You are on page 1of 12

MECHANICS

RESEARCH COMMUNICATIONS

Mechanics Research Communications 30 (2003) 651662


www.elsevier.com/locate/mechrescom

Drag forces of interacting spheres in power-law uids


Chao Zhu
a

a,*

, Kit Lam b, Hoi-Hung Chu b, Xu-Dong Tang c, Guangliang Liu

Department of Mechanical Engineering, New Jersey Institute of Technology, Newark, NJ 07102, USA
b
Department of Mechanical Engineering, Hong Kong Polytechnic University, Hong Kong
c
Department of Engineering Mechanics, Tsinghua University, Beijing 100084, PR China
Received 18 December 2001

Abstract
Drag forces of interacting particles suspended in power-law uid ows were investigated in this study. The drag
forces of interacting spheres were directly measured by using a micro-force measuring system. The tested particles
include a pair of interacting spheres in tandem and individual spheres in a cubic matrix of multi-sphere in ows with the
particle Reynolds number from 0.7 to 23. Aqueous carboxymethycellulose (CMC) solutions and glycerin solutions were
used as the uid media in which the interacting spheres were suspended. The range of power-law index varied from 0.6
to 1.0. In conjunction to the drag force measurements, the ow patterns and velocity elds of power-law ows over a
pair of interacting spheres were also obtained from the laser assisted ow visualization and numerical simulation.
Both experimental and computational results suggest that, while the drag force of an isolated sphere depends on the
power-index, the drag coecient ratio of an interacting sphere is independent from the power-law index but strongly
depends on the separation distance and the particle Reynolds number. Our study also shows that the drag force of a
particle in an assemblage is strongly positions dependent, with a maximum dierence up to 38%.
2003 Published by Elsevier Ltd.

1. Introduction
An extensive and rapidly increased number of industrial applications of multiphase ows involve particle
motions in non-Newtonian ows. Typical cases are exemplied by crude oil ows with rocks, sands or
natural gas; bubble entrainment and migration in plastic casting processes; polymeric ows with catalytic
particles; biouid ows in three-phase uidized beds; and aseptic processing of particulate food in liquid.
All of the above applications call for an in-depth understanding of particle dynamics in non-Newtonian
ows, especially the information of drag forces of interacting particles.
The earliest theoretical work on the force experienced by a sphere suspended in a Newtonian uid was
dated back in 1901 when Stokes (1901) rigorously derived the famous Stokes law to calculate the drag force

Corresponding author. Tel.: +1-973-642-7624; fax: +1-973-642-4282.


E-mail address: zhu@adm.njit.edu (C. Zhu).

0093-6413/$ - see front matter 2003 Published by Elsevier Ltd.


