Dans le cadre du projet ASTRID ((Advanced Sodium Technological Reactor for Industrial Demonstration), les interactions fluide-structure mettant en jeu la dynamique du coeur (gerbage), tels qu'elles peuvent survenir lors d'un séisme, sont d'un grand intérêt. Le gerbage du coeur est également reconnu comme l'événement le plus plausible pour expliquer les quatre AURN (Arrêt d'Urgence pour Radioactivité Négative) survenus dans le réacteur Phénix, durant les années 1989 et 1990. L'objectif poursuivi est d'améliorer, pour leurs aspects dynamiques, la compréhension des interactions fluide-structure susceptibles de se produire dans un SFR (Sodium Fast Reactor). Le centre d'intérêt principal étant phénomène de dissipation visqueuse, cette thèse entreprend trois approches expérimentales, numérique et analytique, en s'appuyant sur des expériences de vibrations libres menées sur deux installations, PISE1A, mono-assemblage et PISE2C, multi-assemblages. Deux séries d'expériences de vibrations libres ont été menées sur PISE1A, en faisant varier la hauteur d'eau et en utilisant un mélange d'eau et de glycérol, dans des proportions variables. Le but est d'examiner l'influence des variations de masse ajoutée et de viscosité sur la dynamique des oscillations de l'assemblage. Les simulations numériques correspondantes, développées dans le code CAST3M, se sont appuyées sur la résolution des équations de Navier-Stokes 3D. Les écarts entre les résultats numériques et expérimentaux sont présentés et analysés. En particulier, les effets d'extrémité se sont révélés être d'une importance marginale. Des expériences de vibrations libres ont également été effectuées sur PISE2C en sollicitant l'installation de trois façons différentes : mise en mouvement globale, mise en mouvement par la couronne externe puis par la couronne interne. Un modèle réticulé, fondé sur des hypothèses de symétrie et de linéarité a été développé parallèlement. Les résultats expérimentaux ont permis de confirmer les symétries mais ont remis en cause les hypothèses de linéarité. Ce résultat encourage à persévérer dans la voie des modèles déterministes. / In the scheme of French ASTRID (Advanced Sodium Technological Reactor for Industrial Demonstration) project, fluid-structure interaction phenomena involved in the dynamic behaviour of core flowering, which could happen during seismic events, are of high interest. Also core flowering behaviour is considered as the main initiating event for the four SCRAMs that happened in Phénix reactor during 1989 and 1990. In objective to improve the knowledge of fluid-structure interaction phenomena of dynamic issues in a SFR core, especially focused on damping, this Ph.D. thesis have been conducted in experimental, numerical and analytical approaches based on free-vibration experiments on mono-assembly test facility PISE-1A and multi-assembly test facility PISE-2C. Two series of free-vibration experiments have been performed on PISE-1A with different water heights and different mass fractions of water-glycerol mixtures to examine the dynamic behaviours with respect to different added mass, different densities and viscosities. Corresponding numerical interpretations have been conducted with 3D Navier-Stokes model in CAST3M code. Sources of uncertainties are discussed to explain the discrepancies between the numerical computation and experimental results. Edge effects are not found to have an important impact on the dynamic behaviours of the system. On PISE-2C, free-vibration experiments with different modes of excitations have been conducted, including total flowering, partial flowering with internal crown excited and partial flowering with external crown excited. A reticulate model with homogenised linear hypothesis has been developed to interpret PISE-2C experiments. Good symmetries are found in PISE-2C suggesting that the deterministic tool is valid for the analysis.

The need for energy is growing in the world and the market of nuclear power is now once more expanding. Some issues of the current light-water reactors can be solved by the next generation of nuclear power, Generation IV, where sodium-cooled reactors are one of the candidates. Phénix was a French prototype sodium-cooled reactor, which is seen as a success. Although it did encounter an earlier unexperienced phenomenon, A.U.R.N., in which a negative reactivity transient followed by an oscillating behavior forced an automatic emergency shutdown of the reactor. This phenomenon lead to a lot of downtime of the reactor and is still unsolved. However, the most probable cause of the transients is radial movements of the core, referred to as core-flowering. This study has investigated the available documentation of the A.U.R.N. events. A simplified model of core-flowering was also created in order to simulate how radial expansion affects the reactivity of a sodium-cooled core. Serpent, which is a Monte-Carlo based simulation code, was chosen as calculation tool. Furthermore, a model of the Phénix core was successfully created and partly validated. The model of the core has a k_eff = 1.00298 and a neutron flux of (8.43+-0.02)!10^15 neutrons/cm^2 at normal state. The result obtained from the simulations shows that an expansion of the core radius decreases the reactivity. A linear approximation of the result gave the relation: change in k_eff/core extension = - 60 pcm/mm. This value corresponds remarkably well to the around - 60 pcm/mm that was obtained from the dedicated core-flowering experiments in Phénix made by the CEA. Core-flowering can recreate similar signals to those registered during the A.U.R.N. events, though the absence of trace of core movements in Phénix speaks against this. However, if core-flowering is the sought answer, it can be avoided by design. The equipment that registered the A.U.R.N. events have proved to be insensitive to noise. Though, the high amplitude of the transients and their rapidness have made some researcher believe that the events are a combination of interference in the equipment of Phénix and a mechanical phenomenon. Regardless, the origin of A.U.R.N. seems to be bound to some specific parameter of Phénix due to the fact that the transients only have occurred in this reactor. A safety analysis made by an expert committee, appointed by CEA, showed that the A.U.R.N. events are not a threat to the safety of Phénix. However, the origin of these negative transients has to be found before any construction of a commercial size sodium-cooled fast reactor can begin. Thus, further research is needed.