Development of a computational platform for the simulation of low Prandtl number turbulent flows

Mathematical modeling of physical phenomena is at the basis of many scientific field researches. Complex systems show multiscale and multiphysics aspects that cannot be always taken into account in detail. In the past many numerical codes have been developed and specialized to solve different aspects of turbulence and, in general, fluid motion for a very wide range of engineering applications. Nowadays, numerical code coupling and computational platforms are gaining a lot of interest for the simulation of very complex phenomena. This PhD study focuses on modeling physical systems with coupled simulations, in particular turbulent heat transfer for liquid metals. This type of fluids, known as low Prandtl number fluids, requires more sophisticated turbulent heat transfer models since those used to simulate fluids such as air or water lead to a sensible heat transfer overestimation. Seeking an increased numerical stability, a four logarithmic parameter turbulence model is proposed, starting from a model that has already been validated with simulations of Lead-Bismuth-Eutectic (LBE) fully developed turbulent flows. The turbulence model has been implemented in the finite element code FEMuS to perform an extensive validation by comparing obtained results with Direct Numerical Simulations and experimental data. Many simulations are performed, for fully developed turbulent flows in plane channels, cylindrical pipes and 19 pin nuclear reactor bundles and for turbulent forced and mixed convection over a backward facing step. When conservation equations of mass, momentum and energy need be coupled with dynamic two-equation or thermal turbulence four-equation models the use of numerical coupling becomes important. In order to dispose of a greater choice of dynamical turbulence models, a computational platform containing OpenFOAM and FEMuS codes has been developed.