OTC 22951

Accountin
ng for VIV
V in Wak
ke Induce
ed Motio
on of Risers in Ta
andem att High
Re
eynolds Numberr
Jean-Francois SAINT-MAR
RCOUX Subssea 7, Roberrt D Blevins Consultant
C

Copyright 2012, Offshore

e Technology Confere
ence
d for presentation at th
he Offshore Technolo
ogy Conference held in Houston, Texas, USA,
U
30 April3 May 2012.
Thiss paper was prepared
a OTC program com
mmittee following review of information co
ontained in an abstracct submitted by the author(s).
a
Contents of the paper have not been
Thiss paper was selected for presentation by an
revie
ewed by the Offshore
e Technology Confere
ence and are subjectt to correction by the author(s). The material does not necessa
arily reflect any positiion of the Offshore Technology Conferencce, its
officcers, or members. Electronic reproduction
n, distribution, or storage of any part of this
t
paper without the written consent of the Offshore Technology Conference is prohibited. Permissiion to
reproduce in print is restrricted to an abstract of
o not more than 300 words; illustrations may
m not be copied. Th
he abstract must contain conspicuous ackn
nowledgment of OTC
C copyright.

Ab
bstract
Ass production oof oil and gas from deepwatter and ultra deepwater
d
incrreases, riser in
nterference beecomes criticaal for engineerring
dessign and anallysis. The use of seafloor space must be
b optimized and the size of the floatinng units reducced. For exam
mple
SC
CRs are closeely spaced at the
t FPU and lay in close raange at the seaafloor. Similarrly for decouppled riser systems, jumpers are
loccated at short ddistances at thhe FPU level.
Bleevins provided the seminall paper on thee forces and stability
s
of a cylinder
c
in a wake. One off the major coonclusions off his
pap
per was that aaccounting forr drag as per Huse
H
should be
b refined to be
b able to asseess the potentiial instability of a downstreeam
cyllinder in a wakke.
To
o be complete it was necesssary to expand the theory to
t the case off an upstream cylinder in crross vibrationn about its steady
position. This toook more tim
me because thee wake behindd a vibrating cylinder doess not follow the
t classical bbell shape currve.
Ob
btaining reliabble experimenttal data to be able
a to describbe it was diffiicult, especiallly at high Rey
ynolds numberr (105).
Th
he paper preseents experimeental data for a wake behinnd a vibratingg (and - for benchmarking
b
g - a fixed) cylinder
c
from 2D
dow
wnstream to 50D downstrream. A relattively simple extension off the Blevins original theoory allows acccounting for the
mo
odified shape of
o the wake.

Ro
ole of Intterferencces in the
e Design o
of Deepw
water Risser System
ms
Deeepwater and ultra deepwaater developm
ments require a close spaciing of the risers and flowllines in orderr to optimize the
seaafloor. Exampples of such prractical configgurations are:
- Decouppled risers (Buundle-HRT annd SHR);
o interferencee between the BHRT or SHR risers themsselves as theyy are located close to each other
o
w
Bundle Hybrid Riserr Towers (Buundle-HRT); interference between the top
o Turret-moored FPSOs with
jumpers thaat run from thee top of the Buundle HRT to the turret-FPSO
ween
o For submerged buoy suchh as the BSR (Boia de Supoorte de Riser risers suppoort buoy), interrference betw
the risers ass the spacing governs
g
the lenngth of the buuoy (this appliies to SCRs as
a well as Flexxibles)
- Coupleed risers (SCR
Rs, Flexibles):
o For spread-moored FPSO
Os; interferennce between thhe risers at th
he FPSO side,, as the spacinng of the riserrs is
way from the center of the vessel, the higgher is the excitation provided
critical. Thhis is because the further aw
by the pitchh and roll of thhe FPSO
his is also refeerred to as Wake
W
Induced Motion (WIM
M). As a geneeral guideline DNV (2009)), recommendds analyzing riser
r
Th
intterference in thhe range of up
p to 25 diameeters downstreeam. For a typpical riser of 0.3m
0
(outer diameter), this corresponds to
t a
7.5
5m range, whhereas this coorresponds too about 60m for a Bundlee-HRT. Clearrly those are ranges of acctual engineerring
app
plications.
Th
he general casse of interfereence of arrayss of risers is simplified in practice by using
u
strip th
heory and connsidering the two
t
clo
osest risers at aany given tim
me. Also the tim
me-average hyydrodynamic excitation forcces are considdered.
Th
he wake induced hydrodynaamic load behiind a fixed cyylinder was firrst introduced to the offshorre industry byy Huse (1993 and
19996) in 1993. IIn the Huse model only the shielding effeect is accounteed for: the fielld of hydrodynnamic forces is limited to drag
d

