beginning of the section along the right wheelpath.
2. When the end of the section is reached, go across the lane toward the left wheelpath.
3. Perform measurements along the left wheelpath toward the beginning of the section.
4. After reaching the start of the section, go across the lane and terminate data collection at the right wheelpath.

It is not known whether this procedure for Dipstick data collection was followed when data were collected for the 1990 study. However, this procedure for Dipstick data collection was used for all other profiler comparison studies, except for the study performed in 1991 in Michigan. In the Michigan study, Dipstick measurements were first made along a wheelpath from the beginning of the section to the end of the section and, thereafter, measurements were made along the same path from the end of the section to the beginning of the section. This resulted in two profiles being available for each wheelpath. (Note: This was the Dipstick measurement procedure used in the LTPP program at that time.)

Dipstick measurements on PCC test sections were performed in the afternoon, at the same approximate time of day as expected for the collection of profiler data for all comparison studies. This procedure was followed to avoid the slab curling effects that may be present in the PCC pavements in the morning.

Profiler Data Collection

Current test procedures indicate that data collection at the test sections should be performed at a test speed of 80 km/h (50 mi/h).⁽⁶⁾ Data collection at PCC sections was performed in the afternoon, at the same approximate time of day as when Dipstick data were collected at the sections. At each test section, each profiler collects a set of measurements following the normal operating procedures that are used during data collection at LTPP sections.

During the 1990 profiler comparison, profile testing was performed at test speeds of 56 and 80 km/h (35 and 50 mi/h). In the profiler comparison studies that were performed in 1991 and 1992, profile testing was performed at test speeds of 64 and 80 km/h (40 and 50 mi/h). For all other comparison studies, profile testing was performed only at 80 km/h (50 mi/h).

The left wheelpath for all test sections was marked using paint dots for the profiler comparisons that were conducted in 1990 and 1991. In these studies, the profiler driver was asked to align the vehicle along the left wheelpath when profiling the test sections. In the 2000 comparison, the wheelpaths of two sections were marked with paint dots. In the 2003 comparison, both wheelpaths on all test sections were marked with paint dots. The location of the wheelpaths was not marked for the comparisons that were held in 1992 and 1998. In these studies, the profiler driver judged the location of the wheelpaths and aligned the profiler along the wheelpaths when profiling the test sections.

Computation of IRI Values

The computation of IRI values from the profiler data was performed by each region using the ProQual software.^(7,8) The number of runs that were used in the analysis for the different comparison studies was either five or six. Each RSC selected the profile runs that were to be used in the analysis from all available repeat runs. Each RSC prepared a table that included the left- and right-wheelpath IRI values for all selected runs and submitted the table and a brief report to FHWA and its technical support contractor.

Analysis of Data from LTPP Comparison Studies

The analysis of data obtained from the LTPP profiler comparison studies has been performed by the LTPP technical support contractor. Reports documenting the analyses and findings for all profiler comparison studies are available. (See references 10, 11, 12, 13, 14, and 15.) Summaries of the findings from each profiler comparison study are presented in appendix A.

LTPP PROFILER VERIFICATION STUDIES

Profiler equipment changes occurred in the LTPP program in 1996 and 2002. In 1996, each RSC replaced their K.J. Law Engineers DNC 690 profiler with a K.J. Law Engineers T-6600 profiler. In 2002, each RSC replaced their T-6600 profiler with an ICC MDR 4086L3 profiler. On each of these occasions, after accepting delivery of the new profiler, each RSC performed a comparison of the old and new profilers before collecting data with the new profiler. Details and findings from these two verification studies are presented in appendix B.

OTHER PROFILER COMPARISON/ANALYTICAL STUDIES

A literature review was performed to gather information on other studies that have been performed to compare IRI from profilers and reference devices. The purpose of obtaining information about other studies was to determine if they contained any explanations regarding the differences in IRI between inertial profilers and reference devices that could be useful for this research project. In addition, reports on other studies that have analyzed LTPP profile data were reviewed.

PIARC Comparison

In 1998, World Road Association (PIARC) Committee C1 on Surface Characteristics conducted a global experiment to investigate the performance of various high-speed profiling equipment.(16) Test sections were established in the United States, Japan, and Europe (The Netherlands and Germany). Reference profile measurements at the test sections were conducted using Dipstick and the ARRB walking profiler. The number of high-speed profilers that performed measurements in the United States, Japan, and Europe were 4, 7, and 25, respectively. The profilers that took measurements at the test sections in the United States generally showed good agreement in IRI. However, there was wide variability in the IRI values that were obtained by the different profilers at the test sections located in Japan and Europe. In addition, there were some large discrepancies in the IRI values obtained from the reference devices and the high-speed profilers at many sections. The causes of the differences in IRI were not investigated in this project.

Road Profiler User Group Comparisons: 1993 and 1994

Four regional test centers were used for these studies performed in 1993 and 1994-Pennsylvania, South Dakota, Nevada, and Mississippi.^(17,18) Eight test sections were used at each regional center, except for Nevada, where six test sections were used. Dipstick measurements were obtained at all test sections. LTPP‘s DNC 690 profilers were used in these two comparisons. An evaluation of the results indicated that at each of the regional test centers, LTPP‘s profilers were among the profilers that best matched the Dipstick IRI.

LTPP Profile Variability Analysis

This study visually reviewed all LTPP profile data that were collected between June 1989 and October 1997 for saturation spikes, lost lock, shifted start, wrong location, out of study, and other equipment- and operator-related problems.⁽¹⁹⁾ The data for the review were obtained from the LTPP database. The profile data for the period under review were collected by the DNC 690 profilers, except for a few data sets that were collected by the T-6600 profilers. The profiles that exhibited problems were divided into two groups-reparable and irreparable profiles. Reparable profiles included profiles with saturation spikes that were not marked. These data were reprocessed, and the IRI values in the LTPP database were updated. Data that exhibited problems that could not be repaired were deleted from the LTPP database.