WIDEBAND MICROWAVE IMAGING OF ... - Semantic Scholar
WIDEBAND MICROWAVE IMAGING OF CONCRETE NONDESTRUCTIVE TESTING
FOR
By Hong C. Rhim1 and Oral Bu¨yu¨ko¨ztu¨rk2 ABSTRACT: Radar has the potential of becoming a powerful and effective tool for the nondestructive testing of concrete structures. An advancement of the method can be achieved through the understanding of the interaction between electromagnetic waves and concrete, and the identification of optimum radar measurement parameters for probing concrete. For the determination of optimum parameters, systematic radar measurements are needed in tandem with the implementation of proper signal processing techniques for imaging. This paper presents the results of such radar measurements, which are made using a wideband imaging radar with different frequency ranges. Inverse synthetic aperture radar is used for the measurements of laboratory-size concrete specimens having three different types of internal configurations. The signal processing algorithms implemented to obtain two-dimensional and three-dimensional imagery of concrete targets from radar measurements are also discussed. For the concrete specimens and measurement setup used in this study, 2–3.4 GHz waveforms are found to be adequate for the concrete thickness measurement, 3.4–5.8 GHz waveforms are adequate for the detection of delamination, and 8–12 GHz waveforms are adequate for the detection of inclusions embedded inside concrete.
INTRODUCTION Structures deteriorate over time. This deterioration can occur slowly over the life cycle of a structure or suddenly, as damage from an earthquake. In either case, proper inspection and evaluation are needed prior to any rehabilitation, retrofit, repair, or replacement action being taken. Because of the growing number of aging infrastructures worldwide, there is an increasing need to develop reliable nondestructive testing (NDT) techniques for ascertaining the integrity of these infrastructures (Chong et al. 1990). The goal of any NDT technique is to detect deterioration and/or objects located at a certain distance below the surface in an optically opaque medium. In the radar method, the detectability of anomalies and the capability of a wave to penetrate concrete are affected by a variety of measurement parameters—center frequency, frequency bandwidth, polarization of the wave, measurement distance and angle, and the geometric and material properties of a target. This necessitates a study to examine the influence of each parameter on radar measurement results and to establish an optimum combination of the parameters. The application areas of the radar method used for concrete systems range from concrete thickness determination to condition monitoring and the imaging of inclusions inside concrete systems (Rhim 1995b; Rhim et al. 1995). The objective of the work described in this paper was to determine the optimum radar frequency ranges for concrete slab thickness determination, delamination detection, and rebar detection. The following is a presentation of the parameters for radar measurements, the concrete specimens used for the measurements, radar measurement setup, the imaging algorithms implemented, and radar measurement results. RADAR MEASUREMENT PARAMETERS Parameters that affect radar measurement results can be categorized into the following three groups: (1) characteristics 1
Assoc. Prof., Nondestructive Evaluation and Struct. Lab., Dept. of Arch. Engrg., Yonsei Univ., Seoul 120-749, Korea. E-mail: herhim@ yonsei.ac.kr 2 Prof., Dept. of Civ. and Envir. Engrg., Massachusetts Inst. of Technology, Cambridge, MA 02139. Note. Associate Editor: David Rosowsky. Discussion open until May 1, 2001. To extend the closing date one month, a written request must be filed with the ASCE Manager of Journals. The manuscript for this paper was submitted for review and possible publication on July 30, 1999. This paper is part of the Journal of Structural Engineering, Vol. 126, No. 12, December, 2000. 䉷ASCE, ISSN 0733-9445/00/0012-1451–1457/ $8.00 ⫹ $.50 per page. Paper No. 21584.
related to the incident waves of radar, such as center frequency, frequency bandwidth, and polarization; (2) characteristics related to the geometric and material properties of a target, such as size, shape, and electromagnetic properties; and (3) parameters related to measurement setup, such as measurement distance and angle (Rhim 1995a). The parameters of the center frequency ( fc), frequency bandwidth (B) of the incident wave, and electromagnetic properties (real and imaginary parts of a complex permittivity of concrete) are important, as they determine detection and penetration capabilities. According to the principles of radar measurement, the resolution of closely spaced objects can be obtained by narrowing the transmitted pulse width T in time and increasing the system bandwidth B in frequency so that BT ⬇ 1, thus yielding the range resolution r ⬵ c/2B, where c is the speed of light in space (3 ⫻ 108 m/s). A time-bandwidth product approximating unity is inherent in the class of pulse radars in which a carrier is amplitude modulated by a pulsed waveform. Resolution is the ability to distinguish between closely spaced objects. An arbitrarily high degree of range accuracy can be obtained by using signals with a large bandwidth or high level of energy (Skolnik 1979). The range resolution r, which can be achieved by a wideband radar in concrete, is determined by the following relationship (Mensa 1981): r in concrete = (c/兹εr)/2B
(1)
where r = range resolution; and εr = real part of the complex permittivity of concrete. In addition to the range resolution, synthetic aperture radar (SAR) or inverse synthetic aperture radar (ISAR) can generate cross-range imagery (Menon et al. 1993; Novak et al. 1993). The SAR concept employs a coherent radar system based on a single moving antenna to simulate the function of all of the antennas in a real linear antenna array. The SAR system produces high cross-range resolution imagery by processing the stored backscatter amplitude and phase information obtained by sequentially transmitting and receiving electromagnetic energy along a linear aperture path, re-creating in time a long linear antenna array. If a target itself is rotating and the radar is stationary, return signals can be processed in a manner similar to the one using SAR to re-create a high cross-range resolution target image. This technique is often referred to as inverse synthetic aperture radar. ISAR techniques have been used to produce target images for objects on rotating platforms, for satellites rotating in orbit, and even for ships at sea, which roll rhythmically on the sea surface (Eaves and Reedy JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1451
TABLE 1.
