Aeroacoustic source diagnosis for turbulent subsonic jets using Canonical Correlation Decomposition

Common flow decompositions methods such as POD are widely used to diagnose the main aeroacoustic sources of turbulent jets. But the most energic modes extracted from POD are not necessarily the most acoustically-dominant flow structures. In this work, we use the recently-proposed Canonical Correlation Decomposition (CCD) method to probe the flow structures that are most correlated with acoustic pressure fluctuations of a turbulent subsonic jet. Both far-field and near-field pressure fluctuations are used as observables. In the near-field case, it is found that CCD is able to extract the well-known large-scale structures connected with K-H instabilities, which agrees current understanding on the jet’s near-field dynamics. The resulting CCD spectrum shows a significant low-rank behavior, and the near-field pressure fluctuations can be efficiently reconstructed using only the first two modes. In the far-field case, CCD spectrum also exhibit a much quicker decay. The leading-order modes are of large scale and low frequencies. As the mode number increases, CCD modes are of increasingly short scales and high frequencies. It is shown that some of the resulting CCD modes are directly connected with the spectral features of the far-field sound. It is shown that CCD can be a useful tool for observable diagnosis such as aeroacoustic source diagnosis for turbulent flows.