Oral Session 10a. Mass Transfer. Chairman : P. de CANNIERE

Transcription

1 Oral Session 10a Mass Transfer Chairman : P. de CANNIERE

2 O-10a-1 LARGE SCALE TRANSPORT BEHAVIOUR OF A JURASSIC ARGILLACEOUS ROCK AS INFERRED FROM PROFILES OF STABLE WATER ISOTOPES AND CHLORIDE Thomas Gimmi 1, H. Niklaus Waber 1, André Rübel 2, Andreas Gautschi 3 1. Bern University, Switzerland 2. GRS Braunschweig, Germany 3. Nagra, Wettingen, Switzerland Low-permeability formations are envisioned in many countries as potential host rocks for the disposal of radioactive or other hazardous waste. Nagra, the Swiss cooperative for disposal of radioactive waste, is currently investigating a Jurassic sediment, the so-called Opalinus Clay, at Benken in northeast Switzerland. Transport properties of host rocks are often assessed by measuring material properties on small samples, and by mathematically scaling these properties up to scales of interest. In the case of a repository, an experimental verification of upscaled transport properties is precluded by the very large temporal and spatial scales (thousands to millions of years and hundreds of metres). In this situation, the analyses of natural tracers, like stable water isotopes or chloride, offer unique possibilities. Looking back to what has happened in the past allows model concepts and material properties at relevant scales to be reliably evaluated. In the deep borehole at Benken, a sequence of claystones and marls with hydraulic conductivities generally below m s -1 was encountered at a depth of about 450 to 700 m below ground. The low-permeability formations are sandwiched between two aquifers. Profiles of stable isotope ratios ( 18 O and 2 H) as well as of Cl in these formations were obtained with various methods, namely diffusive exchange and distillation for stable water isotopes, and aqueous extraction and squeezing of core samples for Cl. Profiles of all three tracers have similar shapes. Maximum values occur near the centre or upper half of the Dogger units, and concentrations decrease clearly towards the underlying aquifer. The shapes of the profiles hint at diffusive transport processes. Mass exchange seems to have happened mainly between the underlying aquifer and the low-permeability zone, whereas the upper aquifer obviously was less important. To evaluate possible mechanisms and time scales of evolution, we performed a series of advective-dispersive model calculations. In view of the almost undisturbed geologic stratification, a one-dimensional approach was chosen. The aquifers located above and below the low-permeability zones defined upper and lower boundaries of the modelled domain. Our basic hypothesis was that the dominant features of the profiles originate from a relatively late change in the lower boundary condition. Diffusive, and possibly also advective, transport has then propagated the perturbation of concentrations at the lower boundary into the Dogger units. The concentration drop at the lower boundary was modelled in two different ways. The first was simply to assume constant concentrations according to the values measured today. The second takes into account a gradual decrease from the initial values to those observed today by introducing a mixing cell at the boundary. The concentration drop in this mixing cell results from the interplay of advective and diffusive mass transfer across the boundary to the low-permeability zone and advective input and output of aquifer water. Clays In Natural And Engineered Barriers For Radioactive Waste Confinement Page 135

3 O-10a-1 The following conclusions could be drawn : - the methods applied seemed to be well suited to estimate concentrations of stable water isotopes and chloride of low-permeability samples, - the observed profiles of natural tracers developed mainly under the influence of molecular diffusion. No signatures of advection could be detected, - the chloride data show more scatter than the stable water isotope data, which makes the interpretation slightly more difficult. A possible reason for the larger scatter is, among others, that the fraction of chloride accessible pore space is not exactly known, - from the profiles of stable water isotopes and a laboratory diffusion coefficient, an evolution time of about 1 to 2 Ma was estimated. The estimated evolution time is plausible if compared against geological evidence. This, and the absence of advective signatures, allows an additional, more general, conclusion to be drawn : - at least to within an order of magnitude, small-scale (centimetres or metres, months) parameters of the investigated Jurassic sediments, such as diffusion coefficients, are relevant also at the formation scale (tens of metres, millions of years). Page 136 Clays In Natural And Engineered Barriers For Radioactive Waste Confinement

