Prospects for use of probabilistic safety criteria

Design against accidents is so far mainly based on deterministic criteria. Deterministic criteria consist of a set of engineering principles (e.g. defense-in-depth, redundancy, diversity, fail-safe) and design against a set of pre-selected, so-called design base accidents. To analyse the remaining risk and to find optimal ways to reduce risks is called risk management. The management of risks is based on two principles: - risk can be reduced below any given level, however, not to zero, and - the cost of further risk reduction increases with the level of safety achieved. It is thus necessary to find a balanced approach for cost-effective risk reduction. There are three paths to reduce risk from accidents both for using prevention and mitigation strategies: a) more extended application of engineering principles, b) extension of the set of design base accidents to include more unlikely events, and c) analysis and quantification of the remaining risk, and risk reduction based on decision criteria (i.e. probabilistic safety criteria). It is clear that all three options have to be followed and that the deterministic approaches (a and b) and the probabilistic approach (c) have to complement each other. The probabilistic approach also provides a rational tool to check on the balance and risk significance of measures taken based on a) or b). This paper concentrates on the probabilistic approach especially as it relates to nuclear plants