Ultrasound Tomography Testing at the National Airport Testing Facility (NAPTF) Dr. Lev Khazanovich, Principle Investigator

Transcription

1 Ultrasound Tomography Testing at the National Airport Testing Facility (NAPTF) Dr. Lev Khazanovich, Principle Investigator Kyle Hoegh, M.S, Research Assistant

2 Ultrasonic Tomography Testing at NAPTF Executive Summary Surface distress was observed at the National Airport Pavement Testing Facility (NAPTF) at certain locations where strain gauges were embedded. Since the pavement sections exhibiting this distress had not been exposed to load applications at the time of the appearance of the distress it is hypothesized that the aluminum structure used to hold the gauges in place has had an adverse reaction with the pavement. All of the testing in this report was conducted on the CC6 slabs of the NAPTF. Nondestructive testing has been conducted at the locations where gauges have been installed to determine the presence and/or extent of the subsurface damage prior to load applications. In this case, the effect of the pre-loading damage, if any, can be differentiated from the effects caused by load applications. Additional tests were conducted at locations where gauges are not installed to examine if the observed distress is unique to the gauge locations. Ultrasonic tomography testing of the CC6 section of the NAPTF and subsequent analysis showed that locations with and without sensors both experienced a high percentage of sound concrete condition (87% and 98%, respectively). Therefore, the results suggest that the popouts and surface damage experienced at various locations prior to load applications is not a systematic problem and should not be generalized to all locations with those types of sensors installed. Nonetheless, while both locations with and without sensors showed a low percentage of damage, it is important to note that a significantly higher percentage of PCC locations with embedded sensors were diagnosed to be damaged. The analysis indicates that approximately 13% of the slab locations with embedded sensors experienced portions of subsurface concrete with at least a moderate level of damage. Testing at locations without the presence of sensors indicated that approximately 2% of the slab locations without embedded sensors experienced at least a moderate level of damage. This indicates that while the embedded sensors did not inherently damage the surrounding concrete it did increase the likelihood of damage. Recent improvements of the data analysis tools (SAFT-Panoramic and SAFT-3D) allowed for more detailed analysis of the tested locations. It would also be useful to apply similar measurements and analysis with MIRA at various stages during and after the load applications to get a time history characterization of the subsurface damage accumulation with load repetitions with respect to the pre-loading condition. 2

3 Technical Background Equipment and On-Site Measurement Analysis A recently developed ultrasonic tomography device (MIRA) offers high accuracy measurements of concrete thicknesses, reinforcement or inclusion location, debonding between layers, subsurface cracking or defects, and presence of honeycombing or poor consolidation. MIRA is comprised of 40 (10x4 array) touch and go transmitting and receiving dry point contact (DPC) transducers utilizing 45 transducer pair measurements (see figure 1) in each scan resulting in a 2D depth profile. The DPC transducers provide the necessary consistency of impact and wavefront penetration for diagnositcs up to 3 ft. deep. The multiple sensor pairs in each scan allows for the required redundancy of information to evaluate heterogeneous mediums such as concrete. Figure 1 shows a manual MIRA measurement as well as the high redundancy of information when using a multi-static array of transmitting and receiving transducers. Mira 45 pairs per measurement Figure 1. MIRA ultrasonic tomography device and illustration of transmitting and receiving pair measurements. MIRA is an ultrasonic tomography device that uses an array of transducers to send and receive sound waves at the surface. Each approximately 1 second MIRA scan gives a 2 dimensional depth cross-section (SAFT B-scan) with the vertical axis indicating the depth of any reflection (caused by any change in acoustic impedance), and the horizontal axis indicating the location along the 16 inch aperture of the MIRA with 0 being the center of the scan location. Figure 2 shows an example on site SAFT B-scan output that was taken during testing on an undamaged CC6 location over the top of a dowel. It can be observed that there is a red planar shape at approximately 12 in. depth at the interface between the PCC and the base and a red circle at approximately half of that depth. Any changes in acoustic impedance (material stiffness/density changes) will cause a high intensity of reflection (red) associated with the depth of this anomaly. The dowel reflection was caused due to the higher acoustic impedance of metal as compared to concrete. The reflection was caused at the concrete-base interface due to the lower acoustic impedance of the base material. This method can be used to detect cracks in concrete among other subsurface anomalies because air has a much lower acoustic impedance than concrete. More background about MIRA analysis can be found by consulting the documents in the references section. 3

4 0 Depth, in. Dowel 16 in. PCC/Base Interface 12 MM Figure 2. Example on-site SAFT B-scan indicating the location of a dowel and depth of the PCC/Base interface. CC6 Section Background A construction cycle at the FAA s National Airport Pavement Test Facility (NAPTF) includes test pavement construction with embedded instrumentation, traffic tests to failure, post traffic testing, and ultimately pavement removal. The current rigid pavement in the area of the medium strength subgrade was designated as Construction Cycle 6 (CC6). The three test items in CC6 have been constructed with identical cross-sections, but with three different concrete mixes 4

