Numerical modeling of large-scale solid-source diffusion experiments in Callovo-Oxfordian clay

Abstract Diffusion and retention parameters of clays are commonly derived from in situ diffusion experiments which are performed by injecting dissolved tracers in packed sections of boreholes drilled in clay formations. Interpretation of these experiments is prone to uncertainties caused by the filter, the gap and the drilling-disturbed zone. The large-scale solid-source diffusion experiment (SSDE) overcomes the uncertainties of in situ diffusion experiments because tracers are emplaced in a clay plug at the center of a cylindrical clay block (30 cm high and 30 cm in diameter). Sampling of the clay block along vertical boreholes at the end of the experiment provides a 3-D tracer distribution from which the effective diffusion, D e , and K d are estimated. Numerical models of SSDE performed in samples of Callovo-Oxfordian clay with HTO and 85 Sr 2+ are presented here. Experiments are modelled with 2-D axi-symmetric finite element grids because clay bedding is parallel to the bases of the blocks. The model takes into account diffusion anisotropy and is solved with CORE 2D V4. Relevant diffusion and retention parameters are identified by sensitivity analyses. Parameters are estimated by solving the inverse problem. Uncertainties in parameters of the clay plug do not affect the estimation of clay parameters because tracer activities are not sensitive to clay plug parameters. Tritium data show axial symmetry with mild fluctuations and allow the accurate estimation of horizontal D e which ranges from 4.04 × 10 −11 to 4.51 × 10 −11  m 2 /s. Vertical D e is smaller than horizontal D e by a factor (anisotropy ratio) which ranges from 1.79 to 3.8. 85 Sr 2+ data show a large scatter and provide only crude estimates of K d and horizontal D e .