Effect of large-eddy simulation fidelity on predicted mechanisms of jet noise reduction

The aeroacoustic control of a Mach 1.3 turbulent jet demonstrates an unexpectedly subtle importance of accurate numerics for predicting and reducing turbulent jet noise. We observe that two nominally high-wave-number filters (as used to stabilize large-eddy simulations), one explicit that is optimized based upon an integral criterion and the other implicit with a tunable parameter, lead to significant changes in the near-field turbulence with concomitant changes in the acoustic field on the order of 2-4 dB. An apparent consequence is that an adjoint-based optimization procedure identifies different controls for the two simulated jets despite similar reductions of radiated sound. This suggests that sound-generating mechanisms of the Mach 1.3 turbulent jet may be more fundamentally modified by subtle changes in numerics than previously expected.