Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Home Search Collections Journals About Contact us My IOPscience

Numerical study of two side-by-side cylinders with unequal diameters at low Reynolds number

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2012 IOP Conf. Ser.: Earth Environ. Sci. 15 062041

(http://iopscience.iop.org/1755-1315/15/6/062041)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 155.69.2.13
The article was downloaded on 29/11/2012 at 05:46

Please note that terms and conditions apply.


26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

Numerical study of two side-by-side cylinders with unequal


diameters at low Reynolds number

Y Y Gao, X K Wang and S K Tan


Maritime Research Centre and Nanyang Environment and Water Research Institute
School of Civil and Environmental Engineering, College of Engineering, Nanyang
Technological University, Singapore

Email: YYGao@ntu.edu.sg

Abstract. Two-dimensional laminar flow about two side-by-side unequal cylinders with
different diameter ratios d/D and centre-to-centre spacing ratios T/D at Re=300 (based on the
larger cylinder diameter) was simulated using a CFD software. Comparisons of experimental
and numerical results were made to elucidate the degree of interference due to d/D and T/D and
their effects on the flow patterns and vortex shedding frequencies. The findings showed that
the flow patterns behind two unequal cylinders were distinctly different from that behind two
equal side-by-side cylinders, with distinct in-phase and anti-phase vortex shedding, and
random switching of modes of vortex shedding.

1. Introduction
Many experiments have been carried out to elucidate the wake characteristics behind a pair of side-by-
side circular cylinders with equal diameter [1]-[6]. Zdravkovich [7, 8] reviewed the phenomenon of flow
interference when two cylinders are arranged in tandem, side-by-side or staggered configurations in a
steady current. He found that when more than one bluff body is placed in a fluid flow, the resulting
forces and vortex shedding patterns may be completely different from those found on a single body at
the same Reynolds number. The spacing between the two bodies affects the flow patterns significantly.
Numerical simulations on the flow about two side-by-side cylinders of equal diameters have also
been carried out. Meneghini et al. [9] performed a two-dimensional finite element study on the flow past
two tandem and side-by-side cylinders at a Reynolds number of 200. Chen et al. [10] investigated the
turbulent wake flows behind two side-by-side cylinders using the large-eddy-simulations (LES) model.
They predicted the wake dynamics at Re=750 at intermediate and large ratios T/d=1.7 and 3.0 (T was
the cylinders’ centre-to-centre spacing, d was the cylinder diameter) and found that the results were in
accordance with experimental observations. Agrawal et al. [11] used the lattice Boltzmann method
(LBM) to investigate the flow around two square cylinders in side-by-side arrangement at low
Reynolds number. They found that vortex shedding occurs either in-phase or in anti-phase in the
synchronized regime, whereby in-phase locking is the predominant mode using the linear stochastic
estimate (LSE) calculations. Mizushima and Ino [12] investigated the stability and transition of flow
past a pair of circular cylinders in a side-by-side arrangement by numerical simulations and linear
stability analyses. They found that the deflected oscillatory flow arises from the steady symmetric
flow through sequential instabilities due to stationary and oscillatory unstable modes.
However, relatively little attention has been paid to the configuration of two circular cylinders with
unequal diameters. Palmer and Keffer [13] measured an asymmetric turbulent wake generated by two

Published under licence by IOP Publishing Ltd 1


26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

side-by-side cylinders of unequal diameters at different downstream distances from the cylinders. The
result showed that the wake width grew asymmetrically in the downstream direction, with the spread
rate and entrainment coefficients being larger on the smaller diameter cylinder. Lam and To [14]
investigated flow-induced vibration of a flexible circular cylinder located in the vicinity of a larger
cylinder and subjected to cross-flow. These authors found that the response was different from that of
two equal cylinders. Alam and Zhou [15] considered the diameter ratio effects on the Strouhal numbers,
forces and flow structures in the wake of two tandem cylinders of different diameters. They found that
with decreasing d/D (d was the upstream cylinder diameter and D was the downstream cylinder
diameter), the time-averaged drag on the downstream cylinder increases and the width of the wake
between the cylinders appears to narrow. Gao et al. [16] investigated the flow patterns behind two side-
by-side circular cylinders with unequal diameters using particle image velocimetry (PIV) technology,
they studied the bistable flow at small spacing ratios and provided explanation for the generation
mechanism from the velocity distribution. For the numerical study, Zhao et al. [17] simulated viscous
flow past two circular cylinders of different diameters using the finite element method, and found that
there exist, at medium gap ratios, strong interactions between the vortices from the cylinders.
This paper presented the 2D numerical simulations study on the flow around two unequal cylinders
in side-by-side configuration in laminar flow regime using a finite-volume method. The effects of
spacing and diameter ratios on the flow patterns and vortex shedding frequencies were investigated.
Of special interest was the changeover phenomenon of vortex shedding from that of in-phase mode to
anti-phase mode.

