Performance comparison of particle tracking velocimetry (PTV) and particle image velocimetry (PIV) with long-exposure particle streaks

Abstract

Abstract Particle tracking velocimetry (PTV) and particle image velocimetry (PIV) are popular experimental methods to quantitatively measure flow fields. In many practical applications, hardware limitations result in longer exposure times, causing particle images to elongate into particle streak images. In this study, the performances of PTV and PIV in relation to particle streak images are evaluated systematically by means of both synthetic and experimental images. For the synthetic images, particle streak images are created via the integration over time of the standard Gaussian approximation of particle images, plus the effects of exposure time, and parameters such as particle image diameter, intensity, and density were investigated in terms of the velocity fields of a 1D uniform flow and the rotational 2D Hill’s vortex. The results show that PTV performs well for short exposure times, and its peak-finding criteria can be verified for nondimensional exposure times E T ∗ up to 10. As the E T ∗ increases from 10 to 70, the reliability of the PTV algorithm drops significantly, while the yield drops only slightly. Longer exposure times cause an increase in the number of RMS displacement errors; the PTV algorithm is more likely to fail when particle diameter, image intensity and particle density are larger. The displacement error of PTV RMS ranges from 0.064 pixels to 0.157 pixels. In comparison to PTV, PIV is more robust in relation to streak images with long-exposure times, and not sensitive to the effects of particle image diameter and intensity. The RMS dispersion errors for PIV range from 0.015 pixels to 0.316 pixels. A comparison of the RMS displacement errors exhibited by PIV and PTV shows that where the seeding of particles is not an issue, PIV is more robust, and can handle long-exposure streak images. The results of experimental streak images show that PIV produces lower RMS dispersion error values at low Reynolds numbers, whereas it produces more significant errors for high Reynolds numbers. At regions of low seeding density, PTV can resolve local fluid motions with superior accuracy, but it is vulnerable to high exposure time. Based on these results, where a longer exposure time is a requirement for an application, the use of PIV is recommended, by virtue of its robustness to image streaking, provided that a good seeding and high particle image density are available. For applications such as some microscale flows or flows with separation or recirculation regions, PTV is more capable of fulfilling their requirements.