4 O-10a-2 GAS TRANSPORT PROCESSES IN OPALINUS CLAY - CONCEPTUAL CONSIDERATIONS AND EXPERIMENTAL EVIDENCE Paul Marschall 1, Steve Horseman 2, R. Senger 3 1. NAGRA, Hardstrasse 73, Wettingen CH British Geological Survey, Kingsley Dunham Centre, Keyworth, Nottingham, UK 3. Intera Inc., 9111A Research Blvd., Austin, Texas The investigation of gas migration processes through argillaceous formations is a key issue in the assessment of repository performance. As part of the Swiss waste disposal programme, gas transport processes in Opalinus Clay were studied in-situ and by means of laboratory experiments. Of particular value are the geoscientific data from a 1000 m deep investigation borehole at Benken (Northern Switzerland) and the studies at the Mont Terri rock laboratory. Gas transport in the Opalinus Clay is controlled not only by the hydro-mechanical properties of the rock mass (porosity, intrinsic permeability, rock strength) but also by the gas pressure at the generation locus and the hydro-mechanical state of the rock (saturation, porewater pressure, rock stress). Gas release through the Opalinus Clay may be accommodated by the following mechanisms : - advective-dispersive transport of gas dissolved in the porewater, - generalized Darcy flow, (two-phase flow), - microscopic pathway dilation driven by the gas production rate (quasi-stationary fracture growth), - macroscopic gas / hydro-fracturing (fracture initiation when frac pressure is exceeded). From the perspective of longterm repository performance, assessment of these gas transport processes may be linked (i) to gas transport capacity and (ii) to possible impact on the hydraulic barrier of the rock formation. Advective and dispersive transport of gas in porewater is a straightforward migration mechanism which is characterised by Darcy s law, Fick s law and Henry s law. Due to the low hydraulic conductivity of the Opalinus Clay in the order of to m/s, the efficiency of this transport mechanism is very limited. Typical diffusion coefficients in the Opalinus Clay are in the order of m 2 /s. Two-phase flow is described as the process of porewater displacement by the gas phase under the influence of visco-capillary forces. The controlling factor for this mechanism is the capillary gas entry pressure. The gas entry pressure of Opalinus clay has been investigated through in-situ gas threshold pressure tests in Mont Terri and in the Benken borehole and by gas permeability tests with core specimen from both sites. The determined entry pressures were in the range of 1 10 MPa. It is believed, that gas-water displacement preferentially takes place in the connected system of macropores of the rock matrix (> 30 nm). Microstructural analyses of the Opalinus clay at Benken exhibit a total porosity of 0.11, whereby the fraction of macropores amounts to 20%. Due to the low tensile strength of the Opalinus Clay it is most likely that the rock will not withstand long-term gas pressures with a level higher than the minimum principal stress σ 3. In fact, it seems plausible that microfractures will form before the macroscopic tensile strength of the formation is exceeded. This microscopic pathway dilation causes an inelastic deformation of the pore space and a detectable, stress-dependent increase in intrinsic permeability. Microfracturing occurs when the gas pressure is allowed to grow slowly: the additional pore volume which is the result of dilation has to hold a steady balance with the Clays In Natural And Engineered Barriers For Radioactive Waste Confinement Page 137

5 O-10a-2 volumetric gas production rate so that a quasi-stationary gas flow may evolve along the newly opened gas flow path. Qualitative and quantitative evidence for dilatant gas transport mechanisms at gas pressures below minimum stress was gained through a series of in-situ experiments at Mont Terri. Hydraulic tests before and after the in-situ gas tests did not show significant changes of hydraulic conductivity of the rock mass due to the injection. A macroscopic tensile fracture develops when the gas pressure is larger than the sum of the minimum principal stress σ 3 and the tensile strength of the rock (frac-pressure). The macroscopic frac is initiated quasi instantaneously. The propagation comes to a halt when the gas pressure in the macroscopic fracture becomes less than the value of the minimum principal stress (shut-in pressure). At Mont Terri hydraulic fracs and gas re-fracs were created. Frac pressure was surprisingly high at about 9 MPa and refrac pressure was approximately 5 MPa. In the light of performance assessment, two-phase flow and microscopic pathway dilation are believed to be the most likely processes to accommodate the release of repository generated gas in Opalinus clay. The transition between the two processes is not sharp - due to the high gas entry pressures the on-set of microscopic pathway dilation during pressure build-up may be expected before water displacement starts. In fact, both gas transport processes will not give rise to significant alteration of the rock properties (e.g. enhancement of intrinsic permeability) during the passage of a gas phase. Due to the low gas generation rates, the occurence of macroscopic hydro- and gas fracs can be excluded. Page 138 Clays In Natural And Engineered Barriers For Radioactive Waste Confinement

