Sunteți pe pagina 1din 12

1 Copyright 2013 by ASME

Proceedings of the 32
rd
International Conference on Ocean, Offshore and Arctic Engineering
OMAE2013
9-14 June 2013, Nantes, France
OMAE2013- 10635
NUMERICAL SIMULATION OF VORTEX INDUCED VIBRATIONS
FOR MARINE RISERS SUBJECTED TO SHEARED FLOW


F. Van den Abeele
Cranfield University
Cranfield, UK
J. Vande Voorde
OCAS N.V.
Ghent, Belgium

F. Kara
Cranfield University
Cranfield, UK


ABSTRACT
The increasing demand for oil and gas, currently estimated at
135 million barrels of oil equivalent per day, keeps pushing the
boundaries of offshore engineering into ever deeper waters.
Exploration and production activities in the Gulf of Mexico, for
instance, are performed in water depths exceeding 3000 meters.
For such deepwater developments, the suspended length of the
marine risers adds up to several kilometers. When designing
and installing risers in (ultra)deep water, the length/diameter
aspect ratio of the marine riser can exceed L/D >1000, and the
features of the fluid flow in depth direction can no longer be
neglected. Indeed, both the magnitude and the direction of the
current change with water depth, giving rise to higher
harmonics in the VIV response.

The prediction of vortex induced vibrations for deepwater risers
is very challenging, owing to the fact that the incident flows are
non-uniform and the associated fluid structure interaction
phenomena are highly complex. These complex conditions give
rise to a non linear coupled system with a large number of
degrees of freedom, which depends on several physical and
mechanical parameters.

In this paper, 3D CFD calculations are performed to evaluate
the effect of the third dimension for risers subjected to uniform
flow and sheared currents. For a uniform current velocity at the
inlet boundary, it is shown that vortex shedding in the wake of
long slender tubulars can give rise to the development of
vortices with horizontal axis, resulting in a fluctuation of the
flow in the Z-direction. These three dimensional vortices are
strong enough to modulate the vortex shedding on the riser as a
function of depth. The 3D simulations with uniformcurrent
velocity are then compared to marine risers subjected to
sheared currents. It is shown that the presence of sheared
currents invokes a shift in both phase and frequency of the
vortex shedding.

VORTEX INDUCED VIBRATIONS IN MARINE RISERS
Wall thickness design for marine risers is based on Barlows
formula [1]

o
h
=
p

2 t
k o

(1)

which states that the hoop stress o
h
expressed as a function of
the internal pressure p

, mean diameter and wall thickness t,


is limited by the specified minimum yield stress o

of the
material, multiplied by a safety factor k = u.6 for hazardous
service. Additional design guidelines are applied to account for
corrosion allowance and continuity of the internal diameter.

Barlows formula (1) indicates that a smaller diameter riser
can convey hydrocarbons at a higher internal pressure. Hence,
multiple small diameter risers are typically preferred over one
single large diameter riser. During the design of floating
production platforms in deepwater, it has been recognized [2]
that there is a risk of interference between adjacent production
and export risers, or possibly between other combinations of
tendons, drilling risers and production risers.

This paper presents numerical analyses to predict vortex
induced vibrations in marine risers subjected to sheared flow.
The paper subsequently addresses

Wake interference and proximity effects. First, 2D
simulations on fixed rigid cylinders are performed to
investigate wake interference and proximity effects for
multiple marine risers in tandem arrangement. The
influence of end spacing on the flow pattern is studied,
and the drag and lift coefficients for both the upstream
and downstream riser are compared to evaluate
proximity effects.

2 Copyright 2013 by ASME
Multiphysics modeling of fluid structure interaction. In
order to predict the displacements of marine risers
experiencing vortex induced vibrations, multiphysics
modeling of fluid structure interaction is needed. Fluid
structure interaction requires co-simulation of a
structural solver and a Computational Fluid Dynamics
(CFD) code. In this paper, a weakly coupled solution
is presented to estimate the VIV response of marine
risers in close proximity.

Slender risers subjected to sheared flow. When
designing and installing risers in (ultra)deep water, the
length/diameter aspect ratio of the marine riser can
exceed I > 1uuu, and the features of the fluid flow
in depth direction can no longer be neglected. Both the
magnitude and the direction of the current change with
water depth, giving rise to higher harmonics in the
VIV response. At the end of this paper, 3D CFD
calculations are performed to evaluate the effect of the
third dimension for risers subjected to uniform flow
and sheared currents.

COMPUTATIONAL FLUID DYNAMICS FOR RISERS

The simulations, reported in this paper, have been performed
with OpenFOAM, an open source CFD solver. This solver uses
the generalized version of the Navier-Stokes equations [3],
solving for the velocity field u = (u, :) and the pressure p.
When the fluid flows past a fixed cylinder like a marine riser, a
region of disturbed flow is formed, like schematically shown on
Figure 1.

