Comparison of image masking methods for pump impeller blade region PIV experiments

Due to the blade reflection effect when applying the PIV technology to the internal flow field measurement of rotating machinery, the blade area requires special treatment. In this paper, zero-value, particle image, random number, and mean-value masking methods are compared and analyzed based on particle images from the PIV challenge and rotating impeller. The peak significance and coordinates deviations are used as the masking effect evaluation criteria. The results indicate that an increase in the interrogation window masking area proportion causes a weakening of the PIV calculation particle cluster displacement peak value. The peak position is shifted, which leads to the wrong velocity vector. The masking area proportion should not be higher than 0.7, 0.5, 0.19, and 0.22 for mean-value, random number, zero-value, and particle image masking methods, respectively. The results show that flow field calculation results near the blade boundary are more accurate for mean-value masking.