NUREG/CR-6823
SAND2003-3348P
Sandia National Laboratories
U.S. Nuclear Regulatory Commission
Office of Nuclear Regulatory Research
Washington, DC 20555-0001
2003
Prepared by:
C.L. Atwood JL. LaChance, H.F. Martz, D.L Anderson, M. Englehardte, D. Whitehead, T. Wheeler
Probabilistic risk assessment (PRA) is a mature technology that can provide a quantitative assessment of the risk from accidents in nuclear power plants. It involves the development of models that delineate the response of systems and operators to accident initiating events. Additional models are generated to identify the component failure modes required to cause the accident mitigating systems to fail. Each component failure mode is represented as an individual "basic event" in the systems models. Estimates of risk are obtained by propagating the uncertainty distributions for each of the parameters through the PRA models. The data analysis portion of a nuclear power plant PRA provides estimates of the parameters used to determine the frequencies and probabilities of the various events modeled in a PRA. This handbook provides guidance on sources of information and methods for estimating the parameters used in PRA models and for quantifying the uncertainties in the estimates. This includes determination of both plant-specific and generic estimates for initiating event frequencies, component failure rates and unavailabilities, and equipment non-recovery probabilities.