Mapping Voids, Debonding, Delaminations, Moisture, and Other Defects Behind or Within Tunnel Linings (2013)

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.

18 C h a p t e r 3 an Investigation for Detecting Delaminations, Voids, and Water Intrusion Introduction As indicated in Chapter 2, this investigation used several NDT techniques—the ultrasonic linear array system, air-coupled GPR, ground-coupled GPR, thermal camera, and the portable seismic property analyzer—to detect defects in concrete, shot- crete, and steel test specimens. Ultrasonic Tomography The results of the ultrasonic tomography testing are contained in Appendices M and N. As indicated in Appendices D and M, the team concluded that the system is effective in detecting defects but with the following limitations: • Speed of data acquisition is low (0.8 to 2.3 min/sq ft). • The system provides no phase change information to infer defect type. • No information deeper than initial air interfaces is discernible. • The system has difficulty detecting reinforcement below two layers of reinforcement mesh. • For a 50-kHz use, defects under 2 in. from the surface are not directly detected. • For a 50-kHz use, reinforcement under No. 5 (0.625-in. diameter) is not typically detected. Air-Coupled GPR For this investigation, TTI personnel used the specimens described previously (see Tables 2.1 through 2.3). Details of the results are in Appendix K. The team used a 1-GHz central frequency device owned by TTI and determined that the equipment could detect only three simulated voids, all of them located in the shotcrete sections. Those specimens are the following: • Specimen D, air-filled void placed 7.625 in. from the surface; • Specimen F, air-filled void placed 3 in. from the surface; and • Specimen G, water-filled void placed 3 in. from the surface. The equipment could not detect delaminations or voids in the other specimens. The delaminations in the specimens did not contain significant air pockets or moisture, so GPR would not be effective in any case. The team estimated the depth to the defect using air-coupled GPR analysis software developed by TTI. For Specimen D, the estimated depth is 7.7 in. For Specimen F, the estimated depth is 2.6 in. For Specimen G, the estimated depth is 2.7 in. The team also collected air-coupled GPR data on a 12-in.- thick plain concrete specimen placed on top of a steel plate with a 1-sq ft void in the center of the plate. The team deter- mined that the equipment could not locate this defect. The team repeated the test with a 15-in.-thick specimen with two layers of reinforcement. Again, the team determined that the equipment could not locate the defect. Although layer depth information, areas of moisture, and areas of low material density can possibly be measured with air-coupled GPR, the team recommends using surface dielec- tric measurements from this device to determine areas to test with other devices. Normal concrete has a dielectric value usu- ally between 8 and 12. Values above this range indicate exces- sive moisture; values below this range indicate lower than normal material density (i.e., more air voids). Air has a dielec- tric value of 1; water has a dielectric value of 81. Ground-Coupled GPR The research team collected data by using a 900-MHz ground- coupled GPR on five reinforced concrete specimens. Three specimens had simulated 1-sq-ft delaminations, one speci- men had a simulated 1-sq-ft air-filled void placed 8 in. from Findings and Applications

19 the surface, and the final specimen had a 1-sq-ft water-filled void placed 8 in. from the surface. The team determined that the equipment could not locate the defects. The delaminations in the three specimens did not contain significant air pockets or moisture, so GPR would not be effective in any case. The voids in the other two specimens were located under a layer of reinforcement that consisted of No. 9 rebar placed at an 8 in. spacing in both directions. The ground-coupled GPR could not see through this layer of reinforcement. However, as docu- mented in the literature, ground-coupled GPR is effective in detecting voids and significant delaminations in concrete, provided the correct device is used. The team also collected ground-coupled GPR data on a 12-in.-thick plain concrete specimen placed on top of a steel plate with a 1-sq-ft void in the center of the plate. The team determined that the equipment could not locate this defect and repeated the test with a 15-in.