Seismic risk evaluation within the technology neutral framework

The NRC Office of Nuclear Regulatory Research has proposed a risk-informed and performance-based licensing process that is referred to as the technology neutral framework (TNF). In the TNF, licensing basis events (LBEs), determined using probabilistic risk assessment methods, take the place of design basis accidents. These LBEs are constructed by grouping together accident sequences with similar phenomenology. All event sequences with a mean frequency greater than 10−7 per reactor year are to be considered as part of the licensing basis. ng such a limit would require that earthquakes with a mean return period of ten million years be considered as part of the licensing basis. It is difficult to get seismic hazards (i.e., ground accelerations) from expert seismologists at such low frequencies. This is because it is difficult or impossible to confidently say what the seismic hazard might be at these extremely low frequencies. A linear extrapolation in log-log space of hazard curves at the Clinton site down to 10−7 per year leads to a peak ground acceleration of about 4.5 g. A Weibull distribution is also used to fit the curve leading to a peak ground acceleration of about 2.6 g. These extrapolations demonstrate the extreme nature of rare earthquakes. Even when seismic isolation is implemented, the TNF goal is not met. The problem appears to be that there is no limit on initiating event frequency in the TNF. Demonstrating that a design meets the goals of the TNF would be nearly impossible. uency limit for earthquakes could be imposed at a frequency of about 10−5 per year to focus on events that may be designed against. This type of initiating event frequency screening is consistent with current practice.