Compilation and Consolidation of Field and Laboratory Data for Hydrogeological Properties

Transcription

1 Technical Report Title: Document ID: Author: Compilation and Consolidation of Field and Laboratory Data for Hydrogeological Properties TR Robert Walsh Revision: 1 Date: April 1, 2011 DGR Site Characterization Document Geofirma Engineering Project

2 Geofirma Engineering DGR Site Characterization Document Title: Document ID: Compilation and Consolidation of Field and Laboratory Data for Hydrogeological Properties TR Revision Number: 1 Date: April 1, 2011 Author: Technical Review: QA Review: Robert Walsh John Avis, Kenneth Raven; Dylan Luhowy, NWMO John Avis Approved by: Kenneth Raven Document Revision History Revision Effective Date Description of Changes 0 April 21, 2009 Initial release 1 April 1, 2011 Includes new data from boreholes DGR-3, DGR-4, DGR-5 and DGR-6; added sections on rock density and effective diffusion coefficients; revisions to compressibility and storage data to reflect uncertainty. April 1, 2011 ii

3 TABLE OF CONTENTS 1 INTRODUCTION HYDRAULIC CONDUCTIVITY Shallow System Hydraulic Conductivities Straddle-Packer Testing of DGR-1/2, DGR-3, DGR-4, DGR-5, and DGR Formation Hydraulic Conductivity Estimates Hydraulic Conductivity Anisotropy POROSITY ELASTIC MODULI Calculating the Low Compressibility Estimate Calculating the High Compressibility Estimate SPECIFIC STORAGE ROCK DENSITY EFFECTIVE DIFFUSION COEFFICIENTS DATA USE AND CONCLUSIONS REFERENCES LIST OF FIGURES Figure 1 Straddle-Packer Test Results and Formation Hydraulic Conductivities. Formation Tops are based on DGR-1 and DGR-2 Reference Stratigraphy Figure 2 Anisotropy in Core Permeability from Gas Permeability Testing of DGR-2 Cores and Brine Permeability Testing of DGR-3 and DGR-4 Cores Figure 3 Measured Liquid and Total Porosity Values and Formation Average Values. The Shaded Regions of the Plot Show Which Values Were Used to Generate Table Figure 4 Calculated Low and High Compressibility Values and Formation Average Values Figure 5 Specific Storage Data, Formation Average Ranges from Geomechanical Testing, and Fitted Specific Storage from Straddle-Packer Testing. Note that the Point Values are Plotted for Illustrative Purposes but were not used for Calculating the Averages Figure 6 Grain and Bulk Dry Density, Measured Data and Formation Averages Figure 7 Measured Effective Iodide Diffusion Coefficient and Formation Average Values, Normal to Bedding Planes April 1, 2011 iii

4 LIST OF TABLES Table 1 Summary of Hydraulic Conductivities for Lucas, Amherstburg, Bois Blanc and Bass Islands Formations... 2 Table 2 Horizontal Hydraulic Conductivity Values of Formations... 4 Table 3 Hydraulic Conductivity Anisotropy of Formations... 7 Table 4 Average Porosity Values of Formations Table 5 Low and High Estimates of Average Rock Compressibility of Formations Table 6 Typical Poroelastic Parameters for Various Rock Types. Source: Wang (2000) Table 7 Formation Averages of Fluid Density and Specific Storage Table 8 Average Dry Bulk Density and Grain Density of Formations Table 9 Formation Averages of Effective Diffusion Coefficient LIST OF APPENDICES APPENDIX A Detailed Mechanical Testing Results April 1, 2011 iv

5 1 Introduction Geofirma Engineering Ltd. (formerly Intera Engineering Ltd.) has been contracted by the Nuclear Waste Management Organization (NWMO), on behalf of Ontario Power Generation, to implement the Geoscientific Site Characterization Plan (GSCP) for the Bruce site located near Tiverton, Ontario. The purpose of this site characterization work is to assess the suitability of the Bruce site to construct a Deep Geologic Repository (DGR) to store low-level and intermediate-level radioactive waste. The GSCP is described by Intera Engineering Ltd. (2006, 2008). This Technical Report presents a compilation and interpretation of the available site-specific hydrogeological data from boreholes DGR-1, DGR-2, DGR-3, DGR-4, DGR-5, and DGR-6, as well as from pre-existing data. This summary may be useful for future Safety Assessment modelling work at the DGR site and preparation of the Descriptive Geosphere Site Model Report for the Bruce nuclear site. Data on the following parameters are presented: Hydraulic conductivity Porosity Elastic moduli (including compressibility) Specific storage coefficients Rock density (bulk dry density and grain density) Effective diffusion coefficients As this Technical Report is intended for modelling work, parameters will be needed for all formations, even if measurements are not currently available. Where a given measurement is not available for a particular formation, expert opinion and/or extrapolation from nearby formations with similar properties is used to replace omissions in the data. Work described in this Technical Report was completed following the general requirements of the DGR Project Quality Plan (Intera Engineering Ltd., 2009a). 2 Hydraulic Conductivity Site specific hydraulic conductivity data at the Bruce site consist of the results of straddle-packer hydraulic testing performed in isolated intervals of the DGR-1 borehole in the summer and fall of 2007, straddle-packer testing in DGR-2, DGR-3, and DGR-4 in the spring and summer of 2009, and tests in DGR-5 and DGR-6 in the spring and summer of 2010 (Geofirma Engineering Ltd., 2011a, TR-08-32). Lab scale petrophysical tests on core samples were also performed (Intera Engineering Ltd., 2010a, TR-07-18; Intera Engineering Ltd., 2010b, TR-08-28). Additionally, data on drilling fluid losses were used to estimate hydraulic conductivities in portions of the Devonian sequence. 2.1 Shallow System Hydraulic Conductivities Data on formation hydraulic conductivity for the shallow bedrock (Lucas, Amherstburg, Bois Blanc and Bass Islands Formations) are available from summaries of geotechnical bedrock investigations and Bruce A and B cooling water intake tunnelling experience (Golder Associates Ltd, 2003), straddle packer testing of US-1 to US-7 (106 tests - Lukajic, 1988), slug testing of Westbay test intervals in US-5 and US-6 (14 tests - Golder April 1,

