- Email: [email protected]
Non-invasive detection of fractures, fracture zones, and rock damage in a hard rock excavation — Experience from the Äspö Hard Rock Laboratory in Sweden
Engineering Geology 196 (2015) 210–221
Contents lists available at ScienceDirect
Engineering Geology journal homepage: www.elsevier.com/locate/enggeo
Non-invasive detection of fractures, fracture zones, and rock damage in a hard rock excavation — Experience from the Äspö Hard Rock Laboratory in Sweden G. Walton a,⁎, M. Lato b, H. Anschütz c, M.A. Perras d, M.S. Diederichs e a
Colorado School of Mines, Golden, Colorado RockSense GeoSolutions, Ottawa, Canada c Norwegian Geotechnical Institute, Oslo, Norway d ETH Zurich, Zurich, Switzerland e Queen's University, Kingston, Canada b
a r t i c l e
i n f o
Article history: Received 9 July 2014 Received in revised form 27 February 2015 Accepted 15 July 2015 Available online 21 July 2015 Keywords: Excavation Damage Zone Resistivity survey Ground Penetrating Radar LiDAR Geophysical mapping
a b s t r a c t A key requirement for licensing of the construction of underground repositories for nuclear waste is the demonstrated capability to verify design assumptions involving the presence and extent of the excavation damage zone around tunnels, shafts, emplacement holes and caverns. As part of ongoing work to select and refine key technologies and techniques towards this end, geophysical surveys were performed at two locations within the Äspö Hard Rock Laboratory in Sweden. Earth resistivity (RES), induced polarization (IP), and Ground Penetrating Radar (GPR) data were collected using a variety of survey parameters; Light Detection and Ranging (LiDAR) data were collected as a reference for surface structures, surface topography, and site geology. Based on an analysis of the data, models for the Highly Damaged Zone (HDZ) and Excavation Damage Zone (EDZ) at both sites were developed. The HDZ was found to be approximately 5 to 10 cm in thickness, and the EDZ was found to extend between 15 and 35 cm below the excavation surface. Two-dimensional (2D) RES profiling generated the most reliable assessment of the HDZ, whereas chargeability data and GPR data were more useful in the estimation of the EDZ dimensions. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The long term safety case for an underground nuclear waste repository is based in large part on the management of the zones of damaged rock around excavations and shafts. These can form pathways for advective or diffusive radionuclide release and transport beyond the repository rooms and can create short circuit pathways along backfilled shafts. The Excavation Damage Zone (EDZ) and Highly Damaged Zone (HDZ) are areas where micro-scale and macroscopic fractures develop, respectively, around underground openings due to construction-induced damage and stress re-distribution. The definitions of EDZ and HDZ used in this paper are (Diederichs et al., 2013): • Excavation Damaged Zone (EDZ) — region of inelastic but discontinuous induced fractures or non-interacting displacements on existing structure. Intensity of damage increases towards the excavation surface. • Highly Damaged Zone (HDZ) — region of discrete and continuous macro-fractures and/or opening or slip along existing structures due to unloading and associated strains during void creation. May exist without any construction induced damage. ⁎ Corresponding author at: Department of Geology and Geological Engineering, Berthoud Hall, 1516 Illinois Street, Golden, Colorardo, 80401. E-mail address: [email protected] (G. Walton).
http://dx.doi.org/10.1016/j.enggeo.2015.07.010 0013-7952/© 2015 Elsevier B.V. All rights reserved.
The understanding, prediction, identification, management, and monitoring of these zones are of importance to those developing excavations with strict design requirements, as relatively small amounts of damage can lead to issues. In particular, the development of EDZ and HDZ contributes to the decreased excavation stability in both the short and long terms, as well as increased rockmass permeability (Bossart et al., 2002). In the case of nuclear waste storage in deep geological repositories, these issues are of significant concern due to the potential flow pathway along the HDZ/EDZ (Diederichs et al., 2013; Tsang et al., 2005) and the degree of damage needs to be identified to verify design parameters. In a perfectly homogenous, isotropic medium, where all damage is induced by stress redistribution (i.e. the construction process itself induces negligible damage), gradational zones of damage can be expected to develop, with the degree of damage decreasing away from the excavation as the stress concentration decreases (see Fig. 1). Where natural fractures exist and/or damage is generated by the construction process, it is difficult to predict how new fractured zones might develop. In particular, micro-scale EDZ damage may irregularly clump around macroscale HDZ fractures or natural fractures, and this damage may be due to highly local stress concentrations, or due to dynamic loads associated with construction. Drilling, either to obtain core samples from near the boundary of an excavation or for the installation of down-hole equipment to determine
G. Walton et al. / Engineering Geology 196 (2015) 210–221
211
Fig. 1. Schematic diagram illustrating a simple conceptual model for HDZ/EDZ with potential influences on some physical properties of interest described; note that σ1 and σ3 represent the major and minor principal stresses, respectively.
HDZ/EDZ dimensions can lead to further damage (Souley et al., 2001). Therefore, there is a desire to develop efficient and non-invasive methods for characterizing and monitoring the EDZ and HDZ. Geophysical methods have already been demonstrated as useful tools for noninvasive EDZ/HDZ detection. A large amount of research has been conducted on the use of active and passive seismic methods for this purpose (see Carlson and Young, 1993; Cabrera et al., 1999; Backblom and Martin, 1999; Cosma et al., 2001; Alheid et al., 2002; Malmgren et al., 2007, for example). Ground Penetrating Radar (GPR) and resistivity methods have shown potential in a few isolated studies, although their use requires further testing and development (Scott et al., 1968; Kruschwitz and Yaramanci, 2004; Suzuki et al., 2004; Gibert et al., 2006; Silvast and Wiljanen, 2008; Kantia et al., 2013; Lesparre et al., 2013). This study aims to further test these methods through the comparison of multiple different data sources to test for consistency of the results
obtained. Also, by providing a systematic evaluation of the influence of some resistivity survey parameters, the authors aim to provide guidelines for future geophysical surveys used for similar applications. 2. Test sites In October, 2012, geophysical surveys were performed in two niches at the Äspö Hard Rock Laboratory (HRL), 20 km North of Oskarshamn, Sweden. The niches (NASA 2376A and NASA 2715A) are located at depths of 315 m and 358 m below surface level, respectively (Fig. 2). Both sites were constructed in Äspö diorite, a fine to medium grained igneous rock with ~1 cm potassium-feldspar megacrysts, and some finegrained granite present in dykes and veins. The diorite has 0.4% porosity and the granitic dykes have 0.2% porosity (Johansson et al., 1998). Although it is difficult to estimate the porosity increase expected in the
Fig. 2. Perspective view of Äspö laboratory layout with the two niches (NASA 2715 and 2376) marked for this study.
212
G. Walton et al. / Engineering Geology 196 (2015) 210–221
HDZ/EDZ, it has been demonstrated that bulk hydraulic conductivities in the damaged zone around excavations can increase to over 1000 times their baseline value (Pusch, 1989). Saline groundwater inflow was noted at both of these sites, although inflow was notably more significant at the lower site (NASA 2715A), where a pump was required to minimize the amount of water pooled on the floor during the investigation. At NASA 2715A, several blocks of rock along the excavation wall are bounded by open fractures visible at surface. Striking these blocks with a rock hammer indicated that they are partially detached from the intact portion of the rockmass, suggesting that the fractures observed at surface persist deeper into the rockmass. At the sites of interest, no damage was expected to be induced by the concentration of in-situ stresses around the excavation. This is based on the high strength of both major rock units (crack initiation threshold of 80–90 MPa and uniaxial compressive strength of approximately 200 MPa from Jacobsson, 2006) relative to the magnitude of the insitu stresses near the test sites (major principal stress ~ 10 MPa in the horizontal plane from Ask, 2003). This conclusion is consistent with the empirical damage prediction charts presented by Martin et al. (1999) & Diederichs (2007). Instead, it is likely that the majority of the damage at both sites was induced by the construction process. Because of the natural blockiness of the rockmass, a likely source of the damage present at the two study sites is the opening of natural fractures during unloading. Previous investigations elsewhere in Äspö showed that in an area constructed using the same excavation method (drill and blast), the EDZ was determined to be approximately 30 cm (Emsley et al., 1997). This value was used as an initial expected value to set survey depth targets for NASA 2715A and NASA 2376A. Deviations from this value are expected, however, due to the potential differences in stress, structure, and particularly the degree of damaged rock removed prior to the investigation, which is unknown for the current study. The geophysical surveys conducted at the sites consisted of resistivity (RES)/time-domain induced polarization (IP) and GPR. In addition to these techniques for detecting the HDZ/EDZ, the authors have identified Light Detection and Ranging (LiDAR) as a useful tool to augment the capabilities of geophysical surveys. In particular, the LiDAR data has been used as a reference for geology and surface structures at the test sites, and has been used to locate the true three-dimensional (3D) positions of the resistivity array electrodes for use in the resistivity data inversion. Both RES/IP and GPR surveys were conducted along the same horizontal line at each site. In both cases, the line was parallel to the long axis of the niche, and was located along the sidewall of the niche. Figs. 3 (NASA 2376A) and 4 (NASA 2715A) illustrate the locations of the profiles. In addition, boreholes were cored at both sites (three at NASA 2376A and four at NASA 2715A). The data obtained from the boreholes were compared to the geophysical results to produce an integrated interpretation.
Fig. 4. Annotated LiDAR image of the NASA 2715A site with line chainages indicated (0 cm to 480 cm); borehole locations are labeled and are marked with an “x”.
3. Resistivity/IP surveys RES/IP surveys are based on the principle that by measuring the potential difference induced by current injection at the surface and then calculating the apparent resistive and capacitive properties of the subsurface based on these measurements, a map of true resistivity and chargeability can be generated for the subsurface. For more details on this type of survey and its practical applications, the reader is referred to Kearey et al. (2002) and Loke et al. (2013). The key materials of interest at the Äspö HRL site are intermediatefelsic intrusive igneous rocks (ρ ≈ 1,000 Ω•m–1,000,000 Ω•m) and saline groundwater (ρ ≈ 0.2 Ω•m–1 Ω•m) (resistivity values from Sharma, 1997 and Kirsch, 2006). Because of the vast differences between the groundwater and rock matrix resistivities, as well as the fact that current strongly follows pathways of minimal resistance, the resistivities will generally be lower in damaged zones. Even if the microscopic damage present in the rock has not been 100% saturated with groundwater at the time of surveying, there should still be a resistivity decrease associated with the flow of any new volume of water into the damaged zone (Archie, 1942). The resistivity data will also be more sensitive to water present in macroscopic fractures as these present relatively continuous, interconnected pathways for current flow — as compared to grain-scale pores associated with damage to the rock matrix. In general, it cannot be assumed that any given individual fracture will be clearly identifiable in the resistivity data, as joint saturation may vary between full saturation and localized areas of flow depending on the groundwater flow regime. In cases where portions of the rockmass are drained, the resistivity in areas of damage is greater than or equal to those of the intact rockmass, due to an increased presence of high resistivity voids (Suzuki et al., 2004). With respect to the IP effect, a decreased resistivity in the presence of salt-water will correspond to an increasingly free movement of charge, and therefore a decreased chargeability; this phenomenon can be conceptualized in terms of the electrical double layer, and the influence of increased ion concentration on reducing the double layer's thickness (Kearey et al., 2002). 3.1. Survey setup
Fig. 3. Annotated LiDAR image of the NASA 2376A site with line chainages indicated (0 cm to 600 cm); borehole locations are labeled and marked with an “x”.
