Large eddy simulation of airflow in human vocal folds

Human phonation is a complex physiological process involving flow-induced oscillations of the vocal folds and aeroacoustic sound generation. The flow fields encountered in phonation are highly unsteady, feature massive flow separation and recirculation in the supraglottal spaces and generation of coherent vortex structures from the shear layer of the jet. For the sake of computational aeroacoustic modeling of human voice generation, an accurate resolution of the airflow through the vocal folds is essential. The Reynolds-averaged Navier-Stokes turbulence modeling is inappropriate, since it provides only the averaged flow field. The paper presents the first results obtained with a large-eddy simulation of flow through a model of human vocal folds using a second-order finite volume discretization of incompressible Navier-Stokes equations. In the first step, the flow field was resolved on a fine 2D mesh covering a short subglottal region, the glottis and a part of the supraglottal channel. The simulation was parallelized using domain decomposition method and run in parallel on a shared-memory supercomputer. The results compare two large eddy simulations using the algebraic Smagorinsky and one-equation sub-grid scale models against a simulation without turbulence model