You are on page 1of 9

Engineering Turbulence Modelling and Experiments - 5 657

W. Rodi and N. Fueyo (Editors)


2002 Elsevier Science Ltd. All rights reserved.
NUMERICAL SIMULATION OF THE FLOW AROUND A CIRCULAR
CYLINDER AT HIGH REYNOLDS NUMBER
P. Catalano\ M. Wang^, G. laccarino^, and P. Moin^
^ CIRA - Italian Aerospace Research Center,
81043 Capua (CE), ITALY
^ Center for Turbulence Research
Stanford University/NASA Ames Research Center,
Stanford, CA 94305-3030, USA
ABSTRACT
The viability and accuracy of Large Eddy Simulation (LES) with wall modeling for high Reynolds num-
ber complex turbulent flows is investigated by considering the flow around a circular cylinder in the super
critical regime. A simple wall stress model is employed to provide approximate boundary conditions to
the LES. The results are compared with those obtained from steady and unsteady RANS and the avail-
able experimental data. The LES simulations are shown to be considerably more accurate than the RANS
results. They capture correctly the delayed boundary layer separation and reduced drag coefficients con-
sistent with experimental measurements after the drag crisis. The mean pressure distribution is predicted
correctly. However, the Reynolds number dependence is not predicted accurately due to the limited grid
resolution.
KEYWORDS
Large eddy simulation, Wall modeling. Unsteady RANS, High Reynolds number flows. Circular cyUn-
der, Navier-Stokes equations
MOTIVATION AND OBJECTIVES
Large eddy simulation of wall bounded flows becomes prohibitively expensive at high Reynolds numbers
since the number of grid points required to resolve the vortical structures (streaks) in the near wall region
scales as the square of the friction Reynolds number [2]. This is nearly the same as for direct numerical
simulation.
To circumvent the severe near wall resolution requirements, LES can be combined with a wall layer
model. In this approach, LES is conducted on a relatively coarse mesh with the first off-wall grid point
located in the logarithmic region. The dynamic effects of energy-containing eddies in the wall layer (vis-
658
cous and buffer region) are determined from a wall stress model that provides to the outer LES a set of
approximate boundary conditions, often in the form of wall shear stresses.
In recent years, the wall models based on Turbulent Boundary Layer (TBL) equations and their simpli-
fied forms [3, 5] have received much attention. These models, used with a Reynolds Averaged Navier
Stokes (RANS) type eddy viscosity, have shown promise for complex flow predictions [15]. To compute
the wall stress, the turbulent boundary layer equations are solved on an embedded near-wall mesh. The
tangential velocities obtained by the LES solution are imposed at the outer boundary, and the no-slip con-
ditions are applied at the wall. The turbulent eddy viscosity is modeled through a mixing length model
[14,5].
The main objectives of this paper are to further assess the viability and accuracy of large eddy simulation
with wall modeling for high Reynolds number complex turbulent flows and to compare this technology
with RANS models. To this end, the flow around a circular cylinder at Reynolds number (based on the
cylinder diameter D) of 0.5,1 and 2 x 10^ has been computed using LES with a wall stress model, and
the results have been compared to those achieved by steady and unsteady RANS, and to the available ex-
perimental data. The flow around a circular cylinder, with its complex features such as separating shear
layers and vortex formation and shedding, represents a canonical problem to validate new approaches in
computational fluid dynamics. To take the best advantage of wall modeling, we have concentrated on
the super critical flow regime in which the boundary layer becomes turbulent prior to separation. This
is, to the authors' knowledge, the first such attempt using LES. A related method, known as detached
eddy simulation (DES), in which the entire boundary layer is modeled, has been tested for this type of
flow [13]. Recently an LES study [4] has been conducted at a Reynolds number of 1.4 x 10^, and a good
comparison with the experimental data, especially in the near wake, has been shown.
NUMERICAL METHOD
The numerical method for LES employs an energy-conservative scheme of hybrid finite difference/spectral
type written for C meshes [6,10]. The fractional step approach, in combination with the Crank-Nicholson
method for viscous terms and third order Runge-Kutta scheme for the convective terms, is used for the
time advancement. The continuity constraint is imposed at each Runge-Kutta substep by solving a pres-
sure Poisson equation using a multigrid iterative procedure. The subgrid scale stress (SGS) tensor is mod-
eled by the dynamic model [7] in combination with a least-square contraction and spanwise averaging [9].