-thick specimen with two layers of reinforcement. Again, the team determined that the equipment could not locate the defect. However, as described in Appendix Q, the ground-coupled GPR data showed defects in tunnel linings relatively near the tunnel lining surface. Thermal Camera In this investigation, the TTI team used a FLIR T300 infrared camera owned by TTI. The team collected infrared images on the specimens during the daytime and nighttime. Details of the results are in Appendix L. The camera images indicated defects in three shotcrete spec- imens: (1) the 3-in.-deep air-filled void (Specimen F), (2) the 3-in.-deep water-filled void (Specimen G), and (3) the 1-in.- deep delamination (Specimen L). The image for Specimen F was the most distinct. The images did not indicate the defects in the other specimens. The team noted that surface texture influenced the surface temperature measured by the camera. Portable Seismic Property Analyzer—Impact Echo and Ultrasonic Surface Waves The IE and USW results on the TTI specimens are shown in Appendix U. As an example, a USW planar contour map and an IE spectral B-scan on selected intact concrete and shotcrete slabs are shown in Figure 3.1. In spite of the Figure 3.1. PSPA results on 12-in.-thick intact concrete and shotcrete slabs. Longitudinal Distance from Center (in) Tr an sv er se D is ta nc e fro m C en te r ( in) -20 -16 -12 -8 -4 0 4 8 12 16 20 -20 -16 -12 -8 -4 0 4 8 12 16 20 4000 4500 5000 5500 6000 6500 7000 7500 8000 Longitudinal Distance from Center (in) Tr an sv er se D is ta nc e fro m C en te r ( in) -20 -16 -12 -8 -4 0 4 8 12 16 20 -20 -16 -12 -8 -4 0 4 8 12 16 20 2000 2500 3000 3500 4000 4500 5000 5500 6000 Longitudinal Distance from Center (in) Fr eq ue nc y (H z) -20 -16 -12 -8 -4 0 4 8 12 16 20 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Longitudinal Distance from Center (in) Fr eq ue nc y (H z) -20 -16 -12 -8 -4 0 4 8 12 16 20 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (a) USW average modulus in concrete (b) USW average modulus in shotcrete (c) IE B-scan in concrete (d) IE B-scan in shotcrete

20 heterogeneity of the shotcrete slabs, the contour maps of the variations in average USW modulus and dominant IE fre- quency exhibited reasonable uniformity for intact slabs (both concrete and shotcrete). In most cases the variation in modu- lus with depth was quite small. The reported thicknesses from spectral B-scan agreed well with the actual slab thicknesses. However, the peak frequency along the centerline varies more significantly in the shotcrete slab than the concrete slab, mostly because of the heterogeneity of shotcrete. In spite of the effectiveness of the IE method in estimating the slab thickness, this method, as configured in the PSPA, cannot estimate the thickness of slabs that are thicker than 18 in. or thinner than 6 in. The manifestations of shallow delaminated zones or voids were quite apparent on the time records collected by the device. USW and IE contour maps on selected defective con- crete and shotcrete slabs are shown in Figure 3.2. Planar maps of both methods provided confirmed shallow (3-in.) defects. When defects were deeper, the USW average modulus became less sensitive to the presence of defects, while the thickness mode (as opposed to the flexural mode) of the IE method became more effective. This occurs because surface waves propagate along a cylindrical front and thus become less sensitive to horizontal discontinuities with depth. Deep defects (deeper than 6 in.) were not readily detectable from the USW results. However, they could be readily identified through the IE results. Because of the size of the specimens, reflections from the vertical boundaries sometimes affect the frequency content of the signal. The PSPA software contains appropriate filters to minimize the effect of these reflections as long as the slab is not very thick and the PSPA is located at an adequate dis- tance from the boundary. Field Validation testing of NDt Devices by Using actual tunnels Introduction This section summarizes the results of the following: • A pilot project for the SPACETEC equipment; • Initial tests with air-coupled GPR and thermal cameras in Finland; and • Tunnel testing in Texas, Virginia, and Colorado. Figure 3.2. Contour maps of USW average modulus and IE dominant frequency in concrete and shotcrete slabs with embedded delamination 3 in. from top surface. (a) USW results in 15-inch-thick concrete (b) USW results in 12-inch-thick shotcrete (c) IE results in 15-inch-thick concrete (d) IE results in 12-inch-thick shotcrete

