You are on page 1of 12

Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39

https://doi.org/10.1007/s40430-018-1554-1

TECHNICAL PAPER

Flow of power‑law fluid past a circular cylinder in the vicinity


of a moving wall
P. J. Zhang1 · J. Z. Lin1 · X. K. Ku1

Received: 7 August 2018 / Accepted: 21 December 2018


© The Brazilian Society of Mechanical Sciences and Engineering 2018

Abstract
In this paper, the flow of power-law fluid over a circular cylinder near a moving wall is simulated numerically using a
finite volume method for different Reynolds numbers (Re = 1, 10, 40), gap ratios (G/D = 0.2–1.0), and power-law indices
(n = 0.5–1.5). The effects of fluid inertia, wall proximity, and rheological property on the drag and lift coefficient, gap flow
characteristics, and recirculation modes are studied. Possible mechanisms for the variation of the drag and lift coefficient and
the change in recirculation mode are addressed. The results show that the decrease in G/D or Re results in the increase in drag
and lift coefficient, while the effect of n is dependent on Re and G/D. For the drag coefficient, shear-thickening fluid is more
sensitive to the change of G/D. The variation of the lift coefficient can be explained by the movement of angular position of
the front stagnation point. The redistribution of the flow around the cylinder results in different recirculation modes. Five
distinct modes are found and the critical value of G/D for the mode change decreases with the decrease in n and increase in
Re. Variation of the relative vortex intensity and the relative flow intensity nearby leads to the change of recirculation mode.

Keywords  Power-law fluid · Flow past a near-wall cylinder · Wall proximity effect · Recirculation mode · Numerical
simulation

1 Introduction only existed for G/D > 0.3 and the Strouhal number (St) did
not change within the gap ratio studied. Grass et al. [2] also
The flow past a circular cylinder has attracted a vast amount investigated the variation of St at different gap ratios and
of attention recently owing to its theoretical and practical found that St increased to its peak slightly and then declined
significance. The interactions between the cylinder and the as G/D decreased, which is contrary to the results given by
wall will result in a more complex flow when the cylinder Bearman and Zdravkovich [1]. Besides, they thought that jet
is placed in the proximity of a solid wall. In such flows, the flow deflected by the downstream separation zone inhibited
Reynolds number and the gap ratio G/D (where G is the gap the large-scale vortex roll-up and canceled the vorticity in
height, D is the cylinder diameter) are the dominant param- the free shear layer arising from the cylinder surface, which
eters affecting the flow. Previous researches have focused on resulted in the suppression of vortex shedding. Similarly,
the stationary wall where the boundary layer on the wall also Taniguchi and Miyakoshi [3] attributed the suppression of
exerted great influence on the flow characteristics. Bearman vortex shedding to the loss of high concentration of vorticity
and Zdravkovich [1] studied experimentally the influence in the shear layer under the cylinder through absorption by
of gap ratio on the vortex shedding of the cylinder near a the opposite vorticity and penetration of highly turbulent
stationary wall and found that the regular vortex shedding eddies in the boundary layer on the wall. Price et al. [4]
reported that the inhibition of shear layer’s growth accounted
for the vortex shedding suppression and shear layers on the
Technical Editor: Cezar Negrao, PhD. cylinder and the wall did not cancel each other out, which
was contrary to the explanation of Grass et al. [2] and Tani-
* J. Z. Lin
mecjzlin@public.zju.edu.cn guchi and Miyakoshi [3]. Lin et al. [5] studied flow charac-
teristics around a near-wall circular cylinder experimentally
1
Department of Mechanics, State Key Laboratory of Fluid and found that the gap flow exhibited wall-jet characteristics
Power and Mechatronic Systems, Zhejiang University, and the non-dimensional velocity profile presented excellent
Hangzhou 310027, People’s Republic of China

