Numerical modelling of fluid flow and particle transport in rough rock fracture during shear

Sammanfattning: The effects of different shearing processes and sample sizes on the fluid flow anisotropy and its impact on particle transport process in rough rock fractures are significant factors that need to be considered in the performance and safety assessments of underground nuclear waste repositories. The subjects, however, have not been adequately investigated previously in either laboratory experiments or numerical modeling. This thesis addresses these problems using numerical modeling approaches. The modeling consists of two parts: 1) fluid flow simulations considering more complex but realistic flow boundary conditions during shear processes that cannot be realized readily in laboratory experiments, using digitalized fracture surfaces scanned in the laboratory, so that anisotropic fluid flow induced by shearing with channeling phenomenon can be directly simulated and quantified; 2) particle tracking simulations to demonstrate the impacts of such channeling effects on characteristic properties of particle transport. The numerical method chosen for the simulations is the Finite Element Method (FEM). Scale effects were considered in the simulations by using fracture surface samples of different sizes. The distributions of fracture aperture during shear were obtained by numerically generating relative translational and rotary movements between two digitalized surfaces of a rock fracture replica without considering normal loading. From the evolutions of the aperture distributions during the shearing processes, the evolutions of the transmissivity fields were determined by assuming the validity of the cubic law locally. A geostatistical approach was used to quantify the scale effects of the aperture and transmissivity fields. The fluid flow was simulated using different flow boundary conditions, corresponding to translational and rotary shear processes. Corresponding to translational shear (with a 1 mm shear displacement interval up to a maximum shear displacement of 20 mm), three different flow patterns, i.e., unidirectional (flow parallel with and perpendicular to the shear direction), bi-directional and radial, were taken into account. Corresponding to rotary shear (with a 0.5o shear angle interval up to 90o), only the radial flow pattern was considered. The particle transport was simulated using the Particle Tracking Method, with the particles motion following the fluid velocity fields during shear, as calculated by FEM. For the unidirectional particle transport, the breakthrough curves were analyzed by fitting to an analytical solution of 1-D advection-dispersion equation. The dispersivity, Péclet number and tracer velocity, as well as their evolutions during shear, were determined numerically. The results show that the fracture aperture increases anisotropically during translational shear, with a more pronounced increase in the direction perpendicular to the shear displacement, causing significant fluid flow channelling. A more significant increase of flow rate and decrease in travel time of the particles in the direction perpendicular to the shear direction is predicted. The particle travel time and characteristics are, correspondingly, much different when such effects caused by shear are included. This finding may have an important impact on the interpretation of the results of coupled hydro-mechanical and tracer experiments for measurements of hydraulic properties of rock fractures, because hydraulic properties are usually calculated from flow test results along the shear directions, with the effects of the significant anisotropic flow perpendicular to the shear direction ignored. The results also show that safety assessment of a nuclear repository, without considering the effects of stress/deformation of rocks on fluid flow and transport processes, may have significant risk potential. The results obtained from numerical simulations show that fluid flow through a single rough fracture changes with increasing sample size, indicating that representativehydro-mechanical properties of the fractures in the field can only be accurately determined using samples of representative sizes beyond their stationarity thresholds.