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.
42 The project included three equipment comparison experiments. The first experiment evaluated the main available technologies in terms of repeatability and reproducibility. These were sub sequently verified and validated in the second and third experiments, respectively. 3.1 Initial Equipment Comparison The initial equipment comparison was conducted at the Virginia Smart Road in Blacksburg, Virginia. 3.1.1 Location and Procedures Stationary, walkingÂspeed, and highÂspeed macrotexture measuring equipment was used to make measurements along the 2.2Âmile, researchÂdedicated test track. The facility offers a variety of surfaces, including denseÂgraded HMA, OGFC, proprietary HFST, as well as grooved, tined, or ground PCC sections, as is shown in Figure 21. Figure 22 shows surface photos of each of the surfaces studied. All photos were taken from approximately a 3 ft. focal length. Each photo has been scaled to show relative textures. 3.1.2 Stationary and Walk-Along Devices The objective of the experiment was to assess the repeatability (ability to produce equal measure ments) and agreement (analysis of difference in measurements between devices) of several non contacting laser devices. These devices were both placed (static) and pushed (walkingÂspeed) along the pavement surface. These devices could be considered reference devices for network level macrotexture measurement devices. Stationary Devices The experiment was carried out at the Heavy Vehicle Simulator (HVS) and the test track at the Virginia Tech Transportation Institute (VTTI). These locations were used because the enclosed environment of the HVS shielded the measurement device from unwanted weather effects, and measurements could be made on the test track without the influence or danger of traffic. The HVS was not running and no vehicles were moving on the test track at the time of the experiment to ensure the testing apparatuses were unaffected by potential movement. Figure 23 shows the experiment setup within the HVS. The test track was marked with a chalk line for walkingÂspeed devices to follow and painted with boxed areas for static device placement (Figure 24). Walking speed devices collected data along the chalk line continuously and static devices collected a single profile in each of five boxes painted on each pavement section. C H A P T E R  3 Data Collection
Data Collection 43  Figure 21. Virginia Smart Road sections and surface types. Properties for the static devices can be found in Table 8. The repeatability of Device 6 and Device 12 was tested inside the HVS. The repeatability of device 8 was not determined due to unavailability at the time of testing. For Device 12, scans were made at maximum resolution, with a 0.00635 mm longitudinal (direction of traffic) sampling interval and a 0.0247 mm trans verse sampling interval. This resulted in a 3D pavement surface point cloud with 16,380 à 2,917 (over 47 million) texture height measurements. Five replicates were created by pressing the âscanâ button on the top of the machine five times while the device remained in place. Figure 25 shows an example of the surface profile measured.
44 Protocols for Network-Level Macrotexture Measurement Figure 22. Virginia Smart Road pavement surfaces.
Data Collection 45  (a) HVS site at VTTI (b) Alignment template in place Figure 23. Experiment location. Figure 24. Test track setup. Device ID Description Laser Type Measurement Patch Sample Distance Vertical Resolution 6 C.T. Meter Single-SpotLaser 300 mm (circular) 0.087 mm 0.003 mm 8 Ames Rapid Laser Texture Scanner 9500 Line Laser 101.6 mm L à 101.6 mm W 0.0496 mm L à 0.0415 mm W 0.01 mm 10 Ames Reference Beam Single-SpotLaser < 0.003 mm 12 Ames Rapid Laser Texture Scanner 9400HD Line Laser 107.95 mm L à 72.01 mm W 0.00635 mm L à 0.0247 mm W 0.003 mm Table 8. Static measurement device characteristics.