The resistivity surveys were performed using a multi-gradient array measurement protocol with an ABEM Terrameter system (Dahlin and Zhou, 2006). The multi-gradient array consists of measurements taken with a potential dipole that moves laterally along the survey line between two current electrodes. By varying the surficial spacing of the current electrodes, information can be obtained with respect to resistivity changes into the rockmass. Measurements were collected along a 6 m long line at NASA 2376A and a 4.8 m long line at NASA 2715A. The line lengths were as long as possible without risking the possibility of significant off-line influences from either end of the niches. Given the
G. Walton et al. / Engineering Geology 196 (2015) 210–221
small scale of the damaged zones, a small electrode spacing (6 cm) was used to achieve a high survey resolution. It must be noted that the use of electrodes which have non-negligible size compared to their spacing in this case was a significant source of potential error. As can be seen from the results presented in Sections 3.2 and 6, however, this did not appear to lead to any significant artifacts, and did not prevent reasonable results from being obtained. Standard half-space geometrical factors were used in the conversion of resistance to resistivity. These factors have been shown to be reasonable for Wenner array line lengths up to 18 m in the case of a 3 m diameter excavation (Kruschwitz and Yaramanci, 2004). As such, a line length of 4.8 m–6 m for an excavation with a 5 m diameter justifies the use of half-space values for most array types. To couple the measurement system to the rockmass, small holes (8 mm diameter, ~1 cm deep) were drilled into the rockmass at the desired electrode locations. Metal anchors were then inserted into these holes, their bases were expanded to press against the surface of the hole, and screws were threaded into the anchors to serve as the electrodes. The resistivity survey setup at NASA 2376A is illustrated in Fig. 5, with the coupling method described above illustrated in the inset. The data cables were strategically positioned without crossing sequential connections that could cause errors in the IP data. This was unavoidable at NASA 2715A, and as such the IP data from this site was discarded. The setup described above is the baseline survey. Other survey parameters were tested at both sites in an effort to better constrain what parameters provide an optimal survey design for EDZ/HDZ detection. These parameters are discussed individually in Section 3.3. 3.2. Data processing The data were inverted using RES2DINV (Loke, 2001). The inversion method selected is a cell-based method which minimizes the L1 norm of the error vector; this effectively minimizes the normalized absolute differences between synthetic (model) data and measured data (hereafter referred to as the “absolute error” of the data residuals) (Ellis et al., 1993; Ellis and Oldenburg, 1994). The irregular topography of the excavation surface was extracted from the LiDAR data and input into RES2DINV for incorporation into the inversion model. Six iterations were used for the resistivity inversion. Performing iterations beyond this was found to have a minimal effect on the absolute error of the data residuals associated with the resistivity model, which was found to typically be around 5%. Qualitatively speaking, the results obtained for the sides and the lower portion of the model can be considered the most uncertain. Because the features of interest are very shallow relative to the length of
Fig. 5. Resistivity line setup at NASA 2376A site with anchor and screw electrode combination and installed electrodes shown in the inset image.
213
the line (b30 cm vs. 600 cm) and there are very few lateral variations observed, none of the results were considered to be adversely affected by a loss in signal sensitivity. In other cases where significant changes in resistivity may be associated with deeper features, it is appropriate to evaluate the sensitivity of the result as a function of depth (e.g. Oldenburg and Li, 1999; Caterina et al., 2014). 3.3. Results The baseline inversion results for resistivity and chargeability at NASA 2376A are shown in Fig. 6. The main feature that appears in the resistivity model is a very low resistivity layer (~ 200 Ω•m to 1,500 Ω•m) near surface, which extends to depths of approximately 5 cm to 8 cm beneath the surface. This result is consistent with the presence of an HDZ with fluid-carrying macro-fractures near the excavation wall. Many of the deeper changes in resistivity are caused by edge effects in the inversion, and for the purposes of delineating the HDZ/ EDZ, this zone can generally be considered as intact rock. The exception is the low resistivity area around 320 cm along the line (near the center) which has been interpreted as larger HDZ (see Fig. 6b). In observing the results, the largest gradient in the logarithm of resistivity was used to interpret the limits of the damage zones (Nguyen et al., 2005). The chargeability results for NASA 2376A show a similar trend to the resistivity results — a low chargeability layer exists to a depth between 15 cm and 20 cm beneath the surface, and is underlain by a higher chargeability material. As stated earlier, the correlation between low resistivities and low chargeabilities was hypothesized based on the presence of water-bearing fractures in this zone, although a clear discrepancy between the size of these zones requires further explanation. The chargeability response may be more sensitive to the opening of grain-scale isolated cracks, whereas the resistivity response requires larger, interconnected fractures and/or fracture systems to generate lower values. As such, the chargeability results are considered to represent the EDZ rather than the HDZ. Given the dense survey arrangement, however, there is also a possibility that the results obtained are an artifact of an inductive coupling effect. The resistivity model generated for NASA 2715A (Fig. 7a) shows similar features to the model from the NASA 2376A (Fig. 6a), with a low resistivity layer extending to 5 cm to 12 cm below the surface. Note the decreased resistivity values in this zone relative to the low resistivity zone at NASA 2376A; this is consistent with the observed difference in the saline water inflow rate. Although the difference in HDZ depth is minimal, a difference in the water content in the HDZ explains the lower resistivity values. The implication of this finding is that, through careful calibration, the absolute values of model resistivities could be used to obtain formation factors, which could potentially be used as a relative measure of HDZ permeability in areas where saline water is present. One more feature to note is the low resistivity zone below the near surface high resistivity zone around 240 cm along the line (see Fig. 7a). This zone corresponds with a large, joint-bound block which was deemed to be loose during the data collection. In this case, the resistivity model shows the presence of an open fracture or fracture system ~15 cm to 25 cm below surface. 3.3.1. Array type influence To help provide guidelines for future surveys, the multi-gradient array was compared to a dipole–dipole array. The multi-gradient and dipole–dipole inversion results for NASA 2715A (Fig. 7a and b, respectively) show similar features for the entire extent of the profile. The key difference is that the dipole–dipole model shows greater contrasts in resistivity between high and low resistivity zones, but for the purposes of interpretation, there is negligible value added. Given that the dipole–dipole model has a higher degree of error associated with it than the multi-gradient array, and that the dipole–dipole data collection was more time consuming, the authors have concluded that the multi-
214
G. Walton et al. / Engineering Geology 196 (2015) 210–221
Fig. 6. Multi-gradient resistivity (a) and induced polarization inversion (c) results at NASA 2376A (note that the error associated with the induced polarization result is presented in chargeability units, not as a percentage); (b) shows the interpreted HDZ/EDZ boundary on a zoomed section (area indicated in (a)) of the resistivity results.
gradient array is more appropriate for collecting 2D resistivity data for damage detection. 3.3.2. Electrode spacing influence It is of practical importance to understand how survey resolution deteriorates with increasing electrode spacing, since tighter electrode spacings require more data to be collected for a given line length. The results obtained using a multi-gradient array with 6 cm and 12 cm electrode spacings at NASA 2376A are illustrated in Fig. 8a and b, respectively. It appears that the transition from the lower model resistivities to higher model resistivities (transition at approximately 7,000 Ω•m)
occurs at depths ranging from ~6 cm to ~15 cm in the lower resolution case (as opposed to ~5 cm to ~8 cm in the high resolution case). In this case, it appears that the effective smoothing of the model induced by using a larger electrode spacing has resulted in a deeper mapped HDZ/ EDZ boundary. There is a risk, however, that if the electrode spacing is made too large relative to the size of the HDZ that the smaller lower resistivity zone may be missed in the inversion. This is particularly likely to occur where there is a relatively small contrast between the HDZ resistivity and the intact rock resistivity. At NASA 2715A, data were collected using 6 cm and 12 cm spacing, as well as at 3 cm spacing for the middle portion of the line. The
Fig. 7. Comparison of multi-gradient (a) and dipole–dipole (b) resistivity inversion results for NASA 2715A.
G. Walton et al. / Engineering Geology 196 (2015) 210–221
215
Fig. 8. Multi-gradient resistivity inversion results at NASA 2376A using 6 cm electrode spacing (a) and 12 cm electrode spacing (b).
difference in the apparent HDZ as detected by the 3 cm spacing and 6 cm spacing surveys is insignificant compared to the difference between the 6 cm spacing and 12 cm spacing results, as illustrated in Fig. 9. The data corresponding to the largest electrode spacing resulted in an inverted model with a relatively large HDZ thickness estimation (~10 cm to ~16 cm). Based on these results, it is clear that there is a maximum acceptable electrode spacing for damage detection surveys to obtain accurate results, and that below this threshold there is minimal improvement in
damage zone resolution as electrode spacing is further decreased. The authors suggest that for similar conditions, an electrode spacing on the order of the expected damage zone thickness (based on numerical modelling or field observations) should be used. 4. GPR surveys Ground Penetrating Radar technology operates by emitting an electromagnetic (EM) pulse from an antenna and recording the arrival time
Fig. 9. Multi-gradient resistivity inversion results at NASA 2715A using 6 cm electrode spacing (a) and 12 cm electrode spacing (b). The black boxes indicate the area for which a model was generated based on limited data acquired using 3 cm electrode spacing. Inversion results for this zone are shown for 3 cm (c), 6 cm (d), and 12 cm (e) electrode spacing (these three models share a common scale bar).
216
G. Walton et al. / Engineering Geology 196 (2015) 210–221
of the reflected pulses that return to a receiver antenna. The recorded travel-times of reflections can then be converted into estimated reflector depths. GPR pulses are reflected whenever they encounter a discontinuity in wave velocity. In general, differences in conductivity, magnetic susceptibility, and dielectric permittivity can cause such a discontinuity. For most geomaterials with low magnetic susceptibility, the relative dielectric permittivity (εr) is the key parameter that controls the wave velocity and therefore the reflectivity of material interfaces (Saksa et al., 2005). In terms of identifying individual HDZ fractures, previous researchers have published successful results mapping subsurface discontinuities using GPR. The specific applications of the technique include the investigation of quarries (Grandjean and Gourry, 1996; Grasmueck, 1996), exposed cliffs (Dussauge-Pessier et al., 2003), medieval masonry (Arias et al., 2007), pavement damage (Saarenketo and Scullion, 2000; Bendetto and Pensa, 2007), and excavation damage (Silvast and Wiljanen, 2008). The reflectivity of subsurface fractures is caused by the notable permittivity contrast between a rock matrix and either air or water in an open fracture. A key limitation of the GPR surveys with respect to identifying HDZ fractures at this site is that fractures infilled by mineralization are likely to generate reflections similar to those generated by open fractures. It is also difficult to delineate fractures in the damaged zone due to the high potential for wave scattering and irregular or out-of-plane reflectors. Frequency dependent conductive losses and wave scattering are both key controls on EM wave attenuation. The relatively high electrical conductivity of pore water in damaged zones and the increased potential for wave scattering losses both suggest that the EDZ should be detectable as an area of high wave scattering and energy loss, particularly in higher frequency signal components. Silvast and Wiljanen (2008) have demonstrated the practical use of a proprietary dispersion algorithm to delineate the EDZ at a Finnish spent nuclear fuel repository test site, ONKALO. Based on their results, the outer edge of the near surface scattering zone observed in high frequency data may be correlated with the boundary between the EDZ and intact rockmass. 4.1. Survey setup The GPR surveys at Äspö HRL were conducted using a custom-built step-frequency GPR system developed at the Norwegian Geotechnical Institute. Two antennae were used with the system in a monostatic configuration. The lower frequency antenna collected data using frequencies between 600 MHz and 2,600 MHz, with a peak in signal power around 1,500 MHz. The higher frequency antenna collected data using frequencies between 800 MHz and 3,000 MHz, with a peak in signal power around 2,500 MHz. Sampling frequencies of approximately 10.23 GHz and 11.26 GHz were used for the low and high frequency antennae, respectively. The GPR system recorded data in a continuous measurement mode (one trace per second), so to appropriately locate each trace, the trace numbers corresponding to chalk marks made at 3 cm intervals were recorded by the operator. An average trace spacing of approximately 1.5 cm was obtained for each line completed. Data were recorded a minimum of two times along each line using each antenna for two perpendicular antenna orientations. Due to the somewhat irregular surface topography at both sites, maintaining consistent coupling during data collection was challenging. For this reason, the use of air-coupled systems which do not require direct contact with the rock surface is recommended for future studies.