6 Associates Ltd, 2003) and from drilling fluid loss observations made during drilling of DGR-1 and DGR-2 (Intera Engineering Ltd., 2010c, TR-07-06) and US-8 (Intera Engineering Ltd., 2009b TR-07-19). Packer test flowrates reported by Lukajic (1988) and injection pressures and drilling fluid loss rates and heads were converted to equivalent hydraulic conductivities assuming conditions of horizontal, confined, steady radial flow using the Thiem (1906) equation. Steady injection flow rate and head data presented by Lukajic in units of Igpm and psi were converted to SI units of m 3 /s and m for use in the Thiem equation. Packer test data of Lukajic were converted to hydraulic conductivity values using the measured test interval lengths (typically 3 and 6 m) and the assumption of radius of test influence equal to 10 m. Measurements of drilling fluid loss in permeable horizons reported in m 3 /core run were converted to m 3 /s based on reported time to core the horizon. Injection heads were calculated based on the elevation of the top of the casing overflow at ground surface and the estimated formation hydraulic head based on head measurements recorded in nearby US-series monitoring wells. For DGR-1 the heads were estimated based on formation pressures measured in US-3. Hydraulic conductivities were calculated from drilling fluid losses using the estimated test interval length determined from core observations and the assumption of radius of test influence of 50 m. Table 1 summarizes best estimates of hydraulic conductivities for the Lucas, Amherstburg, Bois Blanc and Bass Islands Formations, the basis/rationale for the estimate and the data source for their inclusion in this report. Of note are the very permeable sections (1x10-4 m/s) of the upper 20 m of Bass Islands Formation that created significant drilling fluid losses during drilling of DGR-1, DGR-2 and US-8. Table 1 Summary of Hydraulic Conductivities for Lucas, Amherstburg, Bois Blanc and Bass Islands Formations Formation Lucas Lucas Amherstburg Amherstburg Bois Blanc (to 100 m) Combined Amherstburg and Bois Blanc Bass Islands (upper 20 m) Hydraulic Conductivity (m/s) Basis/Rationale Source 6x10-9 to 3x10-5 Range, geometric mean from packer Analysis of Lukajic 2x10-6 tests in US boreholes (1988) Data 4x10-9 to 2x10-4 Range, geometric mean from Bruce Golder Associates 5x10-7 A site investigations Ltd. (2003) 8x10-10 to 8x10-5 Range, geometric mean from packer Analysis of Lukajic 8x10-8 tests in US boreholes (1988) Data 1x10-8 to 2x10-5 Range, geometric mean from Bruce Golder Associates 2x10-7 A site investigations Ltd. (2003) 6x10-10 to 1x10-5 Range, geometric mean from packer Analysis of Lukajic 1x10-7 tests in US boreholes (1988) Data Range, geometric mean from tunnel 1x10-6 to 1x10-4 dewatering experience and slug Golder Associates 1x10-5 testing of US-5 & US-6 casings Ltd (2003) 1x10-5 to 3x10-4 Analysis of drilling fluid losses in TR-07-06, 1x10-4 US-8 and DGR-1 TR Golder Associates Ltd (2003) Bass Islands 1x10-5 value Estimated average representative Data from Lukajic (1988) and Golder Associates Ltd, (2003) were not collected as part of the DGR site characterisation plan, and therefore may not meet the general requirements of the DGR Project Quality Plan (Intera Engineering Ltd., 2009a). April 1,

7 2.2 Straddle-Packer Testing of DGR-1/2, DGR-3, DGR-4, DGR-5, and DGR-6 Hydraulic testing of intervals isolated by straddle packers was performed in all exploration boreholes (i.e., DGR-2, DGR-3, DGR-4, DGR-5 and DGR-6). A total of 89 straddle packer tests were performed: 3 in DGR-1, 15 in DGR-2, 23 in DGR-3, 25 in DGR-4, 11 in DGR-5, and 12 in DGR-6. Representative formation values for hydraulic conductivity based on hydraulic testing were calculated in Geofirma Engineering Ltd. (2011a, TR-08-32), where a detailed description of the test results and analysis can be found. Typically best estimates of formation hydraulic conductivity from DGR borehole straddle-packer testing are geometric means of available test data. 2.3 Formation Hydraulic Conductivity Estimates Table 2 shows the best estimate formation horizontal hydraulic conductivity values. Figure 1 shows the measured hydraulic conductivities (straddle-packer results) and the formation average values. Where no hydraulic conductivity data were available for a given unit, the permeability of an adjacent unit of similar rock type was used. The choice of which units to use for estimates was informed by fluid electrical conductivity logs from Intera Engineering Ltd. (2009c, TR-07-14). 2.4 Hydraulic Conductivity Anisotropy Currently, the only measurements of vertical hydraulic conductivity (K v ) are from the lab-scale samples on cores (Intera Engineering Ltd., 2010a, TR-07-18; Intera Engineering Ltd., 2010b TR-08-28). In the Ordovician sequences the average lab scale hydraulic conductivity values are typically higher than the straddle-packer tests from the same formations. This is particularly the case in the shale dominated Blue Mountain, Georgian Bay, and Queenston formations. This discrepancy may be explained by core damage during coring, extraction, or transport and would have had a greater impact in the softer shale dominated formations. In the more durable limestone, damage to the core was probably of a lesser magnitude. Overall, the reliability of the lab-scale permeability data are somewhat compromised. Nevertheless, the lab scale horizontal and vertical permeabilities plotted in Figure 2 do show vertical hydraulic conductivities (K v ) between 1 and 10 times less than the horizontal conductivity (K h ). The field-scale straddle-packer tests measured the horizontal transmissivity of the system. On a macro scale, given the horizontal layering in these sedimentary formations, K v is expected to be lower than the K h. The straddle-packer results represent the arithmetic average response of the formation within the packer interval. The arithmetic average will tend to emphasize the response of any high conductivity beds in the packer interval. Ignoring the potential impact of a sparse fracture network, the effective vertical conductivity of a horizontally bedded formation will be much closer to the harmonic average, which will emphasize any lower conductivity beds or sub-layers. On a micro scale, hydraulic conductivity anisotropy of sedimentary rock due to the orientation of mineral grains will also contribute to a lower vertical hydraulic conductivity, especially in shale or mudstone. Thus, an estimated horizontal to vertical hydraulic conductivity anisotropy of 10 to 1 for most units is considered conservative. Table 3 shows the best estimate of hydraulic conductivity anisotropy for the formations (Intera Engineering Ltd., 2011, DGR-TR ). April 1,

