KSCE Journal of Civil Engineering (2016) 20(6):2559-2568 Water Engineering

Copyright ⓒ2016 Korean Society of Civil Engineers

DOI 10.1007/s12205-015-0209-3 pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

Numerical Investigation of Flow Around Circular Cylinder with Splitter Plate

Huy Cong Vu*, Jungkyu Ahn**, and Jin Hwan Hwang***
Received March 13, 2015/Revised July 23, 2015/Accepted September 3, 2015/Published Online October 9, 2015

··································································································································································································································

Abstract

The flow around a single circular cylinder with splitter plate was investigated using Computational Fluid Dynamics. The present
study investigated the effects of varying splitter plate length and the Reynolds numbers on flow characteristic, such as vortex
shedding, drag force, lift force, separation point, pressure, and friction coefficients of the cylinder. It was revealed that the vortex
shedding behind cylinder is completely suppressed when the splitter plate length is longer than the critical value, which is
proportional to the Reynolds number. The variation of drag and lift coefficients with respect to splitter length can be classified into
two patterns of a monotonic decrease and a general decrease but with small increase. We found that the first pattern occurs at
Reynolds number ≤ 180 and higher Reynolds number gives the second pattern. When the splitter plate length is similar to the
diameter of the cylinder, the vortex frequency, drag, and lift coefficients reach local minimum depending on the Reynolds number. A
functional relationship between drag coefficient, length of splitter plate, and Reynolds number was determined.
Keywords: cylinder, splitter plate, drag force, drag coefficient, vortex shedding, separation point
··································································································································································································································

1. Introduction these structures. For example, vortex shedding was indicated as

one of the major cause of the failure of three towers at Ferry bridge
The flow around circular cylinder has been a subject of many Power Station C in United Kingdom in 1965 (Bandyopadhyay,
studies because of its common occurrence in many forms and 1986); and a thrill ride “VertiGo” at Cedar Point in Sandusky,
diverse applications. The appearance of circular cylinder changes Ohio, suffered from vortex shedding during the winter of 2001,
the flow structure significantly in particular by generating the causing one of the three towers collapse (Byko, 2002).
vortex shedding or wakes. Kumar and Mittal (2006) noted that Therefore, the controlling the vortex shedding, which can
the vortex shedding appears behind the cylinder and flow vibrate the bluff body (Mittal and Raghuvanshi, 2001) is necessary
becomes unsteady when the Reynolds number is higher than 47. for design the structures. For example, the suppression vortex for
Even though the vortex shedding has been studied intensively, long flexible cylindrical structures enduring ocean currents is
most of the previous studies focused on measuring frequency of common in the offshore industry (Mukundan, 2008). There were
vortex, in terms of the Strouhal number, St, (e.g. Goswami et al., several works on controlling the vortex shedding exerted by
1992; Williamson, 1989 and 1996; Jain and Goel, 1976; Fey et cylinder. Ozono (1999) investigated flow behavior by placing a
al., 1998) but the role of vortex shedding in the formation of detached short splitter-plate at different distances from cylinders.
resistance force is not fully investigated yet; or force mechanism The variation in Strouhal number significantly depends on
has still not been studied in detail. location of short splitter plate. Strykowski and Sreenivasan
However, it is well known that the appearance of vortex (1990) conducted experiment and concluded that the vortex
shedding behind a circular body results in the change of flow shedding past a circular cylinder can be controlled by a proper
structure, such as pressure distribution around the body, and placement of a (smaller) control cylinder close to the main one.
increases the drag force. The shedding is also responsible for the They reported that the suppression of vortex shedding is
fluctuation of lift force, which is normal to the stream direction accompanied by a significant reduction in the mean drag coefficient.
(Kundu, 1990) and the fluctuated shedding of vortices forces Hwang and Yang (2007) investigated the drag reduction resulted
alternatingly on the structures. This occurs any time on a long from the use of dual short splitter-plates, one in the upstream and
structure, which is exposed to flowing fluid, such as bridge the other in the downstream of the cylinder. In a certain case,
peers, pipelines, and marine risers. The fluctuating force shortens they achieved 38.6% reduction of the drag coefficient. In
the life of the structure, in some cases, has a significant effect on addition, splitter plate is known as a passive method and one of

*Ph.D Candidate, Dept. of Civil & Environmental Engineering, Dongguk University, Seoul 08826, Korea (E-mail: [email protected])
**Member, Assistant Professor. Dept. of Civil and Environmental Engineering, Incheon National University, Incheon 22012, Korea (Corresponding
Author, E-mail: [email protected])
***Member, Associate Professor, Dept. of Civil & Environmental Engineering, Seoul National University, Seoul 08826, Korea (E-mail: [email protected])

