Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Three-Dimensional Numerical

Simulations of Circular Cylinders


Undergoing Two Degree-of-
Juan P. Pontaza
Department of Mechanical Engineering,
Texas A&M University,
Freedom Vortex-Induced
College Station, Texas 77843
Vibrations
Hamn-Ching Chen
Department of Civil Engineering, In an effort to gain a better understanding of vortex-induced vibrations (VIV), we present
Ocean Engineering Program, three-dimensional numerical simulations of VIV of circular cylinders. We consider oper-
Texas A&M University, ating conditions that correspond to a Reynolds number of 105, low structural mass and
College Station, Texas 77843 damping (m* = 1.0, ␨* = 0.005), a reduced velocity of U* = 6.0, and allow for two degree-
of-freedom (X and Y) motion. The numerical implementation makes use of overset (Chi-
mera) grids, in a multiple block environment where the workload associated with the
blocks is distributed among multiple processors working in parallel. The three-
dimensional grid around the cylinder is allowed to undergo arbitrary motions with re-
spect to fixed background grids, eliminating the need for grid regeneration as the struc-
ture moves on the fluid mesh. 关DOI: 10.1115/1.2746396兴

Introduction drag coefficient and base pressure coefficient are overpredicted


due to the omission of spanwise wake effects with short correla-
Long flexible circular cylinders, known as marine risers, linking tion lengths.
the seabed to an offshore platform are exposed to strong sea cur- VIV numerical simulation results under the assumption of two-
rents that cause the flow around the risers to separate and initiate dimensionality using the present overset grid approach were pre-
vortex shedding. Because of low structural damping, the resultant sented by Pontaza et al. 关1–3兴, for isolated and multicylinder con-
lift and drag forces induce forced oscillations of the cylinder, figurations under high Reynolds number flow conditions,
known as vortex-induced vibrations 共VIV兲. High-amplitude vibra- including close surface encounters in the multicylinder configura-
tions of the riser could potentially result in a high level of fatigue tions. Nevertheless, it has been shown by Blackburn et al. 关4兴 that,
damage in a relatively short period of time. even at low Reynolds numbers, three-dimensional effects on VIV
Design principles for deepwater risers are typically drawn from are important and cannot be neglected.
empirical knowledge and existing experimental data. These data- The present study focuses on numerical simulations of three-
dimensional VIV allowing for two-degree-of-freedom motions
bases would further benefit from reliable computational studies on
共XY motions兲 at high Reynolds number and low structural mass
fluid-induced motions of structures. The focus of this work is to
and damping. Previously published three-dimensional numerical
present such a study; in particular, we present three-dimensional simulation results for VIV are predominantly for single degree-of-
numerical simulation results for an isolated cylinder allowed to freedom forced 共i.e., prescribed兲 motions 关4,5兴. The Reynolds
undergo two-degree-of-freedom motion 共XY motion兲. number considered here is Re= 105, which is typical in practical
In practical applications, the Reynolds number is of the order applications but in the low end.
105 and higher. At these flow conditions, the flow around and
behind the cylinder is transitional or turbulent, and highly three- Numerical Solution Method
dimensional. These conditions make analysis difficult and compu-
tationally expensive for direct numerical simulations of the gov- From the computational viewpoint, simulations of riser VIV are
fluid-structure interaction problems posed at high Reynolds num-
erning incompressible Navier-Stokes equations. In this work, we
bers. Standard discretization methods, such as finite volumes or
study the flow past a cylinder of aspect ratio L / D = 3.0 and use a Galerkin finite elements, are not robust with respect to high Rey-
large eddy simulation 共LES兲 Smagorinsky model to account for nolds numbers and tend to develop node-to-node spurious oscil-
unresolved scales in the flow field. lations when the cell Reynolds number exceeds a certain thresh-
The flow past a fixed isolated cylinder has been the subject of old. Stabilization techniques 共upwinding or subgrid stabilization兲
extensive experimental and numerical studies, and is relatively may be invoked to render the schemes stable and are popular in
well understood in terms of its wake dynamics. Given the high practice. In addition, these methods typically need grid regenera-
demand on computational resources for three-dimensional simu- tion and/or control of element distortions as the structure moves
lations, the majority of numerical simulations have been per- on the fluid mesh.
formed assuming a two-dimensional flow field. In this case, the In the present work, the governing incompressible Navier-
Stokes equations are discretized using the local-analytic-based
discretization procedure of Pontaza et al. 关6兴, whereby local ana-
Contributed by the Ocean Offshore and Arctic Engineering Division of ASME for lytic solutions of the linearized advection-diffusion part of the
publication in the JOURNAL OF OFFSHORE MECHANICS AND ARCTIC ENGINEERING. Manu- momentum equations are obtained at the local 共element兲 level and
script received April 17, 2006; final manuscript received November 12, 2006. Re-
view conducted by Dan Valentine. Paper presented at the 25th International Confer-
used to construct finite-difference-like stencils in terms of neigh-
ence on Offshore Mechanics and Arctic Engineering 共OMAE2006兲, June 4–9, 2006, boring nodal degrees of freedom. The local analytic nature gives
Hamburg, Germany. the resulting stencils suitable upwinding according to the local