2. Numerical simulation
All numerical simulations were performed with a finite volume method using CFD software. The
governing equations are the incompressible unsteady Navier-Stokes equations and the continuity
equation, which can be written as equations 1-3.
u v
 0 (1)
x y

u u u p   2u  2u 
u v    v 2  2  (2)
t x y x  x y 

v v v p   2v  2v 
 u  v    v 2  2  (3)
t x y y  x y 

where u , v are the velocity vectors in x, y directions, respectively in the Cartesian coordinate system
(x, y) for the 2D simulation, (x, y) denote the coordinates along the stream direction, the transverse
direction, respectively. p is pressure,  is the fluid kinematic viscosity. A finite-volume method, based
on the control volume technique, is used to obtain a solution of the general integral conservation from
of the Navier-Stokes equation. The SIMPLE technique is used to resolve the coupling between the
pressure and the velocity fields. Discretization of the convection terms in the conservation equations is
accomplished through a second-order accurate upwind differencing scheme, and the diffusion term is
discretized using a central-difference technique, which is second-order accurate and is sufficient in the
present study.
The mesh for a centre-to-centre distance of T/D =2.4(where T is the center-to-center distance, D is
the larger cylinder diameter) in a rectangular flow field of 30D  60D is shown in Figure 1. The total
number of nodal point was 28,500 and number of elements was 56,600. The two cylinders were
located at mid-height of the computational domain; the entrance, top and bottom boundaries were 15D
from the cylinder center whereas the outflow boundary was located 45D downstream.
A total of 160 nodal points were placed on the circumference of the cylinder. The structured four-
node quadrilateral elements were used near the cylinder surface. The minimum mesh size in the radial

2
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

direction near the cylinder was about 0.01D. The non-dimensional time step of 0.0025 was used.
Symmetry conditions were applied at the top and bottom boundaries. Free outflow boundary condition
was imposed at the exit (boundary on the right).

(a) a overall view of the mesh (b) local mesh near the cylinder surface
Figure 1. Flow past two unequal cylinders for d/D=2/3 in side-by-side configuration
In order to validate the accuracy of the numerical model, simulations of uniform flow past a single
cylinder at the Reynolds number Re=200 were repeated based on three meshes. Details of the meshes
are given in Table 1.
Table 1. The effect of mesh refinement on the calculation results of the flow around a single
cylinder at Re=200.
Number of circumferential Number of nodes in the
Cd St
nodes mesh
120 13770 1.345 0.207
160 18110 1.360 0.194
200 21000 1.401 0.190
Table 2. Comparisons of mean drag coefficients and Strouhal numbers for a single cylinder at
Re=200 with literature data.

St Cd
Current simulation result 0.194 1.36
Braza et al. [18] 0.2 1.35
Sa & Chang [19] 0.186 1.13
Meneghini [20] 0.196 1.25
Arkell [21] 0.196 1.30
Saltara [22] 0.190 1.25
Meneghini et al. [9] 0.196 1.30
Zhao et al. [17] 0.196 1.36
The largest deviation of the average drag coefficient is about 4%, which may be deemed as of
adequate accuracy for the present study. Comparisons of the average drag coefficients and Strouhal
numbers for a single cylinder at Re=200 with published data from the literature are shown in Table 2.
From tables 1 and 2 it can be seen that the values of the averaged drag coefficient and Strouhal

3
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

number (based on 160 circumferential nodes) are 1.360 and 0.194, which are very close to those
obtained by other researchers. All subsequent simulations presented in this paper were based on 160
circumferential nodes.