− 2559 −
 Huy Cong Vu, Jungkyu Ahn, and Jin Hwan Hwang

the most successful mean to control the vortex and so has been pressure coefficient and separation point, under the effect of
studied in the several works (e.g., Roshko, 1955; Akilli et al., splitter plate with respect to the length and the Reynolds
2008; Gerrard, 1966). numbers. We simulated the flow around cylinder without splitter
Roshko (1955) used a splitter plate to observe the change of plate and compared to previous experimental and numerical data
flow at 1.45×104 of the Reynolds number, Re, with respect to the for a validation of the numerical model. Then, simulations were
ratio between the length of splitter, L, and the diameter of conducted with the cases of splitter plate. The length of the
cylinder, D. He found that a splitter plate of L = 5D reduced the splitter attached on the cylinder varied L/D of from 0.5 to 6 under
vortex shedding and pressure drag on the cylinder by 63% Re of 60~240. Considering the suppression of vortex shedding
compared to the one without the splitter. Gerrard (1966) by splitter plate, we present shedding frequency, reduction of
measured the frequency of vortex shedding from a circular fluid drag and lift forces (coefficient), pressure distribution, and
cylinder at Re = 2.0×104 by changing the length of the splitter flow separation point. Determination of the critical length of the
plate up to 2D. In his study, the Strouhal number decreases when splitter plate, which minimizes the vibration and fluid forces, is
L < D, but it increases for D < L < 2D. Apelt et al. (1973 and presented also.
1975) conducted experiment at Re from 1.0×104 to 5.0×104 for 0
< L < 2D in 1973 and L > 2D in 1975 to investigate the effect of 2. Numerical Model and Validation
a splitter plate. They found that the drag coefficient decreased by
31% at L/D = 1 and the vortex shedding disappeared at L/D > 5. 2.1 Governing equations
Kwon and Choi (1996) simulated cylinder with a splitter plate at Flows around two cylinders were simulated using a Computational
low Re (= 80~160) to investigate St and flow modification. They Fluid Dynamic code, Fluent. The finite volume method was
found two types of St profiles with respect to L/D, one is applied to obtain a solution of the Navier-Stokes equations and
monotonic decrease and the other is general decrease with small the coupling of pressure and velocity field was solved by the
increase, depending on Re. However, other important feature, Semi-Implicit Pressure Linked Equation (SIMPLE) scheme.
such as lift force on the cylinder, was not fully investigated. Reynolds Averaged Navier-Stokes, RANS, models use various
Grove et al. (1964) conducted laboratory experiments for Re = closure schemes such as k-ε or k-ω to create the turbulent energy
25~177 focusing on the separated flow around a circular cylinder field, and they are widely used because of their low computational
with and without splitter plate. They found that the appearance or cost (Chang et al., 2015). The Shear Stress Transport (SST) k-w
length of the splitter plate had no noticeable effect on the flow model was used for turbulence closure in this study. The reason
separation point. is attributed to the improvement in SST k-w model in prediction
Most of the previous studies mainly focused on the frequency the flow separation under adverse pressure gradient, which is
of the vortex shedding and the drag coefficient at high Reynolds important behavior of flow when it passes circular cylinder and
numbers so far. However, the information on lift force coefficient (Bardina et al., 1997; Vu et al., 2015). The continuity and
was hardly discussed regard with Re. In fact, the alternating lift momentum equations are as follows, (ANSYS, 2010):
force is an important factor resulting in the vibration of structure.
∂ρ
Flow features on the surface of cylinder, such as pressure ------ + ∇ ⋅ ( ρv ) = 0 (1)
∂t
coefficients were not fully investigated although this is required
to interpret the variation of fluid force. Therefore, more work where ρ is density of water, t is time, and v is velocity.
needs to be done to achieve a deep understanding of flow around
∂
cylinder with splitter plate, which includes fundamental features, ---- ( ρv ) + ∇ ⋅ ( ρvv ) = –∇p + ∇ ⋅ ( τ ) + ρg + F (2)
∂t
such as the wake, flow separation, vortices, and force. In
addition, the influence of Re on the flow characteristic needs where p is the static pressure, τ is the stress tensor, and ρg and
further investigations. Several previous studies found the local F are the gravitational and external body forces, respectively.
minimum of the drag coefficient, CD, at around L/D = 1 in the The stress tensor τ is given by:
range of high Reynolds number (Anderson and Szewczyk, 1997
τ = μ ⎛ ∇v + ∇v ⎞ – --- ∇ ⋅ vI
T 2
at Re = 46,000, and Apelt et al., 1973 at Re = 1.0 × 104~5.0 × (3)
⎝ ⎠ 3
104). However, CD decreases with L/D without having local
minimum for low Reynolds numbers of 80~160 (Kwon and where the superscript of T represents a transpose of matrix, μ is
Choi, 1996). the dynamic viscosity and I is the unit tensor. The second term on
We can reduce the alternating fluid force by suppressing vortex the right hand side is the effect of volume dilation (ANSYS,
shedding. Information on the flow around a cylinder like 2010).
structure with a splitter, if excessive force and/or vibrations are
expected, is required from the designing stage of the structure. 2.2 Computational Domain and Boundary Conditions
To address these issues, this study focused on a systematical The schematic diagrams of the computational domain for a
investigation on the relationships between flow characteristics cylinder with and without splitter plate are shown in Fig. 1. The
around cylinder, such as vortex shedding, drag and lift forces, dimension of the computational domain was selected to eliminate

− 2560 − KSCE Journal of Civil Engineering

 Numerical Investigation of Flow Around Circular Cylinder with Splitter Plate

shown in Eq. (5)~(9), which can be found in many text books

and papers, such as Kundu (1990), Fox et al. (2008), Lam et al.
(2008), and White and Nepf (2003).
2π 2π

FD = ∫ R p cosθdθ + ∫ Rτwsinθ dθ (5)

0 0
Fig. 1. Computational Domain for Circular Cylinder: (a) Without
where FD is the drag force acting on the cylinder, R (= D/2) is the
Splitter Plate, (b) With Splitter Plate
radius of cylinder, τw is the local wall shear stress, θ is the
angular displacement from the front stagnation point (θ = 0o).
boundary effects on the computational results. The inlet and outlet The lift force, perpendicular to the direction of flow, can be
boundaries locate at 8D and 24D from the center of cylinder, determined similarly.
respectively. The lateral walls are located at 10D each from the 2π 2π

center of cylinder. These distances were chosen similarly to studies FL = ∫ R psinθ dθ + ∫ Rτw cosθ dθ (6)
of Mahir and Altac (2008), Meneghini and Saltara (2001), Seo and 0 0

Song (2012). A splitter plate was attached to the cylinder and where FL is the lift force.
placed along the center of the wake. Akilli et al. (2008) conducted The drag force coefficient can be determined from its original
the laboratory experiments with the splitter thicknesses of 0.016D, definition.
0.04D, and 0.08D. They noted that the plate thickness has no 2π 2π
2FD 1 1
considerable effect on the flow structure. The thickness of splitter 2
- = --- ∫ Cp cosθdθ + ---------2 ∫ τw sinθ dθ
CD = ------------- (7)
plate was selected to be 0.025D for this study. ρUo D 2
0
ρU o 0
We used no-slip boundary condition for both the cylinder wall where Cp is named after Leonhard Euler and often called the
and the splitter plate. The periodic boundaries were used for the pressure coefficient (Fox et al., 2008). It is as defined as:
side walls. The uniform velocity and pressure outlet were used as
Cp = ( p – p0 ) ⁄ ⎛ --- ρU0⎞
1 2
upstream and downstream boundary conditions, respectively, see (8)
⎝2 ⎠
Fig. 1. The velocity at the inlet boundary was imposed to obtain
the expected Reynolds number as Eq. (4). where p0 is the free stream pressure.
The same as drag force, the drag coefficient can be divided
ρUo D
Re = -------------
- (4) into pressure drag coefficient (CDp), the first term on the right
μ
hand side of Eq. (7), and friction one (CDf), the second term.
where Uo is the velocity of free stream. Similarly, the lift force coefficient is:
A multi-block grid system was applied to make computational 2π 2π
2FL
mesh. Domains near the cylinder and splitter plate were meshed - = 1--- ∫ Cp sinθdθ + ---------
CL = -------------
2
1
τ cosθ dθ
2∫ w
(9)
with finer body-fitted grids, whereas coarser and rectangular ρUo D 2 0 ρUo 0
ones were used for the rest of domain, see Fig. 2. As the vortices shed, drag and lift forces periodically exert on
the cylinder. The lift force, which is in the transverse direction,
2.3 Numerical Parameters (fluid forces acting on cylinder) has the same frequency as the vortex-shedding cycle, while the
In this study, the drag and lift forces were calculated by frequency of the drag force, which is in the stream direction, is
integrating the pressure and skin drags around the cylinder, as twice of the vortex shedding frequency (Behzad and Hamed,
2010; Wanderley et al., 2002). The drag and lift force coefficients
were shown to be continuous over time in Fig. 3. The fluctuations

