Lieux de thèse : IRSN (Fontenay-aux-Roses), déplacements ponctuels à Grenoble et Raleigh (États-Unis) pour la seconde partie des études
Date de début : octobre 2018
Écoles d’ingénieurs ou Master 2 physique et mathématique
Sujet de thèse
In the global trend of the increase of the use of simulation within nuclear systems design, safety analysis and exploitation the experimental validation becomes a specific scientific discipline with its own conceptual basis. While sufficient progress is achieved in the single physics validation and in the separate fields of expertise, the validation of the coupled, multi-physics and multi-scale models remains challengeable. The works performed in the past – in the frameworks of the NEA projects ICSBEP and IRPhE - give a good pattern ; it is proposed to extend the approved methodology of the data evaluation to other physics and even multi-scale, multi-physics and coupled phenomena. The challenges for the coupled, multi-physics experimental data evaluation are numerous as follows: (1) the available practice for different fields of expertise is not uniform – the paradigms applied over there are different – from stringent hierarchic approaches in reactor physics and in the thermal-hydraulics to widely used PIRT analysis in others, (2) the non-linear processes are impacted by divergence and bifurcation phenomena losing the information. The evaluation of the Integral Experimental Data requires numerous sophisticated computational and analytical works. Very often it requires even development of the specific tools to manage the uncertainties, discontinuities etc. The problem is even more complicated in the case of non-stable systems or of phenomena for which where we have not yet the robust theoretical models.
The proposed study should result in the merging of different practices and in the development of the guidance on the data evaluation isolating the bifurcations. The PhD work will consist in the development of the methodology to evaluate the complex integral experiments data, the establishment of the representative high fidelity benchmarks based on experimental data, the elaboration of the guidance for comprehensive multi-physics tools validation, and the application of the validation process on a limited number of demonstrations to confirm the feasibility of the approaches. The PhD student will use modern computational coupled multi-physics code systems (MOOSE-BIZON, SERPENT2 coupled with OPENFOAM. Specific computational tools or/and computational approaches will be used for the experimental data evaluations. Such tools could include as specific workbenches intended on the errors’ propagation (RAVEN, DAKOTA, URANIE, PROMETEE).