The effect of particle image blur on the correlation map and velocity measurement in PIV

In PIV particle image blur is usually observed near fluid optical interfaces, i.e. shock waves, and thin flow structure with large density variations, e.g. shear layers and boundary layers. In such an environment the particle image is not only subject to blur, but is also displaced from its actual position due to refraction, which is denoted as optical displacement. In this study particle image blur near a shock wave is investigated in relation to the auto- and cross-correlation map, measurement accuracy and confidence level. The results from a numerical study are supported by PIV measurements of a shock wave in a supersonic wind tunnel. It is demonstrated that particle images are blurred in the direction of lower refractive index (directional blurring). The particle images are also skewed. Therefore particle image blur not only causes correlation peak broadening due to the fact that the particle images increase in size, but more importantly can introduce an asymmetry in the correlation peak and in turn introduce a small bias error in the measured velocity. However, experimental results indicate that particle image blur itself is not the main cause for the increase in measurement uncertainty near shock waves, but that the reduced accuracy can be attributed to the optical displacement. The observation of particle image blur can be used as a detection criterion for a qualitative assessment of the optical displacement. Certain combinations of experimental parameters (viewing angle, f/# and interrogation window size) yield significant errors in the measured velocity. Under certain circumstances optical distortion can become so strong to introduce an unphysical acceleration within the shock wave, visualized as an inflection point with positive slope in the velocity profile across the shock. The study provides some practical suggestions to limit the effect of aero-optical distortion on the velocity measurement.