OTC 22951

forces. Drag alone however cannot account for instability observed in downstream cylinders by Fontaine (207). Blevins (2005)
proposed and experimentally justified, a model of hydrodynamic forces including both drag and lift forces1 behind a fixed
cylinder. The more recent model by Blevins (2011) expands the model to the case where the upstream cylinder is moving
under Vortex Induced Vibration (VIV).
Therefore the possible effects are (see Table 1):
- Wake shielding behind a fixed cylinder
- Wake instability behind a fixed cylinder
- Wake shielding behind a cylinder under VIV
- Wake instability behind a cylinder under VIV
The proposed theory accounts for all effects and can reduce to the other models by setting to zero some of its parameters.

Quasi static Upstream Cylinder

Upstream cylinder under VIV

Wake shielding
Huse (1993/6)[2a/b]
Blevins (2011)[4]

Wake instability
Blevins (2005)[3]
Blevins (2011)[4]

Table 1. Scope of the models.

It is found that self-similarity, as defined by Schlichting (1979), is playing a role in the range of 2D to about 30D, downstream
of the cylinder center, which is the domain of many engineering applications, and allows a close form description of the wake.
It was also necessary to assess the limit of validity of this non-classical shape, as it had to merge at some point downstream
into the shape predicted by the classical theory. Experiments were carried out until 50D downstream, range at which the nonclassical shape started evolving.
The high Reynolds number selected (105) is representative of insulated production risers. It can also be expected that data so
obtained would be conservative when used for higher Reynolds number regimes.

Description of the studies from 2006 to present

In 2006 it was felt necessary for offshore applications to experimentally confirm the drag and lift theory at higher Reynolds
numbers Blevins (2008). In order to do so the equilibrium position of a downstream cylinder in the wake of an upstream
cylinder was investigated in a basin.
The results are shown in Fig. 1 for the case of a downstream cylinder initially located two diameter downstream and two
diameter transverse of a fixed cylinder2. The downstream cylinder is constrained by linear springs in the downstream and
transverse direction. The velocity of the fluid is increased from 0 to 1.2 m/s in 0.1m/s increments.
- For a fixed upstream cylinder, corresponding to the square black dots
o At low velocities, the downstream cylinder is outside of the wake and experiences only drag effects
o As velocity increases the riser moves downstream into the expanding wake and is pulled inwards by the lift
forces. As the downstream cylinder gets closer to the centerline of the wake the lift force decreases
- For a vibrating upstream cylinder corresponding to the white squares and triangles (two sets of experiments)
o The effect of lift starts at very low velocities
o The effect of lift is smaller at higher velocities

1
2

DNV (2009) acknowledges that a lift force formulation is considered essential for describing the instabilities in the wake.
The outside diameters of both cylinders were 0.106 m. The range of Reynolds number investigated was up to 120,000

TRANSVERSE POSITON, DIAMETER

OTC 22951

2.5

UpStream Fixed

increasing velocity

UpStreamVIV
2

UpStreamVIV
region of higher
lift with VIV

1.5

region of lower
lift with VIV

1
0.5

Upstream
cylinder

0
0

2
3
4
INLINE POSITION, DIAMETER

Fig. 1 Equilibrium position of a cylinder in a wake as current increases (initial position X=2, Y=2)
According to the C.B. Rawlins postulate cited in Blevins (2005) and Price (1976), the lift force is proportional to the
transverse derivative of the drag coefficient. See formula below (where the subscript d denotes the downstream cylinder):

C Ld

C Dd
(Y / Dd )

This provided the rational for experimentally measuring very precisely the velocity in the wake, in order to, at least
qualitatively, confirm the validity of underlying assumption of the model.
The shape of the velocity profile of the wake as measured in 2006, in the plane perpendicular to the free stream velocity and at
6D downstream, is given for reference in Fig. 2 below:
- For a fixed cylinder the classical bell-shape curve is well observed (black diamonds)
- For a vibrating cylinder the area where the velocity gradient is high is effectively further away from the centerline
than for a fixed cylinder (white diamonds),
70
Ups tream cylinder fixed
Ups treeam Cylinder VIV

