T-book. Reliability data of components in Nordic nuclear power plants. 6. ed

The main objective of the T-Book is to provide reliability data for the unavailability computations that are made for each component that is considered in the compulsory, probabilistic safety assessments (PSA) of nuclear power plants. As the use of PSA is large in the normal safety work at the NPPs, there is a need for easily accessible and reliable failure data. The failure characteristics presented in the T-Book are primarily based on the failure reports stored in the central database TUD and the Licensee Event Reports delivered to the Swedish Nuclear Power Inspectorate (SKI). Fortunately, the TUD database was started already in the middle of the seventies by the Swedish power companies. In 1981, the Finnish power company TVO, operating two reactor units of Swedish design, joined the data collection system. Before the TUD data are statistically treated they are carefully examined with respect to the consistency and correctness. This T-Book comprises only critical failures, i.e. failures that stops the function of components or leads to repair. The first edition of the T-Book was issued in 1982 encompassing operational statistics from 21 reactor years. The second edition was published 1985, based on operating data covering about 40 reactor years. The T-Book 3 was published in 1992 and included data up to the operating year 1987 (108 reactor years). Edition 4 was published 1994 containing information up to and including 1992 (178 reactor years). Edition 5 was published year 2000 containing information up to and including 1996 (234 reactor years). This edition 6 contains information including year 2002 (315 reactor years). At the same time as the amount of data has increased with the successive editions of the T-Book there has been a continuous work to improve the methods for the statistical inference and related program tools, required to derive the reliability parameters from the operational data in the database. Already in the initial edition there was a Bayesian reasoning in the description of the uncertainty associated with the failure rates or demand-related failure probabilities. Analytically attractive distributions, gamma distributions for the failure rate, were used to model this uncertainty, and straightforward statistical methods were used to estimate the parameters of these distributions. The reliability parameters nowadays are calculated according to a method called 'Bayes empirical Bayes'. According to this method also the parameters of the uncertainty distributions are estimated in a Bayesian way. Of great interest to the community of nuclear safety analysts is the application of the qo+λst model, where qo is derived from failures occurring at the demand occasion while λs characterises failure mechanisms that are active during the standby time. This model is applicable on component groups where several test intervals are represented in the operational data. The marginal distributions of qo and λs are shown for both generic and unit specific values. The T-Book should be considered as an important step in the continuous effort to collect and analyse data on the reliability of safety components at NPPs. There will be future revisions of the T-Book with incorporation of new data and implementation of user experiences. The English edition of the T-Book has been brought forward in response to a large interest outside the Nordic countries. The contents are essentially the same as the Swedish one