5 designed to give different values of flexural strength, ranging from nominal 3.5MPa (500 psi) to 6.9MPa (1000psi). In addition, two different base materials (econocrete and hot-mix asphalt) provide a total of 6 combinations of concrete strength and base type. The steel dowels, measured 25mm in diameter and 450mm in length, were placed in both longitudinal and transverse directions. Water soaked burlap strips were placed over the surface of the concrete at the end of the placement day. The burlap was in turn covered with plastic sheets in order to retain the moisture for curing the concrete over a 28-day period. Surface distress was observed atcertain locations within CC6 where strain gauges were embedded (see Figure 3). Since the pavement sections exhibiting this distress had not been exposed to load applications at the time of the appearance of the distress it is hypothesized that the aluminum structure used to hold the gauges in place has had an adverse reaction with the pavement. Figure 3. Observed distress at a strain gauge location. To investigate if this observed distress was a systematic problem at the sensor locations, ultrasonic tomography testing was conducted in 2 mode types including express mode testing at locations without embedded sensors and detailed testing at locations with embedded sensors. Figure 4shows express mode testing where 3 semi-overlapping scans are taken in each orientation for a total of 6 scans centered at each location of testing. Figure 5shows the 45 scan locations for detailed mode testing which was conducted at the embedded gage locations at transverse and longitudinal joints. The detailed mode testing was conducted where instrumentation is present to ensure that information at and around the instrumentation will be available. Testing occurred from left to right and typically started centered between lines 4 and 5 5

6 moving towards the joint along each subsequent line. The express mode testing with less MIRA scans allowed for the necessary information to determine the subsurface concrete condition at locations where instrumentation is not present. Testing occurred from left to right starting from the furthest location followed by testing closer to the joint. 6

7 C L NOTE: ALL EXPRESS LOCATIONS HAVE THE SAME DIMENSIONS SHOWN BELOW. TESTING OCCURRED AT THE NUMBERED LOCATIONS IN ORDER. 15' 3" 2" #1 9" #4 #2 #5 15' #3 #6 JOINT Figure 4. Illustration of the 6 scans in relation to the testing location in express mode.

8 JOINT LINE-1 LINE0 EMBEDDED GAUGE LINE1 LINE2 LINE3 LINE DIAGRAM 2 Figure 5. Detailed testing locations with 45 scans at embedded gages (EG) LINE5 8

9 Analysis methods The data collected during testing allowed for identification of dowels and defects when present at and around each testing location. The following analysis was applied at each scanned location to obtain this information: SAFT B-scan: Synthetic aperture focusing technique cross-sections based on the absolute intensity of reflection in each 2D-depth profile (see Figure 6). This type of analysis was conducted at every scanned location. SAFT-Panoramic: Synthetic aperture focusing technique panoramic reconstructions based on fusing absolute intensity of reflection from multiple overlapping scans (see Figure 7). This type of analysis was conducted at every scanned location. SAFT-3D: Synthetic aperture focusing technique 3-dimensional reconstructions based on interpolation of threshold intensity reflection locations (see Figure 8). This type of analysis was conducted at all embedded gauge locations 2D-UTSA: Two-dimensional ultrasonic tomography signature analysis method comparing SAFT B-scans using Pearson s correlation to create a Corellogram of each scan as compared to the sound concrete condition. This type of analysis was conducted at the unreinforced express mode testing locations. Table 1 gives an explanation of the different subsurface damage level designations with higher numbers indicating greater subsurface damage extent. The damage at the various locations was categorized by using some of the following evaluation methods as applied to the analysis methods described above: Direct reflection evaluation - The depth and lateral location of strong reflections within the PCC layer were evaluated using all three SAFT analysis methods described above. A comparison of reflection locations with the expected as-designed instrumentation or reinforcement locations gave information about the condition of the concrete around the instrumentation and reinforcements when present. If no high intensity of reflection locations were observed within the concrete layer other than at reinforcement or instrumented positions, this indicated a sound concrete condition. Locations where high intensity reflections were observed other than locations were reinforcement or sensors were located were reported as indicative of a certain extent of damage. Shadowing effect - The reflection at the depth of the PCC layer at the interface with the base also gave information about the condition of the subsurface PCC layer. If the concrete is in sound condition a continuous oblong reflection ( backwall reflection) should be expected at the depth of the PCC. If there is a discontinuity in this backwall reflection it can be an indicator that a flaw within the PCC layer is blocking the wave from propogating to this interface and back to the surface. This phenomenon is referred to as a shadowing effect in this report. Drop in Pearson s correlation - For the locations where no reinforcement was present, two-dimensional ultrasonic tomography signature analysis (2D-UTSA) was conducted to 9