6 DIFFUSION OF TRACERS HTO, Na +, Cl - AND Sr 2+ THROUGH COMPACTED BENTONITE : UNDERSTANDING THE EFFECT OF IONIC STRENGTH AND BULK DRY DENSITY Ian C. Bourg 1, 2, *, Alain C.M. Bourg 1, Garrison Sposito 2 O-10a-3 1. Environmental Hydrogeochemistry Group, UMR 5034, Univ. of Pau-CNRS, Hélioparc Pau Pyrénées, Pau Cedex 9, France 2. Environmental Geochemistry Group, Hilgard Hall, #3110, Univ. of California, Berkeley, CA , USA Current performance assessment models for radionuclide diffusion through compacted bentonite use a "black box" approach with many adjustable parameters which cannot be independently determined (surface diffusion coefficients for cations, effective porosities for anions) or are difficult to explain mechanistically (very high tortuosity coefficients). This may be problematic when extrapolating these semi-empirical descriptions to larger scales of distance and time (in a waste disposal situation) and in the presence of coupled fluxes (e.g., electromigration caused by groundwater of different ionic strengths on opposite sides of the clay barrier). A mechanistic model, if it were capable of predicting the results of small scale diffusion experiments, would be more suited to such a task. A transport model adapted to the structure of smectite clays should probably describe the dual nature of its porosity (interlayer/inter-aggregate pores), and the properties of the interlayer pore water (high viscosity, low electrostatic potential). Such a model would need to be solved numerically. Simplified versions are derived here and solved analytically in two cases: diffusion of uncharged species (no electrostatic effects) and diffusion at high degrees of compaction (no inter-aggregate pores). The apparent diffusivity of water in compacted Kunipia-F bentonite is predicted from 0.7 to 2.0 kg/l of bulk dry density, using independently determined data of interlayer spacing vs. bulk dry density and interlayer water diffusivity vs. interlayer spacing, and a theoretical interlayer tortuosity of 3. The inter-aggregate pores were assumed to be flattened enough by compaction that a tortuosity of 3 also applied to them. The agreement to experimental HTO diffusion data is good within 0.3 orders of magnitude. The shape of the predicted curve of apparent diffusivity vs. bulk dry density is also in agreement with the experimental data. The steady-state effective diffusivity of Na +, Sr 2+, and Cl - through compacted MX-80 bentonite at 1.8 kg/l bulk dry density is then predicted, in the range of to 1 mol/l background ionic strength on both sides of the clay slice. The predictions are in excellent agreement with experimental data if all species diffuse 15 times slower in the interlayers than in pure water. This is the only unknown parameter and fits within the range of independently determined diffusion coefficients for sodium in the interlayers of a two-layer hydrate of Namontmorillonite. Clays In Natural And Engineered Barriers For Radioactive Waste Confinement Page 139

7 O-10a-3 These two successful predictions of experimental data show that 1. the mechanics of diffusion through compacted clay (porosity, tortuosity) can be described by accounting for the dual nature of its porosity and the higher viscosity of interlayer water, and 2. the electrostatics of diffusion through compacted clay (diffusion of charged species) can be described by a membrane description of the montmorillonite interlayers. A key point here is that the model is consistent in describing the diffusion of both cations and anions (instead of using semi-empirical coefficients of "surface diffusion" for cations and "effective porosity" for anions). Another key point is that the model agrees well with independent studies of clay properties (notably studies of clay interlayers), thus allowing for very few unknown parameters. Considering these advances, the agreement to the experimental data is quite encouraging. Page 140 Clays In Natural And Engineered Barriers For Radioactive Waste Confinement

8 O-10a-4 THREE - DIMENSIONAL MODELING OF DIFFUSION IN THE MAIN FAULT OF THE OPALINUS CLAY IN THE MONT - TERRI LABORATORY J. Jaime Gómez-Hernández 1, Carolina Guardiola-Albert 1, Yvon Droullier 2, J. Carlos Mayor 3, P. de Cannière 4 1. Universidad Politécnica de Valencia, Spain 2. Andra, France 3. ENRESA, Spain 4. SCK-CEN, Belgium Between November 1999 and February 2001, a diffusion experiment was carried out in borehole BFM-C3. The diffusion experiment was carried out in three phases : - phase A: 18 th November th June Installation of downhole equipment, and pressure monitoring in the section where the tracer will be introduced. In-diffusion monitoring of 4 He, and out-diffusion experiment of 3 He, - phase B: 13 th June th February Installation of tracer test equipment, tracer injections and sampling, - phase C: 19 th Feburary 2001 overcoring of the borehole FMC-3 and tracer analysis of the core samples. Prior to the introduction of iodine and tritium into the tracer interva, the diffusion from the formation into the borehole of the naturally occurring 4 He was monitored. Subsequently, 3 He was introduced and its evolution monitored. This is the only diffusion experiment in the Mont Terri laboratory that has used a gas tracer; its larger diffusivity would favor the monitoring of its migration to a nearby borehole. A two-dimensional model was used to predict and later verify the Helium diffusion. Iodine and tritium were introduced in the tracer interval of borehole BFM-C3 in June The experiment was designed as a pulse test. The evolution of the tracers in the observation section was followed by weekly samples. Overcoring took place on February 19 th The rock volume around the test interval was recovered and samples taken. The analysis of these samples produced tracer profiles away from the borehole yielding information on the threedimensional distribution of the tracers at the end of the test. A model was constructed aiming to reproduce the temporal evolution at the tracer interval and the spatial distribution at the end of the test. First, a two-dimensional model provided initial approximations of the diffusivity of the tracers and the porosity of the formation. Then, a three-dimensional finite-difference model was built to reproduce the 3-D spatial profiles. Clays In Natural And Engineered Barriers For Radioactive Waste Confinement Page 141