158 / Vol. 129, AUGUST 2007 Copyright © 2007 by ASME Transactions of the ASME

Downloaded From: http://offshoremechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


velocity field and element aspect ratio. In essence, the amount of
upwinding is determined by virtue of the local analytic solution
and not in a heuristic manner along gridlines. The three-
dimensional 19-point stencil allows for skew upwinding, depend-
ing on local flow conditions.
The incompressibility constraint is enforced by a discrete pro-
jection method that ensures proper velocity-pressure coupling.
Time stepping is performed using a second-order stiffly stable
backward differentiation scheme, popularly know as a BDF2
scheme. For low to moderate cell Reynolds numbers, the scheme
is second-order accurate in velocities and pressure. At arbitrarily
high cell Reynolds number, the scheme is always asymptotically
second-order accurate and unconditionally stable. Details of the
discretization procedure, including verification and validation, are
presented in Pontaza et al. 关6兴.
The numerical method is implemented in an overset 共Chimera兲
grid environment. This allows for great flexibility in representing
complex geometries and allows for surgical mesh refinements, in
the form of embedded and overlapping groups of grid compo-
nents. Individual structured grids are allowed to overlap arbi-
trarily, and intergrid communication is achieved by mass conser-
vative Lagrange interpolation at the fringes.
Figure 1 depicts a cross section in the XY plane 共a “Z-slice”兲 of
the Chimera grid used for the simulations. The Z-slice shown has
close to 35,000 cells. The cylindrical grid wrapped around the
cylinder can be refined independently from neighboring grids to
any desired level, resulting in improved local resolution and over-
all computational savings 共as the other grid components remain
unrefined兲. When the cylinder undergoes two-degree-of-freedom
VIV, the cylindrical mesh wrapped around it moves on top of the
earth-fixed Cartesian mesh, thus avoiding tedious grid-
regeneration and/or mesh distortion monitoring. The interpolation
stencils that are needed to communicate across grid components
are simply updated as the cylindrical grid moves.
We make further use of the Chimera technique and allow for
variable spanwise resolution within each grid component. For ex-
ample, the cylindrical grid wrapped around the cylinder has 120
cells along its span, while the Cartesian mesh on which it is em-
bedded has 60 cells along its span, and the far-field blocks have
only 30 cells along their span.
Figure 2 shows a spanwise cut of the grid to show the grid
distribution along the spanwise length of the cylinder, L / D = 3.0.
In total, the mesh we use for the computations has close to 2.1
million cells.
For the Reynolds number considered, a direct numerical simu- Fig. 1 A “Z-slice” of the computational domain showing the
multiple-block structure of the Chimera grid in the XY plane
lation would be prohibitively expensive and our grid would be
deemed inappropriate because it would not satisfy the strict reso-
lution requirements. We use a LES Smagorinsky model to account
for unresolved scales. In addition, we note that we are mostly
interested in the accurate prediction of the instantaneous drag and As seen from Fig. 1, the size of the computational domain is
lift forces exerted on the cylinder by the oncoming fluid, which taken sufficiently large to preclude unwanted effects on computed
essentially will determine the response of the cylinder to VIV. We flow metrics due to blockage or location of inflow and outflow
are not necessarily interested in resolving the far-wake flow field, boundaries. The computational domain extends from −15D at the
for which our grid resolution is inappropriate, even with the LES inlet to 20D at the outlet and from −16D to 16D in the cross-flow
model. The grid shown is the finest we have considered and the direction.
one for which we present our results. Computations were also Boundary conditions include freestream values assigned at the
performed in coarser meshes, with 1.2 and 1.6 million grid points. inflow boundary, a no-flux or symmetry boundary condition at the
The turnaround time of the simulations is accelerated by further lateral boundaries, no-slip boundary conditions on the cylinder
exploiting the multiple-block data structure of our code and the surface, and nonreflective outflow boundary conditions. Boundary
Chimera approach. Using MPI bindings, we are able to distribute conditions on the pressure are not explicitly necessary and are
the load among different processors working in parallel. Specifi- recovered as part of the velocity projection step 关6兴. In other
cally, different number of blocks can be assigned to different pro- words, pressure is consistently computed at all boundaries.
cessors. Load balancing is the task of the user, but the structure of In the spanwise direction, we impose periodic boundary condi-
our code facilitates this task by allowing single or arbitrarily large tions. The assumption of periodic behavior in the spanwise direc-
groups of consecutive or nonconsecutive blocks to be assigned to tion is adopted to minimize the effect of the spanwise boundaries
different processors. The simulations performed in the grid shown and essentially takes the spanwise direction as a homogeneous
were performed using eight processors in parallel. Each processor direction. Periodic boundary conditions in the spanwise direction
was loaded with ⬃260,000 grid points. The wall-clock time per are typically the norm in large-scale three-dimensional simula-
time step was around 40 s. tions of flow past a circular cylinder 关5,7兴.