21 SPACETEC Pilot Project The research team conducted the SPACETEC pilot project in the Chesapeake Channel Tunnel during the night of April 11–12, 2011. The TS3 scanner was installed on the roof at the rear of the inspection vehicle, providing an undisturbed 360° mea- surement. The highest resolution—10,000 pixels—was used for an appropriate imaging of fine-scale features. A full traffic closure was not possible. Thus, the recording was performed twice: first in the north–south direction of the lane to Virginia Beach and second in the direction of the east- ern shore of Virginia. Traffic could pass the inspection vehicle, as is visible in the recordings. Appendix I contains the results of this testing. SPACETEC personnel provided a copy of the TuView software that is used to analyze the data from this equipment and indicated areas of concern in the data files that the software displays. The research team was interested in the infrared images from this equipment, but this SHRP 2 study does not involve evaluating profile or visual images. The team discussed the results of this testing with Chesa- peake Bay Bridge-Tunnel (CBBT) personnel using the TuView software. CBBT personnel and the team compared the SPACETEC infrared images with CBBT construction plans for a tunnel tile replacement project. In addition, the SPACETEC equipment operator reviewed the infrared images immedi- ately after collecting the data and noticed an area on the tun- nel wall that appeared to have a defect. That image is shown in Figure 3.3. The team evaluated this area using impact echo and deter- mined that a problem did appear to be present in this area. The team compared the SPACETEC thermal images with the hammer sounding results. Ninety-seven percent of areas covering more than 50 tiles could be detected, compared to 55% for areas covering less than 50 tiles. An additional analysis was performed to investigate why some of the debonded areas were not detected in the SPACETEC data. Small debonded areas covering less than 20 tiles were not reli- ably detectable in the thermal images obtained during this particular survey. When larger debonded areas were missed, the top two factors were found to be reflection of light from the surface of the tiles (at certain scanning angles) and inter- ference with the temperature gradient in front of air vents. The great advantage of such scanning operations is the speed with which they can be performed: the SPACETEC survey took about 1 hour at 1.5 km/h (1 mph) compared with the tedious hammer sounding survey, which took one man- month. Appendix V contains the results of this analysis. To summarize, the team’s analysis suggests that a combina- tion of thermal and visual imaging offers an alternative to the tedious practice of hammer sounding on individual tiles to determine tile debonding. Initial Tests with Air-Coupled GPR, Thermal Cameras, and Laser Scanners in Finland The tests in Finland concentrated on the technical feasibility of air-coupled GPR systems, thermal cameras, and laser scanners— as well as their integrated analysis—for monitoring tunnel lin- ing conditions. The idea was to test whether these systems can provide reliable and repeatable data and to collect information on the potential sources of error in these techniques. Another goal for these tests was to provide basic information on the potential defects, such as moisture problems close to the surface of tunnel lining structures. The tests were carried out in two tunnels in the Helsinki area in Finland. One tunnel has a con- crete lining, and the other tunnel lining is made of shotcrete. The two tunnels were selected to determine whether air- coupled GPR can be used in different types of tunnel lining measurements. The research team used the same air-coupled GPR data collection settings as normally used in pavement thickness and quality control surveys. Preprocessing of the collected data was done using standard methods, including automatic air-coupled elevation and amplitude correction, background removal, and vertical time domain filtering. The standard GPR data analysis consisted of reflection amplitude and dielectric value calculations and their analysis. The same two tunnels were also used to determine how well digital thermal cameras can detect thermal anomalies in tunnel linings, pointing out areas of moisture anomalies, voids, or cracks. The team tried different data collection and analysis techniques to find an optimal survey method. The goal in the laser scanner tests was to determine whether the method could provide valuable information about the tunnel lining condition and shape. Although laser scanning is beyond the scope of this project, the results were of interest to the team. Figure 3.3. Infrared image from SPACETEC indicating area of concern.