10 evaluate the uniformity of the concrete slabs and identify subsurface defects when present. In this type of evaluation a low Pearson s correlation value indicated the presence of a subsurface flaw. Damage Levels Inclusion Table 1. Subsurface damage level explanations. Description Undamaged subsurface condition - only slight level of reflection intensity at locations other than should be expected from reinfocement or gauge locations indicating high probablility that there is no significant suburface damage. Slightly damaged subsurface condition - discernable high intensity of reflection at locations other than should be expected from reinforcement or gauge locations or slight shadowing effects indicating a significant probability that there is a slight level of subsurface distress. Damaged subsurface condition - significant high intensity of reflection at locations other than should be expected from reinforcement or gauge locations or significant shadowing effects indicating the presence of a moderate level of subsurface distress. Extremely damaged subsurface condition - high level and coverage of high intensity of reflection at locations other than should be expected from reinforcement or gauge locations combined with significant shadowing effects indicating largely damaged subsurface concrete. Subsurface inclusion systematic high intensity of reflections indicative of an inclusion (such as a reinforcement mesh) in the concrete rather than damage at locations other than should be expected from gauge locations. The lack of shadowing suggests that there are no flaws at this location, rather the presence of a stiff inclusion such as metal. Figures 6 through 8 show an example of a scanned location (slab 10S; sensors 103 and 104) categorized to have no significant subsurface damage (level 1). In these cases the high level of reflection intensity are observed at lateral and depth locations where there is an expected change in acoustic impedance such as a dowel inclusion or at the interface between the concrete and base material. Figure 6 shows 9 B-scans each centered approximately 5 in. from the transverse joint (between line 1and line 2 positions in Figure 5). It can be observed that round high intensity reflections (red) are located at about half the depth of the more oblong high intensity reflection (red) at a greater depth. The round reflections indicate the lateral location and depth of the dowels while the oblong reflection indicates the depth of the PCC pavement layer. It can be observed that locations other than the doweled and PCC depth have a low intensity of reflection indicating sound concrete. The location where each SAFT B-scan taken at the NAPTF during testing can be accessed in Table A1 of Appendix A. 10

11 L L Scan 1 L L M Scan 5 L M Scan 7 Scan 4 Scan 2 L M M R Scan 8 M M R Scan 3 Scan 6 Scan 9 Figure 6. Example set of 9 overlapping B-scans used to create a SAFT Panoramic. Figure 7 shows the five SAFT Panoramic reconstructions each resulting from nine overlapping SAFT B-scans taken in two inch step sizes after they have been fused together. These scans allowed for an approximately 3 ft. reconstruction along the joint with MIRA centered at the locations listed below with the top scan in Figure 7 corresponding to the first location on this list and so on: 11 in. from the joint - between lines 4 and 5 (see Figure 5) 9 in. from the joint - between lines 3 and 4 (see Figure 5) 7 in. from the joint - between lines 2 and 3 (see Figure 5) 5 in. from the joint - between lines 1 and 2 (see Figure 5) 3 in. from the joint - between lines 0 and 1 (see Figure 5) The resulting panoramic tomograph indicates the subsurface condition of a 3 ft wide section of the pavement. Each panoramic shown in Figure 7 indicates an undamaged condition where the only high intensity of reflection (red) occurs due to features that were as designed including the slightly less than 1 ft concrete depth reflection, circular reflections at the location and depth of the dowels, and reflections at the location and depth of the sensor area. The lack of reflection (blue) at the remaining locations indicates undamaged concrete. Dowel reflections are not observed in the top panoramic because the scans were taken 11 in. from the joint where no dowels are present. Dowels are observed in the remaining SAFT panoramic centered 11

12 9 in. and closer to the joint. The circular red high intensity reflections can be observed at the lateral and depth locations of the dowels under the scan locations. The left, middle, and right dowels in the reconstructions associated with the SAFT B-scans shown in Figure 6 and corresponding SAFT Panoramic are labeled L, M, and R. This shows how the dowels from overlapping scans are treated and how fusing data from multiple B-scans can be used to create a clearer picture of the dowel locations in relation to one another using the 3 ft. SAFT Panoramics as well as for creating a continuous reflection at the depth of the PCC. These strong reflections at the locations of the dowels and strong backwall reflection at the depth of the concrete and rebar locations should be expected if there are no significant flaws or cracking within the concrete layer. The SAFT Panoramic of each tested location at the NAPTF is shown in Appendix B and the location of the file where each panoramic can be accessed is given in Table B1. Reflection at PCC depth 11 Reflection at PCC depth 9 Reflection at Dowels Reflection at PCC depth L M R 7 Reflection at PCC depth 5 Reflection at Sensor Location Reflection at PCC depth 3 Location along joint, in. Figure 7. Five SAFT Panoramic examples at slab 10S sensors 103 and 104 diagnosed to be undamaged. 12

13 Figure 8 shows the SAFT 3D reconstruction resulting from the SAFT Panoramics shown in Figure 7. This reconstruction was created by using threshold and interpolation techniques to make a three dimensional reconstruction of the 3 ft by 1ft tested location. The approximately 9 in. long dowels can be observed in Figure 8 in addition to some reflections at the sensor location and the continuous reflection at the PCC depth. This type of SAFT 3D reconstruction was useful for getting relational information about the high intensity reflections to determine if the reflection was caused by an as designed inclusion or if it indicated damaged concrete. The videos showing a rotating view of each of the SAFT 3D reconstructions can be accessed at the file locations indicated in Table C1 of Appendix C. Sensor Reflection at PCC Depth Dowels Depth, in Figure 8. SAFT 3D reconstruction at slab 10S sensors 103 and 104 using the SAFT Panoramics shown in Figure 7. The SAFT Panoramic and SAFT 3D analysis techniques were primarily used for determination of the subsurface damage level due to the additional redundancy and coverage provided by the methods. Figures 9 and 10 show SAFT Panoramic and SAFT 3D reconstructions, respectively, giving an example of a damage level 4 detailed testing location at slab 22N sensors 23 and 24. Figures 11 and 12 show SAFT Panoramic and SAFT 3D reconstructions, respectively, giving an example of a damage level 3 detailed testing location at slab 2N sensors 3 and 4. Figures 13 and 14 show SAFT panoramic and SAFT 3D reconstructions, respectively, giving an example of a damage level 2 detailed testing location at slab 3S sensors 41 and 42. These figures illustrate example evaluation indicators used to diagnose the various levels of subsurface damage. This includes high intensity reflections in excess of the as-designed subsurface inclusion or layer locations as well as shadowing of the backwall reflection. Figure 9 shows SAFT panoramic reconstructions at slab 22N sensors 23 and 24 categorized as damage level 4. Figure 9 from top to bottom was created with MIRA centered at the following locations: 11 in. from the joint - between lines 4 and 5 (see Figure 5) 13

