3D modeling of solute transport through natural fractures with microrough walls in crystalline host rocks: The effects of asperities and surface characteristics on hydrodynamics

Low-permeability granites are considered as host rocks for nuclear waste repositories. Understanding fluid flow and solute transport in granite fractures are essential in assessing the feasibility and safety of a nuclear waste repository. The internal variability of fractures, such as aperture distribution and asperities, dictates the hydrodynamics of reactive fluid, thus affecting the dispersion and retention of radionuclides. Numerical studies using 2-D models have focused on the heterogeneity of aperture distribution, but the effects of fracture asperities and additional surface features on the evolution of flow paths have not been systematically examined. In this study, the nonreactive solute transport behavior in a single fracture was numerically investigated considering the effects of fracture aperture and surface asperity by comparing 2.5-D and 3-D modeling results on a realistic fracture. The additional motivation here was to identify the limitations of model simplification. The 3-D fracture geometry was extracted from a micro-computed tomography of a natural fracture several centimeters long. Then, 2.5-D models were generated by mapping the aperture distribution of the 2-D fracture geometry on the x-y plane. Flow simulations were performed in both numerical models to detect the respective effects of fracture shape and surface asperities. For validation, we performed a sensitivity analysis by decreasing the 3-D fracture geometry mesh according to the quadric edge-collapse strategy, simulating the solute transport behavior under different fracture surface properties. The size variability of the isometric grid blocks ranges from 6.5 µm to 2.2 mm. Thus, we provide a function that can be used to quantitatively estimate the concentration error due to the simplification of the geometry mesh. The results show which fracture asperities and surface properties can significantly affect the solute transport behavior. Above a certain geometry complexity, the 3-D model results show less retention in the rather stagnant zones and thus better agreement with breakthrough curves (BTCs) of experiments compared to the 2.5-D model approaches. The results of the 3-D models also agree well with previous studies that less pronounced tailing is observed in the case of lower surface roughness. Simplifying the model geometry leads to more distorted results, with the 3-D model being more sensitive than the 2.5-D model. Moreover, based on a function summarized from the BTCs, the error in the simulated concentration due to mesh simplification can be estimated within a certain range that varies with the fracture geometry. The results presented show the capabilities and limitations of using 2.5D models in comparison with more elaborate 3-D models in predicting fluid dispersion in fractured crystalline rocks. Our study can serve as a guideline for the construction of fracture geometry and model design in future reactive transport modeling.