Technical Report NTB 04-01

Grimsel Test SiteInvestigation Phase VModelling the Transport of Solutes and Colloids in a Water-Conducting Shear Zone in the Grimsel Test Site

This report describes modelling of the transport of solutes and colloids in an experimental system comprising an artificial dipole flow field in a water-conducting shear zone at Nagra's Grimsel Test Site (GTS) in the central Swiss Alps. The modelling work forms part of the Colloid and Radionuclide Retardation Project (CRR), which includes a series of field transport experiments and a supporting laboratory programme, as well as modelling studies. Four independent groups representing different organisations or research institutes have conducted the modelling, with each group employing its own modelling approach or approaches. Only the work conducted at the Paul Scherrer Institute (PSI) is described in the present report.

Bentonite, which is widely considered as a potential backfill material for a range of radioactive and chemotoxic wastes, could conceivably provide a source of colloids that could then influence the transport of radionuclides released from a geological repository for radioactive waste. The main objective of CRR is to enhance understanding of the in situ retardation of radionuclides in the presence of bentonite colloids, in a system analogous to the near-field/geosphere interface of a geological repository.

The field transport experiments were carried out by injecting various cocktails of tracers, some of which included bentonite colloids, into the injection borehole of the dipole and measuring the resulting breakthrough curves. Modelling work was carried out in order to assist in the planning of the main experimental runs and to contribute to the interpretation of the results. Three model variants are used in the present study, namely a 1-D advection-dispersion model, similar to that developed in support of the earlier GTS Migration Experiment (MI), a 2-D advection-dispersion model, and a non-Fickian dispersion model: the CTRW (continuous time random walk) model. The 1-D and 2-D models treat dispersion as a diffusion-like process that obeys Fick's laws. They also include the retardation mechanisms of matrix diffusion of solutes and solute sorption on matrix pore surfaces. Colloids are excluded from matrix pores in all the model variants. The CTRW model allows a more general treatment of dispersion, but does not currently include matrix diffusion, and so was only applied to the transport of colloids.

The modelling of preliminary tests carried out in advance of the main CRR experimental runs showed that the 1-D and 2-D advection-dispersion models with matrix diffusion provide similarly good fits for tracers conveyed as aqueous species, using reasonable and consistent sets of parameter values. They were less successful at modelling colloid breakthrough, and various explanations for this have been considered. Of these, the occurrence of non-Fickian dispersion is considered the most likely. The CTRW model, which allows for non-Fickian dispersion, indeed provides an adequate fit in the case of colloids with a consistent set of parameters.

On the basis of the modelling of the preliminary tests, predictions of the breakthrough of Am, Pu, Np, U and Cs, both with and without the addition of bentonite colloids to the injection cocktail, were made for the main experimental runs in advance of the experiments being carried out. The experimental measurements confirm the model assumption that at least part of the injected inventories of Am, Cs, Pu and Th migrates in association with bentonite colloids. Furthermore, discrepancies between predictions and measurements that Am, Pu and Th are transported in colloidal form, even when no bentonite colloids are added to the injection cocktail. The addition of bentonite colloids, however, increases the recovery of these tracers. The characterisation of colloids in the injection cocktails, which was not available at the time that the model predictions were made, enables improved agreement to be obtained between model calculations and measured breakthrough curves.

The CRR experiment and the present modelling study have a number of limitations. For example, there is the possibility that non-Fickian dispersion affects the transport of solutes as well as colloids. It is not, however, possible to discriminate between the impact of this non- Fickian dispersion and matrix diffusion effects by modelling the breakthrough curves. If non- Fickian dispersion of solutes takes place, then this has implications for the derivation of parameter values for safety assessment (and especially sorption coefficients) from field tracer transport experiments. In particular, values derived using advection-dispersion models with matrix diffusion and with dispersion modelled using Fick's laws need to be viewed with caution.

The modelling approaches used in the present study may not be directly applicable to safety assessment problems and the direct implications of the results of this study for safety assessment are limited. It can, however, be said that the study has demonstrated the high degree of mobility of bentonite and other colloids in a system that is at least in some ways comparable to those of interest in safety assessment, and has shown that bentonite colloids can at least potentially affect the transport of some safety relevant radionuclides over longer temporal and spatial scales than those addressed here.