Abstract:
Substantial understanding of the processes involved in fluid flow and tracer transport in geological systems can be gained from the use of numerical simulations. Regardless of the nature or scale of a geological system, the probability of encountering fractures in hard rock units is high. Therefore, modelling flow and transport in fractures and fracture networks is a key component of almost any complex hydrogeological simulation. Over the last few decades, with the increase in computational power and the availability of increasingly sophisticated and powerful specialized modelling software, numerical models have become increasingly commonplace, and are meanwhile regarded as vital tools in the wide field of geoscience.
This work portrays the development of a series of numerical models, with discrete fracture network geometry, based on the knowledge of an actual fracture network gained through prior experiments performed at the Grimsel Test Site (GTS) in the Swiss Alps. A three dimensional reconstruction of the actual fracture network geometry is developed from the extensive photographic data set provided by the Excavation Project (EP) and used as the basis for purely advective particle tracking models, as well as for fully advective-dispersive transport models. Realistic fracture apertures based on actual fracture measurements are generated through a geostatistical method and their influence upon flow and transport is studied. Various degrees of heterogeneity with regard to aperture distributions are generated and used in a Monte Carlo approach to examine their influence. Further, the retardation effects of matrix diffusion are also considered and incorporated into the models by means of a new semi-analytical method included in the FEM code Rockflow/Geosys V4. The resulting breakthrough curves are analysed and fitted with several analytical solutions, including advective dispersive transport with matrix diffusion and continuous time random walks (CTRW). It is shown that the standard advection dispersion equation (ADE) is not suitable for approximating or predicting breakthrough curves from heterogeneous models.
The development of the various models portrayed herein depended strongly on the ongoing advances made in the available modelling codes and in new and dedicated preprocessing tools. Several newly developed methods and tools were tested and applied during the project.
For the first time, a group of numerical simulations is presented, that incorporates complex fracture network geometry based on actual field data, geostatistically generated realistic apertures and the effect of matrix diffusion. This new level of realism provides the basis for new insights into the role of heterogeneity, matrix diffusion and the combination of both in the transition toward anomalous transport.