Thomas Gimmi

In-situ diffusion of HTO, 22Na+, Cs+ and I- in Opalinus Clay at the Mont Terri underground rock laboratory

Summary The diffusion properties of the Opalinus Clay were studied in the underground research laboratory at Mont Terri (Canton Jura, Switzerland) and the results were compared with diffusion data measured in the laboratory on small-scale samples. The diffusion of HTO, 22Na+, Cs+ and I- were investigated for a period of 10 months. The diffusion equipment used in the field experiment was designed in such a way that a solution of tracers was circulated through a sintered metal screen placed at the end of a borehole drilled in the formation. The concentration decrease caused by the diffusion of tracers into the rock could be followed with time and allowed first estimations of the effective diffusion coefficient. After 10 months, the diffusion zone was overcored and the tracer profiles measured. From these profiles, effective diffusion coefficients and rock capacity factors could be extracted by applying a two-dimensional transport model including diffusion and sorption. The simulations were done with the reactive transport code CRUNCH. In addition, results obtained from through-diffusion experiments on small-sized samples with HTO, 36Cl- and 22Na+ are presented and compared with the in situ data. In all cases, excellent agreement between the two data sets exists. Results for Cs+ indicated five times higher diffusion rates relative to HTO. Corresponding laboratory diffusion measurements are still lacking. However, our Cs+ data are in qualitative agreement with through-diffusion data for Callovo–Oxfordian argillite rock samples, which also indicate significantly higher effective diffusivities for Cs+ relative to HTO.

How Mobile Are Sorbed Cations in Clays and Clay Rocks

Diffusion of cations and other contaminants through clays is of central interest, because clays and clay rocks are widely considered as barrier materials for waste disposal sites. An intriguing experimental observation has been made in this context: Often, the diffusive flux of cations at trace concentrations is much larger and the retardation smaller than expected based on their sorption coefficients. So-called surface diffusion of sorbed cations has been invoked to explain the observations but remains a controversial issue. Moreover, the corresponding surface diffusion coefficients are largely unknown. Here we show that, by an appropriate scaling, published diffusion data covering a broad range of cations, clays, and chemical conditions can all be modeled satisfactorily by a surface diffusion model. The average mobility of sorbed cations seems to be primarily an intrinsic property of each cation that follows inversely its sorption affinity. With these surface mobilities, cation diffusion coefficients can now be estimated from those of water tracers. In pure clays at low salinities, surface diffusion can reduce the cation retardation by a factor of more than 1000.

Combined effects of heterogeneity, anisotropy, and saturation on steady state flow and transport: A laboratory sand tank experiment

Field soils show rather different spreading behavior at different water saturations, frequently caused by layering of the soil material. We performed tracer experiments in a laboratory sand tank. Such experiments complement and help comprehension of field investigations. We estimated, by image analysis, the first two moments of small plumes traveling through a two-dimensional, heterogeneous medium with strongly anisotropic correlation structure. Three steady state regimes were analyzed. Two main conclusions were drawn. First, low saturation led to very large heterogeneity and to strong preferential flow. Thus the description of the flow paths and the prediction of the solute arrival times require, in this case, more accurate knowledge about the topological structure. Second, saturation-dependent macroscopic anisotropy is an essential element of transport in unsaturated media. For this reason, small structural soil features should be properly upscaled to give appropriate effective soil parameters to be input in transport models.

Dilution of non-reactive tracers in variably saturated sandy structures

Based on a dye tracer experiment in a sand tank we addressed the problem of local dispersion of conservative tracers in the unsaturated zone. The sand bedding was designed to have a defined spatial heterogeneity with a strong anisotropy. We estimated the parameters that characterize the local dispersion and dilution from concentration maps of a high spatial and temporal resolution obtained by image analysis. The plume spreading and mixing behavior was quantified on the basis of the coefficient of variation of the concentration and of the dilution index. The heterogeneous structure modified the flow pattern depending on water saturation. The shape of the tracer plumes revealed the structural signature of the sand bedding at low saturation only. In this case pronounced preferential flow was observed. At higher flow rates the structure remained hidden by a spatially almost homogeneous behavior of the plumes. In this context, we mainly discuss the mechanism of re-distributing a finite mass of inert solutes over a large volume, due to macro- and micro-heterogeneities of the structure.

