INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS

Int. J. Numer. Meth. Fluids 2012; 68:1538–1560

Published online 17 March 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/fld.2542

Numerical simulation of laminar flow past a circular cylinder

with slip conditions

Il Won Seo∗, †, ‡ and Chang Geun Song

Department of Civil and Environmental Engineering, Seoul National University, 599 Gwanak-ro, Gwanak-gu,
Seoul 151-744, South Korea

SUMMARY
The no-slip condition is an assumption that cannot be derived from first principles and a growing number
of literatures replace the no-slip condition with partial-slip condition, or Navier-slip condition. In this study,
the influence of partial-slip boundary conditions on the laminar flow properties past a circular cylinder
was examined. Shallow-water equations are solved by using the finite element method accommodating
SU/PG scheme. Four Reynolds numbers (20, 40, 80, and 100) and six slip lengths were considered in
the numerical simulation to investigate the effects of slip length and Reynolds number on characteristic
parameters such as wall vorticity, drag coefficient, separation angle, wake length, velocity distributions on
and behind the cylinder, lift coefficient, and Strouhal number. The simulation results revealed that as the
slip length increases, the drag coefficient decreases since the frictional component of drag is reduced, and
the shear layer developed along the cylinder surface tends to push the separation point away toward the
rear stagnation point so that it has larger separation angle than that of the no-slip condition. The length
of the wake bubble zone was shortened by the combined effects of the reduced wall vorticity and wall
shear stress which caused a shift of the reattachment point closer to the cylinder. The frequency of the
asymmetrical vortex formation with partial slip velocity was increased due to the intrinsic inertial effect
of the Navier-slip condition. Copyright 䉷 2011 John Wiley & Sons, Ltd.

Received 6 September 2010; Revised 19 November 2010; Accepted 11 January 2011

KEY WORDS: Navier-slip condition; laminar flow; shallow-water equations; SU/PG; slip length; circular
cylinder

1. INTRODUCTION

The unsteady viscous flow past a cylinder has been a major research topic and remains one
of the most challenging problems in the fluid mechanics and hydraulic engineering fields. The
information of the flow quantities past a cylinder body is of important value in understanding
flow physics and practical applications such as the pier scour problem and the suppression of
flow-induced failure or vortex-induced vibration of structures. In addition, it is a good bench mark
problem for new flow modeling codes and provides a prestudy to the classical problem of fluvial
hydraulics.
Numerical simulation of flow past a cylinder involves the specification of appropriate boundary
conditions to close the problem. The boundary condition is a major factor affecting the flow pattern
near the solid boundary and the formation of recirculation zones because the magnitude of flow

∗ Correspondence to: Il Won Seo, Department of Civil and Environmental Engineering, Seoul National University,
599 Gwanak-ro, Gwanak-gu, Seoul 151-744, South Korea.
† E-mail: seoilwon@snu.ac.kr
‡ Professor.

Copyright 䉷 2011 John Wiley & Sons, Ltd.

 NUMERICAL SIMULATION OF LAMINAR FLOW 1539

separation at a solid body can modify the wall shear stress distribution or initiate instability toward
turbulence [1]. Therefore, a proper understanding of the influence of physical boundaries is of
practical importance since it requires some assumptions about the nature of the fluid motion at
the solid interface. From a physical point of view, the fluid cannot penetrate the solid and thus
its normal velocity is naturally equal to zero. This is the condition of non-penetration. On the
contrary, the absence of slip is not very intuitive and the validity of no-slip condition stating zero
relative velocity between the fluid and solid at the interface has long been debated since the early
days of fluid mechanics. Owing to the rapid progress of current measurement techniques, nonzero
slip velocities at the solid wall has been intensively reported since the average of the bulk velocity
of incoming and reflecting particles is not zero. However, most of the previous works on fluid
past a bluff body assumed either no-slip or zero shear stress, i.e. free-slip boundary conditions.
Recently, a growing number of literatures [2–6] use different boundary conditions concerning the
relative slip motion between the fluid and the solid surface. Such a slip can be accounted for by the
Navier-slip boundary condition, whereby the slip velocity is proportional to the tangential viscous
stress. Actually, fluid slipping at the solid surface has been observed in many phenomena, such as
low-viscosity fluids, flows past chemically reacting walls, large length scale flows, high shear rate
flows and submerged physical materials with small imperfections as well as microflows, where the
classical no-slip condition is no longer valid. Craig et al. [7] observed the occurrence of the partial
boundary slip by direct measurements of hydrodynamic drainage forces and maintained that slip
length is a function of both surface approach velocity and fluid viscosity. The practical motivation
to reduce the drag and the pressure drop is so significant since the shear caused by the difference
between the free-stream velocity and the zero boundary velocity results in drag force, which is
inevitable in any fluid flow [8]. The Navier-slip boundary condition can also be applied in these
types of drag reduction investigations. For example, it has been reported that the slip velocity on
a hydrophobic surface results in a significant drag reduction [9]. Min and Kim [10] suggested that
reduction of skin-friction drag can be achieved in large-scale flows with a hydrophobic surface
treatment. A few other studies regarding slip boundary application to a cylinder are Keh and Wang
[11] and Wang [12]. Keh and Wang [11] obtained solutions for the slow motion of a long circular
cylindrical particle with a slip surface in an arbitrary transverse direction near a large plane wall
and computed the hydrodynamic force and torque acting on the cylinder. Wang [12] studied the
axisymmetric stagnation slip flow on a translating cylinder and the effect of slip factor. However,
previous studies on the Navier-slip condition were restricted to the investigation of force reduction
and slip length determination, and thus the separation phenomena and velocity structure around
the cylinder have not been fully examined.
To conduct an analysis of an unsteady flow past a cylinder experimentally is not easy because
the variation of the approaching flow is difficult to accurately control. In contrast, the application
of numerical simulations does not suffer from such technical difficulties [13]. In this study, a
numerical study was performed to investigate the unsteady flow past a cylinder. The finite element
code was developed based on a two-dimensional shallow-water equations.
In this study, we examine the effects of partial-slip boundary conditions on the laminar flow prop-
erties past a circular cylinder including characteristic parameters such as drag and lift coefficient,
the separation phenomena and velocity distributions on and behind the surface.

2. THEORETICAL BACKGROUND

2.1. Shallow-water flow modeling

Mathematical modeling of hydrodynamics in natural water systems is based on the formulation
and solution of the appropriate equations of continuity and motion of water. Most open-channel
flows can be regarded as shallow-water problems, because the effect of vertical motions is usually
insignificant and depth-averaged equations are generally accepted for analyzing the open channel
flows with reasonable accuracy and efficiency. The shallow-water equations are obtained by aver-
aging the Navier–Stokes equations with respect to time and integrating these over the flow depth