Resolutions of Radar System in Concrete
Resolution (1) Range resolution Cross-range resolution
2–3.4 GHz (mm) (2)
3.4–4.8 GHz (mm) (3)
8–12 GHz (mm) (4)
10.71 15.91
6.25 9.34
3.75 4.30
1987). In ISAR measurements, the rotational angle (⌬rad) of a target determines the cross-range resolution. The cross-range resolution that can be achieved by ISAR is determined by the following relationship: xr in concrete = [c/( fc兹εr)]/[2(⌬rad)]
(2)
where xr = cross-range resolution; fc = center frequency of an incident wave; and ⌬rad = angular rotation of a target in radians during measurement. The range and cross-range resolutions obtained from (1) and (2) are tabulated in Table 1. One of the important objects of using radar for NDT is to identify the proper combination of center frequency and frequency bandwidth suitable for a specific application. This must be considered because there is a tradeoff between detection and penetration capabilities as frequency changes. The frequency range and bandwidth available for the ISAR measurements used in this research are determined by the antenna used. In general, a bandwidth increases as a center frequency increases. The importance of finding the proper center frequency and frequency bandwidth suitable for a specific application of the radar method for the NDT of concrete is thus a primary objective of this research. Sets of wideband radar measurements are made using an inverse synthetic aperture radar. ISAR is used because of its ease and versatility in conducting measurements on laboratory size concrete specimens. The ISAR used for the measurements is capable of transmitting waves in a wide frequency range, from 0.1 to 18 GHz, and is fully polarimetric. For the research described, the following three different frequency bands are used: 2–3.4 GHz (S/C band), 3.4–5.8 GHz (C band), and 8– 12 GHz (X band). The frequency range and center frequency selected for each measurement are based on the standard range-frequency combination inherent in the antenna system used. Due to the nature of hardware, as center frequency increases, frequency bandwidth also increases. The three frequency bands used here represent low, medium, and high frequency levels in the microwave frequency spectrum. The output power of the ISAR is 20 dBm (a decibel relative to 1 mW) and the dynamic range is 50 dB.
to make sure that no moisture is present inside. As the specimens are dried inside a humidity-controlled laboratory, no microcracks appeared at the time of the radar measurements. The concrete specimens used for the radar measurements are shown in Fig. 1. A plain concrete specimen is used for the thickness determination of different sets of center frequency and bandwidth. A specimen with delamination is used to examine the possibility of assessing condition change inside concrete by using radar. The present specimen has a relatively large size of delamination in the initial stage of the research work. The size of the delaminations is determined by the frequency range used by the radar, as described in this paper. Finally, a specimen with three 25.4 mm diameter steel bars is measured to investigate the detectability of the bars. These measurement cases are listed in Table 2. Every type of material has a unique set of electromagnetic properties affecting the way in which it interacts with the electric and magnetic fields of electromagnetic waves. Concrete is a dielectric (nonmetallic) material. A dielectric material can be characterized essentially as having two independent electromagnetic properties—the complex permittivity ε* and the complex (magnetic) permeability *. In general, four independent measurements are necessary to establish the quantities of both the real and the imaginary parts of ε* and *. However, most common dielectric materials including concrete are nonmagnetic, making the permeability * very close to the permeability of free space (0 = 4 ⫻ 10⫺7 H/m). Thus, the focus herein is on the complex permittivity ε*. Electromagnetic properties affect the velocity, wavelength, and amount of reflection and attenuation (Kong 1990). It is important to note
CONCRETE SPECIMENS Laboratory-size concrete specimens with different internal configurations are used as targets for the measurements. The concrete specimens were cast with a water/cement/sand/coarse aggregate mix ratio of 1:2.22:5.61:7.12 (by weight). Type I portland cement was used. Coarse aggregates have a maximum size of 38.1 mm. The aggregates are round shaped and from the river. They are properly stored and used prior to the mixing. Concrete specimens have dimensions of 304.8 mm (width) ⫻ 304.8 mm (height) ⫻ 101.6 mm (depth). The age of the specimens at the time of the measurements was four weeks. The uniaxial compression strength of the specimen was 21 MPa at 28 days. Since the work presented here focuses on the radar measurements and signal processing techniques used for the nondestructive testing of concrete rather than on the mechanical or structural aspects of concrete, the strength of the concrete specimens is not a relevant factor. However, the specimens can be made to satisfy a certain level of strength in future studies. The specimens are air dried for two months 1452 / JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000
FIG. 1. Laboratory Size of Concrete Specimens: (a) Perspective of Specimen; (b) Section of Plain Concrete Specimen; (c) Section of Concrete Specimen with Delamination; (d) Section of Concrete Specimen with Three Steel Bars TABLE 2.
Radar Measurement Cases
Concrete specimens (1)
2–3.4 GHz (2)
3.4–5.8 GHz (3)
8–12 GHz (4)
101.6 mm thick plain concrete Concrete with 25.4 mm delamination at 50.8 mm depth Concrete with three 25.4 mm diameter steel bars at 50.8 mm depth
—
—
—
—
—
—
—
—
—
that these electromagnetic properties are not constant. They change with the frequency of incoming waves, temperature, moisture, and mixture of the material (Rhim and Bu¨yu¨ko¨ztu¨rk 1998). For the study presented in this paper, dry concrete specimens are used for the radar measurements. RADAR MEASUREMENT SETUP Sets of wideband radar measurements are made using an ISAR. The ISAR is used because of its capability in generating 2D imagery, and the ease and versatility with which it can be used to conduct measurements on laboratory-size concrete specimens. A schematic view of a measurement setup is shown in Fig. 2. A concrete specimen is placed vertically as a target at the top of a turntable made of Styrofoam. The turntable is capable of rotating the target. A monostatic wideband radar located at a distance sends and receives the wave. The distance selected has measurements meeting a far field criterion, in which both the incident wave and the reflected wave become plane waves for the measurements. The measurements were made by a sweeping frequency, starting from frequency f1 and ending with frequency f2, with 0.1 GHz increments at a fixed angle. The target is rotated one degree at a time for each frequency sweeping. This generates two-dimensional data as a function of angle and frequency. For typical measurements, the target is rotated 20⬚. The frequency sweeping provides range resolution, while the rotation of the target provides cross-range resolution. RADAR IMAGING OF CONCRETE SPECIMENS Upon the radar measurements of concrete specimens, sets of raw data are collected that contain information on the signals returned from the targets. By processing the raw data, one- and two-dimensional imagery can be obtained. The incident wave generated by the antenna of the ISAR system is a stepped frequency continuous wave. It has a pulse width of 20 ns and a power of 20 dBm. The antenna transmits the waves swept from f1 to f2. The raw measured data contain information on a target received by an antenna as a function of frequency at a fixed incident angle for one-dimensional imaging. Each entry in the data has the amplitude and phase change of the received signal at each frequency. This information is collected in frequency. To have an image, the data need to be converted into a time domain with an appropriate processing scheme (Rhim 1995a). First, the raw data are converted into a format of real and imaginary parts to form a complex number real = 10amplitude/20 ⭈ cos(phase) amplitude/20
imaginary = 10
⭈ sin(phase)
complex number = real ⫹ j ⭈ imaginary
(3a) (3b) (3c)
The converted reflected signal is windowed in a frequency domain using a Hanning window, which has the following property (Press et al. 1988): w[n] = 0.54 ⫺ 0.46 cos(2n/M); 0 ⱕ n ⱕ M
(4a)
w[n] = 0 otherwise
(4b)
FIG. 2.
