Engineering aspects of probabilistic risk assessment

Abstract

Listen

Translate article

Over the last decade, the use of probabilistic risk assessment (PRA) in the nuclear industry has expanded significantly. In these analyses the probabilities of experiencing certain undesired events (for example, a plant accident which results in damage to the nuclear fuel) are estimated and the consequences of these events are evaluated in terms of some common measure. These probabilities and consequences are then combined to form a representation of the risk associated with the plant studied. In the relatively short history of probabilistic risk assessment of nuclear power plants, the primary motivation for these studies has been the quantitative assessment of public risk associated with a single plant or group of plants. Accordingly, the primary product of most PRAs performed to date has been a 'risk curve' in which the probability (or expected frequency) of exceeding a certain consequence level is plotted against that consequence. The most common goal of these assessments has been to demonstrate the 'acceptability' of the calculated risk by comparison of the resultant risk curve to risk curves associated with other plants or with other societal risks. The first large scale application of probabilistic risk techniques to evaluate the safety of nuclear power plants was the Reactor Safety Study (also known as WASH-1400, or the Rasmussen Report)) This study produced risk curves for a BWR and a PWR and compared these curves to those associated with other sources of societal risks (e.g. airplane crashes, floods, meteorites, etc.). Subsequent to the Reactor Safety Study, the Nuclear Regulatory Commission (NRC) requested that certain plants in high population areas perform PRAs to ensure that these plants do not represent a 'disproportionately high segment of the total societal risks '2. The recent introduction of'safety goals' and 'quantitative design objectives' by the NRC 3 will ensure that a major motivation behind and focus of PRA will remain generation of a risk curve and the performance of a PRA will be analogous to performing an (analytical) test on a plant which must be acceptably passed before operation (or continued operation) is allowed. While the use of PRA in this manner can provide considerable benefits to the entire industry by providing a rational limit to an otherwise open-ended regulatory 'ratcheting' process, it is extremely important to recognize that there are a number of other (and perhaps better) ways of using PRA models and techniques to improve the safety of plants. Recognition of these alternative PRA applications is important to ensure that maximum benefit is achieved from use of the models and techniques, to ensure that the credibility and usefulness of PRA models and techniques are not judged solely on the basis of one application, and to ensure that resources devoted to PRA methodology development activities are expended where they can do the most good. Presented below are brief descriptions of some of these alternate applications of PRAs, a discussion of how these other applications compare or contrast with the currently popular uses of PRA, and a discussion of the relative benefits of each.