14 o High intensity reflections can be observed at a shallow depth although no inclusions were designed to be present at 11 in. from the joint indicating the presence of damage. o There is a discontinuity in the backwall reflection at the depth of the PCC indicating that a shallower flaw is blocking the wave propagation above this lateral location. 9 in. from the joint - between lines 3 and 4 (see Figure 5) o High intensity reflections can be observed in addition to the reflections at asdesigned dowel locations indicating the presence of damage. o There is a discontinuity in the backwall reflection at the depth of the PCC indicating that a shallower flaw is blocking the wave propagation above this lateral location. 7 in. from the joint - between lines 2 and 3 (see Figure 5) o High intensity reflections can be observed in addition to the reflections at asdesigned dowel locations indicating the presence of damage. o There is a discontinuity in the backwall reflection at the depth of the PCC indicating that a shallower flaw is blocking the wave propagation above this lateral location. 5 in. from the joint - between lines 1 and 2 (see Figure 5) o High intensity reflections can be observed in addition to the reflections at asdesigned dowel locations indicating the presence of damage. o There is a discontinuity in the backwall reflection at the depth of the PCC indicating that a shallower flaw is blocking the wave propagation above this lateral location. 3 in. from the joint - between lines 0 and 1 (see Figure 5) o High intensity reflections can be observed in addition to the reflections at asdesigned dowel and sensor locations indicating the presence of damage. o There is a discontinuity in the backwall reflection at the depth of the PCC indicating that a shallower flaw is blocking the wave propagation above this lateral location. 14

15 Shadowing due to damage 11 Shadowing due to damage 9 Reflection at Dowels Damage Shadowing due to damage 7 Reflection at Dowels Shadowing due to damage 5 Shadowing due to damage 3 Location along Joint, in. Figure 9. SAFT Panoramics at slab 22N sensors 23 and 24 indicating a damage level of 4. Figure 10 shows two views of the 3D reconstruction at this location. It can be observed that the shallow reflection and shadowing to the right of the center dowel extends from the joint to over 15

16 10 in. away from the joint indicating a high level of damage (damage level 4) at this sensor location. damage View (a) Depth, in. Distance from Joint, in. Distance from Joint, in. damage Shadowing View (b) Depth, in. Location along Joint, in. Figure 10. SAFT 3D reconstructions at slab 22N sensors 23 and 24 indicating a damage level of 4. Figure 11 shows SAFT panoramic reconstructions at slab 2N sensors 3 and 4 categorized as damage level 3. From top to bottom in the figure these panoramic reconstructions were created with MIRA centered at the following locations: 7 in. from the joint - between lines 2 and 3 (see Figure 5) 16

17 o High intensity reflections can be observed at the designed dowel locations. A slight amount of reflection at the close upper left portion of the middle dowel area indicates a moderate level of damage. 5 in. from the joint - between lines 1 and 2 (see Figure 5) o High intensity reflections can be observed at the designed dowel locations. Significant reflection intensity can be observed to the upper left portion of the middle dowel indicating damage. o There is a discontinuity in the backwall reflection at the depth of the PCC below the center dowel indicating that a shallower flaw is blocking the wave propagation above this lateral location. 3 in. from the joint - between lines 0 and 1 (see Figure 5) o High intensity reflections can be observed at the designed dowel and sensor locations. Significant reflection intensity can be observed to the upper left portion of the middle dowel indicating damage. o There is a discontinuity in the backwall reflection at the depth of the PCC below the center dowel indicating that a shallower flaw is blocking the wave propagation above this lateral location. 17

18 7 Reflection at Dowels Damage Shadowing due to damage 5 Reflection at Dowels Shadowing due to damage 3 Location along the Joint, in. Figure 11. Three SAFT Panoramic reconstructions at slab 2N sensors 3 and 4 diagnosed with a damage level of 3. Figure 12 shows two views of the 3D reconstruction at this location. It can be observed that the shallow reflection and shaddowing to the left of the center dowel extends from the joint to about 5 in. away from the joint and a moderate level of damage up to 7 in. away from the joint resulting in the diagnosis of damage (damage level 3) at this sensor location. 18