In this simulation of laminar flow, the free stream velocity
is shown in green. Lower velocities are depicted in blue,
whereas yellow indicates values higher than the stream
velocity. Evidently, the velocity varies in terms of magnitude,
direction and time, and four regions can be distinguished:

1. The retarded flow is a narrow region in front of the
cylinder, where the local (time-averaged) velocity is
lower than the free stream velocity
2. Two boundary layers attached to the surface of the
cylinder
3. Two sideway regions where the local (time-averaged)
velocity is higher than the free stream velocity
4. The wake, which is the downstream region of
separated flow where the local (time-averaged)
velocity is less than the free stream velocity

The fluid flow around a circular cylinder, as shown on Figure 1,
is a well-known and documented [4-6] problem in
computational fluid dynamics, and often used as a benchmark
for CFD solvers [7].

Figure 1: Regions of disturbed flow

The flow pattern in the wake of the cylinder is primarily
governed by the Reynolds number

Rc =
I
o
u
(2)

which expresses the ratio of inertia forces to viscous forces,
with I the fluid flow velocity,
o
the total outer diameter and

u =
p
p
w
(3)

the kinematic viscosity as a ratio of the dynamic viscosity p
with the density p
w
. A detailed analysis of the different flow
regimes around subsea structures can be found in [8-9]. In
summary, the regimes of fluid flow across a smooth subsea
structure can be divided in
Unseparated flow for very low (Rc < S) Reynolds
numbers
The regime for S < Rc < 4u , where a pair of Fppl
vortices develop in the wake
The transition range (1Su < Rc < Suu) from laminar
flow to turbulence
The regime where the vortex street is fully turbulent
(Suu < Rc < S 1u
5
)
For even higher numbers (S 1u
5
< Rc < S 1u
6
),
the laminar boundary layer undergoes turbulent
transition, and the wake will be narrower and
disorganized
At very high Reynolds numbers (Rc > S 1u
6
), re-
establishment of a turbulent vortex street occurs

For the range of Reynolds numbers relevant to offshore
pipeline engineering, the flow is fully turbulent, and it becomes
increasingly difficult if not impossible- to predict the transient
flow behavior with a laminar solver for the Navier Stokes
equations.



3 Copyright 2013 by ASME
The possible options for CFD simulations at very high
Reynolds numbers are:
Direct Numerical Simulation (DNS), which solves the
Navier Stokes equations for the pressure and the
velocity components in a time-dependent domain. This
approach requires a very fine mesh size and very small
time steps to resolve the smallest eddies and capture
the fluctuations in the turbulent flow [10]. As a result,
this approach is not economically feasible for pipeline
design.

Large Eddy Simulation (LES), where large turbulent
eddies are computed in a time-dependent simulation,
whereas small eddies are predicted with a compact
model. Indeed, smaller eddies have an isotropic (and
hence more universal) behavior, but larger eddies in
the turbulent flow tend to be anisotropic, and their
behavior is directly influenced by the problem
geometry. The viability and accuracy of Large Eddy
Simulation for complex turbulent flows at high
Reynolds numbers is investigated in [11], but has
proven to be not feasible for full 3D analysis of
offshore structures [12].

Reynolds Averaged Navier Stokes (RANS) turbulence
model. In the RANS approach, all flow characteristics
are decomposed as the sum of a steady (mean) value
and a fluctuating term. This decomposition gives rise
to a Reynolds stress tensor, which adds six unknowns
to the system of equations. As a result, turbulence
models are required to provide additional transport
equations to close the system [13]. In this paper, an
enhanced k - e model is used to simulate vortex
induced vibrations in multiple marine risers. The
mathematical details of the turbulence model applied
are given in the Appendix.
WAKE INTERFERENCE FOR TANDEM RISERS

During the design of floating production platforms in
deepwater, it has been recognized [2] that there is a risk of
interference between adjacent production or export risers, or
possibly between other combinations of tendons, drilling risers
and production risers. The consequences of most concern are
the possible increase in fatigue damage due to vortex induced
vibrations (VIV), and the likelihood of contact between
adjacent risers.

A large body of work has been published addressing
measurement, modeling and analysis of marine risers in tandem
arrangement [14]. A careful review of flow interference
between two circular cylinders in various arrangements has
been presented by Zdravkovich [15-16], including an extensive
list of references on this subject. He has also introduced a
classification of flow regimes around two circular cylinders,
depending on their relative position.
Different studies for the tandem arrangement of two
adjacent risers [2, 17-19] have shown that the changes in drag,
lift and vortex shedding are not continuous. Instead, an abrupt
change for all flow characteristics is observed at a critical
spacing between the risers. An exhaustive description on
proximity effects and wake interference can be found in [20],
and a comprehensive summary of VIV in tandem risers is
provided in [2]. Recent research results have been published in
a.o. [21-23].

In this paper, the published data on riser interference tests
for flexible tubulars [2] will be used as experimental validation.
To simulate these experiments, a 2D CFD model is constructed,
assuming fixed rigid cylinders with an outer diameter of 114.3
mm. The simulation setup, with a grid of Su by 1S, is
shown on Figure 2.


