TY - JOUR
AU1 - Muller,, Wayne
AU2 - Scheuermann,, Alexander
AB - Abstract Measuring the electrical permittivity of civil engineering materials is important for a range of ground penetrating radar (GPR) and pavement moisture measurement applications. Compacted unbound granular (UBG) pavement materials present a number of preparation and measurement challenges using conventional characterisation techniques. As an alternative to these methods, a modified free-space (MFS) characterisation approach has previously been investigated. This paper describes recent work to optimise and validate the MFS technique. The research included finite difference time domain (FDTD) modelling to better understand the nature of wave propagation within material samples and the test apparatus. This research led to improvements in the test approach and optimisation of sample sizes. The influence of antenna spacing and sample thickness on the permittivity results was investigated by a series of experiments separating antennas and measuring samples of nylon and water. Permittivity measurements of samples of nylon and water approximately 100 mm and 170 mm thick were also compared, showing consistent results. These measurements also agreed well with surface probe measurements of the nylon sample and literature values for water. The results indicate permittivity estimates of acceptable accuracy can be obtained using the proposed approach, apparatus and sample sizes. permittivity characterization, unbound granular road pavements, GPR 1. Introduction Unbound granular (UBG) pavements are widely used within Australia as economical road construction materials. The structural performance of these materials is highly moisture sensitive and so methods of monitoring pavement moisture are useful for a range of pavement engineering applications. The permittivity of these materials is strongly influenced by, and can be related to, the volumetric moisture content. An approach using multi-offset noise-modulated ground penetrating radar (NM-GPR) is currently being developed to quantify moisture within UBG pavements by measuring the permittivity of pavement layers (Muller et al2012, Muller 2014). To enable moisture estimates from these measurements, moisture-permittivity relations need to be developed and calibrated by measuring the bulk permittivity of UBG samples prepared to mimic a range of field moisture and density conditions in the laboratory. Existing characterisation methods, however, have difficulty accommodating these unbonded coarse-grained materials; only measure small material volumes; use delicate apparatus that may be damaged during sample compaction or lack measurement precision. A more precise and practical approach better suited to measuring larger, more representative samples of compacted UBG materials is therefore required. 2. Methods 2.1. Free-space characterisation There are several approaches to free-space permittivity characterisation of materials. One method involves measuring the variation in phase and amplitude of electromagnetic (EM) signals passing between a fixed pair of antennas when a known thickness of a sample material placed in between. These measurements are normally collected using a vector network analyser (VNA) attached to the antennas, which collects measurements before and after the sample is inserted. Assuming far-field conditions for low-loss materials the frequency-dependent real and imaginary components of complex relative permittivity, herein referred to as real permittivity ( εr′ ⁠) and imaginary permittivity ( εr″ ⁠), can be determined via (Trabelsi et al2000): εr′≈(1+ΔΦλ0360d)21 εr″≈ΔAλ0εr′8.686πd2 where A  =  attenuation due to sample insertion (dB); λ0  =  free space wavelength for a given frequency; d  =  material thickness; and ΔΦ  =  measured phase-shift (degrees). The measured phase-shift occurs due to slowing of the EM wave within the sample compared to the initial measurement through air. For non-magnetic and low loss ( εr′≫εr″ ⁠) materials the phase velocity (v) of propagating EM waves is related to the real permittivity via: v=cεr′3 where c is the speed of light in a vacuum. Free-space techniques have been used to a limited extent for the characterisation of civil engineering materials. Examples in the literature have involved more complicated analysis methods, measured relatively thin samples or measured at frequencies well above that used for pavement GPR (e.g. Büyüköztürk et al2006, Panzner et al2010, Jamil et al2013, Pellinen et al2015). Such measurements are normally undertaken using horn antennas