22 The following findings are of particular interest for this study: • GPR horn antenna data provided good quality structural information from the concrete tunnel but could not be used in the shotcrete tunnel where steel fibers were used in the shotcrete. The GPR data provided useful information on structures behind the tunnel linings. • The optimum distance from the air-coupled GPR antenna to the tunnel lining surface is 0.5 m (19.7 in.). • The thermal camera gave excellent results in the shotcrete tunnel. However, in the new concrete tunnel, hardly any anomalies could be detected. One reason for this may be a lack of problems close to the surface. • The thermal camera results are repeatable, but tunnel wall surface temperature can change during the day, which could affect the results. • Thermal anomalies can be seen in different ways when the surveys are conducted in summer, fall, and winter. The best time to survey is early summer. However, results, surpris- ingly, showed that moisture anomalies could always be seen as colder areas. • The thermal camera is sensitive to the survey direction to the tunnel wall and roof, and focusing the camera on white tiles can be difficult. Also, the survey van can cause unwanted thermal reflections. • Laser scanning systems provided useful data on the shape and condition of the tunnel linings. The results were excel- lent, especially in the shotcrete tunnel, but interesting and valuable information was also detected in the concrete tunnel. Although layer depth information, areas of moisture, and areas of low material density can possibly be measured with air-coupled GPR, the researchers used surface dielectric mea- surements from this device to determine areas to test with in-depth devices. Normal concrete has a dielectric value usu- ally between 8 and 12. Air has a dielectric value of 1; water has a dielectric value of 81. Values above this range indicate exces- sive moisture; values below this range indicate lower than normal material density (i.e., more air voids). Appendix J reports the details of the tunnel testing in Finland. Tunnel Testing in Texas, Virginia, and Colorado The team conducted nondestructive testing in the following tunnels: • Washburn Tunnel, located under the Ship Channel east of Houston, Texas: The TTI team collected air-coupled GPR, ultrasonic tomography, and acoustic sounding data in this tunnel in September 2011. • Chesapeake Channel Tunnel, located east of Norfolk, Virginia: The team collected NDT data in this tunnel in September and October 2011. • Hanging Lake Tunnel, located on I-70 west of Denver, Colorado: The team collected NDT data in this tunnel in October 2011. • No Name Tunnel, located on I-70 west of Denver, Colo- rado: The TTI team collected air-coupled GPR data in this tunnel in October 2011. The following is a summary of the results from the tunnel testing. Air-Coupled GPR The team used the TTI 1-GHz air-coupled GPR system for collecting data in the tunnels listed above. In particular, the team collected data at 1-ft spacing in the plenums of the Ches- apeake Channel, Eisenhower Memorial, and Hanging Lake tunnels; and along the tiled roadway sections in the Chesa- peake Channel, Hanging Lake, and No Name tunnels. As men- tioned earlier, the research team was most interested in the surface dielectric measurements from this device. The team mounted the equipment on a cart for testing in the plenums and on a vehicle with a crane for testing in the roadway. Figure 3.4 shows results from testing on the Chesapeake Channel Tunnel roof. As shown in the figure, the surface dielec- tric varies, with significant peaks occurring in several areas. The research team was not able to test all areas because of time constraints. However, one high dielectric area selected for test- ing did contain a shallow delamination, though no visual dis- tress was present. Researchers could only collect air-coupled GPR data along the top of the tunnel roof. The presence of cables and conduits on the sides of the tunnel roof made it impossible to collect GPR data in those areas. Figure 3.5 shows results from testing on the Hanging Lake Tunnel roof. In this case, none of the surface dielectric values exceeded 11. However, peaks in the values occurred at several locations; these areas should be inspected more closely. This tunnel roof contained many cracks with moisture; however, the moisture usually was outside the GPR testing area. Again, the presence of cables and conduits on the sides of the tunnel roof made it impossible to collect GPR data in those areas. To summarize, the team recommends that the surface dielectric measurements from air-coupled GPR be used for scanning purposes to determine where more in-depth inspec- tion and testing may be useful. The team noted surface dielec- tric changes in both concrete and tile-lined tunnels. In general, the researchers recommend inspecting areas where the surface dielectric is greater than 11 or where significant