Figure 2: Simulation setup to study wake interference

For the simulations of fluid flow around marine risers in
tandem arrangement, the computational grid comprises some
250 000 cells. Depending on the end spacing, the dimensionless
wall distance is in the range of

2u < y
+
=

w
u
1
y

< Su (4)
with y the distance to the nearest wall, and u
:
the friction
velocity defined by

u
1
= _

w
(5)

where
w
is the average wall shear stress. As long as (4) is
satisfied, the problem is well conditioned. The enhanced k -e
model, presented in [24] and detailed in the Appendix, was
used to simulate vortex induced vibrations in multiple marine
risers in tandem arrangement.

On Figure 3, the turbulent eddy viscosity is shown for very
high (Rc = 2.S 1u
6
) Reynolds numbers, clearly indicating
that this enhanced k -e eddy viscosity model is capable of
simulating a turbulent wake with significant separation.


4 Copyright 2013 by ASME

Figure 3: Distribution of turbulent eddy viscosity

The parametric approach, suggested in Figure 2, enables
the investigation of risers in staggered arrangements as well, for
c = u. In this paper, we focus on risers in tandem arrangement
(c = u) with different end spacings 2 < I < 6. It has been
shown experimentally [16-18] that there is strong interference
between two cylinders in tandem arrangement for spacing
ratios with I < S.S. At a spacing I = S.S, a sudden
change of the flow pattern in the gap between the adjacent
risers is observed.

On Figure 4, the influence of the end spacing on the fluid
flow pattern in the wake of the tandem risers is shown for a
Reynolds number Rc = 1u
5
, i.e. the two-bubble regime of the
transition in the boundary layers. These simulation results
indeed endorse the experimental observations of Allen [2],
Zdravkovich [16] and King [17]:

For small end spacing (I < S), vortex shedding
only occurs in the wake of the downstream riser: the
free shear layers which separate from the upstream
riser are permanently re-attached to the downstream
riser. In [25], Zdravkovich refers to this type of wake
interference as quasi-steady re-attachment.

Figure 4: Tandem risers with different end spacing

5 Copyright 2013 by ASME
When increasing the gap (I > S) between both
risers, a turbulent vortex street appears in the wake of
both the upstream and the downstream riser. The
vortices shed by the upstream riser coalesce with the
vortex street of the downstream riser, and binary eddy
streets are observed. It can be clearly seen that there is
no re-attachment of the free shear layers separated
from the upstreamriser to the downstreamone.

Drag coefficient data [16, 18] shows that the upstream riser
takes the brunt of the burden, and that the downstream riser has
little or no effect on the upstream one. For different values of
spacing I , the drag coefficient is shown on Figure 5.

Figure 5: Drag coefficients at Re = 10
5


Apparently, the drag coefficient on the upstream riser is not
significantly influenced by the downstream one, but a
significant change in drag is observed on the downstream
cylinder for I > S. In [2], drag coefficients are measured on
risers in tandem arrangement with increasing end spacing for
Reynolds numbers from Rc = 1 1u
5
up to Rc = 2.S 1u
5
. On
Figure 8, for instance, the measured drag coefficients for both
upstream and downstream riser are shown for a spacing
I = S. The drag coefficients, predicted by the CFD
simulations at Rc = 1 1u
5
, are indicated as well, showing a
very good agreement with the experimental data.


Figure 6: Drag coefficients for L = 3D [2]

Figure 6 shows that for the upstreamcylinder, the drag
crisis occurs somewhat earlier (i.e. at a lower Reynolds
number) than traditional measurements of this phenomenon
[17, 18], which could be attributed to the combined effects of
free-stream turbulence and cylinder displacement. The
combination of an early drag crisis on the upstreamriser and
large displacements of the downstream riser produces a larger
total drag force on the downstream riser for Rc > 1.7 1u
5
.
MULTIPHYSICS: FLUID-STRUCTURE INTERACTION

The CFD simulations, presented in the previous section,
were performed on fixed, rigid cylinders. Although such
simulations are capable of identifying the proximity effects
between adjacent risers by revealing their influence on drag
coefficients and flow pattern in the wake, they cannot predict
the VIV response of the riser.

Blevins [26] pointed out that the cross-flow cylinder
vibration can significantly affect the vortex shedding. In
summary, the cylinder displacement tends to

Increase the d rag on the cylinder
Shift the vortex shedding frequency to the cylinders
vibration frequency
Increase the strength of the vortices
Alter the vortex pattern and hence the vortex shedding
frequency

Some quantitative details can be found in [26-27]. In
addition to drag coefficients, Allen [2] reports measurements
for the transverse displacements of the upstream and
downstream cylinders as well. In order to predict the
displacements of marine risers experiencing vortex induced
vibrations, multi-physics modeling of fluid structure interaction
is needed.