doi:10.1016/S0093-6413(03)00067-3

652

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

on a single rigid sphere in an unbounded creeping Newtonian ow. Drag forces of multi-spheres in a
viscous uid ow have been extensively investigated. Stimson and Jeery (1926) presented a complete
solution for the slow motion of two spheres parallel to their line of centers in an unbounded viscous uid
while Kynch (1958) delivered a theoretical treatment of Stokes motions of three or more spheres. Payne and
Pell (1960) solved the Stokes equations for the slow motions of several axially symmetric bodies in an
unbounded viscous uid. Zick and Homsy (1982) numerically investigated the drag forces on the simple
and periodic arrays of spheres in a Stokes ow by formulating the problem as a set of two-dimensional
integral equations. In order to account for the wall eect, Greenstein (1980) numerically computed the drag
forces on two spherical particles translating in a cylindrical tube lled with an incompressible viscous uid.
From his calculation, interaction and wall correction factors based on the distance between particles and
the distance between the particles and the tube wall were proposed.
Kaneda and Ishii (1982) pointed that, if particles are randomly located in a particulate uid system, the
most important hydrodynamic interactions are those between a pair of particles. However, in the case of a
pair of particles sedimenting vertically one above the other in an unbounded uid, the dierence between
the forces on the leading and trailing spheres cannot be explained by an analysis based on the Stokes
equation. Hence, Kaneda and Ishii further proposed an asymptotic analysis to treat the hydrodynamic
interaction of two spheres moving in an unbounded uid at small but nite Reynolds number, in which the
inertia eect was taken into account. However, these treatments can only be applied to cases with large the
separation distance between particles in small Reynolds numbers. It is realized that the typical distance of
strong interaction of a pair of particles is less than twice particle diameter (Zhu et al., 1994). In addition, for
most applications in particulate multiphase ows, the particle Reynolds number based on the isolatedparticle terminal velocity and particle diameter is typically in a range from tens to several hundreds. For
example, at a particle Reynolds number from 10 to 200, the corresponding sizes of glass beads vary from
230 lm to 1.1 mm in water and from 130 to 650 lm in air. Hence, the Reynolds number range of practical
signicance to a multiphase ow system may be far beyond the Stokes regime. For such a case, inertia eect
and wake eect must be taken into account.
Happel and Pfeer (1960) experimentally investigated on the two spheres falling in viscous liquids. They
observed that the terminal velocity of two spheres was greater than that of an isolated sphere, which
suggests that the drag force of any one of the two spheres is reduced by the sphere interaction. Realizing the
importance of the direct measurements of drag force of interacting particles, Rowe and Henwood (1961),
Lee (1979) and Tsuji et al. (1982) used the pendulum method and water channel ow to measure the drag
force at a particle Reynolds number ranged from 500 to 10,000. However, the data were so scattered that
only a general trend of the interactions could be reected. Zhu et al. (1994) developed a micro-force
measuring system to directly measure the drag forces on two interacting particles at a particle Reynolds
number varying from 20 to 130. It was found that the particle Reynolds number aects not only the
magnitude of the drag forces of an interacting particle but also its variation with the separation distance. By
using Zhus experimental technique, Liang et al. (1996) and Chen and Wu (2000) have also measured the
drag forces of interacting spheres in Newtonian uids.
The early study of particle dynamics in non-Newtonian ows can be dated back in early 1960s. However,
the study of particulate non-Newtonian multiphase ows was not very active until 1990s. Slattery and Bird
(1961) performed terminal velocity experiments of single spheres in various sized tubes lled with carboxymethycellulose (CMC) solutions to determine the drag coecients of a single sphere in such kinds of pipe
ows. Acharya et al. (1976) presented an approximated momentum integral boundary layer analysis to
determine the drag coecient of a slow moving sphere in the creeping ow regime through a power-law
non-Newtonian uid in the presence of a at wall. An active yet extensive study of a single sphere in various
non-Newtonian uids can be found in several recent reports (Briscoe et al., 1993; Graham and Jones, 1994;
Ribeiro et al., 1994; Machac and Lecjaks, 1995). Settling of a single but non-spherical particle in a nonNewtonian uid was also reported (Madhav and Chhabra, 1994; Feng et al., 1995).

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

653

Similar to Newtonian multiphase ows, groups of multi-particles do not behave like those of isolated
particles unless the ow is extremely dilute. Hence the study of the multi-particle group behavior and
hydrodynamic interactions among particles is of great importance. Kawase and Ulbrecht (1981) used a free
surface model and the boundary layer theory to estimate the motion of an assemblage of spheres moving in
a power-law non-Newtonian uid at high Reynolds numbers. Staish and Zhu (1992) and Jaiswal et al.
(1993) numerically investigated unbounded slow ows of a power-law non-Newtonian uid through an
assemblage of spheres. Considering the eect of wall and the ow disturbance induced by upstream and
downstream particle groups, Subramaniam and Zuritz (1994) determined the drag force on multiple assemblies of spheres suspended in CMC solutions, from which an averaged drag force on each particle was
deduced. However, this averaged drag force should not be used in the calculation of dynamic motions of
particles with strong wake interactions. As measured by Zhu et al. (1994), for a pair of particles, the drag
force of the trailing particle can be less than 20% of that of the leading particle. An averaged drag force of
the two would lead to a biased account of the dynamic motion of the pair, especially of the trailing particle.
So far, there appears no report on direct drag force measurements of a single particle inside an assemblage
in Newtonian and/or power-law ows. In addition, due to the strong particleparticle interaction, it is
unlikely to have a stable structure of particle assemblage in actual ows. Hence, a direct drag force
measurement of a single particle in various positions inside an assemblage would be of interest.
The objectives in this study were to investigate the drag forces of interacting particles suspended in
power-law ows and to understand the eect of power-law index on the drag forces. These objectives were
approached by both direct measurements and numerical computations on drag forces of interacting particles suspended in power-law pipe ows. In conjunction to the drag force measurements, ow patterns and
velocity elds of power-law ows over a pair of interacting spheres were also obtained from laser assisted
ow visualization and numerical simulations.

