Abstract
This phD is in the frame of nuclear waste storage . The purpose is to test the ability of coupled chemical transport code to study the impact of porosity changes on solute transport, by designing clogging/dissolution experiments .To do this, experimental setups "through-diffusion" have been adapted to study the various porous media with increasing complexity: the compacted sand, glass frit , the chalk, sandstone close up to the materials used in the context of a storage, argillites Tournemire. Together with an inert tracer, diffusion experiment, precipitation reactions (calcium oxalate, gypsum or barite) or dissolution (acid attack chalk) were performed to assess their influence on the diffusive flux of these tracers. At the end of experience, the porous media and precipitates were characterized by MEB-EDS. Finally, this large data set was used to test one of the coupled chemical transport codes used in this phD, the Crunch code. The results from the first diffusion / clogging experiments performed through sand and glass frits show that the first experimental setup was not suited to such highly permeable porous media. The flux Curves are very noisy, and SEM observations do not reveal the existence of a precipitate front, as expected, but only new scattered phases in porous media. This is due to possible parasitic flow solution during sampling. In contrast, experiments through the chalk, have not been touched by these artifacts. In this case, it was clearly observed by SEM a precipitated front within the chalk, the effect on diffusive flux of water tracers (HTO and HDO) is more or less important depending on the type of precipitate. Thus, on the one hand, the cell with barite precipitation is impacted from the start by clogging , with a continued decrease of the HDO flux , which can be up to 40 times lower than the flux measured in the porous media without clogging. However, the cell with gypsum precipitation is affected much later by the clogging (70 days after the start of the experiment), and less markely, with flux values lowering a factor of 3 compared to those measured in healthy environment. All this tends to suggest that the efficiency of the clogging is more related to the nature of mineral than the precipitated amount, barite is probably denser and less porous than gypsum. In addition, after more than two months of experience, the HTO flux of the cell containing the Argilite Tournemire sample shows no impact related to the possible gypsum precipitation, which is consistent with the very low diffusion coefficients measured in this rock. In addition, for glass frits, there has been a slight effect of the barite precipitation on the flux. With regard to the simulation, the tests carried out with Crunch did not permit to reproduce the experimental results, particularly those from the diffusion / precipitation barite experiment. This can be explained by the way are considered kinetic laws in the code, but more probably to empirical formulation of the Archie law, used to link the evolution of diffusion coefficient with the porosity .