Journal of Offshore Mechanics and Arctic Engineering AUGUST 2007, Vol. 129 / 159

Downloaded From: http://offshoremechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


siderable scatter is present in measurements at high Reynolds
numbers due to a variety of influencing factors. Among these are
end conditions, freestream turbulence level, and cylinder rough-
ness. Nevertheless, the drawing of comparisons to experimental
data is needed to establish confidence in the numerical method.
The drag coefficient, averaged over 100 convective time units is
CD = 1.05. Experimental measurements report a time-averaged
value around CD ⬇ 1.20 关10,11兴, in fair agreement with the com-
puted value. Simulations without the LES model give a time-
averaged drag coefficient of CD = 0.95. Thus, we see improvement
in this flow metric by using the LES model. The predicted rms lift
coefficient is CL,rms = 0.62, also in fair agreement with measure-
ments 关12兴, CL,rms ⬇ 0.55.
Figure 4 shows the power spectra of the streamwise and cross-
flow velocities at two points in the near wake of the cylinder, on
the plane Z = 1.5. The spectra of the cross-flow velocities have a
peak at the Strouhal number frequency. The small and some me-
dium scales are being suppressed due to insufficient grid resolu-
tion. Additional mesh refinement would be needed to adequately
resolve these scales. Nevertheless, the large energy content of the
spectra is resolved and the grid cutoff takes place after the energy
has dropped by four orders of magnitude. The spectra qualita-
Fig. 2 Spanwise cut along Y = 0.0 showing the grid distribution tively follow the −5 / 3 slope in the resolved part of the inertial
on the spanwise direction. Blocks in the vicinity of the cylinder subrange.
have higher spanwise resolution.
Two-Degree-of-Freedom VIV
Regarding the spanwise length, taken here to be L / D = 3.0, we The cylinder is now left free to respond to the flow field. The
consider it to be sufficiently large to capture relevant spanwise structural response of the cylinders is assumed to be governed by
wake effects. We have only access to limited computational re- a mass-spring-damper system in the X and Y directions as follows:

