Session L21: Turbulent Boundary Layers VIII: Rough Walls II

Roughness effects on the control of laminar separated boundary layers

The effect of roughness positioned upstream of a separation bubble is studied by direct numerical simulation using a high resolution numerical scheme, and the immersed boundary method. The two-dimensional roughness elements have rectangular shapes, and a height $h$ between 0.9 and 1.8$\theta_0$, $\theta_0$ being the inflow momentum thickness. The preliminary 2D studies show a decrease in the extent of the separation bubble as well as the separation position compared with the baseline case without surface roughness. The 3D studies show an earlier transition which eliminates the laminar separation bubble completely. Furthermore, it shows that downstream of the position where the original separation bubble reattaches, the skin friction, $C_f$ follows the same functional behavior as the $C_f$ in the uncontrolled flow, although having a higher value. Further work will be done in trying to optimize the roughness based control, which consists in minimizing the profile losses by finding a balance between the reduction of the extent of the separation bubble and the transition to turbulence.   

Direct numerical simulation of turbulent boundary layer separation under unsteady pressure gradients

Direct numerical simulations of attached, separated, and unsteady separated turbulent boundary layers are performed. Blowing-suction velocity distributions are imposed along the upper boundary to introduce adverse pressure gradients to the turbulent boundary layer. A time varying adverse pressure gradient induces unsteady separation of the turbulent boundary layer. Comparing unsteady and steady separated cases demonstrates significant differences in distributions of the average velocity, vorticity, and kinetic energy budget despite similar average wall pressure and skin friction distributions. The behavior of the unsteady separated turbulent boundary layer is described using instantaneous flow visualization, frequency of flow reversal, turbulent kinetic energy budget, vorticity distributions, Reynolds stress events, and auto-correlation of velocity fluctuation in time and space. Complex flow phenomena such as division of the recirculation zone during separation bubble collapse, motion of vortical structures into and over the bubble, and characteristics of detachment point and reattachment point variation in stream and span are revealed.   

Turbulent flow around a wall mounted square cylinder: evaluating the LES and Reynolds Stress turbulence models in predicting the negative turbulence productions upstream the obstacle

The airflow field around a square cylinder was simulated using the Reynolds Stress Turbulence Model (RSTM) as well as the Large Eddy Simulation (LES). Particular attention was given to the case with Reynolds number of 5610 for which the Direct Numerical Simulation (DNS) data of Yakhot et al (2006) were available. They found that the unsteadiness due to the unstable interaction of the flow at the upstream and the sides of the cube generated the vortex shedding downstream the cube. Also, they argued that the inaccuracy of some of the RANS and LES turbulence models was because of their inability to predict the negative turbulence production in front of the cube, which was the source of the horseshoe vortex. Therefore, in this paper the accuracy of the RSTM and LES turbulence models in predicting the turbulence production at the upstream of the wall mounted square cylinder was investigated. The nature of the 3D wakes behind the cube as well as the vortices in front and at the back of the cube were analyzed. The simulation results were compared with the DNS data and the accuracy of the two turbulence models were underlined.   

Large eddy simulation of channel flows with strip roughness

We describe a large-eddy-simulation (LES) study, at Reynolds number up to $Re_\tau=O(10^{8})$, of turbulent flow in a long channel with the walls consisting of roughness strips. The strips are oriented either parallel or perpendicular to the flow resulting in repeated transitions of smooth and rough surfaces in the span-wise or stream-wise directions respectively. The present LES uses a wall model which contains Colebrook's empirical formula as a roughness correction to both the local, dynamic calculation of $u_\tau$ and also the LES wall boundary condition. This operates point-wise across wall surfaces, and hence changes in the outer flow can be viewed as a response to the overall roughness distribution. The results indicate that for the strips perpendicular to the flow, the recovery of the flow beyond a smooth to rough transition is faster than that of a rough to smooth transition.   

Direct numerical simulations in turbulent boundary layers over cube-roughened walls with varying spanwise spacing

Direct numerical simulations of turbulent boundary layers over three-dimensional cube-roughened walls were performed to investigate the effects of the spanwise spacing ($p_{z}/k)$ on the turbulent statistics. The spanwise extent between the cubes was varied $p_{z}/k$=2, 3, 4 and 6 at the fixed streamwise extent ($p_{x}/k$=3), where $k$ is the roughness height. Lee \textit{et al}. (2012) examined by varying the streamwise extent (2$\le p_{x}/k\le $10) that the roughness function has the maximum contribution at $p_{x}/k$=4 and the outer value of the Reynolds stresses at \textit{y/$\delta $}$\approx $0.4 increases with increasing the streamwise pitch ($p_{x}/k)$. The roughness function has the local maximum at $p_{z}/k$=3 and the value of the Reynolds stresses at the same outer location increases linearly with increasing the spanwise pitch ($p_{z}/k)$. This implies that the wall friction is closely correlated with the roughness density because the roughness function has the maximum at the similar roughness density and there is also a strong interaction between the inner and outer regions at large spacing values of $p_{z}/k$. Furthermore, we can understand the roughness effects on the turbulence structures.   

Large Eddy Simulation of fully developed turbulent flow over a moving wavy surface

In order to investigate the vertical profile of wind speed over waves during typhoons for wind engineering applications, Large eddy simulation is performed to analyze the fully developed turbulent flow over a moving wavy surface. The Reynolds number Re is 10170, where Re is based on the channel depth $\delta $ and bulk velocity $U$. The wave steepness ${2\pi \alpha } \mathord{\left/ {\vphantom {{2\pi \alpha } \lambda }} \right. \kern-\nulldelimiterspace} \lambda $ ($\alpha $ and $\lambda $ is wave amplitude and wavelength respectively) is 0.25. Different wave age $c$/$U$ that ranges from -1.0 to 2.0 is considered to model the influence of $c$/$U$ on flow separation, drag force and vertical profiles of mean velocity and turbulence intensity. The simulation considered two kinds of flow motions, translation and undulation, in other words, whether the rigid wavy surface translates downstream with velocity $c$ ($>$0), or the wary surface moves downstream as a deformable undulation with phase speed $c$. Comparisons of the mean and turbulent velocity characteristics between these two flow motions are made to illustrate the difference of turbulence generation mechanism of these two motions, while shedding lights on the clarification of wind profile for wind engineering applications.