8 Table 2 Horizontal Hydraulic Conductivity Values of Formations Formation Depth of Top in DGR-1/2 K h Comments (mbgs) (m/s) Lucas E-06 Geometric mean of Golder and Lukajic mean Amherstburg (top 20 m) E-06 Geometric mean of Golder and Lukajic mean Amherstburg (lower E-07 m) Geometric mean of Golder and Lukajic mean Bois Blanc E-07 Bass Islands (upper 20m) E-04 Bass Islands (lower 25 m) E-05 Salina G Unit E-11 No testing available, based on core observation Salina F Unit E-14 Salina E Unit E-13 Salina D Unit E-13 Set equal to Salina E Salina C Unit E-13 Salina B Unit E-13 Salina B evaporite E-13 Salina A2 carbonate E-10 Salina A2 evaporite E-13 Set equal to Salina B evaporite Salina A1 Upper E-07 carbonate Salina A1 carbonate E-12 Salina A1 evaporite E-13 Set equal to Salina B evaporite Salina A0 Unit E-13 Set equal to Salina B evaporite Guelph E-08 Goat Island E-12 Gasport E-12 Lions Head E-12 Fossil Hill E-12 Cabot Head E-14 Manitoulin E-13 Queenston E-14 Georgian Bay E-14 Blue Mountain E-14 Collingwood E-14 Cobourg E-14 Sherman Fall E-15 Kirkfield E-15 Coboconk E-11 Gull River E-12 Shadow Lake E-09 No testing available, based on core observation Cambrian E-06 Based on flow test in DGR-2, Intera Engineering Ltd. (2010d,) TR April 1,

9 Figure 1 Straddle-Packer Test Results and Formation Hydraulic Conductivities. Formation Tops are based on DGR-1 and DGR-2 Reference Stratigraphy. April 1,

10 Figure 2 Anisotropy in Core Permeability from Gas Permeability Testing of DGR-2 Cores and Brine Permeability Testing of DGR-3 and DGR-4 Cores. April 1,

11 Table 3 Hydraulic Conductivity Anisotropy of Formations Formation Depth of Top in DGR- 1/2 K h : K v (mbgs) (-) Lucas : 1 Amherstburg (top 20 m) : 1 Amherstburg (lower m) 10 : 1 Bois Blanc : 1 Bass Islands (upper 20m) : 1 Bass Islands (lower 25 m) : 1 Salina G Unit : 1 Salina F Unit : 1 Salina E Unit : 1 Salina D Unit : 1 Salina C Unit : 1 Salina B Unit : 1 Salina B evaporite : 1 Salina A2 carbonate : 1 Salina A2 evaporite : 1 Salina A1 Upper carbonate 1 : 1 Salina A1 carbonate : 1 Salina A1 evaporite : 1 Salina A0 Unit : 1 Guelph : 1 Goat Island : 1 Gasport : 1 Lions Head : 1 Fossil Hill : 1 Cabot Head : 1 Manitoulin : 1 Queenston : 1 Georgian Bay : 1 Blue Mountain : 1 Collingwood : 1 Cobourg : 1 Sherman Fall : 1 Kirkfield : 1 Coboconk : 1 Gull River : 1 Shadow Lake : 1 Cambrian : 1 April 1,

12 3 Porosity Core samples were analysed for physical and/or liquid porosity by four laboratories. Data from all porosity reports has been compiled and further analysed in TR (Geofirma Engineering Ltd., 2011b), which is the primary reference for this section of the report. In TR-08-34, water-loss porosity values from supporting reports were corrected based on formation average brine density and TDS, to yield liquid porosities (Geofirma Engineering Ltd., 2011b). The recommended porosities for hydrogeological modelling are based on both physical (total) porosity measurements and liquid porosity measurements. In general, total porosity measurements are preferred. However, at this time the quantity and consistency of liquid porosity measurements exceeds that of the total porosity measurements in many formations. Porosity measurements based on gravimetric techniques (i.e. water loss by oven drying and in some cases vacuum distillation) can be used as proxy measurements of the actual porosity. The mass of water extracted is used to calculate the volumetric water content, from which porosity is estimated assuming negligible gas saturations. The presence of significant quantities of gypsum in some samples (particularly the Salina G through A2 units) could lead to significant overestimates of the porosity using this method. The water released from these samples may include waters of crystallization from gypsum (CaSO 4 2H 2 O) which is present in these formations but would not contribute to the porosity of the rock sample, as has been suggested by some authors (i.e. Intera Engineering Ltd., 2010a). Total porosity measurements are not subject to this error. For formations Salina G Unit through Salina A2 Unit carbonate, total porosities are used in preference to liquid porosities In the deeper Silurian and the Ordovician formations there is little evidence for significant quantities of gypsum. Furthermore, in many of these formations the number of liquid porosity samples greatly exceeds the number of total porosity samples. For this reason, liquid porosity numbers are thought to be more representative for many of these formations. Formation average porosities of the geological units were calculated using an arithmetic average of the porosities of each core sample. The core sample porosities were all given an equal weight when calculating the formation average, regardless of the porosity determination method used. Where no porosity data was available for a given unit, the porosity of an adjacent unit of similar rock type was used to estimate the formation average. Sample and formation average porosities are shown Figure 3. The shaded regions in Figure 3 indicate which porosity average (liquid or total) has been judged more representative. The average porosity values of the rock formations are summarised in Table 4. In the Descriptive Geosphere Site Model (Intera Engineering Ltd., 2011, DGR-TR ), porosities in the Ordovician shale are also calculated in an alternative fashion. The sample porosity values are grouped into high porosity massive shale samples with liquid porosities ranging between approximately 6 and 11%, and 'hard beds' consisting largely of limestone and/or siltstone and with porosities ranging from less than 1% up to 5%. To calculate the formation averages of Ordovician shale units in Table 4, all porosities within a particular unit were included in one formation average, and no distinction was made between massive shale and 'hard bed' samples. This was necessary to obtain a single formation average for modelling. A more refined representation of the heterogeneity within these formations would require a better knowledge of the proportion, typical thicknesses, and continuity of the shale and 'hard bed' units within these formations, which is unavailable at this time. April 1,