Radar Measurement Setup
When a run of N-sampled points of periodogram spectral estimation is selected, an infinite run of sampled data cj is, in effect, multiplied by a window function in time; here, one is zero, except during the total sampling time N⌬, when it is unity. Because the square window function turns on and off so rapidly, there is a leakage at high frequencies. To remedy this situation, the input data cj, j = 0, . . . , N ⫺ 1 are multiplied by the window function wj, which changes more gradually from zero to a maximum and back to zero as j ranges from 0 to N ⫺ 1. Now that these data have values in the frequency domain, an inverse Fourier transform (IFT) is carried out to express the data in terms of range and cross range. As the aspect angle in two-dimensional imaging has a constant value according to frequency, a one-dimensional IFT is performed and a twodimensional IFT is applied to three-dimensional imaging in which data consist of two variables, such as aspect angle and frequency. RADAR MEASUREMENT RESULTS AND DISCUSSION In general, the radar method has the advantages of remote sensing, fast measurement, high resolution, sensitivity to metallic objects, and safety during measurements. In addition, the radar system used in this work can generate cross-range imagery. This can enhance the usefulness of the radar by providing imagery similar to optic images. Second, the radar system used here has versatility, in that both its center frequency and its bandwidth are adjustable. This can be effective in optimizing detection and penetration capabilities in the NDT of concrete. Based on these features, three possible application areas of the radar method have been identified and presented in the following text. Determination of Concrete Thickness The measurement of concrete thickness is a basic but important application in the use of radar for the NDT of concrete systems. This thickness measurement can be used as a tool for quality-control purposes after the construction of pavement, for example. Also, by measuring the remaining thickness of concrete components such as bridge decks, foundations, abutments, and retaining walls, the degree of deterioration occurring can be assessed. Radar measurement results on a 304.8 mm (width) ⫻ 304.8 mm (height) ⫻ 101.6 mm (depth) dry concrete specimen without any inclusion are shown in Figs. 3(a)–5(b). The measurements are made at the following three different frequency ranges: 2–3.4, 3.4–5.8, and 8–12 GHz. Three-dimensional images are obtained by rotating the target 20⬚. The 3D images are shown in Figs. 3(b), 4(b), and 5(b). Two-dimensional images are obtained from the 3D images by selecting data at a normal incident angle, and are shown in Figs. 3(a), 4(a), and 5(a). By examining the results in Figs. 3(a)–5(b), the peaks in Figs. 3(a) and 3(b) are found to be wider than the ones in Figs. 4(a)–5(b) due to the narrower bandwidth of 1.4 GHz in a lowfrequency range, compared to the 2.4 and 4 GHz bandwidth in the medium- and high-frequency ranges, respectively. The narrower bandwidth gives less resolution, as in (1). It is also evident that the amplitude of the largest peak, which comes from the front surface in Figs. 3(a) and 3(b), is smaller than the ones in Figs. 4(a)–5(b) due to less scattering in the lowerfrequency range. However, the reflection from the back surface is clearly captured in the 2–3.4 GHz measurement. This is due to the fact that less attenuation occurs through the concrete thickness during wave travel, and that the waves are less sensitive to the edges of the specimens in the lower-frequency JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1453
turned signals from the specimens with delaminations compared to sound concrete specimens is related to the condition change occurring inside the concrete. Radar measurement results of concrete specimens with a delamination are shown in Figs. 6(a)–8(b). These results show the feasibility of using radar for monitoring condition change inside concrete at a distance. The existence of the 25.4 mm thick delamination at 50.8 mm depth is clearly seen as a peak in between the front and back surface reflections in Figs. 7(a)– 8(b). The results are obtained from 3.4–5.8 GHz and 8–12 GHz measurements, respectively. However, the measurements
FIG. 3(a). Two-Dimensional Plot of 1D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 2–3.4 GHz
FIG. 4(b). Three-Dimensional Plot of 2D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 3.4–5.8 GHz
FIG. 3(b). Three-Dimensional Plot of 2D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 2–3.4 GHz
FIG. 5(a). Two-Dimensional Plot of 1D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 8–12 GHz
FIG. 4(a). Two-Dimensional Plot of 1D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 3.4–5.8 GHz
range. Thus, 2–3.4 GHz waveforms are found to be adequate for concrete thickness measurement. Condition Monitoring of Concrete Deterioration in concrete severely affects the service life, safety, and maintenance costs of concrete structures. Remote monitoring of a deterioration process in concrete, including the detection of delaminations, can be studied through the radar measurements of laboratory size concrete specimens with artificially created delaminations inside. The change in the re1454 / JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000
FIG. 5(b). Three-Dimensional Plot of 2D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 8–12 GHz
This technique can be further developed and applied to the detection of subsurface condition change. Detection of Steel Bars inside Concrete The detection of steel bars embedded inside concrete for reinforcement has been one of the major goals in nondestructive testing methods for concrete. The use of radar in detecting
FIG. 6(a). Two-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 2–3.4 GHz
FIG. 7(b). Three-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 3.4–5.8 GHz
FIG. 6(b). Three-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 2–3.4 GHz
FIG. 8(a). Two-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 8–12 GHz
FIG. 7(a). Two-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 3.4–5.8 GHz
at 2–3.4 GHz in Figs. 6(a) and 6(b) were not sensitive enough to pick up the change. More measurements with delaminations at different depths indicate that 3.4–5.8 GHz waveforms, as well as 8–12 GHz waveforms, are effective in monitoring condition change. It can be suggested that the condition monitoring of a concrete target over a period of time is possible because the change in the reflected signal represents a change in quality.
