James G. Brasseur
Department of Mechanical Engineering, Pennsylvania State University
April 28, 2009
Foothills laboratory 2, Room 1022
Application of Interactive Visualization-based Data Interrogation to Explore Local Dynamics of Vorticity in Shear-dominated Turbulence
The local dynamics of small-scale turbulence structure has a strong impact on clustering of particles, aerosol dynamics, cloud formation, reaction-rate chemistry, etc. Small-scale evolution centers on the interactions between local concentrations of vorticity and strain-rate fluctuations modulated by mean gradients. I shall discuss the incorporation of an interactive visualization-based data analysis environment into the statistical analysis of the local dynamics that causes isotropic small-scale turbulence to transition to shear-dominated small-scale turbulence. The data are from direct numerical simulations of initially isotropic turbulence under mean shear. The novel aspect of the analysis environment is the interactive integration of visualization with quantification of the interactions between individual vorticity and strain-dominated structures. Combined visualization-quantification originates from the “extraction” (i.e., identification of surface coordinates) of individual “structures” based on concentration, and the separation of the vorticity and strain-rate fields into objectively defined higher vs. lower intensity structures surrounded by low intensity fluctuations. Unlike traditional thresholding, our extraction algorithm includes lower-magnitude content that is part of the coherent structure. The structure-extraction algorithm was automated, and also integrated within an interactive environment where the user can extract potentially interesting structures by hand. We used the latter to follow the development of a single hairpin vortex in the shear-dominated state backwards in time to learn its origin in the isotropic state and the local dynamics that created it. The visualization of structures is integrated with statistical quantification. I shall present a number of interesting results from the integration of the visualization-based analysis environment with more classical statistics.