19 Distance away from joint, in. Distance away from joint, in. damage damage Slight Shadowing Distance along joint, in. Slight Shadowing View (a) View (b) Depth, in. Location along joint, in. Figure 12. SAFT 3D reconstructions at slab 2N sensors 3 and 4 with two views indicating a subsurface damage level of 3. Figure 13 shows SAFT panoramic reconstructions at slab 3S sensors 41 and 42 categorized as damage level 2. From top to bottom these SAFT panoramic reconstructions were created with MIRA centered at the following locations: 11 in. from the joint - between lines 4 and 5 (see Figure 5) o A moderate amount of reflection intensity at the upper left portion of the panoramic reconstruction indicates a moderate level of damage. 9 in. from the joint - between lines 3 and 4 (see Figure 5) o High intensity reflections can be observed at the designed dowel locations. A moderate amount of reflection intensity can be observed at the same lateral and depth location observed at 11 in. from the joint indicating a moderate level of damage. 19

20 3 in. from the joint - between lines 0 and 1 (see Figure 5) o High intensity reflections can be observed at the designed dowel and sensor locations. Reflection intensity can be observed in a larger area than the expected sensor location indicating moderate damage in the concrete surrounding the sensor. o There is a discontinuity in the backwall reflection at the depth of the PCC below the center dowel indicating that there is moderate damage above this lateral location. 11 Reflection at Dowels Damage Shadowing due to damage 9 Reflection at Dowels Shadowing due to damage 3 Location along joint, in. Figure 13. SAFT panoramic reconstructions at slab 3S sensors 41 and 42 indicating a damage level of 2. Figure 14 shows the 3D reconstruction at this location. It can be observed that the shallow reflections observed at 9 and 11 in. from the joint were not at a significant enough intensity to be include in the 3D reconstruction which only shows reflections at a threshold of over 200. However, the reflection intensity observed in a larger area surrounding the sensor location was also observed in the 3D reconstruction near the joint. These observations resulted in the diagnosis of moderate damage (damage level 2) at this sensor location. 20

21 damage Depth, in. Figure 14. A SAFT 3D reconstruction at slab 3S sensors 41 and 42 indicating a damage Level of 2. There were also locations such as slab 6S sensors 51 and 52 where the sensor area was categorized as having a systematic metal inclusion. Figure 15 shows an example SAFT panoramic reconstruction centered at 11 in. from the joint where the area was categorized as inclusion. It can be observed that there is a planar high intensity reflection shallower than the reflection at the designed PCC thickness. In addition, it is observed that this high intensity reflection does not cause a shadowing of the backwall reflection indicating that the inclusion is most likely not a flaw blocking the waves from propagating to the PCC depth. This type of reflection was observed consistently in each SAFT panoramic reconstruction at various sensor locations. In these cases no flaw indications were observed. While these locations indicated sound concrete they were labeled as inclusion to indicate the presence of this different type of signal. Inclusion 11 Location along Joint, in. Figure 15. Inclusion diagnosis location diagnosis example at slab 6S sensors 51 and 52 at 11 in. from the joint. 21

22 Pearson s correlation was used to compare SAFT B-scans in locations were inclusions were not present to determine any variability from a sound concrete condition. The Pearson s coefficient was determined using the following equation: C j XY 2D N M j j ( xik x2dmean )( yik = Cov[ X 2D, Y2D ] i= 1 k= 1 = j N M N Var[ X M D Var Y 2 2 ] [ 2D ] ( xik x2dmean ) i= 1 k= 1 i= 1 k = 1 y ( y j ik j 2Dmean y ) j 2Dmean (2) where X 2D and Y j 2D are the matrices of reflection intensity for the reference B-scan and current B-scan, respectively; x ik and y j ik are the single intensity values of the reference signal and current signal, respectively, with depth below the measurement location increasing with i and the location along the aperture of the scan increasing with k; X 2Dmean and Y j 2Dmean are the mean intensities of the reference B-scan and current B-scan, respectively; N and M are the number of intensity values in the depth and device aperture direction, respectively; and C j XY2D is Pearson s correlation coefficient, which measures the strength of the linear dependence between X 2D and Y j 2D. ) 2 Thus, if a SAFT B-scan taken on relatively sound concrete with similar structural geometry is used as the reference scan, flawed concrete locations can be identified with a low Pearson s correlation. On the extremes, a C j XY2D value of 0 would indicate no correlation and a C j XY2D value of 1 would indicate that the two B-scans are related linearly. Therefore, a higher C j XY2D would indicate similar scans or sound concrete, and a significant decrease in the correlation coefficient would indicate dissimilar B-scans, or flawed concrete, especially if observed in a group of adjacent scans. This type of analysis will be referred to as the 2D-UTSA method. There is generally little variation between B-scans of concrete in relatively good condition at different locations if the same instrument settings are used, while there is a significant variation between scans where flaws are present at different locations. Therefore, sound concrete will have a similar level of correlation with the reference B-scan, whereas the correlation of scans with flaws at different locations will fluctuate. It should be noted that systematic flaws or inclusions (such as metal reinforcements) that extend horizontally at similar depths could potentially affect the reference scan selection. In these cases, as well as cases where the structural geometry is inconsistent within the scanning area, the current formulation of the 2D- UTSA method is not applicable. The 2D-UTSA method requires selection of a reference B-scan based on engineering judgment, which raises concerns of sensitivity of the 2D-UTSA analysis to the choice of reference scan. Therefore, a procedure for generating a reference scan was used in the analysis. In this procedure, the reference scan is taken as the average of all of the B-scans in the set being compared. Because each B-scan has an intensity value associated with each pixel location, and the dimensions of matrices containing the intensity values is the same for all B-scans, the resulting reference scan associated with this procedure is simply the average intensity value for all pixel locations in the scanned area. It is expected that sound concrete may not necessarily have as high of a correlation with the generated reference scan as is the case for a manually selected reference scan. However, if a significant portion of measurements are made on sound 22

