Numerical Predictions of Mixing Phenomena in Turbulent Variable Density Flows with Second Order Closure Turbulence Models

Abstract

This work is devoted to the prediction of turbulent variable density flows which is considered as an important step towards the prediction of turbulent flames. Indeed, this step is of particular interest as far as many practical devices are often characterized by a complex aerodynamics: recirculation zones, impinging jets, swirling flows, … that determines the efficiency of the fuel-air mixing. Combustion modifies the dynamic field but in general, the flow pattern is qualitatively recovered between the case of inert mixing of the reactants and the reactive flow. In this context, tests of the turbulence models which are implemented for three dimensional computations of industrial applications [1] in non reactive flows may give precious indications for the simulation of the corresponding flames; in particular about the widely used Boussinesq assumption for the scalar turbulent flux and the assumed proportionality law between the characteristic time of the scalar dissipation and that one of the turbulent kinetic energy. These questions are addressed here through numerical predictions of two basic turbulent variable density flows. The first one is an axisymmetric turbulent jet (diameter D= 26 mm) surrounded by a low velocity (0.9 m/s) coflow of air of diameter 285 mm. The central jet is fed by slightly heated air, carbon dioxide or helium. For this latter case, density variations are of the order of magnitude of those encountered in flames. The inlet momentum is kept constant for the three experiments (see Table 1). Detailed measurements [2,3] of the dynamic and scalar fields, including probability density functions, were performed at the IRPHE (Marseille). The second study is related to a more complex turbulent flow.