2. Experimental methodology
In order to perform the direct measurements of drag forces and ow velocity elds, an experimental
setup was built up, which consisted of a ow-circulating loop; a micro-force measuring system; and a PIV
ow visualization and velocity measurement system, as shown in Fig. 1. In addition, selection and calibration of easy-to-make, not highly viscous, transparent, simple (in viscosity formulation), and time-independent non-Newtonian uids were also crucial to this experiment.
The ow-circulating loop aims to provide a steady and isothermal ow for designated measurements
without time restraint. The loop includes a gear pump (Oriental, Model GB-200); two controlling valves; a
temperature control system; a ow reduction section atop of which the test section (a vertical rectangular
column) is located; a uid reservoir; and an overow end section with a ow return pipe for reducing the
bubble generation. Hence, a steady upward ow is yielded in the vertical test section where the interacting
spheres are suspended. Various ow velocities in the test section can be obtained by adjusting the controlling valves.
The drag force of an interacting particle can be directly measured using the micro-force measuring
system, as illustrated in Fig. 2. The test particle (whose drag force is to be measured) is connected to a frame
with a thin rod and the frame is directly put on top of a precision electronic balance. Under the ow
condition, the drag force on the test particle and the connecting rod leads to a reduction in the loading
weight of the frame on the balance, as shown in Fig. 2(a). Hence, by deducting the drag force of the rod, the
drag force of the test particle can be obtained. In this experiment, a precision top-loading electronic balance
(Sartorius, Model LC211S) with the accuracy of 1 mg and response frequency of 2 Hz is used. The balance
is further connected to an IBM PC586 for data acquisition and analysis. It must be noted that the drag
force on the thin rod which is partially submerged into the uid is not negligibly small (about 2540% of the

654

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

Fig. 1. Schematic diagram of experimental apparatus.

Balance

Stainless
Steel frame

(a)

(b)

Fig. 2. Micro-force measuring system.

total drag in our experiments). Therefore, in order to yield the drag force of the particle, drag forces of the
rod without particle connected must be independently measured, as shown in Fig. 2(b). Schematic description of drag force measurements for cases where the test particle is located inside or in the downstream
region of a cubic matrix of multi-particle assemblage is provided in Fig. 3.
Flow visualizations and velocity eld measurements of ows over a pair of interacting particles are
carried out by using an optical measuring system, as depicted in Fig. 1. The laser sheet, for the purpose of
illuminating the measurement area of interest, is generated by a laser beam sweeping unit whose sweeping
range and frequency are governed by a scanning beam box and controller. In this study, the sweeping range
is 150 mm and frequency is up to 2000 Hz. The laser sheet thickness, adjustable by the collimating optics,

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

655

Fig. 3. Cubically arranged multi-particle assemblage.

