Quantitative safety assessments for geologic disposal comprise an assessment of the overall level of performance of a repository and an analysis of the associated uncertainties. Such assessments rely on system-level models that are applied to predict the behavior of the whole system under consideration, which in turn are supported by process-driven submodels that assess individual components of the repository system. Many factors defining the evolution of a repository involve complex processes that are quite challenging to predict over short and long time scales, thus demonstrating model fidelity and accounting for prediction uncertainties is an unavoidable aspect of a safety case. Furthermore, results from uncertainty analyses need to be formulated in a clear way understandable to the public and other stakeholders.
This session seeks contributions on how model uncertainties of different nature - scenario uncertainty, conceptual and numerical model uncertainty, and parameter uncertainty – can be evaluated and potentially reduced, how they can be best incorporated in both system and submodels to avoid error propagation, and how uncertainty evaluations can be best communicated to achieve and maintain acceptance. Possible topics include:
• Model comparison against experimental data from lab- and field-based studies as well as natural analogs
• Benchmarking and model comparison studies
• Uncertainty quantification via probabilistic and deterministic approaches
• Transferring information from submodels to system-level models: Upscaling, simplifications, and abstractions
• Propagation of uncertainty between submodels and system-level models
• Assessing uncertainty over very long time frames
• Communicating model uncertainties to the public and other stakeholders
• Case studies from national and international programs and initiatives