13
Vol.:(0123456789)
39   Page 2 of 12 Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39

similarity at small gap ratio. Harichandan and Roy [6] per- on a wall for Reynolds number varying from 20 to 500 with
formed simulations for single and tandem cylinders with the spectral element method. The prograde rolling was found
finite volume method for Re = 100 and 200, and analyzed to destabilize the wake flow, while the retrograde rotation
the complex interaction between boundary layer on the wall increased the critical Reynolds number for vortex shedding
and the shear layer on the cylinder surface. Besides, there and transition.
are other researches on flow past tandem cylinders [7] and In light of the earlier works, the flow characteristics in
rotating cylinder [8] near a stationary wall. the unsteady flow regime such as vortex shedding suppres-
To study the wall proximity effect alone without the sion, hydrodynamic coefficient, and transition mechanism
interference of the boundary layer on the wall, researches were researched extensively and well understood for the
turned to investigate the flow around the circular cylinder flow past a circular cylinder near a stationary wall. How-
in the vicinity of a moving wall with the same velocity of ever, the steady flow over a circular cylinder near a moving
the free stream. One of the earliest experiments on the cyl- wall has not been studied systematically and there are even
inder translating parallel to a wall, which was equivalent some contradictions among them as shown above. Hence,
to a cylinder placed near a moving wall, was performed by in this paper, the steady flow over a circular cylinder in the
Taneda [9]. In the experiment, the G/D was set to 0.1 and vicinity of a moving wall is investigated thoroughly at three
0.6 at Re = 170. For G/D = 0.1, a single row of vortex street fixed Reynolds number, i.e., 1, 10, and 40, to find out effects
was formed and the wavelength of the vortex street, which of wall proximity and fluid inertia on flow characteristics
means the distance between two adjacent vortices, increased and the mechanism behind them. Besides, the researches
as G/D further decreased. Nishino et al. [10] investigated mentioned above only involve the flow of Newtonian fluid,
experimentally the flow around a circular cylinder near a and there is a lack of research on the non-Newtonian fluid,
moving wall at a sub-critical Re. It was observed that, for while it is often encountered in practical applications, for
a cylinder with end-plates, the flow was essentially two- example, the pipelines on the seabed where the water car-
dimensional. At the small gap (G/D < 0.35), the vortex shed- ries mud and sand and shows non-Newtonian property. For
ding was totally ceased and the drag coefficient was nearly complement, the fluid rheology property is also taken into
constant. However, for the cylinder without end-plates, the account. Among the non-Newtonian fluid, the power-law
Karman-type vortices were not generated and a nearly con- fluid is a kind of typical fluid because its rheological prop-
stant drag coefficient was observed in a wide gap ratio rang- erty such as shear-shinning and shear-thickening has an
ing from 2.0 to 0.05. Huang and Sung [11] employed the effect on free convection [17], natural convection [18], and
immersed boundary method (IBM) to study moving wall’s heat transfer enhancement [19]. Therefore, the purpose of
effect for Re < 600. They identified different vortex shed- this paper is to study the hydrodynamic coefficient, gap flow
ding regimes and studied the flow instability through the characteristics, and recirculation mode for the steady flow
Rayleigh’s theorem and Fjørtoft’s condition. The critical gap of power-law fluid with different power-law indices over a
ratio to suppress the alternate vortex shedding was found to circular cylinder near a moving wall for different gap ratios
decrease as Re increased and became constant for Re > 500. and explain the formation mechanism by investigating the
Furthermore, a single St-G/D curve could be observed at vortex characteristics in the wake behind the cylinder.
different Reynolds number if the separation velocity defined
by Roshko [12] was used to define the Strouhal number. Li
et al. [13] performed a similar numerical simulation with
immersed interface method (IIM) at a low Reynolds num- 2 Numerical model
ber regime (Re = 20–100). In their study, the lift coefficient
presented a monotonic decrease as G/D increased and the 2.1 Problem definition
drag force reached its peak at G/D = 0.5 for various Reynolds
numbers. However, in a recent study of Jiang et al. [14], The sketch of the power-law fluid flowing past a circular cyl-
the drag coefficient was found to decrease monotonically inder in the vicinity of a moving wall is shown in Fig. 1. A
with the increase in G/D in the steady regime. Through the circular cylinder of diameter D is placed at a gap height of
relative principle, Rao et al. [15] investigated the flow over G over a moving wall which moves with the same velocity
a cylinder near a moving wall with the spectral element of the uniform incoming flow so that the interference of the
method, which was equivalent to a circular cylinder trans- boundary layer on the wall is eliminated. In this study, the
lating at different heights above a wall. When the flow kept key non-dimensional parameters are gap ratio G/D, Reynolds
steady, a monotonic decrease for both the drag coefficient number and the power-law index n, which indicates the rate of
and the lift coefficient was observed as G/D increased for shear-thinning or shear-thickening against the shear rate. For
Re < 200. As for the rotating cylinder, Stewart et al. [16] the power-law fluid, the Reynolds number is generalized by
investigated the flow of circular cylinder rolling and sliding Metzner and Reed [20], i.e. Re = ρU2−nDn/K, where ρ is fluid