can be as thin as 1 mm. The image capturing and analysis system consists of a digital CCD camera with a
resolution of 1k by 1k, an image processing system (Matrox), an IBM PC586, and the PIV software
(Optical Flow Systems, VidPIV) for velocity eld analysis.
In this experiment, the non-Newtonian uids are desired to be easy-to-make, not highly viscous,
transparent, simple (in viscosity formulation), and time-independent. A reasonably low viscosity of the uid
is preferred because of the limitations in the available power of the gear pump for ow circulation.
Transparency of uid is required for the sake of ow visualization and optical measurement techniques
involved. Simplicity in uid viscosity formulation is desired for the benet of simple mathematical formulations in numerical computations. The time-independence is sought due to the requirement in repeatability of experimental measurements and time duration needed for taking the measurements. The
simplest non-Newtonian uids are power-law uids. For convenience, CMC uids that are transparent,
power-law, shear thinning and time-independent are selected. The CMC solutions of various concentrations are prepared by mixing and dissolving dierent amounts of dry CMC powders into water. A powerful
blender (Toshiba, Model BMP-150B) is used for the mixing and a precision viscometer (Brookeld, Model
LDV II+) is used for the determination of power-law index and consistency coecient of a CMC uid.

3. Fundamentals of numerical modeling


Numerical simulations of power-law ows over a pair of interacting spheres suspended on the axis in a
circular duct are performed to provide theoretical explanation and extend understanding of the experimental results. These simulations are aimed at the calculations of drag forces of the spheres and ow eld
under the particle interactions. The calculated drag forces and ow elds are then compared with the
experimental measurements and visualizations.
In the numerical modeling approach, it is assumed that the ows are laminar, steady, incompressible,
and axisymmetric. The body force, i.e., gravity, is omitted because its eect has already been deducted from
the experimental measurements. The constitutive relationship between the shear stress and the rate of shear
deformation for one-dimensional shear ow of power-law liquids is determined from the viscometer
measurements as follows:
s kc_

where s and c are shear stress and the shear strain respectively, while k and n are consistency coecient and
power-law index respectively. The above scalar expression can be expressed by (William, 1978)

656

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

Fig. 4. Grid distributions.

s lA k

1
trA2
2

n1=n
2

where s now represents the shear stress tensor and A is the shear strain tensor, and trA2 denotes for the
trace of the dot product of A and A.
For steady, incompressible, axisymmetric and laminar ows, the continuity and momentum equations
are expressed as
*

r  qV 0
**

3
$

r  qV V rp r  s f

4
*

where q presents the density of uid, V is the ow velocity, p is the pressure, and f implies the body force.
Computational domain and grid distributions for numerical simulations are illustrated in Fig. 4. For the
convenience of computation, the entire domain consists of four sub-domains, respectively for entrance, exit,
near eld, and far eld (wall side). In the near eld sub-domain, the boundary tting grids are used to
match perfectly the contour of the interacting spheres. The inlet velocity distributions are set either according to experimental measurements (if available) or assuming fully developed ow conditions. At the
exit, it is always assumed that the ow reaches fully developed. On the central line, the axisymmetric
condition is used while non-slip condition is applied to the surfaces of spheres. On the pipe wall, the nonslip condition is applied for the simulation of pipe ows while slip condition is adopted when simulating
unbounded ows (with a large ratio of pipe diameter to sphere diameter). It is noted that, if only the ratio of
drag coecient concerned, the computational results are weakly dependent on the size of computational
domain and inlet velocity distributions. Hence, the results obtained could be extended to more general
applications. From the non-dimensional governing equations and their boundary conditions, it is shown
that there are only three major parameters aecting on the ratio of drag forces, namely, the particle
Reynolds number, the power-law index and the relative separation distance between the two interacting
particles. The eects of these parameters on the ratio of drag forces will be discussed later.

