You are on page 1of 11

Experimental evaluation of the virtual mass

of two solid spheres accelerating in uids


Abdullah Abbas Kendoush
a,
*
, Abbas H. Sulaymon
b
, Sawsan A.M. Mohammed
b
a
Center of Engineering Physics, Ministry of Science and Technology, Baghdad, Iraq
b
Chemical Engineering Department, University of Baghdad, Baghdad, Iraq
Received 12 April 2006; received in revised form 16 August 2006; accepted 23 August 2006
Abstract
The virtual mass force of two equal spheres is investigated experimentally when both spheres were moving side by side and along the
line joining their centers. The velocities of two accelerated spheres rising along the axis of a glass cylindrical column of a 2 m long and a
diameter of 0.3 m, under the action of a driving weight, are measured as a function of time and separation distance between spheres
centers ranging from about 2 to 7 radii. During the accelerated motion of the spheres, Reynolds number varied between 10 and 10
3
.
The spheres employed are made of stainless steel of the following four dierent diameters: 9.4, 12, 13.5 and 15.4 mm. A main conclusion
of this work is that the virtual mass coecient (C
VM
) was found to decrease with increasing the instantaneous Reynolds number. For two
equal spheres moving side by side, C
VM
> 0.5 when the spheres touched. As the distance between them increased, C
VM
decreased and
approached the single sphere value of 0.5 when the spheres were separated by more than 3 radii. For two equal spheres moving along
the line joining their centers, when the spheres touched, C
VM
< 0.5. As the distance between them increased, C
VM
increased and
approached the value of 0.5 when the spheres were separated by more than 5 radii. The experimental results compared well with theo-
retical results found in the literature.
2006 Elsevier Inc. All rights reserved.
Keywords: Hydrodynamic interaction; Two spheres; Drag coecient; Virtual mass coecient
1. Introduction
The ow of bubbles, droplets or solid particles in uids
play an important role in many natural and industrial pro-
cesses where energy, mass and momentum are exchanged
between the bubbles, droplets or particles with uids.
Hydrodynamic interaction and momentum exchange
become increasingly important as the inter-particle dis-
tance becomes smaller.
Hydrodynamic interaction of a pair of three-dimen-
sional bodies moving in a uniform ow has been investi-
gated by a number of researcher, Hicks [1] and Herman
[2] analyzed the kinetic energy of the uid, due to the
motion of two spheres along the line joining their centers,
and obtained solutions of added masses in terms of dou-
blets interior to each body.
An exact solution of the linearized NavierStokes equa-
tions for steady axisymmetric motion of viscous uid at
low Reynolds number around two spheres was obtained
by Stimson and Jeery [3].
For the time-dependent motion, Oseen [4] studied the
problem of a falling spherical rigid inclusion in a Newto-
nian uid. The equation of motion of the sphere developed
by Boussinesq, Basset and Oseen (BBO equation) may be
written as follows:
F t 6pal
f
u v
1
2
m
f
u v 6a
2
pl
f
q
f

1=2

_
t
0
d
ds
u v
t s
1=2
ds 1
0894-1777/$ - see front matter 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.expthermusci.2006.08.007
*
Corresponding author. Address: Department of Mechanical and
Aerospace Engineering, University of Florida, 237 MAE Building B
Gainesville, FL 32611-6300, USA.
E-mail address: kendoush@u.edu (A.A. Kendoush).
www.elsevier.com/locate/etfs
Experimental Thermal and Fluid Science 31 (2007) 813823
Some approximate solutions for two-sphere problems or
a sphere in the neighborhood of a plane can be found in
classics like Lamb [5] and Milne-Thomson [6], where image
methods are mainly used.
Michael [7] solved the problem of an ideal uid owing
parallel to the line of centers of an innite row of spheres in
an unbounded ow and derived the dependence of the
added mass coecient of a sphere on the spacing. He found
that the added mass coecient varies from 0.173 with
touching spheres to 0.5 with spheres far apart.
Voinov [8] computed the kinetic energy and hydrody-
namic forces for two spheres only small distance apart.
To examine the acceleration eects on the motion of
the particle, Odar and Hamilton [9], proposed modifying
the BBO equation by multiplying respectively the three
terms in right side of Eq. (1) by empirical coecients C
D
,
C
VM
and C
H
. These coecients depend on the particle
Reynolds number (Re
s
) and the acceleration number
(Ac), where
Ac
u
2
2a
du
dt
_ _ 2
C
D

24
Re
s
1 0:15Re
0:687
s
3
C
VM
1:05
0:066
Ac
2
0:12
4
and
C
H
2:88
3:12
Ac 1
3
5
The modied form of Eq. (1) becomes the following:
F t
1
2
qC
D
pa
2
u
2
C
VM
m
f
du
dt
C
H
a
2
pl
f
q
f