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1540 I. W. SEO AND C. G. SONG

with some assumptions: (1) vertical velocity and acceleration are negligible; (2) pressure distri-
bution in the vertical direction is hydrostatic; (3) depth-averaged values are sufficient to describe
properties that vary over flow fields; and (4) the bottom slopes are small in longitudinal and
transverse directions. The solution of shallow-water equations is of considerable importance for
a variety of practical flow problems in water resources engineering. Numerical computation of
water levels and mean velocities by shallow-water approximation dates back to the 1960s. During
the first period (1960–1970), mathematical models based on finite differences under regular grids
were developed while models based on finite element techniques and irregular grids have been
appearing in the literature since Grotkop [14]. Recently, finite-volume methods [15], Godunov-
type methods [16], lattice Boltzmann methods [17], boundary element methods [18], and spectral
element methods [19] have been successfully applied. However, finite element methods are still
popular from an implementation perspective for a number of reasons: grids are easily refined in
shallow areas or in areas with changing topography; irregular boundaries can be described more
accurately by using triangular or curved elements; Neumann boundary conditions enter the weak
statement of the problem naturally [20].
Physically, as a flow passes a cylinder with a blunt shape, separation generally occurs due to
the effect of negative pressure gradients. The separation lines can roll and form the so-called
Karman vortices. Consequently, the flow around the body always behaves unsteadily, even when
the approaching flow is steady and uniform [13]. The bottom friction and the vertical extent of the
water depth play a significant role in the dynamics of shallow wakes [21]. In particular, the bottom
friction affects the stability of the wake similar to viscosity in laminar flow [22]. For shallow-water
flows dominated by bottom friction, the regimes of the unsteady wake are interpreted in terms
of a stability parameter S = c f D/ h, in which c f is the friction coefficient at the bed, D is the
cylinder diameter, and h is the water depth [23]. The wake momentum defect is reduced by the
bed friction force, which will eventually nullify the wake in the far region. Chen and Jirka [24]
have classified three forms of unsteady wake patterns: (1) the vortex street wake with unsteady
separation at the cylinder for S0.2; (2) the unsteady bubble wake with a re-circulating bubble
attached to the cylinder that becomes unstable with sinuous oscillations further downstream for
0.2S0.5; and (3) the steady bubblewake that shows no oscillations for S0.5. Therefore, when-
ever the shallow wake parameter S becomes large, bottom friction suppresses the vortex shedding
process and any subsequent transverse instabilities so that the classical Kármán vortex street is
not formed [25]. Linear stability analysis of the depth-averaged shallow-water equations with bed
friction using a time-averaged transverse near-wake velocity profile as input has shown that vortex
shedding corresponds to an absolute instability, where temporal/spatial evolution of a disturbance
occurs, and a steady bubble corresponds to a convective instability, where only spatial evolution
occurs [26].
In this study, the stability parameter is less than 0.1 and thus vortex street-like wake is expected
to evolve. In addition, applied Reynolds numbers based on cylinder diameter range from 20 to
100 and the corresponding Reynolds numbers based on height of the shallow layer are between 45
and 227. To avoid the additional complexities associated with turbulence, results for predominant
laminar flow are presented in this study. Under the prescribed stability parameter and Reynolds
number, the shallow layer approaching the cylinder is laminar and fully developed so that both
the inflow and near wake are stable. In the case of the laminar flow past a circular cylinder with
Reynolds number less than 160, two-dimensionality prevails over the domain, according to the
experimental measurements by Nishioka and Sato [27] and Mansy et al. [28] and to the conclusive
analysis given by Williamson [29]. To describe the velocity field and vortex structure around a
cylinder, the depth-averaged shallow flow model has been widely adopted and successfully applied
in many researches [30–35]. The shallow-water equations are thereby used to describe the flow
properties around a cylinder in this study.

2.2. Slip length

In 1823, Navier [36] proposed a general boundary condition which states that the tangential
component of the velocity at the surface is proportional to the rate of strain at the surface. Using

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1541

u∞

Separation point
Mean recirculation region
2
1

ition
lip cond Wake bubble
ier-s
Nav
Front stagnation point Rear stagnation point Reattatchment point

Figure 1. Navier-slip condition and slip length (b).

Einstein’s summation convention, the tangential stress at the surface is expressed as

 
b
ti i j n j +u si = 0 (1)


where ti and n j are the unit tangential and normal vectors to the boundary, respectively;  is the
kinematic viscosity; i j is the stress tensor; u si is the velocity components at the boundary; and b
is the slip length defined by the corresponding distance extrapolated from nonzero to zero velocity
in the direction normal to the wall (Figure 1). A higher slip length means a larger slip and lower
friction at the boundary. Both free-slip and no-slip boundary conditions are special cases of the
Navier-slip boundary condition; a zero-valued slip length corresponds to a no-slip condition while
an infinite value yields the free-slip condition.
Considering the characteristic size of a computational domain, boundaries could be considered
as a source of disturbance, and the main issue is to what extent the boundaries may affect the flow.
This is of particular relevance in numerical computations, where the small variation of boundaries
is difficult and costly to capture by the discretization schemes [37]. A numerical simulation of the
flow problems in the presence of irregular boundaries is difficult since it requires many mesh nodes
and handling of many data. For computational purposes, an artificial smooth boundary, close to
the original one, is usually taken and the equations are solved in the new domain. It is clear that
the non-penetration condition should be retained at the new boundary, but there are no reasons
to keep the no-slip [38]. In order to deal with this problem, the Navier-slip boundary condition
has been applied to replace the classical boundary condition prescribed on the idealized artificial
boundary placed inside the physical domain. Physically, a nonzero slip length arises from unequal
wall and fluid densities, a weak wall–fluid interaction, and a high temperature [39]. Therefore,
simulations that involve low-viscosity fluids, large length scale [40], high shear rate flow [41] and
the hydrophobic coated surface to reduce drag force [8, 42] as well as microflow often call for the
application of the Navier-slip boundary condition at the solid boundary.

3. MATHEMATICAL MODEL

3.1. Governing equations

The present mathematical model of a free surface flow consists of the 2D shallow-water equations,
which are obtained by averaging momentum and mass balance equations along the vertical direction
under barotropic and hydrostatic assumptions. Surface shear stress due to wind is ignored in this
study since it is important only if the wind speed exceeds 100 times the flow velocity, which
will not often be the case [43]. To consider the bed friction, the shear stress at the bottom
is modeled using the quadratic friction law based on Manning’s bottom roughness, depth, and

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1542 I. W. SEO AND C. G. SONG

gravitational acceleration. Then, the set of variables u i (x, t) and h(x, t) is a solution of the following
equations:
  √
*u i *u i *(H +h) 1 * *u i 2 ui u j u j
+u j = −g + h − gn (2)
*t *x j *x i h *x j *x j h 4/3

*h *u j *h
+h +u j =0 (3)
*t *x j *x j
where t is the time; u 1 , u 2 is the vertically averaged velocity components corresponding to the
x-, y-directions, respectively; g is the acceleration of gravity; H is the bottom elevation; h is the
flow depth;  is the kinematic viscosity; and n is Manning’s roughness coefficient. The above
equations are written using the indicial notation and the usual summation convention for repeated
indices. The unknowns, u i and h, are defined in the bounded domain  ⊂ R N with N (1 or 2), the
dimension of the physical space, and for a time interval [0, T ].
In order for the boundary conditions to supplement Equations (2) and (3), the velocity is
postulated at the inflow boundary (*in ) and the flow depth at the outflow boundary (*out ) as
the essential condition; u i |*in = û i and h|*out = ĥ, where overhat refers to prescribed value. In
this study, the flow symmetry condition is specified at the lateral boundaries and the Navier-slip
boundary condition is imposed at the solid surface, which was discussed in detail in the previous
section. Together with the boundary conditions, the initial condition is prescribed as the hydrostatic
condition, (u i (x, 0) = 0, h(x, 0) = constant) in order to initiate the solution set.