antenna remained static. The true trace locations were recorded every 3 cm; within each 3 cm bin, traces were stacked to improve the signal to noise ratio. Multiple data sets collected at the same site using the same antenna were also stacked. Since a preliminary analysis of the data showed no difference between data collected using different antenna orientations, both orientations were used for the stacking. The number of traces stacked for a given location was variable due to the variable number of traces collected within each 3 cm bin, but an average value of eight stacks is representative for most locations (four data collection instances times two traces per bin per instance). To aid in the interpretation of potential fracture locations, a bandpass filter (500 MHz to 3 GHz) was applied to the lower frequency data followed by an automatic envelope correction (3.6 ns window length). It was found that for the higher frequency data, amplitude scaling primarily amplified noise at later times; since the outer envelope of the zone of high reflectivity may be representative of the EDZ in the high frequency data, the final sections were produced without any amplitude scaling. For the purposes of visualization, data were clipped two standard deviations from the mean amplitude. Time to depth conversions were performed using an assumed dielectric permittivity of 7, which is typical of granites in this region (Saksa et al., 2005; Kantia et al., 2013). Because of the monostatic configuration of the GPR system, it was not possible to collect common midpoint (CMP) data to improve the velocity estimate for the rock. The velocity used, however, appears to be relatively reliable based on previous experience in the area, and the good correlation between interpreted fracture locations and fractures observed in boreholes (see Section 6). 4.3. Results & interpretation The stacked GPR sections produced for NASA 2376A and NASA 2715A are shown in Figs. 10 and 11, respectively. The lower frequency data was found to provide a clearer indication of potential fracture locations at both sites. As can be seen in the data, the interpretation of fractures from the data is relatively difficult due to the presence of noise (likely due to the irregularity of the surveyed surface). A rough interpretation can still be made, however, and dashed lines have been placed on the profiles to delineate what were initially identified as potential fracture locations independent of the results of the other surveys (Figs. 10b and 11b). These are likely not all of the fractures within the rockmass, and, as stated above, many of these fractures may be natural, infilled, low porosity fractures. In GPR data, the extent of the EDZ should correspond to a near surface high reflector density zone, particularly for higher frequency components. This is because the large number of small-scale permittivity discontinuities present in damaged rock causes large amounts of the original signal to scatter back to the antenna. There are some issues associated with attenuation effects when making a direct comparison between the extent of the reflective zone and the EDZ, but studies have shown that it can still serve as a good approximation of the EDZ (Silvast and Wiljanen, 2008; Kantia et al., 2013). If the edge of the reflective zone in the higher frequency data is assumed to correspond exactly to the edge of the EDZ, the edge of the EDZ is interpreted as being at approximately 25 cm depth at NASA 2376A and 30 cm depth at NASA 2715A. These results correlate roughly with the chargeability results at NASA 2376A (EDZ boundary between 15 cm and 25 cm). 5. Borehole data
4.2. Data processing The raw GPR data were imported into MATLAB (MathWorks, 2010). Using the operator's field notes, the data were assigned spatial locations. The mean trace was subtracted from each data set, as approximated by the first 30 traces which were recorded at the start of each line while the
Based on a preliminary analysis of the data performed on-site, borehole locations were selected. Boreholes were drilled to depths of 100 cm at 240 cm, 326 cm, and 468 cm along the line at NASA 2376A and to depths of 80 cm at 120 cm, 243 cm, 297 cm, and 364 cm along the line at NASA 2715A. The exact locations of the core drillings are
G. Walton et al. / Engineering Geology 196 (2015) 210–221
217
Fig. 10. Stacked GPR sections for the NASA 2376A site. Panel (a) shows the filtered and amplitude corrected lower frequency data; panel (b) shows an interpretation of potential fracture locations identified prior to the analysis of the borehole data; panel (c) shows the unfiltered higher frequency data.
illustrated in Figs. 3 and 4. Core was recovered from all holes, and the observed fractures were logged for depth, orientation, and condition. Using a borehole camera, fractures observed in the core were correlated with in-situ fractures to verify the logged fracture conditions, and also to better constrain the three-dimensional orientation of the fractures. Unfortunately, no information could be reliably determined on the saturation condition of the fractures, as the use of drilling fluid in the coring process disturbed the in-situ hydrogeological condition. Images obtained using the borehole camera showing the variety of fracture conditions encountered are presented in Fig. 12. Although all of the fractures depicted (natural, sealed, and open) have the potential
to appear as reflectors in the GPR data, only the more permeable fractures are suspected to have a significant influence on the resistivity data.
6. Integrated interpretation In an effort to validate the hypotheses formed based on the geophysical data alone, the fracture locations and conditions were plotted and overlaid with the geophysical data. The orientations of the fractures are projected onto the survey planes, since many of the fractures were found to have an out-of-plane component to their attitude.
Fig. 11. Stacked GPR sections for the NASA 2715A site. Panel (a) shows the filtered and amplitude corrected lower frequency data; panel (b) shows an interpretation of potential fracture locations identified prior to the analysis of the borehole data; panel(c) shows the unfiltered higher frequency data.
218
G. Walton et al. / Engineering Geology 196 (2015) 210–221
Fig. 12. Different fractures observed using the borehole camera. The upper row illustrates fractures which have been sealed/infilled, the middle row illustrates a transition to increasingly open fractures, and the fractures on the bottom row are open with apertures approaching 1 mm. The resistivities associated with these fractures are expected to decrease from the top left corner to the bottom right corner.
6.1. NASA 2376A The multi-gradient profile and lower frequency GPR stacked section for the NASA 2376A site are shown in Fig. 13 with mapped fractures overlain. In the resistivity data (Fig. 13a and b), the key finding is that each borehole contained at least one open and/or weathered fracture within the upper 10 cm. Borehole A contained three open fractures at depths of 3 cm, 5 cm, and 8 cm, all of which were roughly excavation parallel. These findings suggest that the HDZ extends to a depth of 5 cm to 8 cm, which is consistent with the size of the zones imaged as having low resistivity. Similar open fractures within the upper 5 cm were seen in boreholes B and C. In the GPR data (Fig. 13c) it is difficult to correlate individual reflectors to individual fractures because of the high number of reflected arrivals in the damaged zone. The fractures mapped deeper in all three
boreholes correlate well with nearby reflectors, however, which is consistent with the use of an appropriate dielectric permittivity. Also, a comparison of the individual stacked traces at the locations of each borehole and the mapped fractures in the boreholes shows that there is good correlation between reflector peaks and mapped fractures (Fig. 13d). The borehole observations (fracture location, orientation, and condition) provide no information with respect to the micro-scale damage which was potentially detected by the GPR and chargeability surveys. Based on the consistency of the results achieved using these distinct methods, however, it can be concluded that the EDZ at NASA 2376A is variable in extent, and exists to depths within the range of 15 cm to 35 cm along the profile. This is also consistent with the expected value based on the work of Emsley et al. (1997).
G. Walton et al. / Engineering Geology 196 (2015) 210–221
219
Fig. 13. (a) Inverted resistivity model with interpreted HDZ/EDZ limits and (b) zoomed views of the borehole locations with observed fractures indicated; (c) stacked GPR data (lower frequency) overlain with known fracture locations from boreholes at NASA 2376A and interpreted reflectors; (d) individual GPR traces at borehole locations with mapped fractures illustrated as vertical lines.
6.2. NASA 2715A The multi-gradient profile and the lower frequency GPR stacked section for the NASA 2715A site are illustrated in Fig. 14 with mapped fractures overlain. In the resistivity data (Fig. 14a and b), the presence of an HDZ in the upper 10 cm is supported by the presence of near surface fractures in boreholes D and F. In borehole G, the slightly weathered fractures may be related to the lower resistivity features in the model that lie just above and just below their locations, or they may have sufficiently low water content that they had little influence on the data. In this case, the lower resistivity zones would be associated with smaller scale damage which was not visually detectable in the core. The most interesting results are from borehole E. These results show that there is indeed an open fracture which separates the relatively intact but loose block at this location from the intact rockmass behind. There is also a second open fracture at the base of the relatively thick low resistivity zone which exists behind this block. It is possible that this low resistivity zone corresponds to a smaller-scale damage induced preferentially between the two open fractures. Again, in the GPR data (Fig. 14c) it is difficult to correlate individual fractures with any degree of certainty, but there are some notable
correlations between fracture and reflector locations. In particular, the borehole and GPR data are consistent for fractures lower in holes E & F and near surface in hole D. Again, the individual trace reflector peaks along the borehole alignments tend to agree with the mapped fracture locations (Fig. 14d). Based on the GPR survey, it can be concluded that the EDZ at NASA 2715A extends to variable depths between 15 cm and 35 cm below the excavation surface. 7. Conclusions & recommendations The analysis of the geophysical research conducted at the Äspö HRL site demonstrates that the RES/IP and GPR methods tested have value in detecting the HDZ/EDZ in a hard rock environment. In particular, 2D resistivity profiling proved most useful in delineating areas where macroscopic open fractures were observed (HDZ), whereas the GPR proved useful in identifying individual fractures and in estimating the EDZ dimensions. When performing 2D resistivity profiling, the multi-gradient array was found to be suitable, with electrode spacings on the order of the expected HDZ thickness being optimal. The lower frequency GPR antenna (centered around ~ 1,500 MHz) was found to be most useful in
220
G. Walton et al. / Engineering Geology 196 (2015) 210–221
Fig. 14. (a) Inverted resistivity model with interpreted HDZ/EDZ limits and (b) zoomed views of the borehole locations with observed fractures indicated; (c) stacked GPR data (lower frequency) overlain with known fracture locations from boreholes at NASA 2376A and interpreted reflectors; (d) individual GPR traces at borehole locations with mapped fractures illustrated as vertical lines.
identifying individual fractures, whereas the higher frequency antenna (centered around ~ 2,500 MHz) provided EDZ dimension estimates more consistent with those provided by other methods (as shown in Figs. 10 and 11). With respect to underground nuclear waste disposal, the most significant impact of the EDZ and HDZ is their potential to act as dominant groundwater flow pathways in the long term. To design effective measures such as cut-off seals to mitigate the effects of these zones following construction requires a knowledge of their geometries. As has been demonstrated in this study, RES/IP and GPR surveying methods both have the potential to provide valuable information on the HDZ/EDZ without the risk of inducing further damage in the rockmass. In the future, it may be of interest to use a combination of approaches as demonstrated in this study for repeated surveys to characterize changes in the HDZ/EDZ over time (as demonstrated using the resistivity method by Gibert et al., 2006 and Lesparre et al., 2013). Such surveys conducted over a large time-scale could provide information on the potential for
HDZ/EDZ self-sealing with time, which could have significant implications for deep geological repository performance. Acknowledgments The authors would like to thank SKB for providing access to the test site for this study. Tomas Lehtimaki played a critical role in coordinating the data collection activities. This work would not have been possible without funding provided by the Nuclear Waste Management Organization of Canada (NWMO) (NWMO-EDZ-2), the Natural Sciences and Engineering Research Council of Canada (NSERC) (CGS-D), and the Norwegian Research Council. References Alheid, H.-J., Schuster, K., Kruchsitz, S., Yaramanci, U., 2002. Seismic and geolectic measurements for the characterization of the excavation damage zone around tunnels in Opalinus Clay. Z. Angew. Geol. 48 (2), 48–55.