1=2

_
t
0
du
ds
ds
t s
1=2
6
Each of the dynamic force contributions in Eq. (6), must be
modied to take into account particle interaction eects.
The drag force can be modied by introducing an interac-
tion parameter k, dened as the ratio of the drag coecient
of an interactive particle C
D
, to the one of an isolated sin-
gle particle, C
Do
, i.e. k = C
D
/C
Do
, which is a function of
the particle spacings [10].
For two spheres moving side by side in a uid, Legendre
et al. [11] used a correction factor for the drag force in the
moderate to large Reynolds number regime as
k 1
1
8
l
d
_ _
3
7
For two spheres moving along their line of centers, Zhu
et al. [12] proposed an empirical equation for the drag force
correction for Reynolds number from 20 to 130 as
C
D
C
Do
1 1 A exp B
l
d
_ _
8
where
A 1 exp0:483 3:45 10
3
Re
s
1:07 10
5
Re
2
s

9
and
B 0:115 8:75 10
4
Re
s
5:61 10
7
Re
2
s
10
Nomenclature
a radius of sphere (m)
Ac acceleration number ()
C
D
drag coecient ()
C
Do
drag coecient of an isolated sphere ()
C
H
history coecient ()
C
VM
virtual mass coecient ()
d sphere diameter (m)
F(t) hydrodynamic force (N)
F
D
drag force (N)
F
H
history force (N)
F
VM
virtual mass force (N)
g gravitational acceleration (m/s
2
)
KE total kinetic energy of the system (N m)
KE
1
kinetic energy of the body (N m)
KE
2
kinetic energy of the uid (N m)
l the distance between the centers of spheres (m)
M mass of the body (kg)
M
0
mass of the uid displaced by the body (kg)
M
drive
drive weight (kg)
m
f
mass of the uid (kg)
m
s
mass of the sphere (kg)
Re
s
Reynolds number based on the sphere diameter
(ud/m) ()
t time (s)
u sphere velocity (m/s)
v uid velocity (m/s)
W uncertainty ()
Greek letters
Dt time step (s)
k drag interaction parameter (C
D
/C
Do
) ()
l
f
dynamic viscosity of uid (kg/m s)
q
f
density of uid (kg/m
3
)
q
s
density of sphere (kg/m
3
)
s dummy time variable (s)
/ velocity potential (m
2
/s)
814 A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823
The results of Leichtberg et al. [10] indicated that
the Basset (history) force is not signicantly altered by
multi-particle interaction eects and thus adequately
approximated by the single particle representation for most
engineering purposes.
For more than one particle in line with uniform spacing,
two numerical results were obtained by Helnstine and
Dalton [13]. First at larger spacing, all particles have the
same and constant value of the added mass coecient
equal to 0.5. Second, at low spacing the added mass coe-
cient depends on distance and is slightly lower than 0.5.
For two spheres moving in line with the ow Van
Wijngaarden [14] found
C
VM

1
2
1 3
a
l
_ _
3
3
a
l
_ _
6
9
a
l
_ _
8
3
a
l
_ _
9

_ _
11
and for two spheres where the line connecting their centers
is perpendicular to the ow
C
VM

