This paper presents a snapshot of how numerical modelling of groundwater is being used in the interpretation of hydrogeological data for the proposed Sellafield radioactive waste repository. The main objective of the modelling at the present stage of the investigation is to help develop a conceptual understanding of the site hydrogeology, by testing whether particular concepts are quantitatively feasible and whether they are able to explain observations.
Models have been constructed with differing levels of detail: they each contribute to building a consistent conceptual model of the behaviour of the site as a whole, but they do not individually provide such an integrated conceptual model. They are presented here to illustrate the approach to developing the conceptual understanding, and to illustrate the range of quantitative approaches being applied to different aspects of the site hydrogeology.
Four groups of models are described in outline, incorporating variously porous medium and fracture flow, complex geology, anisotropic formations, and dense solutes from more than one source. These are modelled in two and three dimensions.
A two-dimensional areal model reproduces head variation observed in shallow wells penetrating the Sherwood Sandstone Group, identifying recharge and discharge areas, and constraining the hydraulic conductivity of the sandstone. Vertical section models through the geological sequence, incorporating one or two sources of dense saline water were able to reproduce qualitatively the main features observed in deep boreholes, in terms of pressure and salinity. The two-source model produces the better match to the salinity distribution.
A three-dimensional model, using simplified geology, was able to demonstrate the interaction of salinity and geology, and demonstrate how pressures and salinity may evolve with time. Three-dimensional fracture network models of the Borrowdale Volcanic Group show head variability with depth similar to that observed in some deep boreholes. A number of fracture concepts and parameter distributions are compatible with the data. The modelling work presented shows progress towards a numerically consistent conceptual hydrogeological model of the Sellafield region that can accommodate the site characterization data, and further work is under way to continue this progress.