3.2. Navier-slip condition

In this study, Equation (1) together with the following complementary equation, which represents
no flow through a solid wall, are used to impose the slip wall boundary condition
u si n i = 0 (4)
The normal and tangential unit vectors can be related through the following equations:
ty = n x , tx = −n y (5)
After recasting Equations (1) and (4) in terms of their components and making use of Equation (5),
the slip-wall velocity components at the solid boundary are found as [44]:
     
*u 1 *u 2 *u 2 *u 1
u s1 = −bn y 2 − nx n y + + (n 2y −n 2x ) (6)
*x *y *x *y
     
*u 1 *u 2 *u 2 *u 1 2 2
u s2 = bn x 2 − nx n y + + (n y −n x ) (7)
*x *y *x *y
In practice, the free-slip boundary condition is imposed on the lateral boundary and the no-slip
or Navier-slip condition is imposed on the other boundary because the no-slip boundary condition
on the side walls would maintain the fixed nature of the nodes of the interface on the side walls,
which would lead to large distortions of the mesh [45].

3.3. Numerical formulation

In the computations that follow, the finite element method was employed for the spatial discretiza-
tion and the flow field  to be analyzed was divided into the subdomain e , e = 1, 2, . . . , n el , where
n el is the number of elements. The momentum equation was discretized using the SU/PG scheme
and the continuity equation using the Galerkin method. After multiplying shallow-water equations
by the weighting function wk ∈ H 1 (k = 1, 2, 3 for linear triangular element and k = 1, 2, 3, 4 for
bilinear quadrilateral element) and integrating over , we applied Green’s theorem to the viscous
stress terms in the momentum equation to arrive at weak formulation, where H 1 denotes the

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1543

Sobolev space of vector functions defined on the spatial domain . The problem can then be stated
as follows:
Find u i (x, t) and h(x, t) ∈ H 1 such that ∀wk and Nk ∈ H 1
   √  
*u i *u i *(H +h) 2 ui u j u j *wk *u i
wk +u j +g + gn + d = 0 (8)
 *t *x j *x i h 4/3 *x j *x j
  
*h *u j *h
Nk +h +u j d = 0 (9)
 *t *x j *x j

where i = 1, 2, k = 1, 2, 3 or 1, 2, 3, 4 depending on the element shape; and j follows Einstein

summation convention. Line integrals in Equation (8) resulting from integration by part have
been omitted since it is vanished for every side of an element as well as those boundaries where
Dirichlet conditions are imposed [46]. The finite element approximation of Equation (8) and (9)
involves finding an approximate solution in the finite-dimensional subspace H h of the space H 1 .
In this study, H h represent the piecewise bilinear finite element function spaces of the form u hi =
n j n
j =1 u i (t)N (x) and h =
j h j j j
j =1 h (t)N (x), where N (x) is the shape function. Conventional
Galerkin discretization involving piecewise linear interpolation results in the central difference
approximation so that spurious wiggles often appear in convection-dominated Galerkin finite
element solutions. One of the preferred methods to preclude this problem is the SU/PG scheme
achieved by adding a term to the Galerkin formulation, with each term in the form of an integral
over a different element. Thus, SU/PG can be viewed as a modification to the classical Galerkin
method. The physical idea is to increase control over the advective-derivative term by adding an
artificial diffusion term which acts only in streamline direction. This control can be introduced
within a weighted residual format, thus maintaining consistency [47]. With these statements, the
above problem can be recast as:
Find u hi (x, t) and h h (x, t) ∈ H h such that Nkh ∈ H h

⎡ ⎛  ⎞ ⎤
 u hi u hj u hj
*u h *u h *(H +h h ) *N h *u h
⎣ Nkh ⎝ i +u hj i +g + gn 2 ⎠ + k i ⎦
d
 *t *x j *x i (h h )4/3 *x j *x j
⎛  ⎞
 u hi u hj u hj

n el *u h *u h *(H +h h )
+ pkh ⎝ i +u hj i +g + gn 2 ⎠ de = 0 (10)
e=1 e *t *x j *x i (h h )4/3

  
*h h *u hj *h h
Nkh +h h
+u hj d = 0 (11)
 *t *x j *x j

where the perturbation weighting function is defined by

 
h̄ h *Nk
pkh = uj
2uh  *x j

where uh  is the Euclid norm of velocity; h̄ is the element characteristic length [48];  =
coth(/2)−2/ is the quadrature points; and  = uh h̄/ is the element Reynolds number.
Formulations (10) and (11) have been time-discretized with a fully implicit method. There are
no restrictive compatibility conditions, such as the LBB condition [49–51] on the discrete spaces;
thus, piecewise bilinear interpolations are used for all fields in the computations reported here and
three points quadrature is employed for numerical integration. Bilinear quadrilateral elements with

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1544 I. W. SEO AND C. G. SONG

non-orthogonal edges are transformed to straight-sided orthogonal elements by introducing the

natural coordinates  = (x, y) and  = (x, y). The shape functions Nk in the natural coordinate
system are defined as

(1+k )(1+k )
Nk = , k = 1, 2, 3, 4 (12)
4

where k and k are corner nodes of the square stretching from (−1, −1) to (1, 1).
The non-linear equation system resulting from the finite element discretization of the flow
equations is solved using the Newton–Raphson method and the linear set of equations is solved
by the frontal method. At any time step, iteration is repeated until the difference of flow properties
between two iterations is smaller than the preset tolerances.

4. NUMERICAL SIMULATIONS

4.1. Simulation conditions

In this study, the developed numerical model was applied to the laminar flow problem around a
circular cylinder with varying slip length conditions. Table I summarizes the simulation conditions
considered in this study. The upstream and downstream boundary conditions were assigned as
freestream velocity (u ∞ ) and depth (h ∞ ), respectively, with a constant cylinder diameter (D =
0.22 m). The Reynolds number, Re, is associated with the diameter of the cylinder, freestream
velocity and viscosity of the fluid so that constant kinematic viscosity was adopted for each
of the four cases to facilitate the Reynolds number assignments. Steady simulations with four
Reynolds numbers and six slip lengths were performed to investigate the effect of slip length and
Reynolds number on time-averaged parameters such as wall vorticity, drag coefficient, separation
angle, wake length and velocity distributions. Slip length was chosen to be as large as possible
so that it could produce distinguishable differences in flow behaviors while maintaining stable
computation. Unsteady simulations were intended to determine lift coefficient, Strouhal number,
and vortex formation in which the time domain must be considered. In addition, a time step of
tu ∞ /D = 0.0227, providing a resolution of approximately 280 time steps per vortex shedding
period, was selected. The relative tolerance limit of 10−4 has been used.
If the meshes near the cylinder wall are too coarse, the prediction of the Strouhal number and
drag coefficient might be low, as reported by Beaudan and Moin [52]. Therefore, to ensure the
mesh independency, computations at Re = 100 were performed and mesh sensitivity was tested in
terms of the drag coefficient and Strouhal number. The mesh independency was achieved at 180
points around a cylinder wall with 14 143 nodes and 14 778 elements in total, which was used
for the detailed analysis that follows. The dimension of the computational domain was chosen
such that the lateral wall and upstream/downstream boundaries had no effect on the computational
results. The analysis of Behr et al. [53] revealed that a distance from the lateral wall to the cylinder
of 16D was sufficient. In addition, the distances of upstream and downstream boundaries from
the cylinder placed at 8D and 22.5D, respectively, were sufficient based on the computational
results by Tezduyar and Shih [54] and Behr et al. [55]. The computational domain reflecting these
conditions is displayed in Figure 2.