Upscaling of anisotropy in unsaturated Miller-similar porous media

Geological and pedological processes rarely form isotropic media as is usually assumed in transport studies. Anisotropy at the Darcy or field scale may be detected directly by measuring flow parameters or may become indirectly evident from movement and shape of solute plumes. Anisotropic behavior of a soil at one scale may, in many cases, be related to the presence of lower-scale directional structures. Miller similitude with different pore-scale geometries of the basic element is used to model macroscopic flow and transport behavior. Analytical expressions for the anisotropic conductivity tensor are derived based on the dynamic law that governs the flow problem at the pore scale. The effects of anisotropy on transport parameters are estimated by numerical modeling.

Transport of volatile chlorinated hydrocarbons in unsaturated aggregated media

Transport of volatile hydrocarbons in soils is largely controlled by interactions of vapours with the liquid and solid phase. Sorption on solids of gaseous or dissolved compounds may be important. Since the contact time between a chemical and a specific sorption site can be rather short, kinetic or mass-transfer resistance effects may be relevant.An existing mathematical model describing advection and diffusion in the gas phase and diffusional transport from the gaseous phase into an intra-aggregate water phase is modified to include linear kinetic sorption on gas-solid and water-solid interfaces. The model accounts for kinetic mass transfer between all three phases in a soil. The solution of the Laplace-transformed equations is inverted numerically.We performed transient column experiments with 1,1,2-Trichloroethane, Trichloroethylene, and Tetrachloroethylene using air-dry solid and water-saturated porous glass beads. The breakthrough curves were calculated based on independently estimated parameters. The model calculations agree well with experimental data. The different transport behaviour of the three compounds in our system primarily depends on Henrys constants.

Translational diffusion of water and its dependence on temperature in charged and uncharged clays: A neutron scattering study

The water diffusion in four different, highly compacted clays [montmorillonite in the Na- and Ca-forms, illite in the Na- and Ca-forms, kaolinite, and pyrophyllite (bulk dry density rho(b)=1.85+/-0.05 gcm(3))] was studied at the atomic level by means of quasielastic neutron scattering. The experiments were performed on two time-of-flight spectrometers and at three different energy resolutions [FOCUS at SINQ, PSI (3.65 and 5.75 A), and TOFTOF at FRM II (10 A)] for reliable data analysis and at temperatures between 27 and 95 degrees C. Two different jump diffusion models were used to describe the translational motion. Both models describe the data equally well and give the following ranking of diffusion coefficients: Na-montmorillonite<or=Ca-montmorilloniteor=Na-montmorillonite>Ca-illite>Na-illite>or=kaolinite>pyrophyllite>or=water, in both jump diffusion models. For clays with a permanent layer charge (montmorillonite and illite) a reduction in the water content by a factor of 2 resulted in a decrease in the self-diffusion coefficients and an increase in the time between jumps as compared to the full saturation. The uncharged clay kaolinite exhibited no change in the water mobility between the two hydration states. The rotational relaxation time of water was affected by the charged clay surfaces, especially in the case of montmorillonite; the uncharged clays presented a waterlike behavior. The activation energies for translational diffusion were calculated from the Arrhenius law, which adequately describes the systems in the studied temperature range. Na- and Ca-montmorillonite (approximately 11-12 kJmol), Na-illite (approximately 13 kJmol), kaolinite and pyrophyllite (approximately 14 kJmol), and Ca-illite (approximately 15 kJmol) all had lower activation energies than bulk water (approximately 17 kJmol in this study). This may originate from the reduced number and strength of the H-bonds between water and the clay surfaces, or ions, as compared to those in bulk water. Our comparative study suggests that the compensating cations in swelling clays have only a minor effect on the water diffusion rates at these high densities, whereas these cations influence the water motion in non-swelling clays.