with relatively large samples; or using focussed antennas or high measurement frequencies to achieve far-field focussing and minimise the effects of material boundaries (Chen et al2004, Kaatze and Hübner 2010). The need for specialist focussing antennas or a controlled measurement environment, however, limits the practicality of these methods for routine use in a soil laboratory. Furthermore, high frequency measurements require relatively expensive VNA equipment and may limit the thickness of samples due to focussing considerations and losses due to signal attenuation and scattering. Modified free-space (MFS) is an adaptation of the free-space approach that uses a compact arrangement comprising of a pair of antennas and material coupling similar to that used in a typical ground-coupled GPR system. The approach was initially trialled on moist compacted samples of UBG materials and compared to GPR permittivity measurements (Muller et al2012). It was later compared to results from a large one-port coaxial cell for a range of dielectric materials (Muller and Dérobert 2013) and to time-domain reflectometry (TDR) and GPR measurements for UBG material samples prepared to a range of moisture content and density conditions (Muller et al2015), overall showing good agreement with these other techniques. MFS is intended as a means of calibrating GPR estimates of layer depth and moisture content. The similarity of the MFS and GPR equipment provides the potential advantage that any permittivity measurement errors due to material properties (for example, when measuring lossy materials) should equally affect both methods and so should not affect the accuracy of field estimates. Moreover, as the VNA equipment can be used as a GPR (e.g. Kong et al2012), reflected signals viewed in the time-domain can be directly compared to GPR measurements and the change in arrival times of direct and reflected signals can be used as an additional means of calculating the sample permittivity. The ability to measure through material samples enables characterisation of larger, more representative sample volumes compared to typical surface probes or small coaxial cells. The use of a removable dielectric container enables easier compaction of loose UBG materials to better replicate field conditions without the risk of damaging delicate or expensive test apparatus. Furthermore the measurement of phase-shift is more precise compared to conventional time-domain techniques such as TDR or GPR measurements through a sample of known thickness. There are, however, a number of limitations of the MFS approach. For one, it is not possible to reliably measure the imaginary permittivity due to changes in antenna performance with coupling (Muller and Dérobert 2013). Although the effect of sample edges was previously investigated and found to be minor (Muller and Dérobert 2013), reflections within the sample or apparatus remain a potential concern. Furthermore, due to the relatively close antenna separation, near-field effects have potential to influence the measurements. 2.2. Approach The aim of this paper is to better understand and optimise the MFS test configuration, to investigate concerns regarding near-field effects and to further assess the measurement accuracy of the MFS approach. To this end numerical techniques are used to model wave propagation within the sample and apparatus to enable better understanding and optimisation of the test procedure and the size of material samples. The influence of near-field effects is investigated in a series of laboratory experiments that involve assessing the linearity of phase with increasing antenna separation and assessing the accuracy of real permittivity measurements for known materials (water, air) and for an independently characterised material (nylon). 2.3. Equipment A FieldFox N9923A portable VNA was used for the investigation. It was configured to measure at 1001 frequency steps over the range 7.984 MHz to 4.0 GHz using the VNA’s high power setting and taking the average of five frequency sweeps per measurement. The VNA’s inbuilt ‘QuickCal’ calibration procedure was used prior to these measurements. VNA measurements, called scattering parameters (S-parameters), are denoted Sxy, where x and y denote the receiving and transmitting ports of the VNA, respectively. While equation (1) only requires the phase change of the transmitted signal to determine ε′, the magnitude and phase of both transmitted (S21) and reflected (S11) signals were collected to enable additional time-domain analysis and signal filtering