Methodology - PSA Regulatory handbook. Comparisons to a modern PSA study

The regulatory handbook is applicable to all types of initiating events and all operating conditions. It should be noted that it does not make the traditional subdivision of PSA into internal and external events, level 1 and level 2 PSA, or power operation and shut-down. The reason for this is that this has given the regulatory handbook a more logical structure, and that this approach underlines the integrated character of PSA when it comes to creating the plan risk profile. The regulatory handbook has been structured following the requirements on a PSA for a nuclear power plant, as this is the most demanding application. However, it is applicable also to the analysis of other nuclear installations. The purpose of the comparative review presented in this report has been to, as part of a quality review establish the PSA Handbook, compare (parts of) the handbook and its criteria with a recent PSA analysis, and to identify major discrepancies. Considerable weight has also been allocated to a review of the plant model (Risk Spectrum event trees and fault trees). The results presented in the report are not based on a complete review of the PSA in question (or of the complete PSA Handbook). Following discussions between the SKI and SwedPower, and based on the experience of the SwedPower reviewers, the following issues were chosen to be the main parts of the project: 1) General comparison according to content and transparency - Levels of ambition in PSA Handbook, PSA method description and actual PSA report. 2) Detailed comparison of: Selected component failure data - Assumptions regarding room events - CCI frequencies, realism, identification, categorisation - Taking credit for non-safety classified systems - Event tree modelling - Presentation of results 3) Fault tree model, specifically - Time frame for crediting of battery capacity - Modelling of regulators - Modelling of dependencies for room events - general quality, like how the paper documentation and the logic fault tree model correspond. The main focus of the review has been on the plant model. The comments and conclusions are dominated by issues concerning modelling. As of part 1.5, the criteria for the review as presented in the PSA Handbook are not to be seen as strict requirements. The layout of the Handbook does not follow the traditional PSA with its subdivision of into internal and external events, level 1 and level 2 PSA, or power operation and shut-down. The reason for this is that this has given the regulatory handbook a more logical structure, and that this approach underlines the integrated character of PSA when it comes to creating the plan risk profile. When the Handbook is used as the reference for a review of a PSA this has to be taken into consideration. One partial analysis of a PSA program can not be expected to meet all requirements of the PSA Handbook. On a general level, the PSA in question in many ways seem to be on a satisfactory level with modern PSA standard. As expected though, in relation to the level of ambition as described in the Handbook, some areas could be improved. The main conclusions can be summarised as: The PSA Handbook (PSAHB) is not intended to be used for detailed review of analyses, but for an overall judgement of a complete PSA program - The PSAHB pictures an 'ideal' version of a complete PSA. During a comparative review, as the one described in this report, considerable discrepancies show up concerning level of ambition etc. - The PSAHB, and in particular its chapters on presentation and analysis of results are well written and worth considering for any PSA analyst. - The specific PSA analysis studied shows a high level of ambition, but as - A thorough analysis of the results is missing. - Some of the analyses are overly conservative. - The CCI-analysis is presented only in a 'first screening' status. - No uncertainty analysis is presented. - The criteria for the sensitivity analysis are not shown. - The models used for internal fire- and flooding analyses are too conservative to be applied for older, poorly separated plants. - The above shortcomings put severe limitations regarding the use for decision making for the intended applications