Uncertainties propagation in the UAM numerical rod ejection benchmark. (15th June 2020)

• Record Type: Journal Article
• Authors: Sargeni, A.; Fouet, F.; Ivanov, E.; Probst, P.
• Is Part Of: Annals of nuclear energy. Volume 141(2020)
• Publication Date: 2020-06-15
• Subjects: Uncertainties propagation -- Rod ejection -- UAM; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2020.107339
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: Nuclear data uncertainties were propagated through UAM rod ejection model w/o feedback. Relative standard deviation (driven by the fast group) follows and grows faster than the mean power curve. The dispersion of a power peak can become very sensitive to small variations of the nuclear data. How far can we rely on sampling-based methodology characterizing accuracy of a transient? A conservative curve has to include all the extremes points of an uncertainties propagation. Abstract: This paper presents the execution and the results of a rod ejection benchmark inspired by the UAM benchmarks II-2b: cross sections uncertainties propagation during a rod ejection transient in a mini-core (3 × 3 PWR-type assemblies), without feedback. Performing transient computations in two groups time-dependent neutron diffusion approximation while starting from a steady-state power of 0.1409 MW, we obtained a mean power peak around 0.25 MW with a standard deviation of, nearly, 1.5%. The time-dependent standard deviation curve follows the time-dependent mean power curve and the relative standard deviation, in a percentage of the mean power, increases with the mean power peak value. We found, too, that the first-group diffusion coefficient and the first-group absorption cross section are, practically, the only contributors to the total power peak variance. Drawbacks and findings from the exercise provide essential input for uncertainty propagation in a further multi-physics simulation of a postulated rod ejection accident.