13 Figure 3 Measured Liquid and Total Porosity Values and Formation Average Values. The Shaded Regions of the Plot Show Which Values Were Used to Generate Table 4. April 1,

14 Table 4 Average Porosity Values of Formations. Formation Estimated Total Porosity (%) Comments Lucas 7.8 set equal to Bois Blanc Amherstburg (top 20 m) 7.8 set equal to Bois Blanc Amherstburg (lower 25 m) 7.8 set equal to Bois Blanc Bois Blanc 7.8 only liquid porosity available Bass Islands (upper 20m) 5.5 only liquid porosity available Bass Islands (lower 25 m) 5.5 only liquid porosity available Salina G Unit 16.7 only liquid porosity available Salina F Unit 10.7 Salina E Unit 11.9 Salina D Unit 8.9 set equal to Salina A2 evaporite Salina C Unit 19.4 Salina B Unit 15.8 Salina B evaporite 8.9 set equal to Salina A2 evaporite Salina A2 carbonate 12.4 Salina A2 evaporite 8.9 Salina A1 Upper carbonate 6.3 only liquid porosity available Salina A1 carbonate 4.0 use liquid porosity* Salina A1 evaporite 1.2 Salina A0 Unit 5.4 Guelph 13.1 use liquid porosity* Goat Island 2.8 use liquid porosity* Gasport 1.9 only liquid porosity available Lions Head 8.3 only liquid porosity available Fossil Hill 0.5 only liquid porosity available Cabot Head 10.4 use liquid porosity* Manitoulin 3.1 Queenston 7.5 use liquid porosity* Georgian Bay 7.1 use liquid porosity* Blue Mountain 7.1 use liquid porosity* Collingwood 2.3 Cobourg 1.9 Sherman Fall 2.9 Kirkfield 2.3 use liquid porosity* Coboconk 0.9 use liquid porosity* Gull River 2.2 use liquid porosity* Shadow Lake 8.9 use liquid porosity* Cambrian 10.1 * Due to more representative sampling. April 1,

15 4 Elastic Moduli Uniaxial compressive testing has been performed on samples from all of the exploratory boreholes at the DGR. Mechanical test results of DGR-1 and DGR-2 cores are presented in TR (Intera Engineering Ltd., 2009d) while mechanical test results of DGR-3 and DGR-4 cores are available from TR (Intera Engineering Ltd., 2010e) and TR (Intera Engineering Ltd., 2010f). Geomechanical testing of core from the inclined boreholes DGR-5 and DGR-6 is documented in TR (Geofirma Engineering Ltd., 2011c). TR also includes results from supplementary testing on cores collected from DGR-2, DGR-3, and DGR-4. A common characteristic of many of the rock samples that were tested is that the slope of the stress strain curve is not linear. Test results where the slope increases at lower stresses, enters an approximately linear phase at intermediate stresses, and then decreases as the sample begins to fail are commonly observed. The increasing slope at low stresses is likely due to the elastic closure of high aspect ratio voids and micro-fractures. It is common practice in rock mechanics to assume the roughly linear phase of the stress-strain curve is representative of the deformation behaviour of the in-situ rock. Table A.1 through Table A.5 in Appendix A show the results of a preliminary analysis of the undrained uniaxial tests carried out by CANMET (Intera Engineering Ltd., 2009d, 2010e, 2010f; Geofirma Engineering Ltd., 2011c). For the majority of samples there is a large variation between the dynamic moduli, the static moduli measured with LVDT transducers, and the static moduli measured using bonded strain gauges. Dynamic moduli are generally considered less reliable for engineering purposes than statically determined moduli, and are more likely to be affected by anisotropy and the presence of water in the system. Therefore, the dynamic results are not reported here and have been discounted. Of the two different static measurements, the LVDT data are thought to be more reliable for the following reasons (Gorski, 2009, personal communication from lead author of TR-07-03, TR-08-24, TR-08-39, and TR-09-07): The rock samples were frequently quite heterogeneous. Strain gauges may have been bonded to a part of the rock with different characteristics to those of the overall sample. Bonding the gauges to wet samples was challenging and may be a potential source of error. During the compression tests, the pore fluid became very pressurized and may have damaged the underside of the bonded gauges. Given these arguments, particularly regarding the sample heterogeneity, it is reasonable to assume that the LVDT data are probably the most reliable. Indeed, for TR (Intera Engineering Ltd., 2010e), TR (Intera Engineering Ltd., 2010f), and TR (Geofirma Engineering Ltd., 2011c) bonded strain gauges were not used to measure static deformation. In TR-07-03, TR-08-24, TR and TR the Young s modulus was calculated from the slope of the stress-strain curve at 40% of the uniaxial compressive strength (σ c ). For many of the samples, 0.4*σ c was much higher than the likely in-situ compressive stress. In some of the samples, (TR-07-03, TR and TR-08-39) reported Poisson s ratios exceeded 0.5, indicating that the rock was becoming dilatant (i.e. the sample volume was increasing with increased compressional loading). This is caused by the development of microcracks, and indicates that the sample is undergoing some plastic deformation. Elastic properties measured at this stress will not be representative of the behaviour of the undisturbed rock mass. In order to obtain representative values, some of the uniaxial compression tests performed by CANMET and presented in TR-07-03, TR and TR were reanalysed (numbers from TR were not reanalysed). April 1,

