Risk-informed safety margin characterization methods

The concept of "margin" has a long history in nuclear licensing and in the codification of good engineering practices. However, some regulatory applications of "margin" are based on surrogate scenarios (such as design basis scenarios), without regard to the actual frequencies of those scenarios, and have been carried out in a systematically conservative fashion that supports a regulatory finding of sufficiency of system performance, but is sub-optimal for plant decision-making. Beginning with the traditional discussion of "margin" in terms of a "load" (a physical challenge to system or component function) and a "capacity" (the capability of that system or component to accommodate the challenge), we are working to improve the capability to characterize probabilistic load and capacity spectra, reflecting both aleatory and epistemic uncertainty in system response. For example, the probabilistic load spectrum will reflect the frequency of challenges of a particular severity. Such a characterization is required if decision-making is to be informed optimally from a plant owner's point of view. Previous work has shown that developing probabilistic load and capacity spectra is possible using available tools, but difficult. Partly for that reason, the INL is developing a safety analysis approach that will significantly enhance the capability to do this kind of next generation probabilistic risk assessment. This paper will discuss technical challenges in multi-resolution system modeling and multi-physics simulation, and resulting evolution of requirements for a Risk Informed Safety Margins Characterization (RISMC) computational engine and the formulation of load and capacity metrics on which to base this kind of analysis.