Refine your search:     
Report No.

Overview of recent methods for the modeling of the uncertainties on the calculations of consequences of a nuclear power plant severe accident

Chevalier-Jabet, K.*; Zheng, X. ; Mabrouk, A.*; Maruyama, Yu ; Baccou, J.*

Severe accident phenomenology in light water nuclear power plants is complex. For the past decades, extensive experimental programs have been conducted to gain knowledge and computational tools have been built to predict accident progressions and consequences. Nevertheless remained uncertainties directly affect the predictability of severe accidents consequences. Monte-Carlo techniques are widely used in previous uncertainty analysis and the shortcomings are addressed by JAEA and IRSN. The first part of the article deals with uncertainty propagation. Possibilist formalisms are presented with an example. In the second part, JAEA has developed a method for source term assessment using a Dirichlet process. The implementation of the method is described from the Bayesian nonparametric model to the cross-validation process. As results, corresponding computational cost and importance measure of inputs are addressed. The third part describes the current research at IRSN. The combination of Bayesian formalism and graph theory is applied to modeling severe accident uncertainties. The method allows the information to propagate in any direction of the graph, making inference easy to perform. Bayesian networks allow the representation of a complex model in an integrated environment.



- Accesses





[CLARIVATE ANALYTICS], [WEB OF SCIENCE], [HIGHLY CITED PAPER & CUP LOGO] and [HOT PAPER & FIRE LOGO] are trademarks of Clarivate Analytics, and/or its affiliated company or companies, and used herein by permission and/or license.