Table I. Simulation conditions.

Attribute u ∞ (m/s) h ∞ (m) D (m) Re b/D(×10−3 )
Steady solver 0.1 0.5 0.22 20, 40, 80, 100 0, 4.5, 9.0
13.5, 18.0, ∞
Unsteady solver 80, 100 0, 13.5

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1545

16D

D
8D 22.5D

Figure 2. Computational domain.

4.2. Velocity and depth distributions

Simulation results of velocity structures and distributions around the cylinder wall for Re = 100 are
plotted in Figures 3 and 4. Figure 3 shows contours of longitudinal velocity. Owing to the velocity
constraint around the cylinder, separation occurred behind the cylinder at the b/D = 0.0 and 0.018
cases. If the slip length, b is equal to infinity, it corresponds to the free-slip condition and vortex
is not evolved since the boundary layer separation did not occur, and thus in the case of b/D = ∞,
attenuation of velocity behind the cylinder was observed but reverse velocity had not evolved. The
result by b/D = 0.018 had a higher velocity gradient along the longitudinal direction than that of
b/D = 0.0. In the cases of b/D = 0.0 and 0.018, the maximum velocities occurred as passing beside
the cylinder while for b/D = ∞ maximum velocity occurred on the walls perpendicular to the flow
direction. There was a considerable difference between the maximum velocities; the maximum
velocity by b/D = ∞ was about 50% higher than that of b/D = 0.0 and 0.018. The contour of
transverse velocity is presented in Figure 4. In contrast to the case of b/D = ∞ (Figure 4(c)), the
direction of transverse velocity was reversed about 3D away from the center of the cylinder in
b/D = 0.0 and 0.018. More detailed analysis is given as follows.
The longitudinal and transverse velocity distributions at the cylinder wall for the case of Re = 100
are plotted in Figure 5. When the free-slip condition was applied, the maximum velocity was
observed around the cylinder surface tangent to the flow direction and the minimum velocity was
observed around the cylinder surface normal to the flow direction. Thus, the largest and smallest
velocities were formed on the circumference of the cylinder surface and solid lines denoting the
free-slip condition showed this behavior. However, if the partial slip condition was applied, the
velocity distribution in the whole computational domain was completely changed due to the shear
layer separation. By the slip length impositions of b/D = 0.009 and 0.018, the maximum velocities
that occurred along the cylinder surface showed 10 and 20% values of freestream velocity and the
locations of these points were rotated 20◦ and 12◦ toward the front stagnation point for longitudinal
and transverse velocities, respectively.
The mean velocity distributions along the spanwise sections behind the cylinder at Re = 100 are
illustrated in Figures 6 and 7. The x axis was taken in the flow direction and the y axis was taken
perpendicular to the flow direction, with the origin at the center of the cylinder. In Figure 6, the
part of the curves below the x axis in the cases of x/D = 2.0 and 5.0 corresponded to the returning
flow in the wake region and thus showed the presence of eddies. The distribution showed a velocity
overshoot between |y/D| = 1 and 5 due to the acceleration at the shoulder of the cylinder so that
the velocity deficit due to the wake effect was clearly obtained. The result by b/D = 0.018 had
relatively large minimum velocity along each spanwise section compared with that of b/D = 0.0
while maximum velocity showed opposite trend. In Figure 7, the direction of transverse velocity

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1546 I. W. SEO AND C. G. SONG

(a) (b)

(c)

Figure 3. Iso-velocity contours of longitudinal velocity at Re = 100 (Unit: m/s):

(a) b/D = 0.0; (b) b/D = 0.018; and (c) b/D = ∞.

was reversed near x/D = 3 and this asymmetric configuration illustrated clearly the effects of the
mean recirculation region. The magnitude of transverse velocity by b/D = 0.018 was smaller than
that of b/D = 0.0 because of reduced acceleration of flow passing the cylinder.
The contours of water depth distribution around the cylinder at Re = 100 are given in Figure 8
and the wall depth distribution according to varying slip lengths is illustrated in Figure 9. As the
flow is blocked by the cylinder, the water surface rises in front of the cylinder, called the bow-wave,
and drops noticeably around the front side of the cylinder reaching the minimum value at the
right angle ( = 90◦) of the cylinder, and then rises again around the rear part of the cylinder. This
phenomenon was well observed in the results by b/D = 0.0 and 0.0135. There was no significant
difference between the results of b/D = 0.0 and 0.0135. However, the free-slip imposition gave
physically inappropriate results at the rear of the cylinder in the sense that it had higher value than
the freestream depth.

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1547

(a) (b)

(c)

Figure 4. Iso-velocity contours of transverse velocity at Re = 100

(Unit: m/s): (a) b/D = 0.0; (b) b/D = 0.018; and (c) b/D = ∞.

4.3. Effect of slip length on time-averaged parameters

The asymmetric and periodic behavior of laminar vortex shedding for Reynolds number less than
150 is usually averaged with respect to time to obtain time-averaged parameters such as wall
vorticity ( ≡ *u 2 /*x −*u 1 /*y) or separation angle on the cylinder surface and mean recirculation
region in the wake. This approach gives mean characteristic of instantaneous behavior of the
unsteady physical properties. However, the researchers engaged in computational fluid dynamics
have their own mathematical model, spatial/temporal discretization method, numerical scheme,
and algorithm. In particular, most of the numerical researches referred here for comparisons with
the result of this study based their studies on Navier–Stokes equations or on their variant forms such
as stream-vorticity equations, not shallow-water equations which are relevant for describing the
free surface flow. Inconsistent computational domains, boundary conditions, input parameters and

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1548 I. W. SEO AND C. G. SONG

2.0

1.8

1.6
b/D=0.0
1.4 b/D=0.009
b/D=0.018
1.2 b/D=

us1/u
1.0

0.8

0.6

0.4 20 20

0.2

0.0
0 30 60 90 120 150 180 210 240 270 300 330 360
(a)
1.2

1.0

0.8

0.6
12°
0.4

0.2
us2/u

0.0

-0.2
12°

-0.4

-0.6

-0.8

-1.0
0 30 60 90 120 150 180 210 240 270 300 330 360
(b)

Figure 5. Velocity distributions at the cylinder wall (Re = 100):

(a) Longitudinal velocity and (b) transverse velocity.

post-processing methods also diversify the computational result for the same variables or properties
so that the various numerical models show discrepancies under the same Reynolds number or
flow conditions. Despite these discrepancies, they give some banded ranges for desired results and
numerical results by present study were compared with the previous numerical and experimental
results. The non-dimensionalized wall vorticity, D/(2u ∞ ), along the upper semicircle of the
cylinder surface for Re = 40 and 100 is shown in Figure 10. The angle was measured from front
stagnation point as presented in Figure 1, which is a convention used throughout this research.
Clearly, the non-dimensionalized quantity became zero at three points: front stagnation point at 0◦ ;
separation points located within 110◦ and 135◦ ; and rear stagnation point at 180◦ . However, in
the case of b/D = 0.018, the wall vorticity at the front stagnation point showed small negative

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1549

1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8

0.6 0.6
u1/u

0.4 b/D=0.018 b/D=0.0 0.4

x/D=2.0 x/D=2.0
x/D=5.0 x/D=5.0
0.2 x/D=7.5 x/D=7.5 0.2
x/D=10.0 x/D=10.0

0.0 0.0

-0.2 -0.2

-0.4 -0.4
-5 -3 -1 1 3 5
y/D

Figure 6. Mean longitudinal velocity distributions along the spanwise sections at Re = 100.