冉 冊
sources, which precludes simulations with a larger spanwise
length with adequate spanwise resolution. 4␲␰* 2␲ 2
2 CD共t兲
ẍ + ẋ + x= 共1兲
U* U* ␲ m*

冉 冊
Flow Past a Fixed Cylinder
4␲␰* 2␲ 2
2 CL共t兲
First, we consider the flow past a fixed cylinder at Re= 105. In ÿ + ẏ + y= 共2兲
this range of Reynolds number, the only three-dimensional simu- U* U* ␲ m*
lation results are those reported by Breuer 关7兴 at Re= 1.4⫻ 105, where the dimensionless parameters m*, ␨*, and U* are the struc-
using LES. Still higher Reynolds numbers have been recently at- tural mass ratio, damping ratio, and reduced velocity, respectively.
tempted by Catalano et al. 关8兴 using LES and a wall model, with CD is the instantaneous drag coefficient, and CL is the instanta-
not very accurate results toward the higher end of their Reynolds neous lift coefficient.
number range. Direct numerical simulations have been performed Thus, the problem of VIV is essentially a fluid-structure inter-
at Re= 104 only recently by Dong and Karniadakis 关5兴, using over action problem, where the structure responds instantaneously to
1500 processors in parallel. the fluid-induced lift and drag forces. These forces are computed
At this Reynolds number, the flow is still subcritical, with the using the instantaneous velocity and pressure fields at each time
boundary layer separating laminarly and transition taking place in step, by performing a numerical integration around the circumfer-
the free shear layers close to the wall. As alluded to earlier, we ence of the cylinder. The cylinder is then displaced at each time
avoid the use of any kind of wall functions or transition models step, by numerically solving Eqs. 共1兲 and 共2兲, and the flow field is
and resolve the boundary layer by using fine grid spacing near the time advanced and recomputed to yield new instantaneous lift and
cylinder surface and using the no-slip boundary condition. The drag forces.
near-wall spacing in the radial direction of the cylindrical grid The overset grid capabilities allow for relative motion between
wrapped around the cylinder is taken as ⌬r / D = 10/ Re= 10−4, grid components, allowing for arbitrarily large motions of the cyl-
which is four times smaller than the near-wall spacing used by inder without the need for tedious and costly grid regeneration
Breuer 关7兴. In all our computations, we take the Smagorinsky and/or mesh deformation monitoring. In essence, the grid wrapped
constant in the LES model, as Cs = 0.10. around and/or attached to the structure moves freely inside earth-
As expected, the flow is highly three-dimensional and unsteady, fixed background meshes. Boundary interpolation stencils to com-
exhibiting the well-known von Karman vortex street. Figures 3共a兲 municate between grid components are recomputed every time
and 3共b兲 show instantaneous contours of the spanwise velocity step to ensure a continuous solution across grid fringes. In the
component and of the cross-flow vorticity on the center plane Y present study, the structure’s equations of motion are numerically
= 0.0, from which the three-dimensional effects are evident. Also integrated in time using a fourth-order accurate Runge-Kutta
shown, in Fig. 3共c兲, are isosurfaces of the streamwise vorticity scheme.
component. The cross-flow and streamwise vortices exhibit a co- We are interested in the structure’s response for low mass and
herent structure in the spanwise direction 共Figs. 3共b兲 and 3共c兲兲. damping. The values for the mass ratio, damping ratio, and re-
Figure 3共d兲 shows spanwise vorticity isosurfaces, showing a two- duced velocity used for the simulations in this study are m* = 1.0,
dimensional flow field in-front of the cylinder and an increasingly ␨* = 0.005, and U* = 6.0.
three-dimensional flow field downstream. The cylinder is first held fixed until the shedding has reached
The generally measured and accepted value of the Strouhal full strength, at which point the cylinder is “released” and allowed
number at this Reynolds number is St⬇ 0.20 关9,10兴, in good to respond to the flow field. Figure 5 shows the XY response of the
agreement with the predicted value of St= 0.19. Before presenting cylinder for 100 convective time units, where time is implicit in
additional comparisons to experimental data, we remark that con- the plot.