FIELD-SCALE WATER TRANSPORT IN UNSATURATED CRYSTALLINE ROCK

Safe disposal of toxic wastes in geologic formations requires minimal water and gas movement in the vicinity of storage areas, Ventilation of repository tunnels or caverns built in solid rock can desaturate the near field up to a distance of meters from the rock surface, even when the surrounding geological formation is saturated and under hydrostatic pressures. A tunnel segment at the Grimsel test site located in the Aare granite of the Bernese Alps (central Switzerland) has been subjected to a resaturation and, subsequently, to a controlled desaturation, Using thermocouple psychrometers (TP) and time domain reflectometry (TDR), the water potentials psi and water contents theta were measured within the unsaturated granodiorite matrix near the tunnel wall at depths between 0 and 160 cm. During the resaturation the water potentials in the first 30 cm from the rock surface changed within weeks from values of less than -1.5 MPa to near saturation. They returned to the negative initial values during desaturation, The dynamics of this saturation-desaturation regime could be monitored very sensitively using the thermocouple psychrometers, The TDR measurements indicated that water contents changed dose to the surface, but at deeper installation depths the observed changes were within the experimental noise. The field-measured data of the desaturation cycle were used to test the predictive capabilities of the hydraulic parameter functions that were derived from the water retention characteristics psi(theta) determined in the laboratory. A depth-invariant saturated hydraulic conductivity k(s) = 3.0 x 10(-11) m s(-1) was estimated from the psi(t) data at all measurement depths, using the one-dimensional, unsaturated water flow and transport model HYDRUS Vogel er al., 1996, For individual measurement depths, the estimated k(s) varied between 9.8 x 10(-12) and 6.1 x 10(-11) The fitted k(s) values fell within the range of previously estimated k(s) for this location and led to a satisfactory description of the data, even though the model did not include transport of water vapor.

Flux and Resident Injection in Gaseous Advection Experiments

The occurrence of gaseous pollutants in soils has stimulated many experimental activities, including forced ventilation in the field as well as laboratory transport experiments with gases. The dispersion coefficient in advective-dispersive gas phase transport is often dominated by molecular diffusion, which leads to a large overall dispersivity gamma. Under such conditions it is important to distinguish between flux and resident modes of solute injection and detection. The influence of the inlet type oil the macroscopic injection mode was tested in two series of column experiments with gases at different mean flow velocities nu. First we compared infinite resident and flux injections, and second, semi-infinite resident and flux injections. It is shown that the macroscopically apparent injection condition depends on the geometry of the inlet section. A reduction of the cross-sectional area of the inlet relative to that of the column is very effective in excluding the diffusive solute input, thus allowing us to use the solutions for a flux Injection also at rather low mean flow velocities nu. If the whole cross section of a column is exposed to a large reservoir like that of ambient air, a semi-infinite resident injection is established, which can be distinguished from a flux injection even at relatively high velocities nu, depending on the mechanical dispersivity of the porous medium.

Mixing-cell boundary conditions and apparent mass balance errors for advective-dispersive solute transport

In many field or laboratory situations, well-mixed reservoirs like, for instance, injection or detection wells and gas distribution or sampling chambers define boundaries of transport domains. Exchange of solutes or gases across such boundaries can occur through advective or diffusive processes. First we analyzed situations, where the inlet region consists of a well-mixed reservoir, in a systematic way by interpreting them in terms of injection type. Second, we discussed the mass balance errors that seem to appear in case of resident injections. Mixing cells (MC) can be coupled mathematically in different ways to a domain where advective-dispersive transport occurs: by assuming a continuous solute flux at the interface (flux injection, MC-FI), or by assuming a continuous resident concentration (resident injection). In the latter case, the flux leaving the mixing cell can be defined in two ways: either as the value when the interface is approached from the mixing-cell side (MC-RT -), or as the value when it is approached from the column side (MC-RT +). Solutions of these injection types with constant or-in one case-distance-dependent transport parameters were compared to each other as well as to a solution of a two-layer system, where the first layer was characterized by a large dispersion coefficient. These solutions differ mainly at small Peclet numbers. For most real situations, the model for resident injection MC-RI + is considered to be relevant. This type of injection was modeled with a constant or with an exponentially varying dispersion coefficient within the porous medium. A constant dispersion coefficient will be appropriate for gases because of the Eulerian nature of the usually dominating gaseous diffusion coefficient, whereas the asymptotically growing dispersion coefficient will be more appropriate for solutes due to the Lagrangian nature of mechanical dispersion, which evolves only with the fluid flow. Assuming a continuous resident concentration at the interface between a mixing cell and a column, as in case of the MC-RI + model, entails a flux discontinuity. This flux discontinuity arises inherently from the definition of a mixing cell: the mixing process is included in the balance equation, but does not appear in the description of the flux through the mixing cell. There, only convection appears because of the homogeneous concentration within the mixing cell. Thus, the solute flux through a mixing cell in close contact with a transport domain is generally underestimated. This leads to (apparent) mass balance errors, which are often reported for similar situations and erroneously used to judge the validity of such models. Finally, the mixing cell model MC-RI + defines a universal basis regarding the type of solute injection at a boundary. Depending on the mixing cell parameters, it represents, in its limits, flux as well as resident injections