13
Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39 Page 3 of 12  39

( 2
𝜕 ux 𝜕 2 ux
)
𝜕ux 𝜕u 𝜕u 1 𝜕p
+ u x x + uy x = − +v + , (2)
𝜕t 𝜕x 𝜕y 𝜌 𝜕x 𝜕x2 𝜕y2

( 2 )
𝜕uy 𝜕uy 𝜕uy 1 𝜕p 𝜕 uy 𝜕 2 uy
+ ux + uy =− +v + , (3)
𝜕t 𝜕x 𝜕y 𝜌 𝜕y 𝜕x2 𝜕y2

where u and p are the velocity and pressure, respectively;


the apparent kinematic viscosity v is calculated through a
power-law model [21]:
[ ( )] n−1
𝜕ux 2 𝜕uy 2 𝜕ux 𝜕uy 2
) ( ) (
K
v= 2 +2 + + . (4)
Fig. 1  Sketch of the flow past a circular cylinder near a moving wall
𝜌 𝜕x 𝜕y 𝜕y 𝜕x

The computational domain is presented in Fig. 2a in


density and K is the consistency coefficient of the power-law which the coordinate origin is set at the center of the cylin-
fluid. der. A uniform horizontal velocity is specified at the inlet,
and the outlet is set fully developed. Upper and lower bound-
2.2 Governing equations and boundary conditions aries are given a fixed velocity same as the free stream to
simulate the infinite boundary and the moving wall, respec-
Simulations are performed with Open FOAM to solve the tively. The no-slip boundary condition is employed on the
two-dimensional incompressible continuity and Navier–Stokes cylinder’s surface. The upstream, downstream, and upward
equations: distances are set in the condition of no evident boundaries’
influence on the flow characteristics. To build structured
𝜕ux 𝜕uy
+ = 0, (1)
𝜕x 𝜕y

Fig. 2  Schematics of the computational domain

13
39   Page 4 of 12 Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39

grids, the computational domain is partitioned into several Table 1  Hydrodynamic coefficients for cases with different grid den-
blocks as shown in Fig. 2b. The mesh is gradually refined sity
as grids get closer to the cylinder and the height of the first Grid density CD CL
layer of mesh on the cylinder surface is approximately
Coarse 2.276653 (0.059%) 1.382346 (0.14%)
0.003D.
Medium 2.275309 (0.00040%) 1.383437 (0.058%)
Fine 2.275300 1.384245

