Ocean Engineering 94 (2015) 36–50

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Influence of wall proximity on flow around two tandem

circular cylinders
X.K. Wang a,n, J.-X. Zhang b, Z. Hao c, B. Zhou a, S.K. Tan a
a
Maritime Research Centre, Nanyang Technological University, 639798, Singapore
b
School of Naval Architecture, Ocean and Civil Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
c
College of Logistics Engineering, Shanghai Maritime University, Shanghai 201306, China

art ic l e i nf o a b s t r a c t

Article history: An experimental study was conducted to investigate the flow around two tandem cylinders placed near
Received 3 August 2013 and parallel to a plane wall. The Reynolds number based on the cylinder diameter (D) was 6300. The
Accepted 30 November 2014 cylinder centre-to-centre spacing ratio (Ln ¼L=D) was varied from 1.5 to 6, and the gap-height-to-
Available online 16 December 2014
cylinder-diameter ratio (Gn ¼G=D) from 0.15 to 2. The flow fields were measured using Particle Image
Keywords: Velocimetry (PIV), in conjunction with measurements of fluid dynamic forces (drag and lift) on the
Tandem cylinders downstream cylinder using load cell. The flow strongly depends on the combined value of Gn and Ln .
Wall proximity With reference to Gn , the flow could be classified as vortex-shedding suppression regime (Gn o 0.3),
Vortex shedding intermediate-gap regime (0.3o Gn o1) where vortex shedding occurs but is influenced by wall
Drag and lift coefficients
proximity, and large-gap regime (Gn 4 1) where the wall influence becomes negligible. Similarly, three
Spectrum of lift coefficient
categories can be identified as a function of Ln , namely, extended-body regime 1 o Ln o2, reattachment
Particle image velocimetry (PIV)
regime at 2 oLn o4, and impinging regime at Ln 44. Variations of dynamic drag and lift coefficients,
spectra, Strouhal numbers, and Reynolds shear stress are also presented to characterize the different
flow regimes in the Gn –Ln plane.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction sub-divided into two distinct categories, for which the reattachment
is on the rear and leading surfaces of the downstream cylinder,
The interference of flow around two circular cylinders is of both respectively (see Fig. 1). The exact values of Ln to delineate the
academic interest and practical importance, see Sumner (2010) for boundaries between different regimes depend on the value of
a comprehensive review. Among the many possible arrangements Reynolds number (Carmo et al., 2010) and free-stream turbulence
of the two cylinders to be positioned in relative to the flow direction, intensity (Ljungkrona et al., 1991). The critical spacing ratio (Ln cr ), at
the tandem configuration has been extensively studied. This type of which periodic vortex shedding begins to occur from the upstream
interference, referred to as ‘wake interference’ by Zdrvkovich (1987), cylinder, varies from Ln cr ¼ 3 to 5 in the literature (e.g., Lee et al.,
is a function of the inter-cylinder distance (expressed as the ratio 2009). Correspondingly, the fluid forces on the cylinders would
between the centre-to-centre spacing and the cylinder diameter, experience a discontinuous ‘jump’ at about Ln cr (Zdravkovich and
Ln ¼L=D, thereafter abbreviated as the spacing ratio). Zdravkovich Pridden, 1977). Moreover, Xu and Zhou (2004) showed that the
(1987) proposed that the flow can be classified into three basic types: vortex shedding frequency is dependent on Reynolds number over
(i) single bluff-body regime at small Ln (1oLn o1.2 1.8), where the range Re¼800–4.2  104 (Re¼ UD=ν, where ν is the kinematic
periodic von Kármán vortex shedding is observed only in the wake of viscosity of fluid).
the downstream cylinder; (ii) reattachment regime at moderate Ln On the other hand, there are a number of engineering practices
(1.2 1.8oLn o3.4 3.8), where the shear layers emanating from the in which cylindrical structures are placed near a plane wall, such as
upstream cylinder reattach onto the surface of the downstream submarine pipelines, risers and cables on seabed. To date, many
cylinder; (iii) impinging regime at large Ln (Ln 43.4 3.8), where von researchers have examined the influence of wall proximity on a
Kármán vortices are shed from the upstream cylinder and periodi- single cylinder with the cross-section of either circular (e.g.,
cally impinge on the downstream cylinder. Zhou and Yiu (2006) Bearman and Zdravkovich, 1978; Lei et al., 1999; Price et al., 2002;
showed that the reattachment regime (2oLn o5) can be further Dipankar and Sengupta, 2005; Nishino et al., 2007; Wang and Tan,
2008a; Lin et al., 2009; Sarkar and Sarkar, 2010; Ong et al., 2012;
Wang et al., 2013), square (e.g., Wang and Tan 2008b; Mahir, 2009)
n
Corresponding author. Tel.: þ 65 67906619; fax: þ65 67906620. or rectangular (e.g., Maiti, 2012; Maiti and Bhatt, 2014). The nearby
E-mail address: [email protected] (X.K. Wang). wall affects not only the dynamic pressure and forces on the

http://dx.doi.org/10.1016/j.oceaneng.2014.11.018
0029-8018/& 2014 Elsevier Ltd. All rights reserved.
 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50 37