Fluid structure interaction requires co-simulation of a
structural solver and a CFD code. In a strongly coupled
solution, the fluid flow will dictate the displacements, which in
turn will influence the flow pattern. The structural
displacements are used as an input for the CFD simulation, and
the resulting pressure distribution is fed back to the structural
solver [28]. When simulating large displacements (e.g. vortex
induced vibrations), the moving mesh is severely distorted and
the strongly coupled solution procedure is prone to numerical
instabilities.

In a weakly coupled solution, the structural solver and the
CFD solver are executed sequentially. This approach provides a
better balance between accuracy and computational expense,
but is only applicable when the structural response does not
significantly influence the fluid flow. In this paper, weakly
coupled simulations of fluid structure interaction are conducted
to estimate the VIV response of marine risers in close
proximity.
6 Copyright 2013 by ASME
WEAK COUPLING
STRONG COUPLING

Figure 7: Comparison between weak and strong coupling

In the sequentially coupled simulations, we first calculate
the flow patterns to estimate the lift and drag forces exerted on
the structure. The CFD simulations are performed according to
the approach described in the previous section: the fluid domain
is modelled in 2D, and the cylinders are fixed and assumed to
be rigid (cfr. Figure 2).

On Figure 8, the calculated lift and drag forces are shown
for the downstream riser at a spacing I = S, for a Reynolds
number Rc = 1 1u
5
. The oscillating signals reflect a fully
developed turbulent wake. Note that the average lift force is
zero, while the average drag force is a measure for the
resistance against fluid flow.

On Figure 9, the Fast Fourier Transform (FFT) of the lift
and drag forces is shown, to reveal the frequency content of the
signals. Clearly, the dominant frequency of the drag force is
twice the lift frequency:
d
= 2
I
= u.42 Hz.


Figure 8: Lift and drag forces on the downstream riser

For the top-tensioned risers used in [2], the n-th eigenfrequency
can be estimated by [29]

f
n
=
n
2 I
_
I
m
+ [
n n
I

2

EI
m

(6)

with I the length, I the tension, m the distributed mass and EI
the bending stiffness. In (6), the tubes are assumed to be simply
supported, straight and with constant tension. As a result, the
computed eigenfrequency is omni-directional and corresponds
to the first bending mode of the cylinder. Note that the natural
frequency (6) is a function of the tension, but is predominantly
controlled by the bending stiffness [30].

The lowest natural frequency is calculated as
0
=2.35 Hz,
while the measured frequency (by means of Pluck tests) was
found to be
0
=2.23 Hz. Numerical modal analysis (solving
the eigenvalue problem) confirmed a lowest natural frequency
of
0
=2.225 Hz.


Figure 9: Frequency spectrum of lift and drag forces

As indicated in Figure 9, the drag frequency
d
= u.42 Hz
is still significantly lower than the first natural frequency of the
test tubes, so only moderate displacements are expected.
7 Copyright 2013 by ASME
To predict the response of the downstream riser when
subjected to the lift and drag forces shown on Figure 8, the
principle of virtual work [31] is applied:

_
]
n
]
ou

JS
S
= _p u

ou

JI
v
+ _o
]
oe
]
JI
v
- _

ou

JI
v
(7)

where S is the boundary of the Lagrangian body I, o
]
are the
stress components, e
]
the strain components,

the external
forces (including lift I(t) and drag (t)), and u

the unknown
displacements. The top-tensioned test tube is modeled as a
compliant structure, with Youngs modulus E =2700 MPa and
density p =1050 kg/m. The displacements are simulated with
a finite element code using a transient dynamic explicit solver.


Figure 10: Predicted cross-flow displacements

The predicted cross-flow displacements y of the
downstream riser are shown on Figure 10. After some 20
seconds, the signal reaches its maximum amplitude

y
mux

= u.1S (8)

For the same situation (end spacing 3 diameters and
Reynolds number 10
5
), the maximum measured transverse
displacement [2] was y
mux
= u.16S.

On Figure 11, the transverse RMS displacement
measurements for both the upstreamand downstream test tube
are shown for the range 1u
5
< Rc < 2.S 1u
5
. Apparently, for
the lower Reynolds numbers (Rc < 1.7 1u
5
), the
displacement of the downstream cylinder is smaller than for the
upstream cylinder. For higher Reynolds numbers, the down-
streamcylinder vibrates at larger transverse amplitudes than the
downstreamcylinder. However, other tests have shown [32]
that for higher vibration modes [33], the downstream cylinder
always vibrates less than the upstreamone, when the magnitude
of the displacements is sufficiently large.

Figure 11: Transverse displacements for L = 3D [2]

The prediction of the fluid structure interaction (FSI)
simulation is included in Figure 11 as well, showing a good
agreement with the experimental observation.