G. Walton et al. / Engineering Geology 196 (2015) 210–221 Archie, G.E., 1942. The electrical resistivity log as an aid in determining some reservoir characteristics. Trans. Am. Inst. Min. Metall. Eng. 146, 54–62. http://dx.doi.org/10. 2118/942054-G. Arias, P., Armesto, J., Di-Capua, D., Gonzalez-Drigo, R., Lorenzo, H., Perez-Gracia, V., 2007. Digital photogrammetry, GPR and computational analysis of structural damages in a mediaeval bridge. Eng. Fail. Anal. 14, 1444–1457. http://dx.doi.org/10.1016/j. engfailanal.2007.02.001. Ask, D., 2003. Evaluation of measurement-related uncertainties in the analysis of overcoring rock stress data from Aspo HRL, Sweden: a case study. Int. J. Rock Mech. Min. Sci. 40, 1173–1187. http://dx.doi.org/10.1016/s1365-1609(03)00114-x. Backblom, G., Martin, C.D., 1999. Recent experiments in hard rocks to study the excavation response: implications for the performance of a nuclear waste geological repository. Tunn. Undergr. Space Technol. 14 (3), 377–394. http://dx.doi.org/10.1016/ s0886-7798(99)00053-x. Bendetto, A., Pensa, S., 2007. Indirect diagnosis of pavement structural damages using surface GPR reflection techniques. J. Appl. Geophys. 62, 107–123. http://dx.doi.org/10. 1016/j.jappgeo.2006.09.001. Bossart, P., Meier, P.M., Moeri, A., Trick, T., Mayor, J.C., 2002. Geological and hydraulic characterization of the excavation disturbed zone in the Opalinus Clay of the Mont Terri Rock Laboratory. Eng. Geol. 66, 19–38. http://dx.doi.org/10.1016/s00137952(01)00140-5. Cabrera, J., Volant, P., Baker, C., Pettitt, W., Young, R.P., 1999. Structural and geophysical investigations of the EDZ (Excavation Disturbed Zone) in indurated argillaceous media: the tunnel and the galleries of the IPSN Tournemire site (France). 37th US Symposium on Rock Mechanics vol. 2. Taylor & Francis, US. Carlson, S.R., Young, R.P., 1993. Acoustic emission and ultrasonic velocity study of excavation-induced microcrack damage at the Underground Research Laboratory. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 30, 901–907. http://dx.doi.org/10.1016/ 0148-9062(93)90042-c. Caterina, D., Hermans, T., Nguyen, F., 2014. Case studies of incorporation of prior information in electrical resistivity tomography: comparison of different approaches. Near Surf. Geophys. 12 (4), 451–465. Cosma, C., Olsson, O., Keskinen, J., Heikkinen, P., 2001. Seismic characterization of fracturing at the Äspö Hard Rock Laboratory, from the kilometer scale to the meter scale. Int. J. Rock Mech. Min. Sci. 38, 859–865. http://dx.doi.org/10.1016/s13651609(01)00051-x. Dahlin, T., Zhou, B., 2006. Multi-gradient array measurements for multichannel 2D resistivity imaging. Near Surf. Geophys. 4 (2), 113–123. http://dx.doi.org/10.3997/18730604.2005037. Diederichs, M.S., 2007. The 2003 Canadian Geotechnical Colloquium: mechanistic interpretation and practical application of damage and spalling prediction criteria for deep tunnelling. Can. Geotech. J. 44, 1082–1116. http://dx.doi.org/10.1139/t07-033. Diederichs, M., Lam, T., Jensen, M., Damjanac, B., Martin, D., 2013. Ultra-long term geomechanics design for a deep geological repository in sedimentary rock. Proc. 47th US Rock Mechanics/Geomechanics Symposium, San Francisco, California. Dussauge-Pessier, C., Wathelet, M., Jongmans, D., Hantz, D., Couturier, B., Sintes, M., 2003. Investigation of a fractured limestone cliff (Chartreuse Massif, France) using seismic tomography and ground-penetrating radar. Near Surf. Geophys. 1, 161–170. http:// dx.doi.org/10.3997/1873-0604.2003007. Ellis, R.G., Oldenburg, D.W., 1994. Applied geophysical inversion. Geophys. J. Int. 116, 5–11. Ellis, R.G., Farquharson, C.G., Oldenburg, D.W., 1993. Approximate inverse mapping inversion of the COPROD2 data. J. Geomagn. Geoelectr. 45, 1001–1002. Emsley, S., Olsson, O., Stenberg, L., Alheid, H.J., Falls, S., 1997. ZEDEX — a study of damage and disturbance from tunnel excavation by blasting and tunnel boring. Technical Report 97-30. Swedish Nuclear Fuel and Waste Management Co., p. 198. Gibert, D., Nicollin, F., Kergosien, B., Bossart, P., Nussbaum, C., Grislin-Mouezy, A., Conil, F., Hoteit, N., 2006. Electrical tomography monitoring of the excavation damaged zone of the Gallery 04 in the Mont Terri rock laboratory: field experiments, modelling, and relationship with structural geology. Appl. Clay Sci. 33, 21–34. http://dx.doi. org/10.1016/j.clay.2006.03.008. Grandjean, G., Gourry, J.C., 1996. GPR data processing for 3D fracture mapping in a marble quarry (Thassos, Greece). J. Appl. Geophys. 36, 19–30. http://dx.doi.org/10.1016/ s0926-9851(96)00029-8.
221
Grasmueck, M., 1996. 3-D ground-penetrating radar applied to fracture imaging in gneiss. Geophysics 61 (4), 1050–1064. http://dx.doi.org/10.1190/1.1444026. Jacobsson, L., 2006. Oskarshamn site investigation — borehole KLX10: uniaxial compression test of intact rock. SKB Technical Report: P-06-37. Johansson, H., Siitari-Kauppi, M., Skalberg, M., Tullborg, E.L., 1998. Diffusion pathways in crystalline rock — examples from Aspo-diorite and fine-grained granite. J. Contam. Hydrol. 35, 41–53. Kantia, P., Heikkenen, E., Cosma, C., Enescu, N., Lehtimaki, T., Kemppainen, K., 2013. Nondestructive testing of rock mass deformation in spent nuclear fuel disposal research facilities. Proc. 47th US Rock Mechanics/Geomechanics Symposium, San Francisco, California. Kearey, P., Brooks, M., Hill, I., 2002. An Introduction to Geophysical Exploration. 3rd ed. Blackwell Science, Cornwall, Great Britain. Kirsch, R., 2006. Grounwater Geophysics — A Tool for Hydrogeology. Springer. Kruschwitz, S., Yaramanci, U., 2004. Detection and characterization of the disturbed rock zone in claystone with the complex resistivity method. J. Appl. Geophys. 57, 63–79. http://dx.doi.org/10.1016/j.jappgeo.2004.09.003. Lesparre, N., Gibert, D., Nicollin, F., Nussbaum, C., Adler, A., 2013. Monitoring the excavation damaged zone by three-dimensional reconstruction of electrical resistivity. Geophys. J. Int. 195 (2), 972–984. http://dx.doi.org/10.1093/gji/ggt282. Loke, M.H., 2001. Electrical imaging surveys for environmental and engineering studies. A practical guide to 2-D and 3-D surveys. RES2DINV Manual. IRIS Instruments (www. iris-instruments.com). Loke, M.H., Chambers, J.E., Rucker, D.F., Kuras, O., Wilkinson, P.B., 2013. Recent developments in the direct-current geolectrical imaging method. J. Appl. Geophys. 95, 135–156. http://dx.doi.org/10.1016/j.jappgeo.2013.02.017. Malmgren, L., Saiang, D., Toyra, J., Bodare, A., 2007. The excavation disturbed zone (EDZ) at Kiirunavaara mine, Sweden — by seismic measurements. J. Appl. Geophys. 61, 1–15. http://dx.doi.org/10.1016/j.jappgeo.2006.04.004. Martin, C.D., Kaiser, P.K., McCreath, D.R., 1999. Hoek–Brown parameters for predicting the depth of brittle failure around tunnels. Can. Geotech. J. 36, 136–151. http://dx.doi.org/ 10.1139/cgj-36-1-136. MathWorks, 2010. MATLAB (Version 7.10.0.499) — Software. Nguyen, F., Garambois, S., Jongmans, D., Pirard, E., Loke, M.H., 2005. Image processing of 2D resistivity data for imaging faults. J. Appl. Geophys. 57, 260–277. Oldenburg, D.W., Li, Y., 1999. Estimating depth of investigation in DC resistivity and IP surveys. Geophysics 64 (2), 403–416. Pusch, R., 1989. Alteration of the hydraulic conductivity of rock by tunnel excavation. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 26 (1), 79–83. Saarenketo, T., Scullion, T., 2000. Road evaluation with ground penetrating radar. J. Appl. Geophys. 43, 119–138. http://dx.doi.org/10.1016/s0926-9851(99)00052-x. Saksa, P., Heikkinen, E., Lehtimäki, T., 2005. Geophysical radar method for safeguards application at Olkiluoto spent fuel disposal site in Finland. STUK Technical Report. STUK-YTO-TR 213. Scott, J.H., Lee, F.T., Carroll, R.B., Robinson, C.S., 1968. The relationship of geophysical measurements to engineering and construction parameters in the straight creek tunnel pilot bore, Colorado. Int. J. Rock Mech. Min. Sci. 5, 1–30. http://dx.doi.org/10.1016/ 0148-9062(68)90020-x. Sharma, P.V., 1997. Environmental and Engineering Geophysics. Cambridge University Press. Silvast, M., Wiljanen, B., 2008. ONKALO EDZ-measurements using Ground Penetrating Radar (GPR) method. Posiva Working Report 2008-58. Souley, M., Homand, F., Pepa, S., Hoxha, D., 2001. Damage-induced permeability changes in granite: a case example at the URL in Canada. Int. J. Rock Mech. Min. Sci. 38, 297–310. http://dx.doi.org/10.1016/s1365-1609(01)00002-8. Suzuki, K., Nakata, E., Minami, M., Hibino, E., Tani, T., Sakakibara, J., Yamada, N., 2004. Estimation of the zone of excavation disturbance around tunnels, using resistivity and acoustic tomography. Explor. Geophys. 35, 62–69. http://dx.doi.org/10.1071/ eg04062. Tsang, C.F., Bernier, F., Davies, C., 2005. Geohydromechanical processes in the Excavation Damaged Zone in crystalline rock, rock salt, and indurated and plastic clays—in the context of radioactive waste disposal. Int. J. Rock Mech. Min. Sci. 42 (1), 109–125. http://dx.doi.org/10.1016/j.ijrmms.2004.08.003.