3 Methodology

3.1 Numerical schemes
Table 2  Hydrodynamic coefficients for cases with different domain
sizes
There exist some different methods and models of numerical
simulation for the flow past a circular cylinder, for example, LU/D LD/D H/D CD CL
finite difference method and Unsteady Double Wake Model. 10 30 10 2.350765 (4.62%) 1.424221 (4.10%)
In this paper, Eqs. (1)–(4) are solved with the finite volume 15 35 15 2.275309 (1.27%) 1.383437 (1.12%)
method (FVM) and the pressure implicit with splitting of 20 40 20 2.246757 1.368118
operators (PISO) algorithm. The unsteady term is discretized
with first-order Euler implicit scheme, while the surface
normal gradient in the diffusion term is discretized with a
central difference scheme together with the non-orthogonal simulation results and their relative differences to the finest
correction. For the convection term, a linear interpolation is case are listed in Table 1. Obviously, as the mesh is refined,
applied to calculate the quantities on the cell surface. Note the relative difference of the force coefficients to the ones in
that, even though all flows involved are in the steady regime, the fine mesh is below 0.2%. For the sake of accuracy and
transient solver is still adopted here to present the develop- low cost of computation resources, the medium grid density
ment of the wake. is adopted in the subsequent sections.

3.2 Grid independence test


3.3 Independence study of domain size
The mesh convergence is checked by three mesh cases
with different grid density for G/D = 0.2 and Re = 40. In the To study the effect of the domain size, the computational
coarse one, there are 240 grids uniformly distributed on the domains with different upstream distance (L U), down-
cylinder’s surface and 25 layers of mesh in the near-cylinder stream distance (LD), and upward distance (H) are investi-
square zone. The total number of mesh cell is 15,900. In the gated. The simulation results are summarized in Table 2.
latter two cases, the grid number on the surface increases to It is found that the difference declines as the domain size
320 and 400, while the mesh layers increase to 35 and 40, increases. To save the computational cost, L U  = 15D,
so that the total cell number becomes 23,500 and 30,700, LD = 35D and H = 15D are used in the subsequent sections.
respectively.
To verify the grid independence, two hydrodynamic coef-
ficients are investigated, namely the drag coefficient CD and 3.4 Validation
the lift coefficient CL, which are defined as:
To further verify the accuracy of numerical schemes in
FD simulating the flow past a circular cylinder near a moving
CD = / , (5)
𝜌U 2 D 2 wall, the hydrodynamic coefficients obtained in Newtonian
fluid at Re = 25 and Re = 45 are compared with the results
given by Rao et al. [15] and Jiang et al. [14] as shown in
FL Fig. 3. For Re = 25, the relative errors for the drag coef-
CL = / , (6)
𝜌U 2 D 2 ficient in all cases except at G/D = 2.0 are below 4%, while
the relative errors for the lift coefficient in all cases are
where FD and FL are the drag force and lift force obtained smaller than 2.5%. As for the case at Re = 45, all the rela-
by projecting the integrated total force to x and y directions. tive errors for hydrodynamic coefficients are below 2%.
Besides, CD is used as an indicator for the steady state. If Thus, it is evident that the hydrodynamic coefficients agree
the relative error of CD between two sequent time steps is well that given by the former researches and the accuracy
less than 1­ 0−6, the numerical simulation is finished. The of numerical schemes is verified.

