Aeroacoustic Source Diagnosis of Subsonic Jet Noise Using Canonical Correlation Decompositions

The recently developed canonical correlation decomposition (CCD) is used to diagnose the aeroacoustic sources of a subsonic jet. The method is applied to a numerical jet dataset obtained using high-fidelity large eddy simulations (LES). The acoustic pressure at $x=0$ and $r=4D$ is used as the observable. We first use the pressure and streamwise velocity as the flow variables to find the coherent pressure/velocity structures that are most correlated with the acoustic pressure. We then use Lighthill's stress tensor as the flow variable to examine the dominant equivalent source structures that are most correlated with the acoustic pressure. Results show that the decomposition of pressure and velocity exhibits good low-rank behavior. The leading mode for both fields is a large-scale coherent structure. Crucially, the second and third CCD modes for pressure manifest as well-defined wave packets. Furthermore, the decomposition of normal stress tensors exhibits more pronounced low-rank characteristics than that of shear tensors. Notably, the streamwise normal stress ($T_{uu}$) is the most dominant component that is correlated with the acoustic pressure. Its leading mode is characterized by a large-scale, axisymmetric structure located downstream of the potential core. The application of CCD effectively extracts dominant, large-scale structures correlated with the acoustic pressure, offering clear insights into jet noise mechanisms.