4. Results and discussions


In this study, investigations of drag forces of interacting spheres in power-law ows were conducted via
both direct measurements and numerical simulations. The power-law ow range of the experimental investigation is considerably limited due to two primary restraints, which are the limited pump power and the
problems of bubble entrainment at high velocity. The bubble entrainment from various free surfaces in a
highly viscous uid ow would create a serious problem to all experimental measurements concerned in this
study. For example, the bubble attachment to the test spheres yields various errors in drag force mea-

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

657

surements. The bubble-contained uid would cause a reduced and uncontrollable apparent uid viscosity,
which may signicantly change the ow conditions of the experiment. In addition, bubbles would strongly
aect the PIV measurements, even make the ow visualization impossible. To make things worse, in the
ow circulation process, the pump would further break up the large bubble into many tiny bubbles that are
hard to get rid of. As a result, the experiments were performed using CMC solutions with the range of
power-law index from 0.9 to 1.0 and consistency coecient from 0.089 to 0.375 Pa sn . The ow velocity
varies from 0.009 to 0.07 m/s in the test section, which results in a particle Reynolds number range from 0.7
to 23. In order to expand the investigation range and to provide a theoretical comparison, a numerical
study (which can overcome such technical diculties in experiments) of power-law ows over a pair of
interacting spheres in tandem was also performed, from which drag forces of interacting spheres were
calculated. In the numerical computation, the power-law index varies from 0.6 to 1.0.
The comparison of drag coecient between our experimental results and reported results (Chhabra,
1990; Anubav et al., 1994) for an isolated sphere in power-law ows is presented in Fig. 5. It is shown that
the experimental results are bounded within the standard curve of n 1 (Newtonian ows) and the numerical computational curve of n 0:6. As consistent with the literature results, the eect of power-law
index appears to be strong and is particle Reynolds number dependent. The drag coecients of power-law
uids are higher than that of Newtonian ows when Re is less than a critical Reynolds number (say,
Re 23) while the drag coecient are less than that of Newtonian ows when Re is beyond the critical
value. The higher the power-index deviation, the larger the dierence.
Drag force measurements of a pair of interacting spheres in tandem were carried to illustrate the eect of
power-law index on the particleparticle interaction. As shown in Fig. 6, the drag reduction of the trailing
sphere due to the particle wake interaction is exponentially correlated to the separation distance and the
particle Reynolds number. It is interesting to further examine the eect of power-law index on this drag
reduction. At the similar Re conditions, little eect of power-law index on the curve(s) of drag coecient
ratio of the trailing sphere is observed, as illustrated in Fig. 7. However, it should be realized that, due to

Fig. 5. Single sphere drag coecient.

658

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

Fig. 6. Drag coecient ratio versus separation distance.

Fig. 7. Eect of power-law index on drag coecient ratio.

the experimental limitations, the variation range of power-law index of the experiments might be too
narrow to yield conclusive results. In order to oset this experimental limitation, a wider range of investigation on power-law index is completed through the numerical simulation. As shown in Fig. 7, the numerical results with the power-law index from 0.6 to 1.0 also suggest that the drag coecient ratio is very
weakly dependent on the power-law index. Hence, for engineering application purposes, it may be deduced
that the ratio of drag coecient in power-law ows is exponentially dependent on the separation distance
and is strongly related to the particle Reynolds number but nearly independent of the power-law index
within the range of our investigation.
As mentioned before, there appears no report on direct drag force measurements of a single particle
inside an assemblage in Newtonian and/or power-law ows. In addition, to reveal the unstable nature of the
assemblage structure in actual ows, a direct drag force measurement of a single particle in various positions inside an assemblage is of interest. Drag forces of a single particle at various positions inside or in
the downstream of a cubic assemblage (refer to Fig. 3) are presented in Fig. 8. It is shown that the drag
force of a particle in an assemblage is position dependent. Similar to the case of two-particle interaction, the

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

Fig. 8. Drag coecient of a sphere in an assemblage: (a) inside an assemblage and (b) downstream of an assemblage.

Fig. 9. Velocity eld from PIV measurement.