16 Where Poisson s ratios were unrealistically high the Young s modulus and Poisson s ratio were recalculated at a lower stress level, typically at 20% of σ c. Estimating the deformation modulus at a lower stress is also justified by the fact that the likely in-situ vertical confining stress is typically much lower than 0.4*σ c. In one case the static measures of Poisson s ratio were not reliable so the dynamically measured Poisson s ratio was used. In two of the DGR-2 tests, at depths of and mbgs, the test results were deemed unusable. The original and recalculated undrained elastic parameters are shown in Appendix A. The Young s modulus of a rock mass is usually less than the modulus measured for a small intact sample. There are a number of empirical methods to convert values from the intact rock Young s modulus to a rock mass modulus. One method, based on data from a large number of in-situ measurements is provided by Hoek and Deiderichs (2006). Their expression is: /2 1 (1) where: E rm = rock mass Young s modulus E i = intact rock Young s modulus D = disturbance factor (equals 0 for undisturbed rock) GSI = geological strength index (a function of rock mass structure and surface weathering, see Hoek and Brown, 1998) Assuming a relatively undisturbed and unweathered rock mass, a disturbance factor (D) of 0 and a GSI of 70 were used for all rock formations. This is consistent with the high rock quality designation (RQD) and low fracture frequency observed during coring (Intera Engineering Ltd., 2010c; 2010g). Tables A.9 through A.13 show the corrected rock mass moduli for the core samples. The calculation of the elastic moduli is intended to lead to the calculation of bulk modulus, and thereafter bulk compressibility and storage coefficient. However, the derivation of bulk modulus from Young s modulus is uncertain, and depends on a number of assumptions about the applicability of these numbers. Two key questions are: 1. Is the uniaxial compression test drained or undrained? 2. What mechanical boundary conditions constrain the in-situ rock? Unfortunately, neither of these questions can be answered unequivocally. The uniaxial compression tests were performed on unjacketed samples, and could therefore be considered drained test results. On the other hand, the duration of a typical test was on the order of a few hours. Given the very low permeability of many of the samples, the test may be more representative of an undrained response. It s likely that the truth lies somewhere between these bounding cases, i.e. a partially drained response. The assumed in-situ mechanical boundary condition is another important parameter. The calculation of the insitu compressibility and therefore storage coefficient hinges upon this assumption. In the majority of hydrogeological applications, it is presumed that all deformation of the rock mass due to changes in pore fluid pressure occurs by vertical deformation only, in other words changes in pore space volume occur as a result of April 1,

17 uniaxial strain in the rock. This assumption is quite appropriate in the near surface (down to a few hundred metres deep), as horizontal stresses commonly greatly exceed vertical stresses (Brown and Hoek, 1978). At greater depths this assumption no longer necessarily holds true. At the DGR site, the horizontal stress in the deeper buried units is estimated to exceed the vertical stress by a factor of 1.5 to 2.5 (Lam, 2007). Given these constraints, it may be more appropriate to assume that the in-situ formations will deform triaxially in response to changes in pressure. Given the uncertainty in these assumptions, it makes sense to produce two sets of bulk moduli representing high and low estimates of formation compressibility. The high-end estimate assumes: Lab-scale tests represent the undrained response of the sample. In-situ mechanical constraints allow for triaxial deformation. The low-end estimate assumes: Lab-scale tests represent the drained response of the sample. In-situ mechanical constraints allow for vertical deformation only. 4.1 Calculating the Low Compressibility Estimate Assuming an isotropic material, and having representative values for Young s modulus (E rm ) and Poisson s ratio (ν ) it is possible to calculate the bulk modulus (K) of the rock sample using the following expression: (2) where: K = triaxial bulk modulus (GPa) E rm = rock mass Young s modulus (GPa) ν = Poisson s ratio ( - ) The bulk modulus calculated in equation 2 assumes the rock is free to deform in all directions. The vertical bulk modulus (K ) can be calculated using the following expression: (3) where: K' = uniaxial bulk modulus (GPa) K = triaxial bulk modulus (GPa) ν = Poisson's ratio ( - ) The low bulk compressibility estimate (C L ) can be calculated as the inverse of the uniaxial bulk modulus (K'). Tables A.9 through A.13 show the calculated rock mechanical parameters, including the uniaxial bulk modulus of the rock samples from DGR-1, DGR-2, DGR-3, DGR-4, DGR-5, and DGR-6. Formation averages were April 1,

18 calculated using a geometric mean. Where no data exists for a formation, the average was assumed to be equal to nearby formations with similar lithologies. Figure 4 shows sample and formation log average values for the low compressibility estimate. Table 5 summarises the same data. Figure 4 Calculated Low and High Compressibility Values and Formation Average Values. April 1,

19 Table 5 Low and High Estimates of Average Rock Compressibility of Formations Formation C L (Pa -1 ) C H (Pa -1 ) Comments Lucas 2.3E E-11 Amherstburg (top 20 m) 4.7E E-10 Amherstburg (lower 25 m) 4.7E E-10 Bois Blanc 3.5E E-11 Bass Islands (upper 20m) 8.4E E-10 Bass Islands (lower 25 m) 8.4E E-10 Salina G Unit 7.5E E-10 Salina F Unit 1.0E E-10 Salina E Unit 1.0E E-10 Set equal to Salina F Unit Salina D Unit 1.9E E-11 Set equal to Salina A1 evaporite Salina C Unit 1.3E E-09 Salina B Unit 3.9E E-09 Salina B evaporite 1.9E E-11 Set equal to Salina A1 evaporite Salina A2 carbonate 5.2E E-10 Salina A2 evaporite 1.9E E-11 Set equal to Salina A1 evaporite Salina A1 Upper carbonate 3.3E E-11 Set equal to Salina A1 carbonate Salina A1 carbonate 3.3E E-11 Salina A1 evaporite 1.9E E-11 Salina A0 Unit 7.6E E-11 Guelph 3.5E E-11 Goat Island 2.0E E-11 Gasport 2.0E E-11 Set equal to Goat Island Lions Head 2.0E E-11 Set equal to Goat Island Fossil Hill 2.0E E-11 Set equal to Goat Island Cabot Head 3.3E E-09 Manitoulin 5.3E E-10 Queenston 6.6E E-10 Georgian Bay 1.4E E-09 Blue Mountain 2.6E E-09 Collingwood 3.7E E-11 Cobourg 2.4E E-11 Sherman Fall 5.9E E-10 Kirkfield 5.3E E-10 Coboconk 1.4E E-11 Gull River 2.0E E-11 Shadow Lake 4.1E E-11 Set equal to Cambrian Cambrian 4.1E E-11 April 1,