Fig. 2. Computational Mesh for Cylinder with Splitter Plate: (a)

Entire Domain, (b) Around Cylinder, (c) Near a Quarter of
Cylinder (subplot (a) and (b), every sixth grid points are dis- Fig. 3. Temporal Variation of Drag and Lift Force Coefficients of
played) Circular Cylinder Without Splitter Plate at Re = 200

Vol. 20, No. 6 / September 2016 − 2561 −

 Huy Cong Vu, Jungkyu Ahn, and Jin Hwan Hwang

of these forces lead the vibration of structure as discussed in Vu et al. (2015), two grid resolution schemes with the number
previously. The amplitude of the variation of drag coefficient is of grid cells around cylinder of 160 and 320 were tested under
smaller than that of lift coefficient. From now, the drag force Re = 60 and 100. With these two mesh resolutions, the grid cells
coefficients are presented by the time-averaged values and the in the domain were 173760 and 215680, respectively. Table 1
lift force coefficients are represented by root mean squared value shows the result of grid independency test in terms of drag
(Hwang et al., 2006). coefficient and the Strouhal number. The results of CD and St
The Strouhal number, as defined as Eq. (10), is a dimensionless show good agreements with the previous results in particular 240
frequency of vortex shedding, which is obtained based on the grid cells for all investigated Reynolds number as expected and
dominant frequency in the time series of lift coefficient using 240 grid cells were chosen or the present simulation.
Fourier analysis (Surmas et al., 2004; Menghini and Saltara, The pressure distribution, Eq. (8), was verified for a single
2001; Vu and Hwang, 2014). cylinder case at the Reynolds number of 100 as shown in Fig. 4.
Because the distribution of Cp is symmetrical, only a half of
fs D
St = ------
- (10) cylinder is presented. In this figure, Cp generally decreases from
Uo
its maximum value at the front stagnation point, θ = 0o, to θ = 90o
where fs is the shedding frequency. and then slightly increase to the rear, θ = 180o, due to the
recirculation of flow behind cylinder. The results show good
2.4 Numerical Validation agreements with results of Lixia et al. (2013) and Sharman
The flow field around single cylinder without splitter plate was (2005).
compared with the numerical and experimental results from the The drag coefficient and the Strouhal number of a cylinder
literatures. A grid independency test around a single cylinder without splitter were examined for the different Reynolds
was presented by Vu et al. (2015), Seo and Song (2012). Vu et numbers. The Strouhal number was calculated and presented
al. (2015) concluded that simulation with 240 grid cells around together with the Reynolds number in Fig. 5(a). In the range of
the cylinder provides the most reasonable computational results the investigated Reynolds number, the Strouhal number increases
under Re = 200. In this study, beside the grid test was presented as the Reynolds number increases. The present results agree well
with the data reported by Sa and Chang (1991) and Williamson
(1989). The relationship between the drag coefficients and the
Table 1. The Calculated and Measured Results for Flow over a Reynolds numbers is shown in Fig. 5(b). In which, the drag
Single Cylinder coefficient decreases as Re increases. Relationship between CD
Re Study CD St and Re also shows a good agreement with the previous
Tritton (1959) (Re = 60,5)a 1.47 -
Ncb = 240 (Vu et al., 2015) 1.468 -
60
Ncb = 160 (This study) 1.479 -
Ncb = 320 (This study) 1.456 -
Su and Kang (1999) 1.36 0.163
Lam et al. (2008) 1.36 0.16
Meneghini and Saltara (2001) 1.37 0.165
Mahir and Altac (2008) 1.368±0.029 0.172
Liu et al. (1998) 1.35±0.012 0.164
100
Harichandan and Roy (2010) 1.352±0.01 0.161
Norberg (2003)a - 0.168
Ncb = 160 (This study) 1.389 0.163
Ncb = 240 (Vu et al., 2015) 1.366 0.16
Ncb = 320 (This study) 1.348 0.155 Fig. 4. Mean Pressure Coefficient Distribution Around Circular Cylin-
Lam et al. (2008) 1.32 0.196 der (without splitter plate) at Re = 100, where θ is the Angle
Meneghini and Saltara (2001) 1.3 0.196 of Points Around Circular Cylinder (modified from Vu et al.,
Williamson (1996)a - 0.183 2015)
Norberg (2003)a - 0.18-0.196
Zhang and Dalton(1998) 1.32 0.198
200
Ding et al. (2007) 1.348±0.05 0.196
Liu et al. (1998) 1.31±0.049 0.192
Ncb = 160 (Vu et al., 2015) 1.38 0.191
Ncb = 240 (Vu et al., 2015) 1.33 0.186
Ncb = 320 (Vu et al., 2015) 1.303 0.182
a : laboratory experimental studies, Ncb : the number of grid cells on the Fig. 5. Reynolds Number Versus: (a) Strouhal Number, (b) Drag
cylinder. Force Coefficient for Cylinder Without Splitter Plate