options. The antennas used for the S-parameter measurements were a custom pair of shielded ground-coupled bow-tie dipole antennas. These passive antennas incorporate a balun at the antenna feed point to improve signal transmission. In terms of radiation pattern, numerical modelling of shielded dipole antennas in free-space indicates a downward lobe directly beneath the antenna (Diamanti and Annan 2013). Bow–tie antennas produce a narrower beam pattern compared to small dipoles (Millard et al2002). The antenna performance changes with the proximity to, and properties of, the ground and other nearby materials (Diamanti and Annan 2013, Millard et al2002). Beam patterns measured in air narrow significantly when in contact with the dielectric material (Millard et al2002) and become narrower and more directional with increasing permittivity (Diamanti and Annan 2013). The vast majority of energy emitted by these antennas is pulled into the ground, which has little to do with the antenna shielding (Diamanti and Annan 2013). 3. Results and discussion 3.1. Numerical modeling To better understand wave propagation within the test apparatus 2D finite difference time domain (FDTD) modelling was undertaken (figure 1). Sample boxes with internal dimension of 300  ×  300  ×  100 mm containing material samples with real permittivity values of 5–15 were modelled using GPRMax 2D software (Giannopoulos 2005). The excitation used for these models was a simple 1.5 GHz point source Ricker wavelet located at the position of the feed-point of the transmitting antenna. The directionality of the bow–tie antenna and effects of shielding were not considered in this modelling. Figure 1. Open in new tabDownload slide Numerical model showing instantaneous electric-field strength (left) and transmitted signal strength measured by the receiving antenna (right): (a) Within a ply box with εr′  =  9 sample material, (b) Within an empty ply box, and (c) With only a ply sheet present. Figure 1. Open in new tabDownload slide Numerical model showing instantaneous electric-field strength (left) and transmitted signal strength measured by the receiving antenna (right): (a) Within a ply box with εr′  =  9 sample material, (b) Within an empty ply box, and (c) With only a ply sheet present. This FDTD modelling was useful as it revealed several key points. The first was that the existing sample box aspect ratio of 3:1 was a good choice as spurious reflections from box sides and the double bounce arrived at a similar time, indicated with an arrow in figure 1(a). That is, this aspect ratio enables the direct wave to be maximally separated from these unwanted reflections, which in turn makes it easier to isolate the direct wave or remove the unwanted effects—for example, by using time-domain windowing as per Muller and Dérobert (2013). A second observation was the detrimental effect the sides of the empty ply box had when collecting the reference signal. That is, when the box is empty, the signal reflecting from the sides of the box arrives soon after the direct wave resulting in a poor reference signal (figure 1(b)). Subsequent modelling of a ply sheet as the reference instead of an empty box showed a significant reduction of such unwanted reflections (figure 1(c)). As the calculation of real permittivity depends on sample thickness, slight variations in the depth of prepared samples will affect the results. Using thicker samples would reduce the relative effect of such variations, however maintaining the 3: 1 sample aspect ratio would significantly increase the mass, making manual handling impractical. To understand the consequences on the VNA measurements of using deeper sample boxes with an altered aspect ratio, 2D FDTD modelling was undertaken for boxes of internal dimensions 170  ×  250  ×  250 mm. This change corresponds to a slight increase in mass from approximately 23 kg to 26 kg. However, as observed in figure 2 a disadvantage of the change is that the side reflections arrive much closer in time and begin to impinge on the arrival of the direct wave signal, as indicated with an arrow. Figure 2. Open in new tabDownload slide 2D FDTD numerical modelling of a 170  ×  250 mm sample box filled with a εr′  =  9 UBG material. Figure 2. Open in new tabDownload slide 2D FDTD numerical modelling of a 170  ×  250 mm sample box filled with a εr′  =  9 UBG material. 