20 4.2 Calculating the High Compressibility Estimate For the high-end compressibility estimate we assume that the uniaxial compression tests described in TR-07-03, TR-08-24, TR and TR were undrained. Thus they provide us with undrained elastic moduli which must be corrected to obtain drained parameters. To calculate the drained bulk modulus (K d ) it is necessary to estimate the Biot-Willis (α) and Skempton (B) coefficients for the rock samples. A good description of α and B can be found in Wang (2000). In the absence of measurements of these parameters it is assumed that α = 0.9 and B = 0.8 for predominantly shale units and the relatively soft Salina B carbonate, and α = 0.7 and B = 0.3 for all other rock types. Table 6 shows typical values for α and B for various rock types. Table 6 Typical Poroelastic Parameters for Various Rock Types. Source: Wang (2000). Rock K d α B K ν φ (GPa) ( ) ( ) (GPa) ( ) ( ) Berea sandstone Boise sandstone Ohio sandstone Pecos sandstone Ruhr sandstone Weber sandstone Tennessee marble Charcoal granite Westerly granite Clay Mudstone Kayenta sandstone Limestone Berea sandstone Indiana limestone The following expression is used to calculate the drained bulk modulus (K d ) (Wang, 2000): 1 (4) where: K d = drained bulk modulus (GPa) K = triaxial bulk modulus (GPa) α = Biot-Willis coefficient B = Skempton s coefficient April 1,

21 K d is always less than K because in undrained conditions the pore fluid resists compression in addition to the rock matrix. A further assumption in calculating the high-end bulk compressibility is that the in-situ rock is able to deform in all three dimensions in response to changes in pore pressure. For this reason the triaxial bulk modulus (see equation 2) is used in Equation 4. The high-end bulk compressibility (C H ) is calculated as the inverse of the drained bulk modulus. Tables A.9 through A.13 show the estimated α and B coefficients, the corresponding drained bulk moduli, and the compressibility of the rock samples from DGR-1, DGR-2, DGR-3, DGR-4, DGR-5, and DGR-6. Formation averages were calculated using a geometric mean. Where no data exists for a formation, the average was assumed to be equal to nearby formations with similar lithologies. Figure 4 shows sample and formation log average values for the high compressibility estimate. Table 5 summarises the same data. 5 Specific Storage Using the formation average compressibility and porosity data presented in the previous sections, specific storage [m -1 ] is calculated as (Wang, 2000): S = ρ g + f ( C φc f ) ( 5 ) where: ρ f = fluid density g = gravitational acceleration (9.806 m/s 2 ) C = drained bulk compressibility φ = porosity C f = the fluid compressibility This expression assumes that the solid grains and pores are incompressible. Please note that this assumption is inconsistent with the derivation of drained compressibility (high compressibility estimate) in the previous section. It is possible to correct the storage number using the same assumed α and B values as in Section 4, but the resulting specific storage values will not be significantly different from those reported here. For C f, a value of 3.3E-10 Pa -1 is used for the brine compressibility (Earlougher, 1977). Fluid density at the DGR varies considerably with depth. Using data from multiple sources, described in Intera Engineering Ltd. (2011, DGR-TR ), formation average fluid densities have been calculated (see Table 7). These formation average fluid densities have been used for the specific storage calculation. The formation average porosities from Table 4 and the drained compressibility from Table 5 were used for calculating the specific storage. Formation average specific storage coefficients are summarised in Table 7 and displayed graphically in Figure 5. Also shown in Figure 5 are formation specific storages fitted by analysis of the straddle-packer hydraulic testing (Geofirma Engineering Ltd., 2011a, TR-08-32). Although there is clearly not a perfect agreement between the storage coefficients determined by lab-scale geomechanical testing and field scale hydraulic testing, the juxtaposition of these results shows that on average they both produce results of a similar order of magnitude. Obtaining similar results from a very different assay increases confidence that the derived storage coefficients are reasonable. In one formation, Salina A0 Unit, the upper and lower geomechanically based estimates are identical. This is due to rounding of the storage coefficient to one significant figure. April 1,

22 Table 7 Formation Averages of Fluid Density and Specific Storage Formation Fluid Density S s low estimate S s high estimate (kg/m) 3 (m 1 ) (m 1 ) Lucas 990 5E-07 7E-07 Amherstburg (top 20 m) 993 7E-07 2E-06 Amherstburg (lower 25 m) 993 7E-07 2E-06 Bois Blanc 993 6E-07 1E-06 Bass Islands (upper 20m) E-06 2E-06 Bass Islands (lower 25 m) E-06 2E-06 Salina G Unit E-06 2E-06 Salina F Unit E-06 7E-06 Salina E Unit E-06 7E-06 Salina D Unit E-07 7E-07 Salina C Unit E-06 1E-05 Salina B Unit E-06 2E-05 Salina B evaporite E-07 7E-07 Salina A2 carbonate E-06 2E-06 Salina A2 evaporite E-07 6E-07 Salina A1 Upper carbonate E-07 1E-06 Salina A1 carbonate E-07 1E-06 Salina A1 evaporite E-07 4E-07 Salina A0 Unit E-07 3E-07 Guelph E-07 1E-06 Goat Island E-07 5E-07 Gasport E-07 5E-07 Lions Head E-07 7E-07 Fossil Hill E-07 4E-07 Cabot Head E-06 3E-05 Manitoulin E-07 1E-06 Queenston E-06 5E-06 Georgian Bay E-06 1E-05 Blue Mountain E-06 3E-05 Collingwood E-07 1E-06 Cobourg E-07 6E-07 Sherman Fall E-07 2E-06 Kirkfield E-07 2E-06 Coboconk E-07 4E-07 Gull River E-07 6E-07 Shadow Lake E-07 1E-06 Cambrian E-07 1E-06 April 1,

23 Figure 5 Specific Storage Data, Formation Average Ranges from Geomechanical Testing, and Fitted Specific Storage from Straddle-Packer Testing. Note that the Point Values are Plotted for Illustrative Purposes but were not used for Calculating the Averages. In a few instances the disagreement between hydraulic testing and geomechanical results is large. For example, in the Cabot Head formation the geomechanically determined storage coefficient is roughly two orders of magnitude higher than the hydraulic testing storage coefficient. However, both straddle packer tests on the Cabot Head also include other formations, which may mask the Cabot Head signal. It is also the case that the geomechanical storage coefficient is based on only one core sample, which may have sustained damage or may not be representative of the formation average. Hydraulic test analysis is less sensitive to specific storage than it is to permeability and formation pressure. Therefore, at present the geomechanically derived values should probably be considered more reliable for use in numerical modelling. However, it is important to remember that April 1,

