Large-scale clustering of coherent fine-scale eddies in a turbulent mixing layer

Abstract

Clustering of coherent fine-scale eddies in a turbulent mixing layer has been analyzed by using direct numerical simulation (DNS) data at Reλ ≃ 250. The coherent fine-scale eddies are defined based on the second invariant of the velocity gradient tensor and the vorticity vector. The clustering is evaluated by the number density of coherent fine-scale eddies, and the large-scale structures are extracted by low-pass filtered velocity fields. Conditional averaging shows that the large-scale enstrophy increases with the number density, whereas the large-scale strain rate stays around the average in the high number density region. The alignments of the vorticity vector and the eigenvectors of the large-scale strain rate tensor are conditioned by the number density or the strain rate magnitude. The eigenvectors and the vorticity vector indicate strong preferential alignments under the intense large-scale strain rate condition. On the other hand, those alignments become weak in the high number density regions. The inter-scale energy transfer between grid and subgrid scales is significantly correlated with the magnitude of the large-scale strain rate while there is no apparent correlation with the number density.