Data-driven Identification of Nonlinear Flow-induced Vibration Sources for a Cavitating Propeller Under Ship Wake

The non-uniform ship wake would trigger propeller cavitation dynamics and hence result in severe vibration on the ship hull. The irregular cavity volume oscillation would bring about frequency modulation and coupling to transfer oscillation energy. This work employs cavitation tunnel experiments, numerical simulations, and data-driven modal analysis to conduct a demodulation and decoupling study on the one-dimensional pressure signal and three-dimensional flow fields, which aims to unravel the flow-induced vibration mechanisms of the cavitating propeller. The synchronous experiments of cavitation images and pressure pulses validate the reliability of propeller cavitating dynamics simulation. For the excitation source identification, spectral proper orthogonal decomposition (SPOD) presents statistical energy representations of rotation harmonic frequency modes. Cyclostationary-SPOD (CS-SPOD) identifies the temporal modulation oscillation patterns and the confined cavitation excitation sources under the ship hull. Bispectral mode decomposition (BMD) results suggest the cavitation would introduce an outstanding line spectrum coupling on the three-dimensional flow fields, and the interaction maps directly characterize the spatial transfer path of oscillation energy. As the demodulation and decoupling analysis illuminated, the flow fields of non-cavitation, incipient cavitation, and developed cavitation present linear, weakly nonlinear, and strongly nonlinear effects, respectively. The identification of propeller cavitation-induced excitation sources contributes to the vibration reduction of the ship hull. The methods adopted in this work show great potential to resolve the cross-frequency coupling of periodic flows in turbomachinery.