Contents lists available at ScienceDirect
Engineering Geology journal homepage: www.elsevier.com/locate/enggeo
Non-invasive detection of fractures, fracture zones, and rock damage in a hard rock excavation — Experience from the Äspö Hard Rock Laboratory in Sweden G. Walton a,⁎, M. Lato b, H. Anschütz c, M.A. Perras d, M.S. Diederichs e a
Colorado School of Mines, Golden, Colorado RockSense GeoSolutions, Ottawa, Canada c Norwegian Geotechnical Institute, Oslo, Norway d ETH Zurich, Zurich, Switzerland e Queen's University, Kingston, Canada b
a r t i c l e
i n f o
Article history: Received 9 July 2014 Received in revised form 27 February 2015 Accepted 15 July 2015 Available online 21 July 2015 Keywords: Excavation Damage Zone Resistivity survey Ground Penetrating Radar LiDAR Geophysical mapping
a b s t r a c t A key requirement for licensing of the construction of underground repositories for nuclear waste is the demonstrated capability to verify design assumptions involving the presence and extent of the excavation damage zone around tunnels, shafts, emplacement holes and caverns. As part of ongoing work to select and refine key technologies and techniques towards this end, geophysical surveys were performed at two locations within the Äspö Hard Rock Laboratory in Sweden. Earth resistivity (RES), induced polarization (IP), and Ground Penetrating Radar (GPR) data were collected using a variety of survey parameters; Light Detection and Ranging (LiDAR) data were collected as a reference for surface structures, surface topography, and site geology. Based on an analysis of the data, models for the Highly Damaged Zone (HDZ) and Excavation Damage Zone (EDZ) at both sites were developed. The HDZ was found to be approximately 5 to 10 cm in thickness, and the EDZ was found to extend between 15 and 35 cm below the excavation surface. Two-dimensional (2D) RES profiling generated the most reliable assessment of the HDZ, whereas chargeability data and GPR data were more useful in the estimation of the EDZ dimensions. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The long term safety case for an underground nuclear waste repository is based in large part on the management of the zones of damaged rock around excavations and shafts. These can form pathways for advective or diffusive radionuclide release and transport beyond the repository rooms and can create short circuit pathways along backfilled shafts. The Excavation Damage Zone (EDZ) and Highly Damaged Zone (HDZ) are areas where micro-scale and macroscopic fractures develop, respectively, around underground openings due to construction-induced damage and stress re-distribution. The definitions of EDZ and HDZ used in this paper are (Diederichs et al., 2013): • Excavation Damaged Zone (EDZ) — region of inelastic but discontinuous induced fractures or non-interacting displacements on existing structure. Intensity of damage increases towards the excavation surface. • Highly Damaged Zone (HDZ) — region of discrete and continuous macro-fractures and/or opening or slip along existing structures due to unloading and associated strains during void creation. May exist without any construction induced damage. ⁎ Corresponding author at: Department of Geology and Geological Engineering, Berthoud Hall, 1516 Illinois Street, Golden, Colorardo, 80401. E-mail address: [email protected] (G. Walton).
http://dx.doi.org/10.1016/j.enggeo.2015.07.010 0013-7952/© 2015 Elsevier B.V. All rights reserved.
The understanding, prediction, identification, management, and monitoring of these zones are of importance to those developing excavations with strict design requirements, as relatively small amounts of damage can lead to issues. In particular, the development of EDZ and HDZ contributes to the decreased excavation stability in both the short and long terms, as well as increased rockmass permeability (Bossart et al., 2002). In the case of nuclear waste storage in deep geological repositories, these issues are of significant concern due to the potential flow pathway along the HDZ/EDZ (Diederichs et al., 2013; Tsang et al., 2005) and the degree of damage needs to be identified to verify design parameters. In a perfectly homogenous, isotropic medium, where all damage is induced by stress redistribution (i.e. the construction process itself induces negligible damage), gradational zones of damage can be expected to develop, with the degree of damage decreasing away from the excavation as the stress concentration decreases (see Fig. 1). Where natural fractures exist and/or damage is generated by the construction process, it is difficult to predict how new fractured zones might develop. In particular, micro-scale EDZ damage may irregularly clump around macroscale HDZ fractures or natural fractures, and this damage may be due to highly local stress concentrations, or due to dynamic loads associated with construction. Drilling, either to obtain core samples from near the boundary of an excavation or for the installation of down-hole equipment to determine
G. Walton et al. / Engineering Geology 196 (2015) 210–221
211
Fig. 1. Schematic diagram illustrating a simple conceptual model for HDZ/EDZ with potential influences on some physical properties of interest described; note that σ1 and σ3 represent the major and minor principal stresses, respectively.
HDZ/EDZ dimensions can lead to further damage (Souley et al., 2001). Therefore, there is a desire to develop efficient and non-invasive methods for characterizing and monitoring the EDZ and HDZ. Geophysical methods have already been demonstrated as useful tools for noninvasive EDZ/HDZ detection. A large amount of research has been conducted on the use of active and passive seismic methods for this purpose (see Carlson and Young, 1993; Cabrera et al., 1999; Backblom and Martin, 1999; Cosma et al., 2001; Alheid et al., 2002; Malmgren et al., 2007, for example). Ground Penetrating Radar (GPR) and resistivity methods have shown potential in a few isolated studies, although their use requires further testing and development (Scott et al., 1968; Kruschwitz and Yaramanci, 2004; Suzuki et al., 2004; Gibert et al., 2006; Silvast and Wiljanen, 2008; Kantia et al., 2013; Lesparre et al., 2013). This study aims to further test these methods through the comparison of multiple different data sources to test for consistency of the results
obtained. Also, by providing a systematic evaluation of the influence of some resistivity survey parameters, the authors aim to provide guidelines for future geophysical surveys used for similar applications. 2. Test sites In October, 2012, geophysical surveys were performed in two niches at the Äspö Hard Rock Laboratory (HRL), 20 km North of Oskarshamn, Sweden. The niches (NASA 2376A and NASA 2715A) are located at depths of 315 m and 358 m below surface level, respectively (Fig. 2). Both sites were constructed in Äspö diorite, a fine to medium grained igneous rock with ~1 cm potassium-feldspar megacrysts, and some finegrained granite present in dykes and veins. The diorite has 0.4% porosity and the granitic dykes have 0.2% porosity (Johansson et al., 1998). Although it is difficult to estimate the porosity increase expected in the
Fig. 2. Perspective view of Äspö laboratory layout with the two niches (NASA 2715 and 2376) marked for this study.
212
G. Walton et al. / Engineering Geology 196 (2015) 210–221
HDZ/EDZ, it has been demonstrated that bulk hydraulic conductivities in the damaged zone around excavations can increase to over 1000 times their baseline value (Pusch, 1989). Saline groundwater inflow was noted at both of these sites, although inflow was notably more significant at the lower site (NASA 2715A), where a pump was required to minimize the amount of water pooled on the floor during the investigation. At NASA 2715A, several blocks of rock along the excavation wall are bounded by open fractures visible at surface. Striking these blocks with a rock hammer indicated that they are partially detached from the intact portion of the rockmass, suggesting that the fractures observed at surface persist deeper into the rockmass. At the sites of interest, no damage was expected to be induced by the concentration of in-situ stresses around the excavation. This is based on the high strength of both major rock units (crack initiation threshold of 80–90 MPa and uniaxial compressive strength of approximately 200 MPa from Jacobsson, 2006) relative to the magnitude of the insitu stresses near the test sites (major principal stress ~ 10 MPa in the horizontal plane from Ask, 2003). This conclusion is consistent with the empirical damage prediction charts presented by Martin et al. (1999) & Diederichs (2007). Instead, it is likely that the majority of the damage at both sites was induced by the construction process. Because of the natural blockiness of the rockmass, a likely source of the damage present at the two study sites is the opening of natural fractures during unloading. Previous investigations elsewhere in Äspö showed that in an area constructed using the same excavation method (drill and blast), the EDZ was determined to be approximately 30 cm (Emsley et al., 1997). This value was used as an initial expected value to set survey depth targets for NASA 2715A and NASA 2376A. Deviations from this value are expected, however, due to the potential differences in stress, structure, and particularly the degree of damaged rock removed prior to the investigation, which is unknown for the current study. The geophysical surveys conducted at the sites consisted of resistivity (RES)/time-domain induced polarization (IP) and GPR. In addition to these techniques for detecting the HDZ/EDZ, the authors have identified Light Detection and Ranging (LiDAR) as a useful tool to augment the capabilities of geophysical surveys. In particular, the LiDAR data has been used as a reference for geology and surface structures at the test sites, and has been used to locate the true three-dimensional (3D) positions of the resistivity array electrodes for use in the resistivity data inversion. Both RES/IP and GPR surveys were conducted along the same horizontal line at each site. In both cases, the line was parallel to the long axis of the niche, and was located along the sidewall of the niche. Figs. 3 (NASA 2376A) and 4 (NASA 2715A) illustrate the locations of the profiles. In addition, boreholes were cored at both sites (three at NASA 2376A and four at NASA 2715A). The data obtained from the boreholes were compared to the geophysical results to produce an integrated interpretation.
Fig. 4. Annotated LiDAR image of the NASA 2715A site with line chainages indicated (0 cm to 480 cm); borehole locations are labeled and are marked with an “x”.
3. Resistivity/IP surveys RES/IP surveys are based on the principle that by measuring the potential difference induced by current injection at the surface and then calculating the apparent resistive and capacitive properties of the subsurface based on these measurements, a map of true resistivity and chargeability can be generated for the subsurface. For more details on this type of survey and its practical applications, the reader is referred to Kearey et al. (2002) and Loke et al. (2013). The key materials of interest at the Äspö HRL site are intermediatefelsic intrusive igneous rocks (ρ ≈ 1,000 Ω•m–1,000,000 Ω•m) and saline groundwater (ρ ≈ 0.2 Ω•m–1 Ω•m) (resistivity values from Sharma, 1997 and Kirsch, 2006). Because of the vast differences between the groundwater and rock matrix resistivities, as well as the fact that current strongly follows pathways of minimal resistance, the resistivities will generally be lower in damaged zones. Even if the microscopic damage present in the rock has not been 100% saturated with groundwater at the time of surveying, there should still be a resistivity decrease associated with the flow of any new volume of water into the damaged zone (Archie, 1942). The resistivity data will also be more sensitive to water present in macroscopic fractures as these present relatively continuous, interconnected pathways for current flow — as compared to grain-scale pores associated with damage to the rock matrix. In general, it cannot be assumed that any given individual fracture will be clearly identifiable in the resistivity data, as joint saturation may vary between full saturation and localized areas of flow depending on the groundwater flow regime. In cases where portions of the rockmass are drained, the resistivity in areas of damage is greater than or equal to those of the intact rockmass, due to an increased presence of high resistivity voids (Suzuki et al., 2004). With respect to the IP effect, a decreased resistivity in the presence of salt-water will correspond to an increasingly free movement of charge, and therefore a decreased chargeability; this phenomenon can be conceptualized in terms of the electrical double layer, and the influence of increased ion concentration on reducing the double layer's thickness (Kearey et al., 2002). 3.1. Survey setup
Fig. 3. Annotated LiDAR image of the NASA 2376A site with line chainages indicated (0 cm to 600 cm); borehole locations are labeled and marked with an “x”.