0.15 0.15
b/D=0.0
x/D=0.5
x/D=1.0
x/D=2.0
0.10 0.10
x/D=5.0
x/D=7.5
x/D=10.0

0.05 0.05
u2/u

0.00 0.00

-0.05 -0.05

b/D=0.018
x/D=0.5
x/D=1.0
-0.10 -0.10
x/D=2.0
x/D=5.0
x/D=7.5
x/D=10.0
-0.15 -0.15
-5 -3 -1 1 3 5
y/D

Figure 7. Mean transverse velocity distributions along the spanwise sections at Re = 100.

value. Owing to the imposition of Navier-slip condition on the cylinder surface, the velocity at the
front stagnation point was not equal to zero. Since the longitudinal velocity component should be
vanished by the cylinder blocking, the transverse velocity evolved and the weak wall vorticity at
the front stagnation point resulted from this velocity gradient. For Re = 40, the result by b/D = 0.0

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1550 I. W. SEO AND C. G. SONG

(a) (b)

Figure 8. Contours of water depth distribution at Re = 100 (Unit: m):

(a) b/D = 0.0 and 0.0135 and (b) b/D = ∞.

1.002
b/D=0.0
b/D=0.0135
b/D=

1.001

1.000
h/h

0.999

0.998

0.997
0 30 60 90 120 150 180

Figure 9. Wall depth distributions at the cylinder surface (Re = 100).

was in good agreement with Dennis and Chang [59] and for Re = 100 was in good agreement with
Fornberg [57]. This is because the numerical results by previous researchers were all obtained with
no-slip condition. The results by b/D = 0.018 had lower value than those of b/D = 0.0 since the
nonzero velocities along the cylinder surface kept the velocity gradient smaller.
In this study, the drag and lift force exerted on the body were determined from
 
FD = Tx d, FL = Ty d (13)
 

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1551

12
11 This study
b/D=0.0
b/D=0.018
10
Numerical researches (b/D=0.0)
9 Braza et al. [56]
Fornberg [57]
8 Martinez [58]
Dennis and Chang [59]
7
6
5
4
3
2
1
0
-1
0 30 60 90 120 150 180
(a)
7
This study
b/D=0.0
6 b/D=0.018
Numerical researches (b/D=0.0)
Braza et al. [56]
5 Fornberg [57]
Dennis and Chang [59]

-1
0 30 60 90 120 150 180
(b)

Figure 10. Wall vorticity distributions: (a) Re = 100 and (b) Re = 40.

where Tx and Ty are the traction components on the surface of the body and are calculated by
  
*u i *u j
Ti = − p
i j + + nj (14)
*x j *x i
where
i j denotes the Kronecker delta. Although the pressure, p disappeared in the shallow-water
equations, it can be obtained by taking the projection of the two-dimensional vorticity equation in
the tangential direction [60, 61] as
 
*
p = p0 − n j d (15)
0 *x j

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1552 I. W. SEO AND C. G. SONG

2.5
This study
b/D=0.0
2.3 b/D=0.0045
b/D=0.009
b/D=0.0135
2.1 b/D=0.018
Numerical researches (b/D=0.0)
Martinez [58]
Dennis and Chang [59]
1.9 Persillon and Braza [62]
Braza [63]
Tuann and Olson [64]
CD 1.7 Ta Phuoc Loc [65]
Hamielec and Raal [66]
Experimental research
1.5 Tritton [67]

1.3
b

1.1

0.9
20 40 60 80 100
Re

Figure 11. Drag coefficient variations by Reynolds numbers.

160
This study
b/D=0.0
155
b/D=0.018
Numerical researches (b/D=0.0)
150 Martinez [58]
Dennis and Chang [59]
Persillon and Braza [62]
145 Braza [63]
Tuann and Olson [64]
Ta Phuoc Loc [65]
140
Hamielec and Raal [66]
Wu et al. [68]
135 Ha Minh [69]
Thoman and Szewczyk [70]
s

130

125
b
120

Experimental researches
115 Coutanceau and Bouard [71]
Dimopoulos and Hanratty [72]
110 Grove et al. [73]
Homann [74]

105
20 40 60 80 100
Re

Figure 12. Time-averaged separation angle and Reynolds number relationship.

where p0 is a reference pressure of the front stagnation of the cylinder. In Figure 11, the drag coef-
ficient C D , defined by C D ≡ 2FD /(u 2∞ D) is plotted against Re. The values given in the literature
for C D were quite dispersed. However, the shaded band formed by the present results b/D = 0.0
and 0.018 covered most of the previous numerical and experimental results. The experimental result
by Tritton [67] lay between b/D = 0.0045 and 0.0135. The lowest drag coefficients were obtained
in the b/D = 0.018. The drag coefficients for Re = 20 and 100 in b/D = 0.018 were reduced by

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1553

9
This study
b/D=0.0
8 b/D=0.018
Numerical researches (b/D=0.0)
Martinez [58]
7 Dennis and Chang [59]
Persillon and Braza [62]
Braza [63]
6
Tuann and Olson [64]
Ta Phuoc Loc [65]
Hamielec and Raal [66]
5
L/D Ha Minh [69]
Thoman and Szewczyk [70] b
4 Son and Hanratty [75]

1
D L

0
20 40 60 80 100
Re

Figure 13. Time-averaged wake lengths.

0.45
b/D=0.0
b/D=0.0135
0.35

0.25

0.15

0.05

-0.05
CL

-0.15

-0.25

-0.35

-0.45
100 110 120 130 140
tu
D

Figure 14. Time variation of lift coefficient at Re = 100.

approximately 26 and 16%, respectively, compared with those in the b/D = 0.0. As the slip length
increased, there was a decreasing tendency of drag coefficient since the frictional component of
drag decreased while the pressure part remained almost the same.
The evolution of the separation angle (s ) for b/D = 0.0 and 0.018 is presented in Figure 12.
There was approximately a 15◦ deviation band for the observed separation angles among various
researchers and the results by this study were entirely included in this band. For the case of

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1554 I. W. SEO AND C. G. SONG

(a)

(b)

(c)

(d)

Figure 15. Streamlines (solid lines) and vorticity (dashed lines) pattern for Karman vortex
street (Re = 100 and b/D = 0.0135): (a) T /4; (b) 2T /4; (c) 3T /4; and (d) 4T /4.

b/D = 0.018, the shear layer that developed along the cylinder surface tended to push the separation
point away toward the rear stagnation point so that it had larger separation angle than the results
by b/D = 0.0. However, the separation angle is not significantly affected by slip lengths. This can
be considered by referring to Figure 10(a). Separation angles,  corresponding to the zero vorticity
locating around 120◦ were not distinctively changed by slip length variations.
The time-averaged wake length corresponding to the distance between the rear stagnation
point and reattachment point in mean recirculation region (Figure 1) is shown in Figure 13.
Comparison with several numerical results was found to be satisfactory. As the Reynolds number
increased, the destabilized recirculation zone expanded, which yielded an almost linear rela-
tion between L/D and Re. The slope by b/D = 0.018 was smaller than b/D = 0.0 since the
length of the wake bubble zone was shortened by the combined effects of the reduced wall
vorticity and wall shear stress which caused a shift of the reattachment point closer to the
cylinder.