cylinder, but also the wake pattern and flow-induced vibrations. of the test section were made of glass to allow for optical access.
The ratio between the gap height to the cylinder diameter The free-stream velocity was uniform to within 1.5% across the test
(Gn ¼ G=D, abbreviated hereafter as the gap ratio) is found to be section, and the turbulence intensity in the free stream was
the predominant parameter. The impermeability of the wall poses below 2%.
an irrotational constraint to the wake development, resulting in Fig. 3 shows a sketch of the two tandem cylinders placed near
suppression of the classical von Kármán vortex shedding that is in and parallel to a plane wall. The cylinder models were made of
absolute instability (Huerre and Monkewitz, 1990) below a critical smooth, transparent acrylic rod with an outer diameter of D¼ 15
gap ratio (Gn cr ). As sketched in Fig. 2(a), when Gn oGn cr , the wake is mm. During the experiments, the free-stream velocity was kept
steady with a long recirculation region: while the gap flow keeps constant at U ¼0.42 m/s (Re¼ 6300). The approach boundary layer
attached on the wall, the upper shear layer emanating from the was fully developed with a thickness of δ ¼7 mm (  0.5D). The
cylinder exhibits as elongated Kelvin-Helmholtz (K-H) type of roll- cylinders’ centre-to-centre spacing was varied as L¼22.5, 30, 45,
ups (in convective instability). However, when Gn 4Gn cr (Fig. 2(b)), 60, 75, 90 and 105 mm (Ln ¼1.5–7), and the gap height G ¼2.25, 6,
the gap flow is strong enough to detach upward from the wall (or 9, 12, 21 and 30 mm (Gn ¼0.15–2). Therefore, totally 42 cases were
upwash), and to interact with the upper shear layer to form discrete considered in the present study.
vortices. It should be noted that the vortex shedding is asymmetric The span (b) of the cylinders was 200 mm, leading to an aspect
about the horizontal wake centerline; also, there is a coupling ratio (AR) of b=D¼13.3. This value was considered to be large
between the lower shear layer and the wall boundary layer, as enough (AR Z10 according to previous finding, for example, Lam
reflected by the phenomenon that each anticlockwise vortex is and Zou, 2010) to ensure a nominally 2D flow in the near wake.
accompanied by a small clockwise vortex in the near wall region. Therefore, the velocity measurements with Particle Image Veloci-
The value of Gn cr E0.3 slightly varies with Re and thickness of the metry (PIV) were performed in the mid-span plane. The origin of
wall boundary layer (e.g., Buresti and Lanciotti, 1992; Price et al., the coordinate system was located at the center of the upstream
2002). cylinder, with x, y and z denoting the streamwise, transverse and
However, little attention has been paid to the configuration of spanwise directions, respectively. The positive drag and lift forces
two tandem cylinders in proximity to a plane wall (see Fig. 3). The are in the x- and y-directions, respectively.
flow interference between the two cylinders is further compli- Velocity measurements were performed using a digital PIV
cated due to the presence of the wall boundary. Bhattacharyya and system (LaVision model). The flow field was illuminated with a
Dhinakaran (2008) numerically studied the 2-dimensional (2D) double cavity Nd:YAG laser light sheet at 532 nm wavelength (Litron
flow around two tandem square cylinders with a linear incident model, power 135 mJ per pulse, duration 5 ns). Sphericels 110P8
velocity profile at Gn ¼0.5 and Ln ¼ 1.5–6. The non-uniform approach hollow glass spheres (neutrally buoyant with a mean diameter of 13
flow causes difference in the strength of the upper and lower shear μm) were seeded in the flow as tracer particles. The images were
layers. The flow can be steady up to Re¼125 depending on the value
of Ln . More recently, Harichandan and Roy (2012) simulated the flow
around two near-wall tandem cylinders (circular/square) at Re¼ 100
and 200, Gn ¼0.5 and 1, and Ln ¼2 and 5. For a given Re, the Strouhal
numbers of the two cylinders are identical, but the lift and drag
coefficients are different.
As described above, there is limited information available on the
flow around two tandem cylinders in proximity to a wall boundary.
Two aspects need attention. Firstly, the only two published studies,
namely, Dhinakaran (2008) and Harichandan and Roy (2012), were
conducted at relatively low Re (up to 200), that is, in the laminar
regime. Yet, in engineering practice the flow is generally in the
subcritical regime. Secondly, both studies considered only a rather
limited number of combinations of Gn and Ln , and hence a complete
picture in the Gn –Ln plane is still unavailable. These motivate the
present relatively systematic investigation for 0.15rGn r2 and
1.5rLn r7 under a constant Reynolds number in subcritical regime
(Re¼6300).

2. Experimental set-up and methodology

The experiments were performed in a re-circulating open

Fig. 2. Schematic of the flow around a near-wall single cylinder: (a) vortex-
channel located at Maritime Research Centre, Nanyang Technolo- shedding-suppression regime at small gap ratio; and (b) vortex-shedding regime
gical University, with a test section of 5 m  0.3 m  0.45 m at moderate gap ratio. Proposed based on the flow measurement results in Wang
(length  width  height). The channel bed and the two side walls and Tan (2008a).

Fig. 1. Schematic of the flow around two tandem cylinders as a function of spacing ratio (Ln ) under free-standing conditions. Modified from Zhou and Yiu (2006).
38 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50

Fig. 3. Schematic of the flow around two near-wall tandem cylinders.