The resistivity surveys were performed using a multi-gradient array measurement protocol with an ABEM Terrameter system (Dahlin and Zhou, 2006). The multi-gradient array consists of measurements taken with a potential dipole that moves laterally along the survey line between two current electrodes. By varying the surficial spacing of the current electrodes, information can be obtained with respect to resistivity changes into the rockmass. Measurements were collected along a 6 m long line at NASA 2376A and a 4.8 m long line at NASA 2715A. The line lengths were as long as possible without risking the possibility of significant off-line influences from either end of the niches. Given the
G. Walton et al. / Engineering Geology 196 (2015) 210–221
small scale of the damaged zones, a small electrode spacing (6 cm) was used to achieve a high survey resolution. It must be noted that the use of electrodes which have non-negligible size compared to their spacing in this case was a significant source of potential error. As can be seen from the results presented in Sections 3.2 and 6, however, this did not appear to lead to any significant artifacts, and did not prevent reasonable results from being obtained. Standard half-space geometrical factors were used in the conversion of resistance to resistivity. These factors have been shown to be reasonable for Wenner array line lengths up to 18 m in the case of a 3 m diameter excavation (Kruschwitz and Yaramanci, 2004). As such, a line length of 4.8 m–6 m for an excavation with a 5 m diameter justifies the use of half-space values for most array types. To couple the measurement system to the rockmass, small holes (8 mm diameter, ~1 cm deep) were drilled into the rockmass at the desired electrode locations. Metal anchors were then inserted into these holes, their bases were expanded to press against the surface of the hole, and screws were threaded into the anchors to serve as the electrodes. The resistivity survey setup at NASA 2376A is illustrated in Fig. 5, with the coupling method described above illustrated in the inset. The data cables were strategically positioned without crossing sequential connections that could cause errors in the IP data. This was unavoidable at NASA 2715A, and as such the IP data from this site was discarded. The setup described above is the baseline survey. Other survey parameters were tested at both sites in an effort to better constrain what parameters provide an optimal survey design for EDZ/HDZ detection. These parameters are discussed individually in Section 3.3. 3.2. Data processing The data were inverted using RES2DINV (Loke, 2001). The inversion method selected is a cell-based method which minimizes the L1 norm of the error vector; this effectively minimizes the normalized absolute differences between synthetic (model) data and measured data (hereafter referred to as the “absolute error” of the data residuals) (Ellis et al., 1993; Ellis and Oldenburg, 1994). The irregular topography of the excavation surface was extracted from the LiDAR data and input into RES2DINV for incorporation into the inversion model. Six iterations were used for the resistivity inversion. Performing iterations beyond this was found to have a minimal effect on the absolute error of the data residuals associated with the resistivity model, which was found to typically be around 5%. Qualitatively speaking, the results obtained for the sides and the lower portion of the model can be considered the most uncertain. Because the features of interest are very shallow relative to the length of
Fig. 5. Resistivity line setup at NASA 2376A site with anchor and screw electrode combination and installed electrodes shown in the inset image.
213
the line (b30 cm vs. 600 cm) and there are very few lateral variations observed, none of the results were considered to be adversely affected by a loss in signal sensitivity. In other cases where significant changes in resistivity may be associated with deeper features, it is appropriate to evaluate the sensitivity of the result as a function of depth (e.g. Oldenburg and Li, 1999; Caterina et al., 2014). 3.3. Results The baseline inversion results for resistivity and chargeability at NASA 2376A are shown in Fig. 6. The main feature that appears in the resistivity model is a very low resistivity layer (~ 200 Ω•m to 1,500 Ω•m) near surface, which extends to depths of approximately 5 cm to 8 cm beneath the surface. This result is consistent with the presence of an HDZ with fluid-carrying macro-fractures near the excavation wall. Many of the deeper changes in resistivity are caused by edge effects in the inversion, and for the purposes of delineating the HDZ/ EDZ, this zone can generally be considered as intact rock. The exception is the low resistivity area around 320 cm along the line (near the center) which has been interpreted as larger HDZ (see Fig. 6b). In observing the results, the largest gradient in the logarithm of resistivity was used to interpret the limits of the damage zones (Nguyen et al., 2005). The chargeability results for NASA 2376A show a similar trend to the resistivity results — a low chargeability layer exists to a depth between 15 cm and 20 cm beneath the surface, and is underlain by a higher chargeability material. As stated earlier, the correlation between low resistivities and low chargeabilities was hypothesized based on the presence of water-bearing fractures in this zone, although a clear discrepancy between the size of these zones requires further explanation. The chargeability response may be more sensitive to the opening of grain-scale isolated cracks, whereas the resistivity response requires larger, interconnected fractures and/or fracture systems to generate lower values. As such, the chargeability results are considered to represent the EDZ rather than the HDZ. Given the dense survey arrangement, however, there is also a possibility that the results obtained are an artifact of an inductive coupling effect. The resistivity model generated for NASA 2715A (Fig. 7a) shows similar features to the model from the NASA 2376A (Fig. 6a), with a low resistivity layer extending to 5 cm to 12 cm below the surface. Note the decreased resistivity values in this zone relative to the low resistivity zone at NASA 2376A; this is consistent with the observed difference in the saline water inflow rate. Although the difference in HDZ depth is minimal, a difference in the water content in the HDZ explains the lower resistivity values. The implication of this finding is that, through careful calibration, the absolute values of model resistivities could be used to obtain formation factors, which could potentially be used as a relative measure of HDZ permeability in areas where saline water is present. One more feature to note is the low resistivity zone below the near surface high resistivity zone around 240 cm along the line (see Fig. 7a). This zone corresponds with a large, joint-bound block which was deemed to be loose during the data collection. In this case, the resistivity model shows the presence of an open fracture or fracture system ~15 cm to 25 cm below surface. 3.3.1. Array type influence To help provide guidelines for future surveys, the multi-gradient array was compared to a dipole–dipole array. The multi-gradient and dipole–dipole inversion results for NASA 2715A (Fig. 7a and b, respectively) show similar features for the entire extent of the profile. The key difference is that the dipole–dipole model shows greater contrasts in resistivity between high and low resistivity zones, but for the purposes of interpretation, there is negligible value added. Given that the dipole–dipole model has a higher degree of error associated with it than the multi-gradient array, and that the dipole–dipole data collection was more time consuming, the authors have concluded that the multi-
214
G. Walton et al. / Engineering Geology 196 (2015) 210–221
Fig. 6. Multi-gradient resistivity (a) and induced polarization inversion (c) results at NASA 2376A (note that the error associated with the induced polarization result is presented in chargeability units, not as a percentage); (b) shows the interpreted HDZ/EDZ boundary on a zoomed section (area indicated in (a)) of the resistivity results.
gradient array is more appropriate for collecting 2D resistivity data for damage detection. 3.3.2. Electrode spacing influence It is of practical importance to understand how survey resolution deteriorates with increasing electrode spacing, since tighter electrode spacings require more data to be collected for a given line length. The results obtained using a multi-gradient array with 6 cm and 12 cm electrode spacings at NASA 2376A are illustrated in Fig. 8a and b, respectively. It appears that the transition from the lower model resistivities to higher model resistivities (transition at approximately 7,000 Ω•m)
occurs at depths ranging from ~6 cm to ~15 cm in the lower resolution case (as opposed to ~5 cm to ~8 cm in the high resolution case). In this case, it appears that the effective smoothing of the model induced by using a larger electrode spacing has resulted in a deeper mapped HDZ/ EDZ boundary. There is a risk, however, that if the electrode spacing is made too large relative to the size of the HDZ that the smaller lower resistivity zone may be missed in the inversion. This is particularly likely to occur where there is a relatively small contrast between the HDZ resistivity and the intact rock resistivity. At NASA 2715A, data were collected using 6 cm and 12 cm spacing, as well as at 3 cm spacing for the middle portion of the line. The
Fig. 7. Comparison of multi-gradient (a) and dipole–dipole (b) resistivity inversion results for NASA 2715A.
G. Walton et al. / Engineering Geology 196 (2015) 210–221
215
Fig. 8. Multi-gradient resistivity inversion results at NASA 2376A using 6 cm electrode spacing (a) and 12 cm electrode spacing (b).
difference in the apparent HDZ as detected by the 3 cm spacing and 6 cm spacing surveys is insignificant compared to the difference between the 6 cm spacing and 12 cm spacing results, as illustrated in Fig. 9. The data corresponding to the largest electrode spacing resulted in an inverted model with a relatively large HDZ thickness estimation (~10 cm to ~16 cm). Based on these results, it is clear that there is a maximum acceptable electrode spacing for damage detection surveys to obtain accurate results, and that below this threshold there is minimal improvement in
damage zone resolution as electrode spacing is further decreased. The authors suggest that for similar conditions, an electrode spacing on the order of the expected damage zone thickness (based on numerical modelling or field observations) should be used. 4. GPR surveys Ground Penetrating Radar technology operates by emitting an electromagnetic (EM) pulse from an antenna and recording the arrival time
Fig. 9. Multi-gradient resistivity inversion results at NASA 2715A using 6 cm electrode spacing (a) and 12 cm electrode spacing (b). The black boxes indicate the area for which a model was generated based on limited data acquired using 3 cm electrode spacing. Inversion results for this zone are shown for 3 cm (c), 6 cm (d), and 12 cm (e) electrode spacing (these three models share a common scale bar).
216
G. Walton et al. / Engineering Geology 196 (2015) 210–221
of the reflected pulses that return to a receiver antenna. The recorded travel-times of reflections can then be converted into estimated reflector depths. GPR pulses are reflected whenever they encounter a discontinuity in wave velocity. In general, differences in conductivity, magnetic susceptibility, and dielectric permittivity can cause such a discontinuity. For most geomaterials with low magnetic susceptibility, the relative dielectric permittivity (εr) is the key parameter that controls the wave velocity and therefore the reflectivity of material interfaces (Saksa et al., 2005). In terms of identifying individual HDZ fractures, previous researchers have published successful results mapping subsurface discontinuities using GPR. The specific applications of the technique include the investigation of quarries (Grandjean and Gourry, 1996; Grasmueck, 1996), exposed cliffs (Dussauge-Pessier et al., 2003), medieval masonry (Arias et al., 2007), pavement damage (Saarenketo and Scullion, 2000; Bendetto and Pensa, 2007), and excavation damage (Silvast and Wiljanen, 2008). The reflectivity of subsurface fractures is caused by the notable permittivity contrast between a rock matrix and either air or water in an open fracture. A key limitation of the GPR surveys with respect to identifying HDZ fractures at this site is that fractures infilled by mineralization are likely to generate reflections similar to those generated by open fractures. It is also difficult to delineate fractures in the damaged zone due to the high potential for wave scattering and irregular or out-of-plane reflectors. Frequency dependent conductive losses and wave scattering are both key controls on EM wave attenuation. The relatively high electrical conductivity of pore water in damaged zones and the increased potential for wave scattering losses both suggest that the EDZ should be detectable as an area of high wave scattering and energy loss, particularly in higher frequency signal components. Silvast and Wiljanen (2008) have demonstrated the practical use of a proprietary dispersion algorithm to delineate the EDZ at a Finnish spent nuclear fuel repository test site, ONKALO. Based on their results, the outer edge of the near surface scattering zone observed in high frequency data may be correlated with the boundary between the EDZ and intact rockmass. 4.1. Survey setup The GPR surveys at Äspö HRL were conducted using a custom-built step-frequency GPR system developed at the Norwegian Geotechnical Institute. Two antennae were used with the system in a monostatic configuration. The lower frequency antenna collected data using frequencies between 600 MHz and 2,600 MHz, with a peak in signal power around 1,500 MHz. The higher frequency antenna collected data using frequencies between 800 MHz and 3,000 MHz, with a peak in signal power around 2,500 MHz. Sampling frequencies of approximately 10.23 GHz and 11.26 GHz were used for the low and high frequency antennae, respectively. The GPR system recorded data in a continuous measurement mode (one trace per second), so to appropriately locate each trace, the trace numbers corresponding to chalk marks made at 3 cm intervals were recorded by the operator. An average trace spacing of approximately 1.5 cm was obtained for each line completed. Data were recorded a minimum of two times along each line using each antenna for two perpendicular antenna orientations. Due to the somewhat irregular surface topography at both sites, maintaining consistent coupling during data collection was challenging. For this reason, the use of air-coupled systems which do not require direct contact with the rock surface is recommended for future studies.