3.2. Investigation of near-field effects Equation (1) assumes plane waves. These occur in the far-field region of the antenna (Millard et al2002). There are different methods of estimating the distance to the transition between near-field and far-field regions, including those described by Johnson et al (1973) and Millard et al (2002). Numerical modelling indicates that this transition may in fact be much further away than predicted by conventional methods of estimating the distance to the far-field region (Diamanti and Annan 2013). However the boundaries between the near-field and far-field regions are not well defined and the suitability of the means of estimation depends on the intended application (Johnson et al1973). Using any of these estimation methods, it is unlikely that far-field conditions are being unambiguously met using the current MFS test configuration. Previous investigations, however, produced permittivity estimates that match relatively well with measurements using other techniques (Muller et al2012, 2015, Muller and Dérobert 2013), indicating the influence of near-field effects may be relatively limited for the test configurations used to date. To further investigate and better understand the influence of these effects on the MFS measurements, a series of experiments was undertaken. 3.2.1. Experiment 1. The first experiment was a simple test to determine the linearity of phase measurements in air using the MFS antennas. The experiment involved collecting VNA S21 phase measurements while separating the antennas in 100 mm increments from zero to one metre apart. The phase-shift between a reference measurement at zero offset and each subsequent antenna position was used to calculate the real permittivity of air ( εr′  ≈  1) using equation (4), a modification of equation (1) that accounts for the varying antenna separation. The real permittivity values were then recalculated taking the measurements at 100 mm, 200 mm and 300 mm as the reference. The calculated values are shown in figure 3. While care was taken during the measurements, the results can only be considered approximate as the antennas were manually separated and the offset was determined using a simple tape measure. εr′≈(ΔΦλ0360d)24 Figure 3. Open in new tabDownload slide Real permittivity calculated for air using different antenna reference offsets. Figure 3. Open in new tabDownload slide Real permittivity calculated for air using different antenna reference offsets. The results in figure 3 indicate there is a significant error in the real permittivity calculation using the zero spacing measurement as the reference, most likely due to the influence of near-field effects. These errors decrease when measurements at 100 mm and 200 mm were used as the reference, however there was little difference when increasing from 200 mm to 300 mm. This indicates the phase-change with increasing distance is relatively linear at 200 mm, though may be acceptable at some point between 100 mm and 200 mm. However, as sample boxes of size 300  ×  300  ×  100 mm already weigh more than 23 kg when filled with compacted UBG materials, maintaining the same 3: 1 aspect ratio while increasing the depth to 200 mm would make manual handling impossible (~187 kg). 3.2.2. Experiment 2. The next experiment involved initially fixing the pair of antennas at an offset of 125 mm. A reference transmission (S21) measurement was then collected through a 400  ×  400  ×  18 mm ply sheet placed on top of the lower receiving antenna. As shown in figure 4, a series of 350  ×  350 mm nylon sheets were then incrementally placed on top of the ply sheet and between the antennas. After each sheet was added the phase of the transmitted signal passing between the antennas was measured and the phase-shift relative to the initial calibration was determined. The antenna spacing was then increased to 190 mm and an updated reference measurement was collected through the ply sheet. Nylon sheets 170 mm thick were then placed on top of the ply sheet. The phase measurement of the transmitted signal was collected and the phase-shift was determined compared to the updated calibration measurement. The real permittivity was then determined for each sample thickness and antenna separation, without time-domain windowing (figure 5). The results were then compared to an independent measurement of one of the nylon sheets using the dielectric probe described by Wagner et al (2014) and an Agilent E5061B-3L5 network analyser. Figure 4. Open in new tabDownload slide VNA measurements with fixed antenna separation through samples of nylon and a ply sheet. Figure 4. Open in new tabDownload slide VNA measurements with fixed antenna separation through samples of nylon and a ply sheet. Figure 5. Open in new