recorded using a 12- bit CCD camera with a resolution of 3. Results and discussion
1600  1200 pixels. LaVision Davis software (Version 7.2) was used
to process the particle images and determine the velocity vectors. 3.1. Near-wall single cylinder
Particle displacement was calculated using the fast-Fourier-transform
(FFT) based cross-correlation algorithm with standard Gaussian sub- The effects of wall proximity on a single cylinder has been
pixel fit structured as an iterative multi-grid method. The processing examined in this section. Fig. 4 shows the variation of the dynamic
procedure included two passes, starting with a grid size of 64  64 force coefficients (C D , C L , C D' and C L' ) versus Gn , together with the
pixels, stepping down to 32  32 pixels overlapping by 50%, which data reported by Roshko et al. (1975) and Lei et al. (1999). As
resulted in a set of 7500 vectors (100  75) for a typical field. In shown in Fig. 4(a), a prominent feature is that the most dramatic
between the two passes, the vector maps were filtered by using a 3 change of C D occurs from small- to intermediate-gap ratios (e.g.,
 3 median filter in order to remove possible outliers. The number Gn o0.75); when Gn 41, by contrast, it remains approximately
of particles in a 32  32 pixel window was of the order of 1015 to constant at C D E1.1 (asymptotically approaching the value for an
yield strong correlations. The field of view was set at 190 mm  143 isolated cylinder), indicating that the wall effects become negli-
mm, therefore the spatial resolution was 1.9 mm  1.9 mm (i.e., gible. A similar trend is found for the RMS coefficients (C D' and C L' )
0.13D  0.13D). For each case, a series of 1050 instantaneous flow as shown in Fig. 4(c). However, it is noted that the asymptotic
fields was acquired at the sampling frequency of 15 Hz (or 70 s values are considerably lower than those reported in Roshko et al.
recordings), in order to achieve a reasonably statistical convergence (1975) and Lei et al. (1999), namely, C D E1.1 versus 1.3 and C L' E
of the measured quantities, such as Reynolds shear stress. The 0.075 versus 0.6. The discrepancy is likely attributed to the
uncertainty in the instantaneous velocities (u and v) was estimated difference in measurement techniques and oncoming flow condi-
to be about 3.5% for the present setup. The instantaneous spanwise tions (such as Re and δ). Note the data in Roshko et al. (1975) and
vorticity (ωz ¼ Δv=Δx  Δu=Δy) was calculated using the least Lei et al. (1999) were based on pressure distribution around the
squares extrapolation scheme. The uncertainty in ωz was estimated cylinder circumference for an elemental slice (referred to as sectional
to be about 10% based on the method proposed by Fouras and Soria force by Norberg (2003)), while the present study measured the total
(1998). force on the whole span of the cylinder, which always has a lower
A piezoelectric load cell (Kistler Model 9317B) was used to magnitude due to the end effects (West and Apelt, 1997). In the
directly measure the fluid dynamic forces on the downstream present study, the thickness of boundary layers developed on the
cylinder, drag (F D ) and lift (F L ). The output signal was captured side walls where the two ends of the cylinder were attached was
with a National Instruments A/D card at a sampling rate of 100 Hz about 0.5D, so the length of the cylinder subjected to end effects was
(at least 1 order of magnitude greater than the vortex shedding about 1D (or 7.5% of the total span). Therefore, the difference
frequency, which was about 5–6 Hz). The duration of recording for between the measured total force and the ideal sectional force would
each case was about 200 s, which corresponded to about 1000 be less than 10%. In fact, the present results are in good agreement
cycles of vortex shedding and was sufficiently long according to the with the published data on a single cylinder using similar measure-
criterion proposed by Sakamoto et al. (1987). The dimensionless ment technique (load cell), such as C D E1.186 and C L' E0.089 in Lam
shedding frequency was expressed as Strouhal number (St¼f D=U), et al. (2003) for Re¼ 4.8  104, and C L' E0.08 in Tadrist et al. (1990)
where f is the frequency determined from spectral analysis of the for Re¼ 7000. As depicted in Fig. 4(b), the cylinder experiences a
fluctuating lift coefficient using power spectral density (PSD) fun- positive mean lift (C L 40) at small- to intermediate-Gn , suggesting
ction. Also, the mean and root-mean-square (RMS) values of fluid that the cylinder is pushed upward from the wall. The mean lift
dynamic drag and lift coefficients (C D ¼ 2F D =ρU 2 Db and C L ¼2F L = coefficient has a maximum of C L E0.3 at the smallest gap ratio
ρU 2 Db) were calculated, where ρ is the fluid density. Through a (Gn ¼ 0.15), and thereafter decreases monotonically until reaching the
number of repeated measurements on a single cylinder, the uncer- asymptotic value of C L ¼0 at Gn 41.
tainty in the mean drag was determined to be within 1%. The data Fig. 5(a) shows the time histories of dynamic lift coefficient (C L )
for a free-standing (isolated) single cylinder measured at the same on the cylinder at different gap ratios. As Gn increases, the signal
Reynolds number (Re¼ 6300) served as the benchmark reference: changes from a chaotic pattern at Gn ¼0.15 and 0.25, to a periodic
C D0 ¼ 1.1, C D0' ¼ 0.055, C L0' ¼0.075 and St0 ¼0.2 (where the subscript pattern at Gn 4 0.4 with a much higher magnitude of fluctuation.
0 denotes the isolated cylinder). The chaotic pattern at small-Gn is due to the cessation of periodic
 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50 39

shedding becomes stronger within this range (a similar trend is

inferred from the velocity data in Wang and Tan (2008a)). At
Gn ¼0.15 or 0.25, on the other hand, a weak peak is discernible at a
relatively high frequency, that is, St E0.85, which corresponds to
the K-H roll-ups in shear layer instability. Rajagopalan and Antonia
(2005) proposed an empirical relationship between the shear layer
instability frequency (f sl ) and the vortex shedding frequency (f V )
as a function of Re for an isolated cylinder, namely, f sl =f V ¼0:029
Re0:65 . The predicted value using this equation for Re ¼6300 is
f sl =f V ¼8.55, which is about twice the measured value of 0.85/
0.2 ¼4.25. However it is noted that the working fluid in Rajagopalan
and Antonia (2005) is air (vs. water in the present study), which would
result in relatively thinner shear layers and hence higher f sl , since f sl is
inversely proportional to the shear layer thickness (Gerrard, 1967).
Fig. 5(b) also shows that at Gn ¼0.4 (corresponding to onset of periodic
vortex shedding), the spectrum exhibits co-existence of both peaks
(StE0.2 and 0.85), albeit rather weak, implying a transition/competi-
tion between the two types of instability (von Kármán vs. K-H).

3.2. Near-wall tandem cylinders

3.2.1. Instantaneous flow patterns around the cylinders

Fig. 6 shows a representative snapshot of the instantaneous
vorticity fields around the near-wall tandem cylinders at selected
gap ratios (Gn ¼0.15, 0.4, 0.6 and 1.4) and spacing ratios (Ln ¼2, 3 and
5). It is obvious that the flow patterns depend on both Gn and Ln .
Periodic vortex shedding from both cylinders is suppressed when Gn
is small. For Gn ¼0.15 (1st row), the upper shear layer emanating from
the upstream cylinder is K-H type of roll-ups (elongated in the
streamwise direction with negative vorticity), which pass over (or
overshoot) the downstream cylinder. However, the lower shear layer,
which is evident in the spacing between the cylinders, is rather weak
in magnitude and small in size. For Gn ¼0.4 (2nd row), both shear
layers are still largely in K-H type at small spacing ratios (e.g., Ln ¼ 2 or
3). At Ln ¼5, on the other hand, they begin to display as relatively large,
discrete ‘patches’ of vorticity behind the downstream cylinder, indi-
cative of the occurrence of vortex shedding. However, no vortex
shedding is observed from the upstream cylinder: while the upper
shear layer is still in K-H type, the lower shear layer either reattaches
steadily on the leading surface of the downstream cylinder at Ln ¼2
and 3, or dissipates around x=DE4 at Ln ¼ 5. In addition, flow-induced
separation is found in the near wall region, similar to the case of the
near-wall single cylinder shown in Fig. 2(b). When Gn increases to 0.6
(3rd row), the wall effects still exist, but to a lesser degree. At this gap
ratio, periodic vortex shedding is always observed from the down-
stream cylinder, as well as from the upstream cylinder at wide-spacing
ratios (e.g., Ln ¼5). At Gn ¼1.4 (4th row), the wall effects become almost
negligible such that the flow is similar to the free-standing case.
Obviously, the three cases, Ln ¼2, 3 and 5, belong to the ‘extended-
body’, ‘reattachment’ and ‘impinging’ regimes, respectively, as shown
in Fig. 1. The critical spacing ratio is Ln cr E4.5, which is also consistent
with the reported values of Ln cr ¼3–5 in the literature.
A closer examination of the instantaneous vorticity fields indi-
cates that the effects of wall proximity cannot simply be described
Fig. 4. Variation of dynamic force coefficients with Gn for the near-wall single as inhibiting vortex shedding from the cylinders; instead, it plays a
cylinder: (a) mean drag coefficient (C D ); (b) mean lift coefficient (C L ); and (c) RMS complex role in affecting the shear layer development and interac-
drag (C D' ) and lift (C L' ) coefficients. Present: Re¼ 6300 and δ ¼ 0.5D; Lei et al. (1999):
tion. Take the case of Ln ¼2 at different gap ratios (left column in
Re ¼1.36  104 and δ ¼ 0.14D; Roshko et al. (1975): Re¼ 2  104 and δ ¼ 0.5D.
Fig. 6) as an example. At Gn ¼ 1.4, the flow is in extended-body
vortex shedding from the cylinder (e.g., Price et al., 2002; Wang regime, and the two shear layers separated from the upstream
and Tan, 2008a). Also, the critical gap ratio (Gn cr ) is between 0.25 cylinder are kept nearly horizontally and wrap around the down-
and 0.4, in accordance with the reported value of Gn cr E 0.3 in the stream cylinder. At intermediate gap ratios (Gn ¼ 0.4 and 0.6),
literature. The periodicity of lift signals is reflected in the corre- however, the lower shear layer is broken into two segments. The
sponding spectra shown in Fig. 5(b). For Gn Z0.4, each spectrum one in between the two cylinders deflects upward and reattaches
displays one obvious peak at St E0.2 (similar to the case of an on the leading surface of the downstream cylinder. Similar shear
isolated cylinder), with the magnitude of the peak progressively layer deflection and reattachment are evident for Ln ¼3 at inter-
increasing with Gn until Gn ¼0.8. This indicates that vortex mediate gap ratios.
40 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50

