A numerical study of the wall-modeled large-eddy simulations of attached and separated flows with a semi-coupled synthetic turbulence generator

The predictive accuracy of the fluctuations inside turbulent boundary layers directly relies on the upstream turbulence synthesizer in wall-modeled large-eddy simulations (WMLESs). The conventional fully coupled synthetic turbulence generator (STG) strongly depends on the grid topology and simultaneous advancement of the unsteady Reynolds-averaged region. In this study, the two-stage semi-coupled STG, which is tolerant to the grid topology and is cost-effective due to the compatibility with precursor methods, was investigated. First, the semi-coupled STG efficiently generated artificial turbulence based on the existing benchmark data in the calculations of the fully developed channel flow. The adaptation length of the STG was less than ten boundary layer thicknesses by means of the Reynolds stresses and wall shear stresses, while the downstream turbulence was sufficiently "mature" for the predictions of more complex flows. The semi-coupled method with different wall-modeling strategies was validated for the mildly separated flow over a flat strut. The steady two-dimensional (2D) precursor data at the inlet were effectively utilized by the STG to generate turbulence and the monitored pressure fluctuations agreed well with the experimental data. Finally, the calculations of the flow around a wing-body junction showed that the algebraic wall-modeled large-eddy simulation demonstrated good accuracy in predicting the horseshoe vortex at the symmetry plane, and coherent structures rapidly developed from the artificial turbulence generated by the STG. To conclude, the WMLES with the semi-coupled STG can effectively predict the unsteady fluctuations in the attached and separated boundary layers. The compatibility of the semi-coupled STG with precursor methods and the relatively low computational cost of the WMLES can broaden the application areas of the STG method in complex flows at high Reynolds numbers.