46 Protocols for Network-Level Macrotexture Measurement Pavement profiles were analyzed for their effect on vehicle/road interactions by representing key characteristics in pavement surfaces as parameters. The software package for Device 12 is capable of calculating a wide array of these parameters from a given measured profile (Table 9). For example, in the United States, a pavementâs MPD is typically reported for a given pavement. In this study, repeatability values for Device 12 are given for all the parameters to facilitate inter pretation of results by agencies that may wish to use them. Walk-Along Devices The characteristics of the walkingÂspeed devices used can be seen in Table 10. The test track (Figure 26) was used for the repeatability analysis of the walkingÂspeed devices and the limits of agreement (LOA) analysis of all static and walkingÂspeed devices. Representative surfaces of the test track were selected to provide a variety of texture types and orientations for the analysis. The pavement surfaces tested are listed in Table 11. MPD was selected as the parameter to represent the various pavement surface types tested for most of the repeatability and agreement analysis in this work. As noted in various standards for calculating MPD (ASTM E1845 [2015]; ISO 13473Â1 [2019]), outliers in the data must first be removed. This is especially important for MPD calculations, as they are sensitive to large data variations away from the mean. Figure 25. Example pavement surface profile from Device 12 Parameter Reference/Definition Mean Profile Depth (MPD) ASTM E1845 (2015), ISO 13473-1 (1997) Estimated Texture Depth (ETD) ASTM E1845 (2015), ISO 13473-1 (1997) Root Mean Square (RMS) ISO 13473-2 (2002) Length The length of each scan line (accounting for changes in elevation) Length Ratio The ratio of profile length to device scan length (107.95 mm) Statistical Moments ISO 4288 (1996), ISO 4287 (1997), ASME B46.1 (2009) Mean Square Roughness (Rq) Skewness (Rsk) Kurtosis (Rku) Table 9. Summary of parameters used for Device 12.
Data Collection 47  Device ID Description Laser Type Sample Distance Vertical Resolution 7 WDM TM2 Line Laser 1 mm (Transverse)1 mm (Longitudinal) < 0.05 mm 9 ARRB Walking Profiler 3 Single-Spot Laser 1 mm 0.005 mm 11 SSI Walking Profiler Line Laser 0.3 mm (Transverse)0.5 mm (Longitudinal) 0.015â0.040 mm Table 10. Walking-speed device characteristics. Figure 26. Walk-along device measurements at the Virginia Smart Road facility.
48 Protocols for Network-Level Macrotexture Measurement 3.1.3 High-Speed Equipment HighÂspeed data were collected using five vehicles, each fitted with laser triangulation sensors capable of collecting pavement macrotexture information at the rates shown in Table 12. All data were gathered in the left wheelpath of the âuphillâ lane (the lane runs from east to west). Data were received in the raw spatial form from the operators. Height measurements were given with the fixed sampling interval shown in Table 12. In the case of Device 3, the authors inter polated time domain data into spatialÂdomain data according to the maximum interval observed for the speeds used. The operators performed no filtering or outlier correction. Raw data were requested to enable an objective comparison of all data by applying the same analysis methods to remove variables such as filtering, outlier correction, and method of parameter calculation. Note that Device 15 was mounted directly behind Device 1 on the same vehicle to test the effects of vehicle motion and driver wander. The Virginia Smart Road (Figure 27) was prepared by placing rubber strips at or near the pavement section transitions of the surfaces tested. These were used to accurately identify the boundaries between the different surfaces. Traffic cones with a strip of reflective tape along the entire height were also placed on the roadâs shoulder to trigger photocells to mark where each transition occurred. Operators calibrated their distance measurement equipment at 105 km/h using survey points placed on the road for that purpose. Section Material * Surface Length (m) SRB PCC Transverse Grooved CRCP 610 PCC2 PCC Longitudinally Diamond Ground JPCP 178 PCC1f PCC Longitudinally Grooved and Ground 162 PCC1d HFST Cargill Safe Lane 30 PCC1b HFST EP-5 30 PCC1a PCC Transverse Tined CRCP 69 L1 AC SMA-12.5 91 K AC Open-Graded Friction Course 92 J AC SM-9.5D 89 H AC SM-9.5D 92 * PCC = Portland cement concrete, HFST = high-friction surface treatments, AC = asphalt concrete, SMA = stone matrix asphalt, SM = surface mix. Table 11. Pavement surfaces tested. Device ID Description Laser Type Make Sampling Frequency (kHz) Raw Data Spatial Interval (mm) Vertical Resolution (± mm) 1 Ames Single-SpotLaser Acuity (Custom) 100 0.25 0.020 2 ARRB Single-Spot Laser Limab SR TexRough 32 1.0 0.010 3 ICC Single-Spot Laser LMI Gocator 1300 Series 32 0.9 0.049 4 Ames (VTTI) Single-Spot Laser Acuity (Custom) 100 0.5 0.020 5 SSI Line Laser LMI Gocator 5 0.3 (Transverse) 25 (Longitudinal) 0.015 to 0.040 15 Ames Single-Spot Laser LMI Optocator 62.4 0.25 0.045 Table 12. Data collection equipment information.