Fig. 5. (a) Time histories of dynamic lift coefficient (C L ) on the near-wall single cylinder at different gap ratios; and (b) corresponding spectra based on power spectral
density (PSD) function.

Fig. 6. A representative snapshot of the instantaneous normalized vorticity (ωn ¼ ωD=U) fields around the near-wall tandem cylinders at Ln ¼2, 3 and 5 and Gn ¼0.15, 0.4,
0.6 and 1.4. Positive: solid red lines; negative: dashed blue lines. Cut-off value ωn min ¼ 1; contour interval ¼0.5. (For interpretation of the references to color in this figure,
the reader is referred to the web version of this article.)
 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50 41

Based on the PIV measurement results, a map of flow patterns where periodic vortex shedding occurs, but the strength of vortex
around the near-wall tandem cylinders in the Gn –Ln plane is shedding reduces with decreasing Gn ; and (iii) small-gap regime
proposed in Fig. 7. With reference to Ln , it can be roughly divided (approx. Gn o0.3), where periodic vortex shedding is completely
into three basic types of spacing, that is, close (1oLn o2), moderate suppressed.
(2oLn o4) and wide (Ln 44), which are roughly equivalent to the
‘extended-body’, ‘reattachment’ and ‘impinging’ regimes, respec-
tively, for free-standing tandem cylinders. Similarly, the flow can be 3.2.2. Forces, lift spectra and Strouhal numbers on the downstream
broadly classified as a function of Gn : (i) large-gap regime (approx. cylinder
4 1), the flow and vortex shedding characteristic are similar to the This section presents the time histories of fluctuating lift on the
free-standing case; (ii) intermediate-gap regime (approx. 0.3o1), downstream cylinder and the corresponding spectra for different

Fig. 7. Overview of flow patterns for the near-wall two tandem cylinders as a function of both Gn and Ln .

Fig. 8. (a) Time histories of fluctuating lift coefficient; and (b) corresponding spectra on the downstream cylinder for different spacing ratios at Gn ¼ 0.15.
42 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50

combinations to Ln and Gn . The results for Ln ¼1.5–7 at Gn ¼0.15 are spacing ratios considered. Similar to the case of the near-wall
shown in Fig. 8. It is clear that the flow is in the vortex-shedding single cylinder (Fig. 5), each spectrum displays a peak at StE0.85
suppression regime. Accordingly, the lift signal is irregular for all associated with K-H type of roll-ups. It is noted that for wide-

Fig. 9. (a) Time histories of fluctuating lift coefficient; and (b) corresponding spectra on the downstream cylinder for different spacing ratios at Gn ¼0.4.

Fig. 10. (a) Time histories of fluctuating lift coefficient; and (b) corresponding spectra on the downstream cylinder for different spacing ratios at Gn ¼ 0.6.
 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50 43

spacing configuration (Ln Z4), there is an additional peak around exist at small- and intermediate-Gn disappears completely, so each
StE 0.18 (corresponding to the large-scale von Kármán vortex spectrum is characterized by a well-defined frequency at St¼0.15–
shedding) but it is rather broad-banded. 0.19. Based on the proposed classification, they belong to exte-
For Gn ¼0.4 (Fig. 9(a)), the signal gradually changes from a nded-body regime (Ln ¼1.5 and 2), reattachment regime (Ln ¼3
chaotic pattern at Ln ¼1.5 and 2, to a more periodic pattern at and 4), and impinging regime (Ln ¼5, 6 and 7), respectively. Several
Ln Z4. The amplitude of fluctuation increases sharply with Ln from features can be observed. Firstly, in either extended-body or
close- to moderate-Ln (Ln r4). As shown in Fig. 9(b), each spe- impinging regime, the peak is well-defined, while in reattachment
ctrum always exhibits a peak at StE0.83–0.86 associated with the regime (Ln ¼3 and 4) it is somewhat broad-banded. Secondly, the
K-H instability, implying that the wall effects are still significant. It periodicity of lift signal does not vary monotonically with Ln ;
should be noted that from Ln ¼3 onward, an additional peak at instead, it first achieves a minimum at Ln ¼ 3 (onset of reattach-
StE 0.18 appears as well. The double-peak character indicates the ment regime) and then a maximum at Ln ¼5 (onset of impinging
co-existence of two different fluid dynamic processes, as could be
appreciated from Fig. 6. Located in the lee of the upstream
cylinder, the downstream cylinder is subjected to shear layer
reattachment on its surface. At Gn ¼0.4, the shear layers are
basically in K-H type of roll-ups. However, when Ln is large enough
(e.g., Ln ¼ 5), discrete ‘patches’ of vorticity are formed in the wake
of the downstream cylinder, indicative of occurrence of vortex
shedding at relatively low frequency. Fig. 9(b) shows that in the
case of Ln ¼3, the peak at StE0.19 is rather broad-banded and
small in amplitude; with further increase in Ln , it becomes more
distinct, suggesting that vortex shedding becomes more regular
and stronger.
The results for Gn ¼0.6 are shown in Fig. 10. In this case, the lift
signal becomes significantly more periodic than that of Gn ¼ 0.4 at
the same spacing ratio. On the other hand, the peak for the high-
frequency K-H roll-ups becomes nearly invisible as a tiny hump,
indicating that the wall effects reduce with increasing Gn . All the
spectra except for those at Ln ¼ 3 and 4 display a dominant
frequency of St¼ 0.18–0.2 corresponding to periodic vortex shed-
ding. At Ln ¼ 3 and 4, however, the spectral peak is rather broad-
banded, suggestive of weakened vortex shedding activity.
At Gn 41, the wall effects become nearly negligible, see Fig. 11 Fig. 12. Variation of Strouhal number (St) with Ln at large gap ratios, together with
for Gn ¼ 1.4. The high-frequency component that may otherwise published data on two tandem cylinders under free-standing conditions.

