Thursday, July 28, 2011

NRC Briefing: Severe Accidents and Level 3 PRA

The staff of the US Nuclear Regulatory Commission today held a briefing for the Commissioners on Severe Accidents, and Options for proceeding with Probabilistic Risk Asessment - Level 3 (PRA Level 3).

Traditionally, PRA/PSA Level 3 has not been a strong regulatory requirement because the results of Level 1 (usually the core damage frequency, CDF) and the results of Level 2 (large early release fraction/frequency, LERF) can be used as surrogate proxies for the types of Risk Metrics that a Level 3 PSA/PRA might generate, which could include the following: the number of early fatalities; the number of early injuries, the number of latent cancer fatalities, or the total population dose at different locations; as well as the individual early and latent fatality risk, and the economic cost of mitigation actions taken following a severe accident. The CDF, for example, can proxy for the latent cancer risk, while the LERF could proxy for the prompt fatality risk.

However, a number of potential benefits are foreseen for a full PRA Level 3 analysis, including feedback into risk-informed regulatory guidance for new reactors and use of risk insights in forthcoming SMR design reviews. In addition, capabilities such as modeling of radionuclide aqueous dispersion modes, and multi-unit risk assessment could also be addressed. The modeling of radionuclide dispersal in the event of a severe accident which is initiated by an external event such as an earthquake, tsunami or hurricane also calls for additional modeling efforts, since meteorological variables such as windspeed & direction, ambient precipitation and humidity may not correspond to what is normally expected for that site at that time of year (for example).

NRC Staff plan to use an existing SPAR (Standardized Plant Assessment Risk) model as the basis for proceeding to PSA Level 3. The SPAR model is essentially a plant-specific PSA/PRA Level 1 designed to incorporate both external and internal initiating events, recent modifications of which include capabilities to yield LERFs. Since external hazards are site-specific, much greater value can be expected to be derived if the SPAR model selected for development to PRA Level 3 is for a NPP site that is either representative of the entire population of NPPs, or, has a larger than average number and type of external hazards.