Data Collection 49  Three sets of tests were completed at high speed as summarized in Table 13. The first test, âHigh Speed,â was aimed at studying the overall repeatability and agreement of the various devices. The highÂspeed test commenced with the test vehicle on the Virginia Smart Road turn around (labeled âT4â in Figure 21.) The vehicle subsequently accelerated to the test speed and remained at this speed (typically using cruise control) across all the surfaces tested. To study the possible effect of speed on a device, the second test, âConstant Speed,â started at location T3 and ended at T2. This test involved a PCC section and an asphalt concrete (AC) section. The vehicle commenced at T3, accelerated to the test speed before crossing into the test Figure 27. High-speed measurements at the Virginia Smart Road facility. Test Name Speed # Runs SectionTested High Speed (HS) 55 mph (89 km/h) 5 Smart Road Bridge through Section C Constant Speed (CS) 15 mph to 65 mph (24 km/h to 105 km/h), in 10 mph (16 km/h) increments 4 PCC1a, Section L Variable Speed (VS) 25 mph (40 km/h), speed up to 50 mph (80 km/h); 50 mph (80 km/h), slow to 25 mph (40 km/h); 50 mph (80 km/h), full stop, 50 mph (80 km/h); and 50 mph (80 km/h), slow to 25 mph (40 km/h), and then speed up to 50 mph (80 km/h) 4 Sections H through F Table 13. Summary of tests performed.
50 Protocols for Network-Level Macrotexture Measurement sections, and remained at that speed for the duration of the test. The speeds listed in Table 13 were used in the constantÂspeed test. The third test, âVariable Speed,â evaluated the effect of acceleration on measurements. It com menced at Section H and ran through Section F. As was shown in Figure 21, the surface type is the same for these sections, so they can be considered as one section. Four speed profiles were used, and the drivers were instructed to achieve the initial speed from T3 and then perform the test action (acceleration or deceleration) by the midÂpoint of the test section, which was marked by a cone. The vehicles then proceeded through the remainder of the test section at that speed or accelerated to the next required speed by the end of the test section. For the highÂspeed runs, data were collected without interruption along the length of the road. To break these datasets into individual sections (i.e., various surface types), reflective cones were placed to trigger event markers using the various devicesâ photocell triggers, and bump strips of known dimensions were placed across the width of the lane. The reflective cones did not always trigger an event marker for all devices, so the bump strips were used to locate the beginning and end of each section definitively. These bumps were meticulously located in the data for each of five runs of the 18 sections of interest for all six devices. This ensured that each sectionâs set of data had a common starting point across devices and runs to minimize the effect of vehicle wander on profile lengths compared to simply using the overall test start and stop points. The start/stop bumps were similarly located for the constantÂspeed and variable speed tests data. Once sections were parceled out, 3 m were cut off at the beginning and end of all data. This was done to ensure that the bump strip would not affect the data analysis. The length of 3 m was also selected to minimize the effect of inconsistencies in the road surface. For example, steel bridge expansion joints, areas of PCC near the start of a section that could not receive grinding/ grooving to match subsequent pavement, and abrupt transitions between PCC/AC or surface treatments were all removed in this manner. This provided a more consistent surface profile from which to calculate macrotexture parameters. The MPD was again selected as the parameter to represent the macrotexture, as it is widely used both in the United States and abroad. An initial analysis of variance (ANOVA) of repeated runs of the same devices showed that aggregating MPD to a 1Âmeter or greater reporting length resulted in failure to reject the null hypothesis, indicating that means were equal for each run. For this reason, MSDs aggregated to 1 m to get an overall MPD were used for the entirety of the analysis in this work. Aggregation reduces the negative impacts of slight vehicle wander, operator experience, and misalignment of individual runs. As noted in various standards for calculating MPD (ASTM E1845 [2015]; ISO 13473Â1 [2019]), outliers in the data must first be removed. This is especially important for MPD calculations, as this parameter is sensitive to large data variations. All data from singleÂspot laser devices were treated with the same adaptive outlier removal routine used for the walking devices. Fig ure 28 provides an example of profiles before and after outlier removal. The data from the line laser (Device 5) was treated in the same manner as the walkingÂspeed macrotexture measure ment devices. 3.2 Verification Experiment The second equipment comparison aimed to verify some of the first comparisonâs findings and to further assess and refine the most promising approaches for collecting data and characterizing pavement macrotexture identified from that first experiment.
