Studies in wall turbulence using dual plane particle image velocimetry and direct numerical simulation data.

Abstract

Wall-bounded turbulent flows play a critical role in a variety of engineering applications. A detailed understanding of the fundamental processes underlying these flows is crucial to the accurate modeling and design of efficient and practical systems. The present studies are directed towards demonstrating the utility of experimental and numerical approaches in tandem to elucidate the structure and dynamics of wall-bounded turbulent flows over a range of Reynolds numbers. The first part of this thesis deals with the use of Dual Plane Particle Image Velocimetry (DPPIV) to understand the organization of vortex structures in the logarithmic region of turbulent boundary layer and channel flows. DPPIV provides the full velocity gradient tensor in a plane parallel to the wall and was used to calculate the projection angles of vortex structures with the three coordinate directions. In order to validate the experimental technique for identification of vortex cores, Direct Numerical Simulation (DNS) data at a comparable Reynolds number and higher resolution were used. The DNS data were averaged to the resolution of the DPPIV data and a vortex core identification routine was implemented to compare the raw DNS, averaged DNS and DPPIV data. It was observed that results from the DPPIV data match those from the raw and averaged DNS data very well. This confirms that DPPIV is a robust technique for calculating vortex core statistics, and the resolution of the measurements is sufficient to adequately resolve these structures. The second part of the thesis examines the effect of Reynolds number on the scale energy budget in wall-bounded turbulent flows. An understanding of the distribution of turbulent kinetic energy across the momentum deficit region in wall-bounded turbulent flows is critical to the complete understanding of the energy dynamics in these flows. The scale energy budget provides a tool to simultaneously assess the influence of spatial location and scales in the flow on the distribution of turbulent energy. This analysis was conducted using three DNS data sets across a range of Reynolds numbers, and results from these data were compared to results from an earlier study at a smaller Reynolds number. It is observed that the previous low Reynolds number study did not sufficiently resolve all the quantities of the energy budget in the near-wall region, primarily due of the lack of a distinct logarithmic region in the low Reynolds number simulation. Close to the wall, no effects of Reynolds number were observed on the terms of the scale energy analysis across the datasets studied. Upon moving away from the wall, the turbulent production term did not display any effects of Reynolds number, while the scale transfer term increased with increasing Reynolds number. As a result the cross-over scale, a quantity related to the shear scale in turbulent flows, increased with increasing Reynolds numbers for all datasets. The shear scale provides the scale at which the change from an isotropy-recovering range to an anisotropic region is observed, while the cross-over scale provides a transition between a transfer dominated range to a production dominated range of scales. The plot of cross-over scale versus wall-normal location revealed that the slope of the best fit lines increased with increasing Reynolds number. Further, the slope of the best fit line decreased with increasing wall-normal location. DPPIV data were also used to conduct the same analysis in turbulent boundary layers across a range of Reynolds numbers. The aim of using DPPIV data was to quantify the effects of Reynolds number on the scale energy analysis, using larger Reynolds number experimental data. The effects of resolution and Reynolds number on the DPPIV data were assessed and described in detail. The effect of resolution was most dominant on the wall-normal gradients of the streamwise and spanwise velocities, and hence the transfer of energy in physical space and the transfer of energy in scale space were overestimated in the lower resolution data. The effect of resolution was smallest on the turbulent production term, since it does not contain of any fluctuating velocity gradients. With increasing Reynolds numbers, the production term did not change significantly in the logarithmic layer, while the scale transfer term increased, resulting in a larger range of transfer-dominated scales. The value of the cross-over scale between the effective production and scale transfer term increased with increasing Reynolds number, suggesting a larger range of isotropic-type transfer dominated scales. In conclusion, it was demonstrated that DPPIV in tandem with DNS can be used to reliably assess the scale energy budget in wall-bounded turbulent flows. i