13
Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39 Page 5 of 12  39

Fig. 3  Variation of the hydrodynamic coefficient

4 Results and discussion When the gap is small enough, a monotonous increase in the
lift coefficient as n increases is observed. As Re increases
4.1 Hydrodynamic force coefficient to 10 and 40, the lift coefficient increases monotonously as
n increases for the whole power-law index range. Generally
After validating the numerical method, simulations for speaking, decrease in G/D and Re results in an increase in
power-law fluid in the steady regime are carried out at the hydrodynamic coefficient, while variation trend of the
three different Reynolds numbers for n = 0.5–1.5 and hydrodynamic coefficient against the power-law index varies
G/D = 0.2–1.0 to study the effect of fluid rheological prop- with different G/D and Re.
erty and the wall proximity on the hydrodynamic force
coefficient. The variation of the drag coefficient and the lift 4.2 Gap flow characteristics
coefficient are plotted against the power-law index in Fig. 4.
For comparison, the results for the flow of power-law fluid To study the blockage effect of the wall, the variation of the
over a unconfined cylinder [22] are also presented. It is evi- gap flow ratio is investigated. The gap flow ratio, which is
dent that wall proximity results in an obvious monotonous defined by Jiang et al. [14], is the ratio between the total gap
increase in both the drag coefficient and the lift coefficient, flow rate and the total incoming flow rate below the cylinder
irrespective of the power-law index and the Reynolds num- centerline (y = 0). The gap flow ratio is always equal to 1
ber. Similar trend can be found in the numerical simula- for an isolated circular cylinder and decreases if part of the
tions given by Jiang et al. [14] and Rao et al. [15]. Besides, fluid is deflected to the upper side of the cylinder. Figure 5
the sensitivity of the hydrodynamic force coefficient to the shows the variation of the gap flow ratio against the gap
change of the gap distance varies with the power-law index. ratio for different power-law fluids at fixed Reynolds num-
As can be seen from the diagram, the amplification of the ber. Results indicate that increase in n and decrease in G/D
drag coefficient is larger in the shear-thickening fluid (n > 1) result in a decrease in the gap flow ratio. Besides, the gap
than those in the Newtonian fluid (n = 1) and shear-thinning flow ratio decreases with the decrease of Re. Shear layer on
fluid (n < 1), which leads to the difference in the variation the moving wall plays an important role in restricting the
trend at different gap ratios. For Re = 1 and Re = 10, the trend flow through the gap, and the shear layer becomes thicker
of the drag coefficient against the power-law index changes with the decrease in Re. As for the power-law index, the flow
gradually as G/D decreases, from a monotonous decrease to resistance is larger in the shear-thickening fluid than that
a monotonous increase. As for Re = 40, the drag coefficient in the shear-thinning fluid under the same shear rate. Con-
increases monotonously with the increase in n, and the CD–n sequently, the fluid with a larger power-law index applies
curve becomes steeper as G/D decreases. Unlike the drag severer resistance to the flow and deflects a larger percentage
coefficient, the lift coefficient of the cylinder arises from of fluid to the upper side, which results in a decrease in the
the wall proximity, i.e., so-called ground effect. For Re = 1, gap flow ratio.
the lift coefficient increases to its peak at n = 1.0 and then Wall proximity redistributes the incoming flow around
decrease as n further increases for several large gap ratios. the circular cylinder and alters the front stagnation point on

13
39   Page 6 of 12 Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39

Fig. 4  Variation of the drag coefficient (left) and lift coefficient (right) against the power-law index

the cylinder surface. Since the stagnation point is the posi- stagnation point angle, i.e., the included acute angle between
tion where the incoming flow splits into two streams, the the x-axis and the radius pointing to the front stagnation
factors that contribute to the decrease in the gap flow ratio point, as shown in Fig. 6. Furthermore, the variation of the
lead to a downward movement of the front stagnation point lift coefficient against the gap ratio for different power-law
with respect to the front point of the cylinder. That is to say, fluid (omitted for simplification) is analogous to that of the
increase in Re or G/D and decrease in n reduce the front angular position of the front stagnation point at Re = 10 and

13
Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39 Page 7 of 12  39

Fig. 5  Variation of the gap flow ratio against the gap ratio

Re = 40. Similar trend is found by Jiang et al. [14] at finite monotonously with the increase in n or Re within the steady
Reynolds numbers. However, two variation trends are quite regime. As referred in Sect. 4.2, wall proximity redistrib-
different at Re = 1. This is because the lift force arises from utes the flow around the cylinder. Consequently, the recir-
the flow asymmetry and the front stagnation point is closely culation zone will exhibit several distinct modes as shown
related to the pressure distribution since it is the maximum- in Fig. 7. The phase diagram of the recirculation modes at
pressure-coefficient point. In creep flow, the viscous force is different Reynolds number is presented in Fig. 8. It is evi-
comparable to the inertial force and makes considerable con- dent that as G/D decreases, the recirculation mode changes
tributions to the lift force. As Re further increases, inertial gradually from Mode 2 to Mode 5 at Re = 10 or from Mode
forces outweigh the viscous forces and become the dominant 1 to Mode 4 at Re = 40 within the gap ratio range investi-
source of the lift force. Consequently, variation of the lift gated. The critical gap ratio for the mode change decreases
coefficient and movement of the front stagnation point are as n decreases, which means that the shear-thickening fluid
increasing alike as Re increases. changes the recirculation mode more easily than the other
two types under wall proximity. Comparison of two-phase
4.3 Recirculation mode diagrams indicates that for a certain type of power-law fluid,
the critical gap ratio for mode change decreases as the Re
For the power-law fluid flowing past an isolated cylinder, increase.
there are two separate recirculation zones behind the cyl- In Mode 1, fluid in the upper vortex near the junc-
inder if the Reynolds number exceeds the critical value tion of two recirculation zones is partly attracted by the
for flow separation. The recirculation length increases lower vortex, which makes the lower recirculation zone