Data Collection 51  3.2.1 Location and Procedures The comparison was conducted on September 24â26, 2018, at the Minnesota DOT MnROAD facility in Albertville, Minnesota. To facilitate the alignment and processing of the data, the research team marked each section with reflective tape (Figure 29). Figure 30 shows pictures of the surfaces tested, which were located in the mainline and in the lowÂvolume loop, as indicated in Table 14, which summarizes the sections selected for the experiment. 3.2.2 Reference Measurements Reference measurements were obtained with one of the walkÂalong devices (Figure 31). This particular device was selected because it allowed measurements to be taken at a higher speed than the other devices from the first comparison, provided appropriate data, and the manufacturer (WDM Ltd.) was willing to participate in the experiment. Figure 28. Example outlier removal of single-spot lasers. Figure 29. Section demarcation.
52 Protocols for Network-Level Macrotexture Measurement Figure 30. Surfaces tested at MnROAD.
Data Collection 53  3.2.3 High-Speed Equipment The devices included in the comparison experiment are listed in Table 15. Figure 32 includes some pictures taken during the experiment. Note that for part of the testing, the surface was wet; measurements under these conditions were taken only to assess the impact of moisture on the surface. 3.3 Validation Experiments The third comparison, conducted at the Texas A&M Universityâs RELLIS Campus, aimed to validate the recommended method for networkÂlevel macrotexture data collection and pro cessing. To accomplish this objective, Texas A&M Transportation Institute (TTI) researchers included test speed and laser exposure time in the comparison test matrix, fabricated a measure ment beam to collect static reference texture data using a highÂresolution laser, and compared texture measurements from different devices with corresponding reference values. Figure 31. Reference measurements at MnROAD. Section # Location Lane Wearing Course (using MnROAD map) Cell Speed (mph) Length (ft) 1 Mainline O Dense-Graded AC 20 50 500 2 Low Volume I Open-Graded Friction Course 55/58 40 150 3 Mainline O Gap-Graded AC 3 50 454 4 Mainline O Microsurface 115 50 496 5 Low Volume O Chip Seal 233 50 425 6 Low Volume O PCC, Transverse Tined 36 50 480 7 Mainline I PCC, Longitudinally Tined 613 50 500 8 Mainline I PCC, Longitudinally Diamond Ground 71/73 50 477 9 Mainline O PCC, Longitudinally Grooved 71/73 50 477 10 Mainline O PCC, Transverse Tined 12 50 499 I = inside lane; O = outside lane. Table 14. MnROAD sections tested.
