Using Data-Driven Reduced Order Methods as a Post Processing Tool to Understand Cavitating Flow Induced Noise

Wakes behind bluff bodies are characterized by the presence of a Kármán vortex street in the far wake region, and an associated vortex shedding frequency as discussed in [1]. A cavitating wake behind a bluff body can experience significant change in the dynamics of vortex shedding depending upon the extent of cavitation. The shedding frequency of a cavitating wake can increase from a non-cavitating condition to attain a peak value, followed by a reduction in shedding frequency of a super-cavity, as discussed in [2]. It has been recently shown in [3] that in such cavitating bubbly flows, the effect of compressibility is important, and the shedding dynamics change when the local wake Mach number approaches unity, as shown in Figure 1. In this study, the acoustic noise of a cavitating wake at conditions corresponding to different shedding frequencies are presented. In addition, flow structures responsible for the observed changes in the acoustic behavior at different frequencies are identified by performing Proper Orthogonal Decomposition (POD), and Dynamic Mode Decomposition (DMD) on time synchronized time-resolved void fraction fields. Based on the findings, the effect of compressibility on the observed cavitation noise spectrum is presented.