− 2562 − KSCE Journal of Civil Engineering

 Numerical Investigation of Flow Around Circular Cylinder with Splitter Plate

Fig. 6. Mean Separation Angle, θs , and Amplitude of the Fluctua-

tion of the Separation Angle, Δθs
Fig. 7. Variation of Strouhal Number with L/D

researches.
The separation point is defined as a point between the forward
and backward flows near the cylinder, where the local shear
stress, τw, vanishes (Kundu, 1990; Taamneh, 2011). The angle of
separation, θs, was measured from the stagnation point θ = 0o to
the point of flow separated from the cylinder. The shedding of
vortices from the cylinder also results in the oscillating
separation angle. The separation point varies between the points
“a”(θa) and “b”(θb) as shown in Fig. 6. The time averaged value
of angle of separation, θs , was obtained during about 30 vortex
shedding cycles. Fig. 6 shows the averaged separation angle and
its variation with respect to the Reynolds number. As the
Reynolds number increases, the location of mean separation
shifts to the front and the amplitude of variation, Δθs = θs – θa , Fig. 8. Vorticity Contour Versus L/D at Re = 160
increases due to the stronger vortex behind the cylinder. The
mean separation angle and its amplitude of variation were in a
good agreement with the previous results from Park et al. (1998) number at L/D = 1 was also reported by Cimbala and Garg
and Wu et al. (2004). Therefore, it was concluded that our (1990) for Re = 7.5 × 103. This implies that St has a local
numerical model has capability of simulating flow around a minimum value at L/D = 1 regardless of Re, if Re is larger
single cylinder without splitter. The capability of the numerical than 140. Such behaviors of St depending on the Reynolds
model around a cylinder even with splitter plate is shown in the numbers were also reported by Kwon and Choi (1996) and
following sections by comparing our numerical results to the Fig. 7 includes their data for the cases of Re = 100, 120, and
literatures. 160. The minimum St was found at the slightly different
Reynolds numbers, but it could be due to the different
3. Results and Discussion numerical approach and overall trends seem to agree well
with the present observations.
3.1 Shedding Frequency (Strouhal number) Figure 8 shows the variations of vortex formation with respect
The splitter plate changes the frequency of vortex shedding to the various lengths of splitter plate at Re = 160. The shear
behind the cylinder, which is expressed with the Strouhal layers from both sides of the cylinder roll and form vortex
number. Therefore, we investigated the effects of L/D on the shedding at the downstream. With the presence of the splitter
Strouhal number at various Reynolds (Fig. 7). When L/D ≤ 1, plate, the shape of vortex shedding varies significantly depending
St decreases significantly for all Reynolds numbers. When L/ on L/D. When L/D = 0.5 and 1, the vorticity layers elongate in
D > 1, however, the Strouhal numbers show the different stream direction and the vortex formed at the edge of the splitter
patterns depending on the Reynolds numbers. For Re < 140, plate as shown in Figs. 8(b) and (c) but still very similar to the
the Strouhal number decreases monotonically as L/D increases. case of no-splitter. As L/D increases, the interaction of shear
However, for Re > 140, St grows gradually until reaches local layers on both sides of the cylinder is blocked by the splitter plate
maximum around L/D = 2.2. At the higher Reynolds number, and the wake region significantly elongates streamwise as shown
this trend was also found in other studies (e.g., Cimbala and in Figs. 8(e), (f), and (g). The elongation of the wake region is an
Garg (1990) for Re = 7.5 × 103, Apelt and West (1975) for Re indication of the substantial attenuation of vortex formation
= 1.0 × 104 – 5.0 × 104). The local minimum of the Strouhal (Akilli et al., 2008). At L/D = 4.5, the vortex shedding is

Vol. 20, No. 6 / September 2016 − 2563 −

 Huy Cong Vu, Jungkyu Ahn, and Jin Hwan Hwang

Fig. 9. Variation of Lift Force Coefficient with the Length of Splitter

Plate

Fig. 10. Comparison of Lift Force Coefficient from Eq. (11) and
completely suppressed for Re = 160, Fig. 8(h) Numerical Result

3.2 Lift Force Coefficient

Figure 9 presents the variation of root mean squared value of defined as the critical length of a splitter plate, (L/D)c, in which
the lift force coefficient, CL, with respect to L/D depending on the vortex shedding begins to disappear. The critical length ratios
the Reynolds numbers. There are two obvious ranges of CLs at Re = 100 and 160 are consistent with previous studies. Ribeiro
depending on the Reynolds number, which are distinguished by et al. (2002), reported that the suppression of vortex shedding at L/
the appearance of local minimum around L/D = 1. For the cases D = 3 for Re = 100 and L/D = 4.6 for Re = 160 while Kwon and
of Re >180, CL significantly decreases in the range of L/D = 0~1 Choi, (1996) found that (L/D)c are approximately 3 and 5 for Re
and reaches to the local minimum around L/D = 1. It is noted that = 100 and 160 which are similar to the present experiments.
L/D of 1 coincides with the ratio that has the local minimum When the L/D is smaller than (L/D)c, a functional relationship
value of the Strouhal number in the previous section, Fig. 7. In between CL -L/D-Re was determined as:
fact, at L/D = 1, CL reduces significantly by 84% and 82% for Re
CL = – 0.15682 ( L ⁄ D ) + 0.005214Re
0.367975 0.797327

= 240 and 200, respectively. Beyond this ratio, CL increases (11)