FIG. 8(b). Three-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 8–12 GHz JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1455
presence of the widest bandwidth of 4 GHz, as shown in Figs. 11(a) and 11(b). Discussion The radar method has several advantages in NDT. These include remote sensing without contacting objects, fast measurement at the speed of light, achievement of high resolution, sensitivity to metallic objects, and safety during measurements. A radar system can also be versatile because it adjusts
FIG. 9(a). Two-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 2–3.4 GHz
FIG. 10(b). Three-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 3.4–5.8 GHz
FIG. 9(b). Three-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 2–3.4 GHz
FIG. 11(a). Two-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 8–12 GHz
FIG. 10(a). Two-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 3.4–5.8 GHz
steel bars is advantageous due to the sensitivity of electromagnetic waves to metallic objects and the versatility of using appropriate electromagnetic wave polarization with respect to the orientation of the bars. In Figs. 9(a)–11(b), radar measurement results on a concrete specimen with steel bars are presented. The waveforms with 2–3.4 GHz and 3.4–5.8 GHz were not sensitive enough to detect bars, as shown in Figs. 9(a)–10(b). However, 8–12 GHz waveforms detected bars with good resolution due to the 1456 / JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000
FIG. 11(b). Three-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 8– 12 GHz
its center frequency and bandwidth, which can optimize detection and penetration capabilities. A comparison of radar measurement results to a computer simulation based on numerical modeling suggests that simulation provides reference imagery for both the prediction and the interpretation of radar measurement results (Bu¨yu¨ko¨ztu¨rk and Rhim 1995). Expanding a database for computer simulation and radar measurements can identify signals for unknown targets. Also, by examining reflection coefficients, one can determine whether a specific peak belongs to a reflection from the front surface, the back surface, or the delamination inside concrete. In this, dielectric constants of concrete play an important role because they can determine the distance between the peaks. The radar measurement results reported in this paper imply that certain requirements be considered for hardware system design. As shown in the results, a specific combination of radar measurement parameters can provide or achieve a specific NDT goal. For example, if the purpose of the measurements is thickness detection, and a typical dimension of the thickness is known, a proper set of measurement parameters can be used instead of having to use all of the features in a hardware system. Handy, portable radar measurement equipment should be considered as a simple, practical means of employing this method, rather than having to use complex machinery. CONCLUSION A series of radar measurements are made on laboratory-size concrete specimens using ISAR. The measurement setup was such that the antenna used was located at a distance from a concrete target for remote measurement. Three different frequency ranges are used for the measurements at 2–3.4 GHz, 3.4–5.8 GHz, and 8–12 GHz. An application of the radar method is studied for concrete thickness detection, condition monitoring, and steel bar detection. As results, suitable frequency ranges are identified for each of the three specific problem areas. For thickness detection using the laboratory-size specimens and radar setup described in this paper, 2–3.4 GHz waveforms provide good results; moreover, 3.4–5.8 GHz waveforms are good for detecting back surface reflection with a reduced peak. For condition monitoring inside concrete, 3.4–5.8 GHz and 8–12 GHz waveforms are shown to be sensitive. For steel bar detection, 8–12 GHz waveforms are suitable because of their high resolution. ACKNOWLEDGMENTS This work was partially supported by the Korea Earthquake Engineering Research Center, funded by the Korea Science and Engineering Foundation (KOSEF). Computational facilities were provided by the Advanced Building Science and Technology Research Center in the College of Engineering, Yonsei University, Seoul, Korea. The writers would like to thank You Sok Kim of the Nondestructive Evaluation and Structures Laboratory at Yonsei University for his help in preparing this paper through an internship provided by KOSEF. Special thanks are due to Prof. Jin A. Kong in the Department of Electrical Engineering and Computer
Science at Massachusetts Institute of Technology (MIT), and Dr. Robert T. Shin at MIT Lincoln Laboratory. The raw data of the ISAR measurements were obtained at MIT Lincoln Laboratory through a research project supported by the U.S. Army Corps of Engineers, Waterways Experiment Station, Vicksburg, Mississippi through contract number DACW39-92-K-0029. The writers would like to thank Dr. Tony C. Liu, Mitch Alexander, and Kenneth Saucier at the U.S. Army Corps of Engineers for the supported provided.
APPENDIX I.
REFERENCES
Bu¨yu¨ko¨ztu¨rk, O., and Rhim, H. C. (1995). ‘‘Modeling of electromagnetic wave scattering by concrete specimens.’’ Int. J. Cement and Concrete Res., 25(5), 1011–1022. Chong, K. P., Scalzi, J. B., and Dillon, O. W. (1990). ‘‘Overview of nondestructive evaluation projects and initiative at NSF.’’ J. Intelligent Mat., Sys., and Struct., 1(1), 422–431. Eaves, J. L., and Reedy, E. K. (1987). Principles of modern radar, Van Nostrand Reinhold, New York. Kong, J. A. (1990). Electromagnetic wave theory, 2nd Ed., Wiley, New York. Menon, M. M., Boudreau, E. R., and Kolodzy, P. J. (1993). ‘‘An automated ship classification system for ISAR imagery.’’ The Lincoln Lab. J., 6(2), 289–308. Mensa, D. L. (1981). High resolution radar imaging, Artech House, Dedham, Mass. Novak, L. M., Owirka, G. J., and Netishen, C. M. (1993). ‘‘Performance of a high resolution polarimetric SAR automatic target recognition system.’’ The Lincoln Lab. J., 6(1), 1–24. Rhim, H. C. (1995a). ‘‘Nondestructive evaluation of concrete using wideband microwave technique.’’ PhD thesis, Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, Mass. Rhim, H. C. (1995b). ‘‘Radar imaging of a cylindrical concrete specimen for nondestructive testing.’’ Experimental Techniques, 19(1), 21–22. Rhim, H. C., and Bu¨yu¨ko¨ztu¨rk, O. (1998). ‘‘Electromagnetic properties of concrete at microwave frequency range.’’ ACI Mat. J., 95(3), 262– 271. Rhim, H. C., Bu¨yu¨ko¨ztu¨rk, O., and Blejer, D. J. (1995). ‘‘Remove radar imaging of concrete slabs with and without a rebar.’’ Mat. Evaluation, 52(2), 295–299. Skolnik, M. I. (1979). Radar handbook, McGraw-Hill, New York.
APPENDIX II. NOTATION The following symbols are used in this paper: B c e f1 f2 fc kn N T ⌬
⌬rad
= = = = = = = = = = =
εr ε* 0 * r xr
= = = = = =
frequency bandwidth; speed of light in space; 2.718281828; starting frequency; ending frequency; center frequency; imaginary part of complex wave number; number of samples of periodogram spectral estimation; transmitted pulse width; sampling interval; rotational angle of concrete specimen in ISAR measurement; real part of complex permittivity of concrete; complex permittivity; permeability of free space; complex permeability; range resolution; and cross range.
JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1457
FOR
By Hong C. Rhim1 and Oral Bu¨yu¨ko¨ztu¨rk2 ABSTRACT: Radar has the potential of becoming a powerful and effective tool for the nondestructive testing of concrete structures. An advancement of the method can be achieved through the understanding of the interaction between electromagnetic waves and concrete, and the identification of optimum radar measurement parameters for probing concrete. For the determination of optimum parameters, systematic radar measurements are needed in tandem with the implementation of proper signal processing techniques for imaging. This paper presents the results of such radar measurements, which are made using a wideband imaging radar with different frequency ranges. Inverse synthetic aperture radar is used for the measurements of laboratory-size concrete specimens having three different types of internal configurations. The signal processing algorithms implemented to obtain two-dimensional and three-dimensional imagery of concrete targets from radar measurements are also discussed. For the concrete specimens and measurement setup used in this study, 2–3.4 GHz waveforms are found to be adequate for the concrete thickness measurement, 3.4–5.8 GHz waveforms are adequate for the detection of delamination, and 8–12 GHz waveforms are adequate for the detection of inclusions embedded inside concrete.
INTRODUCTION Structures deteriorate over time. This deterioration can occur slowly over the life cycle of a structure or suddenly, as damage from an earthquake. In either case, proper inspection and evaluation are needed prior to any rehabilitation, retrofit, repair, or replacement action being taken. Because of the growing number of aging infrastructures worldwide, there is an increasing need to develop reliable nondestructive testing (NDT) techniques for ascertaining the integrity of these infrastructures (Chong et al. 1990). The goal of any NDT technique is to detect deterioration and/or objects located at a certain distance below the surface in an optically opaque medium. In the radar method, the detectability of anomalies and the capability of a wave to penetrate concrete are affected by a variety of measurement parameters—center frequency, frequency bandwidth, polarization of the wave, measurement distance and angle, and the geometric and material properties of a target. This necessitates a study to examine the influence of each parameter on radar measurement results and to establish an optimum combination of the parameters. The application areas of the radar method used for concrete systems range from concrete thickness determination to condition monitoring and the imaging of inclusions inside concrete systems (Rhim 1995b; Rhim et al. 1995). The objective of the work described in this paper was to determine the optimum radar frequency ranges for concrete slab thickness determination, delamination detection, and rebar detection. The following is a presentation of the parameters for radar measurements, the concrete specimens used for the measurements, radar measurement setup, the imaging algorithms implemented, and radar measurement results. RADAR MEASUREMENT PARAMETERS Parameters that affect radar measurement results can be categorized into the following three groups: (1) characteristics 1
Assoc. Prof., Nondestructive Evaluation and Struct. Lab., Dept. of Arch. Engrg., Yonsei Univ., Seoul 120-749, Korea. E-mail: herhim@ yonsei.ac.kr 2 Prof., Dept. of Civ. and Envir. Engrg., Massachusetts Inst. of Technology, Cambridge, MA 02139. Note. Associate Editor: David Rosowsky. Discussion open until May 1, 2001. To extend the closing date one month, a written request must be filed with the ASCE Manager of Journals. The manuscript for this paper was submitted for review and possible publication on July 30, 1999. This paper is part of the Journal of Structural Engineering, Vol. 126, No. 12, December, 2000. 䉷ASCE, ISSN 0733-9445/00/0012-1451–1457/ $8.00 ⫹ $.50 per page. Paper No. 21584.
related to the incident waves of radar, such as center frequency, frequency bandwidth, and polarization; (2) characteristics related to the geometric and material properties of a target, such as size, shape, and electromagnetic properties; and (3) parameters related to measurement setup, such as measurement distance and angle (Rhim 1995a). The parameters of the center frequency ( fc), frequency bandwidth (B) of the incident wave, and electromagnetic properties (real and imaginary parts of a complex permittivity of concrete) are important, as they determine detection and penetration capabilities. According to the principles of radar measurement, the resolution of closely spaced objects can be obtained by narrowing the transmitted pulse width T in time and increasing the system bandwidth B in frequency so that BT ⬇ 1, thus yielding the range resolution r ⬵ c/2B, where c is the speed of light in space (3 ⫻ 108 m/s). A time-bandwidth product approximating unity is inherent in the class of pulse radars in which a carrier is amplitude modulated by a pulsed waveform. Resolution is the ability to distinguish between closely spaced objects. An arbitrarily high degree of range accuracy can be obtained by using signals with a large bandwidth or high level of energy (Skolnik 1979). The range resolution r, which can be achieved by a wideband radar in concrete, is determined by the following relationship (Mensa 1981): r in concrete = (c/兹εr)/2B
(1)
where r = range resolution; and εr = real part of the complex permittivity of concrete. In addition to the range resolution, synthetic aperture radar (SAR) or inverse synthetic aperture radar (ISAR) can generate cross-range imagery (Menon et al. 1993; Novak et al. 1993). The SAR concept employs a coherent radar system based on a single moving antenna to simulate the function of all of the antennas in a real linear antenna array. The SAR system produces high cross-range resolution imagery by processing the stored backscatter amplitude and phase information obtained by sequentially transmitting and receiving electromagnetic energy along a linear aperture path, re-creating in time a long linear antenna array. If a target itself is rotating and the radar is stationary, return signals can be processed in a manner similar to the one using SAR to re-create a high cross-range resolution target image. This technique is often referred to as inverse synthetic aperture radar. ISAR techniques have been used to produce target images for objects on rotating platforms, for satellites rotating in orbit, and even for ships at sea, which roll rhythmically on the sea surface (Eaves and Reedy JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1451
TABLE 1.