Approximate boundary conditions are imposed on the cylinder surface in terms of wall shear stress com-
ponents Ty,i{i = 1,3) estimated through a simplified form of the TBL equation model [3, 14, 5] :
C/X2 UXi
where in general
_ 1 ^p dui d
pdxi dt dxj ^ ''
The pressure is assumed to be X2 independent, and equal to the value from the outer flow LES solution.
If the substantial derivative term in Fi is neglected, Eq. (1) can be integrated to obtain a closed form
expression for the wall shear stress component [14]
dui
X2=0
Jo u -
Jo ly + i^t J
usi - Fi ;dx2 > (3)
659
where usi denotes the outer flow velocity from LES at the first off-wall node X2 = S. The eddy viscosity
vt is obtained from a RANS type mixing length model with a near wall damping
i^t
:/.y+(l-e^-/^f (4)
where y:^ is the distance to the wall in wall units, K, is the von Karman constant, and A = 19. In the
present work Fi = -f^. has been employed. At the inflow boundary the potential flow solution is im-
posed, while at the outflow a convective boundary condition is used.
RESULTS AND DISCUSSION
Large eddy simulations of the flow around a circular cyhnder at Reynolds numbers of 0.5 x 10^, 1 x 10^,
and 2 x 10^ have been performed. The computational domain has a spanwise size of 2D, and the C mesh
extends about 22D upstream, 17D downstream of the cylinder and 24JD into the far field. The flow is
assumed to be periodic in the spanwise direction and 401 x 120 x 48 grid points are used. The simula-
tion has been advanced for more than 300 dimensional time units AtUoo/D, and the statistics have been
collected over the last 200 time units. For comparison, steady and unsteady RANS simulations have also
been performed using the same computational grid. A commercial CFD code, using the k-e turbulence
model and the classical wall function approach, has been employed. The discussion is focused on the
case of Ren = 1 x 10^, with emphasis on important flow parameters, such as the Strouhal number, the
drag coefficient, and the base pressure coefficient, and their dependence on the Reynolds number.
The mean pressure distribution on the cylinder surface is compared to two set of experimental data in
figure 1. A very good agreement is observed between the LES at Reo = 1 x 10^ and the experiments by
Warschauer & Leene which were performed at Reo = 1.2 x 10^ [16]. The unsteady RANS also provides
a mean pressure coefficient in satisfactory agreement with both LES and experimental data, while, as ex-
pected, the steady RANS yields a poor result. The original data of Warschauer & Leene exhibit some
spanwise variations (see [17]), and for purpose of comparison the average values are plotted. Relative to
the measurements of Falchsbart at RCD = 6.7 X 10^, the numerical results show smaller values in the base
region. It is worth noting that the Falchsbart's data contain a kink near 0 = 110 indicating the presence
of a separation bubble.
The contours of the vorticity magnitude, as computed by LES and URANS, for RCD = 1 x 10^ at a
given time instant and spanwise plane are plotted in figure 2. In the LES results, large coherent struc-
tures are visible in the wake, but they are not as well organized as in typical Karman streets at sub-critical
and post-critical Reynolds numbers. The rather thick layers along the cylinder surface consist mostly of
vorticity contours of small magnitude. These levels are necessary for visualizing the wake structure, but
are not representative of the boundary layer thickness. The true boundary layer, with strong vorticity, is
extremely thin in the attached flow region. The shear layers are more coherent in the URANS than in the
LES. A clear vortex shedding is visible in the URANS results. The axial velocity distribution (time and
spanwise averaged), obtained by LES, is presented in the lower half of figure 3. Compared to flows at
lower Reynolds numbers [8,4], the boundary layer separation is much delayed, and the wake is narrower
resulting in a smaller drag coefficient. The time-averaged URANS velocity distribution is also plotted in
the upper half of figure 3.; this shows a thicker wake, resulting in a higher drag coefficient.
The drag coefficient, the base pressure coefficient, and the Strouhal number for the flow at Reynolds num-
ber of 1 X 10^ are summarized in table 1. The agreement with the measurements of Shih et al. [12] is
reasonably good. The LES overpredicts the drag coefficient compared to Shih et al, but underpredicts
the CD relative to Achenbach [1] (cf. fig. 4). The Strouhal number of 0.22 from Shih et al is for a rough
cylinder. It is generally accepted that periodic vortex shedding does not exist in the super critical regime
660
F \
F \
F
r 1 1 1 1
\
\
\
\
\A.