tabDownload slide Effect of increasing the thickness of nylon on (a) phase-shift linearity and (b) real permittivity calculations. Figure 5. Open in new tabDownload slide Effect of increasing the thickness of nylon on (a) phase-shift linearity and (b) real permittivity calculations. As shown in figure 5, the phase-shift linearity and consistency of real permittivity predictions improve with increasing sample thickness. When the sample thickness has reached 106 mm the measurements are becoming relatively consistent over the range 1.2 to 3.5 GHz and are similar to the measurement through the 170 mm thick nylon sample. These results, which averaged εr′  =  3.00 for the 106 mm thick nylon sample and 3.04 for the 170 mm sample over the frequency range 900 MHz to 3.5 GHz compare relatively well with the dielectric probe results which reported a flat response that averaged εr′  =  2.82 and εr″  =  0.07 over the range 900 MHz to 3.0 GHz. This corresponds to a difference of 6 to 8% between the methods; but the probe produced more consistent results over this range. Another aspect to note in the MFS data is the spike in the phase plots and corresponding real permittivity results at around 2.6–2.8 GHz. This feature corresponds to the first frequency null of the antenna, clearly visible in the S21 amplitude results. Consequently, for these antennas it is recommended to limit the measurement frequency range to between 900 MHz to 2 GHz to avoid potential phase measurement errors in the vicinity of the null. 3.2.3. Experiment 3. One issue with this experiment is that because the antennas were fixed while the sample thickness was increased the antenna coupling will have varied. The approach also introduces a notable asymmetry in the measurement setup. When measuring UBG samples a small asymmetry is also present due to the floor of the sample box. To investigate these effects, an experiment was undertaken that again measured the phase-shift through an increasing thickness of nylon sheets, but this time the upper antenna was moved to maintain consistent sample coupling and a symmetric test configuration. The upper antenna was fixed at a separation that enabled the nylon to be inserted with a small gap of approximately 5 mm. Prior to inserting the nylon a reference measurement was collected. The nylon was then slid in between the antennas and a second measurement was made. The antenna separation was increased to accommodate the next thickness of nylon and the measurements were repeated. This process continued until a total nylon thickness of 170 mm was measured. In the next part of the experiment the measurements were repeated, but this time the ply sheet used in the previous experiment was inserted for each reference measurement and was placed under the nylon sheets for each sample measurement. Real permittivity values calculated over the frequency range 900 MHz to 2.0 GHz using these measurements are shown in figure 6. Figure 6. Open in new tabDownload slide Real permittivity measured for an increasing thickness of nylon sheets (a) using a symmetric test configuration; and (b) the same measurements including a ply sheet placed against the lower antenna. Figure 6. Open in new tabDownload slide Real permittivity measured for an increasing thickness of nylon sheets (a) using a symmetric test configuration; and (b) the same measurements including a ply sheet placed against the lower antenna. As observed in figure 6(a), the stability of measurements over this frequency range improves with the increasing sample thickness. When the sample thickness has reached 106 mm is it is relatively stable and consistent with measurements of thicker samples. Comparing figures 6(a) and (b), for samples 106 mm thick and greater the ply sheet appears to have only a small effect on the calculated permittivity values. 3.2.4. Experiment 4. In the final experiment, permittivity measurements were undertaken using two different thicknesses of nylon, distilled water (via reverse osmosis) and tap water. Both the distilled and tap water samples were at a temperature of 21.3 °C at the time of testing and were poured into the plywood sample boxes normally used for UBG materials (figure 7). The sample boxes used for the water measurements had internal dimension of 300  ×  300  ×  100 mm and 250  ×  250  ×  170 mm. The corners of the ply boxes were sealed with silicone to prevent leakage. S21 measurements were first collected through a ply sheet, 106 mm of nylon placed on the ply sheet and the 100 mm sample box filled to the top with water. These initial measurements