3. Discussion and results

3.1. Flow patterns


Figure 2 shows different flow patterns behind two side-by-side unequal cylinders with different
diameter ratios d/D at a fixed spacing ratio of T/D=1.2. Figures 2(a), (b) represent the PIV
measurements at T/D=1.2 and Re=1200, while figures 2(c)-(h) represent the numerical results for
different diameter ratios d/D from 1/3 to 1. At small diameter ratio, say when d/D=1/3, the
interference effect between the upper and lower cylinder is weak. Two vortex streets with different
scales can be seen behind the two unequal cylinders. As seen from figures 2(c) and (d), vortex
shedding occurs from both cylinders, with smaller-scale vortices from the upper and smaller cylinder,
and larger-scale vortices from the lower and larger cylinder. Downstream of the wake, the vortices
amalgamate into the larger-scale vortex.
When the diameter ratio d/D increases to 2/3, it can be seen the phenomenon of changeover of the
gap flow from the numerical simulation (figures 2(e)-(f)), which is similar to that observed in the
experiment (figures 2(a)-(b)). Detailed discussion of the changeover of the gap flow can be found in
Gao et al. [16].
At large diameter ratio d/D=1, a single vortex street can be observed, see figures 2(g) and (f), as in
the combined wake of the two cylinders. A cycle before, as seen from figure 2(g), the negative inside
shear layer (blue color) from the lower cylinder has amalgamated into the negative outer shear layer
from the upper cylinder, while the positive inner shear layer (red color) from the upper cylinder has
shrunk and disappeared in the combined wake. The opposite phenomena can be seen from figure 2(h).
In figures 2(g) and (h), vortex shedding can only be observed from the outer shear layers.

(a) (b)

(c) (d)

(e) (f)

4
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

(g) (h)

Figure 2. Changeover of the gap flow for two side-by-side cylinders with unequal diameters, T/D=1.2
(a), (b) PIV results at Re=1200; (c)-(f) numerical results at Re=300.

t=464.2s t=464.6s

t=465s t=465.4s

t=465.8s t=466.2s

Figure 3. Changeover of in-phase vortex shedding to anti-phase vortex shedding for two side-by-side
cylinders with unequal diameters at Re=300, T/D=2.4.
Figure 3 shows the evolution of vortex shedding from in-phase shedding to anti-phase shedding for
the case of d/D=2/3 at Re=300 and T/D=2.4. Both in-phase vortex shedding and anti-phase vortex
shedding can be observed at this spacing ratio. In figure 3, symbols A1-D1 represent the vortices shed
from the top smaller cylinder and A2-D2 represent those from the larger cylinder. As time progresses,
vortices A1 and A2 begin to amalgamate, and A1 gradually diminishes, finally coalesces into a large
scale vortex. At t=464.6s, new negative vortex (E1) is shed from the smaller cylinder, while the one

5
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

from the larger cylinder has not developed fully; no new negative vortex is observable from the larger
cylinder till t=465.4s (E2). At the same time, a new positive vortex (F1) is shed from the smaller
cylinder, and only until t=466.2s, a new corresponding positive vortex (F2) is shed from the larger
cylinder. Meanwhile, a new negative vortex (G1) is shed from the smaller cylinder. As a result, it can
be concluded that there may be two possible reasons for the alternate switch of the in-phase shedding
mode and anti-phase shedding mode: i) there is a phase deviation from the in-phase vortex shedding
between the smaller cylinder and larger cylinder; ii) the different scales of the Karman vortex behind
the two cylinders, where a larger Karman vortex formed behind the larger cylinder and a smaller one
behind the smaller cylinder. However, it is difficult to determine which mode is more “stable”. We
may conjecture that the switch of two modes is a random progress.

Figure 4. Schematic development of a binary-vortex street due to in-


phase vortex shedding behind two side-by-side cylinders of equal
diameter at Re=100, T/D=4.0 ([23]).

(a)

(b)Non-dimensional length x/d


y

(c)Non-dimensional length x/D

6
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

Figure 5. Evolution of the in-phase vortex shedding behind two side-


by-side cylinders at Re=300, T/D=3.6, (a) PIV results in a water
channel (Re=1200); (b) numerical results (Re=300); (c) schematic
development.
Williamson [23] defined a binary vortex as a pair of same-signed vortices which rotate around one
another, and is distinguishable from a normal vortex pair which consists of two counter-rotating
vortices, as shown in Figure 4. He also suggested four regions (A, B, C and D) for in-phase vortex
shedding. In region A, close to the cylinders, there are two parallel in-phase streets; region B is a
transition region between region A and C which consists of combined binary street; further
downstream is region D where the binary vortices coalesce and form a large-scale Karman street
represented by region D, see figure 4.
Figure 5 shows the characteristics of vortex streets of our experiments and numerical simulation,
which is distinctly different from that in figure 4. Note that the flow in figure 5 is from right to left, to
match the scheme adapted from Williamson [23]. As shown in figures 5(a) and (b), due to the larger
scale vortex shedding from the larger cylinder, there is a mis-match between the vortex shed from
small and large cylinder. In order to gain an insight of the flow characteristics behind the two unequal
diameters, we adopt the non-dimensional length to avoid the effect of the vortex scale, using x/D for
the larger cylinder and x/d for the smaller cylinder. The schematic development of the in-phase vortex
shedding is shown in figure 5(c). Clearly, the characteristics are different from the case of equal
diameter, with no clear binary-vortex street. Only two regions, namely A and B, can be seen behind
the two cylinders. In region A, there are two parallel in-phase vorticies, which is similar to the one
behind that of equal cylinders. In region B, two vortex streets can be observed behind the two
cylinders, respectively.

