Post-Processing Methods of PIV Instantaneous Flow Fields for Unsteady Flows in Turbomachines

Abstract

Among the experimental techniques, the particle image velocimetry (PIV) is undoubtedly one of the most attractive modern methods to investigate the fluid flow in a non-intrusive way and allows to obtain instantaneous fluid flow fields by correlating at least two sequential exposures. This technique was successfully applied in several fields in order to study high complex three-dimensional flow velocity fields and to provide a significant experimental data base for the validation of combined numerical analysis models. However, on one side, experimental limits and possible perturbing phenomena could negatively affect the PIV experimental accuracy, altering the real physics of the studied fluid flow field. On the other side, the huge amount of data obtainable by means of the PIV technique requires properly post-processing tools to be exploited in an in-depth study of the fluid-dynamical phenomena. To avoid misinterpretation of the phenomena, complex cleaning techniques were developed and applied at the different steps of the PIV processing, starting from the acquired images (background subtraction, mask application, etc.) so as to increase the signal to noise ratio, and finishing to the instantaneous flow fields by means of statistical methods applied in order to identify residual spurious vectors [Raffel et al., 2002]. Even though all these methods allows to obtain a good filtering of the instantaneous flow fields, however they are not able to completely eliminate all the outliers in the results since the removal criteria are always dependent on the choice of a threshold value [Heinz et al., 2004; Westerweeel, 1994; Westerweel and Scarano, 2005]. To overcome this problem, the most common approach is to average the instantaneous PIV flow fields so as to improve the quality of the resulting flow field reconstruction and to more easily identify the flow field characteristics in the investigated area.