Simulation and modelling of fluid-structure interactions in two-phase flows

Steam generators are a key component of nuclear power reactors, and an in-depth knowledge of their mechanisms is a major industrial challenge for the designer AREVA-NP and the operator EDF. Vibration of tube bundles induced by cross-flow is one of the problems encountered by the designer, thus needing to assess the vibration response to the excitation generated by the flow. The critical region is the U shape of the bundle (upper part of the steam generator), where two-phase cross-flow occurs with an important void fraction. In order to measure excitation induced by flow fluctuations on the tube bundle, some physical parameters have to be identified. For single-phase flows, it seems possible to link load on tubular structure to turbulence intensity of the flow, thanks to experimental data reduction methods together with numerical simulation methods. For two-phase flows, it is believed that forces induced on the tubes by the flow have other origins, and might be connected to dynamic contribution of each phase together with interfacial transfers. Nevertheless, relevant physical parameters which could predict the amplitude of the forces remain a subject of debate (void fraction, flow regime, etc.) and physical processes not yet fully understood. In order to study mechanical instabilities in two-phase flows, some analytic experiments have been and continue to be conducted at CEA. These analytic experiments focuses on isolated tube or tube bundles (rigid or flexible), and on a large regime flow range (AMOVI and DIVA mockups at CEA). They aim to describe these mechanical instabilities (forces measurement on the obstacle) based upon average parameters of the flow (gas and liquid flow rates, 'mean' void fraction, etc.), but also local parameters (local void fraction, bubble size, gas velocity, etc.). These measured or locally estimated parameters are used to conduct relevant non-dimensionalization, both on the random excitation forces (two-phase excitation spectrum on a rigid tube) and the fluid-elastic coupling forces (single flexible tube or flexible bundle). Nonetheless, some dispersion remains on the results, physical mechanisms are not well understood, and the non dimensionalization process remains dependent on metrology. The aim of this PhD thesis is to conduct numerical simulations with front tracking in configurations close to the experiments conducted at CEA in order to expand the knowledge on phenomena leading to vibration of tube bundle in steam generators