Theory and simulation of particle image velocimetry

As particle image velocimetry has been developed as a useful experimental method for measuring instantaneous velocity vector fields in turbulent fluid flows, a systematic and comprehensive study of the procedure to assess its accuracy, spatial resolution, reliability and success rate for known velocity fields, experimental parameters and interrogation procedures has been undertaken. A systematic study of the parameters which affect the measurement of velocity vector fields in fluid mechanics experiments using the particle image velocimetry procedure has been carried out for a range of alternative methods of PIV recording and analysis. This study has developed an analytic model and a Monte Carlo simulation to analyze experimental and interrogation procedures for a range of high image-density PIV systems. The study has provided a number of criteria which are used to determine which experimental and interrogation parameters are most important for a range of fluid motions and a range of alternative methods of PIV recording and interrogation. In addition, it has been shown that velocity gradients within the flow field result in biased experimental measurements. The extent of the bias has been quantified and some recommendations to minimize bias have been made. The original analysis of the double-pulse PIV using either Young's fringe method or direct imaging method has been generalized to multiple-pulse systems where multiple exposures are made on a single recording frame and interrogated by autocorrelation methods. The cross-correlation algorithm for separate interrogation windows of a double-exposure single-frame recording has been modeled to illustrate the features of the method as an alternative to auto-correlation methods, particularly in minimizing bias effects and in improving signal-to-noise ratio. Although photographic data recording offers the highest spatial resolution, direct digital imaging on CCD arrays has improved in resolution with larger arrays being developed. With the generalization from photographic film to CCD arrays in mind, the performance of multiple-frame systems in which single images are recorded sequentially on each CCD frame has been considered and compared with the performance of single-frame systems. The cross-correlation algorithms, which determine the image separation between successive single-exposure frames, have been compared with auto-correlation algorithms.