23 peaks or troughs in the dielectric value are observed. The team noted that the data analysis could indicate lining inter- faces and lining thickness estimates; however, actual defects within or behind the tunnel lining could not be readily deter- mined from the analysis. In addition, more work is needed to keep the antenna at a relatively constant distance from the lining to calculate reasonable surface dielectric values. Ideally, this distance should not vary more than 4 in. from the recom- mended distance (usually 19.7 in.). Appendix K contains the data analysis of air-coupled GPR data collected in tunnels. Thermal Cameras The team collected thermal images using both handheld and vehicle-mounted thermal cameras in all of the tunnels tested in this project. Both cameras were able to detect significant ther- mal changes that indicated possible problems at those locations, on both concrete surfaces and tile-lined surfaces. Figure 3.6 shows a thermal camera image from the top of the Eisenhower Memorial Tunnel. Cracks and stalactites containing moisture are indicated in light blue. The team recommends that the handheld thermal camera be used for scanning purposes where more in-depth inspection and testing may be desired. In particular, areas with images that contain significant thermal differences from the surrounding lining should be investigated. Appendix L contains more images from these devices. Ultrasonic Tomography Field evaluations of four public tunnels were conducted using the UST technique to evaluate natural structural defects within actual tunnel linings. The tunnels tested were the Eisenhower Memorial Tunnel, Hanging Lake Tunnel, Chesa- peake Channel Tunnel, and Washburn Tunnel. Because the UST technique does not have a testing methodology that is field ready, the system was first evaluated on the basis of its ability to detect simulated defects in specimens as well as other available sites (e.g., pavements, airport runways, bridge decks) where ground truth validation was available. After that testing, the system was taken to the field to evaluate natural structural defects within actual tunnel linings. The conclu- sions of the tunnel testing are as follows: • The UST system is exceptional at locating horizontal delam- inations ranging in thickness from 0.05 mm to 2.0 mm (0.002 in. to 0.079 in.) and is able to differentiate between Figure 3.4. Air-coupled GPR data for Chesapeake Channel Tunnel roof. Figure 3.5. Air-coupled GPR data for Hanging Lake Tunnel roof.

24 fully debonded and partially bonded areas within a single map based on the color distribution. It is not, however, able to measure the thickness of the delaminations directly. • Cracks were only clearly characterized when they formed nonperpendicular to the testing surface; however, the pres- ence of perpendicular cracks could be assumed by the omission of surface detail. Note that no crack depths were confirmed by ground truth validation, and that should be a focus of further research. • Backwall surfaces up to a depth of 965 mm (38 in.) were successfully and accurately determined. Assuming the plan details were correct (no ground truth validation was avail- able), the UST system predicted this depth within an accu- racy of 5 mm (0.3 in.). • Both air- and water-filled voids ranging from 76 mm to 203 mm (3 in. to 8 in.) in depth could be detected; but differentiation between the two was difficult because shear waves are not supported by air or water, and almost all of the acoustic energy is reflected by these types of voids. Fur- ther study could be done to analyze the difference between phase changes involving these two types of voids. • Reinforcement layout and depth was also successfully deter- mined, as long as the device was polarized in the correct direction. The only exception was some shotcrete applica- tions. When potentially porous materials such as the shot- crete specimens were evaluated, the presence of very small air voids made internal inspection difficult. • With the exception of some medium-size clay lumps (with a diameter of approximately 102 mm, or 4 in.) surrounding reinforcement, clay lump testing was highly successful. • The research team used two A1040 MIRA systems to com- pare the system’s abilities to reproduce the same wave speed. For a test involving 16 specimens, a strong positive correlation existed (with a coefficient of determination of 0.952), with a standard error of approximately 33 m/s (108 ft/s). • When detecting the depth of delaminations using the same device with the same testing procedures and input param- eters (e.g., wave speed, frequency, gain selection), measure- ments typically varied by 1 mm to 3 mm (0.04 in. to 0.12 in.). That variation is more likely to be explained by user error or user interpretation than by device error. The same is true for water-filled and air-filled voids. • The minimum area the MIRA system could test is tied to the size of the device: 370 mm by 170 mm (14.6 in. by 6.7 in.). Figure 3.7 shows an example of a scan from the Hanging Lake Tunnel. The researchers believe that the MIRA system is especially effective for mapping deeper defects and is recommended for situations where such deep defects are suspected. Results of Figure 3.6. FLIR T300 infrared image of the top of Eisenhower Memorial Tunnel. Figure 3.7. Ultrasonic tomography scan from Hanging Lake Tunnel.

