Tri-band ground penetrating radar for subsurface structural condition assessments and utility mapping D. Huston *1, T. Xia 1, Y. Zhang 1, T. Fan 1, J. Razinger 1, D. Burns 1 1 University of Vermont, Burlington, VT, USA * ABSTRACT This paper describes the operating principles and use of an impulse ground penetrating radar (GPR) system that operates with three center frequency bands so as to enable more agile sensing of structural conditions and utility mapping. The operation of a GPR is typically a tradeoff between low-frequency waves that penetrate well, but lack good spatial resolution with high-frequency waves that do not penetrate well but have better spatial resolution. A complication is that many GPRs use short-duration impulse-shaped signals with mixed time and frequency content. The electronic generation of impulse signals is hardware dependent with most units being capable of producing only a single type of electromagnetic wave with fixed frequency content. The result is a system with limited sensing capability. Many structures and subsurface conditions are unpredictable and better sensed with a variety of different wavelengths. A workaround presented here is a GPR system that combines three fixed frequency sensing bands with center frequencies of 400 MHz, 1.6 GHz and 2.3 GHz. This system can sense at three different depths and resolutions so as to provide a wider coverage of subsurface conditions. Results from tests aimed at sensing subsurface conditions in concrete structures, roadways and buried utility location will be presented. In each of these three cases the system performs differently and better or worse for different frequency bands. Also presented will be preliminary results from the development of a frequency-agile multi-static system that has the ability to adjust the pulse shape and frequency content of different waves, along with tracking phase-shifted time-domain methods for more rapid 3-D reconstructions of subsurface conditions. 1 INTRODUCTION Ground penetrating radar (GPR) is a nondestructive method of assessing subsurface conditions by launching, receiving and processing transient electromagnetic (EM) waves. Non-metallic structures, such as those made of reinforced concrete, pavement or soil can be well-suited for inspection with GPR. Since EM waves travel at the speed of light in a dielectric medium, the time, frequency, amplitude and spatial characteristics of these waves plays a large role in the sensitivity, penetration depth and spatial resolution In general, it is desirable for the GPR signals to have the following characteristics: 1. Short duration waveform transients for high-resolution down-range measurement The length of individual waveforms affects the spatial resolution in the downrange, i.e. depth, direction. This affects the ability to measure the thickness of thin layers and to identify small objects. 2. Short duration waveform bundles to reduce overall measurement times A typical single GPR measurement involves launching and receiving a bundle of individual waveforms. Reducing the overall duration of the measurement waveform bundles reduces the overall site measurement times and allows for increasing the speed of the GPR (Xu, 2013). 3. Wide frequency bandwidth for superior penetration depth and spatial resolution High-frequency waves impart superior spatial resolution, but tend to attenuate quickly with depth in lossy media, such as concrete or soil. For the
same amount of loss per cycle low-frequency waves penetrate deeper. Frequency-dependent absorption can lead to frequency-dependent penetration depth. Wide bandwidth signals can penetrate through a wide range of depth and frequency-dependent media. 4. Small amplitude GPRs inevitably emit electromagnetic waves into the environment. These radiated emissions can interfere with other instruments and are often the subject of government regulations. Reducing the launched signal amplitude reduces the radiated emissions. 5. Large dynamic range Many structures amenable to GPR testing contain elements made of lossy dielectric media. This leads to return signals with small amplitudes, which prompt the need for GPR instruments to measure over wide amplitude ranges without distortion. Many test environments contain ambient electromagnetic noise that can contaminate the return signals, especially those with small amplitudes. Launching and receiving of waves with precise control and sensing over large dynamic ranges helps to alleviate these concerns. 6. High cross-range spatial resolution The ideal EM plane wave has infinite lateral extent and no available cross-range resolution. EM waves that have a limited lateral spread have discernable cross range resolution that improves with the reduction in the launched wave spread. Diffraction effects place limits on the cross-range resolution using simple signal processing methods. More complicated sub-diffraction methods are possible. 7. Polarization sometimes it is important to control the polarization of the launched waveforms and sensitivity to polarization on the received signals to discern polarization-dependent subsurface features, such as the direction of an array of reinforcing bars. These desired characteristics conflict with one another. Building GPR systems requires compromises so that the properties of the test waves are best suited for a particular application. Most GPR systems are either impulse (I_GPR) or step-frequency (SF-GPR) (Huston, 2002). I-GPRs launch short-duration impulse signals. Ideally the impulse is a delta-function with infinite frequency-domain bandwidth, but in reality is an impulse with finite duration, T i, and finite frequency domain bandwidth, BW i, with upper and lower frequency cutoffs f u and f l, respectively, Figure 1. Nominally, the time-frequency behavior is BW i = f u f l 1 (1) T i Hardware limitations, such as antennas, impose additional restrictions on the upper and lower ends of the frequency spectrum. Testing with I-GPR requires instruments