13
39   Page 8 of 12 Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39

Fig. 6  Variation of the angle position of the front stagnation point

no longer attached to the cylinder surface. Two recircula- Mode 3. After that, the upper recirculation zone continues
tion zones are nearly equal in size within Mode 1, and the shrinking as G/D decreases.
lower recirculation reattaches to the cylinder’s surface at
the critical point between Mode 1 and Mode 2. As G/D 4.4 Wake characteristics
further decreases, the trend mentioned above is reversed.
Part of fluid in the lower recirculation is attracted by the To understand the mechanism behind the change of recircu-
upper vortex, and the upper recirculation departs from lation mode, the wake dynamics in the recirculation zone is
the cylinder surface. The core of two recirculation zones, investigated. Since the basic flow characteristics are analo-
in which the streamlines are closed, shrinks at first and gous at Re = 10 and 40, which is obvious from the phase
then grows with the decrease in G/D as shown in Fig. 9. diagram, we just consider cases at Re = 40 in the follow-
At the critical G/D for Mode 2 and Mode 3, the upper ing paragraph. Figure 10 shows the variation of the non-
recirculation zone reattaches to the cylinder surface and dimensional vorticity (wD/U) at the vortex centroid against
becomes much bigger than the lower recirculation zone. the gap ratio for different power-law fluids at Re = 40. At
The recirculation zone changes to Mode 3 as the cylinder first, two non-dimensional vorticity increases to the peak
further approaches the wall. The lower recirculation zone at the end of Mode 1 as G/D decreases, which means wall
departs from the cylinder surface again, and two recircula- proximity promotes the growth of shear layer on the cylinder
tion zones shrink until the lower vortex diminishes. Note surface at large gap ratio. As G/D further decreases, non-
that, Mode 1 and Mode 3 are similar to each other but the dimensional vorticity at the lower vortex centroid decreases
sizes of two recirculation zones are obviously different in to the minimum at the end of Mode 2. According to [14],

13
Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39 Page 9 of 12  39

Fig. 7  Vector diagram and streamline profile behind the circular cylinder

Fig. 8  Phase diagram of the recirculation mode

13
39   Page 10 of 12 Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39

Fig. 9  Streamline profile of the recirculation zone behind the cylinder for Re = 40 and n = 1.0 (the length of red centerline is 4D)

Fig. 10  Variation of non-dimensional vorticity at vortex centroid at Re = 40

the shear layer generated on the moving wall weakens the the cylinder is too weak to form a vortex, the lower vortex
shear layer on the lower side of the cylinder. However, the is induced by the upper vortex at small gap ratios.
results indicate that the gap flow also weakens the shear As referred in Sect. 4.3, part of fluid near the boundary
layer on the upper side of the cylinder. Interestingly, both is attracted by either lower vortex or the upper vortex in
non-dimensional vorticity increases again at small gap ratio. different recirculation modes. The relative intensity of two
As shown in Fig. 5c, the gap flow ratio declines faster at vortices is one key factor in determining the recirculation
small gap ratio, which overweighs the weakening effect of mode. Since the mode variation and the vorticity characteris-
the gap flow and enhances the shear layer on the upper side tics are analogous for different power-law fluids, we conduct
of the cylinder. Since the shear layer on the lower side of further study of cases at Re = 40, n = 0.8. Variation of the

13
Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39 Page 11 of 12  39

3, though the intensity of the upper vortex increases, a more


rapid increase in the flow intensity near the upper shear layer
makes it hard for the upper vortex to full roll-up. Thus, part
of the fluid is attracted by the lower vortex.