\ A
A
L"-
(
1 1 1 1 1
90
e
0.5
0
-0.5
-1.5
-2
-2.5
-3
r\
N
F
F
\
\
\
N y\
/ / /
/ / /
//o
//
o
p
Figure 1: Mean pressure distribution on the cylinder: - LES at RtD = 1 x 10^; RANS at Rert =
1X 10^ URANS at Reo = 1 x 10^ o Experiments by Warschauer & Leene [16] at Reo = 1.2 x 10^
(spanwise averaged); A Experiments by Falchsbart (in [17]) at Rep 6.7 x 10^
TABLE 1
DRAG, BASE PRESSURE COEFFICIENT AND STROUHAL NUMBER
FOR THE FLOW AROUND A CIRCULAR CYLINDER AT Rev = 1x10^
LES
RANS
URANS
Exp. (Shihera/. [12])
Exp. (Others, see [17])
CD
0.307
0.385
0.401
0.24
0.17-0.40
- ^ n a s e
0.32
0.33
0.41
0 . 3 3 ^
-
St
0.28
-
0.31
0.22
0.18-0.50
for smooth cylinders [12, 17]. In the present simulations, a broad spectral peak of the unsteady lift at
St ^ 0.28 is found. It can be argued that the discretization of the cylinder surface and the numerical
errors due to the under resolution may act as equivalent surface roughness, causing the flow field to ac-
quire some rough cylinder characteristics. The wide scatter of Co and St among various experiments in
the literature [17], listed at the bottom of the table 1, gives evidence of the high sensitivity of the flow
to perturbations due to surface roughness and free-stream turbulence in the super critical regime. The
lack of detailed experimental data in the super critical flow regime makes a more complete comparison
impossible.
An additional comparison between the LES and URANS is reported in Fig. 4 in terms of lift and drag
time histories. It is again clear that the URANS predicts a very well organized and periodic flow at this
Reynolds number, whereas the LES results (especially in terms of drag coefficients) have a distinct tur-
bulent character. The over-dissipative nature of the URANS calculations is also evident by observing the
limited amplitude of the lift and, especially, drag oscillations.
To assess the robustness of the computational method, large eddy simulations at Rep = 5 x 10^ and
661
LES
URANS
Figure 2: Instantaneous vorticity magnitude at a given span wise cut for flow over a circular cylinder at
Reo = 1 X 10^. 50 contour levels from uD/Uoo = 1 to uD/Uoo = 20 (exponential distribution) are
plotted.
2 X 10^, have also been performed. The predicted mean drag coefficients are plotted in figure 5 along
with the drag curve of Achenbach [1]. The CD at the two lower Reynolds number is predicted rather
well, and the discrepancy becomes larger at Ren = 2 x 10^. More significantly the LES solutions show
relative insensitivity to the Reynolds number, in contrast to the experimental data that exhibit an increase
in CD after the drag crisis. Similar Reynolds number insensitivity has been shown also for other quanti-
ties such as the base pressure coefficient. Poor grid resolution, which becomes increasingly severe as the
Reynolds number increases, is the primary suspect.
The skin friction coefficients predicted by the wall model employed in the LES computations are pre-
sented in figure 6 together with the experimental data of Achenbach [1] at Reo = 3.6 x 10^. The levels
are very different on the front half of the cylinder but are in reasonable agreement on the back half. The
boundary layer separation and the recirculation region are captured rather well, indicating that they are
not strongly affected by the upstream errors. The different Reynolds numbers between the LES and the
experiments can account for only a small fraction of the discrepancy. The computed Cf values are compa-
rable to those reported by Travin et al. [13] using the detached eddy simulation. Travin et al. attribute the
662
Figure 3: Axial velocity distribution (time and spanwise averaged). 45 contour levels from U/Uoc
-0.2tot//[/oo = 1.7.
Figure 4: Time histories of lift and drag coefficients. - LES, URANS.
663
Figure 5: Drag coefficient as a function of the Reynolds number. Achenbach [1]; LES; URANS
overprediction of the C/ before the separation to the largely laminar boundary layer that has not been ad-
equately modeled in either the simulation. In the present work grid resolution is another potential culprit.
In addition, an overprediction of the skin friction by the wall model adopted in the present LES compu-
tations has also been observed by Wang & Moin in the acceleration region of the trailing edge flow [15],
suggesting that this simplified model may have difficulty with strong favorable pressure gradients.