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1555

tu /D=134

tu /D=137

(a)
tu /D=134

tu /D=137

(b)

Figure 16. Iso-velocity contours at tu ∞ /D = 134 and 137 (Re = 100): (a) b/D = 0.0 and (b) b/D = 0.0135.

4.4. Representation of periodic properties

The periodic properties of the flow are clearly shown by the time-dependent evolution of the
vortex and this section is mainly concerned with the numerical results of amplitude of the lift
coefficient C L ≡ 2FL /(u 2∞ D), instantaneous streamline and vorticity and the Strouhal number
(St ≡ f D/u ∞ ), where f denotes the frequency of vortex generation. In this study, onset of vortex
shedding occurs after Re = 47, which agrees with the critical Reynolds number ranging 46±1
[76, 77]. The time history of the lift coefficient for Re = 100 is shown in Figure 14. After suffering
from irregular oscillations for the initial stage, the lift coefficient reached periodic motion after
dimensionless time (tu ∞ /D) 95. It was observed that the amplitude and period experienced by
b/D = 0.0135 were smaller and shorter than no-slip case. The instantaneous streamlines (solid lines)
and vorticities (dashed lines) for Re = 100 and b/D = 0.0135 are presented in Figure 15 to show how
the streamlines and vorticity field are transformed within a period of vortex shedding. Alternating
fashion of streamline and vorticity from each side of the cylinder is clearly shown. Figure 16 shows
the iso-velocity contours at tu ∞ /D = 134 and 137, which are the times corresponding to the local
maximum and minimum of lift coefficient for b/D = 0.0135 (Figure 14). Referring to Figure 16(b),

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1556 I. W. SEO AND C. G. SONG

0.40

0.35 This study

b/D=0.0
b/D=0.0135
0.30 Numerical researches (b/D=0.0)
Beaudan and Moin [52]
Pinol and Grau [78] b
0.25 Zhang et al. [79]
Green and Gerrad [80]

Amplitude of CL
Li et al. [81]
0.20 Rogers [82]
Experimental research
Tanida et al. [83]

0.15

0.10

0.05

0.00
50 60 70 80 90 100
(a) Re

0.18

0.17

b
0.16

0.15
St

This study
0.14
b/D=0.0
b/D=0.0135
Numerical researches (b/D=0.0)
0.13 Behr et al. [53]
Persillon and Braza [62]
Zhang et al. [79]
0.12 Experimental researches
Williamson [84]
Berger and Wille [85]
Roshko [86]
0.11
50 60 70 80 90 100
(b) Re

Figure 17. Amplitude of lift coefficient and the Strouhal number versus Re:
(a) Amplitude of lift coefficient and (b) St number.

lower part of the cylinder has higher velocity at tu ∞ /D = 134 so that upward lift force evolves
and achieves local maximum of lift coefficient. Similar analysis can be applied to the minimum lift
coefficient at tu ∞ /D = 137. Owing to no-slip assignment, iso-velocity contours around the cylinder
in case of b/D = 0.0 shown in Figure 16(a) have completely different figure compared with the
case of b/D = 0.0135 at those times. Figure 17 was plotted by extracting the amplitude and period
of the cyclic behavior represented in Figure 14. The amplitude of lift coefficient versus Reynolds
number is presented in Figure 17(a). The amplitude of the lift coefficient compared well with the
values reported by Beaudan and Moin [52] and Li et al. [81]. However, experimental results of
Tanida et al. [83] showed that all the numerical computations tended to give overpredictions. This

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1557

is because their experiment was performed under unsteady loading on a cylinder oscillating in the
in-line direction at small amplitude. There is a lack of experimental data concerning this amplitude
issue. The amplitude of lift coefficient by b/D = 0.0135 had relatively small value compared with
that of b/D = 0.0 since the velocity imposition along the circumference of the cylinder face reduced
the wall shear stress exerted on the cylinder. In Figure 14, the time gap between local peaks of
the two results increased as periodic motion progressed and this gave a different frequency and
Strouhal number. Figure 17(b) shows the relation between the Strouhal number and Reynolds
number. It was observed that the values from the present computations by b/D = 0.0 and 0.0135
matched well with the experimental results from Williamson [84] and Roshko [86], respectively.
The frequency of the asymmetrical vortex formation with partial slip velocity was increased due
to the intrinsic inertial effect of Navier-slip condition.

5. CONCLUSIONS

Previous studies on the Navier-slip condition were restricted to the investigation of force reduction
and slip length determination, and thus the separation phenomena and velocity structure around the
cylinder have not been fully examined. In this study, a finite element model based on the SU/PG
scheme was developed to solve shallow-water equations and the influence of partial-slip boundary
conditions and Reynolds numbers on the laminar flow properties past a circular including the wall
vorticity, drag coefficient, separation angle, wake length, velocity distributions on and behind the
cylinder, lift coefficient, and Strouhal number has been analyzed. The simulation results showed
that the wall vorticity with partial slip imposition showed a lower distribution than that of no-slip
condition since the nonzero velocities along the cylinder surface kept velocity gradient smaller.
The shear layer developed along the cylinder surface by the Navier-slip condition tended to push
the separation point away toward the rear stagnation point so that it had larger separation angle
than that of no-slip condition. The length of the wake bubble zone was shortened by the combined
effects of the reduced wall vorticity and wall shear stress which caused a shift of the reattachment
point closer to the cylinder. It was found that the simulated drag coefficient by Navier-slip condition
with moderate slip length was closer to the experimental results compared with the simulated
results with no-slip condition The ability to engineer slip could have dramatic influences on flow
since the viscous dominated motion can lead to large pressure drops and large axial dispersion. By
the slip length control, no-slip, partial-slip and free-slip boundary conditions are tunable and the
velocity distributions at the wall, vortex formation and wake pattern including the amplitude of
lift coefficient and frequency were significantly affected by slip length parameter. Further research
will be carried out to elucidate the effect of physical parameters such as surface roughness and
shear rate upon which slip has been found to depend.

ACKNOWLEDGEMENTS
This research was supported by the SNU SIR Group of the BK21 research program funded by the
Ministry of Education, Science and Technology and a grant (Code No.: 2-3-3) from the Sustainable Water
Resources Research Center of the 21st Century Frontier Research Program. This work was conducted at
the Engineering Research Institute of Seoul National University, Seoul, Korea.

REFERENCES
1. Niavarani A, Priezjev NV. The effective slip length and vortex formation in laminar flow over a rough surface.
Physics of Fluids 2009; 21:52105.
2. Tophøj L, Møller S, Brøns S. Streamline patterns and their bifurcations near a wall with Navier slip boundary
conditions. Physics of Fluids 2006; 18:083102.
3. Hron J, Roux CL, Málek J, Rajagopal KR. Flows of incompressible fluids subject to Navier’s slip on the
boundary. Computers and Mathematics with Applications 2008; 56:2128–2143.
4. Daniello RJ, Waterhouse NE, Rothstein JP. Drag reduction in turbulent flows over superhydrophobic surfaces.
Physics of Fluids 2009; 21:085103.