23 concrete, the sound concrete locations should result in similar correlation values, while unsound concrete will result in lower values due to the randomness of flaws. Thus, decreases are still present in the correlogram even when the reference scan includes contributions from the flawed concrete locations. This method is not overly sensitive to selection of the reference scan, and can be generally applied to locate areas of flawed concrete. Results Figure 16 shows the results of 2D-UTSA Pearson s correlation testing at all express locations where inclusions were not designed to be present. It can be observed that only 2 significant dips were observed in Pearson s correlation. These dips occurred at express locations measured on the North of slab 2N and the East of slab 3S. All other locations had a Pearson s correlation higher than 0.90 and were considered to be in sound concrete condition. Thus, over 98 percent of the 2D-UTSA diagnosed locations were considered to be in sound concrete condition. Table D1 in Appendix D can be referenced for each individual Pearson s correlation and the associated testing location Pearson's Correlation location: 3S-E location: 2N-N Scan location (X-axis # is associate with row # in the express mode table) Figure 16. Pearson s Correlation from 2D-UTSA testing at express unreinforced locations (See Table D1 for locations and numbers). Table 2 shows the results of the detailed testing locations where: Column A identifies test locations for detailed grid testing. Grid testing occurred over all the embedded gauges within CC6. The denotation used for these tests was the CC6 23

24 slab and the gauge numbers. For example, (1N 1,2) signifies a grid test location on slab 1N over gauges 1 and 2. Figure 5 depicts the grid layout for grid test items. Column B gives the subsurface damage level as diagnosed using the SAFT panoramic and SAFT 3D analysis methods described above. The damage levels 1-4 and inclusion are defined in Table 1 and the text in the analysis section. Column C gives the surface condition including whether or not there is damage or a surface gauge present. The video files given in Column G can be viewed for additional visual information about the surface. Column D gives the date of testing Column E gives the time of testing Column F gives the file name where the SAFT B-scans and SAFT Panoramic pictures, as well as the 3D reconstruction videos at each location can be found. Column G gives the file name where the video showing each MIRA scan location and surface condition can be found. Column H gives any pertinent comments at each testing location. 24

25 Table 2. Sensor location results. Column A Column B Column C Column D Column E Column F Column G Grid Location Subsurface condition Surface condition Date Time Stamp Name root Video location 1N 1, 2 1 sound 7/19/ : _12_59_01r04_B NAPTF12 2N 3, 4 3 surface gauge 7/19/ : _13_04_01r04_B NAPTF13 2N 5, 6 1 sound 7/19/ : _13_09_01r05_B NAPTF14 2N 15, 16 1 sound 7/19/ : _15_17_01r02_B NAPTF30 3N 19, 20 1 surface gauge 7/19/ : _15_22_01r02_B NAPTF31 3N 21, 22 1 sound 7/19/ : _15_25_01r03_B NAPTF32 3S 27, 28 1 surface gauge 7/19/ : _14_30_01r01_B NAPTF24 3S 29, 30 1 sound 7/19/ : _14_35_01r01_B NAPTF25 3S 37, 38 1 surface gauge 7/19/ : _14_13_01r03_B NAPTF22 3S 41, 42 2 surface gauge 7/19/ : _14_18_01r01_B NAPTF23 6N 47, 48 inclusion sound 7/19/ : _15_42_01r02_B NAPTF33 6S 45, 46 1 sound 7/19/ : _16_01_01r02_B NAPTF37 6S 51, 52 inclusion sound 7/19/ : _15_57_01r02_B NAPTF36 7N 55, 56 inclusion sound 7/21/2011 8: _08_05_01r01_B vid S 59, 60 inclusion sound 7/21/2011 7: _07_55_01r01_B vid S 65, 66 1 sound 7/21/2011 7: _07_51_01r01_B vid N 67, 68 2 sound 7/20/ : _16_03_01r01_B naptfday3_ N 69, 70 1 surface gauge 7/20/ : _15_56_01r01_B naptfday3_ N 71, 72 1 sound 7/20/ : _15_59_01r01_B naptfday3_ N 81, 82 1 sound 7/20/ : _15_29_01r01_B naptfday3_ N 85, 86 1 surface gauge 7/20/ : _13_14_01r01_B naptday N 87, 88 1 sound 7/20/ : _13_18_01r01_B naptday S 103, sound 7/20/ : _15_02_01r01_B naptfday3_ S 107, sound 7/20/ : _14_07_01r01_B naptfday3_2 001 Column H Comments systematic inclusion systematic inclusion systematic inclusion systematic inclusion 25