160 / Vol. 129, AUGUST 2007 Transactions of the ASME

Downloaded From: http://offshoremechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 3 „a… Instantaneous spanwise velocity contours for the flow past a fixed cylinder. Contour range †−0.2, 0.2‡. „b… Instan-
taneous cross-flow vorticity contours for the flow past a fixed cylinder. Contour range †−2 , 2‡. „c… Instantaneous streamwise
vorticity isosurfaces for the flow past a fixed cylinder. Isosurfaces +0.5 and −0.5. „d… Instantaneous spanwise vorticity isosur-
faces for flow past a fixed cylinder. Isosurfaces range †−4 , 4‡.

Initially, the cylinder is pushed downstream due to the acting x共t兲 = AX sin共2␻t + ␪兲 共4兲
drag force exerted by the oncoming fluid. Almost immediately, the
cylinder responds to the acting lift forces and starts displaying
strong cross-stream oscillations. Once the pattern of motion has where AY is the average cross-stream amplitude of vibration, AX is
settled, the XY response displays average peak-to-peak cross- the average streamwise amplitude of vibration, ␻ = 2␲ f Y , and ␪ is
stream amplitudes of 1.65 cylinder diameters. There is also weak the phase angle. The following coefficients, for use in Eqs. 共3兲 and
in-line motion with average peak-to-peak amplitudes of 0.160 cyl- 共4兲, were extracted from the predicted time history response of our
inder diameters. The response is remarkably stable and corre- simulation: AY = 0.824, AX = 0.0795, f Y = 0.1757, and ␪ = −40 deg.
sponds to a “figure-of-eight” pattern, which was maintained In an experimental study, Jeon and Gharib 关13兴 forced a cylin-
throughout the simulation time of 250 convective time units. der to move in the X and Y direction according to Eqs. 共3兲 and 共4兲,
Analysis of the power spectrum of the time history of body as they suggested that nature prefers the “figure-of-eight” pattern.
motion gives f X / f Y = 2.0, i.e., the streamwise oscillation fre- Our simulations confirm that this is indeed the case.
quency is twice that for the transverse 共cross-stream兲 direction. In We believe the response falls toward the start of the recently
addition, we find that a good representation of the average dis- discovered “super-upper” branch 关14兴 for light bodies 共m* ⬍ 6.0兲
placement pattern in the simulation is given by
undergoing two degree-of-freedom VIV. Although the experi-
y共t兲 = AY sin共␻t兲 共3兲 ments of Jauvtis and Williamson 关14兴 were conducted at lower

Journal of Offshore Mechanics and Arctic Engineering AUGUST 2007, Vol. 129 / 161

Downloaded From: http://offshoremechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 4 Power spectra of streamwise and cross-flow velocities
in the near wake of the cylinder

