Professional Documents
Culture Documents
Estimation of Hindered Settling Velocity of Suspensions
Estimation of Hindered Settling Velocity of Suspensions
A R T I C L E I N F O A B S T R A C T
Article history: Four effective-medium models (EM-I, II, III, IV) are utilized and compared for determining hindered
Received 13 May 2008 settling velocity of equi-sized particles in a viscous fluid. Among the models, EM-IV model is found to
Accepted 3 August 2008 accurately predict the effective viscosity and the hindered settling velocity of monodisperse
suspensions. In EM-IV model which was developed for determining the diffusivity of proteins in a
Keywords: biological membrane by Dodd et al. [T.L. Dodd, D. A. Hammer, A.S. Sangani, D.L. Koch, J. Fluid Mech. 293
Hindered settling velocity (1995) 147], the effective-medium region begins at the distance R = a[(1 S(0))/f]1/3 from the origin
Sedimentation
where the center of the test particle is located, where a is the radius of the particle, f is the volume
Suspension
fraction of the particles in the suspension, and S(0) is the zero wavenumber limit of the structure factor.
Effective-medium theory
Structure factor The estimations by EM-IV model agree very well with the exact calculations and the experimental
observations. The hindered settling velocity U of the particles is given, in Richardson–Zaki form, by U/
U0 = (1 f)5.5, where U0 is the settling velocity for an isolated particle.
ß 2009 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights
reserved.
1. Introduction interactions have been presented by Brady and Bossis [5], Ladd [6],
and Mo and Sangani [7]. Their calculation schemes, however, are
Sedimentation of suspensions has been widely used for too complicated to be available for further or other relevant
separating the particles from the fluid in a suspension. In particular analysis. Therefore it is necessary to develop a simple theory which
the sedimentation of suspensions in very low particle Reynolds also matches with the rigorous calculations. The theory should also
number based on the radii and the average settling velocity of the properly accounts for important physics of the multiparticle
particles has received the greatest attention. A major issue in the interactions. Effective-medium theories have been developed for
study of the sedimentation is to determine the settling velocity of predicting the transport properties of suspensions or composite
the particles in suspension as a function of the particle materials consisting of random arrays of particles in a fluid or a
concentration. When an isolated particle falls through a viscous medium. The advantage of effective-medium theories is to use the
fluid due to gravity, the terminal velocity of the particle follows concept of ensemble average, that is, the average over all
Stokes settling velocity. In case that many particles settle, each configurations of all particles. A variety of effective-medium
particle is hindered by other particles and the settling of each models have been proposed for determining various transport
particle is influenced by the existence of other particles. This properties of physical or biochemical processes. Some examples
influence, called multiparticle hydrodynamic interaction, is are the cases of determining the diffusivity of integral proteins in a
reached to quite long particle–particle distance. The evaluation bilipid membrane system [11], the thermal conductivity of
of the multiparticle effect requires complicated computations composite materials [12], and the mass transfer coefficient in a
whose cost increases with the number of particles. Moreover the hollow-fiber membrane system [13].
calculation should satisfy the randomness of particle distribution Some of these effective-medium models are utilized in the
by taking averages for quite large number of particle configura- present study for determining hindered settling velocity of
tions. Since Batchelor’s [1] work, there have been many monodisperse suspensions. Specifically four effective-medium
investigations to determine the sedimentation velocity using models (EM-I, II, III, IV) are used and the predictions of the
microscopic analysis of the suspension [2–10]. Focus has been on hindered settling velocity using these models are compared with
developing numerical methods for accurately computing the rigorous calculation results and experimental observations.
multiparticle interactions over wider range of concentrations of
particles. Remarkable theoretical calculations for the multiparticle 2. Theory and calculations
1226-086X/$ – see front matter ß 2009 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.jiec.2008.08.013
46 S. Koo / Journal of Industrial and Engineering Chemistry 15 (2009) 45–49
origin
where P(x0 j0) is the probability density for finding a particle with
R distance from the origin to the effective-medium
its center at x0 given the presence of a particle at the origin. Note
boundary
that x approaches f as r becomes infinity. For suspensions with
S(0) structure factor at zero wavenumber limit for hard- isotropic pair probability density a closure relation for the stress is
sphere suspension introduced:
U hindered settling velocity of particles in a fluid
U0 settling velocity of an isolated particle in a fluid hsi1 ¼ h pi1 I þ mðxÞ b rhuic þ ðrhuic ÞT c (5)
u velocity of the fluid
where hpi1 is the conditionally averaged pressure and m(x) is
x position vector
viscosity of the suspension at x. And I is the isotropic unit tensor.
