Data-assimilation of transitional flows to predict aerodynamic instabilities of airfoils at low angles of attack

Overview & Synopsis

The laminar-turbulent transition is known to be a particularly challenging physics to capture for Reynolds-Averaged Navier-Stokes models. We use here the Spalart-Allmaras turbulence model on a NACA0012 airfoil at Re=50000 and alpha=2° and highlight its inaccuracies with respect to high-fidelity data from DNS. Employing a data-assimilation approach informed with the full velocity field from the DNS we build a correction accurately matching higher fidelity data. We perform a linear stability analysis around the assimilated flow and find a stable branch of high frequency vortex shedding modes. Finally, we try different combination of correction-observations for the assimilation. We evaluate their efficacy in accurately representing the desired spectrum and we show the significant sensitivity of the high frequency branch to the reconstructed mean flow and its corresponding correction.