Reynolds numbers 共Re= 103 − 104兲, the predicted response clearly


displays the high cross-stream amplitudes associated with this
branch.
Unlike our previous two-dimensional simulations, where the 2T
mode of vortex shedding was observed 关2,3,6兴, here we see a 2S
mode of vortex shedding. In this mode, the cylinder sheds two
single vortices per cycle of body motion, whereas in the 2T mode
two triplets of vortices are shed per cycle of body motion.
Figure 6 shows instantaneous spanwise vorticity contours on
the XY plane corresponding to the spanwise location Z = 1.5,
showing the 2S mode of vortex shedding. Figure 6 depicts the
cylinder at the bottom and half-cycle of body motion. We conjec-
Fig. 6 Spanwise vorticity contours on the XY plane Z = 1.5,
showing the 2S mode of vortex shedding

ture that the 2T mode will be seen at higher reduced velocities, as


one climbs the super-upper branch and the cross-stream ampli-
tudes of vibration increase in magnitude.
Our previous two-dimensional numerical simulation results, at
the same flow and structural conditions, display much stronger
cross-stream and streamwise amplitudes of vibration: AY = 1.54,
AX = 0.54 关2,3,6兴. However, due to the omission of spanwise wake
effects in the two-dimensional simulations, the oscillatory lift
forces are much stronger. Consequently, the cross-stream ampli-
tudes of vibration are nearly double and the streamwise ampli-
tudes nearly seven times larger than those seen here for the three-
dimensional case. This indicates that three-dimensional effects are
important and contribute significantly to the VIV response.
The flow field remains highly three-dimensional during VIV.
Figure 7 shows spanwise vorticity isosurfaces at an instant in the
“figure-of-eight” cycle of motion when the cylinder’s cross-flow
velocity is maximum and the cylinder is moving 共upward兲 in the
positive Y direction. The 2S mode of vortex shedding is also
clearly visible from this plot.
Fig. 5 XY trajectory swept by the cylinder due to VIV. The re- Figure 8 shows the time history of the transverse displacement,
sponse is a “figure-of-eight” pattern of body motion. the lift coefficient, and the corresponding phase plot. Our simula-

162 / Vol. 129, AUGUST 2007 Transactions of the ASME

Downloaded From: http://offshoremechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 9 Power spectrum of the time history of the lift coeffi-
cient, showing a strong 3Ã harmonic

Fig. 7 Instantaneous spanwise vorticity isosurfaces for two- tions show that despite the fact that the fluid-induced forcing is
degree-of-freedom VIV of a circular cylinder. Isosurfaces range nonsinusoidal 共it has higher harmonics兲, the displacement remains
†−4 , 4‡.
remarkably close to sinusoidal.
The power spectrum of the time history of the lift coefficient is
shown in Fig. 9. The frequency is scaled by the first fundamental
frequency, and we see that there is a strong higher harmonic at
three times the first fundamental frequency. The excitation of the
3⫻ harmonic is typical of responses in the supper-upper branch,
although we did not see the 2T mode of vortex shedding at this
reduced velocity.
An increase in the drag coefficient is observed, with a time-
average drag of CD = 2.02. Compared to the fixed case, the mean
drag is doubled. The phase plot of the drag and lift coefficients in
Fig. 10 exhibit an interesting shape. The plot depicts the strong
periodicity of the flow-induced forces.

Concluding Remarks
This work documents three-dimensional numerical simulation
results for two-degree-of-freedom vortex-induced vibrations of a
circular cylinder at low mass ratio, low damping, and a high Rey-
nolds number of Re= 105. The response is characterized by a

Fig. 8 Time histories of transverse „cross-stream… displace-


ment and lift coefficient. The corresponding phase plot is also Fig. 10 Phase plot of the drag and lift coefficients. Average
shown. and fluctuating values can be obtained from the plot.

