Method of data-driven mode decomposition for cavitating flow in a Venturi nozzle

Cavitation is a common phenomenon that occurs in ocean engineering. In the present work, experimental measurement of cavitating flow in a Venturi nozzle is carried out at cavitation number σ = 1.5, and then the simulation based on the Zwart cavitation model is conducted with validation of experimental data. Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD) are introduced to investigate the three-dimensional coherent structures of cavitation and flow fields. For decomposition of vapor volume fraction, both POD and DMD results can characterize cavitation structures and associated frequencies, while different dominant structures can be obtained. For decomposition of streamwise velocity, POD results extract the coherent structures with converse polarities induced by the vortices related to cavity evolution, and DMD results shed light on the structures related to the overall periodic dynamics of cloud cavitation under shedding frequency. Based on the mode decomposition of vapor volume fraction and streamwise velocity, the cavitation-velocity interaction is revealed. Under the studied cavitation condition, cavitation and velocity mainly interact in the range of 0-3Lref downstream the Venturi throat, where the modes of vapor volume fraction and streamwise velocity show similar coherent structures and mode value fluctuations.