Taking into account the information on uncertainties in nuclear safety :some recent advances in modelling, propagation and synthesis
Evaluating and enhancing safety in nuclear industry is a very important issue and led to huge research programs. Among the efforts to benefit from the knowledge gained during these programs, we can mention the development of best-estimate computer codes. Such codes attempt to calculate any accidental transients in a realistic way. In order to take into account all the uncertainty sources to assess safety quantities are below critical values, uncertainty analyses are performed. From a methodological point of view, we have to cope with three key issues: the first one is related to information modelling and to the choice of a mathematical representation of uncertainty adapted to the real state of knowledge, the second one concerns the propagation of this information through a computer code and the third one deals with information synthesis in case of different uncertainty analyses using different codes and implying different experts. This work is devoted to some new developments proposed by IRSN to handle the three previous problems in order to design robust tools in nuclear studies for the accident management procedures and licensing. Since it will become clear that the classical probabilistic framework used by many analysts is not fully satisfactory, a special attention is devoted to a more flexible approach respecting the real state of knowledge on uncertainties. Its advantages are illustrated in the frame of the OECD/CSNI benchmark BEMUSE.
Uncertainty analysis aleatory/epistemic uncertainties RaFu propagation information fusion BEMUSE program.
J. Baccou E. Chojnacki S. Destercke
Institut de Radioprotection et de S?reté Nucléaire, Centre de Cadarache, 13115 St Paul Lez Durance, France
国际会议
上海
英文
2479-2491
2010-10-10（万方平台首次上网日期，不代表论文的发表时间）