Session G17: Separated Flows

Benchmark Smooth Body Flow Separation Experiments

The accurate and repeatable computational fluid dynamics (CFD)-based prediction of turbulent flow separation relevant to off-design aerodynamic configurations remains an important challenge. There is a clear need for improved models that can accurately capture the essential flow physics associated with smooth body, adverse pressure gradient (APG) generated flow separation and reattachment. In this paper archival benchmark smooth body flow separation experiments are described that will be used for model development. These include: (1) attached smooth body APG turbulent boundary layer (TBL) flow, (2) APG TBL flow with incipient separation, (3) APG TBL separation with a small-scale separation bubble and (4) TBL separation with a large region of separated flow extending over a significant fraction of the smooth body surface. The model geometry is fully two-dimensional and consists of a flat TBL development plate followed by a large convex ramp surface with APG controlled by a fully adjustable flexible top tunnel ceiling. Experimental results characterizing the large-scale flow separation case are highlighted.   

Effects of surface roughness on an adverse-pressure-gradient separating turbulent boundary layer

Separating turbulent boundary layers over smooth and rough flat plates are investigated by large-eddy simulations. A suction-blowing velocity distribution is imposed at the top boundary to produce an adverse-to-favourable pressure gradient and a closed separation bubble. Sandgrain roughness in the fully-rough regime is modelled by an immersed boundary method. In the rough-wall case, streamline detachment occurs earlier and the separation region is substantially larger due to the momentum deficit caused by the roughness. The adverse pressure gradient decreases the form drag and causes a thin reversed-flow region below the roughness crest, so that $C_f = 0$ does not coincide with the detachment of the flow from the surface. The wake regions behind roughness elements affect the intermittency of the near-wall flow, so that upstream of the detachment point the flow can be reversed half of the time, but its average is positive. The separated shear layer exhibits higher turbulent kinetic energy (TKE); the growth of the TKE there begins earlier relative to the separation point, and the peak TKE occurs close to the separation point. The momentum deficit caused by the roughness, again, plays a critical role in these changes.   

Exact Theory of Material Spike Formation in Flow Separation

We develop a frame-invariant theory of material spike formation during flow separation over a no-slip boundary in two-dimensional flows with arbitrary time dependence. This theory identifies both fixed and moving separation, is effective also over short-time intervals, and admits a rigorous instantaneous limit. Our theory is based on topological properties of material lines, combining objectively stretching- and rotation-based kinematic quantities. The separation profile identified here serves as the theoretical backbone for the material spike from its birth to its fully developed shape, and remains hidden to existing approaches. Finally, our theory can be used to rigorously explain the perception of off-wall separation in unsteady flows, and more importantly, provide the conditions under which such a perception is justified. We illustrate our results in several examples including steady, time-periodic and unsteady analytic velocity fields with flat and curved boundaries, and an experimental dataset.   

Low-frequency dynamics of pressure-induced turbulent separation bubbles

We experimentally investigate a pressure-induced turbulent separation bubble (TSB), which is generated on a flat test surface through a combination of adverse and favorable pressure gradients imposed on a nominally two-dimensional, incompressible, turbulent boundary layer. We probe the flow using piezo-resistive pressure transducers, MEMS shear-stress sensors, and high-speed, 2D-2C, PIV measurements. Through the use of Fourier analysis of the wall-pressure fluctuations and Proper Orthogonal Decomposition of the velocity fields, we show that this type of flow is characterized by a self-induced, low-frequency contraction and expansion - called breathing - of the TSB. The dominant Strouhal number of this motion, based on the TSB length and the incoming velocity in the potential flow, is of the order of 0.01. We compare this motion to the low-frequency dynamics observed in laminar separation bubbles (LSBs), geometry-induced TSBs, and shock-induced separated flows.   

