Validation of the uncertainty propagation method for the decay heat within the DARWIN2.3 package
Résumé
The DARWIN2.3 package is dedicated to spent fuel characterization in general and to decay heat computations in particular, and benefits from a Verification, Validation and Uncertainty Quantification process (VVetUQ). The current experimental results available are insufficient to accurately assess the biases and uncertainties due to nuclear data for the decay heat for the fuel managements of the French reactor fleet. This is the reason why an uncertainty propagation method has been recently implemented within the CYRUS tool (a mockup for the DARWIN2.3 package), and later on within the DARWIN2.3 package in order to compute a priori uncertainty due to nuclear data variance/covariance matrices. The method implemented is the quadratic summation method, which allows in-depth analysis of the uncertainty through sensitivity and variance analysis. Nevertheless, this method relies on a fundamental hypothesis of linearity, which is not straightforward in the case of the decay heat. Moreover, due to the specificity of the DARWIN2.3 package, approximations have been made when dealing with cross-sections, such as the collapsing of the multigroup covariance matrices, and the neglecting of the Boltzmann/Bateman coupling when computing the sensitivity profiles. These three items have been studied in details. This paper shows some of the results that accredit the legitimacy of the linearity hypothesis and collapsing method, and provide some recommendations in dealing with the Boltzmann/Bateman coupling.
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...