Transition of counter-rotating plasma-induced vortices into a jet

Two low-frequency-modulated plasma actuators are symmetrically positioned on either side of a thin flat plate to continuously generate pairs of counter-rotating vortices, with velocity fields captured using time-resolved particle image velocimetry (PIV). Convolutional neural networks with a U-Net architecture are adopted to generate the phase-averaged velocity field of plasma-induced vortices from a single PIV snapshot to eliminate the requirement of multiple measurements. The influence of the number of input features and samples on the model accuracy is examined. The model-predicted results match well with the measurements, accurately restoring the vortex dynamics and capturing the strength variation. The proposed model is then utilized to reconstruct the phase-averaged results of the starting vortex for continuous plasma jets, revealing that the evolution of plasma-induced vortices and their transition into a jet are characterized by four distinct stages: formation, boosting, distortion, and jetting, all governed by the vortex convection velocity. The vortices exhibit a marked increase in circulation after a delayed development, moving linearly downstream while diverging from the centerline. Consequently, the vortex-induced effect significantly enhances the centerline velocity. The vortices are subsequently stretched and distorted from a circular into a chain-like shape due to the large velocity gradient between the vortex pair, leading to the vortex breakdown and transition into a jet, accompanied by a collapse in the velocity magnitude. Insights into vortex-to-jet transition inform the optimal placement of plasma actuators, thereby enhancing control efficiency in active flow control applications.