54 Protocols for Network-Level Macrotexture Measurement 3.3.1 Equipment Used Table 16 summarizes the equipment used for the comparison. Two highÂspeed profilers were included, one with a singleÂspot laser and the other with a line laser. A walking profiler and a new static reference device (the laser analyzer for pavement surfaces, or LAPS) also were included. The highÂspeed devices were each equipped with two lasers. 3.3.2 Location and Procedures Figure 33 shows pictures of the pavement surfaces used in this last field experiment: dense graded HMA, OGFC, stoneÂmatrix asphalt (SMA), ground and polished HMA, ground and longitudinally grooved PCC, and ground and transversely grooved PCC. Figure 32. High-speed measurements at MnROAD. Device ID Company Device Measurement Speed Sampling Frequency Raw Data Interval 1 WDM Ltd. TM2 Walking NA 1 mm (Transverse) 1 mm (Longitudinal) 2 ICC Single-Spot Laser 50 mph 32 kHz TBD 3 VTTI Single-Spot Laser 50 mph 100 kHz 0.5 mm 4 Ames Single-Spot Laser 50 mph 100 kHz 0.25 mm 5 LTPP Single-Spot Laser 50 mph 62.4 kHz 0.5 mm 6 SSI @ 45° Line Laser 50 mph 5 kHz 0.5 mm (Transverse) 25.4 mm (Longitudinal) 7 Ames Single-Spot Laser 50 mph 62.4 kHz 0.25 mm 8 ICC FTM * Static NA NA 9 ICC Line Laser 50 mph 5 kHz TBD 10 SSI Transversal Line Laser 50 mph 5 kHz 0.5 mm (Transverse) 25.4 mm (Longitudinal) * FTM = Fast Texture Meter. Table 15. Devices included at MnROAD.
Data Collection 55  High-Speed Measurements HighÂspeed textureÂmeasurement systems from Ames and SSI were included in this experi ment. A single, highÂspeed line laser was mounted in three different configurations: longitudinal (LLL), transverse (LLT), and at an angle (LLA), as shown in Figure 34. In the LLL configuration, the laser line was oriented parallel to the direction of travel; in the LLT configuration, the laser line was oriented perpendicular to the direction of travel; and in the LLA configuration, the angle of the laser was set at 45°. To meet the stated objectives, specific vehicle speeds and sensor exposure settings were used, as listed in Table 17. Speeds for all vehicles were the same and were held constant by using inÂvehicle cruise control. Sensor exposure time describes the length of time photons are col lected by the triangulation sensor via the electronic shutter. Exposure can be adjusted in the device software to the lengths listed in the table. Note that the line lasers have an âauto expo sureâ setting that adjusts the exposure time to best expose the height data collected per the manufacturerâs algorithm. SingleÂspot devices do not have an auto exposure setting; therefore, ranges of exposure times were selected for use in the singleÂspot equipment tested. The normal operating condition for the singleÂspot device tested is 5â12 µs. The line laser use a diffuse laser projection; therefore, fewer photons are available per unit area as the return light is collected. For this reason, the exposure times selected for the lineÂlaser equipment were higher than those used with the singleÂspot equipment, allowing more photons to reflect off the surface to read the height. Because the exposure times differed for the singleÂspot and lineÂlaser technologies, levels of Short, Medium, Long, and Automatic (Auto) were established as exposure attributes. Reference Measurements The test plan included measurements with the WDM TM2 rolling texture meter. The TM2 is equipped with a line laser having a 100 mm wide footprint oriented perpendicular to the direction of travel. At RELLIS, the TM2 operator collected line scans at 1 mm intervals along the longitudinal direction and processed the data to report the MPD and RMS macro texture indices at 1 m, 10 m, and 50 m intervals. These indices were determined using the laser displacement readings on each line scan. For texture measurements, the TM2 operator walked the device along the test wheelpath, making three repeat runs on each of the seven RELLIS test sections using the default TM2 laser exposure time setting. To guide the operator, TTI researchers delineated the test wheelpath Device ID LaserOrientation Make Sampling Frequency (kHz) Raw Data Spatial Interval (mm) Vertical Resolution (± mm) SS Single Spot Acuity (Custom) AR550-200 * 100 0.25 0.020 Single Spot Optocator * 62.4 0.25 0.045 LLL Longitudinal LMI Gocator 2342 * 5 25 (Transverse) 0.5 (Longitudinal) 0.015 to 0.040 LLT Transverse LLA 45° WDM Line Laser (Transverse) -- 0.3 (Transverse) 0.5 (Longitudinal) 0.015 to 0.040 LAPS Line Laser(Transverse) Keyence LJ- V7200 0.5 (tested) 0.02 (Transverse) 0.112 (Longitudinal) 0.001 * High-speed profiler. Table 16. Equipment used to gather data at RELLIS.