VELOCITY, CM /S

65
60
55
50
45
40
35
-4

-3

-2

-1

DISTANCE FROM CENTERLINE, DIAMETERS

Fig. 2 Velocity profile in the wake of a fixed (black squares) and a vibrating cylinder
The conclusion was that, at least qualitatively, the Rawlins assumption was consistent with the experimental velocity profile
and equilibrium positions.
A detailed investigation of the velocity profile required a dedicated experimental set-up, which was completed in 2010.

OTC 22951

Selfsimilarity applied to a vibrating cylinder

Self-similarity is a well established technique for solving the Navier-Stokes equation in the far fields of jets and wakes
Schlichting (1979) and (2000). It was initially applied to laminar flow and extended to fully turbulent flow Schlichting (1979
section XXIV3).
Before it is proved experimentally, the role of self-similarity in the near field is merely an assumption. This assumption
however is attractive because it allows a closed form expression of the time-average drag and lift forces. Many engineering
software can readily include hydrodynamic loads so expressed. It also avoids running large CFD models4.
It is possible to model the wake using the fully developed self-similar turbulence model. The assumptions for the model are:
- The model is a 2D local model (strip theory): as such it is applicable to a long cylinder
- The cylinders are rigid; however this is applicable to a long cylinder with bending stiffness, provided that the size of
the strips is consistent with the mode being excited
- The flow is fully developed turbulent: the flow can be described by replacing the dynamic viscosity with a turbulent
dynamic viscosity, defined experimentally.
- The velocity profile is self similar; only experimental evidence may validate this assumption
The above assumptions are those of the classical wake theory and therefore lead to the same equation (see Eq. 9.33 of
Schlichting (1979), Eq. 22.18 of Schlichting and Gersten (2000) :
1
1
g ' ' g ' g 0
2
2

Eq. 1

Where g() is a dimensionless function of lateral profile

g ()

U 0 U ( X , Y ) CD D

U0
X

1
2

And:
Y

4a 2
C D DX

As Eq. 1 is of linear and of second order, the general solution is made of two independent solutions. One of them is the well
known even, symmetrical solution [exp (-2/4)], but there is also another solution which is usually not considered (see Blevins
(2011)) for the derivation of the second solution). The general solution of Eq. (1) can be written as.

g ( ) e

1
2
4

1
2
4

e 4 d

Eq. 2

From Eq. 2 the velocity can be expressed as the sum of two terms:
1

a 2Y

U ( X ,Y )
C D 2 C D DX
1 a1 D
e
U 0
X
1
2

C D
' D
e
X

a 2Y 2

C D DX

4a2
D DX

Eq. 3

d ]

Where:

'

a1

It is convenient to use the following notation, where s1 is the even symmetrical solution and s2 is the odd asymmetrical
3

In Section XXIV of Schlichting (1979), the introduction of the - turbulent - virtual kinematic viscosity shows the identity of
the laminar wake profile (based on the molecular viscosity) with the turbulent wake profile (based on the virtual kinematic
viscosity) (see Schlichting (1979) p. 742).
4
The use of CFD is of course welcome, but often difficult to conciliate with the demanding schedule of a conceptual study or
a project.

OTC 22951

solution:

s1 ( ) e
s 2 ( ) e

1
2
4

1
2
4

[ e

Eq. 4
1 2

d ]

Eq. 5

Then it is possible to re-write Eq. 3 as a sum of the two solutions:

1

'
U ( X ,Y )
C D 2
1 a1 D ( s1 s 2 )
U0
a1
X

(6)

The formula for the drag and lift forces resulting from Eq. 6 are provided in Annex A.