1
2
1
3
2
a
l
_ _
3

3
4
a
l
_ _
6
3
a
l
_ _
8

3
4
a
l
_ _
9

_ _
12
Cai and Wallis [15] presented theoretical predictions and
experimental measurements of the added mass coecient
for rows of spheres in a tube and for rectangular arrays
of spheres.
Ruzicka [16] studied the dynamic behavior of equal-sized
spherical gas bubbles rising in vertical line numerically at
Reynolds number 50200. He suggested a force-law model
for hydrodynamic interaction within the bubble chain.
Legendre and Magnaudet [11] studied numerically the
interaction between two spherical bubbles rising side by
side in a stagnant viscous uid.
Recently, Kendoush [17] showed analytically that the
virtual mass coecient of the rotating sphere in uids is
equal to ve.
It can be concluded from literature survey that most of
the two-sphere studies were theoretical and there is a lack
of experimental investigation that was the main motivation
behind the present paper.
2. Experimental apparatus and measurement system
The experimental apparatus is shown in Fig. 1. It con-
sisted of a borosilicate glass cylindrical column of length
2.0 m and a diameter of 0.3 m. The spheres used in the
experiments were made of stainless steel of diameters 9.4,
12.0, 13.5 and 15.4 mm.
Each sphere pair was connected by using a thin steel rod
with diameter approximately 1/50 of the sphere diameter.
The eect of the existence of the connecting rod on the
motion of the spheres may be considered to be negligible
[18]. The distances between spheres (l) were varied from 2
to 7 radii.
The spheres were suspended by a shing string of
0.18 mm diameter, which passed over an aluminum pulley
to a drive weight that, provided the driving force. External
friction was reduced to a minimum with ball bearings on
the pulleys shaft.
The spheres were submerged in the water of the column
at an initial position of approximately 0.5 m from the
Flange
Column 30 cm
in diameter
Drain
Light source
Photocell
Interface
Computer
Liquid
Pulley
Drive weight
Test objects (spheres) 2.0 m
Light blocks
Fig. 1. Schematic diagram of experimental apparatus.
A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823 815
bottom, considering the eects of the hydrodynamic inter-
actions of the lower bottom wall in the motion of the two
spheres. Upon release of the string the spheres rose in the
column under the action of falling weights. Measurements
of the velocity of the pairs of spheres were carried out for
dierent sphere diameters and sphere separation distances.
On the top of the column there was a system of light
source and a photo-cell. A small pieces of eight light blocks
were xed on the part of string which was un-submerged.
As the spheres moved in the water, the light blocks also
moved up through the collimator, which made the light
intensity seen by the photo-cell varied and hence its resis-
tance. This causes a variable voltage drop across the
photo-cell.
An electronic circuit was constructed to measure the
time elapsed between two successive light blocks. The elec-
tronic circuit components consists of light source, photo-
cell detector and the interface unit connected to the remote
computer. The details of the electronic circuit and the pro-
gram which drives the interface unit is given in the thesis of
the third author [19].
3. Experimental procedure
3.1. Single sphere acceleration measurement
This part included study on the accelerated motion of a
single sphere for calibrating the system. A single sphere was
submerged in the liquid column. It rose in the column
under the action of a falling weight which was slightly
heavier than the sphere. As the sphere moved up, the light
blocks also moved up and passed between the light and the
photo-cell. The interface unit fed the response to the com-
puter until all light blocks were passed. The online com-
puter printed the velocity of the spheres versus time on
the screen.
The virtual mass coecient of the isolated sphere was
found to be in the range of 0.50.55. This result was consid-
ered as a check on the consistency of the experimental
system.
3.2. Two spheres acceleration measurement
Similar to the experiment with single sphere, the motion
of two identical solid spheres rising along their line of cen-
ters and side by side were carried out for dierent sphere
diameters and separation distances.
4. Results and discussion
4.1. Velocity measurements and the qualications of the drag
and history equations
The calculations were preformed with MATLAB, the
program algorithm and source code are available in [19].
Measurement of instantaneous velocity with time.
The online computer printed a velocity versus time plots
for dierent spheres diameters, separation distance and
arrangements. Fig. 2 shows one of such plots. The two
in-line spheres move faster in uids than the two side-by-
side spheres. This is akin to the two cases of dropping a
rod axially in a uid and dropping the same rod horizon-
tally. Surely the rst rod will drop faster than the second
one.
Drag force evaluation.
Modication of the Stokes drag coecient to account
for higher Reynolds number.
For the considered range of spheres Reynolds number
from 10 to 10
3
, the widely used empirical form of drag cor-
rection with Reynolds number employed is
C
D

24
Re
s
1
1
6
Re
2=3
s
_ _
13
This equation has been compared with the main models
and experimental data of the literature as shown in Fig. 3.
The above equation has been proposed by Shiller and
Nauman mentioned in Ref. [20].
0
0.01
0.02
0.03
0.04
0.05
0 10 20 30
Time, s
V
e
l
o
c
i
t
y
,

m
/
s
in line
side by side
Fig. 2. Velocity versus time for two spheres (d = 12 mm, l/a = 2) in water.
Fig. 3. Comparison of Eq. (13) with the various drag models and the
experimental data of the literature.
816 A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823
Drag force correction due to interaction.
Drag force correction for two spheres moving side by
side.
The interaction parameter (k = C
D
/C
Do
) was calculated
using Eq. (7). The variation in the drag force of interacting
spheres with distance between the two spheres using Eq. (7)
is compared with Tsuji et al. [21] experimental results as
shown in Fig. 4.
The drag ratio curve shows that when the spheres are
very close to each other, the drag increases but the eect
disappears at a distance larger than l/d = 23, and the drag
ratio reaches that of a single sphere (i.e. C
D
/C
Do
= 1).
Drag force correction for two spheres moving along
their line of centers.
In the present research, for the considered range of Rey-
nolds number, no reliable correction exists to describe this
interaction.
Tsujis results [21] are used in this study to predict an
expression that relates the drag ratio to the inter-particle
distance, which takes the exponential form
C
D
C
Do
1 exp 0:5
l
d
_ _
14
The data of Tsujis et al. agrees well with the predicted
correlation as shown in Fig. 5.
Eq. (14) is considered an average drag equation for the
two in-line spheres, due to the fact that these spheres are
not completely free to move because of the connecting
rod between them.
The drag is aected by the interaction, it decreases with
decreasing distance between the spheres, but the eect of
interaction disappears at a distance larger than l/d = 510
and asymptotically approaches the single sphere value.
History force evaluation.
The history force coecient (C
H
) is calculated by using
Eq. (5) after obtaining the acceleration number (Ac) from
Eq. (2) by using experimental values of the instantaneous
velocity and acceleration.
_
t
0
du
ds
ds
t s
1=2