25 tunnel testing using ultrasonic tomography are contained in Appendices M and N. Portable Seismic Property Analyzer The UTEP team used the PSPA, which can perform IE and USW tests simultaneously. USW Method (PSPA). After testing each tunnel point by point with the PSPA, the cross sections of variation of modulus with wavelength (or depth) were obtained for each tested sec- tion. As shown in Figure 3.8a, intact areas exhibit more or less constant modulus with depth. The average modulus was around 4,500 ksi. Figure 3.9a shows an example of USW results in a defective area of one of the tested tunnels. In this figure, the problematic areas manifested themselves as areas with lower average moduli. The depth of delamination could be approxi- mated through the B-scan in Figure 3.9a. In Figure 3.10a, the crack was recognized through high average moduli in the USW B-scan when the crack was between the source and the first receiver (because of the travel path of the wave). When the crack was between the two receivers, the reported USW modu- lus was lower than normal. The results for these points agreed well with the actual condition that was documented during visual inspection. The rest of the USW results for the tested tunnels are shown in Appendix P. IE Method (PSPA). Similar to the USW method, the IE results, in the form of a spectral B-scan, were visualized in con- tour maps. As shown in Figures 3.8b and 3.9b, a thickness fre- quency (around 3 kHz) governed the response of intact test points. Other points in Figure 3.9b exhibit either a lower or higher dominant frequency. The low-frequency flexural mode results from a shallow or a deep but an extensive delamination. Thus, its peak frequency does not correspond to any thickness measurement, and the depth of defect can be estimated from a USW B-scan (Figure 3.9a). Alternatively, the high frequency response is attributable to the onset of delamination. In that case, the depth of delamination can be estimated and confirmed with the USW B-scan. When a crack is present, data analysis is more complicated. As shown in Figure 3.10b, multiple frequen- cies were present in the response when a crack was between the source and receiver in an IE B-scan. The remaining IE results are shown in Appendix P. In most cases, the calculated depth and location of delamination agreed well with the USW results. Some exceptions occurred. Where the IE and USW analyses were not consistent, the dif- ferences were attributed to the edge effect near a crack and placement of the PSPA sensor unit relative to the crack. Ultrasonic Echo, Ground-Coupled GPR, and Impact Echo Testing Field testing using three nondestructive testing techniques was carried out between October 3, 2011, and October 12, 2011, in three tunnels in the United States: two in Colorado (Eisenhower Memorial Tunnel and Hanging Lake Tunnel) and one in Virginia (Chesapeake Channel Tunnel). In each tunnel, selected areas were tested using three nondestructive testing (NDT) techniques: GPR, ultrasonic echo, and IE. The allocated testing time in each tunnel was limited. The number and location of the test areas were selected based on either previous analysis or the existence of visual distress. The on- site working conditions were also taken into account. The different measurement techniques used by the Federal Institute for Materials Research and Testing (BAM) for this (a) USW dispersion curve (b) IE frequency spectrum Figure 3.8. PSPA results on an intact area in the Chesapeake Channel Tunnel.

