Key points are not available for this paper at this time.
The secure disposal of nuclear waste is of high societal concern, necessitating the development of deep geological repositories as a reliable solution. A key aspect of repository safety lies in understanding the far field, particularly the host rock, to predict the long-term behavior and migration of radionuclides within the geological environment from the deposit up to the ecosphere. This study addresses the specific challenges associated with crystalline host rocks. Crystalline host rocks could be on the one hand of granitic composition and texture, but the term is also used for host rocks of metamorphic origin. While inside a large granitic intrusion there is little petrological variation expected, metamorphic rocks or the intrusion rim can exhibit complex structures in terms of structural geology as well as mineral composition, especially along potential fluid migration pathways. Consequently, this leads to a multitude of possible rock-composition/fluid-composition interactions and thus significantly affects the retention potential of radionuclides as opposed to the simplified model of an isotropic, uniform granite. The results of the study will allow to determine which components of the host rock are important to be included in geostatistical models which in turn serve as basis to estimate uncertainties of reactive transport through crystalline rocks. Our study involves the development of Python code to feed chemical modelling software like PHREEQC or Geochemists Workbench with varying mineral compositions and chemical conditions of the aqueous phase, following a specific Quasi-Monte-Carlo sampling scheme. The application of compositional data analysis principles is essential to guarantee a meaningful sampling of constraint concentration data, such as mineralogical rock compositions or element concentrations in aqueous phases. Given that compositional data sum to a fixed total, each mineral content becomes a dependent variable in relation to the other contents. Recognizing and accounting for these interdependencies is crucial to ensuring the integrity of the sampling. The chemical modelling software relies on Surface Complexation Models (SCM) for each mineral phase to calculate the distribution coefficient (Kd-value) for the radionuclide (here: uranium) in the respective setting. Furthermore, a global sensitivity analysis is employed to investigate the complex interactions between mineralogical variations and radionuclide behavior. In this study, two techniques are employed namely, High-Dimensional Model Representation (HDMR) and Cumulative Sum of Univariate Nonlinear Regression (CUSUNORO) plots. The application of HDMR allows for a detailed investigation of high-dimensional parameter spaces, while CUSUNORO plots provide a visual representation of cumulative sensitivity effects. This study presents a complete workflow of modelling how petrological variations in crystalline host rocks, including both granitic and metamorphic compositions, affects radionuclide retention. This approach advances the understanding of nuclear waste disposal and provides valuable tools for assessing the retention potential of radionuclides in diverse geological settings.
Pospiech et al. (Fri,) studied this question.