Resolutions of Radar System in Concrete
Resolution (1) Range resolution Cross-range resolution
2–3.4 GHz (mm) (2)
3.4–4.8 GHz (mm) (3)
8–12 GHz (mm) (4)
10.71 15.91
6.25 9.34
3.75 4.30
1987). In ISAR measurements, the rotational angle (⌬rad) of a target determines the cross-range resolution. The cross-range resolution that can be achieved by ISAR is determined by the following relationship: xr in concrete = [c/( fc兹εr)]/[2(⌬rad)]
(2)
where xr = cross-range resolution; fc = center frequency of an incident wave; and ⌬rad = angular rotation of a target in radians during measurement. The range and cross-range resolutions obtained from (1) and (2) are tabulated in Table 1. One of the important objects of using radar for NDT is to identify the proper combination of center frequency and frequency bandwidth suitable for a specific application. This must be considered because there is a tradeoff between detection and penetration capabilities as frequency changes. The frequency range and bandwidth available for the ISAR measurements used in this research are determined by the antenna used. In general, a bandwidth increases as a center frequency increases. The importance of finding the proper center frequency and frequency bandwidth suitable for a specific application of the radar method for the NDT of concrete is thus a primary objective of this research. Sets of wideband radar measurements are made using an inverse synthetic aperture radar. ISAR is used because of its ease and versatility in conducting measurements on laboratory size concrete specimens. The ISAR used for the measurements is capable of transmitting waves in a wide frequency range, from 0.1 to 18 GHz, and is fully polarimetric. For the research described, the following three different frequency bands are used: 2–3.4 GHz (S/C band), 3.4–5.8 GHz (C band), and 8– 12 GHz (X band). The frequency range and center frequency selected for each measurement are based on the standard range-frequency combination inherent in the antenna system used. Due to the nature of hardware, as center frequency increases, frequency bandwidth also increases. The three frequency bands used here represent low, medium, and high frequency levels in the microwave frequency spectrum. The output power of the ISAR is 20 dBm (a decibel relative to 1 mW) and the dynamic range is 50 dB.
to make sure that no moisture is present inside. As the specimens are dried inside a humidity-controlled laboratory, no microcracks appeared at the time of the radar measurements. The concrete specimens used for the radar measurements are shown in Fig. 1. A plain concrete specimen is used for the thickness determination of different sets of center frequency and bandwidth. A specimen with delamination is used to examine the possibility of assessing condition change inside concrete by using radar. The present specimen has a relatively large size of delamination in the initial stage of the research work. The size of the delaminations is determined by the frequency range used by the radar, as described in this paper. Finally, a specimen with three 25.4 mm diameter steel bars is measured to investigate the detectability of the bars. These measurement cases are listed in Table 2. Every type of material has a unique set of electromagnetic properties affecting the way in which it interacts with the electric and magnetic fields of electromagnetic waves. Concrete is a dielectric (nonmetallic) material. A dielectric material can be characterized essentially as having two independent electromagnetic properties—the complex permittivity ε* and the complex (magnetic) permeability *. In general, four independent measurements are necessary to establish the quantities of both the real and the imaginary parts of ε* and *. However, most common dielectric materials including concrete are nonmagnetic, making the permeability * very close to the permeability of free space (0 = 4 ⫻ 10⫺7 H/m). Thus, the focus herein is on the complex permittivity ε*. Electromagnetic properties affect the velocity, wavelength, and amount of reflection and attenuation (Kong 1990). It is important to note
CONCRETE SPECIMENS Laboratory-size concrete specimens with different internal configurations are used as targets for the measurements. The concrete specimens were cast with a water/cement/sand/coarse aggregate mix ratio of 1:2.22:5.61:7.12 (by weight). Type I portland cement was used. Coarse aggregates have a maximum size of 38.1 mm. The aggregates are round shaped and from the river. They are properly stored and used prior to the mixing. Concrete specimens have dimensions of 304.8 mm (width) ⫻ 304.8 mm (height) ⫻ 101.6 mm (depth). The age of the specimens at the time of the measurements was four weeks. The uniaxial compression strength of the specimen was 21 MPa at 28 days. Since the work presented here focuses on the radar measurements and signal processing techniques used for the nondestructive testing of concrete rather than on the mechanical or structural aspects of concrete, the strength of the concrete specimens is not a relevant factor. However, the specimens can be made to satisfy a certain level of strength in future studies. The specimens are air dried for two months 1452 / JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000
FIG. 1. Laboratory Size of Concrete Specimens: (a) Perspective of Specimen; (b) Section of Plain Concrete Specimen; (c) Section of Concrete Specimen with Delamination; (d) Section of Concrete Specimen with Three Steel Bars TABLE 2.
Radar Measurement Cases
Concrete specimens (1)
2–3.4 GHz (2)
3.4–5.8 GHz (3)
8–12 GHz (4)
101.6 mm thick plain concrete Concrete with 25.4 mm delamination at 50.8 mm depth Concrete with three 25.4 mm diameter steel bars at 50.8 mm depth
—
—
—
—
—
—
—
—
—
that these electromagnetic properties are not constant. They change with the frequency of incoming waves, temperature, moisture, and mixture of the material (Rhim and Bu¨yu¨ko¨ztu¨rk 1998). For the study presented in this paper, dry concrete specimens are used for the radar measurements. RADAR MEASUREMENT SETUP Sets of wideband radar measurements are made using an ISAR. The ISAR is used because of its capability in generating 2D imagery, and the ease and versatility with which it can be used to conduct measurements on laboratory-size concrete specimens. A schematic view of a measurement setup is shown in Fig. 2. A concrete specimen is placed vertically as a target at the top of a turntable made of Styrofoam. The turntable is capable of rotating the target. A monostatic wideband radar located at a distance sends and receives the wave. The distance selected has measurements meeting a far field criterion, in which both the incident wave and the reflected wave become plane waves for the measurements. The measurements were made by a sweeping frequency, starting from frequency f1 and ending with frequency f2, with 0.1 GHz increments at a fixed angle. The target is rotated one degree at a time for each frequency sweeping. This generates two-dimensional data as a function of angle and frequency. For typical measurements, the target is rotated 20⬚. The frequency sweeping provides range resolution, while the rotation of the target provides cross-range resolution. RADAR IMAGING OF CONCRETE SPECIMENS Upon the radar measurements of concrete specimens, sets of raw data are collected that contain information on the signals returned from the targets. By processing the raw data, one- and two-dimensional imagery can be obtained. The incident wave generated by the antenna of the ISAR system is a stepped frequency continuous wave. It has a pulse width of 20 ns and a power of 20 dBm. The antenna transmits the waves swept from f1 to f2. The raw measured data contain information on a target received by an antenna as a function of frequency at a fixed incident angle for one-dimensional imaging. Each entry in the data has the amplitude and phase change of the received signal at each frequency. This information is collected in frequency. To have an image, the data need to be converted into a time domain with an appropriate processing scheme (Rhim 1995a). First, the raw data are converted into a format of real and imaginary parts to form a complex number real = 10amplitude/20 ⭈ cos(phase) amplitude/20
imaginary = 10
⭈ sin(phase)
complex number = real ⫹ j ⭈ imaginary
(3a) (3b) (3c)
The converted reflected signal is windowed in a frequency domain using a Hanning window, which has the following property (Press et al. 1988): w[n] = 0.54 ⫺ 0.46 cos(2n/M); 0 ⱕ n ⱕ M
(4a)
w[n] = 0 otherwise
(4b)
FIG. 2.