were collected with a fixed antenna separation of 125 mm. The antenna separation was then increased to 190 mm and the measurements were repeated through the ply sheet, 170 mm of nylon placed on the ply sheet and 170 mm of distilled and tap water placed in the deeper sample box. For each of these measurements the samples was aligned centrally with the antennas and the lower antenna was mounted flush with the surface on which the ply sheet or sample boxes was placed. The upper antenna was aligned with the lower antenna and fixed to achieve the smallest possible air gap above the surface the nylon and water. As a result the measurements included a small asymmetry due to the presence of the ply sheet or base of the sample box. Equivalent time domain signals passing through these samples were later determined from the VNA measurements of amplitude and phase (figure 8). Figure 7. Open in new tabDownload slide MFS test configuration used for measuring a 170 mm deep sample of distilled water shown prior to complete filling of the sample box. Figure 7. Open in new tabDownload slide MFS test configuration used for measuring a 170 mm deep sample of distilled water shown prior to complete filling of the sample box. Figure 8. Open in new tabDownload slide (a) The time domain signal determined from VNA S21 transmission measurements; (b) VNA amplitude measurement; and (c) VNA phase measurement. These measurements were collected through the ply sheet (labelled ‘air’), 106 mm of nylon with the ply sheet and a 100 mm depth of distilled water in a 300  ×  300  ×  100 mm sample box. Figure 8. Open in new tabDownload slide (a) The time domain signal determined from VNA S21 transmission measurements; (b) VNA amplitude measurement; and (c) VNA phase measurement. These measurements were collected through the ply sheet (labelled ‘air’), 106 mm of nylon with the ply sheet and a 100 mm depth of distilled water in a 300  ×  300  ×  100 mm sample box. The permittivity values determined during this experiment are show in figure 9, with measurements undertaken at 125 mm and 190 mm shown as solid and dashed lines, respectively. Time domain windowing has not been applied to these data. The results reveal only a small (1%) difference in the mean permittivity measurement of nylon and water for the different combinations of sample thickness and antenna spacing. A notable oscillation is seen within the 100 mm distilled water permittivity results over the range 900 MHz–1.5 GHz. A corresponding oscillation is also noted below approximately 1.5 GHz in the amplitude spectrum of the 100 mm distilled water S21 measurements, as indicated with the angled arrow in figure 8(b). A similar effect is also seen in the amplitude spectrum of the 170 mm water measurement, though any influence on phase and consequently the permittivity result in figure 8, appears minimal. This oscillation may be due to a low frequency resonance within the water sample, as the significant dielectric contrast at the water-plywood and water-air boundaries results in strong reflections. An indication of this is the relative strength of the double bounce of the transmitted signal, the location of which is indicated with a vertical arrow in figure 8(a). The asymmetry of the test setup may also have contributed to the oscillation, however this is unlikely as the effect was not observed in either the air or nylon measurements which also included a ply sheet. Figure 9. Open in new tabDownload slide Real permittivity determined using the MFS approach for nylon and distilled and tap water using different antenna separations and sample thicknesses. Figure 9. Open in new tabDownload slide Real permittivity determined using the MFS approach for nylon and distilled and tap water using different antenna separations and sample thicknesses. Despite this low frequency irregularity, the overall results for water permittivity are relatively consistent, with average real permittivity values measured through 100 mm and 170 mm sample thicknesses ranging between 79.2 and 80.1. These are within 1% of the value of 79.8 for water at 1.0 GHz interpolated to 21.3 °C based on Buchner et al (1999). As the permittivity of nylon and water are beyond the expected upper and lower bounds for UBG pavement materials, which are typically in the range 5–15, the results indicate that near-field effects are unlikely to have significantly affected the accuracy of real permittivity measurements for the size of samples used to date. As observed in the calculation of permittivity for water, it may be advantageous to use a sample