MARINE RISERS SUBJECTED TO SHEARED FLOW

In the CFD simulations presented in the previous sections,
the risers were assumed to be fixed and rigid, and the length
direction was not taken into account. However, when designing
and installing risers in (ultra)deep water, the length/diameter
aspect ratio of the marine riser can exceed I > 1uuu, and
the features of the fluid flow in depth direction can no longer be
neglected. In this section, 3D CFD calculations are presented to
evaluate the effect of this third dimension for risers subjected to
uniform and sheared currents.

In order to assess the effect of the length direction on the
flow pattern, a 3D CFD calculation was performed on a riser
with a diameter = 1 meter and length I = Su meter. The
calculation grid comprised 616 000 cells. On Figure 12, the
pressure distribution on the riser is shown, and the
corresponding flow pattern is visualized as well.



Figure 12: Fluid flow simulation around 3D riser
8 Copyright 2013 by ASME
Vortex shedding in the wake of the long slender tubular
gives rise to the development of vortices with horizontal axis,
resulting in fluctuation of the flow in the Z-direction. This
could be expected, as it is known [3] that the transport of a
vortex creates in itself a new vortex with perpendicular axis.
The simulation shows that these 3D vortices are strong enough
to modulate the vortex shedding on a riser as a function of
depth, although the inlet boundary has a uniform current
velocity.

On Figure 13, the lift coefficient integrated over the entire
riser length is compared with a 2D simulation for the same riser
diameter. While the dominant frequency is the same, the 3D
signal exhibits higher harmonics, corresponding to a shift in
vortex shedding frequency as a function of depth. The
amplitude of the lift is also lower than expected based on the
2D calculations. A similar trend is observed when comparing
the drag forces for a 2D and 3D simulation. In conclusion, a 2D
simulation will give rise to conservative predictions.


Figure 13: Comparison of lift forces between 2D and 3D

The prediction of vortex induced vibrations for deepwater
risers is very challenging, owing to the fact that the incident
flows are non-uniform and the associated fluid structure
interaction phenomena are highly complex [34]. These complex
conditions give rise to a non-linear, coupled system with a large
number of degrees of freedom, which depends on several
physical and mechanical parameters. While a great deal of
attention has been devoted to riser VIV modeling and
prediction, most of the studies presented in literature [35-38]
only account for a uniform incident current.

A good introduction on the subject of marine risers
subjected to sheared flow is given by Vandiver [39-41], and a
limited set of experiments [29] and simulations [34] have been
published on VIV predictions for linearly sheared currents. At
the end of this paper, a 3D CFD simulation is performed on a
riser span of I = Su meter, subjected to sheared flow. For the
current profile [42], a one-seventh power law

v(z) = I
0
_
J +z
J
_
1 7
(z u)
(9)


Figure 14: Marine riser subjected to sheared current

was chosen, where the flow velocity varies from : = u at the
seabed to : = I
0
at z = I 2 . On Figure 14, the resulting fluid
pattern around the marine riser is shown. The vortex street is
visualized in five horizontal planes, uniformly distributed over
the length of the riser. The influence of the current gradient
over the length of the riser can clearly be observed in Figure 14.

On Figure 15, the lift and drag coefficients are shown as a
function of depth. Close to the seabed (z = S m), the riser
experiences little or no fluctuating lift, and only moderate drag.
When approaching the still water level, the lift and drag
coefficients asymptotically converge towards the 2D solution.
The presence of sheared currents invoke a shift in both phase
and frequency of the vortex shedding. As a result, the lift and
drag will be lower compared to the 3D simulation with a
uniform current speed, shown in Figure 13.


Figure 15: Lift and drag as a function of depth




9 Copyright 2013 by ASME
CONCLUSIONS

In his pioneering paper [43], Prof. Vandiver presented the most
stringent research challenges in the prediction of vortex
induced vibrations for marine risers. In addition to the need of
acquiring high quality full-scale response data and developing
cost effective mitigation measures, he highlights a more
profound understanding of fluid structure interaction and the
simulation of sheared flow as important research topics.

This results, presented in this paper, want to contribute to the
numerical simulation of VIV for marine risers in close
proximity. The main conclusions from this work read:

Given the high Reynolds numbers involved in deep water
riser design (1u
5
< Rc < 1u
6
), turbulence modeling is
required to capture vortex shedding. The enhanced k -e
model, proposed in [24], proves to be the most appropriate
RANS closure to predict VIV.

For two risers in tandem arrangement, there is a sudden
change in flow characteristics for a critical spacing
I = S.S. The upstream riser takes most of the burden,
while the drag coefficient on the downstream riser is lower
at Rc < 1.7 1u
5


Multiphysics modeling of fluid structure interaction allows
predicting the VIV response of marine risers in tandem
arrangement. For high Reynolds numbers, the downstream
riser often experiences higher transverse displacements
than the upstream riser.

For low Reynolds numbers, there is little effect of end
spacing on the drag coefficients and displacements,
whereas the effect of end spacing is obvious and distinct
for Rc > 1.7 1u
5


Fluid flow simulations in 3D indicate that 2D CFD
calculations will yield conservative predictions: the
amplitude of lift and drag are slightly over-estimated in 2D
simulations.