Experimental setup and Investigation

Hereafter the present paper describes experimental measurements of the velocity in the wake of a cylinder with and without
vortex-induced vibration of the upstream cylinder at Re = 105. This is significantly higher Reynolds number than previous
measurements Blevins (2005). A model of cylinder wake with VIV is developed and compared with data.
The experiments were made on a 4 inch (10.16 cm) diameter cylinder in the Stratified Flow Channel at the Scripps Institution
of Oceanography in La Jolla, California (see Fig. 3). The flow channel has a 1.12 m (44 inch) wide by 1.13m (44.4 inch) deep
section, a working test section length of 16 m (52.5 ft), and a maximum velocity of 1.25 m/s (0.49 to 4.1 ft/s). At 1 m/s at mid
section, where the experiments were performed, the root-mean-square turbulence of axial velocity was 1% above 0.2 Hz and
about twice this at lower frequencies where the unsteadiness is associated with surface waves. The vertical profile of axial
velocity was flat to within 4%. A large upstream flow straightener was installed to minimize the channel-induced drift of the
cylinder wake at large downstream distances. Cylinder blockage for the 4 inch cylinder is 0.1016/1.12=9.07%. But with one
diameter amplitude transverse motion, the vibrating cylinder sweeps out a frontal area twice as large.
The cylinder surface was sanded with 220 grit sand paper to give a roughness of approximately 30 micro inch average (Ra)
roughness, e/D= 10 x10-5. The tests were made with a mean flow velocity of 1 m/s. The cylinder Reynolds number is 100000
at 1 m/s based on a water kinematic viscosity of 1.00410-6 m2/s (1.1010-5 ft2/s) at 20 C (68 F). Wake velocity was
measured with an ultrasonic flow meter at mid depth. One hundred seconds of data were taken at 25 samples per second and
averaged. Figure 1 shows the test geometry.
X
W

Uo

Cylinder
vibration

H
D

Fig. 3 Channel flow for measurement of cylinder wake. Uo =1 m/s, D =10.16 cm (4 in.), W = H =1.12 m (44 in.).

OTC 22951

Comparison of Theory and Experiment

Figures 4 and 5 show the measurements of axial velocity in the wake of a cylinder for a fixed cylinder (Fig. 4), and a vibrating
cylinder undergoing large amplitude Vortex Induced Vibration (Fig. 5).
1.1

AXIAL VELOCITY, U/Uo

1
0.9
0.8
axial position

0.7

X/D=2

0.6

X/D=4

0.5

X/D=6

0.4

X/D=12
X/D=24

0.3

X/D=50

0.2
-3

-2

-1
0
1
2
LATERAL WAKE POSITION, Y/D

Fig. 4: Stationary cylinder wake axial velocity as a function of distance X downstream of center of cylinder. Uo = 1 m/s.
1.1

AXIAL VELOCITY, U/Uo .

1
0.9
0.8
0.7

axial position

0.6

X/D=2

0.5

X/D=4
X/D=6

0.4

X/D=12

0.3

X/D=24
X/D=50

0.2
-3

-2

-1
0
1
LATERAL WAKE POSITION, Y/D

Fig. 5: Vibrating cylinder wake axial velocity as a function of distance X downstream of center of cylinder. Uo = 1 m/s.
Figure 4 shows the axial (in line with free stream) velocity in the wake for the fixed 4 inch OD cylinder at 1.10 m/s. The figure
shows very nice looking classical wake profiles at 2, 4, 6, 12, 24, and 50 diameters downstream from the center of the
cylinder. Owing to the cylinder blockage the velocities of the flow just downstream of the cylinder outside the wake are about
9% higher than the corresponding velocities in the far wake.
Figure 5 shows the same wake velocity measurement but with the cylinder vibrating transversely under natural vortex-induced
vibration with an amplitude of 1.25 diameters. The cylinders elastic suspension is discussed in [6]. The peak-to-peak cylinder
displacement is 2.5 diameters (10 inch) and its vibration frequency is 1.76 Hz. The wake is wider and flatter with vortexinduced vibration and the drag coefficient of the cylinder, is approximately twice higher than the one of the fixed cylinder.

