Combining PIV, POD and vortex identification algorithms for the studyof unsteady turbulent swirling flows

Abstract

Particle image velocimetry (PIV) measurements are made in a highlyturbulent swirling flow. In this flow, we observe a coexistence of turbulentfluctuations and an unsteady swirling motion. The proper orthogonaldecomposition (POD) is used to separate these two contributions to the totalenergy. POD is combined with two new vortex identification functions,Γ1 and Γ2. These functions identify the locations of thecentre and boundary of the vortex on the basis of the velocity field. The PODcomputed for the measured velocity fields shows that two spatial modes areresponsible for most of the fluctuations observed in the vicinity of thelocation of the mean vortex centre. These two modes are also responsible forthe large-scale coherence of the fluctuations. The POD computed from theΓ2 scalar field shows that the displacement and deformation of thelarge-scale vortex are correlated to these modes. We suggest the use of sucha method to separate pseudo-fluctuations due to the unsteady nature of thelarge-scale vortices from fluctuations due to small-scale turbulence.