A-posteriori assessment of mixed models for Large Eddy Simulation of polydisperse multiphase flows

This study presents a detailed analysis of various turbulence modeling approaches for the convective subgrid-scale term in the context of Large Eddy Simulation (LES) of a polydisperse two-phase flow. The numerical framework is based on the Volume-of-Fluid (VOF) method. The presented a-posteriori investigation assesses the potential for mixed modeling combining the well-known eddy viscosity model of Smagorinsky and the scale similarity type model of Liu in two different ways. The first approach employs a dynamic blending function based on a subgrid activity sensor, whereas the second approach is based on an explicit scale separation. A preceding analysis of the laminar-turbulent transition for the Taylor-Green vortex reveals that, compared to the standalone formulation of the models, a combination of a functional and a structural model results in a more accurate prediction of the turbulent kinetic energy and its dissipation rate. The main focus of this work is on the simulation of a gas injection into a liquid-filled domain. This study presents a first investigation of advanced mixed LES models for polydisperse bubble-laden flows characterized by a realistic water-to-air density ratio. The distributions of the gas volume fraction and the gas phase velocity obtained in the LES cases employing the mixed models show a good agreement with reference data from both numerical and experimental investigations. These findings indicate that mixed type turbulence modeling is a promising candidate for an efficient and accurate prediction of industry-scale multiphase flows. • Large Eddy Simulation of a turbulent two-phase flow using the Volume-of-Fluid method. • Combined functional and structural models for good alignment and stability. • Taylor–Green vortex indicates superior performance for mixed models. • Mixed modeling demonstrates improvement for various statistics in multiphase flows.