Improved nuclear power plant operations and safety through performance-based safety regulation

This paper illustrates some of the promise and needed future work for risk-informed, performance-based regulation (RIPBR). RIPBR is an evolving alternative to the current prescriptive method of nuclear safety regulation. Prescriptive regulation effectively constitutes a long, fragmented checklist of requirements that safety-related systems in a plant must satisfy. RIPBR, instead, concentrates upon satisfying negotiated performance goals and incentives for judging and rewarding licensee behavior to improve safety and reduce costs. In a project reported here, a case study was conducted concerning a pressurized water reactor (PWR) emergency diesel generator (EDG). Overall, this work has shown that the methods of RIPBR are feasible to use, and capable of justifying simultaneous safety and economic nuclear power improvements. However, it also reveals several areas where the framework of RIPBR should be strengthened. First, researchers need better data and understanding regarding individual component-failure modes that may cause components to fail. Not only are more data needed on failure rates, but more data and understanding are needed to enable analysts to evaluate whether these failures become more likely as the interval between tests is increased. This is because the current state of failure data is not sufficiently finely detailed to define the failure rates of individual component failure modes; such knowledge is needed when changing component-specific regulatory requirements. Second, the role of component testing, given that a component has failed, needs to be strengthened within the context of RIPBR. This includes formulating requirements for updating the prior probability distribution of a component failure rate and conducting additional or more frequent testing. Finally, as a means of compensating for unavoidable uncertainty as an obstacle to regulatory decision-making, limits to knowledge must be treated explicitly and formally. This treatment includes the formulation of probabilities through expert solicitation and the review of risk-informed, performance-based and engineering analyses used to evaluate proposed changes to existing technical specifications.