659

660

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

Fig. 10. Comparison of velocity eld between PIV and simulation.

drag force reaches the minimum at contact with the leading particle. There exists a position where the drag
force of the particle inside the assemblage reaches the maximum due to the combined inuence of both
leading particle and downstream particle. Within the range of the experiment, the maximum dierence
among drag forces at various positions is up to 38% in Fig. 8(a) and up to 30% in Fig. 8(b). This signicant
dierence in drag forces of a particle at various positions inside an assemblage clearly illustrates that the
drag forces of particles in a transient particle group in an actual ow could be quite dierent from the
averaged drag force at a node of an assemblage. Hence, an accurate account of transient dynamic
movements of particles in multiphase ows should be based on the transient drag force information rather
than based on the averaged drag force.
In this study, ow visualization, PIV measurements and numerical computation are attempted all together to investigate the ow led of a pair of interacting spheres in tandem. Fig. 9 illustrates a set of typical
ow patterns in the wake region between two interacting particles with various interacting distances. At low
particle Reynolds numbers, no signicant ow separation or vortex formation is observed in the wake
region. However, velocities in the wake region are greatly reduced. It is noted that when the trailing particle
is away from the leading particle, the velocity between the two interacting particles increases. This partially
explains why the drag force of trailing particle increases with the increase of separation distance. The
comparison between the PIV measurements and numerical simulations in power-law ows, as given in Fig.
10, shows the same velocity tendency in the ow eld.

5. Conclusions
In the present study, drag forces of interacting particles at the particle Reynolds number from 0.7 to 23
were directly measured in ows with power-law index from 0.9 to 1.0. In order to extend the investigation
range of power-law index, numerical simulation method is also performed for ows with power-law index
from 0.6 to 1.0. Both experimental and numerical studies suggest that the ratio of drag coecients in
power-law ows is exponentially dependent on the separation distance and is strongly related to the particle
Reynolds number but nearly independent of the power-law index within the range of our investigation.

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

661

Drag forces of a single particle at node location inside a simple cubic assemblage are directly measured.
This drag force can be regarded as averaged drag force of particles in the assemblage. However, the drag
force measurements of a single particle at various locations inside the assemblage illustrate that the drag
force of a particle in an assemblage is strongly positions dependent, with a maximum dierence up to 38%.
This signicant position dependency suggests that the drag forces of particles in actual transient structure of
particle assemblage may be quite dierent from the averaged drag force at the node. Consequently, advanced modeling of transient dynamic movements of particles in multiphase ows should be based on the
transient drag information rather than based on the averaged drag force of an assemblage.
Study on ow eld of interacting spheres by ow visualization and numerical computation indicates that
the eect of power-law index on ow pattern and hence drag force may be weak, which is in agreement with
our drag force measurements.

Acknowledgements
The authors wish to thank the Research Committee and Department of Mechanical Engineering at the
Hong Kong Polytechnic University for their support. This project was substantially supported by a grant
from the Research Grants Council of the Hong Kong Special Administrative Region (Project No. 350/692).
Advisory comments from Professor Z.C. Zhu at Tsinghua University are also deeply appreciated.