antenna remained static. The true trace locations were recorded every 3 cm; within each 3 cm bin, traces were stacked to improve the signal to noise ratio. Multiple data sets collected at the same site using the same antenna were also stacked. Since a preliminary analysis of the data showed no difference between data collected using different antenna orientations, both orientations were used for the stacking. The number of traces stacked for a given location was variable due to the variable number of traces collected within each 3 cm bin, but an average value of eight stacks is representative for most locations (four data collection instances times two traces per bin per instance). To aid in the interpretation of potential fracture locations, a bandpass filter (500 MHz to 3 GHz) was applied to the lower frequency data followed by an automatic envelope correction (3.6 ns window length). It was found that for the higher frequency data, amplitude scaling primarily amplified noise at later times; since the outer envelope of the zone of high reflectivity may be representative of the EDZ in the high frequency data, the final sections were produced without any amplitude scaling. For the purposes of visualization, data were clipped two standard deviations from the mean amplitude. Time to depth conversions were performed using an assumed dielectric permittivity of 7, which is typical of granites in this region (Saksa et al., 2005; Kantia et al., 2013). Because of the monostatic configuration of the GPR system, it was not possible to collect common midpoint (CMP) data to improve the velocity estimate for the rock. The velocity used, however, appears to be relatively reliable based on previous experience in the area, and the good correlation between interpreted fracture locations and fractures observed in boreholes (see Section 6). 4.3. Results & interpretation The stacked GPR sections produced for NASA 2376A and NASA 2715A are shown in Figs. 10 and 11, respectively. The lower frequency data was found to provide a clearer indication of potential fracture locations at both sites. As can be seen in the data, the interpretation of fractures from the data is relatively difficult due to the presence of noise (likely due to the irregularity of the surveyed surface). A rough interpretation can still be made, however, and dashed lines have been placed on the profiles to delineate what were initially identified as potential fracture locations independent of the results of the other surveys (Figs. 10b and 11b). These are likely not all of the fractures within the rockmass, and, as stated above, many of these fractures may be natural, infilled, low porosity fractures. In GPR data, the extent of the EDZ should correspond to a near surface high reflector density zone, particularly for higher frequency components. This is because the large number of small-scale permittivity discontinuities present in damaged rock causes large amounts of the original signal to scatter back to the antenna. There are some issues associated with attenuation effects when making a direct comparison between the extent of the reflective zone and the EDZ, but studies have shown that it can still serve as a good approximation of the EDZ (Silvast and Wiljanen, 2008; Kantia et al., 2013). If the edge of the reflective zone in the higher frequency data is assumed to correspond exactly to the edge of the EDZ, the edge of the EDZ is interpreted as being at approximately 25 cm depth at NASA 2376A and 30 cm depth at NASA 2715A. These results correlate roughly with the chargeability results at NASA 2376A (EDZ boundary between 15 cm and 25 cm). 5. Borehole data
4.2. Data processing The raw GPR data were imported into MATLAB (MathWorks, 2010). Using the operator's field notes, the data were assigned spatial locations. The mean trace was subtracted from each data set, as approximated by the first 30 traces which were recorded at the start of each line while the
Based on a preliminary analysis of the data performed on-site, borehole locations were selected. Boreholes were drilled to depths of 100 cm at 240 cm, 326 cm, and 468 cm along the line at NASA 2376A and to depths of 80 cm at 120 cm, 243 cm, 297 cm, and 364 cm along the line at NASA 2715A. The exact locations of the core drillings are
G. Walton et al. / Engineering Geology 196 (2015) 210–221
217
Fig. 10. Stacked GPR sections for the NASA 2376A site. Panel (a) shows the filtered and amplitude corrected lower frequency data; panel (b) shows an interpretation of potential fracture locations identified prior to the analysis of the borehole data; panel (c) shows the unfiltered higher frequency data.
illustrated in Figs. 3 and 4. Core was recovered from all holes, and the observed fractures were logged for depth, orientation, and condition. Using a borehole camera, fractures observed in the core were correlated with in-situ fractures to verify the logged fracture conditions, and also to better constrain the three-dimensional orientation of the fractures. Unfortunately, no information could be reliably determined on the saturation condition of the fractures, as the use of drilling fluid in the coring process disturbed the in-situ hydrogeological condition. Images obtained using the borehole camera showing the variety of fracture conditions encountered are presented in Fig. 12. Although all of the fractures depicted (natural, sealed, and open) have the potential
to appear as reflectors in the GPR data, only the more permeable fractures are suspected to have a significant influence on the resistivity data.
6. Integrated interpretation In an effort to validate the hypotheses formed based on the geophysical data alone, the fracture locations and conditions were plotted and overlaid with the geophysical data. The orientations of the fractures are projected onto the survey planes, since many of the fractures were found to have an out-of-plane component to their attitude.
Fig. 11. Stacked GPR sections for the NASA 2715A site. Panel (a) shows the filtered and amplitude corrected lower frequency data; panel (b) shows an interpretation of potential fracture locations identified prior to the analysis of the borehole data; panel(c) shows the unfiltered higher frequency data.
218
G. Walton et al. / Engineering Geology 196 (2015) 210–221
Fig. 12. Different fractures observed using the borehole camera. The upper row illustrates fractures which have been sealed/infilled, the middle row illustrates a transition to increasingly open fractures, and the fractures on the bottom row are open with apertures approaching 1 mm. The resistivities associated with these fractures are expected to decrease from the top left corner to the bottom right corner.
6.1. NASA 2376A The multi-gradient profile and lower frequency GPR stacked section for the NASA 2376A site are shown in Fig. 13 with mapped fractures overlain. In the resistivity data (Fig. 13a and b), the key finding is that each borehole contained at least one open and/or weathered fracture within the upper 10 cm. Borehole A contained three open fractures at depths of 3 cm, 5 cm, and 8 cm, all of which were roughly excavation parallel. These findings suggest that the HDZ extends to a depth of 5 cm to 8 cm, which is consistent with the size of the zones imaged as having low resistivity. Similar open fractures within the upper 5 cm were seen in boreholes B and C. In the GPR data (Fig. 13c) it is difficult to correlate individual reflectors to individual fractures because of the high number of reflected arrivals in the damaged zone. The fractures mapped deeper in all three
boreholes correlate well with nearby reflectors, however, which is consistent with the use of an appropriate dielectric permittivity. Also, a comparison of the individual stacked traces at the locations of each borehole and the mapped fractures in the boreholes shows that there is good correlation between reflector peaks and mapped fractures (Fig. 13d). The borehole observations (fracture location, orientation, and condition) provide no information with respect to the micro-scale damage which was potentially detected by the GPR and chargeability surveys. Based on the consistency of the results achieved using these distinct methods, however, it can be concluded that the EDZ at NASA 2376A is variable in extent, and exists to depths within the range of 15 cm to 35 cm along the profile. This is also consistent with the expected value based on the work of Emsley et al. (1997).
G. Walton et al. / Engineering Geology 196 (2015) 210–221
219
Fig. 13. (a) Inverted resistivity model with interpreted HDZ/EDZ limits and (b) zoomed views of the borehole locations with observed fractures indicated; (c) stacked GPR data (lower frequency) overlain with known fracture locations from boreholes at NASA 2376A and interpreted reflectors; (d) individual GPR traces at borehole locations with mapped fractures illustrated as vertical lines.
6.2. NASA 2715A The multi-gradient profile and the lower frequency GPR stacked section for the NASA 2715A site are illustrated in Fig. 14 with mapped fractures overlain. In the resistivity data (Fig. 14a and b), the presence of an HDZ in the upper 10 cm is supported by the presence of near surface fractures in boreholes D and F. In borehole G, the slightly weathered fractures may be related to the lower resistivity features in the model that lie just above and just below their locations, or they may have sufficiently low water content that they had little influence on the data. In this case, the lower resistivity zones would be associated with smaller scale damage which was not visually detectable in the core. The most interesting results are from borehole E. These results show that there is indeed an open fracture which separates the relatively intact but loose block at this location from the intact rockmass behind. There is also a second open fracture at the base of the relatively thick low resistivity zone which exists behind this block. It is possible that this low resistivity zone corresponds to a smaller-scale damage induced preferentially between the two open fractures. Again, in the GPR data (Fig. 14c) it is difficult to correlate individual fractures with any degree of certainty, but there are some notable
correlations between fracture and reflector locations. In particular, the borehole and GPR data are consistent for fractures lower in holes E & F and near surface in hole D. Again, the individual trace reflector peaks along the borehole alignments tend to agree with the mapped fracture locations (Fig. 14d). Based on the GPR survey, it can be concluded that the EDZ at NASA 2715A extends to variable depths between 15 cm and 35 cm below the excavation surface. 7. Conclusions & recommendations The analysis of the geophysical research conducted at the Äspö HRL site demonstrates that the RES/IP and GPR methods tested have value in detecting the HDZ/EDZ in a hard rock environment. In particular, 2D resistivity profiling proved most useful in delineating areas where macroscopic open fractures were observed (HDZ), whereas the GPR proved useful in identifying individual fractures and in estimating the EDZ dimensions. When performing 2D resistivity profiling, the multi-gradient array was found to be suitable, with electrode spacings on the order of the expected HDZ thickness being optimal. The lower frequency GPR antenna (centered around ~ 1,500 MHz) was found to be most useful in
220
G. Walton et al. / Engineering Geology 196 (2015) 210–221
Fig. 14. (a) Inverted resistivity model with interpreted HDZ/EDZ limits and (b) zoomed views of the borehole locations with observed fractures indicated; (c) stacked GPR data (lower frequency) overlain with known fracture locations from boreholes at NASA 2376A and interpreted reflectors; (d) individual GPR traces at borehole locations with mapped fractures illustrated as vertical lines.
identifying individual fractures, whereas the higher frequency antenna (centered around ~ 2,500 MHz) provided EDZ dimension estimates more consistent with those provided by other methods (as shown in Figs. 10 and 11). With respect to underground nuclear waste disposal, the most significant impact of the EDZ and HDZ is their potential to act as dominant groundwater flow pathways in the long term. To design effective measures such as cut-off seals to mitigate the effects of these zones following construction requires a knowledge of their geometries. As has been demonstrated in this study, RES/IP and GPR surveying methods both have the potential to provide valuable information on the HDZ/EDZ without the risk of inducing further damage in the rockmass. In the future, it may be of interest to use a combination of approaches as demonstrated in this study for repeated surveys to characterize changes in the HDZ/EDZ over time (as demonstrated using the resistivity method by Gibert et al., 2006 and Lesparre et al., 2013). Such surveys conducted over a large time-scale could provide information on the potential for
HDZ/EDZ self-sealing with time, which could have significant implications for deep geological repository performance. Acknowledgments The authors would like to thank SKB for providing access to the test site for this study. Tomas Lehtimaki played a critical role in coordinating the data collection activities. This work would not have been possible without funding provided by the Nuclear Waste Management Organization of Canada (NWMO) (NWMO-EDZ-2), the Natural Sciences and Engineering Research Council of Canada (NSERC) (CGS-D), and the Norwegian Research Council. References Alheid, H.-J., Schuster, K., Kruchsitz, S., Yaramanci, U., 2002. Seismic and geolectic measurements for the characterization of the excavation damage zone around tunnels in Opalinus Clay. Z. Angew. Geol. 48 (2), 48–55.