24 some of the geomechanical storage coefficients are somewhat uncertain, and may require substantial changes if new information becomes available. Uncertainty can also be mitigated through the use of model sensitivity cases. 6 Rock Density There are numerous sources for bulk dry density and grain density in rock at the Bruce DGR site. These include: Intera Engineering Ltd. (2010h, TR-07-17), bulk dry and grain density Intera Engineering Ltd. (2009e, TR-08-06), bulk dry and grain density Intera Engineering Ltd. (2010a, TR-07-18), grain density Intera Engineering Ltd. (2010i, TR-08-27), grain density Intera Engineering Ltd. (2010b, TR-08-28), bulk dry and grain density Geofirma Engineering Ltd. (2011d, TR-08-40), bulk dry and grain density Geofirma Engineering Ltd. (2011e, TR-09-08), grain density In total, there are 106 measurements of bulk dry density and 348 measurements of grain density. The measurements of grain density also cover a larger number of formations. As a result, there are formations for which grain density has been measured, but no corresponding dry density is available. For these formations, bulk dry density was estimated using the following equation: ( 6 ) where: ρ d = bulk dry density (kg/m 3 ) ρ g = grain density (kg/m 3 ) φ = porosity This calculation was done on a formation basis, and formation average liquid porosities were used. Where there is no data for a given formation, grain density was estimated based on adjacent formations with similar mineralogy, and dry density was once again calculated using Equation 6. Formation average densities were calculated using an arithmetic mean. Figure 6 shows the measured values of grain density and bulk dry density as well as the arithmetic averages and estimated averages for each formation. Table 8 lists the average formation values for each formation for dry bulk density and grain density, and comments explaining the basis for presented values. April 1,

25 Figure 6 Grain and Bulk Dry Density, Measured Data and Formation Averages. April 1,

26 Table 8 Average Dry Bulk Density and Grain Density of Formations Formation Dry Bulk Density (kg/m 3 ) Grain Density (kg/m 3 ) Comments Lucas estimated ρ g, ρ d from Equation 6 Amherstburg (top 20 m) estimated ρ g, ρ d from Equation 6 Amherstburg (lower 25 m) estimated ρ g, ρ d from Equation 6 Bois Blanc estimated ρ g, ρ d from Equation 6 Bass Islands (upper 20m) estimated ρ g, ρ d from Equation 6 Bass Islands (lower 25 m) estimated ρ g, ρ d from Equation 6 Salina G Unit ρ d from Equation 6 Salina F Unit Salina E Unit Salina D Unit ρ g set equal to A1 evap, ρ d from Equation 6 Salina C Unit Salina B Unit Salina B evaporite ρ g set equal to A1 evap, ρ d from Equation 6 Salina A2 carbonate Salina A2 evaporite Salina A1 Upper carbonate ρ g set equal to A1 evap., ρ d from Equation 6 Salina A1 carbonate Salina A1 evaporite Salina A0 Unit ρ d from eq. 6 Guelph Goat Island ρ d from eq. 6 Gasport ρ g set equal to Goat Isl., ρ d from Equation 6 Lions Head ρ g set equal to Goat Isl., ρ d from Equation 6 Fossil Hill ρ g set equal to Goat Isl., ρ d from Equation 6 Cabot Head Manitoulin Queenston Georgian Bay Blue Mountain Collingwood Cobourg Sherman Fall Kirkfield Coboconk Gull River Shadow Lake Cambrian April 1,

27 7 Effective Diffusion Coefficients Effective diffusion testing on cores from DGR-2, DGR-3, and DGR-4 was carried out by UNB using throughdiffusion and X-ray radiography techniques. This testing covered the bulk of the Silurian and Ordovician formations. Detailed results are available in the technical reports Intera Engineering Ltd. (2010h, TR-07-17) and Intera Engineering Ltd. (2010i, TR-08-27). The reports provide information on both tritium (HTO) and iodide (NaI) diffusion rates. Measurements of HTO diffusion are less numerous and only encompass the Ordovician shale and limestone units. Where effective diffusion of HTO has been measured, the HTO diffusion coefficient is on average 1.9 times higher than the NaI diffusion coefficients, likely due to the larger size of the hydrated iodide ion compared to tritium and to anion exclusion effects in the negatively charged clay-rich sediments. The reasons for this difference are discussed further in Intera Engineering Ltd. (2011, DGR-TR ). The majority of diffusion measurements determined the diffusion rate normal to the bedding planes (86 NaI, 32 HTO), but a number of measurements were also performed parallel to the bedding planes (24 NaI, 9 HTO). With the exception of two samples in the upper Silurian, diffusion coefficients parallel to the bedding planes were higher. The parallel to normal anisotropy ranged from 1 to 4 for measurements using the NaI tracer, and 1 to 7 for experiments using the HTO tracer. For the purposes of transport modelling, the iodide tracer effective diffusion measurements are more representative. The formation average NaI diffusion coefficients normal to the bedding plane were calculated using a geometric mean, and are shown in Table 9 and Figure 7. Where diffusion measurements for a given formation are not available, the effective diffusion coefficient is assumed to be equal to a nearby formation with similar lithology and porosity. In some cases, a suitable comparison with another formation is not possible. For these formations, the effective diffusion coefficient has been calculated according to Archie s Law (Boving and Grathwohl, 2001): ( 6 ) where: D e = effective diffusion coefficient (m 2 /s) D o = free water diffusion coefficient (m 2 /s) φ = total (physical) porosity m = an empirical coefficient A value of m2/s has been used for the free water diffusion coefficient (Lerman, 1979). A value of 2.0 has been used for m (Intera Engineering Ltd., 2011, DGR-TR ). The formation average total porosities from Table 4 have been used in this calculation. If necessary, the HTO diffusion coefficient can be estimated by multiplying the NaI diffusion coefficient by 2. The estimated average parallel to bedding plane diffusion coefficients are shown in Table 9 (Intera Engineering Ltd., 2011, DGR-TR ). April 1,

