High-Resolution Experiments of Momentum- and Buoyancy-Driven Flows for the Validation and Advancement of Computational Fluid Dynamics Codes

Abstract

In nuclear engineering, fluid dynamics and heat transfer play a crucial role in efficiently and safely removing heat from the nuclear reactor core to produce electricity from thermal energy. Moreover, new designs of nuclear power plants have incorporated passive safety features and passive cooling systems to provide adequate heat removal from the reactor core without the intervention of an operator or the need of external power. These passive systems rely on fluid-dynamic phenomena, such as natural circulation, for their operation. An accurate prediction of such phenomena is important to assess the performance of these systems and ensure that the core cooling is adequate during all accident scenarios.

In recent years, Computational Fluid Dynamics (CFD) is taking a growing role in the assessment of operational and safety features of nuclear power plants. To model turbulence, Reynolds-Averaged Navier-Stokes (RANS) formulations are commonly used due to their low computational cost and robustness; however, experimental data is needed to address some shortcomings of these models.

The primary goal of this thesis is to develop a high-resolution high-fidelity experimental database for the development and improvement of CFD codes, and to gain physical insight into complex phenomena relevant for nuclear power applications. These phenomena include the propagation of stratified fronts and the mixing/interaction of multiple jets in a plenum. To achieve this goal, the High Resolution Jet (HiRJet) and the Reactor Cavity Cooling System (RCCS) experimental facilities were built and equipped with Laser Doppler Velocimetry (LDV) and Particle Image Velocimetry (PIV) systems to obtain velocity fields, and Wire-Mesh Sensor (WMS) units to obtain scalar fields. Both measurement techniques allow for high-resolution experiments in space and time necessary for the validation and advancement of CFD models. The boundary conditions were also monitored with this advanced instrumentation, making these data adequate for CFD validation. Additionally, these measurements incorporate uncertainties due to geometries, image post-processing algorithms, and the measurements' reproducibility and repeatability.

PIV data obtained in the RCCS facility were utilized to study the interaction between multiple rectangular jets. Through time averages and modal decomposition analyses, dominant frequencies and the coherent structures associated with these frequencies were discerned. These structures were determined to be independent of Reynolds numbers. Furthermore, these data exhibited significant differences between separate PIV measurements. It was observed that uncertainty bands that do not consider changes in the measurement conditions and human effects do not always capture these large variations. Through comparisons between experimental data and RANS simulations, under- and over- production of the Reynolds stresses, along with their consequent effects on the velocity profiles, were identified. Furthermore, WMS data obtained in the HiRJet facility were obtained to investigate the propagation of stratified fronts in the presence of positive and negative density gradients. These data were also used to assess the predictive capabilities of RANS simulations. Through these experiments, it was determined that depending on the sign of the density gradient, the simulations over- or under-predicted the extensions of the mixing regions and the speed of the density front propagation. These differences were attributed to an excess of Turbulent Kinetic Energy (TKE) when lighter fluid is injected, and due to a lack of TKE when the denser fluid is injected. Lastly, an experimental technique based on the synchronized use of a WMS and a PIV system was developed to measure turbulent scalar fluxes.