Effects and mechanisms of LES and DDES method on airfoil self-noise prediction at low to moderate Reynolds numbers

The Large Eddy Simulation (LES) method and Delayed Detached Eddy Simulation (DDES) method combined with the Ffowcs Williams and Hawkings (FW-H) analogy are used to predict the self-noise of NACA 65(12)-10 airfoil at low to moderate Reynolds numbers. The computational results have been validated through the comparison with experimental data and NAFNoise predictive data. The results indicate that both LES and DDES methods can output correct sound pressure level (SPL) spectra at a Reynolds number of 2 x 10(5), but only the LES method is capable of providing reasonable SPL spectra at a Reynolds number of 5 x 10(5), which the DDES method has failed to do. The broadband noise below 10 kHz obtained from the DDES method is lower than that from NAFNoise prediction by up to 50 dB. Further analysis shows that the DDES method fails to capture the separation and reattachment on the suction side near the trailing edge. Besides, the magnitude of pressure fluctuation on the airfoil surface obtained from the DDES method is far smaller than that from the LES method, providing incorrect acoustic source information for the FW-H equation. In the near-wall region for the DDES case, the Reynolds-averaged Navier-Stokes method is used to compute the unsteady flow field, providing too large viscosity and too small Reynolds stress near the airfoil surface, which makes it difficult to capture the influence of small-scale eddies in the flow field. The characteristics of the DDES method above could cause errors in simulating the unsteady flow field at a Reynolds number of 10(5) range. On the contrary, the LES method could give better prediction of broadband noise under different incoming flow conditions.