Journal of Offshore Mechanics and Arctic Engineering AUGUST 2007, Vol. 129 / 163

Downloaded From: http://offshoremechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


“figure-of-eight” pattern of body motion with AY = 0.824, AX pp. 1–12.
关4兴 Blackburn, H. M., Govardhan, R. N., and Williamson, C. H. K., 2000, “A
= 0.0795, and vortex shedding in the 2S mode. Presumably, higher Complimentary Numerical and Physical Investigation of Vortex-Induced Vi-
cross-stream amplitudes can be attained in the super-upper branch bration,” J. Fluids Struct., 15, pp. 481–488.
at higher reduced velocities than the one considered in this study. 关5兴 Dong, S., and Karniadakis, G. E., 2005, “DNS of Flow Past a Stationary and
Future work will aim to characterize the range of responses as a Oscillating Cylinder at Re= 10 000,” J. Fluids Struct., 20, pp. 519–531.
function of reduced velocity. 关6兴 Pontaza, J. P., Chen, H. C., and Reddy, J. N., 2005, “A Local-Analytic-Based
Discretization Procedure for the Numerical Solution of Incompressible
Flows,” Int. J. Numer. Methods Fluids, 49, pp. 657–699.
Acknowledgment 关7兴 Breuer, M., 2000, “A Challenging Test for Large Eddy Simulation: High Rey-
Computations were performed using resources of the Texas nolds Number Circular Cylinder Flow,” Int. J. Heat Fluid Flow, 21, pp. 648–
654.
A&M Super Computer Facility; their support is acknowledged. 关8兴 Catalano, P., Wang, M., Iaccarino, G., and Moin, P., 2003, “Numerical Simu-
The support from the Department of Interior, Mineral Manage- lations of Flow Around a Circular Cylinder at High Reynolds Numbers,” Int. J.
ment Services 共MMS兲 and the Offshore Technology Research Heat Fluid Flow, 24, pp. 463–469.
Center 共OTRC兲 is gratefully acknowledged. 关9兴 Bearman, P. W., 1969, “On Vortex Shedding from a Circular Cylinder in the
Critical Reynolds Number Regime,” J. Fluid Mech., 37, pp. 577–585.
关10兴 Schewe, G., 1983, “On the Force Fluctuations Acting on a Circular Cylinder in
References Crossflow From Subcritical up to Transcritical Reynolds Number,” J. Fluid
关1兴 Pontaza, J. P., Chen, C. R., and Chen, H. C., 2004, “Chimera Reynolds- Mech., 133, pp. 265–285.
Averaged Navier-Stokes Simulations of Vortex-Induced Vibration of Circular 关11兴 Achenbach, E., 1968, “Distribution of Local Pressure and Skin Friction
Cylinders,” Proc. of International ASCE Conference: Civil Engineering in the Around a Circular Cylinder in Crossflow up to Re= 5 ⫻ 106,” J. Fluid Mech.,
Oceans VI, ASCE, Baltimore, Maryland, pp. 166–176. 34, pp. 625–639.
关2兴 Pontaza, J. P., Chen, C. R., and Chen, H. C., 2005, “Simulations of High 关12兴 Norberg, C., 2001, “Flow Around a Circular Cylinder: Aspects of Fluctuating
Reynolds Number Flow Past Arrays of Circular Cylinders Undergoing Vortex- Lift,” J. Fluids Struct., 15, pp. 459–469.
Induced Vibrations,” Proc. of 15th International Offshore and Polar Engineer- 关13兴 Jeon, D., and Gharib, M., 2001, “On Circular Cylinders Undergoing Two
ing (ISOPE) Conference, ISOPE, Seoul, Korea, pp. 201–207. Degree-of-Freedom Forced Motions,” J. Fluids Struct., 15, pp. 533–541.
关3兴 Pontaza, J. P., Chen, H. C., and Chen, C. R., 2005, “Numerical Simulations of 关14兴 Jauvtis, N., and Williamson, C. H. K., 2004, “The Effects of Two Degrees of
Riser Vortex-Induced Vibrations,” Proc. of 2005 Society of Marine Engineers Freedom on Vortex-Induced Vibrations at Low Mass and Damping,” J. Fluid
and Naval Architects (SNAME) Conference, SNAME, Houston, Texas, D52, Mech., 509, pp. 23–62.

164 / Vol. 129, AUGUST 2007 Transactions of the ASME

Downloaded From: http://offshoremechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like