The evolution of probabilistic risk assessment in the nuclear industry

Abstract

The use of probabilistic methods for evaluating the performance of plant is now commonplace. In the nuclear industries it has undergone a vigorous period of development and is now considered by its aficionados to be a mature topic. It may be considered technically mature in that methods and data have been refined considerably and its positive and negative points are well understood. However, its breadth of application, especially in its most complex forms when risk rather than reliability is evaluated, has not been as wide as originally hoped, especially as an aid to regulation as in the evaluation of risk acceptance, or tolerability. This paper follows the development of that set of analytical techniques which together form probabilistic risk (or safety) assessment through its most formative years (1975-1985) by means of examples drawn from the definitive calculations of the period - in particular, the Reactor Safety Study (1975), the Zion and Indian Point studies in the USA and the Sizewell B study in the UK (all circa 1983). All of these studies contributed in specific ways to the development of the methods. In addition, the Sizewell B study, through its use in a public enquiry, also precipitated a debate on the use and interpretation of the results in the public domain. This evolution clearly shows both the power of the methods, and their extreme complication. These aspects have contributed to the current status of the methods, both for plant performance and regulatory interpretations, and to the prospects for further developments. © Institution of Chemical Engineers.