Inverse flow and transport modelling to understand contaminant transport experiments in a highly heterogeneous shear zone at the Grimsel Test Site in Switzerland

Abstract:

Hard fractured rock is a possible setting for a deep geological disposal facility for radioactive waste. In such rocks, fractures are likely to provide significant paths for flow and transport. Fracture systems within such rocks are often highly heterogeneous, and difficult to characterise. Data are drawn from a detailed program of field characterisation experiments in a fractured shear zone at the Grimsel Test Site. Tests include short- and long-duration cross-hole pumping tests, single well hydraulic tests and tracer tests between multiple closely located dipoles. Even with such detailed measurements, data do not allow direct geostatistical analysis of transmissivity variation across the shear zone. Previous geostatistical modelling found that transmissivity realisations based upon a wide variety of variograms were able to match cross-hole hydraulic tests. However, many of the realisations were unable to match conservative tracer tests. When calibration was possible, it was non-unique, and resulted in different parameters being needed for each transport experiment rather than determining properties of the shear zone. In contrast, we use inverse modelling to calibrate the 2D model, making simultaneous use of both flow and transport data. Fully distributed contaminant transport modelling is computationally very costly. However, at the scale of resolution of our model we account for the dispersion due to flow path variation explicitly by heterogeneity in the transmissivity. Simple metrics obtainable from particle tracking (25%, 75% and peak particle arrival times) are found to be a computationally inexpensive means to include key features of conservative transport observations as inverse model targets. Such an approach excludes transmissivity distributions that are able to match hydraulic testing but fail to reproduce the observed contaminant recovery. Whilst this gives a more consistent representation of flow and transport in the shear zone, data from field characterisation at different scales are inconsistent. We comment on the extent to which very detailed experimental characterisation of such heterogeneous shear zones is able to yield predictive transport models, and the implications for understanding the detail of radionuclide migration through such a hard fractured rock.