Diffusion Databases for Opalinus Clay, Confining Geological Units and Bentonite: Methods, Concepts and Upscaling of Data

  pdf NTB 23-08 Diffusion Databases for Opalinus Clay, Confining Geological Units and Bentonite: Methods, Concepts and Upscaling of Data(15.42 MB)

Safety analysis for the deep geological disposal of radioactive waste in consolidated sedimentary rocks relies on the use of appropriate parameter values for diffusion and retention properties of dose-relevant radionuclides. The effective diffusion coefficients (D_e) and accessible porosity (e_acc) values for radionuclides are among the most important parameters for a quantitative description of transport processes in clay-rich host rocks for which advective and dispersive transport is assumed to be negligible.

This report describes the methods used for the compilation of diffusion databases (DDB) of element-specific D_e and e_acc values for the Mesozoic rocks in Northern Switzerland as well as compacted bentonite using the appropriate reference porewater compositions. The methods are based on the procedures applied in previous safety analysis but aim at a generic implementation of electrostatic effects of the diffusion of charged species in charged porous media. Such effects are described here using the so-called «Mean-Potential Donnan Layer» (MPDL) approach to quantify the equilibrium distribution of charged species between the bulk porewater and the Donnan phase (a defined volume of charged solution near the charged rock surfaces).

Recent empirical relationships are used here to derive the total porosity and the geometrical factor as a function of the total clay content in rock. These empirical relationships produce narrower uncertainty bands in the prediction of D_e of uncharged tracers in clay rock and are more broadly applicable compared to the extended Archie’s relationship applied in previous safety analysis. For bentonite, the extended Archie’s relationship is still used here because the target bulk dry density of bentonite – and thus the total porosity – is predefined.

The empirical geometrical relationships and the MPDL approach are implemented in the so-called ClaySorDif model (as an extension of the existing ClaySor sorption model) realised in a Python/Jupyter notebook environment based on GEMS codes. The model calibration uses the results of an extensive series of diffusion measurements on rock samples obtained from Nagra’s deep drilling campaign in Northern Switzerland. The use of such a consistent tool and data chain for the derivation of solubility, sorption and diffusion data is thus unique in the literature. The DDBs are generated as look-up tables of diffusion parameters for a range of the total clay content in rock and the selected reference porewaters as the primary input variables. Their applicability is restricted to a temperature of 25 °C and the diffusion direction perpendicular to the bedding. The same procedures as in previous safety analysis are proposed here to consider the effects on diffusion of temperature and the orientation of the clay particles. A comprehensive comparison exercise shows that the results produced by the MPDL model based on GEMS produce almost the same results as a previous implementation of the electrostatic model in PHREEQC. Both models differ slightly in their approaches to quantifying the excess negative charge acting on the Donnan equilibrium distribution of charged species.

The D_e and e_acc values are inherently linked to the speciation of a given radioelement. Low D_e and e_acc values result for elements that are predominantly present as anionic species, which is the case e.g. for Np or Th. The highest values for D_e are found for elements predominantly present as cationic species, such as Cs⁺, alkaline earth metal cations (e.g. Sr²⁺, Ra²⁺) or the trivalent actinides (e.g. Am³⁺). Halogenides (e.g. I^–) generally exhibit rather low D_e and e_acc values.

The predictions by ClaySorDif are compared with the predictions made in previous safety analysis. The results for sediments with moderate to high clay contents do not differ significantly. Only for more calcareous rocks, for which the geometrical relationships are known to be less applicable, are discrepancies noted in some cases. It should also be emphasised that the scope of the previous diffusion models with respect to surface diffusion was limited to radioelements undergoing cation exchange processes. Anion exclusion effects were assumed to be identical for all types of anions, regardless of their charge numbers. The present model is free of such restrictions. It accounts for the full speciation of a given element, irrespective of its redox state, which facilitates the compilation of diffusion data for any desired geochemical constellation of rock and porewater composition without the need to make subjective decisions.

The D_e and e_acc values in the look-up tables computed in ClaySorDif are linked to geophysical borehole log data and produce highly resolved borehole depth profiles of D_e and e_acc, as shown in a dedicated chapter of this report. The profiles calculated are further processed in statistical approaches to estimate not only representative mean values for given abstracted rock units used in transport models for safety and performance analysis, but also their ranges of uncertainty and variability.