Generalized perturbation techniques for uncertainty quantification in lead-cooled fast reactors. (15th December 2021)

• Record Type: Journal Article
• Authors: Abrate, Nicolò; Dulla, Sandra; Ravetto, Piero
• Is Part Of: Annals of nuclear energy. Volume 164(2021)
• Publication Date: 2021-12-15
• Subjects: Uncertainty propagation -- Sensitivity analysis -- Reduced-order models -- GPT -- XGPT -- Monte Carlo -- Serpent 2 code -- Nuclear data -- Lead-cooled fast reactor; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2021.108623
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: GPT and XGPT methods are assessed for the sensitivity analysis and uncertainty quantification in the Monte Carlo framework. GPT and XGPT approaches are presented and discussed. GPT method is employed for an uncertainty quantification study of the Lead-cooled Fast Reactor ALFRED design. XGPT calculations are performed for the two most important nuclides on a fine-group grid structure. Practical indications on method performances are drawn. Abstract: The design of innovative nuclear fission systems requires a careful evaluation of the uncertainties affecting the basic input data. Among them, nuclear data are particularly relevant, due to their dramatic energy dependence. Because of this feature and of the strong spatial heterogeneity of nuclear reactors arrangement, full-core calculations are carried out using energy collapsed and spatially homogenised constants. Nowadays, collapsing is often performed with Monte Carlo codes, which allow a discretisation-free treatment of the neutron transport equation. The most popular method to propagate the uncertainty in the nuclear data libraries throughout the Monte Carlo transport calculation is the Generalised Perturbation Theory (GPT). However, due to its multi-group nature, GPT often blurs the continuous-energy feature of the Monte Carlo method. Therefore, in order to fully exploit its advantages, the XGPT method has been recently proposed. After discussing the main differences between these two approaches, the paper presents the application to an uncertainty quantification study on the lead-cooled fast reactor ALFRED design, performed with GPT and focused on the multi-group cross sections. Afterwards, the two nuclides that most contribute to the overall uncertainties, i.e. Pu-239 and U-238, are considered to compare the GPT results to some XGPT calculations carried out with different multi-group energy structures. This analysis suggests that XGPT is a consistent method for uncertainty quantification in the continuous-energy Monte Carlo framework. Moreover, it can be concluded that an adequate number of low-energy groups is necessary for an accurate uncertainty evaluation in the case of a fast system.