56 Protocols for Network-Level Macrotexture Measurement DGF1: Dense-graded HMA (20-year old HMA surface) lat. 30.629002, long. 96.481648 DGF2: Dense-graded HMA (new surface < 1 year old) lat. 30.627053, long 96.481635 OGFC: Open-graded friction course lat. 30.628968, long. 96.476357 SMAF: Stone-matrix asphalt (fine aggregate gradation) lat. 30.629983, long. 96.476356 HMAP: Ground and polished HMA lat. 30.630992, long. 96.476391 PCCL: Ground and longitudinally grooved PCC lat. 30.628171, long. 96.476331 PCCT: Ground and transversely grooved PCC lat. 30.630935, long. 96.476356 Figure 33. RELLIS comparison test sections.
Data Collection 57  on each RELLIS section with a thin red line stripe over its 100 m length. The pictures given in Figure 35 show these line stripes. Additional reference data to verify the test measurements from other devices were obtained with the laser texture beam illustrated in Figure 35. This LAPS is equipped with a line laser having a 1 µm vertical resolution that is moved along the beam by a stepper motor drive system to measure the surface texture at the given test location. In this regard, swivel casters facilitate positioning the beam along the path to be measured. Once in place, the operator uses the wheel locks to fix the beam at the given position. The system has five main components: notebook computer, operation and setup software, power inlet and signal conditioning module, Keyence laser system with the laser head mounted on a linear bearing carriage, and laser support structure and travel bar. 3.3.3 Engineered Surfaces Commercial test vehicles measure macrotexture using nonÂcontacting laser displacement devices. However, accuracy checks for these vehicles are typically carried out in a static setting using gauge blocks. These conditions do not adequately reproduce the field conditions. No vehicle dynamics exist to perturb the distance measurement equipment, and ambient light and surface metrology do not simulate the pavementâs surface. Furthermore, any measurement of a (a) (b) LLT LLL LLA Figure 34. Reference plates showing (a) segmentation and (b) orientation of the line lasers used. Attribute Single-Spot Device (SS) Line-Laser Device (LLL, LLT, LLA) Speed 25 mph (40 km/hr), 40 mph (64 km/hr), 55 mph (89km/hr) Short Exposure 5â12 μs 40 μs Medium Exposure 10â20 μs 80 μs Long Exposure 30â40 μs 160 μs Auto Exposure N/A Varies Table 17. Experimental travel speeds and exposure settings.
58 Protocols for Network-Level Macrotexture Measurement pavement surface is not valuable in an analysis of accuracy, as the true value of macrotexture is unknown (Izeppi et al. 2012) and because measurement paths cannot be aligned to guarantee the same pavement profile is measured. Given that the true value of a pavementâs macrotexture is unknown, there is no standard against which to compare macrotexture readings from highÂspeed devices. Furthermore, the effect of vehicle speed, exposure settings of test equipment, and signal filtering should be evalu ated to enable recommending reference surface characteristics. One option is to use a rotating disc with shapes cut into it to test device accuracy; however, this requires that a test vehicle be able to simulate a data collection run at a given speed, and not all vehicles are equipped to do so. In addition, a rotating disc is not appropriate for measuring the newer class of line lasers employed by some vendors. Another alternative is to use machined reference plates. This study used several plates developed by Huang et al. (2013) at the Texas DOT and a new plate manufactured at VTTI (Figure 36). The main objective of testing the reference surfaces was to analyze the ability of various devices to reproduce surface profiles at high speed. The experiment evaluated if speed and exposure were significant factors in the determination of derived parameters. All plates were cut on a computer numeric controlÂmachining device from a single billet of aluminum 6061 alloy. The reflective machine marks that could cause unpredictable reflection of laser light were masked using a surface coating of matte red primer paint. Media blasting was considered and tested on other reference surfaces; however, this type of finish is prone to scratching (which would remove the matte finish), and the heat generated in the blasting process can warp the material blasted, resulting in an inconsistent shape. After surface preparation, the plates were measured to determine their final dimensions. Digital calipers were used by two separate techni cians to measure groove depths and peak/valley widths. However, these measurements were inconsistent between the two operators (Table 18). The final reference measurements were obtained with the LAPS system. Figure 35. LAPS in field testing configuration.