5 Conclusions

The power-law fluid past a circular cylinder in the vicin-


ity of a moving wall is studied numerically. The hydrody-
namic coefficients, gap flow characteristics, and recircula-
tion modes are investigated for Reynolds number ranging
from 1 to 40, gap ratio G/D = 0.2–1.0 and power-law index
n = 0.5–1.5. The following main conclusions can be drawn.
Fig. 11  Variation of the ratio of vorticity at two vertex centroid Wall proximity results in an obvious increase in drag and
(Re = 40, n = 0.8) lift coefficients. The sensitivity of both coefficients to the
change of G/D varies with n. The amplification of the drag
coefficient is larger in the shear-thickening fluid than those
in the Newtonian fluid and shear-thinning fluid. The lift coef-
ficient increases to its peak at n = 1.0 and then decrease as
n further increases for large G/D at Re = 1. When G/D is
small enough, the lift coefficient increases monotonously as
n increases. As Re increases to 10 and 40, the lift coefficient
increases monotonously as n increases. A decrease in G/D
and Re results in an increase in drag and lift coefficient,
while variation trend of the coefficient against n varies with
different G/D and Re.
The gap flow ratio decreases with the increase in n and
the decrease in G/D and Re, and declines faster at small n,
which overweighs the weakening effect of the gap flow and
enhances the shear layer on the upper side of the cylinder.
The flow resistance is larger in the shear-thickening fluid
than that in the shear-thinning fluid, so that the fluid with
Fig. 12  Variation of the ratio of mean streamwise velocity below and
above the cylinder within 0.5D (Re = 40, n = 0.8) a larger n deflects a larger percentage of fluid to the upper
side, resulting in a decrease in the gap flow ratio. Wall prox-
imity redistributes the incoming flow around the cylinder
ratio of vorticity at lower vortex centroid and the upper vor- and alters the front stagnation point on the cylinder surface.
tex centroid for Re = 40 and n = 0.8 is presented in Fig. 11. Increase in Re or G/D and decrease in n reduce the front
Besides, the vortex is formed by the rolling-up of flow in stagnation point angle.
the vicinity of the shear layer. Thus, variation of the ratio The redistribution of the flow around the cylinder results
of mean streamwise velocity below and above the cylinder in various recirculation modes. Five distinct modes are found
within 0.5D is also investigated and shown in Fig. 12. A and the recirculation length increases monotonously with the
local extreme value can be found within Mode 1 or Mode increase in n or Re. The critical value of G/D for the mode
2, indicating two reverse variation trends of vortex intensity change decreases with the decrease in n and increase in Re.
on two sides of the extreme point. Near the critical gap ratio The shear-thickening fluid is easier to change the recircula-
for Mode 1 and Mode 2, the intensity of the lower vortex tion mode than the Newtonian fluid and the shear-thinning
is relatively greater than that of the upper vortex. However, fluid when the cylinder approaches the wall. The core of two
the ratio of two mean streamwise velocities almost reaches recirculation zones shrinks at first and then grows with the
its peak, which means that it is hard for the lower vortex to decrease in G/D. Part of fluid near the boundary is attracted
full roll-up. Consequently, part of fluid is attracted by the by either lower vortex or the upper vortex in different recir-
upper vortex and the recirculation zone changes into Mode culation modes. The relative intensity of two vortices is one
2. Similarly, near the critical gap ratio for Mode 2 and Mode key factor in determining the recirculation mode.