m1
n0
u
n1
u
n
Dt
n
2

t t
n
p


t t
n1
p

15
where the counter 0, . . . , m refers to the successive solutions
points in the interval [0, t] and n is the number of time inter-
vals. This equation was developed by Alexander [22]. For
each time interval the value of
du
dt
is obtained from the pres-
ent experimental data and multiplied by the corresponding
2

t t
n
p
. The history force is then calculated from
F
H
C
H
pl
f
q
f

1=2
a
2
_
t
0
du
ds
ds
t s
1=2
16
Fig. 6 shows the change of the history force with time.
The history force decays faster than t
2
, the same behavior
was observed by Lawrence and Mei [23], Mordant and Pin-
ton [24] Coimbra and Rangel [25] and Michaelides [26].
Also, as it could be seen from Fig. 6, the history force is
not signicantly altered by multi-particle interaction
eects.
4.2. Single sphere measurements
The equation of unidirectional acceleration motion of a
single sphere can be written as follows:
M C
VM
M
0

du
dt

F
i
17
1.00
1.05
1.10
1.15
1.20
1.25
1.30
0.00 2.00 4.00 6.00 8.00
l/d
C
D
/
C
D
0
Tsuji's results
Equation (7)
Fig. 4. Drag ratio versus inter-particle distance for two side by side
spheres.
Fig. 5. Drag ratio as a function of inter-particle distance for two spheres
moving along their line of centers.
0.0E+00
1.0E-06
2.0E-06
3.0E-06
0 10 20 30
Time, s
H
i
s
t
o
r
y

f
o
r
c
e
,

N
l/a=2
l/a=3.745
l/a=5.7
l/a=7.638
y=0.0002*x**-2.7
Fig. 6. Change of history force with time for two side by side spheres
(d = 9.4 mm) in water.
A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823 817
where

F
i
= drive weight forcebuoyancy forcedrag
forcehistory force.
Eq. (17) becomes after considering these forces
V
s
q
s
C
VM
q
f

du
dt
M
drive
V
s
q
s
q
f
g
1
2
C
D
pq
f
a
2
u
2
C
H
pl
f
q
f

1=2
a
2
_
t
0
du
dt
ds
t s
1=2
18
The values of C
VM
are obtained form Eq. (18) by substi-
tuting the experimental values of velocity and acceleration
at each time interval. Fig. 7 shows the change of C
VM
with
the instantaneous Reynolds number.
Eq. (13) for the drag coecient and Eq. (15) for the his-
tory force were substituted into Eq. (18) together with the
experimental data of the velocity of the two spheres. The
aim was to get C
VM
from solving Eq. (18).
The virtual mass coecient decreases slightly with
increasing Reynolds number and reaches almost a value
of 0.5 at Reynolds number greater than 400.
The values of the virtual mass coecient are in close
agreement with the general results obtained for C
VM
for a
single sphere (C
VM
= 1/2), presented in the works of Darwin
[27], Leichtberg et al. [10], Magnaudet and Eames [28].
An attempt is made to correlate the experimental values
of C
VM
with the instantaneous Reynolds number. The cor-
relation proposed in the present study is
C
VM
0:5 ARe
B
s
19
where the value of A depends on Reynolds number range
as summarized in Table 1.
4.3. Two spheres measurements
For each pair of spheres, the values of the virtual mass
coecient (C
VM
) are obtained from Eq. (18).
The application of Eq. (18) to the case of the two
spheres rising side by side is quite acceptable due to the
symmetry. However, the application of Eq. (18) to the case
of the two spheres rising in line is not so straightforward.
In fact, the leading sphere is subjected to the driving force
of the string and to a resisting force exerted by the connect-
ing rod, which is unknown. The trailing sphere on the con-
trary, is driven by the force exerted by the connecting rod.
Eq. (18) is thus applicable only to the entire system of the
two spheres, and the resulting C
VM
is simply an averaged
value over the two spheres.
Figs. 8 and 9 show the relation between the virtual mass
coecient and the instantaneous Reynolds number. It can
be seen from the gures that the virtual mass coecient
varies inversely with the instantaneous Reynolds number.
For two equal spheres moving side by side, Fig. 10
shows that when the spheres touch (l/a = 2), the virtual
mass coecient is greater than 0.5. As the distance between
the spheres increases, the virtual mass coecient decreases
and asymptotically approaches the single-sphere value (i.e.
C
VM
= 0.5). This value is attained when the spheres are
separated by more than 3 radii.
For two equal spheres moving along the line joining
their centers, Fig. 11 shows that when the spheres touch,
the virtual mass coecient is less than 0.5. The virtual mass
coecient increases rapidly as the distance between the
spheres increases. At larger separation it asymptotically
approaches the single-sphere value of 0.5. This value is
attained when the spheres are separated by more than 5
radii.
A comparison between the results of this work and the
numerical solution of Helnstine and Dalton [13] is also
shown in Figs. 10 and 11.
Dierences are due to the Reynolds number eect on the
virtual mass coecient which was taken in consideration in
this work. Despite this dierence, one observes that the
curves have similar shapes and that the dierences are less
marked at high Reynolds number.
For two spheres moving side by side in water, the pres-
ent experimental C
VM
can be correlated to the inter-parti-
cle distance and the instantaneous Reynolds number as
follows:
C
VM