Radar Measurement Setup
When a run of N-sampled points of periodogram spectral estimation is selected, an infinite run of sampled data cj is, in effect, multiplied by a window function in time; here, one is zero, except during the total sampling time N⌬, when it is unity. Because the square window function turns on and off so rapidly, there is a leakage at high frequencies. To remedy this situation, the input data cj, j = 0, . . . , N ⫺ 1 are multiplied by the window function wj, which changes more gradually from zero to a maximum and back to zero as j ranges from 0 to N ⫺ 1. Now that these data have values in the frequency domain, an inverse Fourier transform (IFT) is carried out to express the data in terms of range and cross range. As the aspect angle in two-dimensional imaging has a constant value according to frequency, a one-dimensional IFT is performed and a twodimensional IFT is applied to three-dimensional imaging in which data consist of two variables, such as aspect angle and frequency. RADAR MEASUREMENT RESULTS AND DISCUSSION In general, the radar method has the advantages of remote sensing, fast measurement, high resolution, sensitivity to metallic objects, and safety during measurements. In addition, the radar system used in this work can generate cross-range imagery. This can enhance the usefulness of the radar by providing imagery similar to optic images. Second, the radar system used here has versatility, in that both its center frequency and its bandwidth are adjustable. This can be effective in optimizing detection and penetration capabilities in the NDT of concrete. Based on these features, three possible application areas of the radar method have been identified and presented in the following text. Determination of Concrete Thickness The measurement of concrete thickness is a basic but important application in the use of radar for the NDT of concrete systems. This thickness measurement can be used as a tool for quality-control purposes after the construction of pavement, for example. Also, by measuring the remaining thickness of concrete components such as bridge decks, foundations, abutments, and retaining walls, the degree of deterioration occurring can be assessed. Radar measurement results on a 304.8 mm (width) ⫻ 304.8 mm (height) ⫻ 101.6 mm (depth) dry concrete specimen without any inclusion are shown in Figs. 3(a)–5(b). The measurements are made at the following three different frequency ranges: 2–3.4, 3.4–5.8, and 8–12 GHz. Three-dimensional images are obtained by rotating the target 20⬚. The 3D images are shown in Figs. 3(b), 4(b), and 5(b). Two-dimensional images are obtained from the 3D images by selecting data at a normal incident angle, and are shown in Figs. 3(a), 4(a), and 5(a). By examining the results in Figs. 3(a)–5(b), the peaks in Figs. 3(a) and 3(b) are found to be wider than the ones in Figs. 4(a)–5(b) due to the narrower bandwidth of 1.4 GHz in a lowfrequency range, compared to the 2.4 and 4 GHz bandwidth in the medium- and high-frequency ranges, respectively. The narrower bandwidth gives less resolution, as in (1). It is also evident that the amplitude of the largest peak, which comes from the front surface in Figs. 3(a) and 3(b), is smaller than the ones in Figs. 4(a)–5(b) due to less scattering in the lowerfrequency range. However, the reflection from the back surface is clearly captured in the 2–3.4 GHz measurement. This is due to the fact that less attenuation occurs through the concrete thickness during wave travel, and that the waves are less sensitive to the edges of the specimens in the lower-frequency JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1453
turned signals from the specimens with delaminations compared to sound concrete specimens is related to the condition change occurring inside the concrete. Radar measurement results of concrete specimens with a delamination are shown in Figs. 6(a)–8(b). These results show the feasibility of using radar for monitoring condition change inside concrete at a distance. The existence of the 25.4 mm thick delamination at 50.8 mm depth is clearly seen as a peak in between the front and back surface reflections in Figs. 7(a)– 8(b). The results are obtained from 3.4–5.8 GHz and 8–12 GHz measurements, respectively. However, the measurements
FIG. 3(a). Two-Dimensional Plot of 1D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 2–3.4 GHz
FIG. 4(b). Three-Dimensional Plot of 2D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 3.4–5.8 GHz
FIG. 3(b). Three-Dimensional Plot of 2D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 2–3.4 GHz
FIG. 5(a). Two-Dimensional Plot of 1D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 8–12 GHz
FIG. 4(a). Two-Dimensional Plot of 1D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 3.4–5.8 GHz
range. Thus, 2–3.4 GHz waveforms are found to be adequate for concrete thickness measurement. Condition Monitoring of Concrete Deterioration in concrete severely affects the service life, safety, and maintenance costs of concrete structures. Remote monitoring of a deterioration process in concrete, including the detection of delaminations, can be studied through the radar measurements of laboratory size concrete specimens with artificially created delaminations inside. The change in the re1454 / JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000
FIG. 5(b). Three-Dimensional Plot of 2D Measurement of 304.8 mm ⴛ 304.8 mm ⴛ 101.6 mm Dry Concrete Block at 8–12 GHz
This technique can be further developed and applied to the detection of subsurface condition change. Detection of Steel Bars inside Concrete The detection of steel bars embedded inside concrete for reinforcement has been one of the major goals in nondestructive testing methods for concrete. The use of radar in detecting
FIG. 6(a). Two-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 2–3.4 GHz
FIG. 7(b). Three-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 3.4–5.8 GHz
FIG. 6(b). Three-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 2–3.4 GHz
FIG. 8(a). Two-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 8–12 GHz
FIG. 7(a). Two-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 3.4–5.8 GHz
at 2–3.4 GHz in Figs. 6(a) and 6(b) were not sensitive enough to pick up the change. More measurements with delaminations at different depths indicate that 3.4–5.8 GHz waveforms, as well as 8–12 GHz waveforms, are effective in monitoring condition change. It can be suggested that the condition monitoring of a concrete target over a period of time is possible because the change in the reflected signal represents a change in quality.
