Long term redox evolution in granitic rocks: Modelling the redox front propagation in the rock matrix

Advanced plans for the construction of a deep geological repository for highly radioactive wastes from nuclear power plants have evolved during the past decades in many countries including Sweden. As part of the Swedish concept, the waste is to be encapsulated in canisters surrounded by low permeability backfill material. The copper canisters will be deposited at around 500 metres depth in granitic rock, which acts as a natural barrier for the transport of radionuclides to the ground surface. These natural and engineered barriers are chosen and designed to ensure the safety of the repository over hundred of thousands of years. One issue of interest for the safety assessment of such a repository is the redox evolution over long times. An oxidising environment would enhance the corrosion of the copper canisters, and increases the mobility of any released radionuclides. In the first part of the present thesis, the ability of the host rock to ensure a reducing environment at repository depth over long times was studied. A model framework was developed with the aim to capture all processes that are deemed to be important for the scavenging of intruding oxygen from the ground surface over long times. Simplifications allowing for analytical solutions were introduced for transparency reasons so that evaluation of results is straight-forward, and so that uncertain parameter values easily can be adjusted. More complex systems were solved numerically for cases when the analytical simplifications are not applicable, and to validate the simplifications underlying the analytical solutions. Results were presented for prevailing present day conditions as well as for conditions deemed to be likely during the melting phase of a period of glaciation. It was shown that the hydraulic properties have a great influence on the oxygen intrusion length downstream along flow-paths in the rock. An important parameter that determines the extent of interaction between the dissolved oxygen and the reducing minerals in the rock was shown to be the flow-wetted surface to flow-rate ratio. The results show that for an initial period of time, depending on the amount of reducing minerals and reaction rates, chemical reaction kinetics may control the rate of the overall depletion of oxygen. For longer times, internal diffusion resistance in large particles or in the rock matrix become rate limiting for the overall process. It was found that there are many uncertainties that have to be considered in order to make reliable quantitative predictions on the extent of oxygen intrusion. In the second part of the thesis, the impact of intruding oxygen on the corrosion of the copper canisters was explored. Also, a mechanism for the production of sulphide close to the deposition holes was studied. Sulphide is another corroding agent that may be produced microbially in a reducing environment from sulphate in the presence of organic reductants such as methane. From calculation results it was found that corrosion of more than 50 kg of copper is not likely over a period of one million years