International Symposium (NDT-CE 2003)Non-Destructive Testing in Civil Engineering 2003
|Start > Contributions >Posters > Radar:|
GPR ANTENNA MEDIUM COUPLING EFFECTS: EXPERIMENTAL AND 2D FDTD MODELLING RESULTSA Shaari, Department of Physics, Faculty of Science, Universiti Teknologi, Malaysia
S G Millard and J H Bungey, Department of Civil Engineering, University of Liverpool, UK
When planning a survey of a civil engineering structure using ground-penetrating radar (GPR), it is necessary to select an antenna with an appropriate centre frequency. A higher frequency antenna usually has a better ability to resolve closely spaced targets but also has a reduced signal penetration beneath the surface of the structure. The antenna frequency however is modified by coupling effects with the surface of the material being investigated and can be significantly lower than the manufacturer's quoted frequency in air. In addition the same coupling effect also influences the angular spread, or beam-width of the transmitted signal. This paper reports on a study of these two antenna characteristics in air, water and concrete which was carried out on a typical GPR antenna with nominal centre frequency of 900MHz. Results from these measurements are compared with those obtained from 2D finite-difference time domain (2D FDTD) modelling. Basic principles of the experimental and modelling procedures are also discussed. The results show how both the antenna centre frequency and the beam width are influenced by the relative permittivity of the material measured. Some practical implications of these results are discussed.
Ground penetrating radar (GPR) inspection has now become an established part of the range of non-destructive testing (NDT) techniques that can be used to assess the integrity of civil engineering structures. In GPR inspection, the ability to interpret a complex signal is essential and difficulties in interpretation have been identified as a potential drawback associated with the technique. This problem can be partly attributed to inadequate theoretical knowledge and lack of accurate and verifiable experimental data on radar propagation in structural materials. There is also insufficient understanding of the coupling effects of GPR antennas near or on structural materials at relevant frequency ranges. While the former requirement demands more data and better theoretical models of the dielectric properties of materials such as concrete, the latter requires more knowledge on the antenna effective beam-width and the frequency characteristics in these materials [(1)]. There is also a need for means to predict the changes to the characteristics of a GPR antenna since actual testing conditions may vary as the dielectric properties of the structural material change. Numerical modelling of the antenna-medium coupling effects could serve as one of the solutions to this problem. In this paper results on the effective beam-widths and centre-frequency of a 900MHz antenna measured indirectly from reflected signals from targets in materials with different relative permittivity (air, water and concrete) are compared with those from a computer modelling using a two-dimensional finite-difference time domain (2D FDTD) method. Concrete normally has a relative permittivity between 5 (dry) and 12 (wet). Air has a relative permittivity of 1 whilst water has a value of 81.
A direct near-field radiation pattern or beam-width measurement requires a set of highly specialised equipment and an access to the space surrounding the antenna under test [(2)]. Even if the costs of test equipment could be overcome, space restrictions will in most cases rule out the possibility of using a direct measurement method, particularly in solid materials such as concrete. It is necessary therefore to adopt an indirect method to study the changes to the characteristics of an antenna when it is coupled to a material with different dielectric properties. Before any measurement can be made it is important that the dielectric properties of the material (relative permittivity) and both the physical and electrical characteristics of the antenna must be known. As shown in Figure 1a, a typical GPR antenna operating at centre frequencies of 500MHz or more normally comprises broadband bow-tie elements placed in a metal casing to reduce the noise level due to interference with surrounding objects. In this study, a SIR2 radar system from Geophysics Survey System Inc. (GSSI) together with a 900MHz antenna is used. The targets are made of plain steel bars with diameter ranging from 3mm to 20mm with a distance from the antenna of about 250mm. The experimental set-up is shown in Figure 1b. A reflected signal, Vr is recorded as the antenna is moved from one sample point to another. The reflected power in dB is then calculated from the following equation,
where Vr_max is the maximum reflected signal. In order to study the changes to the centre frequency of the signal, a Fourier transform of Vr_max is taken.
Whilst indirect measurements can be used to assess the performance of any antenna, this is time-consuming. An alternative approach is to model the performance of an antenna in any material of interest using computer modelling. The finite-difference time domain (FDTD) technique is a powerful means for modelling of transient or broadband electromagnetic problems. The basis for this technique is the following two continuous time-dependent Maxwell curl equations that are then converted into a set of finite difference equations by replacing the derivatives in these equations with differences .
In the above equations, E and H are the electric and magnetic fields respectively, m and r are the magnetic permeability and conductivity respectively, and e and s are the electrical permittivity and conductivity respectively.
In a two dimensional case, the six components of these equations are reduced to two sets of three scalar equations known as transverse magnetic (TM) mode and transverse electric (TE) mode [(5)]. In this study the TM mode is adopted and the three equations represent the z component of electric field, Ez and the y and x components of magnetic fields, Hx and Hy. The cell size and the arrangement of all field components involved are shown in Figure 2a while the model for the GPR antenna and the target is shown in Figure 2b. In all cases only the reflected Ez component, Vz of the field is computed and Equation 1 is then used to convert it into a reflected power (dB).
Results and Discussions
For beam-width measurements, the reflected power (dB) is plotted against the antenna position (cm) and the criterion used for the antenna beam-width in the medium is the point where the reflected power drops to half of its maximum value. This is known as the 3dB beam-width. An antenna with a narrow beam-width would be expected to have a better resolution (i.e. discrimination between two closely spaced targets) than an antenna with a wider beam-width. In Figures 3 and 4, the radiation patterns of 900MHz GPR antenna both in air and concrete obtained from indirect measurements using the SIR2 radar system are compared with results from 2D FDTD modelling. From these figures a few features can be identified.
From these results, it would be expected that at a distance of 250mm from the antenna it might be possible to resolve two targets at 300mm apart in concrete but not in air.
Figures 5 and 6 show the changes in the centre frequencies of the signals in air, water and concrete. It is seen that there is relatively good agreement between the FDTD results and the actual measurements on the reduction of the centre frequency with materials of increasing relative permittivity.
There is not however quite such a good agreement on the frequency distribution of each signal as it passes through a different dielectric material. There are two possible reasons for this discrepancy. This could be due to the 2D FDTD code used here being unable to model the actual 3D GPR antenna and radiation pattern. Alternatively this could be due to a common problem associated with a process of taking a Fourier transform of a band-limited signal. The problem will be further studied by using a 3D version of the finite difference method. The methods discussed here have been applied to an existing 900 MHz radar antenna, but could be equally applied to any new antenna that becomes available.
From the previous results and discussions, it is seen that the effective beam-width of the 900MHz GPR antenna in concrete and water, has undergone some significant reductions, compared with the value in air. The 2D FDTD modelling is capable of predicting the changes in the antenna beam-widths but it is not quite as effective in modelling changes in the signal frequency distribution. These results, however, can be used as a basis in the interpretation of radar signals particularly with regard to both the horizontal and vertical resolutions of a 900MHz GPR antenna.
The authors would like to acknowledge the contribution from Dr. A Giannopoulos for providing the author with 2D FDTD code. A free version of the software GprMax2D may be downloaded from: http://www.see.ed.ac.uk/~agianno/GprMax/