capable of large amplitude dynamic ranges, fast data acquisition, and often the launching and receiving of bundles of multiple signals for sampling and averaging (Xia, 2013). Most commercial GPRs use the I-GPR technique, primarily due to the overall speed of the signal processing and overall system costs, but have the disadvantage of limited control over frequency domain behavior, especially the operating band. T i Time Figure 1 Time-frequency relation for I-GPR SF-GPRs launch and receive sinusoidal waves. The technique uses bundles of waves with frequencies that step through a frequency band to provide wide bandwidth coverage. An inverse transform synthesizes a time domain impulse from a bundle of frequency domain data extracted from the bundle, Figure 2 and Figure 3. Time Figure 2 Time-frequency relation for single SF-GPR signal Frequency and Phase Bundle Inverse Figure 3 Frequency-time relation of signal bundles in SF-GPR synthesize time-domain impulse f l f u Frequency Frequency and Phase Synthesized Time Impluse
SF-GPRs tend to be slower and more expensive than I-GPRs, but offer greater control over frequencydomain content. SF-GPRs tend to be favored by academic researchers and appear in certain high-performance GPRs. A primary limitation of I-GPRs is that the circuitry and EM hardware tends to be fixed to a specific band of operation. Many testing situations can benefit from a wider bandwidth. In this context, the use of new triband I-GPR provides an enhanced frequency domain coverage. The concept is to combine three I-GPRs into a single instrument with each I-GPR covering a different part of the frequency spectrum, Figure 4. T i1 T i2 Time #1 #3 #2 #1 Figure 6 Trailer with 400 MHz and 1.6GHz antennas 3 TEST RESULTS The tri-band I-GPR provided measurement at a handful of structural and geotechnical sites. The following is a sampling of the results. Figure 7 shows the results of testing the floor of a reinforced concrete building. This floor extends over earth retaining walls and an open room underneath. The corresponding B-scans show different levels of detail and penetration depth corresponding to the different bands. T i3 Time #2 Frequency Time #3 Figure 4 Concept of tri-band I-GPR 2 TRI-BAND I-GPR SYSTEM An attempt at assembling a system with the characteristics of Figure 4 appears in Figure 5. Appearing in the figure is a 400 MHz and 1.6 GHz antenna for frequency bands #1 and #2, respectively. A supplemental 2.3 GHs antenna provided coverage of band #3. The primary components of the system came from commercial vendors and were adapted to the configuration. A key feature is to collect the data simultaneously and to register geometrically the relative positions of the various antennas and the structure under test. a.
b. Figure 8 Asphalt roadway with patch c. Figure 7 Tri-band I-GPR B-scan results from the floor of a reinforced concrete building: a. 400 MHz, b. 1.6 GHz, and c. 2.3 GHZ. The box corresponds to a co-located region in all three plots The next set of tests examined the subsurface conditions of an asphalt bicycle and pedestrian roadway in Burlington, VT, USA. This is a region prone to poor drainage and frost heaving of clay layers in severe winters. From the topside, there is a large patch covering a region of possible subsidence, Figure 8. B- scans of the repair patch at the different bands appear in Figure 9 and Figure 10. At 2.3 GHz, features of the patch appear distinct, but deeper features are not distinguishable. At 1.6 GHz, the patch remains distinct and deeper features appear. At 400 MHz, the patch is visible, but not distinct, while deeper features, such as the pavement-soil interface layer, are visible. Figure 9 B-scans of repair patch at 400 MHz and 1.6 GHz Figure 10 B-scan of repair patch at 2.3 GHz A final set of data concerns the construction site for a new bus station in Burlington, VT. This site, being in a city with modern-era settlement spanning over 250 years has many unknown subsurface features, including the possibility of older archeological sites. A GPR scan of the site turned up many features with the 400 MHz band identifying deeper items, and 1.6 Hz and
2.3 GHz bands successful at near surface depths. As an example of the 400 MHz band performance, Figure 11 shows a B-scan of the site prior to excavation. Features including buried pipes appear as hyperbolas, while a distinct soil layer at a depth of about 2 m. appears. Subsequent excavation revealed a soil layer at the predicted depth, Figure 12. This soil layer is believed to the bed of the ancient Champlain Sea that arose during the receding glaciers in the last ice age, 10,000 B.C.E. (approx.) Figure 11 B-scan of bus station construction site, St. Paul St, Burlington, VT, USA. Detected features include buried pipes and a distinct soil layer at approx. 2 m. 4 CONCLUSIONS This study presents the concept of multi-band I-GPR. The intent is to take advantage of the resolving capability of different frequency bands while retaining the speed, low-cost and ease of use of an I-GPR. ACKNOWLEDGEMENT This research was supported by the VT Agency of Transportation and the U. S. Army Research Laboratory and the U. S. Army Research Office under contract/grant number W911NF-13-1-0301. Figure 12 Soil layer identified with 400 MHz I-GPR, believed to be bed of ancient Champlain Sea REFERENCES Huston, D.R. Fuhr, P.L. Maser, K, Weedon, W.H. 2002. Nondestructive Testing Of Reinforced Concrete Bridges Using Radar Imaging Techniques, Final Research Report NETC 94-2, New England Trans.Consortium, DOI: 10.13140/RG.2.1.4937.2002. Xu, X. Xia, T. Venkatachalam, A. Huston, D. (2013) The Development of a High Speed Ultrawideband Ground Penetrating Radar for Rebar Detection, Journal of Engineering Mechanics, 139, 3, 272-285, DOI:10.1061/(ASCE)EM.1943-7889.0000458. Xia, T. Venkatachalam, A.B. Huston, D. (2012) A High Performance Low Ringing Ultra-Wideband Monocycle Pulse Generator IEEE Transactions on Instrumentation and Measurement, 61, 1, 261-266