A Review of Safety Analysis Philosophies for Nuclear Reactors

Abstract

Various questions can be examined when discussing safety in general. Among these, some key issues are the attitude toward risk and its acceptance, the ways of identifying, analyzing, and quantifying risks, and societal factors and public opinion toward risks. The identification and quantification of risks are central in the regulatory framework and decision making and will be the focus of this article. Various approaches have been used for safety analysis over the years. This paper will survey some of the central attitudes in the nuclear reactor regulation philosophy and discuss the historical background surrounding them. Among these, we mention the “defense-in-depth” approach, the design basis accident (DBA), and beyond design basis accident (BDBA) analyses and discuss the rather subjective nature of their associated decision making. We maintain that it has long been recognized that the natural approach that comes out of the scientific method of inquiry is the probabilistic one, which in today's tools is conducted through the probabilistic safety analysis (PSA) method. This approach unlike the deterministic one, which produced concepts like DBA and defense-in-depth, enables us to put risks into context and to compare different risks such as those posed by different activities in particular or by other industries in general. It has consequently been gaining wide acceptance in regulatory bodies around the world as an effective tool in the inspection and regulation of nuclear reactors. Yet, it is also recognized that despite significant development over the past few decades, PSA still suffers from some well-known deficiencies. Its main benefit at this point is its contribution to identification and prioritization of design features, maintenance, management, and quality assurance (QA) important to safety. PSA has mostly been used in the nuclear power industry, but in recent years it has also started to be incorporated in research reactor (RR) safety analysis, and we therefore cover the subject of PSA usage for this purpose as well.