slightly up to around L/D = 2.2 and then decreases again as L/D [ L ⁄ D < ( L ⁄ D )c, Re ≤ 180 ]
increases. When Re ≤ 180, CL decreases when L/D ≤ 1 for all Re. Equation (11) was determined by fitting the data shown in Fig.
In the conditions of L/D ≥ 1, CL shows a local minimum around 9 to predict the value of CL for the wide arrange of Reynolds
L/D = 1.5 and then increases slightly for Re = 160 and 180. After numbers and the length ratios, L/D. The comparison between Eq.
reaching the local maximum at around L/D = 2.2, CL decreases (11) and modeled results are shown in Fig. 10. Prediction with
gradually as L/D increases. For Re = 60~140, CL monotonically Eq. (11) gives the correlation coefficient of 0.956 and standard
decreases as L/D increases. error of 0.027. It can be seen that very high degree of agreement
In summary, the finding of the critical Reynolds number for was found between the values of CL obtained from the Eq. (11)
two patterns of CL profile implies the sensitivity of the lift force and numerical result.
coefficient to the Reynolds number. The local minimum values
of CL were clearly observed at high Reynolds number, Re >180, 3.3 Drag force Coefficient
and disappears at low Reynolds number, Re ≤ 180. At L/D > 1, Figure 11 shows the variations of the drag coefficient with
the splitter plate does not contribute much to the sudden decrease respect to L/D and Reynolds numbers. Similarly to the Strouhal
of CL but even CL increases when 1< L/D < 2.2 for Re >180. At L/ number and lift force coefficient, the variations of the drag
D < 1, CL decreases more in higher Re. coefficients are observed in two trends depending on the
When the vortex shedding occurs behind the cylinder, it makes Reynolds numbers.
the body to be exposed to the periodic force in the normal When Re is less than 180, the drag force coefficients decrease
direction to the main stream. When the vortex shedding disappears, monotonically along with L/D. In particular for the relatively
the fluctuation of lift force is also suppressed (Dehkordi and larger Reynolds number cases, the drag force coefficient
Jafari, 2010). In other words, the fluctuation of the lift force decreases significantly at L/D ≤ 1. For instance, when the Reynolds
coefficient confirms the existence of periodic vortex shedding number is 180, CD decreases to 88.7 % and 84.2% of that of the
behind the cylinder. Meanwhile, the lift force coefficient becomes cylinder without splitter plate at L/D = 0.5 and 1, respectively.
zero at different L/Ds depending on the Reynolds numbers. CL When the Reynolds number is 80, CD decreases to 95.7% and
equals to zero at L/D = 1, 1.5, 3, 3.5, 4, 4.5, 5 for Re = 60, 80, 92.4% of that of the cylinder without splitter. Therefore, splitter
100, 120, 140, 160, and 180, respectively. These values are plate reduces CD more efficiently in the larger Reynolds number

− 2564 − KSCE Journal of Civil Engineering

 Numerical Investigation of Flow Around Circular Cylinder with Splitter Plate

Fig. 11.Variations of Drag Coefficient with the Length of Splitter

Plate: Circled Points in Each Case Indicate the Conditions Fig. 12. Comparison of Drag Coefficient from Eq. (12) and Numeri-
when the Vortex Shedding Disappears cal Result

cases. In contrary, as L/D increases over 1, the drag coefficients appearance of vortex shedding.
decrease slowly. After L/D reaches a critical spacing ratio, CD Similarly to the lift force, a functional relationship between CD
does not decrease any more, which is consistent with Apelt and -L/D-Re was determined as:
West (1975) although their experiment was performed with the 0.327984 –0.14303
CD = – 0.15251 ( L ⁄ D ) + 2.74379Re
different Reynolds numbers. They concluded that CD is constant (12)
once L/D becomes larger than the critical value. The results also [ L ⁄ D < ( L ⁄ D )c ]
agree well with Kwon and Choi (1996) for Re = 160. The comparison between Eq. (12) and modeled results are
When the Reynolds numbers are larger than 180, a local shown in Fig. 12. Prediction with Eq. (12) gives the correlation
minimum of CD was observed at around L/D = 1. Apelt et al. coefficient of 0.972 and standard error of 0.028. In general, drag
(1973) found the local minimum CD around L/D = 1 in the larger coefficient is one of the main factors for an analysis of flow
Reynolds number experiments of 1.0 × 104~5.0 × 104 and Anderson around structure and the splitter can be used to reduce the drag
and Szewczyk (1997) for Re = 46,000. Hence, Re = 180 could be coefficient. Therefore, the effect of splitter plate can be evaluated
suggested as a critical Reynolds number for the existence of local easily with Eq. (12).
minimum CD at around L/D = 1. At this local minimum, CD
decreases to 82.97% and 81.72% of the plane cylinder without 3.4 Pressure Around Cylinder
splitter for Re = 200 and 240, respectively. Apelt et al. (1973) During structure design works, finding the variation of
reported that for L/D = 1 at Re = 1.0 × 104~ 5.0 × 104, the pressure and its maximum value is important for an appropriate
pressure drag can be reduced to 69% of the plain cylinder values. design, such as determining reinforce material at the location of
This implies that at higher Reynolds number, the effect of splitter high pressure. Fig. 13 shows the variations of the pressure
is more obvious in reducing CD. Beyond this ratio, CD increases coefficient, Cp, around the cylinder with respect to L/D and Re.
and then gradually decreases as L/D increases. Because of the cylindrical symmetry to θ = 180o, values for the
It is well known that beyond the critical spacing ratios, the upper half of the cylinder are only presented. Even though Cp
vortex shedding disappears and the flow field becomes symmetrical
about the splitter plate. The appearance of vortex shedding leads
to fluctuations of pressure behind the cylinder. Depending on the
behavior of vortex shedding, the difference of pressures between
front and real sides of the cylinder can be either large or small,
and this is responsible for the magnitude of pressure drag force.
When vortex shedding disappears, the flow becomes stable and
the drag force does not further decrease as well as alternating
fluctuation. Consequently, using splitter plate behind cylinder,
one can reduce the drag coefficient to maximum of 76, 78, 80,
83, 88, 91, and 96% of that of cylinder without splitter plate, at
Re = 180, 160, 140, 120, 100, 80, and 60, respectively. The drag
coefficients reduce to the minimum and drag force has no longer
fluctuation when no vortex occurs behind the cylinder. These
features corroborate the strong relationship between CD and the Fig. 13. Variation of Pressure Coefficient with L/D at Re = 180