26 Figure 3.9. PSPA results on a defective area in the Chesapeake Channel Tunnel. Longitudinal Axis Fr eq ue nc y (H z) 486+81 1 2 3 4 5 6 7 8 9 10 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 x 104 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1 2 3 4 5 6 7 8 9 10 2 3 4 5 6 7 8 9 10 11 12 Longitudinal Axis D ep th (in .) 486+81 2000 2500 3000 3500 4000 4500 5000 5500 6000 (a) USW dispersion curve (b) IE frequency spectrum Figure 3.10. PSPA results on a cracked area in the Chesapeake Channel Tunnel. Longitudinal Axis Fr eq ue nc y (H z) 473+56 1 2 3 4 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 x 104 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1 2 3 4 2 3 4 5 6 7 8 9 10 11 12 Longitudinal Axis D ep th (in .) 473+56 2000 2500 3000 3500 4000 4500 5000 5500 6000 (a) USW dispersion curve (b) IE frequency spectrum

27 project were mounted on an automated scanning device that BAM developed. Figure 3.11 shows the BAM scanner with the ultrasonic echo device. It can be carried in a relatively small, lightweight package. Its size allows the scanner to be trans- ported in cars and carried through small openings to reach difficult-to-access areas such as the vents above tunnels. The equipment commonly used for NDT of structures—including GPR, ultrasonic echo, and IE devices—can be easily attached to the scanner for testing and detached after completing the measurements. The scanning and NDT data acquisition are controlled by a single notebook. This simplifies the control and reduces the amount of equipment and weight of the mea- surement system. Appendix W summarized the findings and applications of the Federal Institute for Materials Research and Testing (BAM). For the Eisenhower Memorial tunnel plenum, the ground- coupled GPR proved to be the best tool for identifying and locating the reinforcement. However, the ultrasonic echo device was better at locating an anomaly of unknown origin than the ground-coupled GPR. A combination of the two result sets would provide the most detailed and reliable results. Both methods detected the reinforcement and an unknown anomaly. GPR was more effective in detecting the reinforcement, and ultrasonic echo was more effective in detecting the unknown anomaly. The backwall could not be seen with any of the employed techniques here. And the impact echo technique could not register either reinforcement or the anomaly detected by the other two techniques. For the Hanging Lake tunnel plenum, the ground-coupled GPR was the only method able to identify the reinforcement mesh and the reinforcing elements. The fine measurement grid and 3-D data collection allowed detection of reinforcing elements overlapping each other in some views. The ultra- sonic echo technique was able to detect a deeper anomaly and establish that the anomaly under the test area was located at different depths. No reliable information could be extracted from the impact echo data. Again, combining the results of the ground-coupled GPR and ultrasonic echo is desirable. Note that none of these NDT techniques were able to reliably identify the extent of the Hanging Lake tunnel lining. For the Chesapeake Channel tunnel plenum, the ground- coupled GPR proved to be the most reliable NDT method for detecting and identifying reinforcement bars but could not detect a 15-in.-deep localized anomaly. The ultrasonic echo technique was not as clear in detecting the steel bars but did indicate the presence of an anomaly. Both ultrasonic echo and impact echo could detect the thickness of the tunnel lin- ing. A clearer picture of the geometry and condition of the tunnel emerged using all three techniques. For the section of the Chesapeake Channel Tunnel road- way that was lined with tiles, the ground-coupled GPR signals were not disturbed by the presence of the tiles and could image the reinforcement mesh behind the lining. The impact echo signals carried useful information about the bonding condition at the tile-concrete interface and occasionally about the lining itself. The ultrasonic echo device, however, provided no useful information about the condition of the lining. The ultrasonic echo transducer was too large (4 in. by 3 in.) com- pared with the size of the tiles (2 in. by 2 in.). The grid location and spacing had to be adjusted so that meaningful data could be obtained. However, testing was interrupted by an unfore- seen weather condition, and no further measurements could be obtained with the ultrasonic echo device. To summarize, the automated scanning device that BAM used was effective in collecting NDT data in the tunnels with the three techniques. The team recommends that data from both the ground-coupled GPR and the ultrasonic echo devices be collected when conducting in-depth evaluations directly on concrete surfaces. However, for tiled surfaces, data from the ground-coupled GPR and impact echo should be collected together—the ultrasonic echo device may not work on tiled sur- faces because of the tile dimensions. These devices should be effective in collecting data on shotcrete linings as well. Appen- dix Q contains more information on the tunnel testing with these devices. Other Information Appendix R contains depth measurement estimates of appar- ent defects as indicated by the in-depth evaluation devices used in this portion of the research. The appendix also con- tains estimated depth measurements to reinforcing steel or the backwall of the tunnel lining if they were detected. an Investigation for Detecting Loose tiles and Moisture Underneath tiles As mentioned earlier, air-coupled GPR data on tiled linings in the Chesapeake Channel and Hanging Lake tunnels indicated high surface dielectric areas, greater than 11 (see Appendix K). Researchers tested some of those areas with ultrasonic tomog- raphy, impact echo, and hammer sounding. The researchers Figure 3.11. BAM scanner with ultrasonic echo device.