The presence of sheared currents invoke a shift in both
phase and frequency of the vortex shedding. As a result,
the lift and drag will be lower compared to a 3D simulation
with a uniform incident current.
APPENDIX ON TURBULENCE MODELLING

The Navier-Stokes equations for incompressible Newtonian
liquids could be used for turbulent flow simulations. However,
once the flow becomes turbulent, all quantities fluctuate in time
and space with widely varied time scales and length scales. It is
theoretically possible to solve the Navier-Stokes equations for
all scales, yet the required computer resources render this
approach impracticable.
Therefore, the turbulent influence is modeled, and the most
commonly used models are the Reynolds Averaged Navier
Stokes (RANS) models [13]. In this RANS approach, all flow
characteristics are decomposed as the sum of a steady (mean)
value and a fluctuating term. This decomposition gives rise to a
Reynolds stress tensor, which adds six unknowns to the system
of equations. As a result, turbulence models are required to
provide additional transport equations to close the system.

In this paper, the k -e turbulence model was selected to
simulate wake interference in adjacent marine risers. This
model is frequently used to model turbulent flow, and was
identified by [11] as the most appropriate RANS model to
predict vortex induced vibrations in marine risers for Reynolds
numbers up to Rc = 1u
6
.

The k -e turbulence model [44-45] is a two equation
model, providing a transport equation for the kinetic energy k

(
w
k)
t
+
(
w
u
j
k)
x
j
=
Ij
u
I
x
j
-
w

+

x
j
__ +

k
]
k
x
j
_ - 2
k
u
2
(10)

and an additional expression for the viscous dissipation rate e

(
w
)
t
+
(
w
u
j
)
x
j
= C
1
f
1

k

Ij
u
I
x
j
- C
2
f
2

2
k
+

x
j
__ +

s
]

x
j
_ - 2

u
2
exp _-
u
+
2
_
(11)

where J
+
is a non-local function [45] of distance to the wall.
The auxiliary functions read
1
= 1 and

f
2
= 1 -
u.4
1.8
exp _-
Re
T
2
S6
_ (12)
with
Re
T
=

w
k
2

(13)
The turbulent eddy viscosity is computed from

T
= C


w
k
2


(14)

with C

= u.u9 and

= 1 - exp(-u.u11S J
+
). The values
for the other model constants are listed in Table 1.

Table 1: Values for the k- model constants

C
1
= 1.SS C
2
= 1.8u
o
k
= 1.u o
s
= 1.S

10 Copyright 2013 by ASME
This standard k -e model is widely used in computational
fluid dynamics, and was adopted by [11, 46] to predict vortex
shedding around circular cylinders at high Reynolds numbers
(Rc = 1u
6
). The model performs quite well for boundary layer
flows, but is less accurate for risers in which a high mean shear
rate is present or massive separation occurs (which could be
expected for risers in tandem arrangement). In these cases, the
eddy viscosity is over-predicted by the standard formulation.
Moreover, the dissipation rate equation (11) does not always
give the appropriate length scale for turbulence.

To improve the ability of the standard k - e model to
predict complex turbulent flows, an enhanced k -e eddy
viscosity model is proposed in [24]. This model consists of a
new formulation for the viscous dissipation rate

(
w
)
t
+
(
w
u
j
)
x
j
=

x
j
_

x
j
_
+ C
1

w
S -C
2

w

2
k +
(15)

based on the dynamic equation of the mean square vorticity
fluctuation at large turbulent Reynolds numbers. In addition, a
new eddy viscosity formulation is introduced

T
= C


w
k
2


(16)
with
C

=
1
A
0
+A
s
0


(17)

based on the positivity of the normal Reynolds stresses and the
Schwarz inequality for turbulent shear stresses [24]. In (17),
the coefficient C

is determined by

0

= _S
Ij
S
Ij
+

Ij

Ij
(18)
with

Ij
=
Ij
- 2
Ijk

k
(19)

and the parameters A
0
= 4.u and A
s
= 6 cos , where

=
1
S
cos
-1
(6W)
(20)
where

W =
S
Ij
S
jk
S
kI
S

3
(21)
with
S

= _S
Ij
S
Ij
(22)

The other constants, calibrated in [24], are listed in Table 2.