Fig. 11. (a) Time histories of fluctuating lift coefficient; and (b) corresponding spectra on the downstream cylinder for different spacing ratios at Gn ¼1.4.
44 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50

regime), as indicated by the amplitude of lift fluctuation or the Igarashi 1981; Xu and Zhou, 2004), as shown in Fig. 12. This is
magnitude of spectral peak. These observations indicate that the likely attributed to the fact that an increase in Ln allows the shear
downstream cylinder is influenced by the advection and impinge- layers emanated from the upstream cylinder to grow thicker upon
ment of vortices shed from the upstream cylinder. At Ln Z5, the reaching the surface of the downstream cylinder. Accordingly, St
peak continues to decrease in magnitude with increasing Ln , decreases progressively with Ln , because a thicker shear layer leads
suggesting that the interference between the two cylinders is to a lower vortex shedding frequency from a cylinder (Roshko,
reducing. 1954). However, this trend cannot be sustained with further in-
Fig. 11(b) shows that the values of St for the spectral peak in crease in Ln since the flow would change into impinging regime at
impinging regime (Ln Z5) maintain approximately constant. For Ln cr E4, for which the shear layers emanated from the upstream
Ln r4, on the other hand, St first drops with Ln from 0.185 at cylinder will roll up into discrete vortices in between the two
Ln ¼1.5 to a minimum of 0.15 at Ln ¼3, and then recovers to 0.17 at cylinders and are no longer directly connected with the vortex
Ln ¼4. The initial decline of St with Ln at small- to moderate-Ln formation from the downstream cylinder.
seems to be an inherent feature when the wall proximity effects The force data indicate that the wall proximity tends to inhibit
are negligible, since a similar trend is found for Gn ¼2 as well as for periodic vortex shedding from the cylinders. Furthermore, spectral
free-standing tandem cylinders published in the literature (e.g., analysis was applied to the velocity data to illustrate the periodic

Fig. 13. Spectra of transverse velocity (v) in between the two cylinders for the case of Ln ¼5 at: (a) Gn ¼0.15, (b) 0.4, (c) 0.6, (d) 0.8, (e) 1.4 and (f) 2. The velocity signals are
retrieved from: (x; y) ¼ (2:5D;  0:5D), i.e., in the lower shear layer; and (x; y) ¼(2:5D; 0:5D), i.e., in the upper shear layer.
 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50 45

nature of the flow up to f ¼7.5 Hz (i.e., half the PIV sampling rate of coefficient on the downstream cylinder (C D ) remains consistently
15 Hz), or StE0.27. Fig. 13 shows the velocity spectra between the two lower than that of the isolated cylinder. As shown in Fig. 14(a), C D
cylinders for the case of Ln ¼5 at different gap ratios. The velocity increases monotonically with Ln for all gap ratios considered, but
signals are retrieved from (x; y)¼(2:5D;  0:5D) and (2:5D; 0:5D),
which are respectively located in the lower and upper shear layers. At
Gn ¼ 0.15, there is no peak over the measurement range, whereas at
Gn Z0.4, each spectrum begins to exhibit a dominant peak at
St¼0.17–0.2 (which is in accordance with the low-frequency compo-
nent in lift spectra). This confirms the validity of using lift signal as an
indicator of vortex shedding process. The peak magnitude increases
with Gn particularly over the range of Gn r0.8, suggesting that the
effects of wall proximity are decreasing and the strength of vortex
shedding becomes stronger (similar conclusion is inferred from the lift
spectra). It is noteworthy that for a given Gn , the peak in the upper
shear layer (denoted by red line) is generally higher in magnitude than
that in the lower shear layer (denoted by black line) for Gn r1.4,
indicating flow asymmetry about the wake centerline. At large enou
gh gap ratios (e.g., Gn ¼2), the two spectra almost coincide with
each other.
Fig. 14 presents the variations of the mean drag (C D ) and lift
(C L ), RMS drag (C D' ) and lift (C L' ) coefficients on the downstream
cylinder as a function of Ln for different gap ratios, together with
the corresponding values of the isolated single cylinder for com- Fig. 15. Critical gap and spacing ratios (Gn cr and Ln cr ) for onset of vortex shedding
parison. Located in the lee of the upstream cylinder, the mean drag from the downstream cylinder.

Fig. 14. Variation of dynamic force coefficients on the downstream cylinder with Ln at different gap ratios: (a) mean drag coefficient (C D ); (b) mean lift coefficient (C L );
(c) RMS drag coefficient (C D' ); and (d) RMS lift coefficient (C L' ).
46 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50

at different rates (as reflected by slope of the curves). At Gn ¼0.15, tandem cylinders, for instance, Sumner et al. (2005) reported a
C D is about 0.1 and increases slightly with Ln . As Gn increases to 2, minimum of ðC D Þmin ¼  0.55 at Ln ¼ 1.125. The negative (attractive)
the low end of each curve (at Ln ¼ 1.5) decreases until reaching a drag at small-Ln is due to the fact that the downstream cylinder is
minimum of ðC D Þmin ¼  0.13, whereas the high end (at Ln ¼7) completely enwrapped by the shear layers from the upstream
continues to rise up to ðC D Þmax ¼0.84. Therefore, when Gn is cylinder, and hence experiences a negative pressure. The mean lift
relatively large (Gn Z0.8), the downstream cylinder experiences coefficient (C L ), as shown in Fig. 14(b), on the other hand, varies
a drag inverse (from negative to positive) within the range of significantly with both Gn and Ln . Similar to the near-wall single
2 rLn r3. This is a well-known phenomenon for free-standing cylinder, the downstream cylinder generally experiences a positive

Fig. 16. Variation of the mean drag coefficient (C D ) on the downstream cylinder with: (a) Gn at large spacing ratio (Ln ¼ 7), and comparison with published data (Roshko et al.,
1975) and the present measurement data on a near-wall single cylinder; and (b) Ln at large spacing ratio (Gn ¼3), and comparison with published data under free-standing
conditions (Zdravkovich and Pridden, 1977; Harimi and Saghafian, 2012).