28 Figure 3.12. Surface dielectric versus surface rating (using Chesapeake Channel Tunnel results). Table 3.1. Permittivity Values (Real Portion) for a 1-GHz Frequency Water-to- Cement Ratio Relative Humidity (%) 100 85 75 63 43 0.4 17 16 15.5 14.5 12.5 0.5 15 12.7 12 11.8 9.9 0.6 20 15 10.9 10 8.5 Table 3.2. Permittivity Values (Real Portion) for a 2-GHz Frequency Water-to- Cement Ratio Relative Humidity (%) 100 85 75 63 43 0.4 15.5 15 14.5 13.5 12 0.5 14.5 13.5 11.5 11 9 0.6 18 14.9 10.2 9.8 7.5 found debonded tiles and delaminations in those areas. Thus, the team concluded that high surface dielectric measurements on tiled linings can indicate areas of debonded tiles or delami- nations, as well as areas of high moisture behind tiles. Also, as described earlier, the SPACETEC thermal imaging data can be useful for locating loose tiles. Thermal cameras can also indicate areas of loose tiles. As indicated in Appendix G, the TTI team is developing an acoustic sounding test to detect loose tiles. However, this method is still under development and is not recommended for implementation at this time. To summarize, the team suggests that air-coupled GPR, thermal cameras, and the SPACETEC system’s thermal images can be effective scanning devices to locate loose tiles and moisture underneath tiles. Developing NDt for Measuring Concrete permeability Appendix S contains the results of a laboratory study that attempted to correlate dielectric (or permittivity) measure- ments to concrete permeability. As indicated in the appendix, the team determined that the air-coupled GPR cannot mea- sure permeability directly in the field. However, Appendix S does contain information that can be used for future NDT development. In addition, using the results in Appendix L, the TTI team developed Tables 3.1 and 3.2 for the real portion of the permittivity measurement for cement paste. These can be related to the dielectric measurements made with the air- coupled GPR. Table 3.1 is for a 1-GHz frequency. Table 3.2 is for a 2-GHz frequency. The values in these tables can be used as a general guide. Although the measurements were made on cement paste, the team believes that the moisture content in the paste would have the greatest effect on dielectric readings with the GPR. Essentially, the tables suggest that air-coupled GPR dielectric readings above 11 may indicate a potential problem, and read- ings above 15 may indicate excessive moisture in the concrete. The team also attempted to measure resistivity on the con- crete and shotcrete specimens. However, the measured values varied widely. The team concluded that the concrete resistiv- ity device was suitable only for controlled laboratory testing purposes. Based on the observed distress in the Chesapeake Channel Tunnel, the team developed the relationship in Figure 3.12 that relates surface dielectric values measured in the tunnel to surface distress that is assumed to be caused by excessive mois- ture, leading to reinforcing steel corrosion and further distress. Admittedly, significant scatter is apparent in the data shown in Figure 3.12. However, Figure 3.12 could be useful in interpret- ing surface dielectric data for concrete. The surface rating is defined in Table 3.3.

29 Table 3.3. Surface Rating Based on Distress Observed Distress Observed Surface Rating Cracks, no staining 0 Cracks, light staining 1 Cracks, light staining, and light calcium carbonate deposits 2 Cracks, moderate calcium carbonate deposits and staining 3 Cracks, moderate calcium carbonate deposits and staining; potential spalling (<2 in.) 4 Cracks, moderate calcium carbonate deposits and staining; potential spalling (2–6 in.) 5 Cracks, moderate calcium carbonate deposits and staining; potential spalling (6–10 in.) 6 Cracks, moderate to heavy calcium carbonate deposits and staining; spalling (<2 in.) 7 Cracks, moderate to heavy calcium carbonate deposits and staining; spalling (2–6 in.) 8 Cracks, moderate to heavy calcium carbonate deposits and staining; spalling (6–10 in.) 9 Cracks, moderate to heavy calcium carbonate deposits and staining; spalling (>10 in.) 10