2010 Annual Meeting

(522e) Evaluation of SGS Scalar Flux Models in a Turbulent Liquid Flow

Authors

Juan M. Mejía - Presenter, Universidad Nacional de Colombia
Alejandro Molina - Presenter, Universidad Nacional de Colombia
Farid Chejne - Presenter, Universidad Nacional de Colombia
Camilo Parra - Presenter, Universidad Nacional de Colombia
Amsini Sadiki - Presenter, Technischen Universität Darmstadt
Pradeep Pantangi - Presenter, Technischen Universität Darmstadt


It is well known that turbulent mixing plays an important role in various engineering and environmental applications. According to experimental findings, the assumption of isotropy at inertial and dissipation scales is no longer valid for structured functions and derivative skewness of the scalar field occurs in the presence of a mean scalar gradient. Different models have been proposed in order to take into account anisotropy, most of them were developed for gas-phase, and their application to fluids with high Sc numbers was not successful. In the case of Stirred Tank Reactors (STR), turbulent anisotropy plays an important role in the impeller stream, boundary layers and at the separation points, so the use of isotropic-based turbulence models should be treated with caution. In this work, a Large Eddy Simulation (LES) of a mixing process in liquid phase was carried out following three different approaches for the subgrid-scale (SGS) scalar flux models for closing the filtered scalar balance equation: eddy diffusivity model with both constant and dynamically-calculated turbulent Schmidt number (ScT) and the dynamic anisotropy model. The later model is cubic in terms of the scalar gradients and combines the eddy diffusivity model with two additional terms. Simulation results were validated with experimental data available in the literature of a turbulent liquid round jet. It was found that all three models reproduce the autosimilar behavior observed in jet flows, but that the simulation with the anisotropic model (see Figure 1) reproduced best the experimental data for the concentration fluctuations across the jet; however the computational time was increased by 70 % in comparison to the linear eddy diffusivity model. Figure 1. Fluctuation concentration distribution across the jet. Experimental data from Antoine et al (2001) (symbols): + x/D = 70; ● x/D = 80; × x/D = 90. Simulation results (lines): Solid: Eddy; dashed: dynamic; dotted: Anisotropic. Antoine, Y., Lemoine F., Lebouche M., (2001). Turbulent transport of a passive scalar in a round jet discharging into a co-flowing stream European Journal of Mechanics - B/Fluids, 20, 275-301.