CONCLUDING REMARKS
A bold numerical experiment has been performed to compute the flow around a circular cylinder at su-
percritical Reynolds number using LES. The simulations have been made possible by the use of a wall
model that alleviates the grid resolution requirements. Preliminary results are promising in the sense that
they correctly predict the delayed boundary layer separation and reduced drag coefficients consistent with
measurements after the drag crisis. The mean pressure distributions and overall drag coefficients are pre-
dicted reasonably well at Reo = 0.5 x 10^ and 1 x 10^. However the computational solutions are in-
accurate at a higher Reynolds number, and the Reynolds number dependence is not captured well. The
grid used near the surface, particularly before separation, is quite coarse judged by the need to resolve
the outer boundary layer scales. Furthermore the effect of the wall model under coarse grid resolution
and in the laminar boundary layer is not clear. A more systematic investigation is needed to separate the
grid resolution and the wall modeling effects, and to fully validate the numerical methodology for this
challenging flow.
References
[1] Achenbach E. (1968) Distribution of local pressure and skin friction around a circular cylinder in
cross-flow up to i?e = 5 X 10^ J. Fluid Mech., 34, 625-639.
[2] Baggett J. S., Jimenez J. & Kravchenko A. G. (1997) Resolution requirements in large eddy simula-
tions of shear flows. Annual Research Briefs 1997, Center for Turbulence Research, Stanford Univer-
sity/NASA Ames, 51-66.
664
o 0
^i i j I I I
180
0
Figure 6: Skin friction distribution on the cylinder: LES at Reo = 0.5 x 10^; - LES at Reo
1 X 10^ LES at Reo = 2x 10^ o Experiments by Achenbach [1] at Reo = 3.62 x 10^
[3] Balaras E., Benocci C. & Piomelli U. (1996) Two-layer approximate boundary conditions for large
eddy simulations. A/A4 7., 34, 1111-1119.
[4] Breuer M. (2000) A challenging test case for large eddy simulation: high Reynolds number circular
cylinder flow. Int. J. Heat Fluid Flow, 21, 648-654.
[5] Cabot W. & Moin P. (20(X)) Approximate wall boundary conditions in the large eddy simulation of
high Reynolds number flows Flow Turb. Combust., 63, 269-291.
[6] Choi H. (1993) Toward large eddy simulation of turbulent flow over an airfoil. Annual Research
Briefs 1993, Center for Turbulence Research, Stanford University/NASA Ames, 145-149.
[7] Germano M., Piomelli U., Moin P., & Cabot W.H. (1991) A dynamic subgrid-scale eddy viscosity
model. Phys. fluids A. 3, 1760-1765.
[8] Kravchenko A. G. & Moin P. (1998) B-Spline method and zonal grids for numerical simulations of
turbulent flows. Report No. TF-73, Mechanical Engr. Dept., Stanford University, February 1998.
[9] Lilly D.K. (1992) A proposed modification of the Germano subgrid scale closure method. Phys. fluids
A. 3, 2746-2757.
[10] Mittal R. & Moin P. (1997) Suitability of upwind-biased finite difference schemes for large eddy
simulation of turbulent flows. AIAA J., 35, 1415-1417.
[11] Roshko A. (1962) Experiments on the flow past a cylinder at very high Reynolds number. J. Fluid
Mech., Vol. 10, 345-356.
[12] Shih W. C. L., Wang C, Coles D. & Roshko A. (1993) Experiments on flow past rough circular
cylinders at large Reynolds number. Journal of Wind Engineering and Industrial Aerodynamics, 49,
351-368.
[13] Travin A., Shur M., Strelets M. & Spalart P. (1999) Detached eddy simulations past a circular cylin-
der. Flow Turb. Combust. 63, 269-291.
665
[14] Wang M. (1999) LES with wall models for trailing-edge aeroacoustics. Annual Research Briefs
1999, Center for Turbulence Research, Stanford University/NASA Ames, 355-364.
[15] Wang M. & Moin P. (2001) Wall modeling in LES of trailing-edge flow. Proceedings of the Second
International Symposium on Turbulence and Shear Flow Phenomena. Stockholm, Sweden, II, 165-
170.
[16] Warschauer K. A. & Leene J. A. (1971) Experiments on mean and fluctuating pressures of circular
cylinders at cross flow at very high Reynolds number. Proceedings of the International Conference on
Wind Effects on Buildings and Structures, pp. 305-315, Tokyo, Japan Sept. 6-9,1971.
[17] Zdravkovich M. M. (1997), Flow around circular cylinders. Vol.l: Fundamentals. Oxford Univer-
sity Press.

You might also like