Computer simulations are used to understand processes in nuclear waste disposal. The results are used to judge the safety of waste repository. Not all the information needed for such analyses, e.g. rock properties, is precisely known, contributing to uncertainty in the analysis results. We are interested in understanding the effect of the uncertainty of input quantities and of certain simplifications made during model creation on the outcome of computer simulations.

For the deep geological disposal of high-level nuclear waste in rock salt formations, the safety concept includes the backfilling of open cavities with crushed salt. For the prognosis of the sealing function of the backfill for the safe containment of the nuclear waste, it is crucial to have a comprehensive process understanding of the crushed-salt compaction behavior. The KOMPASS projects were initiated to improve the scientific knowledge of using crushed salt as backfill material.

Understanding complex systems such as radioactive waste repositories involves the study of cross-scale coupled processes. We discuss some important concepts and their mutual interactions for interpreting such systems based on complementary model-based analyses at various scales. One goal statement is to explain the formation of drying cracks. Near-field understanding can be used to determine how detailed repository far-field models must be and can lead to more robust analysis results.

The assessment of the integrity of the geological barrier for repositories for nuclear waste can be significantly influenced by uncertainties in the model inputs. Hence, the results of these established integrity analyses should be enriched by statistical information. In this contribution, we present preliminary probabilistic results for rock integrity assessment for the generic repository system in clay rock and the developed tools for stochastic analyses.