Table 2: Values for the enhanced k- model constants

C
1
= max _u.4S,
Sk e
S +Sk e
_ C
2
= 1.9u
o
k
= 1.u o
s
= 1.2

REFERENCES

[1] Mohitpour M., Golshan H. and murray A., Pipeline
Design and Construction A Practical Approach, Third
Edition, ASME Press (2007)
[2] Allen D.W., Henning D.L. and Lee L., Riser Interference
Tests on Flexible Tubulars at Prototype Reynolds
numbers, Proceedings of the Offshore Technology
Conference, OTC-17290 (2005)
[3] Batchelor G.K., An Introduction to Fluid Dynamics,
Cambridge University Press (1967)
[4] Norberg C., Flow Around a Circular Cylinder: Aspects of
Fluctuating Lift, J ournal of Fluids and Structures, vol. 15,
pp. 459-469 (2001)
[5] Rocchi D. and Zasso A., Vortex Shedding from a Circular
Cylinder in a Smooth and Wired Configuration:
Comparison between 3D Large Eddy Simulation and
Experimental Analysis, J ournal of Wind Engineering and
Industrial Aerodynamics, vol. 910, pp. 475-489 (2002)
[6] Norberg C., Fluctuating Lift on a Circular Cylinder:
Review and New Measurements, J ournal of Fluids and
Structures, vol. 15, pp. 57-96 (2003)
[7] Schfer M. and Turek S., Benchmark Computations of
Laminar Flow Around a Cylinder, Flow Simulation with
High Performance Computers II, vol. 52, pp. 547-566
(1996)
[8] Sumer B.M. and Fredsoe J ., Hydrodynamics around
Cylindrical Structures, World Scientific, Signapore (1999)
[9] Gourmel X., Numerical Simulation of the Flow over
Smooth and Straked Riser at High Reynolds Numbers,
M.Sc. Thesis, Cranfield University, School of Applied
Sciences, Offshore and Ocean Technology (2010)
[10] Mainon P. and Larsen C.M., Towards a Time Domain
Finite Element Analysis of Vortex Induced Vibration of
Risers with Staggered Buoyancy, Proceedings of the 30
th