thickness greater than 100 mm to reduce the effect of any unwanted oscillations occurring within the sample or test apparatus. The experiment separating antennas in air indicated that an antenna offset of 100 mm was insufficient as a reliable reference. As the antenna separation used for measuring the 100 mm samples was only 125 mm, it would be preferable to increase this spacing to increase confidence in the validity of the reference phase measurement. Furthermore, the permittivity measurements for water using 170 mm deep sample boxes and the varied aspect ratio produced more consistent results than the 100 mm deep samples, despite the concerns raised from the FDTD modelling that side reflections may affect the measurements. The influence of side reflections would also have been reduced by narrowing of the antenna beam pattern due to the high permittivity of the water sample. 4. Conclusions This paper provides an overview of recent studies aimed at optimising and investigating the influence of near-field effects on the MFS permittivity characterisation approach. The work included FDTD modelling to assist understanding of the nature of wave propagation within the test apparatus. This revealed that a sample aspect ratio of approximately 3:1 (width: depth) was preferable to maximise separation of the direct signal from unwanted reflections from the sides of the sample box and the double bounce from the top of the sample. The modelling also revealed it was best to collect the reference signal through a ply sheet rather than an empty box, as spurious reflections from the box sides complicated the reference measurement. The influence of near-field effects on the measurements was investigated in four experiments. The first indicated that a minimum antenna spacing of between 100 mm and 200 mm in air was required to achieve a linear-phase reference measurement using the custom antennas. The second showed the consistency of permittivity results for nylon improved with increasing sample thickness, though there was little improvement for samples thicker than 106 mm. The third investigated the influence of sample thickness and symmetry of the test configuration, finding the measurements were relatively stable for samples of 106 mm or greater and that the presence of the ply sheet had little effect on permittivity results calculated for these thicker samples. In the final experiment measurements of samples of nylon, distilled water and tap water approximately 100 mm and 170 mm thick were compared, revealing similar results. The average real relative permittivity result for nylon over the frequency range 900 MHz–3.5 GHz was within 6–8% of an independent measurement using a broadband dielectric probe over the range 900 MHz–3.0 GHz. The result for distilled water also agreed well with the literature (within 1%), though a notable oscillation was observed most likely due to internal reflections within the sample. As the samples of nylon and water are beyond the likely upper and lower bounds of permittivity for UBG pavement materials, the investigation indicates measurements through a sample thickness of at least 100 mm using the current MFS apparatus are unlikely to be significantly affected by near-field effects. However, despite concerns raised from FDTD modelling, it appears that the deeper 170  ×  250  ×  250 mm sample boxes may be preferable to enable greater antenna separation and achieve more consistent and accurate permittivity measurements. Furthermore, narrowing of the antenna beam pattern due to coupling with the material sample, particularly for high permittivity materials, may help improve the results by reducing the strength of reflections from the sample edges. Acknowledgments This work was funded under NACOE research project P12 ‘Field-validation of Noise Modulated Ground Penetrating Radar’ by the Queensland Department of Transport and Main Roads. The authors express their sincere thanks to Dr Bryan Reeves for assistance during this work and also to Moritz Schwing and Zhen Chen from UQ for undertaking the independent nylon permittivity measurements. This contribution supports the work of COST Action TU1208 ‘Civil Engineering applications of Ground Penetrating Radar’. The presented research is supported by a Queensland Science Fellowship awarded to A Scheuermann. References Buchner R , Barthel J , Stauber J . , 1999 The dielectric relaxation of water between 0 °C and 35 °C , Chem. Phys. Lett. , vol. 306 (pg. 57 - 63 ) 10.1016/S0009-2614(99)00455-8 Google Scholar Crossref Search ADS WorldCat Crossref Büyüköztürk O , Yu