Nuclear data uncertainties propagation methods in Boltzmann/Bateman coupled problems: Application to reactivity in MTR. (April 2016)

• Record Type: Journal Article
• Authors: Frosio, Thomas; Bonaccorsi, Thomas; Blaise, Patrick
• Is Part Of: Annals of nuclear energy. Volume 90(2016:Apr.)
• Page Start: 303
• Page End: 317
• Publication Date: 2016-04
• Subjects: Uncertainty propagation -- Nuclear data -- Transport -- Depletion -- Sensitivity analysis -- Reactivity; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2015.12.010
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: Hybrid methods are developed for uncertainty propagation. These methods take into account the flux perturbation in the coupled problem. We show that OAT and MC methods give coherent results, except for Pearson correlations. Local sensitivity analysis is performed. Abstract: A novel method has been developed to calculate sensitivity coefficients in coupled Boltzmann/Bateman problem for nuclear data (ND) uncertainties propagation on the reactivity. Different uncertainty propagation methodologies, such as One-At-a-Time (OAT) and hybrid Monte-Carlo/deterministic methods have been tested and are discussed on an actual example of ND uncertainty problem on a Material Testing Reactor (MTR) benchmark. Those methods, unlike total Monte Carlo (MC) sampling for uncertainty propagation and quantification (UQ), allow obtaining sensitivity coefficients, as well as Bravais–Pearson correlations values between Boltzmann and Bateman, during the depletion calculation for global neutronics parameters such as the effective multiplication coefficient. The methodologies are compared to a pure MC sampling method, usually considered as the "reference" method. It is shown that methodologies can seriously underestimate propagated variances, when Bravais–Pearson correlations on ND are not taken into account in the UQ process.