Vol. 20, No. 6 / September 2016 − 2565 −

 Huy Cong Vu, Jungkyu Ahn, and Jin Hwan Hwang

varies with the Reynolds numbers, the trends of its distribution values, no vortex appears and the separation angle remains the
are similar to each other. Therefore, for the case of Re = 180 same. Grove et al. (1964) reported that the appearance or length
only, the profiles of Cp are presented and discussed. of the splitter plate did not have any noticeable effect on the
The splitter plate strongly affects the pressure distribution separation angle by laboratory experiments. However, in present
behind the cylinder (Fig. 13), whereas the pressure distribution in study, the influence of L/D on the separation angle was determined
the front of the cylinder, 0o ≤ θ ≤ 22o, is independent from L/D. although this effect is very small. This discrepancy may be due
The splitter plate increases the static pressure behind cylinder as to the difficulty in measuring small variations of angle when they
compared with the value for the plain cylinder and this can be conducted their experiments. Grove et al. (1964) also mentioned
attributed to the reduction of vortex strength. This feature is the difficulty in measuring the separation point and noted that the
completely consistent with the previous studies, even though error was about ±3o. In this study, small oscillations of the
experiments were performed with Re = 1.0 × 104~5.0 × 104 by separation angle could be detected with numerical simulations.
Apelt et al. (1973) and Re = 14,500 by Roshko (1955). Apelt et al. When flow separates from cylinder, the flow structure around the
(1973) found that the appearance of splitter plate behind cylinder separation point has been recognized as the origin of vortical
increased the base pressure (at θ = 180o) as much as 50%. instabilities in the wake behind cylinder (Wu et al., 2004).
Figure 13 shows that Cps are higher with the longer splitter
plate at θ = 180o. The increase of Cp at the rear side of cylinder 4. Conclusions
results in the decrease of CD, (Vu et al., 2015). This indeed
confirms the reduction of CD as shown in Fig. 11. The order of Flow around circular cylinder with a splitter plate was simulated
magnitude of increase in Cp for long splitters is very small, which with the diverse Reynolds numbers. The flow characteristics
explains small decreases of CD at these length ratios. Fig. 13 also such as the Strouhal number, drag and lift force coefficients,
shows that the location of minimum Cp moves front as L/D pressure coefficients, and location of separation angle were
increases. The minimum Cp are found at θ = 81o, 79.5o, 78o, and investigated with respect to the splitter plate length and the
76.5o when L/D = 0, 0.5, 3, and 5, respectively. Reynolds numbers. This study reached following conclusions.
1. The vortex shedding disappears when the splitter plate is
3.5 The Separation Angle longer than the critical value which was found to be propor-
The variation of separation point affects the shape of the wake tional to the Reynolds number, such as L/D = 1, 1.5, 3, 3.5,
boundary behind cylinder as it separates flow into two distinct 4, 4.5, and 5 for Re = 60, 80, 100, 120, 140, 160, and 180,
follow regions: the inner region with low pressure and the other respectively.
with higher pressure. 2. Vortex shedding affects the formation of drag and lift force
Figure 14 shows the effects of L/D on the separation angle, θs , significantly. The suppression of vortex shedding is accompa-
with the various Reynolds numbers. It was found that the nied by a significant reduction of coefficients. If the length
separation angle decreases as L/D increases. This trend was of splitter is long enough to wipe out vortex shedding, the drag
obvious at L/D < 1and at high Re. For Re = 180, the separation coefficient decreases to 76~96% of that of plain cylinder,
angle changes from θs = 112o to 109o when L/D increases from 0 depending on the Reynolds number. At the higher Reynolds
to 1, which means a net decrease of 3o. For Re = 60, the number, CD decreases more significantly. A functional rela-
separation angle varies between 122o and 121o, a net decrease of tionship between the drag force coefficient, L/D, and Re were
1o. Therefore, when L/D is less than 1, the larger Reynolds determined with high correlation. The lift force vanishes, CL =
number has more influence on θs . When L/D is larger than 1, the 0, if the vortex shedding is suppressed completely.
separation angle slightly decreases with the longer splitter plates 3. For short splitter plate, L/D < 1, the Strouhal number, drag
as far as L/D equals to the critical value. Beyond these critical and lift force coefficients decrease significantly. When L/D
> 1, depending on Reynolds number, CD, CL, and St change
with the different trends. 140 was found as a critical Rey-
nolds number for the Strouhal numbers change in the differ-
ent trends, and 180 was another critical value for the
variation of drag and lift force coefficients. The local mini-
mum values of CD, CL, and St were observed around L/D = 1
if Re > 180.
4. The variation of pressure was found in associated with the
drag force coefficient. This study also found the variation of
separation angle with respect to L/D and the Reynolds number.

Acknowledgements
Fig. 14. Variation of Separation Angle with the Length of Splitter
Plate (L/D) This research was supported by a grant (11-TI-C06) from

− 2566 − KSCE Journal of Civil Engineering

 Numerical Investigation of Flow Around Circular Cylinder with Splitter Plate

Advanced Water Management Research Program funded by Work, USA.

Ministry of Land, Infrastructure and Transport of Korean Gerrard, J. H. (1966). “The mechanics of the formation region of vortices
government. And also this research was a part of the project behind bluff bodies.” Journal of Fluid Mechanics, Vol. 25, No. 2,
pp. 401-413, DOI: 10.1017/S0022112066001721.
entitled “Development of integrated Keum River estuarine
Goswami, I., Scanlan, R. H., and Jones, N. P. (1992). “Vortex shedding
management system” funded by the Ministry of Ocean and from circular cylinders: Experimental data and a new model.”
Fisheries, Korea. Journal of Wind Engineering and Industrial Aerodynamics, Vol. 41,
Nos. 1-3, pp. 763-774, DOI: 10.1016/0167-6105(92)90495-V.
References Grove, A. S., Shair, F. H., Petersen, E. E., and Acrivos, A. (1964). “An
experimental investigation of the steady separated flow past a
Akilli, H., Krakus, C., Akar, A., Sahin, B., and Tumen, N. F. (2008). circular cylinder.” Journal of Fluid Mechanics, Vol. 19, No. 1,
“Control of vortex shedding of circular cylinder in shallow water pp. 60-80, DOI: 10.1017/S0022112064000544.
flow using an attached splitter plate.” Journal of Fluids Engineering, Harichandan, A. B. and Roy, A. (2010). “Numerical investigation of
Vol. 130, No. 4, pp. 041401.1-11, DOI: 10.1115/1.2903813. low Reynolds number flow past two and three circular cylinders
Anderson, A. and Szewczyk, A. A. (1997). “Effects of a splitter plate on using unstructured grid CFR scheme.” International Journal of Heat
the near wake of a circular cylinder in 2 and 3 dimensional Flow and Fluid Flow, Vol. 31, No. 2, pp. 154-171, DOI: 10.1016/
configuration.” Experiments in Fluids, Vol. 23, No. 2, pp. 161-174, j.ijheatfluidflow.2010.01.007.
DOI: 10.1007/s003480050098. Hwang, J. H., Yamazaki, H., and Rehmann, C. R. (2006). “Buoyancy
ANSIS (2010). ANSIS Fluent theory guide, Canonsburg, Pennsylvania, generated turbulence in stably stratified flow with shear.” Physics of
USA. Fluids, Vol 18, No. 4, pp. 045104:1-13, DOI: 10.1063/1.2193472.
Apelt, C. J., West, G. S., and Szewczyk, A. A. (1973). “The effects of Hwang, J. Y. and Yang, K. S. (2007). “Drag reduction on a circular cylinder
wake splitter plates on the flow past a circular cylinder in the range using dual detached splitter-plates.” Journal of Wind Engineering
10 < Re < 5×10 .” Journal of Fluid Mechanics, Vol. 61, No. 1,
4 4
and Industrial Aerodynamics, Vol. 95, No. 7, pp. 551-564, DOI:
pp. 187-198, DOI: 10.1017/S0022112073000649. 10.1016/j.jweia.2006.11.003.
Apelt, C. J. and West, G. S. (1975). “The effects of wake splitter plates Jain, P. C. and Goel, B. S. (1976). “Shedding of vortices behind a
on bluff-body flow in the range 10 < Re < 5×10 . Part 2.” Journal of
4 4
circular cylinder.” Computer and Fluids, Vol. 4, Nos. 3-4, pp. 137-
Fluid Mechanics, Vol. 71, No. 1, pp. 145-160, DOI: 10.1017/ 142, DOI: 10.1016/0045-7930(76)90002-5.
S0022112075002479. Kumar, B. and Mittal, S. (2006). “Prediction of the critical Reynolds
Bandyopadhyay, P. R. (1986). “Aspects of the equilibrium puff in number for flow past a circular cylinder.” Computer Methods in
transitional pipe flow.” Journal of Fluid Mechanics, Vol. 163, Applied Mechanics and Engineering, Vol. 195, Nos. 44-47,
pp. 439-458, DOI: 10.1017/S0022112086002379. pp. 6046-6058, DOI: 10.1016/j.cma.2005.10.009.
Behzad, G. D. and Hamed, H. J. (2010). “On the suppression of vortex Kundu, P. K. (1990). Fluid mechanics, 2 ed. Academic Press. San
nd