13
39   Page 12 of 12 Journal of the Brazilian Society of Mechanical Sciences and Engineering (2019) 41:39

Acknowledgements  The authors would like to thank the Major Pro- 12. Roshko A (1954) On the drag and shedding frequency of two-
gram of National Natural Science Foundation of China (Grant No. dimensional bluff bodies. Technical report archive and image
11632016). library
13. Li Z, Jaiman RK, Khoo BC (2016) An immersed interface method
for flow past circular cylinder in the vicinity of a plane moving
wall. Int J Numer Methods Fluids 81:611–639
References 14. Jiang HY, Cheng L, Draper S, An HW (2017) Two- and three-
dimensional instabilities in the wake of a circular cylinder near a
1. Bearman PW, Zdravkovich MM (1978) Flow around a circular moving wall. J Fluid Mech 812:435–462
cylinder near a plane boundary. J Fluid Mech 89:33–47 15. Rao A, Thompson MC, Leweke T, Hourigan K (2013) The flow
2. Grass AJ, Raven PW, Stuart RJ, Bray JA (1984) The influence of past a circular cylinder translating at different heights above a
boundary layer velocity gradients and bed proximity on shedding wall. J Fluids Struct 41:9–21
from free spanning pipelines. ASME J Energy Resour 106:70–78 16. Stewart BE, Thompson MC, Leweke T, Hourigan K (2010) The
3. Taniguchi S, Miyakoshi K (1990) Fluctuating fluid forces acting wake behind a cylinder rolling on a wall at varying rotation rates.
on a circular cylinder and interference with a plane wall-effects J Fluid Mech 648:225–256
of boundary layer thickness. Exp Fluids 9:197–204 17. Gupta AK, Sasmal C, Sairamu M, Chhabra RP (2014) Laminar
4. Price SJ, Sumner D, Smith JG, Leong K, Paidoussis MP (2002) and steady free convection in power-law fluids from a heated
Flow visualization around a circular cylinder near to a plane wall. spheroidal particle: a numerical study. Int J Heat Mass Transf
J Fluid Mech 16:175–191 75:592–609
5. Lin WJ, Lin C, Hsieh SC, Dey S (2009) Flow characteristics 18. Turan O, Sachdeva A, Poole RJ, Chakraborty N (2012) Laminar
around a circular cylinder placed horizontally above a plane natural convection of power-law fluids in a square enclosure with
boundary. J Eng Mech 135:697–716 differentially heated sidewalls subjected to constant wall heat flux.
6. Harichandan AB, Roy A (2012) Numerical investigation of flow J Heat Transf 134:122504
past single and tandem cylindrical bodies in the vicinity of a plane 19. Sadeghi H, Izadpanah E, Rabiee MB, Hekmat MH (2017)
wall. J Fluids Struct 33:19–43 Effect of cylinder geometry on the heat transfer enhancement
7. Raiola M, Ianiro A, Discetti S (2016) Wake of tandem cylinders of power-law fluid flow inside a channel. J Braz Soc Mech Sci
near a wall. Exp Therm Fluid Sci 78:354–369 39:1695–1707
8. Cheng M, Luo LS (2007) Characteristics of two-dimensional flow 20. Metzner AB, Reed JC (1955) Flow of non-Newtonian fluids: cor-
around a rotating circular cylinder near a plane wall. Phys Fluids relation of the laminar, transition, and turbulent-flow regions.
19(6):93–184 AIChE J 1:434–440
9. Taneda S (1965) Experimental investigation of vortex streets. J 21. D’Alesso SJD, Pascal JP (1996) Steady flow of a power-law fluid
Phys Soc Jpn 20:1714–1721 past a cylinder. Acta Mech 177:87–100
10. Nishino T, Roberts G, Zhang X (2007) Vortex shedding 22. Sivakumar P, Bharti RP, Chhabra RP (2007) Steady flow of
from a circular cylinder near a moving ground. Phys Fluids power-law fluids across an unconfined elliptical cylinder. Chem
19:025103-1–025103-12 Eng Sci 62:1682–1702
11. Huang WX, Sung HJ (2007) Vortex shedding from a circular cyl-
inder near a moving wall. J Fluids Struct 23:1064–1076

13

You might also like