ASME Conference on Ocean, Offshore and Arctic
Engineering, OMAE2011-49046 (2011)
[11] Catalano P., Wang M., Iaccarino G. and Moin P.,
Numerical Simulation of the Flow Around a Circular
Cylinder at High Reynolds Numbers, International J ournal
of Heat and Fluid Flow, vol. 24, pp. 463-469 (2003)
[12] Constantinides Y. and Oakley O., Numerical Prediction of
Bare and Straked Cylinder Vortex Induced Vibrations,
Proceedings of the 25
th
International Conference on
Offshore Mechanics and Arctic Engineering, OMAE2006-
92334, Hamburg, Germany (2006)
[13] Wilcox D.C., Turbulence Modelling for Computational
Fluid Dynamics, Second Edition, DCW Industries (1998)
11 Copyright 2013 by ASME
[14] Kondo N., Three Dimensional Analysis for Vortex Induced
Vibrations of an Upstream Cylinder in Two Tandem
Circular Cylinders, Proceedings of the 30
th
ASME
Conference on Ocean, Offshore and Arctic Engineering,
OMAE2011-49655 (2011)
[15] Zdravkovich M.M., Review of Flow Interference between
Two Circular Cylinders in Various Arrangements, J ournal
of Fluids Engineering, vol. 99 (4), pp. 618-633 (1977)
[16] Zdravkovich M.M., Flow Induced Oscillations of Two
Interfering Circular Cylinders, J ournal of Sound and
Vibration, vol. 101(4), pp. 511-521 (1985)
[17] King R., Wake Interaction Experiments with Two Flexible
Circular Cylinders in Flowing Water, J ournal of Sound
and Vibration, vol. 45(2), pp. 559-583 (1976)
[18] Zhang H. and Melbourne W.H., Interference between Two
Cylinders in Tandem in Turbulent Flow, J ournal of Wind
Engineering and Industrial Aerodynamics, vol. 41-44, pp.
589-600 (1992)
[19] Allen D.W. and Henning D.L., Vortex Induced Vibration
Current Tank Tests of Two Equal Diameter Cylinders in
Tandem, J ournal of Fluids and Structures, vol. 17, pp.
767-781 (2003)
[20] Mittal S. And Kumar V., Flow Induced Oscillations of
Two Cylinders in Tandem and Staggered Arrangements,
J ournal of Fluids and Structures, vol. 15, pp. 717-736
(2001)
[21] Lestang Parade E., Dual String Risers Local Behaviour,
M.Sc. Thesis, Cranfield University, School of Applied
Sciences, Offshore and Ocean Technology (2010)
[22] Li L., Fu S., Yang J ., Ren T. and Wang X., Experimental
Investigation of Vortex Induced Vibration of Risers with
Staggered Buoyancy, Proceedings of the 30th ASME
Conference on Ocean, Offshore and Arctic Engineering,
OMAE2011-49046 (2011)
[23] Corson D., Cosgrove S., Hays P.R., Constantinides Y.,
Oakley O., Mukundan H. And Leung M., CFD
Hydrodynamic Databases for Wake Interference
Assessment, Proceedings of the 30
th
ASME Conference on
Ocean, Offshore and Arctic Engineering, OMAE2011-
49407 (2011)
[24] Shih T.H., Liou W.W., Shabbir A., Yang Z. And Zhu J ., A
New k- Eddy Viscosity Model for High Reynolds
Number Turbulent Flows, Computer Fluids vol. 24(3), pp.
227-238 (1995)
[25] Zdravkovich M.M., Flow Around Circular Cylinders
Applications, Oxford University Press (2003)
[26] Blevins R.D., Flow Induced Vibration, Second Edition,
Krieger Publishing Company, Florida (1994)
[27] Vandiver K.J ., Drag Coefficients of Long Flexible
Cylinders, Proceedings of the Offshore Technology
Conference, Houston, Texas, OTC-4490 (1983)
[28] Tukovic Z., The Finite Volume Method for Fluid Structure
Interaction with Large Structural Displacements
[29] Lie H. and Kaasen K.E., Modal Analysis Measurements
from a Large Scale VIV Model Test of a Riser in Linearly
Sheared Flow, J ournal of Fluids and Structures, vol. 22,
pp. 557-575 (2006)
[30] Weimin C., Zhongqin Z. and Min Li, Effects of Varying
Tension and Stiffness on Dynamic Characteristics and
VIV of a Slender Riser, Proceedings of the 30
th
ASME
Conference on Ocean, Offshore and Arctic Engineering,
OMAE2011-49295 (2011)
[31] Rao S.S., The Finite Element Method in Engineering,
Elsevier Science & Technology Books, 658 pp. (2004)
[32] Allen D.W. and Henning D.L., Vortex Induced Vibration
Current Tank Tests of Two Equal Diameter Cylinders in
Tandem, J ournal of Fluids and Structures, vol. 17, pp.
767-781 (2003)
[33] Ashiru A., Assessment of Vortex Induced Vibration
Response in Higher Harmonics Mode, M.Sc. Thesis,
Cranfield University, School of Applied Ssciences,
Offshore and Ocean Technology (2007)
[34] Srinil N., Analysis and Prediction of Vortex Induced
Vibrations of Variable Tension Vertical Risers in Linearly
Sheared Currents, Applied Ocean Research, vol. 33, pp.
41-53 (2011)
[35] Dixon M. and Charlesworth D., Application of CFD for
Vortex Induced Vibration Analysis of Marine Risers in
Projects, Proceedings of the Offshore Technology
Conference, OTC 18348 (2006)
[36] Huang K., Chen H.C. and Chen C.R., Deepwater Riser
VIV Assessment by Using a Time Domain Simulation
Approach, Proceedings of the Offshore Technology
Conference, OTC 18769 (2007)
[37] Ashuri H., Sadeghi K and Niazi S., A Numerical Model of
Vortex Induced Vibrations on Marine Risers, Proceedings
of the 30
th
ASME Conference on Ocean, Offshore and
Arctic Engineering, OMAE2011-49610 (2011)
[38] Song J .N., Lu L., Teng B., Park H.I., Tang G.Q. and Wu
H., Laboratory Tests of Vortex Induced Vibrations of a
Long Flexible Riser Pipe subjected to Uniform Flow,
Ocean Engineering voL. 38, vol. 1308-1322 (2011)
[39] Kim Y.H., Vandiver K.J . and Holler R., Vortex Induced
Vibration and Drag Coefficients of Long Cables Subjected
to Sheared Flow, Proceedings of the Fourth International
Offshore Mechanics and Arctic Engineering Symposium,
Ed. J .S. Chung, vol. 10185a (1985)
[40] Vandiver K.J . and Chung T.Y., Predicted and Measured
Response of Flexible Cylinders in Sheared Flow,
Proceedings of the ASME Winter Annual Meeting
Symposium on Flow Induced Vibration, Chicago (1988)
[41] Vandiver K.J ., Predicting Lock-in on Drilling Risers in
Sheared Flow, Proceedings of the Flow Induced Vibrations
Conference, Lucerne, Switzerland (2000)
[42] Adams A.J ., Atkins W.S. and Thorogood J .L., On the
Choice of Current Profiles for Deep Water Riser Design,
Proceedings of the SPE Annual Technical Conference,
Society for Petroleum Engineers, SPE 49262 (1998)
[43] Vandiver K.J ., Research Challenges in the Vortex Induced
Vibration Prediction of Marine Risers, Proceedings of the
Offshore Technology Conference, Houston, Texas, USA,
OTC-8698 (1998)
12 Copyright 2013 by ASME
[44] J ones W.P. and Launder B.E, Prediction of Laminarization
with a Two Equation Model of Turbulence, International
J ournal of Heat and Mass Transfer, vol. 15 (1972)
[45] Chien K.Y., Predictions of Channel and Boundary Layer
Flows with a Low Reynolds Number Turbulence Model,
AIAA J ournal, vol. 20(1), pp. 33-38 (1982)
[46] Majumdar S. and Rodi W., Numerical Calculation of
Turbulent Flow Past a Circular Cylinder, Proceedings of
the 7
th
Turbulent Shear Flow Symposium, pp. 3.13-25,
Stanford, USA (1985)

S-ar putea să vă placă și