Laboratoire d'accueil : Laboratoire de neutronique (LN)
Date de début de thèse : septembre 2017
Nom du doctorant : Clément JEANNESSON
Descriptif du sujet
The design of nuclear system requires largely the use of computational tools. The algorithms included are in general based on stochastic and / or determinist solutions of the transport equation, and in any case, final outcomes rely heavily on the nuclear data that describe neutron-matter interactions. Performing independent calculation, in addition to checking the results carried out by others, is part of the IRSN activities. The very first step in the verification procedure is in regarding to the nuclear data used in the calculations. The Évaluation and the processing of nuclear data (translation of the nuclear data from experiments into useful forms for applications) are investigated at IRSN, in the Neutronics Laboratory (LN), where a great deal of effort has been devoted on the development of a nuclear data processing code (GAIA-2). New methods are however required to expand the processing to all energy regions.
The neutron-nuclear interactions are described by cross sections that represent the probability of a reaction to occur. They are functions of the neutron incident energy. At some energies in the epithermal energy range (depending on the target nuclide), neutronnucleus resonances cannot be experimentally distinguished. This energy domain is called the unresolved resonance range (URR), where parameters from which the cross sections can be derived are only given as averaged values. As a direct consequence, cross sections can only be computed as probability tables at some reference energies across the URR. Moreover, existing methodologies in this range have been developed at times where computer resources were very limited. For instance, a crude approximation of the R-matrix theory (that describes neutron-target interaction) is used: the Single-Level Breit-Wigner formalism. One of the motivations of the PhD work is to use a more rigorous formalism, in order to generate consistent probability tables for use in Monte-Carlo calculations.
A module to compute probability tables in the URR has been developed, that uses a slightly different sampling technique than the one in use in most of the existing software, such as NJOY (LANL) or PREPRO (IAEA). This module has been coupled with the IRSN data processing code, GAIA-2, and its results will be soon tested on benchmarks with the IRSN Monte-Carlo code, MORET5.