G. Walton et al. / Engineering Geology 196 (2015) 210–221 Archie, G.E., 1942. The electrical resistivity log as an aid in determining some reservoir characteristics. Trans. Am. Inst. Min. Metall. Eng. 146, 54–62. http://dx.doi.org/10. 2118/942054-G. Arias, P., Armesto, J., Di-Capua, D., Gonzalez-Drigo, R., Lorenzo, H., Perez-Gracia, V., 2007. Digital photogrammetry, GPR and computational analysis of structural damages in a mediaeval bridge. Eng. Fail. Anal. 14, 1444–1457. http://dx.doi.org/10.1016/j. engfailanal.2007.02.001. Ask, D., 2003. Evaluation of measurement-related uncertainties in the analysis of overcoring rock stress data from Aspo HRL, Sweden: a case study. Int. J. Rock Mech. Min. Sci. 40, 1173–1187. http://dx.doi.org/10.1016/s1365-1609(03)00114-x. Backblom, G., Martin, C.D., 1999. Recent experiments in hard rocks to study the excavation response: implications for the performance of a nuclear waste geological repository. Tunn. Undergr. Space Technol. 14 (3), 377–394. http://dx.doi.org/10.1016/ s0886-7798(99)00053-x. Bendetto, A., Pensa, S., 2007. Indirect diagnosis of pavement structural damages using surface GPR reflection techniques. J. Appl. Geophys. 62, 107–123. http://dx.doi.org/10. 1016/j.jappgeo.2006.09.001. Bossart, P., Meier, P.M., Moeri, A., Trick, T., Mayor, J.C., 2002. Geological and hydraulic characterization of the excavation disturbed zone in the Opalinus Clay of the Mont Terri Rock Laboratory. Eng. Geol. 66, 19–38. http://dx.doi.org/10.1016/s00137952(01)00140-5. Cabrera, J., Volant, P., Baker, C., Pettitt, W., Young, R.P., 1999. Structural and geophysical investigations of the EDZ (Excavation Disturbed Zone) in indurated argillaceous media: the tunnel and the galleries of the IPSN Tournemire site (France). 37th US Symposium on Rock Mechanics vol. 2. Taylor & Francis, US. Carlson, S.R., Young, R.P., 1993. Acoustic emission and ultrasonic velocity study of excavation-induced microcrack damage at the Underground Research Laboratory. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 30, 901–907. http://dx.doi.org/10.1016/ 0148-9062(93)90042-c. Caterina, D., Hermans, T., Nguyen, F., 2014. Case studies of incorporation of prior information in electrical resistivity tomography: comparison of different approaches. Near Surf. Geophys. 12 (4), 451–465. Cosma, C., Olsson, O., Keskinen, J., Heikkinen, P., 2001. Seismic characterization of fracturing at the Äspö Hard Rock Laboratory, from the kilometer scale to the meter scale. Int. J. Rock Mech. Min. Sci. 38, 859–865. http://dx.doi.org/10.1016/s13651609(01)00051-x. Dahlin, T., Zhou, B., 2006. Multi-gradient array measurements for multichannel 2D resistivity imaging. Near Surf. Geophys. 4 (2), 113–123. http://dx.doi.org/10.3997/18730604.2005037. Diederichs, M.S., 2007. The 2003 Canadian Geotechnical Colloquium: mechanistic interpretation and practical application of damage and spalling prediction criteria for deep tunnelling. Can. Geotech. J. 44, 1082–1116. http://dx.doi.org/10.1139/t07-033. Diederichs, M., Lam, T., Jensen, M., Damjanac, B., Martin, D., 2013. Ultra-long term geomechanics design for a deep geological repository in sedimentary rock. Proc. 47th US Rock Mechanics/Geomechanics Symposium, San Francisco, California. Dussauge-Pessier, C., Wathelet, M., Jongmans, D., Hantz, D., Couturier, B., Sintes, M., 2003. Investigation of a fractured limestone cliff (Chartreuse Massif, France) using seismic tomography and ground-penetrating radar. Near Surf. Geophys. 1, 161–170. http:// dx.doi.org/10.3997/1873-0604.2003007. Ellis, R.G., Oldenburg, D.W., 1994. Applied geophysical inversion. Geophys. J. Int. 116, 5–11. Ellis, R.G., Farquharson, C.G., Oldenburg, D.W., 1993. Approximate inverse mapping inversion of the COPROD2 data. J. Geomagn. Geoelectr. 45, 1001–1002. Emsley, S., Olsson, O., Stenberg, L., Alheid, H.J., Falls, S., 1997. ZEDEX — a study of damage and disturbance from tunnel excavation by blasting and tunnel boring. Technical Report 97-30. Swedish Nuclear Fuel and Waste Management Co., p. 198. Gibert, D., Nicollin, F., Kergosien, B., Bossart, P., Nussbaum, C., Grislin-Mouezy, A., Conil, F., Hoteit, N., 2006. Electrical tomography monitoring of the excavation damaged zone of the Gallery 04 in the Mont Terri rock laboratory: field experiments, modelling, and relationship with structural geology. Appl. Clay Sci. 33, 21–34. http://dx.doi. org/10.1016/j.clay.2006.03.008. Grandjean, G., Gourry, J.C., 1996. GPR data processing for 3D fracture mapping in a marble quarry (Thassos, Greece). J. Appl. Geophys. 36, 19–30. http://dx.doi.org/10.1016/ s0926-9851(96)00029-8.
221
Grasmueck, M., 1996. 3-D ground-penetrating radar applied to fracture imaging in gneiss. Geophysics 61 (4), 1050–1064. http://dx.doi.org/10.1190/1.1444026. Jacobsson, L., 2006. Oskarshamn site investigation — borehole KLX10: uniaxial compression test of intact rock. SKB Technical Report: P-06-37. Johansson, H., Siitari-Kauppi, M., Skalberg, M., Tullborg, E.L., 1998. Diffusion pathways in crystalline rock — examples from Aspo-diorite and fine-grained granite. J. Contam. Hydrol. 35, 41–53. Kantia, P., Heikkenen, E., Cosma, C., Enescu, N., Lehtimaki, T., Kemppainen, K., 2013. Nondestructive testing of rock mass deformation in spent nuclear fuel disposal research facilities. Proc. 47th US Rock Mechanics/Geomechanics Symposium, San Francisco, California. Kearey, P., Brooks, M., Hill, I., 2002. An Introduction to Geophysical Exploration. 3rd ed. Blackwell Science, Cornwall, Great Britain. Kirsch, R., 2006. Grounwater Geophysics — A Tool for Hydrogeology. Springer. Kruschwitz, S., Yaramanci, U., 2004. Detection and characterization of the disturbed rock zone in claystone with the complex resistivity method. J. Appl. Geophys. 57, 63–79. http://dx.doi.org/10.1016/j.jappgeo.2004.09.003. Lesparre, N., Gibert, D., Nicollin, F., Nussbaum, C., Adler, A., 2013. Monitoring the excavation damaged zone by three-dimensional reconstruction of electrical resistivity. Geophys. J. Int. 195 (2), 972–984. http://dx.doi.org/10.1093/gji/ggt282. Loke, M.H., 2001. Electrical imaging surveys for environmental and engineering studies. A practical guide to 2-D and 3-D surveys. RES2DINV Manual. IRIS Instruments (www. iris-instruments.com). Loke, M.H., Chambers, J.E., Rucker, D.F., Kuras, O., Wilkinson, P.B., 2013. Recent developments in the direct-current geolectrical imaging method. J. Appl. Geophys. 95, 135–156. http://dx.doi.org/10.1016/j.jappgeo.2013.02.017. Malmgren, L., Saiang, D., Toyra, J., Bodare, A., 2007. The excavation disturbed zone (EDZ) at Kiirunavaara mine, Sweden — by seismic measurements. J. Appl. Geophys. 61, 1–15. http://dx.doi.org/10.1016/j.jappgeo.2006.04.004. Martin, C.D., Kaiser, P.K., McCreath, D.R., 1999. Hoek–Brown parameters for predicting the depth of brittle failure around tunnels. Can. Geotech. J. 36, 136–151. http://dx.doi.org/ 10.1139/cgj-36-1-136. MathWorks, 2010. MATLAB (Version 7.10.0.499) — Software. Nguyen, F., Garambois, S., Jongmans, D., Pirard, E., Loke, M.H., 2005. Image processing of 2D resistivity data for imaging faults. J. Appl. Geophys. 57, 260–277. Oldenburg, D.W., Li, Y., 1999. Estimating depth of investigation in DC resistivity and IP surveys. Geophysics 64 (2), 403–416. Pusch, R., 1989. Alteration of the hydraulic conductivity of rock by tunnel excavation. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 26 (1), 79–83. Saarenketo, T., Scullion, T., 2000. Road evaluation with ground penetrating radar. J. Appl. Geophys. 43, 119–138. http://dx.doi.org/10.1016/s0926-9851(99)00052-x. Saksa, P., Heikkinen, E., Lehtimäki, T., 2005. Geophysical radar method for safeguards application at Olkiluoto spent fuel disposal site in Finland. STUK Technical Report. STUK-YTO-TR 213. Scott, J.H., Lee, F.T., Carroll, R.B., Robinson, C.S., 1968. The relationship of geophysical measurements to engineering and construction parameters in the straight creek tunnel pilot bore, Colorado. Int. J. Rock Mech. Min. Sci. 5, 1–30. http://dx.doi.org/10.1016/ 0148-9062(68)90020-x. Sharma, P.V., 1997. Environmental and Engineering Geophysics. Cambridge University Press. Silvast, M., Wiljanen, B., 2008. ONKALO EDZ-measurements using Ground Penetrating Radar (GPR) method. Posiva Working Report 2008-58. Souley, M., Homand, F., Pepa, S., Hoxha, D., 2001. Damage-induced permeability changes in granite: a case example at the URL in Canada. Int. J. Rock Mech. Min. Sci. 38, 297–310. http://dx.doi.org/10.1016/s1365-1609(01)00002-8. Suzuki, K., Nakata, E., Minami, M., Hibino, E., Tani, T., Sakakibara, J., Yamada, N., 2004. Estimation of the zone of excavation disturbance around tunnels, using resistivity and acoustic tomography. Explor. Geophys. 35, 62–69. http://dx.doi.org/10.1071/ eg04062. Tsang, C.F., Bernier, F., Davies, C., 2005. Geohydromechanical processes in the Excavation Damaged Zone in crystalline rock, rock salt, and indurated and plastic clays—in the context of radioactive waste disposal. Int. J. Rock Mech. Min. Sci. 42 (1), 109–125. http://dx.doi.org/10.1016/j.ijrmms.2004.08.003.