1
2
1
3
2
l
a
_ _
3
_ _
ARe
B
s
20
For two spheres moving along the line joining their cen-
ters in water, the experimental values of C
VM
can be corre-
lated to the inter-particle distance and the instantaneous
Reynolds number as follows:
C
VM

1
2
1 3
l
a
_ _
3
_ _
ARe
B
s
21
Fig. 7. Change of virtual mass coecient with the instantaneous Reynolds
number for a single sphere in water.
Table 1
Coecient values of C
VM
proposed correlation
Coecient Value Re
s
range
A 15 100 6 Re
s
6 300
10 300 6 Re
s
6 500
5 Re
s
P500
B 1.125 All ranges
Correlation coecient, R = 0.9835.
818 A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823
where the values of A and B are given in Table 1. These val-
ues are exactly the same for Eqs. (19)(21).
The rst term in Eqs. (20) and (21) show the dependence
of C
VM
on the inter-particle distance. For both arrange-
ments, there is a good agreement with the published corre-
lations. The second term represents the eect of Reynolds
number; it was found that the value of C
VM
varies inversely
with Reynolds number.
The lift force was not considered in the side by side
arrangement simply because the spheres were connected
to each other through the connecting rod and therefore
the spheres were prevented from rotation.
A similar behavior was observed by Moorman [29] and
Takahashi and Endoh [30], the latter had suggested the fol-
lowing correlation for an oscillating sphere:
C
VM
0:7 15Re
0:75
s
for 6 < Re
s
< 4 10
4
22
Kendoush [31] found analytically that C
VM
for the growing
or collapsing spherical bubble is an inverse function of Re
number.
d=9.4 (mm)
0.5
0.6
0.7
0.8
100 200 300 400
Re
s
C
V
M
l/a=2
l/a=3.745
l/a=5.7
l/a=7.638
d=12 (mm)
0.5
0.6
0.7
0.8
100 200 300 400 500
Re
s
C
V
M
l/a=2
l/a=3
l/a=5
l/a=7.833
d=13.5 (mm)
0.5
0.6
0.7
0.8
100 200 300 400 500
Re
s
C
V
M
l/a=2
l/a=3
l/a=4.177
l/a=6.311
d=15.4 (mm)
0.5
0.6
0.7
0.8
100 200 300 400 500 600 700 800 900
Re
s
C
V
M
l/a=2
l/a=2.61
l/a=4.013
l/a=5.354
Fig. 8. Variation of virtual mass coecient with the instantaneous
Reynolds number for two side by side spheres in water.
d=9.4 (mm)
0.3
0.4
0.5
0.6
100 200 300 400 500
Re
s
C
V
M
l/a=2
l/a=3.745
l/a=5.7
l/a=7.638
d=12 (mm)
0.3
0.4
0.5
0.6
0.7
100 200 300 400 500 600
Re
s
C
V
M
l/a=2
l/a=3
l/a=5
l/a=7.833
d=15.4 (mm)
0.3
0.4
0.5
0.6
100 300 500 700 900
Re
s
C
V
M
l/a=2
l/a=2.61
l/a=4.013
l/a=5.354
d=13.5 (mm)
0.3
0.4
0.5
0.6
0.7
100 200 300 400 500
Re
s
C
V
M
l/a=2
l/a=3
l/a=4.177
l/a=6.311
Fig. 9. Variation of virtual mass coecient with the instantaneous
Reynolds number for two in line spheres in water.
A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823 819
4.4. Relative magnitudes of the forces acting on particles
Figs. 12 and 13 present the values of history force com-
pared to that of the virtual mass force for the two spheres
congurations studied.
The gures illustrate that the magnitude of the history
force is small compared to the virtual mass force. It is seven
orders of magnitude smaller approximately at the begin-
ning of motion, the values of the two forces become closer
as time passes until they reach zero. Thus, it can be con-
cluded that the history eects are less important than the
virtual mass force.
4.5. Tube wall correction
In this study the minimum ratio of the diameter of the
cylindrical test column was 20 times the diameter of the
spheres. With such a ratio, no correction for wall eect
was applied to the data.
The wall drag multiplier K
wall
represents the ratio of the
steady drag in the presence of the wall to the Stokes drag
that a sphere experiences in an unbounded ow. K
wall
is
given as follows [26]:
K
wall
1 2:014413a=R 2:08877a=R
3
6:94813a=R
5