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1558 I. W. SEO AND C. G. SONG

5. Liu H, Zhou GJ, Burrows R. Lattice Boltzmann model for shallow water flows in curved and meandering
channels. International Journal of Computational Fluid Dynamics 2009; 23(3):209–220.
6. Niavarani A, Priezjev NV. Modeling the combined effect of surface roughness and shear rate on slip flow of
simple fluids. Physical Review E 2010; 81:011606.
7. Craig VSJ, Neto C, Williams DRM. Shear-dependent boundary slip in an aqueous Newtonian liquid. Physical
Review Letters 2001; 87:054504.
8. Truesdell R, Mammoli A, Vorobieff P, Swol FV, Brinker CJ. Drag reduction on a patterned superhydrophobic
surface. Physical Review Letters 2006; 97:044504.
9. You D, Moin P. Effects of hydrophobic surfaces on the drag and lift of a circular cylinder. Physics of Fluids
2007; 19:081701.
10. Min T, Kim J. Effects of hydrophobic surface on skin-friction drag. Physics of Fluids 2004; 16:L55.
11. Keh HJ, Wang LR. Slow motions of a circular cylinder experiencing slip near a plane wall. Journal of Fluids
and Structures 2008; 24:651–663.
12. Wang CY. Stagnation flow on a cylinder with partial slip—an exact solution of the Navier–Stokes equations.
IMA Journal of Applied Mathematics 2007; 72(3):271–277.
13. Chen C, Fang F, Li Y, Huang L, Chung C. Fluid forces on a square cylinder in oscillating flows with non,-mean
velocities. International Journal for Numerical Methods in Fluids 2009; 60:79–93.
14. Grotkop G. Finite element analysis of long period water waves. Computer Methods in Applied Mechanics and
Engineering 1973; 2:147.
15. Lukácová-Medvid’ová M, Teschke U. Comparison study of some finite volume and finite element methods for
the shallow water equations with bottom topography and friction terms. Zeitschrift für Angewandte Mathematik
und Mechanik 2006; 86(11):874–891.
16. Lee S, Wright NG. Simple and efficient solution of the shallow water equations with source terms. International
Journal for Numerical Methods in Fluids 2010; 63:313–340.
17. Liu H, Zhou JG, Burrows R. Multi-block lattice Boltzmann simulations of subcritical flow in open channel
junctions. Computers and Fluids 2009; 38:1108–1117.
18. Farrant T, Tan M, Price WG. A cell boundary element method applied to laminar vortex-shedding from arrays
of cylinders in various arrangements. Journal of Fluids and Structures 2000; 14:375–402.
19. Blackburn HM, Henderson RD. A study of two-dimensional flow past an oscillating cylinder. Journal of Fluid
Mechanics 1999; 385:255–286.
20. Kolar RL, Westerink JJ, Cantekin ME, Blain CA. Aspects of nonlinear simulations using shallow-water models
based on the wave continuity equation. Computers and Fluids 1994; 23(3):523–538.
21. Ghidaoui MS, Kolyshkin AA, Liang JH, Chan FC, Li Q, Xu K. Linear and nonlinear analysis of shallow wakes.
Journal of Fluid Mechanics 2006; 548:309–340.
22. Negretti ME, Vignoli G, Tubino M, Brocchini M. On shallow-water wakes: an analytical study. Journal of Fluid
Mechanics 2006; 567:457–475.
23. Rockwell D. Vortex formation in shallow flows. Physics of Fluids 2008; 20:031303.
24. Chen D, Jirka GH. Experimental study of plane turbulent wakes in a shallow water layer. Fluid Dynamics
Research 1995; 16:11–41.
25. Jirka GH. Large scale flow structures and mixing processed in shallow flows. Journal of Hydraulic Research
2001; 39(6):567–573.
26. Chen D, Jirka GH. Absolute and convective instabilities of plane turbulent wakes in a shallow water layer.
Journal of Fluid Mechanics 1997; 338:157–172.
27. Nishioka N, Sato H. Measurements of velocity distributions in the wake of a circular cylinder at low Reynolds
numbers. Journal of Fluid Mechanics 1974; 65:97–112.
28. Mansy H, Yang P, Williams DR. Quantitative measurements of three-dimensional structures in the wake of a
circular cylinder. Journal of Fluid Mechanics 1994; 270:277–296.
29. Williamson CHK. Vortex dynamics in the cylinder wake. Annual Review of Fluid Mechanics 1996; 28:477–539.
30. Yulistiyanto B, Zech Y, Graf WH. Flow around a cylinder: shallow-water modeling with diffusion–dispersion.
Journal of Hydraulic Engineering 1998; 124(4):419–429.
31. Zhou JG. A lattice Boltzmann model for the shallow water equations. Computer Methods in Applied Mechanics
and Engineering 2002; 191:3527–3539.
32. Negretti ME. Analysis of the wake behind a circular cylinder in shallow water flow. Master Thesis, Trento
University, Italy, 2003.
33. Chan FC, Ghidaoui MS, Kolyshkin AA. Can the dynamics of shallow wakes be reproduced from a single
time-averaged profile? Physics of Fluids 2006; 18:048105.
34. Liang Q, Zang J, Borthwick AGL, Taylor PH. Shallow flow simulation on dynamically adaptive cut cell quadtree
grids. International Journal for Numerical Methods in Fluids 2007; 53:1777–1799.
35. Liang SJ, Tang JH, Wu MS. Solution of shallow-water equations using least-squares finite-element method. Acta
Mechanica Sinica 2008; 24:523–532.
36. Navier CLMH. Memoire sur les lois du mouvement des fluids. Mémoires de l’Académie Royale des Sciences de
l’Institut de France 1823; 6:389–416.
37. Feireisl E, Málek J. Navier’s slip and incompressible limits in domains with variable bottoms. Discrete and
Continuous Dynamical Systems 2008; 1:427–460.

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
 NUMERICAL SIMULATION OF LAMINAR FLOW 1559