Fig. 17. Mean velocity vector fields for Ln ¼ 2, 3 and 5 and Gn ¼0.15, 0.4, 0.6 and 1.4. Superimposed with contours of the normalized spanwise mean vorticity (ωz D=U).
 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50 47

mean lift (C L 40), but it is considerably smaller in amplitude convex shape is found for Gn ¼1.4 and 2, where the two curves
(C L E 0.05–0.2). The significant variation of C L with Gn and Ln is almost coincide, implying diminishing effects of wall proximity.
probably due to the following two reasons. The first is the possible Depending on the values of Gn and Ln , the shear layers emanated
misalignment of the models, since it is extremely difficult, if not from the upstream cylinder may overshoot, reattach or impinge
impossible, to achieve a perfect alignment experimentally. Sec- upon the downstream cylinder and then separate, perhaps joining
ondly, the flow in the gap may keep steadily reattached on the those developed on the downstream cylinder itself, to form vortices
surface of the downstream cylinder, but may also spontaneously around the downstream cylinder. This results in different behavi-
flip-flop vertically so that the reattachment point varies irregularly ors of fluid dynamic forces on the downstream cylinder. Here, an
over time, which is in analogy to the flow between two side-by- attempt is made to identify the critical gap (Gn cr ) and spacing (Ln cr )
side cylinders at small gap ratios (Sumner et al., 1999). Figs. 14 ratios for vortex shedding based on the periodicity of fluctuating lift
(c) and 14(d) show that the variation trends of C D' and C L' with Ln signals and the intensity of the peak in lift spectra as shown in
are similar, although the latter has a much higher (3  4 times) Figs. 8 – 11 (similar observation can be obtained by analysis of the
magnitude. In general, C D' and C L' display different variation trends velocity signal as shown in Fig. 13). The map for absence/presence of
with respect to Ln depending on the value of Gn , which is vortex shedding from the downstream cylinder in Gn –Ln plane is
consistent with the PIV data. At Gn ¼0.15 when vortex shedding presented in Fig. 15. The vortex-shedding suppression regime is
is suppressed, C D' and C L' increase steadily with Ln over the mainly located at the lower-left corner (i.e., small-Gn and small-Ln ):
measurement range. In intermediate-gap regime (Gn ¼0.4, at the smallest gap ratio (Gn ¼ 0.15), it extends the whole Ln range; as
0.6 and 0.8), they increase rapidly with Ln until reaching a Gn increases, it gradually shrinks in width until completely disap-
maximum at Ln ¼5 (i.e., onset of impinging regime). A similar pears at Gn ¼ 0.8. In addition, the values of C L' can be used to

Fig. 18. Profiles of the normalized streamwise mean velocity (u=U) along two vertical lines (i.e., A-A and B-B, located behind the upstream cylinder and the downstream
cylinder, respectively) for different gap and spacing ratios.
48 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50

qualitatively determine the strength of vortex shedding. It should be ratios (Gn o1), to symmetrical patterns at large gap ratios (Gn 41).
noted that measurements with a higher resolution in Gn –Ln plane are For a given Ln , the recirculation length, defined as the distance
desirable in order to more accurately define the boundaries separat- from the cylinder base to the zero mean streamwise velocity point
ing the different flow regimes. Furthermore, as shown in Figs. 9 – 11, along the wake centerline, increases with Gn . Meanwhile, the gap
due to enhanced activity of shear layer reattachment for Ln ¼3 and flow is deflected upward in y-direction and reattach on the leading
4 in intermediate-Gn regime, the spectral peak is broad-banded in face of the downstream cylinder, most notably in the case of
these cases, which has been highlighted by the shaded region in intermediate-Gn and moderate-Ln (e.g., Gn ¼ 0.6 and Ln ¼3). This
Fig. 15. corresponds to the region of broad-banded peaks in the lift
As shown in Fig. 16(a), variations of the mean drag coefficient spectra, as shown in Fig. 15.
(C D ) on the downstream cylinder with Gn at the largest spacing Profiles of the normalized streamwise mean velocity (u=U)
ratio (Ln ¼7) have been compared with the published data (Roshko along two vertical lines located at 0:5D after the trailing edges of
et al., 1975) and the present measurement on a near-wall single the two cylinders (i.e., A-A and B-B) for different gap and spacing
cylinder. Similar to the case of the single cylinder, C D initially ratios are provided in Fig. 18. One obvious feature is that for a fixed
experiences a sharp increase with Gn before leveling off at large Gn , the profiles at different Ln at A-A almost collapse, whereas
enough gap ratios (Gn Z1). However, at the same Gn it is appre- those at B-B deviate from each other more evidently. This indicates
ciably lower than that of the single cylinder (particular when Gn is that the presence of the downstream cylinder mainly affects the
small), due to the effects from the upstream cylinder. This suggests flow behind it. The presence of the cylinders results in the velocity
that the spacing ratio of Ln ¼7 is still not sufficiently large for the defect behind each cylinder, so that the velocity profiles exhibit as
two tandem cylinders to be considered independently. On the an “S”-shape. When Gn o Gn cr (e.g., Gn ¼0.15), however, the lower
other hand, as shown in Fig. 16(b), the variation of C D with Ln at half of the “S”-shape is not obvious or even completely disappears
Gn ¼ 3 agrees well with the published data under free-standing due to the rather weak gap flow.
conditions (Zdravkovich and Pridden, 1977; Harimi and Saghafian, Furthermore, the shear layer developments can be appreciated
2012), confirming that the gap ratio of Gn ¼3 is large enough for from the contours of the normalized Reynolds shear stress
neglecting the wall effects. (u'v'=U 2 ) in Fig. 19. At small- or intermediate-Gn , the upper shear
layer is both stronger in magnitude and larger in size than the
lower one. For a given Ln , as Gn increases from 0.15 to 1.4, the
3.2.3. Ensemble-averaged flow patterns around the cylinders distributions of u'v'=U 2 gradually become more symmetric about
As shown above, the vortex shedding characteristics from the the wake centerline; meanwhile, regions of significant u'v'=U 2
cylinders depend on both Gn and Ln . This leads to corresponding contract in the streamwise direction (to smaller x), together with
variation in the statistical quantities of the flow, such as distribu- elevated level (or magnitude) of u'v'=U 2 .
tions of mean velocity vectors and Reynolds shear stresses. Similar to the flow classification based on the instantaneous
Consistent with the instantaneous flow structure, the mean vorticity fields, the distributions of u'v'=U 2 can be divided into
velocity vector field (Fig. 17) gradually changes from asymmetrical three different patterns – Pattern 1: regions of significant u'v'=U 2
patterns about the wake centerline at small- and intermediate-gap are found in the wake of the downstream cylinder only (e.g., for all

