Derivation and application of a geologic dataset for flow modelling by discrete fracture networks in low-permeability argillaceous rocks

Argillaceous rock formations are targets of exploration for radioactive waste disposal sites in several countries. Groundwater flow in most indurated argillaceous rocks is very limited and occurs (if at all) mainly in brittle discontinuities, such as faults. On the basis of surface observations, core logging and hydraulic measurements in boreholes penetrating an argillaceous marl formation in the Swiss Alps, the relationships between internal fault architecture, larger-scale arrangement of the fault network and fault transmissivity are explored. Only a fraction of all faults observed in the cores correlates with water inflow points into the boreholes, and this is taken as evidence of variable transmissivity within each fault. Such flow channeling is also supported by geologic evidence. Stochastic discrete fracture network models are used for the upscaling of measured fault transmissivities, namely for the calculation of effective hydraulic conductivities (Keff) of model cubes with lengths of side of 50–500 m. Input data to these models include size, spacing, orientation, heterogeneity (flow channeling) and transmissivity of the faults. Fault size is not well known, but a sensitivity analysis shows that even the extreme assumption of infinite size yields Keff only less than 1 order of magnitude higher when compared to the base case. The effect of different arrangements of flow channels within each fault is also explored but has no appreciable effect on Keff. It is concluded that calculated values for Keff are robust because those input parameters that are least adequately known have only limited effects on the model results.