28 Table 9 Formation Averages of Effective Diffusion Coefficient Formation D e NaI, normal to bedding m 2 /s Horizontal to Vertical anisotropy Comments Lucas 5.9E-12 1:1 calculated using Equation 7 Amherstburg (top 20 m) 5.9E-12 1:1 calculated using Equation 7 Amherstburg (lower 25 m) 5.9E-12 1:1 calculated using Equation 7 Bois Blanc 5.9E-12 1:1 calculated using Equation 7 Bass Islands (upper 20m) 3.2E-12 1:1 calculated using Equation 7 Bass Islands (lower 25 m) 3.2E-12 1:1 calculated using Equation 7 Salina G Unit 4.3E-13 2:1 Salina F Unit 4.1E-12 2:1 Salina E Unit 4.7E-12 2:1 Salina D Unit 4.7E-12 2:1 set equal to Salina E Salina C Unit 1.1E-11 2:1 Salina B Unit 1.2E-11 2:1 Salina B evaporite 7.7E-14 2:1 set equal to Salina A2 evaporite Salina A2 carbonate 1.2E-12 2:1 Salina A2 evaporite 7.7E-14 2:1 Salina A1 Upper carbonate 1.8E-13 1:1 set equal to Salina A1 carbonate Salina A1 carbonate 1.8E-13 2:1 Salina A1 evaporite 3.0E-14 2:1 Salina A0 Unit 3.0E-14 2:1 set equal to Salina A1 evaporite Guelph 3.2E-12 1:1 calculated using Equation 7 Goat Island 1.5E-13 2:1 Gasport 1.5E-13 2:1 set equal to Goat Island Lions Head 9.6E-13 2:1 calculated using Equation 7 Fossil Hill 9.6E-13 2:1 calculated using Equation 7 Cabot Head 3.1E-12 2:1 Manitoulin 1.5E-13 2:1 set equal to Goat Island Queenston 1.0E-12 2:1 Georgian Bay 6.8E-13 7:1 see discussion in PR Blue Mountain 8.2E-13 2:1 Collingwood 4.9E-13 2:1 Cobourg 3.7E-13 2:1 Sherman Fall 2.2E-13 2:1 Kirkfield 4.2E-13 2:1 Coboconk 2.7E-13 2:1 Gull River 2.6E-13 2:1 Shadow Lake 9.4E-12 2:1 calculated using Equation 7 Cambrian 5.0E-12 1:1 calculated using Equation 7 April 1,

29 Figure 7 Measured Effective Iodide Diffusion Coefficient and Formation Average Values, Normal to Bedding Planes. 8 Data Use and Conclusions The limitations on data are discussed and addressed in the individual Technical Reports cited as sources for this report. If the reader wishes to gain a greater understanding of the limitations or potential accuracy of numbers reported here, please refer to the relevant cited references. That notwithstanding, the following comments on data quality and use are provided for the hydrogeologic data presented in this report. The horizontal hydraulic conductivity values presented here represent an average value from a relatively large sample of the total rock formation. The lab-scale hydraulic conductivity values provide limited information on the vertical anisotropy in the system. April 1,

30 Where the effect of gypsum can be discounted, the liquid porosity values provide a reliable estimate of the porosity of the test sample. Where this is not the case, the physical (total) porosity measurements provide a valuable check on the liquid porosity estimates, and are in general agreement with these estimates. In a few formations the average liquid porosity exceeds the average total porosity. A possible explanation for these inconsistent results is that in some formations, there were far fewer samples used for total porosity measurement compared to the number of liquid porosity samples. The reduced number of samples may represent a biased estimate of the formation average porosity The derivation of representative storage data from the lab-scale, uniaxial tests is difficult, requiring a number of assumptions about the poroelastic behaviour and mechanical constraints upon the rock formations. However, in the absence of further information this analysis provides the best available estimate of the likely range of in-situ specific storage values for these rock formations. The general agreement between geomechanically derived specific storage coefficients and specific storage estimated from field-scale hydraulic testing builds some confidence that the predicted storage range likely brackets the actual value in most formations. Storage values will influence the understanding of the genesis and future longevity of the underpressures in the mid-to-upper Ordovician sequences. Uncertainty can be dealt with through sensitivity analysis of models. It is fair to say that grain and bulk densities have been very well characterised in the Ordovician units at the DGR site. In the Silurian units, bulk density measurements are sparser, but provide a reasonable estimate of the true value. The majority of effective diffusion measurements were normal to the bedding planes, using NaI as the tracer. These measurements provide good coverage in the Ordovician and in the Silurian units, although the coverage in the Silurian sequences is less than in the Ordovician rocks which comprise the main upward diffusional barrier for the DGR. The measurements of NaI diffusion parallel to the bedding planes and HTO diffusion are less numerous, and largely limited to Ordovician formations. These measurements allow for estimation of the diffusion coefficient anisotropy, and the impact of anion exclusion and hindrance of the pore walls. Obtaining undisturbed core can be quite difficult, and samples may not be representative of the in-situ formation properties. Core samples may potentially be damaged during coring, during stress relief of the samples, or during transport, causing the formation of microcracks. This microcracking may appreciably affect the permeability, physical porosity, diffusion, and mechanical deformation properties of the core samples. In many cases this potential source of error was mitigated by measuring properties such as porosity or permeability under confining stresses close to the in-situ vertical confining stress at the sample depth. This report is a concise compilation of field and lab scale data from site characterisation experiments. The data shown is relevant to the hydrogeological characterisation of the Bruce DGR site. The parameter values presented in this technical report represent the latest and most accurate data on the likely in-situ parameters at the Bruce DGR site. These data will be considered in the development of the Descriptive Geosphere Site Model Report for the Bruce DGR site. 9 References Boving. T.B. and P. Grathwohl, Tracer diffusion coefficients in sedimentary rocks: correlation to porosity and hydraulic conductivity, Journal of Contaminant Hydrology Vol. 53, pp Brown, E. and E. Hoek Trends and relationships between in-situ stresses and depth, International Journal of Rock Mechanics and Mining Sciences, Vol. 15, pp Golder Associates Ltd., LLW Geotechnical Feasibility Study Western Waste Management Facility Bruce Site Tiverton, Ontario. Golder Associates Ltd. Report Toronto. April 1,