Fig. 19. Contours of the normalized Reynolds shear stress (u'v'=U 2 ) for Ln ¼2, 3 and 5 and Gn ¼ 0.15, 0.4, 0.6 and 1.4. Positive: solid red lines; negative: dashed blue lines. Cut-
off value u'v'=U 2 ¼0.01; contour interval ¼ 0.005 (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.).
min
 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50 49

gap ratios at Ln ¼ 2, and for relatively small gap ratios Gn r 0.6 at Grant no. NRF-CRP5-2009-01 is acknowledged. Hao Z. wishes to
Ln ¼3); Pattern 2: only the upper shear layer is evident, which is, acknowledge the support from Shanghai Municipal Education
however, located not only behind the downstream cylinder, but Commission Project (No. 12ZZ149) and Science and Technology
also in the space between the two cylinders (e.g., for Gn ¼0.15 and Commission of Shanghai Municipality (No. 10PJ1404700).
Ln ¼5); Pattern 3: both the lower and upper shear layers are
observable in between the two cylinders (e.g., for Gn Z 0.4 and
Ln ¼5, and for Gn ¼1.4 and Ln ¼3). References
The distributions of u'v'=U 2 for the case of Gn ¼1.4 and Ln ¼ 5
(where the wall effects are negligible and the flow is in impinging Bearman, P.W., Zdravkovich, M.M., 1978. Flow around a circular cylinder near a
plane boundary. J. Fluid Mech. 89, 33–47.
regime) are noteworthy. In the wake of the upstream cylinder, Bhattacharyya, S., Dhinakaran, S., 2008. Vortex shedding in shear flow past tandem
besides the two large-scale clusters of u'v'=U 2 , there are two square cylinders in the vicinity of a plane wall. J. Fluids Struct 24, 400–417.
additional small-scale clusters of opposite-signed vorticity imme- Buresti, G., Lanciotti, A., 1992. Mean and fluctuating forces on a circular cylinder in
cross-flow near a plane surface. J. Wind Eng. Ind. Aerodyn. 41, 639–650.
diately located upstream of the main shear layers (as highlighted
Carmo, B.S., Meneghini, J.R., Sherwin, S.J., 2010. Possible states in the flow around
by black arrows). This feature agrees with the results of Dong et al. two circular cylinders in tandem with separations in the vicinity of the drag
(2006) on an isolated single cylinder at Re ¼4000 and 10 000. The inversion spacing. Phy. Fluids 22, 003005PHF.
distributions of u'v'=U 2 in the wake of the downstream cylinder Dipankar, A., Sengupta, T.K., 2005. Flow past a circular cylinder in the vicinity of a
plane wall. J. Fluids Struct 20, 403–423.
are confined to a smaller region as compared to those of the Dong, S., Karniadakis, G.E., Ekmekci, A., Rockwell, D., 2006. A combined direct
upstream cylinder. numerical simulation-particle image velocimetry study of the turbulent near
wake. J. Fluid Mech. 569, 185–207.
Fouras, A., Soria, J., 1998. Accuracy of out-of-plane vorticity measurement derived
from in-plane velcotiy field data. Exp. Fluids 25, 409–430.
4. Concluding remarks Gerrard, J.H., 1967. Experimental investigation of separated boundary layer under-
going transition to turbulence. Phy. Fluids 10, S98–S100.
Harichandan, A.B., Roy, A., 2012. Numerical investigation of flow past single and
Flow around two near-wall tandem cylinders at Re ¼6300 has tandem cylindrical bodies in the vicinity of a plane wall. J. Fluids Struct 33,
been investigated extensively for 6 gap ratios (Gn ¼ 0.15, 0.4, 0.6, 19–43.
0.8, 1.4 and 2) and 7 inter-cylinder spacing ratios (Ln ¼1.5, 2, 3, 4, 5, Harimi, I., Saghafian, M., 2012. Numerical simulation of fluid flow and forced
convection heat transfer from tandem circular cylinders using overset grid
6 and 7), making 42 combinations in total. The results show that
method. J. Fluids Struct 28, 309–327.
both the flow patterns (instantaneous and ensemble-average) and Huerre, P., Monkewitz, P.A., 1990. Local and global instabilities in spatially
the dynamic forces (drag and lift) are highly dependent on the developing flows. Annu. Rev. Fluid Mech. 22, 473–537.
combined values of Gn and Ln , due to mixed effects of wall Igarashi, T., 1981. Characteristics of the flow around two circular cylinders arranged
in tandem – 1. B. JSME 24, 323–331.
proximity and mutual interference between the two cylinders. Lam, K., Zou, L., 2010. Three-dimensional numerical simulations of cross-flow
With reference to Ln , the flow interference between the two around four cylinders in an in-line square configuration. J. Fluids Struct 26,
cylinders changes from extended-body regime at close-spacing 482–502.
Lee, K., Yang, K.-S., Yoon, D.-H., 2009. Flow-induced forces on two circular cylinders
ratios (1 oLn o2), reattachment regime at moderate-spacing in proximity. Comput. Fluids 38, 111–120.
ratios (2 oLn o 4), to impinging regime at wide-spacing ratios Lei, C., Cheng, L., Kavanagh, K., 1999. Re-examination of the effect of a plane
(Ln 4 4). However, this classification is valid only for large gap boundary on force and vortex shedding of a circular cylinder. J. Wind Eng. Ind.
Aerodyn. 80, 263–286.
ratios (Gn 41). Lin, W.-J., Lin, C., Hsieh, S.-C., Dey, S., 2009. Flow characteristics around a circular
The nearby wall tends to promote the K-H type of roll-ups and cylinder placed horizontally above a plane boundary. J. Eng. Mech. 135,
inhibit periodic vortex shedding from the cylinders. Therefore, in 697–716.
Ljungkrona, L., Norberg, Ch., Sundén, B., 1991. Free-stream turbulence and tube
intermediate-gap regime (0.3 oGn o1) vortex shedding takes
spacing effects on surface pressure fluctuations for two tubes in an in-line
places but in an asymmetric manner about the wake centerline, arrangement. J. Fluids Struct 5, 701–727.
whereas in small-gap regime (Gn o 0.3) vortex shedding is sup- Mahir, N., 2009. Three-dimensional flow around a square cylinder near a wall.
Ocean Eng. 36, 357–367.
pressed completely. When Gn r0.4, the lift spectrum displays a
Maiti, D.K., 2012. Numerical study on aerodynamic characteristics of rectangular
double-peak mode for Ln Z3, namely, St E0.18–0.2 and 0.83–0.85, cylinders near a wall. Ocean Eng. 54, 251–260.
indicating the co-existence of two instability mechanisms (i.e., von Maiti, D.K., Bhatt, R., 2014. Vortex shedding suppression and aerodynamic char-
Kármán vortex shedding and K-H roll-up). On the other hand, acteristics of square cylinder due to offsetting of rectangular cylinders towards
a plane. Ocean Eng. 82, 91–104.
either an increment or a decrement in Gn would result in single- Nishino, T., Roberts, G.T., Zhang, X., 2007. Vortex shedding from a circular cylinder
peak mode, that is, St E0.84 for Gn ¼0.15, and StE 0.19 for near a moving ground. Phy. Fluids 19, 025103.
Gn Z 0.6. Norberg, C., 2003. Fluctuating lift on a circular cylinder: review and new measure-
ments. J. Fluids Struct 17, 57–96.
For close- and moderate-spacing configurations, the shear Ong, M.C., Holmedal, L.E., Myrhaug, D., 2012. Numerical simulation of suspended
layers emanated from the upstream cylinder (which would other- particles around a circular cylinder close to a plane wall in the upper-transition
wise be kept roughly horizontal) deflect upwards and reattach on flow regime. Coast. Eng. 61, 1–7.
Price, S.J., Sumner, D., Smith, J.G., Leong, K., Païdoussis, M.P., 2002. Flow visualiza-
the leading surface of the downstream cylinder, leads to relatively tion around a circular cylinder near to a plane wall. J. Fluids Struct 16, 175–191.
higher mean drag (C D ) and RMS drag/lift (C D' and C L' ) coefficients Rajagopalan, S., Antonia, R.A., 2005. Flow around a circular cylinder-structure of the
on the latter. The enhanced activity of shear layer reattachment near wake shear layer. Exp. Fluids 38, 393–402.
Roshko, A., 1954. On the drag and shedding frequency of two-dimensional bluff
also results in broadened spectral peak for vortex shedding at bodies. NACA Technical Note No, 3169.
moderate-Ln (Ln ¼3–4) and intermediate-Gn (Gn ¼ 0.4–0.8). The Roshko, A., Steinolfson, A., Chattoorgoon, V., 1975. Flow forces on a cylinder near a
study also attempts to quantitatively identify the critical gap wall or near another cylinder. Proceedings of the Second U.S. National
(Gn cr ) and spacing (Ln cr ) ratios for the onset of vortex shedding Conference on Wind Engineering Research, in Colorado State University, Fort
Collins, IV-15.
from the downstream cylinder based on the fluctuating lift and the Sakamoto, H., Haniu, H., Obata, Y., 1987. Fluctuating forces acting on two square
lift spectra. prisms in a tandem arrangement. J. Wind Eng. Ind. Aerodyn. 26, 85–103.
Sarkar, S., Sarkar, S., 2010. Vortex dynamics of a cylinder wake in proximity to a
wall. J. Fluids Struct 26, 19–40.
Sumner, D., 2010. Two circular cylinders in cross-flow: a review. J. Fluids Struct 26,
Acknowledgments 849–899.
Sumner, D., Richards, M.D., Akosile, O.O., 2005. Two staggered circular cylinders of
equal diameter in cross-flow. J. Fluids Struct 20, 255–276.
Funding support from Singapore National Research Foundation Sumner, D., Wong, S.S.T., Price, S.J., Païdoussis, M.P., 1999. Fluid behavior of side-by-
(NRF) through the Competitive Research Programme (CRP) under side circular cylinders normal to a stream. J. Fluids Struct 13, 309–338.
50 X.K. Wang et al. / Ocean Engineering 94 (2015) 36–50

