Comparison of PIV data using multiple configurations and processing techniques

Particle image velocimetry (PIV) data have been acquired using three different experimental configurations in the far-field of the interaction created by a transverse supersonic jet exhausting from a flat plate into a transonic crossflow. The configurations included two-component PIV in the centerline streamwise plane at two overlapping stations, as well as stereoscopic PIV in both the same streamwise plane and in the crossplane. All measurement planes intersected at a common line. Data from both two-component measurement stations and the stereoscopic streamwise configuration agreed to within the estimated uncertainty, but data from the crossplane exhibited reduced velocity and turbulent stress magnitudes by a small but significant degree. Subsequent reprocessing of the data in nominally the same manner using a newer software package brought all values into close agreement with each other, but produced substantially higher turbulent stresses. The error source associated with the choice of software was traced to the use of image deformation in the newer software to treat velocity gradients, which is shown by synthetic PIV tests to yield a more accurate result for turbulence measurements even for gradients within the recommended limits for classical PIV. These detailed comparisons of replicate data suggest that routine methods of uncertainty quantification used for a turbulent PIV experiment may not fully capture the actual error sources.