Nuclear Power Plant PRA: A Systems Modeling Approach For Addressing Rare Events

The U.S. Nuclear Regulatory Commission (NRC) and its licensees use “probabilistic risk assessment” (PRA) to identify and assess the risk associated with postulated low-probability, high-consequence events associated with the operation of U.S. nuclear power plants. In this application, PRA is performed as a systems modeling exercise; the overall model is comprised of elements representing possible hazards to the plants, the potential (unplanned as well as planned) responses of technical systems and plant operators to these hazards, and the potential consequences to the environment and surrounding population should the technical systems fail to perform as intended. Given the lack of severe accident data and the variability in key nuclear power plant design features, the model structure and parameter values are developed using a variety of sources of information, including severe accident precursor equipment reliability data, physical model predictions, and expert judgment. Special emphasis is placed on the identification and treatment of sources of probabilistic dependency. Recognizing the potential pitfalls associated with the modeling of rare events, the NRC uses the results of PRAs in a risk-informed environment. Thus, decisions are made considering traditional, qualitative attributes (e.g., effect on defense-in-depth) as well as quantitative risk measures.