Technical Report NTB 94-21

INTRAVAL Finnsjon Test Modelling Results for some Tracer Experiments

This report presents the results obtained within Phase II of the INTRAVAL study.

Migration experiments performed at the Finnsjön test site were investigated. The study was done to gain an improved understanding of not only the mechanisms of tracer transport, but also the accuracy and limitations of the model used. The model is based on the concept of a dual porosity medium, taking into account one dimensional advection, longitudinal dispersion, sorption onto the fracture surfaces, diffusion into connected pores of the matrix rock, and sorption onto matrix surfaces. The number of independent water carrying zones, represented either as planar fractures or tube like veins, may be greater than one, and the sorption processes are described either by linear or non-linear Freundlich isotherms assuming instantaneous sorption equilibrium. The diffusion of the tracer out of the water-carrying zones into connected pore space of the adjacent rock is calculated perpendicular to the direction of the advective/dispersive flow. In the analysis, the fluid flow parameters are calibrated by the measured breakthrough curves for the conservative tracer (iodide). Subsequent fits to the experimental data for the two sorbing tracers strontium and caesium then involve element dependent parameters providing information on the sorption processes and on its representation in the model. The methodology of fixing all parameters except those for sorption with breakthrough curves for non-sorbing tracers generally worked well. The investigation clearly demonstrates the necessity of taking into account pump flow rate variations at both boundaries. If this is not done, reliable conclusions on transport mechanisms or geometrical factors can not be achieved. A two flow path model reproduces the measured data much better than a single flow path concept. But, the uniqueness of the resulting best-estimate values is lost when a second preferential flow path is introduced. However, though the mean values differ somewhat for different sets of best-estimate parameters, their ranges often overlap within one standard deviation. Although the transport is dominated by pure advection and dispersion and matrix diffusion has little influence on the tail of the breakthrough curve, matrix diffusion is nevertheless an important mechanism and cannot be neglected. Modelling strontium breakthrough yields best-fit values for the retardation factors and the sorption coefficients which are consistent with Kd values from static experiments. For caesium the best fits are obtained by taking into account non-linear Freundlich isotherms. The extracted values for the Freundlich coefficients are also comparable with those from independent batch sorption experiments. Comparative calculations show that the underlying geometry for flow transport is better modeled by planar fractures and not by tube like veins.