Figure 6. (a) In-phase vortex shedding; (b) Anti-phase vortex shedding at Re=300, T/D=5.0.
At large centre-to-centre spacing ratio T/D=5.0, as seen from Figure 6, the two cylinders with
d/D=2/3 behave as isolated bluff bodies. It can also be observed that anti-phase vortex shedding from
the cylinders changes to in-phase vortex shedding alternately.

3.2. Force coefficients


Special attention is paid on the effect of the centre-to-centre spacing ratio T/D in terms of the force
coefficient. Figure 7 shows the time histories of force coefficients Cl and Cd on the two side-by-side
cylinders with d/D=2/3 at different spacing ratios T/D=1.2-5.0. Different from the cases of equal
diameter observed by Meneghini et al. [20], no negative drag coefficient on the lower cylinder is
observed at small T/D. As shown in figure 7, at T/D=1.2, the drag coefficients for both cylinders are

7
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

irregular. With increasing of T/D, the force coefficients gradually become more regular, plausibly due
to diminishing interactions between the cylinders with increasing T/D. Additionally, it can also be seen
from figure 7 that the drag coefficient Cd on the upper and smaller cylinder is larger than that of the
lower and larger cylinder at small T/D, while at larger T/D, Cd on the small cylinder is smaller than
that of the larger cylinder. Furthermore, different from the cases of equal diameter cylinders, the lift
coefficients on the two unequal cylinders are not in the evident anti-phase or in-phase mode, instead it
appears that there is a phase deviation between them.

8
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

Figure 7. Force coefficients time history for two side-by-side cylinders with d/D=2/3 at different T/D,
at Re=300.
In order to investigate the effect of diameter ratio d/D on the flow behaviors behind two unequal
cylinders, the drag coefficients on the two cylinders with three different diameter ratios d/D=1/3, 2/3
and 1 are considered. Figure 8 shows the variations of mean drag coefficient Cd on the two side-by-
side unequal cylinders with different d / D and T / D , where figure 8(a) shows Cd 1 on the upper and
smaller cylinder, and figure 8(b) shows Cd 2 on the lower and larger cylinder.

Figure 8. Variation of drag coefficient Cd of the two


side-by-side unequal cylinders with d / D and T / D (a)
smaller cylinder; (b) larger cylinder.
It can be seen from figure 8 that both Cd 1 and Cd 2 decrease with increasing T/D from 1.2 to 5.0 for
the three diameter ratios. At the smallest spacing ratio T / D  1.2 , Cd 1 behaves different from those at
larger spacing ratios, which decreases from 2.02 at d/D=1/3 to 1.82 at d/D=2/3, but increases to 2.25 at
d/D=1. At d/D=1/3, Cd 2 is 2.02, which is much higher than Cd 1 , which is 1.59, and the difference
between Cd 1 and Cd 2 decreases with increasing d/D. At larger spacing ratios T / D  2.4 , both Cd 1
and Cd 2 decrease as d / D increases.

9
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

Figure 9. Variation of root mean square


coefficient Cdrms of the two side-by-side unequal
cylinders with d / D and T / D , (a) smaller cylinder; (b)
larger cylinder.
Figures 9(a) and (b) show the root mean square drag coefficients Cdrms on the two cylinders as a
function of T/D, for the three diameter ratios considered. The values of Cdrms for both cylinders
increase with increasing d / D . The r.m.s. drag coefficient Cdrms 2 on the larger cylinder decreases for
small values of T/D and increases for large values of T/D with small d/D.

3.3. Strouhal number


The Strouhal numbers on two side-by-side unequal cylinders are studied. The definitions of the
Strouhal number for the small and large cylinders are based on their respective diameters. Figure 10
shows the variation of Strouhal number with respect to centre-to-centre spacing ratios at different
diameter ratios, associated with the larger (figure 10(a)) and smaller cylinders (figure 10(b)),
respectively.
It can be seen that at smallest spacing ratio T / D  1.2 , the dimensionless vortex shedding
frequency, StD  fD / U of vortex shedding from the lower and larger cylinder decreases progressively
from 0.17 at d/D=1/3 to 0.12 at d/D=1. On the other hand, for larger spacing ratio T / D , the
dimensionless vortex shedding frequency St D from the lower and larger cylinder increases with
d / D and T / D , and eventually the two cylinders behave as two independent bodies and the vortex
shedding frequency reaches to StD  0.22 , close to that from a single cylinder StD  0.21 .
For the smaller and upper cylinder, it can be seen from figure 10(b) that the dimensionless vortex
shedding frequency Std  fd / U increases with d / D , while decreases with T / D . At small spacing
ratio T / D  1.2 , the Strouhal number is larger than that from a single cylinder. With increasing T / D ,
the interactions between the shedding processes from the two cylinders become weaker.

