Long-term Behavior Science: The cornerstone approach for reliably assessing the long-term performance of nuclear waste

Abstract High level waste (HLW) management requires being able to demonstrate the safety over geological timescales, typically 1 My. This can be made possible by using a rigorous, complex and iterative scientific approach called Long-term Behavior Science. The methodology relies on experiments and modelling. A large multi-scale approach is required and involves a mechanistic understanding of the key phenomena controlling the source term (i.e. the flux of radionuclides released from the waste as a function of time), as well as parametric studies, integrated and in situ tests. As a result, it is eventually possible to develop an operational model based on clever simplifications of a very complex reality, ensuring that predictions will always remain conservative despite conceptual and numerical uncertainties. Finally, predictive models must be validated based on the study of natural or archaeological analogues. In this paper, we show how this methodology can be applied by selecting examples of spent nuclear fuel and HLW glass.