26 Column A Column B Column C Column D Column E Column F Column G Grid Location Subsurface condition Surface condition Date Time Stamp Name root Video location 10S 93, 94 1 surface gauge 7/20/ : _15_12_01r01_B naptfday3_ S 95, 96 1 sound 7/20/ : _15_08_01r01_B naptfday3_ N 113, sound 7/20/ : _12_25_01r01_B naptday S 111, sound 7/20/ : _12_48_01r01_B naptday S 117, sound 7/20/ : _12_42_01r01_B naptday N 121, sound 7/20/ : _10_33_01r01_B naptday S 125, sound 7/20/ : _10_47_01r01_B naptday S 127, sound 7/20/ : _11_03_01r01_B naptday S 131, sound 7/20/ : _10_59_01r01_B naptday N 133, sound 7/20/ : _10_13_01r01_B naptday N 135, surface gauge 7/20/2011 9: _09_36_01r01_B naptday N 137, 138 N/A 1 (away from joint) damage 7/20/2011 9: _09_33_01r02_B naptday N 147, sound 7/20/2011 9: _09_29_01r01_B naptday N 151, sound 7/20/2011 9: _09_18_01r02_B naptday N 153, sound 7/20/2011 9: _09_14_01r01_B naptday S 159, surface gauge 7/18/ : _16_29_01r02_B mira testing S 161, sound 7/18/ : _16_25_01r02_B mira testing S 169, surface gauge 7/18/ : _16_19_01r02_B mira testing S 173, surface gauge 7/18/ : _16_14_01r02_B mira testing N 11, 12 3 surface gauge 7/19/ : _13_33_01r04_B NAPTF17 21N 13, 14 3 surface gauge 7/19/ : _13_38_01r04_B NAPTF18 Column H Comments Check location: 33S on Map no measurements close to the joint (damage caused poor surface contact) 26

27 Column A Grid Location Column B Subsurface condition Column C Surface condition Column D Date Column E Time Stamp Column F Name root Column G Video location 21N 7, 8 3 damage 7/19/ : _13_22_01r04_B NAPTF15 21N 9, 10 1 surface gauge 7/19/ : _13_27_01r04_B NAPTF16 21S 17, 18 1 sound 7/19/ : _13_52_01r05_B NAPTF19 22N 23, 24 3 surface gauge 7/19/ : _14_51_01r02_B NAPTF27 22N 25, 26 1 surface gauge 7/19/ : _14_47_01r01_B NAPTF26 22N 35, 36 2 surface gauge 7/19/ : _14_55_01r01_B NAPTF28 22N 39, 40 2 surface gauge 7/19/ : _14_59_01r01_B NAPTF29 22S 31, 32 1 sound 7/19/ : _13_59_01r04_B NAPTF20 22S 33, 34 1 surface gauge 7/19/ : _14_03_01r02_B NAPTF21 25N 43, 44 2 sound 7/19/ : _15_47_01r02_B NAPTF34 25N 49, 50 inclusion sound 7/19/ : _15_52_01r02_B NAPTF35 25S 53, 54 inclusion sound 7/19/ : _16_05_01r02_B NAPTF38 26N 57, 58 inclusion sound 7/21/2011 7: _07_58_01r01_B vid N 63, 64 1 sound 7/21/2011 8: _08_01_01r01_B vid S 61, 62 inclusion sound 7/21/2011 7: _07_48_01r01_B vid N 73, 74 1 sound 7/20/ : _16_07_01r02_B naptfday3_ N 75, 76 1 surface gauge 7/20/ : _16_10_01r01_B naptfday3_ N 77, 78 1 surface gauge 7/20/ : _15_37_01r01_B naptfday3_ N 79, 80 1 surface gauge 7/20/ : _15_33_01r01_B naptfday3_ S 83, 84 1 sound 7/20/ : _15_25_01r01_B naptfday3_2 008 Column H Comments measurements at wheel path of single slab loading measurements at wheel path of single slab loading systematic inclusion systematic inclusion systematic inclusion systematic inclusion 27

28 Column A Column B Column C Column D Column E Column F Column G Grid Location Subsurface condition Surface condition Date Time Stamp Name root Video location 29N 101, surface gauge 7/20/ : _13_28_01r01_B naptday N 105, surface gauge 7/20/ : _13_30_01r02_B naptday N 89, 90 2 sound 7/20/ : _13_24_01r01_B naptday N 91, 92 2 surface gauge 7/20/ : _13_21_01r01_B naptday S 97, 98 1 sound 7/20/ : _15_16_01r01_B naptfday3_ S 99, surface gauge 7/20/ : _15_20_01r01_B naptfday3_ N 109, damage 7/20/ : _12_28_01r01_B naptday N 115, sound 7/20/ : _12_32_01r01_B naptday S 119, sound 7/20/ : _12_52_01r01_B naptday N 123, sound 7/20/ : _10_43_01r01_B naptday N 129, sound 7/20/ : _10_40_01r01_B naptday N 139, sound 7/20/2011 9: _09_54_01r01_B naptday N 141, damage 7/20/2011 9: _09_49_01r01_B naptday N 143, surface gauge 7/20/2011 9: _09_58_01r01_B naptday N 145, surface gauge 7/20/ : _10_02_01r01_B naptday S 149, sound 7/20/ : _10_06_01r01_B naptday N 155, sound 7/18/ : _16_38_01r03_B mira testing N 157, surface gauge 7/18/ : _16_34_01r02_B mira testing N 167, surface gauge 7/20/2011 9: _09_10_01r01_B naptday N 171,172 1 surface gauge 7/18/ : _16_52_01r02_B mira testing S 163, surface gauge 7/18/ : _15_36_01r04_B mira testing S 165, surface gauge 7/18/ : _15_30_01r08_B mira testing 018 Column H Comments 28