shedding from circular cylinders using detached short splitter- Diego, pp. 319-332.
plates.” Journal of Fluids Engineering, Vol. 132, No. 4, (044501). Kwai, H. (1990). “Discrete vortex simulation for flow around a circular
DOI: 10.1115/1.4001384. cylinder with a splitter plate.” Journal of Wind Engineering and
Byko, M. (2002). “Materials give roller coaster enthusiasts a reason to Industrial Aerodynamics, Vol. 33, Nos. 1-2, pp. 153-160, DOI:
scream.” Journal of the Minerals, Metals, & Material Society, 10.1016/0167-6105(90)90031-7.
Vol. 54, No. 5, pp. 16-20, TMS Publications. Kwon, K. K. and Choi, H. (1996). “Control of laminar vortex shedding
Chang, Y. S., Hwang, J. H., and Park, Y. G. (2015). “Numerical simulation behind a circular cylinder using splitter plates.” Physics of Fluid,
of sediment particles released at the edge of the viscous sublayer in Vol. 8, No. 2, pp. 479-486, DOI: 10.1063/1.868801.
steady and oscillating turbulent boundary layers.” Journal of Hydro- Lam, K., Gong, W. Q., and So, R. M. C. (2008). “Numerical simulation
environment Research, Vol. 9, No. 1, pp. 36-48, DOI: 10.1016/ of cross flow around four cylinders in–line square configuration.”
j.jher.2013.07.002. Journal of Fluids and Structures, Vol. 24, No. 1, pp. 34-57, DOI:
Cimbala, J. M. and Garg, S. (1990). “Flow in the wake of a freely 10.1016/j.jfluidstructs.2007.06.003.
rotatable cylinder with splitter plate.” AIAA Journal, Vol. 29, No. 6, Liu, C., Zheng, X., and Sung, C. H. (1998). “Preconditioned multigrid
pp. 1001-1003, DOI: 10.2514/3.10692. methods for unsteady incompressible flow.” Journal computational
Dehkordi, B. R. and Jafari, H. H. (2010). “On the suppression of vortex Physics, Vol. 139, No. 1, pp. 35-57, DOI: 10.1006/jcph.1997.5859.
shedding from circular cylinders using detached short splitter – Lixia, Q., Norberg, C., Davidson, L., Peng, S. H., and Wang, F. (2013).
plates.” Journal of Fluids Engineering, Vol. 132, No. 04, pp. 044501- “Quantitative numerical analysis of flow past a circular cylinder at
1:4, DOI: 10.1115/1.4001384. Reynolds number between 50 and 200.” Journal of Fluids and
Ding, H., Shu, C., Yeo, K. S., and Xu, D. (2007). “Numerical simulation Structures, Vol. 39, pp. 347-370, DOI: 10.1016/j.jfluidstructs.2013.
of flows around two circular cylinders by mesh-free least square 02.007.
based finite difference methods.” International Journal for numerical Mahir, N. and Altac, Z. (2008). “Numerical investigation of convective
method in fluids, Vol. 53, No. 2, pp. 305-332, DOI: 10.1002/fld.1281. heat transfer in unsteady flow past two cylinders in tandem
Fey, U., Konig, M., and Eckelman, H. (1998). “A new Strouhal–Reynolds- arrangements.” International Journal of Heat and Fluid flow, Vol. 29,
number relationship for the circular cylinder in the range 47 < Re < No. 5, pp. 1309-1318, DOI: 10.1016/j.ijheatfluidflow.2008.05.001.
2×10 .” Physics of Fluid, Vol. 10, No. 7, pp. 1547-1549, DOI:
5
Meneghini, J. R. and Saltara, F. (2001). “Numerical simulation of flow
10.1063/1.869675. interference between two circular cylinders in tandem and side by
Fox, R. W., Mcdonald, A. T., and Pritchard, P. J. (2008). Introduction to side arrangements.” Journal of Fluids and Structure, Vol. 15, No. 2,
fluid mechanics (SI version), 7th ed., John Wiley & Sons, New pp. 327-350, DOI: 10.1006/jfls.2000.0343.