Data Collection 59  Reference LAPS Measurements Plates were first scanned with the laser line parallel to the grooves in the plates; however, this proved challenging for the device on some plates, as the abrupt changes from peak to slope and vice versa caused specular reflections that resulted in outlier data that overÂestimated plate depths estimated by the caliper measurements. Final reference measurements were made by sliding the laser line along the length of the plate with the line perpendicular to the grooves (see Figure 37). This method is more similar to the measurement style of the SS and LLL. Engineered Surfaces Studied The reference surfaces engineered to test the macrotexture devices are described in Table 18 and shown in Figure 38. The smallest shape (Plate 1) is a triangular âsawtoothâ pattern. This is representative of fine mixes such as surface dressings and highÂfriction surface treatments. The midÂsized shape (Plate 5) has a hexagonal halfÂshape and represents typical denseÂgraded asphalt mixes. The largest shape (Plate 6) is similar to Plate 5 but has deeper channels, more similar to larger aggregate or tined or grooved concrete. Segmentation of Plates Each plate was approximately 150 mm wide and 600 mm long. Plates were segmented into base lengths approximately 100 mm in length, as shown in Figure 34. These base lengths were used to calculate MPD as specified in ASTM E1845 (2015); however, strict 100 mm base lengths Figure 36. Reference plates tested during the RELLIS field experiment. Attribute Plate 1 Plate 5 Plate 6 Design MPD (mm) 1.25 3.75 5.15 Tech Calibration Depth 1 (mm) 2.6 7.95 10.3 Tech Calibration Depth 2 (mm) 2.0 7.5 10.5 Avg. LAPS MPD (mm) 1.054 3.872 5.161 Standard Deviation (SD) of Avg. LAPS MPD (mm) 0.007 0.037 0.093 Table 18. Engineered reference surface characteristics.
60 Protocols for Network-Level Macrotexture Measurement (a) (b) Figure 37. LAPS reference measurement device showing (a) lab setup and (b) resulting 3D surface profile. 0 50 100 150 200 250 300 350 400 450 Datapoint -2 -1 0 1 2 Datapoint 0 100 200 300 400 500 600 700 800 900 -5 0 5 10 12 14 160 2 4 6 8 Datapoint 104 -6 -4 -2 0 2 4 6 Pl at e #1 Pl at e # 5 Pl at e #6 Te xt ur e H ei gh t ( m m ) Te xt ur e H ei gh t ( m m ) Te xt ur e H ei gh t ( m m ) Figure 38. Cross-sections of reference surfaces.
Data Collection 61  could not be used, as the starting points and end points did not align with repeatable waveform locations. This resulted in incorrect mean line and regression determination, which affected the calculated MPD. As seen in Figure 39, a 100 mm base length resulted in an MPD of 5.18 mm, whereas a base length of 119 mm (which included a multiple of whole waveforms) resulted in an MPD of 5.47 mm. This difference in MPD could differentiate between investigatory and inter vention levels on a road surface, so the level of difference was unacceptable. Accordingly, for this experiment, base lengths as close as possible to 100 mm were selected by taking points at the center of shape peaks to result in vertical and horizontal symmetry and stable, accurate MPD results. The center of each peak at the beginning of a given segment was found in the raw data profile and then a subsequent center of peak was found approximately 100 mm from the first point. This was the base length used, repeated for all segments of each plate, for every test condition used for the LAPS and highÂspeed data collection effort. To test the effect of filtering on the resulting data, the same start and stop points were used on profiles that were first filtered using a lowÂpass infinite impulse response filter conforming to ASTM E1845 (2015). The filter was applied to the entire profile (with several meters of leadÂin and leadÂout data included) for the singleÂspot laser measurements. For all lineÂlaser measure ments, profiles were first unfolded by mirroring at least 2.5 mm (the cutoff wavelength of the filter used) on both sides of the profile and then filtered. 3.4 Additional Measurements to Evaluate Macrotexture Parameters To evaluate existing and newly developed parameters (aimed at improving correlation with pavement surface properties), the research team evaluated their ability to predict a roadwayâs frictional and noise characteristics. Segment Length: 100 mm. Total 6 segments, the first 5 segments are all 100 mm Segment Length: 119 mm. Total 5 segments, all segments are 119 mm Figure 39. Examples of various segment lengths that were tried for the experiment.