Analysis of data for A Vibrating cylinder

Analysis of data for a vibrating cylinder, using all data from 2D to 50D
When VIV is present the velocity profile is described by the values of CD, a1, a2, and from Eq. 6. The data of Fig. 5 have
been put in a non-dimensional form by using g () and as parameters in Fig. 6. The values of these parameters are selected to
minimize the error between the measured and theoretical values:
The optimized parameters are CD=2.0; a1 = 1.0; a2 = 18; = 0.2 (=/a1)

OT
TC 22951

Fig.
F 6 Comparrison of the thheoretical a-dim
mensional waake profile witth the measureements for a vibrating
v
cylinnder at Re = 1005
It is
i interesting to
t analyze thee contribution to the error off each wake profile.
p
The errrors contributted by each prrofile are givenn in
Fig
g. 7:

a Re = 105
Fig. 7 Contrribution to thee error of eachh wake profilee for a cylinder under VIV at
It can
c be observeed from Fig. 7 that there is a larger contrribution to thee total error fo
or the velocity profile at 2D downstream and
mo
ostly at the veelocity profilee at 50D dow
wnstream. In the
t range of 4D
4 to 24 D downstream
d
th
he contributioon to the errorr of
pro
ofiles is relatiively low and
d constant. Thhe error contrributed by thee 2D downstreeam is howev
ver not very large
l
and can
n be
exp
pected as this is very close to the upstreaam cylinder.
Th
he real major ccontribution however
h
is proovided by the 50D downstrream velocity profile. This is
i what was expected, becaause
this is a transitiion profile beetween the neear field and the
t far field. In the far fieeld only the s1 solution rem
mains and thee s2
v
which is well observved here at 50D downstream
m.
sollution has to vanish,

OTC 22
2951

An
nalysis of datta for a vibra
ating cylinde
er, using onlly data from 2D to 24D
Beecause the 50D
D profile is no
ot following the
t same patteern, it is intereesting to analyze the data from
f
2D to 244D excluding the
50D
D profile.
he optimized parameters
p
aree then changedd as follows frrom
Th
CD=2.00; a1 = 1.0; a2 = 18; = 0.22 (=/a1)
To
CD=1.88; a1 = 0.96; a2
a = 14; = 0..16 (=/a1)

mparison of thee theoretical a-dimensional

a
l wake profilee with the meaasurements foor a vibrating cylinder at Re
R =
Fig. 8 Com
105 (data for X/D=50 om
mitted)
he profile of errrors is shownn below in Figg.9:
Th

Fig. 9 Contrribution to thee error of eachh wake profilee for a vibratinng cylinder at Re = 105 (data for X/D=50 omitted)

OTC 22951

It can be concluded that the upper limit of the range of validity of the use of a combination of s1 and s2 to account for VIV is
limited to somewhere between X/D = 30 and X/D = 50. The best accuracy is achieved between X/D = 4 and X/D = 30.

Conclusion,

The wake of a stationary cylinder at Reynolds number 105 conforms well to the classical bell-shape model, but the near wake,
up to about 24 diameters downstream of a vibrating cylinder under going vortex induced vibration has a different shape than
the stationary cylinder - it is wider and flatter. Visualization suggests this is due to a wide, unstable, vortex street spacing in the
near wake.
The self-similarity wake model can be extended by use of a second solution to model the near wake of a cylinder during large
amplitude vortex induced vibration.
Optimized parameters for these models were developed from measurement of the velocity in the wake at 2, 4, 6, 12, 24, and 50
diameters down stream of stationary cylinder and a cylinder undergoing vortex-induced vibration at a Reynolds number of 100
000. This allows the use of the model for practical applications in the field of offshore and deep offshore developments.
The further benefit of this engineering approach is that it can be easily implemented in commercially available software for
conceptual and detailed engineering.

Acknowledgments
The authors acknowledge Subsea 7 for allowing them to perform this study. The contribution of Charles Coughran and Kyle
Lazzarevich for obtaining high quality data is gratefully acknowledged.
The authors would like to thank Subsea 7 for permission to publish this paper. However the paper only reflects the opinion of
the authors and does not imply endorsement by the Company.