1
23
d=9.4 (mm)
0.3
0.35
0.4
0.45
0.5
0.55
0 2 4 6 8 10
l/a
C
V
M
Helfinstine results
Exp. Cvm at Re=200
Exp. Cvm at Re=250
Exp. Cvm at Re=300
Exp. Cvm at Re=350
d=12 (mm)
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0 2 4 6 8 10
l/a
C
V
M
Helfinstine results
Exp. Cvm at Re=200
Exp. Cvm at Re=250
Exp. Cvm at Re=300
Exp. Cvm at Re=350
Exp. Cvm at Re=400
d=13.5 (mm)
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0 2 4 6 8
l/a
C
V
M
Helfinstine results
Exp. Cvm at Re=200
Exp. Cvm at Re=250
Exp. Cvm at Re=300
Exp. Cvm at Re=350
Exp. Cvm at Re=400
d=15.4 (mm)
0.3
0.35
0.4
0.45
0.5
0.55
0 2 4 6
l/a
C
V
M
Helfinstine results
Exp.Cvm at Re=400
Exp. Cvm at Re=500
Exp. Cvm at Re=600
Exp. Cvm at Re=700
Fig. 11. Virtual mass coecient versus inter-particle distance for two in
line spheres in water.
d=9.4 (mm)
0.5
0.6
0.7
0 2 4 6 8 10
l/a
C
V
M
Helfinstine results
Exp. Cvm at Re=200
Exp. Cvm at Re=250
Exp. Cvm at Re=300
Exp. Cvm at Re=350
d=12 (mm)
0.5
0.55
0.6
0.65
0 5 10
l/a
C
V
M
Helfinstine results
Exp. Cvm at Re=200
Exp.Cvm at Re=250
Exp. Cvm at Re=300
Exp. Cvm at Re=350
Exp. Cvm at Re=400
d=13.5 mm
0.5
0.6
0.7
0 2 4 6 8
l/a
C
V
M
Helfinistine results
Exp.Cvm at Re=250
Exp.Cvm at Re=300
Exp.Cvm at Re=350
Exp.Cvm at Re=400
Exp. Cvm at Re=425
d=15.4 (mm)
0.5
0.55
0.6
0.65
0.7
0.75
0 2 4 6
l/a
C
V
M
Helfinstine results
Exp. Cvm at Re=400
Exp. Cvm at Re=500
Exp. Cvm at Re=600
Exp Cvm at Re=700
Exp Cvm at Re=800
Fig. 10. Virtual mass coecient versus inter-particle distance for two side
by side spheres in water.
820 A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823
The value of K
wall
was about 1.1 calculated for the larg-
est-size sphere. However, this equation was derived for
Stokes ow. The present experiments were carried out at
a Re number range of (10103). Fidleris and Whitmore
[32] showed that the correction goes down with increasing
Re number.
4.6. Uncertainty analysis
Let W
C
VM
be the uncertainty in the experimental results
of the C
VM
and w
1
, w
2
, . . . , w
n
be the uncertainties of the
independent variables of the C
VM
. If the uncertainties of
the independent variables are given certain practical devia-
tions, then the percentage error in the C
VM
will be given
according to Holman [33] as follows:
%
W
C
VM
C
VM

oC
VM
=o l=a
C
VM
w
l=a
_ _
2

oC
VM
=oRe
C
VM
w
Re
_ _
2
_ _
1=2
24
Substituting Eq. (19) of the single sphere into Eq. (24)
with w
Re
= 10% yields
%
W
C
VM
C
VM
AB=0:5Re
B1
ARe 25
The results of this equation are plotted in Fig. 14 where the
uncertainties are shown reducing with the increase in Re
number.
For the two spheres moving side by side, we substitute
Eq. (20) into Eq. (24) with w
l/a
= w
Re
= 10% to get the
following:
%
W
C
VM
C
VM

