Twin experiments for data assimilation of cavitating flow around a hydrofoil

Abstract

Twin experiments are conducted to clarify whether assimilation effects can be achieved by data assimilation (DA) with measurement data obtained by existing cavitation flow measurement techniques. The analysis object is the cavitation flow around a Clark-Y11.7% hydrofoil. The pseudo-measurement data are velocity fields from tomographic particle image velocimetry (TPIV) or two-dimensional PIV, both containing missing data in cavity, along with indirectly obtained cavity interface shapes. A large-eddy simulation is used for unsteady simulation of cavitating turbulent flow, and a local ensemble transform Kalman filter is used as the DA method. Visualized flow fields of the ensemble mean show that characteristic phenomena of the pseudo-measurement data are qualitatively reproduced. In addition, the time-series data at an observation point located at the position where the pseudo-measurement data exists converged following the pseudo-measurement data. The velocity inside the cavity, where no pseudo-measurement data exists, is also complemented by CFD that incorporates information from outside of the cavity. However, the complementation performance depends on the accuracy of the cavitation model. This DA program is applied to the real PIV data of single-phase flow and qualitatively reproduces the flow field. Furthermore, the observation noise is reduced and the data outside the measurement domain are complemented.