FIG. 8(b). Three-Dimensional Plot of Dry Concrete Block with 25.4 mm Thick Delamination Located at 50.8 mm Depth from Front Surface at 8–12 GHz JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1455
presence of the widest bandwidth of 4 GHz, as shown in Figs. 11(a) and 11(b). Discussion The radar method has several advantages in NDT. These include remote sensing without contacting objects, fast measurement at the speed of light, achievement of high resolution, sensitivity to metallic objects, and safety during measurements. A radar system can also be versatile because it adjusts
FIG. 9(a). Two-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 2–3.4 GHz
FIG. 10(b). Three-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 3.4–5.8 GHz
FIG. 9(b). Three-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 2–3.4 GHz
FIG. 11(a). Two-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 8–12 GHz
FIG. 10(a). Two-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 3.4–5.8 GHz
steel bars is advantageous due to the sensitivity of electromagnetic waves to metallic objects and the versatility of using appropriate electromagnetic wave polarization with respect to the orientation of the bars. In Figs. 9(a)–11(b), radar measurement results on a concrete specimen with steel bars are presented. The waveforms with 2–3.4 GHz and 3.4–5.8 GHz were not sensitive enough to detect bars, as shown in Figs. 9(a)–10(b). However, 8–12 GHz waveforms detected bars with good resolution due to the 1456 / JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000
FIG. 11(b). Three-Dimensional Plot of Dry Concrete Block with Three 25.4 mm Diameter Steel Bars at 50.8 mm Depth at 8– 12 GHz
its center frequency and bandwidth, which can optimize detection and penetration capabilities. A comparison of radar measurement results to a computer simulation based on numerical modeling suggests that simulation provides reference imagery for both the prediction and the interpretation of radar measurement results (Bu¨yu¨ko¨ztu¨rk and Rhim 1995). Expanding a database for computer simulation and radar measurements can identify signals for unknown targets. Also, by examining reflection coefficients, one can determine whether a specific peak belongs to a reflection from the front surface, the back surface, or the delamination inside concrete. In this, dielectric constants of concrete play an important role because they can determine the distance between the peaks. The radar measurement results reported in this paper imply that certain requirements be considered for hardware system design. As shown in the results, a specific combination of radar measurement parameters can provide or achieve a specific NDT goal. For example, if the purpose of the measurements is thickness detection, and a typical dimension of the thickness is known, a proper set of measurement parameters can be used instead of having to use all of the features in a hardware system. Handy, portable radar measurement equipment should be considered as a simple, practical means of employing this method, rather than having to use complex machinery. CONCLUSION A series of radar measurements are made on laboratory-size concrete specimens using ISAR. The measurement setup was such that the antenna used was located at a distance from a concrete target for remote measurement. Three different frequency ranges are used for the measurements at 2–3.4 GHz, 3.4–5.8 GHz, and 8–12 GHz. An application of the radar method is studied for concrete thickness detection, condition monitoring, and steel bar detection. As results, suitable frequency ranges are identified for each of the three specific problem areas. For thickness detection using the laboratory-size specimens and radar setup described in this paper, 2–3.4 GHz waveforms provide good results; moreover, 3.4–5.8 GHz waveforms are good for detecting back surface reflection with a reduced peak. For condition monitoring inside concrete, 3.4–5.8 GHz and 8–12 GHz waveforms are shown to be sensitive. For steel bar detection, 8–12 GHz waveforms are suitable because of their high resolution. ACKNOWLEDGMENTS This work was partially supported by the Korea Earthquake Engineering Research Center, funded by the Korea Science and Engineering Foundation (KOSEF). Computational facilities were provided by the Advanced Building Science and Technology Research Center in the College of Engineering, Yonsei University, Seoul, Korea. The writers would like to thank You Sok Kim of the Nondestructive Evaluation and Structures Laboratory at Yonsei University for his help in preparing this paper through an internship provided by KOSEF. Special thanks are due to Prof. Jin A. Kong in the Department of Electrical Engineering and Computer
Science at Massachusetts Institute of Technology (MIT), and Dr. Robert T. Shin at MIT Lincoln Laboratory. The raw data of the ISAR measurements were obtained at MIT Lincoln Laboratory through a research project supported by the U.S. Army Corps of Engineers, Waterways Experiment Station, Vicksburg, Mississippi through contract number DACW39-92-K-0029. The writers would like to thank Dr. Tony C. Liu, Mitch Alexander, and Kenneth Saucier at the U.S. Army Corps of Engineers for the supported provided.
APPENDIX I.
REFERENCES
Bu¨yu¨ko¨ztu¨rk, O., and Rhim, H. C. (1995). ‘‘Modeling of electromagnetic wave scattering by concrete specimens.’’ Int. J. Cement and Concrete Res., 25(5), 1011–1022. Chong, K. P., Scalzi, J. B., and Dillon, O. W. (1990). ‘‘Overview of nondestructive evaluation projects and initiative at NSF.’’ J. Intelligent Mat., Sys., and Struct., 1(1), 422–431. Eaves, J. L., and Reedy, E. K. (1987). Principles of modern radar, Van Nostrand Reinhold, New York. Kong, J. A. (1990). Electromagnetic wave theory, 2nd Ed., Wiley, New York. Menon, M. M., Boudreau, E. R., and Kolodzy, P. J. (1993). ‘‘An automated ship classification system for ISAR imagery.’’ The Lincoln Lab. J., 6(2), 289–308. Mensa, D. L. (1981). High resolution radar imaging, Artech House, Dedham, Mass. Novak, L. M., Owirka, G. J., and Netishen, C. M. (1993). ‘‘Performance of a high resolution polarimetric SAR automatic target recognition system.’’ The Lincoln Lab. J., 6(1), 1–24. Rhim, H. C. (1995a). ‘‘Nondestructive evaluation of concrete using wideband microwave technique.’’ PhD thesis, Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, Mass. Rhim, H. C. (1995b). ‘‘Radar imaging of a cylindrical concrete specimen for nondestructive testing.’’ Experimental Techniques, 19(1), 21–22. Rhim, H. C., and Bu¨yu¨ko¨ztu¨rk, O. (1998). ‘‘Electromagnetic properties of concrete at microwave frequency range.’’ ACI Mat. J., 95(3), 262– 271. Rhim, H. C., Bu¨yu¨ko¨ztu¨rk, O., and Blejer, D. J. (1995). ‘‘Remove radar imaging of concrete slabs with and without a rebar.’’ Mat. Evaluation, 52(2), 295–299. Skolnik, M. I. (1979). Radar handbook, McGraw-Hill, New York.
APPENDIX II. NOTATION The following symbols are used in this paper: B c e f1 f2 fc kn N T ⌬
⌬rad
= = = = = = = = = = =
εr ε* 0 * r xr
= = = = = =
frequency bandwidth; speed of light in space; 2.718281828; starting frequency; ending frequency; center frequency; imaginary part of complex wave number; number of samples of periodogram spectral estimation; transmitted pulse width; sampling interval; rotational angle of concrete specimen in ISAR measurement; real part of complex permittivity of concrete; complex permittivity; permeability of free space; complex permeability; range resolution; and cross range.
JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1457
