Desenvolvimento de uma modelagem para escoamentos reativos em malhas adaptativas do tipo bloco-estruturada

Abstract

In most technical applications the flows are turbulent. The use of Navier-Stokes equations for modeling fluid flows allows its characterization in a detailed and precise way, which results in a wide quantity of data. For this reason, the direct resolution of the resulting system of equations for practical cases becomes impossible. As a consequence, Large-eddy Simulations appears as an interesting tool for the characterization of problems involving turbulent and reactive flows. The choice of an appropriate numerical scheme for the advective term of the transport equations description is also very important, as it is directly related to the representation quality of the studied phenomenon. The use of total variation diminishing (TVD) schemes results in oscillation-free solutions, in which problems such as numerical instabilities and generation of unrealistic values are circumvented. In the present work, such schemes were evaluated in canonical situations, such as stationary advection-diffusion and transient linear advection, as well as in complex flows, such as backward-facing step and turbulent jet. According to the transport equations for species and thermal energy, the performed implementation allows the numerical simulation of reactive flows with complex mechanisms, without the need of using the equality hypothesis between thermal and mass diffusivities. The present work is based in the following approaches: Discretization of the fluid dynamics with the use of Finite Volumes Method, flow description based on Large eddy simulations, with the dynamic Smagorinsky turbulence closing model and turbulent inlet conditions, application of TVD schemes for the treatment of the Navier-Stokes equations advective term and, lastly, a library coupling, in order to update all properties related to reactive flows.The use of Random Flow Generation, as a turbulent inlet condition, resulted in a substantial improvement in the characterization of the studied problem when it was compared with white noise or without any source of perturbation. Regarding the advective schemes, CUBISTA methodology was the best option between the evaluated schemes, not presenting substantial dispersion or numerical diffusion. The use of an external library to update thermodynamic, kinetic and transport properties was considered feasible to describe relevant data in the study of reactive flows. Lastly, the use of a formulation based on species diffusivities in the mixture, instead of applying a unity Lewis number modeling, resulted in substantial differences in species properties calculations, which incapacitates the use of Lewis modeling for all situations.