The conditionally averaged pressure and velocity are required to
Greek letters
approach, respectively, the unconditional averages as r becomes
x phase indicator function of particle very large. The average sedimentation velocity of the particles
hxi1 pair probability density, conditional ensemble equals the conditionally averaged velocity evaluated at r = 0.
average of x Effective-medium approximations are developed to determine
f volume fraction of particles the conditionally averaged velocity. Schematic diagrams for the
m(x) viscosity at position x in suspension approximate models are shown in Fig. 1. One simple approxima-
mf viscosity of fluid tion is to model the suspension as the representative (test) particle
m* effective viscosity of suspension at origin and the effective-medium which occupies just outside of
r(x) density at position x in suspension the particle [14]. In this model the region for r > a, r being the
distance from the origin, is considered as the effective-medium.
rf density of fluid
We denote the distance from the origin as R from which the
rp density of particle
effective-medium begins. This model has been proposed by Hill
rs density of suspension [14] for the analysis of composite materials and is referred to as
s stress tensor EM-I in the present study.
Another choice for R can be af1/3. This model is called EM-II
here. In this model a representative particle at origin is surrounded
satisfies: by clear fluid up to the distance R = af1/3 beyond which particle–
r s þ rfg ¼ 0 (1) fluid system is considered as an effective-medium [15]. Hence, for
example, viscosity m is taken to equal the suspending fluid
where s is the stress at point x in the fluid, rf the density of the viscosity mf in the region a < r < R and equal to effective viscosity
fluid, and g the gravitational acceleration. The stress inside the m* of the effective-medium for r > R. Similarly, hxi1 is taken equal
Fig. 1. Schematic diagrams for effective-medium models (R = a for EM-I, R = af1/3 for EM-II, R = a[(1 S(0))/f]1/3 for EM-IV).
S. Koo / Journal of Industrial and Engineering Chemistry 15 (2009) 45–49 47
R3 f ¼ ð1 Sð0ÞÞa3 (9)
where
Z
Sð0Þ ¼ ½Pðxj0Þ PðxÞdV x (10)
3. Results and discussions radius of test particle at origin is chosen as the distance R in EM-I.
This model does not seem to account for the multiparticle effects
The calculation of sedimentation velocity requires the informa- properly. In EM-II and IV, R varies with f. The EM-II model assumes
tion of the effective viscosity of the suspensions. The effective- that the particles are well separated and is further modified to EM-
medium models are used for the calculation of the effective IV which additionally takes the volume exclusion effect due to the
viscosity first. The results are presented in Fig. 3. The estimations presence of the test particle at origin into account. And EM-III
by EM-II, III and IV reconcile with each other and are in excellent model is conceptually closer to rigorous calculations since it is
agreement with the rigorous calculations by Mo and Sangani [7]. based on direct calculation of the conditional average of the
Meanwhile, EM-I yields higher values of the effective viscosity indicator function instead of using relatively simple distinction of
than the others and its predictions are aphysical beyond f of 0.4. the effective-medium region as in EM-I, II, and IV. However EM-III
The difference among the EM model I, II, and IV is the definition of requires somewhat cumbersome calculation for the integration of
the distance R where the effective-medium region begins from the radial distribution function. The predictions by EM-II, III and IV also
origin (r = 0). Fig. 4 shows the distance R for the effective-medium agree with the rigorous calculation results by Mo and Sangani [7]
as a function of the volume fraction f. As mentioned earlier, the and Ladd [6]. It seems that simple EM-II model is sufficiently
accurate for estimating the effective viscosity of the suspensions as
a function of particle volume fraction. However, this trend does not
hold for the case of sedimentation.
The calculation results for the sedimentation velocity of
suspensions with the information of the effective viscosity are
given in Fig. 5. It is shown that EM-II model under-predicts the
sedimentation velocity unlike the effective viscosity problem and
the estimations by EM-III and IV agree well with the rigorous
calculations by Mo and Sangani [7]. It is found that the volume
exclusion effect due to the presence of the test particle at origin
becomes significant in the sedimentation process. It means that
EM-IV model properly reflects the multiparticle interaction which
is largely dependent on the distribution of particles. Therefore it is
seen that EM-IV model is the most suitable for estimating the
sedimentation velocity.
Comparison is also made with previous empirical results. Fig. 6
compares the estimations by EM-IV and the rigorous calculations
with the experimental results by Cheung et al. [20] and Hoyos et al.
[21]. Cheung et al. [20] measured the sedimentation velocities of
polystyrene beads settling in a viscous fluid using NMRI (nuclear
magnetic resonance imaging) method. The particles were of three
mean diameters (274, 655, and 488 mm). The experiment was
separately performed for each size of the particles. The authors
collected the data for each size of the particles in Fig. 2 of their
paper. Hoyos et al. [21] also provided the experimental data for the
sedimentation velocity of glass beads of about 80 mm in diameter
Fig. 3. Normalized effective viscosity m/m0 as a function of f.
U 5:5
¼ ð1 fÞ (18)
U0
4. Summary