9 O-10a-4 HTO in BFMC C/ 0.6 Co C/Co profile Days 0 5 cm C/Co profile cm C/Co profile cm The figures above show the reproduction of some of the experimental profiles, both in time and space for tritium. The parameters used to fit this curves are, effective diffusivity of m 2 /s, and accessible porosity of 0.12 (apparent diffusivity m 2 /s). Page 142 Clays In Natural And Engineered Barriers For Radioactive Waste Confinement

10 O-10a-5 MICROSTRUCTURE AND MICRO - / MACRO - DIFFUSION BEHAVIOR IN BENTONITE Yasuaki Ichikawa 1, Naoki Fujii 2, Katsuyuki Kawamura 3, Kazumi Kitayama 4 1. Div. of Environmental Engineering and Architecture, Graduate School of Environmental Studies, Nagoya University, Chikusa, Nagoya , Japan Tel. (+81) , Fax (+81) , YIchikawa@nucc.cc.nagoya-u.ac.jp 2. Nagoya University, Chikusa, Nagoya , Japan 3. Tokyo Institute of Technology, Meguro, Tokyo , Japan 4. NUMO, Minato, Tokyo , Japan Clay is a microinhomogeneous material with nanoscale microstructure. Key issues to understand the behavior of such a finely microinhomogeneous material are as follows: 1) the microstructure is characterized in details, 2) the local distribution of material properties is identified by experiment or simulation, and 3) the microscale characteristics are related to the macroscale behavior by a seamless manner. For characterizing a microstructure of bentonite, we introduce a conforcal laser scanning microscope (CLSM) together with SEM. By CLSM we can specify a 3-D configuration under atmospheric condition. Properties of watersaturated bentonite are mainly controlled by hydrated montmorillonite, which is the major clay mineral of bentonite. Smectite minerals including montmorillonite are extremely fine and poorly crystallized, so it is difficult to determine the properties by experiment. We inquire into the physicochemical properties by a molecular dynamics (MD) simulation method. Then, we develop a multiscale homogenization analysis (HA) method to extend the microscopic characteristics to the macroscopic behavior. We show numerical examples of a diffusion problem. Results are as follows : 1) by CLSM observation it is possible to specify a 3-D geometry of bentonite under a normal atmospheric condition, 2) a new interatomic potential model and its parameter set are provided for MD simulation of various oxide crystals containing Si, Al, Na, O, Cl, H(OH), O(H 2 O) and H(H 2 O) atoms, 3) MD simulation of water and ice-ih shows to reproduce structures and physical properties well under wide range of temperature, 4) swelling property of beidellite is calculated by MD for various hydration number, which is consistent with experimental data. We observe a step-wise increase of basal spacing as increasing the hydration number. The step-wise behavior corresponds to zero-, one-, twoand three-layer(s) of hydration, 5) by MD simulation of a clay mineral system hydrated with pure water and aqueous solution, we can determine local distributions of structural and physical properties such as diffusivity and viscosity, which can be used for the following HA, 6) applying a multiscale HA, the effective diffusion coefficient of tritium water HTO for bentonite is determined. The effective diffusivity is affected by intermolecular interaction between HTO and smectite minerals and the pore structure. The results are consistent with experimental data well, Clays In Natural And Engineered Barriers For Radioactive Waste Confinement Page 143

11 O-10a-5 7) the effective diffusion coefficient of HTO in pure smectite bentonite (Kunipia-F) is times larger than one in a normal bentonite (Kunigel-V1) under the same dry density 2.0Mg/m 2. If we consider the effects of porosity change and tortuosity, this difference becomes larger, 8) by a multiscale HA we estimate that the average number of montmorillonite lamellae containing in a stack is about 8. Page 144 Clays In Natural And Engineered Barriers For Radioactive Waste Confinement

12 Agence nationale pour la gestion des déchets radioactifs Parc de la Croix Blanche -1/7, rue Jean Monnet Châtenay-Malabry Cedex Tél. : (33) Andra November crédits photos : Office du Tourisme de Reims - conception graphique : adamskidesigns.com