38. Jäger W, Mikelić A. Couette flows over a rough boundary and drag reduction. Communications in Mathematical
Physics 2003; 232:429–455.
39. He Q, Wang X. Numerical study of the effect of Navier slip on the driven cavity flow. Zeitschrift für Angewandte
Mathematik und Mechanik 2009; 89:857–868.
40. Behr M. On the application of slip boundary condition on curved boundaries. International Journal for Numerical
Methods in Fluids 2004; 45:43–51.
41. Choi C, Westin KJA, Breuer KS. Apparent slip flows in hydrophilic and hydrophobic microchannels. Physics of
Fluids 2003; 15(10):2897–2902.
42. Choi C, Kim C. Large slip of aqueous liquid flow over a nanoengineered superhydrophobic surface. Physical
Review Letters 2006; 96:066001.
43. Vreugdenhil CB. Numerical Methods for Shallow-water Flow. Kluwer Academic Publishers: Dordrecht, 1994.
44. Parvazinia M, Nassehi V, Wakeman RJ, Ghoreishy MHR. Finite element modelling of flow through a porous
medium between two parallel plates using the Brinkman equation. Transport in Porous Media 2006; 63:71–90.
45. Gerbeau J, Bris LC, Lelievre T. Mathematical Methods for the Magnetohydrodynamics of Liquid Metals. Oxford
Science Publications, 2006.
46. Heinrich JC, Pepper DW. Intermediate Finite Element Method—Fluid Flow and Heat Transfer Applications.
Taylor & Francis: Philadelphia, 1999.
47. Hughes TJR. Recent progress in the development and understanding of SUPG methods with special reference
to the compressible Euler and Navier-Stokes equations. International Journal for Numerical Methods in Fluids
1987; 7:1261–1275.
48. Yu CC, Heinrich JC. Pertov-Galerkin method for multidimensional, time-dependent, convective-diffusion Equation.
International Journal for Numerical Methods in Engineering 1987; 24:2201–2215.
49. Girault V, Raviart P. Finite Element Methods for Navier-Stokes Equations: Theory and Algorithm. Springer:
Berlin, 1986.
50. Gunzburger MD. Finite Element Methods for Viscous Incompressible Flows. Academic Press: Boston, 1989.
51. Weiyan T. Shallow Water Hydrodynamics-Mathematical Theory and NumericalSolution for a Two-Dimensional
System of Shallow Water Equations. Elsevier: Amsterdam, 1992.
52. Beaudan P, Moin P. Numerical experiments on the flow past a circular cylinder at subcritical Reynolds number.
Report. TF-62, Stanford University, 1994.
53. Behr M, Hastreiter D, Mittal S, Tezduyar TE. Incompressible flow past a circular cylinder: dependence of the
computed flow field on the location of the lateral boundaries. Computer Methods in Applied Mechanics and
Engineering 1995; 123:309–316.
54. Tezduyar TE, Shih R. Numerical experiments on downstream boundary of flow past cylinder. Journal of
Engineering Mechanics 1991; 117:854–871.
55. Behr M, Liou J, Shih R, Tezduyar TE. Vorticity-stream function formulation of unsteady incompressible flow
past a cylinder: sensitivity of the computed flow field to the location of the outflow boundary. International
Journal for Numerical Methods in Fluids 1991; 12:323–342.
56. Braza M, Chassaing P, Ha Minh H. Numerical study and physical analysis of the pressure and velocity fields in
the wake of a circular cylinder. Journal of Fluid Mechanics 1986; 165:79–130.
57. Fornberg B. A numerical study of steady viscous flow past a circular cylinder. Journal of Fluid Mechanics 1980;
98:819–855.
58. Martinez G. Caractéristiques dynamiques et thermiques de l’écoulement autour d’un cylinder circulaire à nombre
de Reynolds modéré. Thèse de Docteur-Ingénieur, Institut National Polytechnique de Toulouse, France, 1978.
59. Dennis SCR, Chang G. Numerical solutions for steady flow past a circular cylinder at Reynolds numbers up to
100. Journal of Fluid Mechanics 1970; 42:471–489.
60. Baranyi L. Lift and drag evaluation in translating and rotating non-inertial systems. Journal of Fluids and
Structures 2005; 20:25–34.
61. Tezduyar TE, Liou J, Ganjoo DK. Incompressible flow computations based on the vorticity-stream function and
velocity–pressure formulations. Computers and Structures 1990; 35(4):445–472.
62. Persillon HP, Braza M. Physical analysis of the transition to turbulence in the wake of a circular cylinder by
three-dimensional Navier–Stokes simulation. Journal of Fluid Mechanics 1998; 365:23–88.
63. Braza M. Simulation numérique du décollement instationnaire externe par une formulation vitesse-pression:
Application à l’écoulement autour d’un Cylindre. Thèse de Docteur-Ingénieur, Institut National Polytechnique
de Toulouse, France, 1981.
64. Tuann SY, Olson M. Numerical studies of the flow around a circular cylinder by a finite-element method.
Computers and Fluids 1978; 6:219–240.
65. Ta Phuoc Loc. Étude numérique de l’écoulement d’un fluide visqueux incompressible autour d’un cylindre fixe
ou en rotation, Effet Magnus. Journal of Méchanics 1975; 14:109–134.
66. Hamielec A, Raal J. Numerical studies of viscous flow around circular cylinders. Physics of Fluids 1969;
12:11–17.
67. Tritton DJ. A note on vortex streets behind circular cylinders at low Reynolds number. Journal of Fluid Mechanics
1971; 45:203–208.
68. Wu M, Wen C, Yen R, Weng M, Wang A. Experimental and numerical study of the separation angle for flow
around a circular cylinder at low Reynolds number. Journal of Fluid Mechanics 2004; 515:233–260.

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld
1560 I. W. SEO AND C. G. SONG

69. Ha Minh H. Application de la méthode implicite des directions alternées (ADI) à la résolution des équations de
Navier–Stokes autour d’un cercle. Report M3-25, Institut de M’ecanique des Fluides de Toulouse, 1979.
70. Thoman DC, Szewczyk AA. Time-dependent viscous flow over a circular cylinder. Physics of Fluids 1969;
12(II):76–86.
71. Coutanceau M, Bouard R. Experimental determination of the main features of the viscous flow in the wake of
a circular cylinder in uniform translation. Part 1. Steady flow. Journal of Fluid Mechanics 1977; 79:231–256.
72. Dimopoulos HG, Hanratty TJ. Velocity gradients at the wall for flow around a cylinder for Reynolds numbers
between 60 and 360. Journal of Fluid Mechanics 1968; 33:303–319.
73. Grove AS, Shair FH, Petersen EE, Acrivos A. An experimental investigation of the steady separated flow past a
circular cylinder. Journal of Fluid Mechanics 1964; 19:60–81.
74. Homann F. Einfluss grsser zähigkeit bei strmung um zylinder. Forschung im Ingenieurwesen 1936; 7:1–9.
75. Son JS, Hanratty TJ. Numerical solution for the flow around a circular cylinder at Reynolds numbers of 40, 200
and 500. Journal of Fluid Mechanics 1969; 35:369–386.
76. Zhang L, Balachandar S. Onset of vortex shedding in a periodic array of circular cylinders. Journal of Fluids
Engineering 2006; 128:1101–1105.
77. Lankadasu A, Vengadesan S. Onset of vortex shedding in planar shear flow past a square cylinder. International
Journal of Heat and Fluid Flow 2008; 29:1054–1059.
78. Piñol S, Grau FX. Influence of the no-slip boundary condition on the prediction of drag, lift, and heat transfer
coefficients in the flow past a 2-D cylinder. Numerical Heat Transfer Part Applications 1998; 34(3):313–330.
79. Zhang H, Fey U, Noack BR, Konig M, Eckelmann H. On the transition of the cylinder wake. Physics of Fluids
1995; 7(4):779–794.
80. Green RB, Gerrard JH. Vorticity measurements in the near wake of a circular cylinder at low Reynolds numbers.
Journal of Fluid Mechanics 1993; 246:675–691.
81. Li J, Chambarel A, Donneaud M, Martin R. Numerical study of laminar flow past one and two circular cylinders.
Computers and Fluids 1991; 19(2):155–170.
82. Rogers SE. NASA Ames Research Center. Private communication with Piñol S, Grau FX, 1989.
83. Tanida Y, Okajima A, Watanabe Y. Stability of a circular cylinder oscillating in uniform flow or in a wake.
Journal of Fluid Mechanics 1973; 61:769–784.
84. Williamson CHK. Defining a universal and continuous Strouhal–Reynolds number relationship for the laminar
vortex shedding of a circular cylinder. Physics of Fluids 1988; 31(10):2742–2744.
85. Berger E, Wille R. Periodic flow phenomena. Annual Review on Fluid Mechanics 1972; 4:313–340.
86. Roshko A. On the development of turbulent wakes from vortex streets. Report 1191, California Institute of
Technology, 1954.

Copyright 䉷 2011 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Fluids 2012; 68:1538–1560
DOI: 10.1002/fld