10
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

(a)behind the larger cylinder (b) behind the smaller cylinder


Figure 10. Variation of Strouhal numbers with T/D and d/DConclusions
Two-dimensional laminar flow about two side-by-side cylinders with unequal diameters was
simulated at a low Reynolds number Re=300 (based on the larger cylinder diameter). The findings of
this study showed that at small center-to-center spacing ratio T / D  1.2 , changeover phenomenon of
the wide wake and narrow wake took place at medium diameter ratio d/D=2/3, as was reported in
experimental studies by others.
At larger spacing ratios T / D  1.2 , both in-phase and anti-phase vortex shedding mode were
observed. The observed alternate switching between the two modes were the results of: i) phase
deviation from the in-phase vortex shedding between the two cylinders; ii) different scales of the
Karman vortex behind the two cylinders.
The findings also indicate that the diameter ratio d/D and spacing ratio T/D produced significant
effects on the flow patterns and vortex shedding frequency.

Acknowledgments
Financial support by DHI-NTU Centre, Maritime Research Centre and Nanyang Environment and
Water Research Institute, Nanyang Technological University (Singapore) are gratefully acknowledged.

References
[1] Kumada M, Hiwada M, Ito M and Mabuchi I 1984 Transactions of the JSME 50 p 1699
[2] Sumner D, Wong SST, Pricem SJ and Paisoussis MP 1999 Phys. Fluids 15 p 1214
[3] Akilli H, Akar A and Karakus C 2004 Flow Meas Instrum. 15 p 187
[4] Zhou Y, So R M C, Liu M and Zhang H J 2000 Int. J. Heat Fluid Flow 21 p 125
[5] Zhou Y, Wang Z J, So R M C, Xu S J and Jin W 2001 J. Fluid Mech. 443 p 197
[6] Sumner D 2010 J.Fluids Struct. 26 p 849
[7] Zdravkovich M M 1977 J. Fluids Eng. 99 p 618
[8] Zdravkovich M M 1987 J.Fluids Struct. 1 p 239
[9] Meneghini J R, Saltara F, Siqueira C L R and Ferrari J R 2001 J.Fluids Struct. 15 p 327
[10] Chen L, Tu J K and Yeoh G H 2003 J.Fluids Struct. 18 p 387
[11] Agrawal A, Djenidi L and Antonia RA 2006 Comput Fluids. 35 p 1093
[12] Mizushima J and Ino Y 2008 J.Fluid Mech. 595 p 491
[13] Palmer M D and Keffer J F 1972 J.Fluid Mech. 53 p 593
[14] Lam K M and To A P 2003 J.Fluids Struct.17 p 1059
[15] Alam M and Zhou Y 2008 J.Fluids Struct. 24 p 505
[16] Gao Y Y, Yu D Y, Tan S K, Wang X K and Hao Z Y 2010 Fluid Dyn.Res. 42
(doi:10.1088/0169-5983/42 /5/055509).
[17] Zhao M, Cheng L, Teng B and Liang DF 2005 Appl. Ocean Res. 27 p 39

11
26th IAHR Symposium on Hydraulic Machinery and Systems IOP Publishing
IOP Conf. Series: Earth and Environmental Science 15 (2012) 062041 doi:10.1088/1755-1315/15/6/062041

[18] Braza M, Chassaing P and Ha Minh H 1986 J.Fluid Mech. 165 p 79


[19] Sa J Y, Chang K S 1991 Int. J. Numer Meth. Fl. 12 p 463
[20] Meneghini J R 1993 Numerical simulation of bluff body flow control using a discrete vortex
method Ph.D thesis (UK: University of London)
[21] Arkell R H 1995 Wake dynamics of cylinders encountering free surface gravity waves Ph.D
thesis (UK: University of London)
[22] Saltrara F 1999 Numerical simulation of the flow about circular cylinders Ph.D thesis (Brazil:
EPUSP University of Sao Paulo)
[23] Williamson CHK 1985 J.Fluid Mech. 159 1-18

12

You might also like