Data assimilation of rotor flow at hovering state using ensemble Kalman filter

To accurately predict the characteristics of the three-dimensional flow fields of rotors, the constants of the shear stress transport model were optimized via data assimilation (DA) using the ensemble Kalman filter algorithm. The unsteady velocity fields of two hovering rotors with a diameter of 300 mm and a speed of 3000 rpm were measured via phase-locked particle image velocimetry, and the mean field was used as the observation. The flow field results for four cross sections (r/R = 0.25, 0.5, 0.75, and 0.93) and two vertical sections (\(\varphi = 0^\circ\) and \(270^\circ\)) were recorded. Two kinds of DA strategy were adopted: DA based on observations of vertical velocities on the vertical section or of horizontal velocities on the cross section. The different DA strategies were assessed via a detailed comparison of their global flow field prediction performances. Among the DA strategies based on vertical-section observations, the strategies based on vertical velocity observations at x/R =  − 0.93 (VS III) and x/R =  − 0.5, − 0.75, − 0.93 (VS I–III) exhibited the best and second-best performances, respectively. The DA strategies with cross-sectional observations exhibited comparable results to the DA strategies with vertical-section observations. The flow field prediction was modulated by all model parameters. The most accurate predicted eddy viscosity was higher than that predicted by the original model, correcting the overadjustment of the original model. This study provides a reference for the correction of the overprediction of boundary layer flow separation under adverse pressure gradients.