9=4 0:1
0:5l=a
4
0:75l=a ARe
B
l=a
4
_ _
2
_
_

ABRe
B1
0:1
0:51 1:5l=a
3
ARe
B
_ _
2
_
_
1=2
26
0.0E+00
5.0E-06
1.0E-05
1.5E-05
2.0E-05
0 10 20 30
Time, s
F
o
r
c
e
,

N
History force
Virtual mass force
Fig. 12. Values of the virtual mass force and the history force, with respect
to time, in the case of two side by side spheres in water (d = 9.4 mm).
0.0E+00
5.0E-06
1.0E-05
1.5E-05
2.0E-05
0.000 5.000 10.000
15.
000
History force
Virtual mass
force
Time, s
F
o
r
c
e
,

N
Fig. 13. Values of the virtual mass force and history force, with respect to
time, in the case of two in line spheres in water (d = 9.4 mm).
Re
VM
C
C
W
VM
%
0
0.001
0.002
0.003
0.004
0.005
100 200 300 400 500
Fig. 14. Uncertainty of the C
VM
as a function of Re number for the single
sphere.
VM
C
C
W
VM
%
a
l
0
0.5
1
1.5
2
2.5
1 3 5 7
Fig. 15. Uncertainty of the C
VM
as a function of l/a for the two spheres
moving side by side at Re = 200.
VM
C
C
W
VM
%
0.158
0.16
0.162
0.164
0.166
0.168
0.17
100 200 300 400 500
Re
Fig. 16. Uncertainty of the C
VM
as a function of Re number for the two
spheres moving side by side at l/a = 4.
A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823 821
The results of this equation are plotted in Figs. 15 and
16. The uncertainty decreases with the sphere separation
while it is increasing with Re number. The invariance
between Figs. 14 and 15 is due to the substantial uncer-
tainty associated with l/a in comparison to that of Re
number.
The uncertainty of the C
VM
for the two spheres moving
in line is obtained in an analogous method and the result is
as follows:
%
W
C
VM
C
VM

