Numerical studies of turbulent flows
Maneesh, Kumar Mishra
Date of Issue2015
School of Mechanical and Aerospace Engineering
Turbulent flows are investigated using numerical simulations. Three aspects of turbulent flows are the focus points for the present work. Firstly, turbulent drag reduction under the effect of spanwise wall oscillations for a boundary layer has been studied using direct numerical simulations. Previous studies focused on channel or pipe flows which are numerically easier due to periodic boundary conditions. Here, we use a novel technique with discontinuous zone of oscillation to study the effect of starting and ending oscillations on boundary layer flows. Spatial transient of Reynolds stresses have been reported for temporal forcing. A predictive relation for deprecation of drag reduction performance with Reynolds number has been proposed. Steady spatial half-square waves have been shown to be energetically more efficient due to lesser power consumption. Optimization of parameters yield ~ 18% net energy savings. Secondly, the effect of wall oscillations on low speed streaks and transition region are investigated. Previous studies have used linear approximation to study wall oscillation effects on streaks whereas in the current work, non-linear interaction terms are included which show a marked difference in results. The wall oscillations are shown to reduce the skin friction which drops below the laminar Blasius flow value (without the presence of streaks) for certain cases of wall oscillations. Influence of wall oscillations on bypass transition has been investigated with both spatial and temporal forcing. A more pronounced delay has been observed with spatial oscillations. Effect of starting position of oscillation, wavenumber and turbulence intensity on skin friction coefficient has been studied. Thirdly, for multi-scale analysis of turbulent flows, a new methodology based on novel optimization techniques for scale decomposition is introduced. The optimization procedure leads to a band-pass filter with prescribed properties in the Kolmogorov energy spectrum. With this filter, scale decomposition using Fourier transform can be performed very efficiently while adequately suppressing Gibbs ringing artifacts. Continuous scale decomposition is particularly important to understand the dynamics of eddies across different wavenumbers. Compared with previous work on multi-scale turbulent flow analysis and visualization, the proposed method enables flexible and efficient turbulence decomposition in a continuous manner.