References
[1] DNV-RP-203 Riser Interference, April 2009
[2a] Huse, E., Interaction in Deepsea Arrays, OTC 7237, Houston, TX, May 1993
[2b]Huse, E. Experimental Investigation of Deep Sea Riser Interaction, OTC 8070, Houston, TX, May 1996
[3] Blevins, R., D., Forces and Stability of a Cylinder in a wake, Journal of Offshore Mechanics and Arctic Engineering,
February 2005, Vol 127, pp.39-45
[4] Blevins, R., D., Saint-Marcoux, J-F., Wake of a vibrating Cylinder at Re=105, OMAE 2011-49075, Rotterdam,
Netherlands, June 2011
[5] Price, S., J., Paidoussis, M., P., The Aerodynamic Forces Acting on a Group of Two or Three Circular Cylinders When
Subjected to a Cross-Flow, J. Wind Eng. Ind. 1976, Aerodyn., 17, pp. 329-347
[6] Blevins, R., D., Saint-Marcoux, J-F., Wu, M., Experimental Investigation of Two Dimensional Motion of an Elastically
Supported Cylinder in a Wake, Journal of Offshore Mechanics and Arctic Engineering, November 2008, Vol.
130/044502-1
[7] Fontaine, E., Capul, J., Rippol, T., Lespinasse, P., (2007) Experimental and Numerical Study of Wake Interference and
Clashing between Steel Catenary Risers, ASME 2007 26th International Conference on Offshore Mechanics and Arctic
Engineering (OMAE2007) June 1015, 2007, San Diego, California, USA
[8] Schlichting, H., (1979) Boundary-Layer Theory, McGraw-Hill, 7th Ed., New-York, NY
[9] Schlichting, H., Gersten, K., (2000) Boundary-Layer Theory, Springer, 8th Ed., Berlin, Germany

10

OTC 22951

Appendix A: Formula for Drag and Lift

The formula for drag is a direct result of Eq.6.
1

C D Du 1
' 2
C Dd ( X , Y ) C Dd 0 1 a1 ( u 0 ) 2 e 4 (1 e 4 d )

X
a1 0

The formula for lift results from the Rawlins postulate (Blevins 2005 and Price 1976).
C Dd
CLd ( X , Y ) Dd
Y
2

1 2
1 a ( C Du 0 Du ) 2 e 4 (1 ' e 4 d )
1

X
a1 0

2
1

C Du 0 Du 12 4
' 2

) e (1 e 4 d )
CLd ( X , Y ) 2C Dd 0 Dd 1 a1 (

X
a1 0
Y

2
1

C D Du 1
' 2
CLd ( X , Y ) 2C Dd 0 Dd 1 a1 ( u 0 ) 2 e 4 (1 e 4 d )

X
a1 0
Y

CLd ( X , Y ) C Dd 0 Dd
Y

1 2
1 a ( C Du 0 Du ) 2 e 4 (1 ' e 4 d )
1

X
a1 0

1 2
a ( C Du 0 Du ) 2 e 4 (1 ' e 4 d )

1
X
a1 0

1
1

C Du 0 Du
C D Du 1
' 2 d 4
' 2
1 a1 ( u 0 ) 2 e 4 (1 e 4 d ) ( )
e (1 e 4 d )

X
X
a1 0
a1 0
y d

2
1
1
2
C Du 0 Du
C D Du 1
4a2
' 2 d 4
' 2
C Ld ( X , Y ) 2C D d 0 Dd a1
1 a1 ( u 0 ) 2 e 4 (1 e 4 d )
e (1 e 4 d )

X
C Du 0 Du x
X
a1 0
a1 0
d

2
1
1
2
C Du 0 Du 12 4
1
' 4 2 d 4
' 2
CLd ( X , Y ) 4C D d 0 Dd a1 a2
1 a1 (
) e (1 e d )
e (1 e 4 d )

X
X
a1 0
a1 0

d
2

C Ld ( X , Y ) 2C D d 0 Dd a1

C L d ( X , Y ) 4a1 a 2 C Dd 0

Dd
X

C D 1
1 2

1 2
1 a ( Du 0 u ) 2 e 4 (1 ' e 4 d ) ( ' e 4 (1 ' e 4 d )
1

X
a1 0
a1 0

a1 2

a1=coefficient of longitudinal decay of the velocity field

a2=coefficient of de transverse decay of the velocity field
a3=coefficient of lift
CDd=drag force coefficient on the downstream cylinder
CDd0=drag force coefficient on the downstream cylinder if it were outside of the wake
CDu0=drag force coefficient on the upstream cylinder in free stream
CLd= lift force coefficient on the downstream cylinder
Dd = diameter of the downstream cylinder (m)
Du = diameter of the upstream cylinder (m)
X = downstream distance (m) from the steady location of the upstream cylinder; identical to L in [3]
Y = Transverse distance (m) from the steady location of the upstream cylinder; identical to T in [3]
= Lift Proportionality constant in the Rawlins postulate
Once the value of has been defined from the wake measurement, the above equations allow predicting the drag and lift
forces acting on the downstream cylinder.