Wall-resolved Large-eddy simulation of flow past a grooved cylinder up to $Re_D=6\times 10^4$

We present wall-resolved, large-eddy simulations (LES) of flow past a groove-walled circular cylinder. Periodic span-wise boundary conditions are implemented with span up to $3\,D$. The stretched-vortex sub-grid scale model is utilized in the whole domain, including regions of large-scale separated flow. The circumferential cylinder surface comprises $32$ sinusoidal, span-wise groves of equal height $\epsilon$. For the first set of LES, with $Re_D=3.9 \times 10^3$ fixed and $0\le \epsilon \le 1/32$, some properties of the mean flow behave similarly to changes in the smooth-cylinder flow when $Re_D$ is increased, such as shrinking mean-flow recirculation length and near-constant pressure coefficient. A second set of LES uses fixed $\epsilon/D=1/32$ and varies $Re_D$ from $3.9 \times 10^3$ to $6 \times 10^4$, the latter value reaching the beginning of the transcritical flow regime. Comparison with experiment as well as wall-friction and Q plots, will be discussed.   

The computation of the drag coefficient of the unbounded flow around a circular cylinder in the limit of zero Reynolds number

The behaviour of the drag coefficient in the flow around a circular cylinder in the limit of zero Reynolds number is still unknown. A numerical experiment has been contacted to investigate this behaviour. This experiment is designed like a laboratory set up with tow tank boundary conditions along the top and the bottom of the computational domain. The two dimensional Navier Stokes equations are solved with Galerkin finite elements. It turns out that the domain should very long at the inlet and the transverse direction of the flow, in order for the results to be valid for the unbounded flow. Hence, supercomputing facilities are required, in order to solve the discretized system of equations which is of the order of 50 million unknowns. Results are given of Reynolds numbers ranging from $1\cdot 10^{-10}$ up to 40., which is the upper limit of the steady state flow conditions for this benchmark. The numerical results are compared with the available experimental measurements in the range of $0.1 \leq Re \leq 40$. The good agreement of our computations with laboratory data add credibility to our investigation in the limit of zero Reynolds number. It is the first time that a code produces results for this flow in such a wide range of values.   

Visualization of Surface Flow on a Prolate Spheroid Model Suspended by Magnetic Suspension and Balance System

In this study, the surface flow on a 6:1 prolate spheroid model was visualized by oil flow method in the magnetic suspension and balance system (MSBS). The MSBS is a support-free system for wind-tunnel test in that a model is levitated by magnetic force. In this experiment, the 0.3-m MSBS was installed in the low-speed wind tunnel. The Reynolds number was 0.5 million and the angle of attack was set 0 and 5 degrees. In addition to free-levitation tests, a thin rod simulating disturbance of a support system was placed on the model surface and the influence of support interference was evaluated. The obtained results indicate that complicated separation patterns are present even at zero angle of attack. At $\alpha \quad =$ 5\textdegree , separation pattern becomes more complicated than that at $\alpha \quad =$ 0\textdegree and the streamlines form a highly three-dimensional structure. A characteristic pattern of open separation is observed and a focal point is formed at the end of the separation line. In evaluation of the support interference, the separation is delayed in the downstream of the rod, suggesting that the change of separation pattern is caused by the transition of laminar boundary layer behind the rod. These results indicate that one must take particular care to the support interference in studying three-dimensional separation on a prolate spheroid.   

Support-Free Measurements of Aerodynamic Characteristics of Axial Circular Cylinders with Fineness Ratio from 0.50 to 0.75

In this study, aerodynamic characteristics of axial circular cylinders having the fineness ratio (length to diameter, $L/D)$ of 0.50, 0.67 and 0.75 were measured using the 1-m Magnetic Suspension and Balance System (MSBS) in the Low-Turbulence Wind Tunnel at Tohoku Univ. The MSBS supports and controls the model using magnetic forces. All the tests were conducted at \textit{Re }$=$ 100,000 and the models were aligned with the free stream. The results of force measurements show that the drag decreases gradually in the $L/D$ range from 0.50 to 0.75 and connects continuously to the data for higher fineness ratios. This indicates that a local maximum of the drag does not exist in this range. The previous studies show that, for axial circular cylinders having $L/D$ from 1.0 to 2.0, the drag measured using a MSBS differs substantially from the value measured with sting support, however this study shows that a circular cylinder of $L/D =$ 0.50 has a drag close to that measured with sting support. This suggests that the influence of support interference is significant when a shear layer separated from the leading edge reattaches on the body or interacts near the base, but is insignificant when a separated shear layer is away from the base area.