T , Ortega J . , 2006 A methodology for determining complex permittivity of construction materials based on transmission-only coherent, wide-bandwidth free-space measurements , Cement Concr. Compos. , vol. 28 (pg. 349 - 359 ) 10.1016/j.cemconcomp.2006.02.004 Google Scholar Crossref Search ADS WorldCat Crossref Chen L F , Ong C K , Neo C P , Varadan V V , Varadan V K . , 2004 , Microwave Electronics: Measurement and Materials Characterization Chichester, West Sussex, UK Wiley Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Diamanti N , Annan A P . , 2013 Characterizing the energy distribution around GPR antennas , J. Appl. Geophys. , vol. 99 (pg. 83 - 90 ) 10.1016/j.jappgeo.2013.08.001 Google Scholar Crossref Search ADS WorldCat Crossref Giannopoulos A . , 2005 Modelling ground penetrating radar by GPRMax , Const. Build. Mater. , vol. 19 (pg. 755 - 762 ) 10.1016/j.conbuildmat.2005.06.007 Google Scholar Crossref Search ADS WorldCat Crossref Jamil M , Hassan M K , Al-Mattarneh H M A , Zain M F M . , 2013 Concrete dielectric properties investigation using microwave nondestructive techniques , Mater. Struct. , vol. 46 (pg. 77 - 87 ) 10.1617/s11527-012-9886-2 Google Scholar Crossref Search ADS WorldCat Crossref Johnson R C , Ecker H A , Hollis J S . , 1973 Determination of far-field antenna patterns from near-field measurements , Proc. IEEE , vol. 61 (pg. 1668 - 1694 ) 10.1109/PROC.1973.9358 Google Scholar Crossref Search ADS WorldCat Crossref Kaatze U , Hübner C . , 2010 Electromagnetic techniques for moisture content determination of materials , Meas. Sci. Technol. , vol. 21 082001 OpenURL Placeholder Text WorldCat Kong F N , Bhasin R K , Wang L Z , Zhang Y G . , 2012 Development of handheld gpr using Fieldfox network analyzer and its application in detecting structure of Qian Tang river Dike, Hang Zhou, China 14th Int. Conf. on Ground Penetrating Radar (GPR-2012) Shanghai, China 4–8 June 2012 IEEE (pg. 146 - 151 ) Millard S G , Shaari A , Bungey J H . , 2002 Field pattern characteristics of GPR antennas , NDT & E Int. , vol. 35 (pg. 473 - 482 ) 10.1016/S0963-8695(02)00023-3 Google Scholar Crossref Search ADS WorldCat Crossref Muller W . , 2014 Self-correcting pavement layer depth estimates using 3D multi-offset ground penetrating radar (GPR) 15th Int. Conf. on Ground Penetrating Radar (GPR-2014) Brussels, Belgium 30 June–4 July 2014 IEEE (pg. 887 - 892 ) Muller W , Dérobert X . , 2013 A comparison of phase-shift and one-port coaxial cell permittivity measurements for GPR applications 7th Int. Workshop on Advanced Ground Penetrating Radar (IWAGPR-2013) 2–5 July 2013 (pg. 1 - 6 ) Muller W B , Bhuyan H , Scheuermann A . , 2015 A comparison of modified free-space (MFS), GPR and TDR techniques for permittivity characterisation of unbound granular pavement materials Int. Symp. Non-Destructive Testing in Civil Engineering (NDT-CE) Berlin, Germany 15–17 September 2015 Wiggenhauser H . Berlin BAM/TU (pg. 421 - 424 ) Muller W B , Scheuermann A , Reeves B . , 2012 Quantitative moisture measurement of road pavements using 3D GPR 14th Int. Conf. on Ground Penetrating Radar (GPR-2012) Shanghai, China IEEE (pg. 517 - 523 ) Panzner B , Jostingmeier A , Abbas O . , 2010 Estimation of soil electromagnetic parameters using frequency domain techniques 13th Int. Conf. on Ground Penetrating Radar (GPR-2010) Lecce, Italy IEEE Pellinen T , Huuskonen-Snicker E , Eskelinen P , Olmos Martinez P . , 2015 Representative volume element of asphalt pavement for electromagnetic measurements , J. Traffic Transp. Eng. , vol. 2 (pg. 30 - 39 ) OpenURL Placeholder Text WorldCat Trabelsi S , Kraszewski A W , Nelson S O . , 2000 Phase-shift ambiguity in microwave dielectric properties measurements , IEEE Trans. Instrum. Meas. , vol. 49 (pg. 56 - 60 ) 10.1109/19.836309 Google Scholar Crossref Search ADS WorldCat Crossref Wagner N , Schwing M , Scheuermann A . , 2014 Numerical 3D Fem and experimental analysis of the open-ended coaxial line technique for microwave dielectric spectroscopy on soil , IEEE Trans. Geosci. Remote Sens. , vol. 52 (pg. 880 - 893 ) 10.1109/TGRS.2013.2245138 Google Scholar Crossref Search ADS WorldCat Crossref © 2016 Sinopec Geophysical Research Institute
TI - Optimising a modified free-space permittivity characterisation method for civil engineering applications
JF - Journal of Geophysics and Engineering
DO - 10.1088/1742-2132/13/2/S9
DA - 2016-04-01
UR - https://www.deepdyve.com/lp/oxford-university-press/optimising-a-modified-free-space-permittivity-characterisation-method-OxhYm0NHPB
SP - S9
VL - 13
IS - 2
DP - DeepDyve
ER -