Continuous energy function sensitivity calculation using GPT in Monte Carlo neutron transport: application to resonance parameters sensitivity study

Abstract : GPT allows calculating sensitivities of one observable (as keff) to all nuclear data in all energy groups with only two neutron transport calculations : one direct and one adjoint. Deterministic transport calculations were the preferred for decades despite the fact that only sensitivities to energy grouped step-wise functions were computable. In deterministic methods, the impact of resonance parameters uncertainty is usually taken into account indirectly through their impact on group cross sections for which the sensitivities are available. The effect of selfshielding needs a specific treatment in the form the calculation of “implicit” effects. The impact of one individual parameter will then be diluted in the one of its energy group data. In this paper, sensitivities of (keff) to individual resonance parameters are calculated thanks to the recent development of GPT capabilities in Serpent Monte Carlo transport code. The implementation allows to easily calculate the sensitivities to the continuous energy function defined as the difference between cross sections calculated using reference and perturbed resonance parameters. Results of this GPT- based method are compared to a few direct simulations. The agreement demonstrates both the linearity of the sensitivities and the performance of the method.
Document type :
Conference papers
Complete list of metadatas

http://hal.in2p3.fr/in2p3-01347441
Contributor : Emmanuelle Vernay <>
Submitted on : Thursday, July 21, 2016 - 9:20:36 AM
Last modification on : Tuesday, May 22, 2018 - 9:48:11 PM

Identifiers

  • HAL Id : in2p3-01347441, version 1

Collections

IN2P3 | LPSC | UGA

Citation

M. Aufiero, A. Bidaud, M. Fratoni. Continuous energy function sensitivity calculation using GPT in Monte Carlo neutron transport: application to resonance parameters sensitivity study. International Congress on Advances in Nuclear Power Plants (ICAPP 2016), Apr 2016, San Francisco, United States. pp.826-830. ⟨in2p3-01347441⟩

Share

Metrics

Record views

103