SharePoint
Aide
Faire avancer la sûreté nucléaire

La Recherchev2

Publications

Initialization bias suppression in Iterative Monte Carlo Calculations: Benchmarks on Criticality Calculation


Fermer

Authentification

Email :

Mot de passe :

Yann RICHET1, Olivier JACQUET2, Xavier BAY3 -Avignon, France, Sept. 12-15, 2005, Mathematics and Computation, Supercomputing, Reactor Physics and Nuclear and Biological Application 2005.

Type de document > *Congrès/colloque

Mots clés > criticité, criticité, Monte Carlo (code)

Unité de recherche > IRSN/DSU/SEC

Auteurs > RICHET Yann

Date de publication > 12/09/2005

Résumé

The accuracy of an Iterative Monte Carlo calculation requires the convergence of the simulation output process. The present paper deals with a post processing algorithm to suppress the transient due to initialization applied on criticality calculations. It should be noticed that this initial transient suppression aims only at obtaining a stationary output series, then the convergence of the calculation needs to be guaranteed independently. The transient suppression algorithm consists in a repeated truncation of the first observations of the output process. The truncation of the first observations is performed as long as a stationarity test based on Brownian bridge theory is negative. This transient suppression method was previously tuned for a simplified model of criticality calculations, although this paper focuses on the efficiency on real criticality calculations. The efficiency test will be based on four benchmarks with strong source convergence problems designed by the NEA Expert Group on Source Convergence in Criticality-Safety Analysis.


1 IRSN
2 Independent Consultant
3 Ecole nationale supérieure des mines de Saint-Etienne

More information