Parallel computing in probabilistic safety assessments of high-level nuclear waste

The influence of rock heterogeneity on contaminant migration is analysed in a Monte Carlo study of transport of Technetium in fractured media. The CRYSTAL3D model developed to address heterogeneity in Monte Carlo transport calculations has been used for this purpose. CPU and memory requirements of this model made it necessary to make use of high-performance computing. Therefore a Monte Carlo driver implemented on a parallel supercomputer has been used to monitor the CRYSTAL3D model. The results of a sensitivity analysis show that the impact of the geometry of the media on the consequence distribution, expressed in this model by the F-ratio number, is a very important parameter. It is also concluded that it is not possible to evaluate the importance of the wet surface area by intercomparison of case simulations with this parameter as the only feature that distinguishes those cases. Therefore, to be able to complete the study of the impact of heterogeneity of fractured media on the estimation of consequences from Monte Carlo calculations of radionuclide transport, it will be necessary to introduce in the parallel driver the ability to sample the wet surface area and the Darcy velocity parameters. (Author)