Tadrist, H., Martin, R., Tadrist, L., Seguin, P., 1990. Experimental investigation of Xu, G., Zhou, Y., 2004. Strouhal numbers in the wake of two inline cylinder. Exp.
fluctuating forces exerted on a cylindrical tube (Reynolds numbers from 3000 Fluids 37, 248–256.
to 30 000). Phys. Fluids A 2, 2176–2182. Zdravkovich, M.M., 1987. The effects of interference between circular cylinders in
Wang, X.K., Hao, Z., Tan, S.K., 2013. Vortex-induced vibrations of a neutrally buoyant cross flow. J. Fluids Struct 1, 239–261.
cylinder near a plane wall. J. Fluids Struct 39, 188–204. Zdravkovich, M.M., Pridden, D.L., 1977. Interference between two circular cylinders;
Wang, X.K., Tan, S.K., 2008a. Near-wake flow characteristics of circular cylinder Series of unexpected discontinuities. J. Ind. Aerodyn 2, 255–270.
close to a wall. J. Fluids Struct 24, 605–627. Zhou, Y., Yiu, M.W., 2006. Flow structure, momentum and heat transport in a two-
Wang, X.K., Tan, S.K., 2008b. Comparison of flow patterns in the near wake of a tandem-cylinder wake. J. Fluid Mech. 548, 17–48.
circular cylinder and a square cylinder placed near a plane wall. Ocean Eng. 35,
458–472.
West, G.S., Apelt, C.J., 1997. Fluctuating lift and drag forces on finite lengths of a
circular cylinder in the subcritical reynolds number range. J. Fluids Struct 11,
135–158.