29 Table 3 shows the distribution of damage levels at the sensor locations. It can be observed that 1 sensor location (slab 16N sensors 137 and 138) testing area was considered not applicable N/A. At this location the surface damage did not allow for coupling of the transducers to the pavement to allow for scans closer than 5 in. from the joint. It should be noted that the incomplete grid testing indicated that the damage had not extended further at the time of MIRA testing. It can also be observed that over 80 percent of the measured locations were diagnosed to be in sound concrete condition (level 1). The remaining 16 locations were diagnosed to be damage levels 2, 3, or 4. Six of these locations were either in the wheel path or in the same slab where loading was applied and bottom-up cracking was purposefully being initiated and propagated in order to determine the strength of the pavement structure. Both level 4 damage locations were observed in the wheel path of this loaded slab. This leaves 10 unloaded sensor locations diagnosed with damage levels of 2 or 3. Including the location where surface damage was observed and MIRA measurements could not be taken close to the joint (slab 16N sensors 137 and 138), about 13% of the sensor locations were diagnosed to have damage not initiated by mechanical loading. Table 3. Summary of subsurface damage at sensor locations. Damage Damage Damage Damage Inclusion level 1 level 2 Level 3 level N/A Figures 17 through 21 show the layout of the slabs with color coded diagnoses to give a view of the distribution of the subsurface damage diagnoses. Sensor locations with a red outline box with no fill indicates a sound concrete (level 1). A solid red filled box indicates a sensor location where level 4 damage was diagnosed. A semi-transparent red filled box indicates a sensor location where level 3 damage was observed. A semi-transparent orange filled box indicates that a level 2 damage level was diagnosed at that sensor location. The Purple boxes indicate locations where the inclusion damage level was diagnosed. The green box indicates the sensor location where the damage at the surface did not allow for a full grid of MIRA testing. The 0.72 and 0.87 locations with a red transparent background indicate the two locations where 2D-UTSA analysis indicated damage at the express testing locations.

30 Figure 17. Slabs 2, 3, 21, and 22 North and South. Figure 18. Slabs 6, 7, 25, and 26 North and South. 30

31 Figure 19. Slabs 8, 9, 10, 27, 28, and 29 North and South. Figure 20. Slabs 13, 14, 32, and 33 North and South. 31

32 Figure 21. Slabs 15, 16, 17, 34, 35, and 36 North and South. 32

33 Conclusions Ultrasonic tomography testing of the CC6 section of the NAPTF and subsequent analysis showed that locations with and without sensors both experienced a high percentage of sound concrete condition (87% and 98% respectively). Therefore, the results suggest that the popouts and surface damage experienced at various locations prior to load applications is not a systematic problem and should not be generalized to all locations with those types of sensors installed. Nonetheless, while both locations with and without sensors showed a low percentage of damage, it is important to note that a significantly higher percentage of PCC locations with embedded sensors were diagnosed to be damaged. The analysis indicates that approximately 13% of the slab locations with embedded sensors experienced portions of subsurface concrete with at least a moderate level of damage prior to mechanical loading. This indicates that while the embedded sensors did not inherently damage the surrounding concrete it did increase the likelihood of damage. The following sensors were categorized to be surrounded by damaged concrete prior to mechanical loading: Slab 3S sensors 41, 42 Slab 8N sensors 67, 68 Slab 17N sensors 153, 154 Slab 25N sensors 43, 44 Slab 29N sensors 89, 90 and 91, 92 Slab 32N sensors 109, 110 Slab 35N sensors 141, 142 and 145, 146 Slab 2N sensors 3, 4 The analysis presented above utilized recently developed data analysis tools (SAFT-Panoramic and SAFT-3D). Although these tools permitted a more comprehensive evaluation than SAFT B- scan analysis, even more detailed analysis have recently become possible due to the introduction of SAFT-Full Waveform analysis. This tool is capable of differentiating between soft (i.e. flaws) and rigid (i.e. metal) inclusions in the concrete and giving automated thickness and dowel location diagnoses. Use of this tool was outside of the scope of this study, but it is possible that additional useful information could be extracted using this tool using the data which has already been collected. It would also be useful to collect additional data after load applications to get a characterization of the subsurface damage accumulation after load repetitions with respect to the pre-loading condition. 33

34 References Hoegh K., Khazanovich L., Yu H.T. Ultrasonic Tomography Technique for Evaluation of Concrete Pavements. Transportation Research Record: Journal of the Transportation Research Board, No. 2232, pp Hoegh, K., Khazanovich, L., Correlation Analysis of 2D Tomographic Images for Flaw Detection in Pavements. ASTM International. Journal of Testing and Evaluation. Volume 40. Issue 2. March Hoegh K., Khazanovich L., Yu H.T. CONCRETE PAVEMENT JOINT DIAGNOSTICS USING ULTRASONIC TOMOGRAPHY. TRB Paper Accepted for publication at the Transportation Research Record: Journal of the Transportation Research Board, Hoegh K., Khazanovich L., Maser K., Tran N. EVALUATION OF AN ULTRASONIC TECHNIQUE FOR DETECTING DELAMINATION IN ASPHALT PAVEMENTS. TRB Paper Accepted for publication at the Transportation Research Record: Journal of the Transportation Research Board,