References
Acharya, A., Mashelkar, R.A., Ulbrecht, J., 1976. Flow of inelastic and viscoelastic uids past a sphereI. Drag coecient in creeping
and boundary-layer ows. Rheol. Acta 15, 454.
Anubav, T., Rajendra, P.C., Thirumalachari, S., 1994. Power law uid ow over spheroidal particles. Ind. Eng. Chem. Res. 33, 403.
Briscoe, B.J., Luckham, P.F., Ren, S.R., 1993. The settling of spheres in clay suspensions. Powder Technol. 76, 165.
Chen, R.C., Wu, J.L., 2000. The ow characteristics between two interactive spheres. Chem. Eng. Sci. 55, 1143.
Chhabra, R.P., 1990. Motion of spheres in power law (viscoinelastic) uid at intermediate Reynolds numbers: A unied approach.
Chem. Eng. Process. 28, 89.
Feng, J., Joseph, D.D., Glowinski, R., Pan, T.W., 1995. A three-dimensional computation of the force and torque on an ellipsoid
settling slowly through a viscoelastic uid. J. Fluid Mech. 283, 1.
Graham, D.I., Jones, T.E.R., 1994. Settling and transport of spherical particles in power-law uids at nite Reynolds number. J. NonNewton. Fluid Mech. 54, 465.
Greenstein, T., 1980. Interaction and wall corrections for the slow motion of two uid or solid particles parallel to the axis of a circular
cylinder through a viscous uid. J. Mech. Eng. Sci. 22, 243.
Happel, J., Pfeer, R., 1960. The motion of two spheres following each other in a viscous uid. AIChE J. 6, 129.
Jaiswal, A.K., Sundararajan, T., Chhabra, R.P., 1993. Hydrodynamics of creeping ow of power law uids through particle
assemblages. Int. J. Eng. Sci. 31, 293.
Kaneda, Y., Ishii, K., 1982. The hydrodynamic interaction of two spheres moving in an unbounded uid at small but nite Reynolds
number. J. Fluid Mech. 124, 209.
Kawase, Y., Ulbrecht, J.J., 1981. Motion of and mass transfer from an assemblage of solid spheres moving in a non-Newtonian uid at
high Reynolds numbers. Chem. Eng. Commun. 8, 233.
Kynch, G.J., 1958. The slow motion of two or more spheres through a viscous uid. J. Fluid Mech. 5, 193.
Lee, K.C., 1979. Aerodynamic interaction between two spheres at Reynolds numbers around 104 . Aerosp. Q. 30, 371.
Liang, S.-C., Hong, T., Fan, L.-S., 1996. Eects of particle arrangements on the drag force of a particle in the intermediate ow regime.
Int. J. Multiphase Flow 22 (2), 285.
Machac, I., Lecjaks, Z., 1995. Wall eect for a sphere falling through a non-Newtonian uid in a rectangular duct. Chem. Eng. Sci. 50,
143.
Madhav, G.V., Chhabra, R.P., 1994. Settling velocities of non-spherical particles in non-Newtonian polymer solutions. Powder
Technol. 78, 77.
Payne, L.E., Pell, W.H., 1960. The stokes ow problem for a class of axially symmetric bodies. J. Fluid Mech. 7, 529.

662

C. Zhu et al. / Mechanics Research Communications 30 (2003) 651662

Ribeiro, G.S., Vargas, A.S., Frota, M.N., 1994. Pressure drop induced by a sphere settling in non-Newtonian uids. Int. J. Multiphase
Flow 20, 355.
Rowe, P.N., Henwood, G.A., 1961. Drag forces in a hydraulic model of a uidised bedPart I. Trans. Instn. Chem. Engrs. 39, 43.
Slattery, J.C., Bird, R.B., 1961. Non-Newtonian ow past a sphere. Chem. Eng. Sci. 16, 231.
Staish, M.G., Zhu, J., 1992. Flow resistance and mass transfer in slow non-Newtonian ow through multiparticle systems. J. Appl.
Mech. 59, 431.
Stimson, M., Jeery, G.B., 1926. Proc. Roy. Soc. A 111, 110.
Stokes, G.G., 1901. Mathematical and Physical Papers III. Cambridge University Press, Cambridge, UK. p. 55.
Subramaniam, G., Zuritz, C.A., 1994. drag on individual cubic assemblies of spheres in non-Newtonian tube ow. Can. J. Chem. Eng.
72, 201.
Tsuji, Y., Morikawa, Y., Terashima, K., 1982. Fluid dynamics interaction between two spheres. Int. J. Multiphase Flow 8, 71.
William, R.S., 1978. Mechanics of Non-Newtonian Fluids. Pergamon Press.
Zhu, C., Liang, S.-C., Fan, L.-S., 1994. Particle wake eects on the drag force of an interactive particle. Int. J. Multiphase Flow 20,
117.
Zick, A.A., Homsy, G.M., 1982. Stokes ow through periodic arrays of spheres. J. Fluid Mech. 115, 13.

You might also like