Volume 74, Issue 3

Increasingly wide application of Probabilistic Safety Assessment (PSA) to support a variety of issues dealing with operation and design of Nuclear Power Plants (NPP) imposes certain requirements on PSA quality in terms of its scope, level of detail, approaches, assumptions, and data used. If a unified set of balanced requirements is developed and consistently implemented, it would give confidence in using PSA results and insights in different applications and Risk-Informed Decision Making (RIDM); in addition, it would be possible to consistently compare PSA results for different plants. Such requirements are currently developed or being developed in several countries. The International Atomic Energy Agency (IAEA) is in the process of development of several publications in the IAEA Safety Standards Series that would provide guidance on the standard technical content of PSA. The paper discusses the latest developments at the IAEA in the area of PSA and provides a brief overview of the publications.

The Open PSA Initiative round table discussion session is to bring the PSA community up-to-date on our meetings, workshops, and working groups during the last year. We will have members of the initiative speak on their point of view of the initiative, what they have gained, and where we as a group want to go during the next year. The main idea is to use appropriate modules of different computer codes for a specific purpose which requires a common platform and a common PSA format. In this short paper we introduce our main project for this year: the Open PSA Model Exchange Format.

Worldwide it can be recognised that the use of probabilistic safety assessment (PSA) in regulatory as well as operational decision-making is state of the art and seen as a successful development. Therefore, in most cases the regulator encourages the performance of PSAs to provide information to complement and support the defence in depth philosophy as well as operational configuration decisions. The main application of the PSA is still as part of integrated safety reviews, in particular in the frame of comprehensive (periodic) safety reviews. Other more specific applications areas of PSA are, among others, design evaluation, event analysis with aid of PSA, evaluation of technical specifications; risk-informed in-service inspection, risk monitoring and accident management. The extent of these applications vary from country to country but has been increasing during the last years.

Precursor analysis is an internationally recognized method for quantifying the safety-relevance of operational events in nuclear power plants. Precursors are operational events in nuclear power plants which had no serious impact, but which could have led to serious impacts, if additional malfunctions had occurred. Examples of such operational events are component failures or transients, for example, the loss of main feedwater. On the basis of the probabilities for the occurrence of additional malfunctions or initiating events precursor analyses determine the probability with which these additional malfunctions during the event occurred would have led to core damage. This conditional probability is a measure for the safety relevance of the operational event occurred. Events, for which the probability of core damages is > 10 –6 , are internationally classified as “precursor”.

Probabilistic Safety Assessments (PSAs) in Germany have essentially been conducted in the framework of recurrent Safety Reviews (SRs). Besides, probabilistic investigations have also been submitted and reviewed in the usual regulatory procedures. Significant safety relevant insights have e. g. been gained from the Human Reliability Assessment (HRA) in the context of Low Power and Shutdown (LPS) PSAs whose performance has been launched by the regulatory body. In this paper major insights and results together with the practical outcome particularly from Human Factor (HF) and safety relevance points of view are described. Thereby, general key issues on HRA as well as substantial insights and conclusions derived from LPS HRAs are addressed. Further, a specific example with respect to the design and review of a complete set of operating procedures specifically dedicated to the shutdown operation is illustrated by outlining the general approach as well as the overall outcome achieved. Finally, conclusions are drawn regarding the relevance of probabilistic investigations, the appropriateness of a symptom oriented proceeding related to shutdown operation and the potential of utilising an advanced HF methodology.

Fire PSA is an essential part of a full scope level 1 PSA. Cable fires play an important role in fire PSA. Usually, cable routing is therefore modeled in detail. During the design of new nuclear power plants the information on cable routing is not yet available. However, for the use of probabilistic safety insights during the design and for licensing purposes a fire PSA may be requested. Therefore a methodology has been developed which makes use of the strictly divisional separation of redundancies in the design of modern nuclear power plants: cable routing is not needed within one division but replaced by the conservative assumption that all equipment fails due to a fire in the concerned division; critical fire areas are defined where components belonging to different divisions may be affected by a fire. For the determination of fire frequencies a component based approach is proposed. The resulting core damage frequencies due to fire are conservative.

The German regulatory body has issued technical documents on methods and data for probabilistic safety assessment, elaborated for a comprehensive integrated safety review of all nuclear power plants in operation. These documents contain, among others, a graded approach for the probabilistic assessment of external flooding. Main aspects are explained such as the underlying probabilistic considerations and the mathematical procedures for the calculation of exceedance frequencies for nuclear power plants at tidal rivers in Germany. Morever, extreme events such as tsunami waves have been investigated.

Ageing is a process, where the properties of systems and processes may degrade through the time and age. The objective of the paper is to analyze the possibilities of introduction of ageing directly into the probabilistic safety assessment. Theoretical models of ageing and examples of their application are investigated, which show, how the current models can be upgraded in sense to directly include the effects of ageing into the models of probabilistic safety assessment. This paper shows that consideration of ageing is theoretically a straightforward issue, which can be dealt with in several ways, which are demonstrated in the paper. The most important problem is the lack of data about the effects of ageing, which would suit to the detailed models of ageing. The results, which are obtained, do not show significantly different results compared to the initial models without direct consideration of ageing. The majority of risk measures change slightly or notably, but no major break through has been found. Contribution of ageing may be smaller then contribution of changes due to other parameters.

A methodology for a Level 2 PSA is described that provides a more efficient structure than the two methods recommended in the German Leitfaden. This is accomplished by merging the top events for the system and accident management functions from the Level 2 space, referred to as the containment system event tree (CSET) as a direct extension of the Level 1 model and by introducing the containment phenomenological event tree (CPET) to represent only the phenomenological Level 2 top events. The interface between the Level 1+ CSET and the CPET is defined as the plant damage states (PDS) with the focused objective to minimize the effects of sequence to sequence variability on the CPET failure fractions. A clear and consequent distinction is made between the statistically random events modeled in the Level 1+ CSET and the deterministic physical processes with large state of knowledge uncertainties modeled in the CPET. It is shown that as a logical consequence the resulting CPET failure fractions are numbers and not distributions, and these numbers are consistent with the interpretation as the degree of confidence probabilities for the outcome of uncertain physical processes. While both methods recommended in the Leitfaden will yield correct results if properly applied, the proposed method will result in a more streamlined model that will be easier to trace, understand and review, and it will result in lower release category uncertainties by minimizing the sequence to sequence variability through the elimination of the intermediate pinch points and through the focused use of the PDS.

On international level, probabilistic safety assessment represents the state of the art in the licensing process for new industrial facilities, but increasingly also for evaluating the safety level of older industrial plants, e. g. as part of periodic safety reviews in case of nuclear installations. Quantitative safety goals or criteria have not yet reached the same level of acceptance. However, the role of such goals/criteria depends on the type of industry. Most of the countries consider such criteria as safety targets rather than as sharply defined boundary values. The Netherlands and the United Kingdom are exceptions because they require demonstration of compliance with legally binding safety goals in the licensing procedure for all types of industrial activities.