62 Protocols for Network-Level Macrotexture Measurement 3.4.1 Surfaces Studied The surfaces selected were the continuous surfaces from the Smart Road Bridge to Pavement Section C. As transitions between different surface types may have adverse effects on any given deviceâs various measurements, the first and last 1 m of each section were removed from the pavement dataset. The left wheelpath of the âuphillâ travel lane of Virginia Smart Road (see Figure 21) was measured with a singleÂspot laser device to capture the pavement surface macrotexture at 55 mph. The same path was measured with two continuous friction measurement devices and a test vehicle capable of measuring road noise. The friction and sound data were then distilled into the parameters for which prediction was attempted (predicted variables) from either a single macrotexture parameter (predictor variable) via singleÂvariable linear regression or several predictor variables via multiple linear regression. 3.4.2 Equipment Used The devices listed in Table 19 were used for the experiment; they gather continuous data of the road network in a single run at high speed. The vehicles traveled at the speed of traffic and used contactless sensors, so no traffic control was necessary. The SCRIM was operated accord ing to normal operating procedures at 50 km/hr (The Highways Agency 1999). This device also included an inertial unit, which allows for measuring grade, crossÂslope, and curvature. The Grip Tester (GT) was likewise operated at 50 km/hr according to the manufacturerâs standard operating procedure (RoadBase operatorâs manual). The onboard sound intensity (OBSI) was measured at 50 km/hr. From the raw data, overall AÂweighted sound pressure levels were calculated. An optical sensor was added to register the one per revolution signal, and the tire revolution signal was used to remove (Feng 2017) the tire tread pattern noise component from the total tire noise. 3.4.3 Predicted Variables The focus of this experiment was to predict friction (as measured by the SCRIM and GT) and noise (as measured by the OBSI) from pavement surface profiles in the macrotexture range. As such, the data collected by the devices to be predicted were treated using the manufactureâs software to derive the predicted variables. The SCRIM was run at 50 km/hr and SCRIM read ings (SR) were output from the system software. Data are typically evaluated at 50 km/hr. To allow Measurement Manufacturer Model Specification Macrotexture Ames Engineering Accutexture 100 100 kHz Single-Spot Laser Triangulation Profiler Friction WDM Ltd. Sideway-force Coefficient Routine Investigation Machine (SCRIM) Continuous Friction Measurement Device (Standard Tire) Friction Findlay Irvine Grip Tester Mark III (GT) Continuous Friction Measurement Device (Smooth Tire) Noise AVEC Inc. Custom Onboard Sound Intensity (OBSI) with Optical Sensor for Tire Noise Separation Table 19. Equipment used to gather parameter data.
Data Collection 63Â Â for small variations in vehicle travel speed, data are corrected (The Highways Agency 1999) to a constant equivalent SCRIM reading at 50Â km/hr (SR50), using the following equation: 100 0.0152 4.77 799 1000 , (18)50 2 SR SR s s( )( )= â + +p where s = instantaneous speed (km/hr) of the individual measurement. GT data were collected at 50Â km/hr along the left wheelpath of the uphill lane of the Virginia Smart Road. The device software outputs Grip Numbers (GN), which range from 0.0 to 1.0 and are dimensionless. These GNs are the pavement coefficient of friction as measured by the GT using the relationship described in Equation 19: = , (19)GN f f d n where fd = drag force (N), and fn = normal force (N). Noise data were collected in the right wheelpath of the uphill lane of the Virginia Smart Road. The parameter distilled from the raw noise data is the Overall AÂWeighted Sound Pressure Level (OASPL). This is essentially the total sound pressure as measured by the OBSI. The AÂweighting is applied to take into account the relative loudness perceived by the human ears. Tire tread pattern noise was removed using the optical sensor data applying the tire noise separation procedure.