Vol. 20, No. 6 / September 2016 − 2567 −

 Huy Cong Vu, Jungkyu Ahn, and Jin Hwan Hwang

Mittal, S. and Raghuvanshi, A. (2001). “Control of vortex shedding behind ACTA MechanicaSinica, Vol. 31, pp.100-105.
circular cylinder for flows at low Reynolds numbers.” International Sumner, D. (2010). “Two circular cylinders in cross-flow: A review.”
Journal for numerical methods in fluids, Vol. 35, No. 04, pp. 421- Journal of Fluids and Structures, Vol. 26, No. 6, pp. 849-899, DOI:
447, DOI: 10.1002/1097-0363(20010228)35:4<421::AID-FLD100> 10.1016/j.jfluidstructs.2010.07.001.
3.0.CO;2-M. Surmas, R., Santos, O. E., and Philippi, P. C. (2004). “Lattice Boltzmann
Mukundan, H. (2008). Vortex-induced vibration of marine risers: simulation of the flow interference in bluff body wakes.” Future
motion and force reconstruction from field and experimental data, Generation Computer System, Vol. 20, No. 6, pp. 951-958, DOI:
PhD Thesis, Massachusetts Institute of Technology. Dept. of 10.1016/j.future.2003.12.007.
Mechanical engineering. Taamneh, Y. (2011). “CFD simulations of drag and separation flow
Noberg, C. (2003). “Fluctuating lift on a circular cylinder: review and around ellipsoids.” Jordan Journal of Mechanical and Industrial
new measurements.” Journal of Fluids and Structures, Vol. 17, Engineering, Vol. 5, No. 2, pp. 12-132.
No. 1, pp 57-96, DOI: 10.1016/S0889-9746(02)00099-3. Tritton, D. J. (1959). “Experiments on the flow past a circular at low
Ozono, S. (1999). “Flow control of vortex shedding by a short splitter Reynolds numbers.” Journal of Fluid Mechanics, Vol. 6, No. 4,
plate asymmetrically arranged downstream of a cylinder.” Physics pp. 547-567, DOI: 10.1017/S0022112059000829.
of Fluids, Vol. 11, No. 10, pp. 2928-2934, DOI: 10.1063/1.870151. Vu, H. C. and Hwang, J. H. (2014). Flow separation around single
Park, J., Kwon, K., and Choi, H. (1998). “Numerical solution of flow circular cylinder at low Reynolds numbers, 40 KSCE Conferences,
th

past a circular cylinder at Reynolds numbers up to 160.” KSME pp 1237-1238.

International Journal, Vol. 12, pp. 1200-1205, DOI: 10.1007/ Vu, H. C., Ahn, J., and Hwang, J. H. (2015). Numerical simulation of
BF02942594. flow past two circular cylinders in tandem and side-by-side arrangement
Ribeiro, P. A. R., Schettini, E. B. C., and Silvestrini, J. H. (2002). “Bluff- at low Reynolds numbers, KSCE Journal of Civil Engineering,
bodies vortex shedding suppression by direct numerical simulation.” (published online), DOI: 10.1007/s12205-015-0602-y.
Brazilian congress of thermal engineering and sciences, CIT02- Wanderley, J. B. V. and Levi, C. A. (2002). “Validation of finite difference
0863. method for the simulation of vortex-induced vibration on a circular
Roshko, A. (1955). “On the wake and drag of bluff bodies.”Journal of cylinder.” Ocean Engineering, Vol. 29, No. 4, pp. 445-460, DOI:
the Aeronautical Sciences (Institute of the Aeronautical Sciences), 10.1016/S0029-8018(01)00014-2.
Vol. 22, No. 2, pp. 124-132, DOI: 10.2514/8.3286. White, B. L. and Nepf, H. M. (2003). “Scalar transport in random
Sa, J. Y. and Chang, K. S. (1991). “Shedding patterns of the near-wake cylinder arrays at moderate Reynolds number.” Journal of Fluid
vortices behind a circular cylinder.” Internal Journal for Numerical Mechanics, Vol. 487, pp. 43-79, DOI: 10.1017/S0022112003004579.
methods in Fluids, Vol. 12, No. 5, pp. 463-474, DOI: 10.1002/ Williamson, C. H. K. (1989). “Oblique and parallel modes of vortex
fld.1650120504. shedding in the wake of a circular cylinder at low Reynolds numbers.”
Seo, I. W. and Song, C. G. (2012). “Numerical simulation of laminar Journal of Fluid Mechanics, Vol. 206, pp. 579-627, DOI: 10.1017/
flow past a circular cylinder with slip conditions.” International S0022112089002429.
Journal for Numerical Methods in Fluids, Vol. 68, No. 12, pp. 1538- Williamson, C. H. K. (1996). “Vortex dynamics in cylinder wake.” Journal
1560, DOI: 10.1002/fld.2542. of Fluid Mechanics, Vol. 28, No. 1, pp. 477-526, DOI: 10.1146/
Sharman, B., Lien, F. S., Davidson, L., and Norberg, C. (2005). “Numerical annurev.fl.28.010196.002401.
predictions of low Reynolds number flow over two tandem circular Wu, M. H., Wen, C. Y., Yen, R. H., Weng, M. C., and Wang, A. B.
cylinders.” International Journal for numerical Method in Fluids, (2004). “Experimental and numerical study of the separation angle
Vol. 47, No. 5, pp. 423-447, DOI: 10.1002/fld.812. for flow around a circular at low Reynolds number.” Journal of Fluid
Strykowski, P. J. and Sreenivasan, K. R. (1990). “On the formation and Mechanics, Vol. 515, pp. 233-260, DOI: 10.1017/S0022112004000436.
suppression of vortex ‘shedding’ at low Reynolds numbers.” Zhang, J. and Dalton, C. (1998). “A three-dimensional simulation of a
Journal of Fluid Mechanics, Vol. 218, pp. 71-107, DOI: 10.1017/ steady approach ow past a circular cylinder at low Reynolds number.”
S0022112090000933. International Journal of Numerical Methods in Fluids, Vol. 26,
Su, M. and Kang, Q. (1999). “Large eddy simulation of the turbulent No. 9, pp. 1003-1022, DOI: 10.1002/(SICI)1097-0363 (19980515)
flow around a circular cylinder at sub-critical Reynolds numbers.” 26:9 <1003::AID-FLD611>3.0.CO;2-W.

− 2568 − KSCE Journal of Civil Engineering