9=40:1
0:5l=a
4
3l=a ARe
B
l=a
4
_ _
2
_
_

ABRe
B1
_ _
0:1
0:5 1 3 l=a
3
_ _
ARe
B
_
_
_
_
2
_
_
1=2
27
The results of this equation are shown plotted in Figs. 17
and 18. In general, higher uncertainty is found in the C
VM
results of the in-line spheres than those of the side-by-side
spheres.
5. Conclusions
From the experimental work that was conducted in this
study, the following conclusions had been attributed:
1. In general, the virtual mass coecient was found to
decrease with increasing Reynolds number.
2. For a single sphere, C
VM
decreased slightly with increas-
ing the instantaneous Reynolds number according to
Eq. (19).
3. For two equal spheres moving side by side the value of
C
VM
> 0.5 when the spheres touched. As distance
between the spheres increasedC
VM
decreased and
approached the single sphere value of 0.5. This value
was attained when the spheres were separated by more
than 3 radii. C
VM
was correlated to the inter-particle dis-
tance and the instantaneous Reynolds number accord-
ing to Eq. (20).
4. For two equal spheres moving along the line joining
their centers, when the spheres touched, C
VM
< 0.5. As
the distance between them increased, C
VM
increased
and approached the value of 0.5 when the spheres were
separated by more than 5 radii. The proposed solution
that relatedC
VM
to the inter-particle distance and the
instantaneous Reynolds number was given in Eq. (21).
5. The magnitude of the history force was small compared
to the virtual mass force. It was seven orders of the mag-
nitude smaller at the beginning of motion.
References
[1] W.M. Hick, On the motion of two spheres in a uid, Phil. Trans. Roy.
Soc. 171 (1880) 455492.
[2] R.A. Herman, On the motion of two spheres in uid, Quart. J. Pure
Appl. Math. 22 (1887) 204216.
[3] M. Stimson, G.B. Jeery, The motion of two spheres in a viscous
uid, Proc. Roy. Soc. A111 (1926) 110116.
[4] C. Oseen, Hydrodynamik, Akademische, Leipzig, 1927.
[5] H. Lamb, Hydrodynamics, Dover, New York, 1932.
[6] L.H. Milne-Thomson, Theoretical Hydrodynamics, fth ed., Mac-
millan, London, 1972.
[7] P. Michael, Ideal ow along a row of spheres, Phys. Fluids 8 (1965)
12631266.
[8] O.V. Voinov, On the motion of two spheres in a perfect uid, PMM
33 (1969) 659667.
[9] F. Odar, W.S. Hamilton, Force on a sphere accelerating in a viscous
uid, J. Fluid Mech. 18 (1964) 302314.
[10] S. Leichtberg, S. Weinbaum, R. Pfeer, M.J. Cluckman, A study of
unsteady forces at low Reynolds number: a strong interaction theory
for the coaxial setting of three or more spheres, Phil. Trans. Roy. Soc.
282 (1976) 610858.
[11] D. Legendre, J. Magnaudet, G. Mougin, Hydrodynamic interaction
between two spherical bubbles rising side by side in a viscous liquid,
J. Fluid Mech. 497 (2003) 133166.
[12] C. Zhu, S.-C. Liang, L.-S. Fan, Particle wake eect on the drag force
of an interactive particle, Int. J. Multiphase Flow 20 (1994) 117129.
[13] R.A. Helnstine, C. Dalton, Unsteady potential ow past a group of
spheres, Comput. Fluids 2 (1974) 99112.
[14] L. Van Wijngaarden, Hydrodynamic interaction between gas bubbles
in liquid, J. Fluid Mech. 77 (1976) 2744.
[15] X. Cai, G.B. Wallis, The added mass coecient for rows and arrays
of spheres oscillating along the axis of tubes, Phys. Fluids A 5 (1993)
16141629.
[16] M.C. Ruzicka, On bubbles rising in line, Int. J. Multiphase Flow 26
(2000) 11411181.
[17] A.A. Kendoush, The virtual mass of a rotating sphere in uids, Trans.
ASME: J. Appl. Mech. 72 (2005) 801802.
VM
C
C
W
VM
%
0
2
4
6
8
10
1 2 3 4 5 6 7
a
l
Fig. 17. Uncertainty of the C
VM
as a function of l/a for the two spheres
moving in line at Re = 200.
Re
0. 176
0. 178
0. 18
0. 182
0. 184
0. 186
0. 188
0. 19
100 200 300 400 500
VM
C
C
W
VM
%
Fig. 18. Uncertainty of the C
VM
as a function of Re number for the two
spheres moving in line at l/a = 4.
822 A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823
[18] T. Kumagai, M. Muraoka, On the motion of spheres in a uid at low
Reynolds number, JSME Int. J. 32 (1989) 309316.
[19] S.A.M. Mohammed, Hydrodynamic interaction between two spheres,
Ph.D. thesis, Dept. of Chem. Eng., College of Engineering, University
of Baghdad, 2006.
[20] R. Clift, J.R. Grace, M.E. Weber, Bubbles, Drops and Particles,
Academic Press, New York, 1978.
[21] Y. Tsuji, Y. Morikawa, K. Terashima, Fluid dynamic interaction
between two spheres, Int. J. Multiphase Flow 8 (1982) 71
82.
[22] P. Alexander, High order computation of the history term in the
equation of motion for a spherical particle in a uid, J. Sci. Comput.
21 (2004) 129134.
[23] C.J. Lawrence, R.W. Mei, Long time behavior of the drag on a body
in an impulsive motion, J. Fluid Mech. 283 (1995) 307327.
[24] N. Mordant, J.-F. Pinton, Velocity measurement of a settling sphere,
Eur. J. Phys. B 18 (2000) 343352.
[25] C.F.M. Coimbra, R.H. Rangel, Spherical particle motion in har-
monic Stokes ows, AIAA J. 39 (2001) 16731682.
[26] E.E. Michaelides, Hydrodynamic force and heat/mass transfer from
particles, bubbles and drops The Freeman scholar lecture, Trans.
ASME: J. Fluids Eng. 125 (2003) 130.
[27] C. Darwin, A note on hydrodynamics, Proc. Camb. Phil. Soc. 49
(1953) 342354.
[28] J. Magnaudet, I. Eames, The motion of high Reynolds number bubbles
in inhomogeneous ows, Ann. Rev. Fluid Mech. 32 (2000) 659708.
[29] R.B. Moorman, Motion of a spherical particle in the accelerated
portion of free fall, Ph.D. dissertation, State University of Iowa,
Iowa, 1955.
[30] K. Takahashi, K. Endoh, A virtual mass and drag coecients for a
oscillating particle, J. Chem. Eng. Jpn. 25 (1992) 583685.
[31] A.A. Kendoush, The virtual mass of a growing and collapsing bubble,
AIChE J. 52 (2006) 20132019.
[32] V. Fidleris, R.L. Whitmore, Experimental determination of the wall
eect for spheres falling axially in cylindrical vessels, Brit. J. Appl.
Phys. 12 (1961) 490494.
[33] J.P. Holman, Experimental Methods for Engineers, McGraw-Hill,
New York, 1984.
A.A. Kendoush et al. / Experimental Thermal and Fluid Science 31 (2007) 813823 823

You might also like