Bulk Density

Published on January 2017 | Categories: Documents | Downloads: 40 | Comments: 0 | Views: 297
of 2
Download PDF   Embed   Report

Comments

Content

Journal of Colloid and Interface Science 302 (2006) 702–703
www.elsevier.com/locate/jcis

Note

The ‘Einstein correction’ to the bulk viscosity in n dimensions
Aditya S. Khair
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125, USA
Received 16 June 2006; accepted 30 July 2006
Available online 2 August 2006

Abstract
We calculate the effective bulk viscosity of a dilute dispersion of rigid n-dimensional hyperspheres in a compressible Newtonian fluid at zero
Reynolds number.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Colloidal dispersion; Bulk viscosity; Hypersphere; Compressible fluid; Low-Reynolds-number flow; Effective properties; Two-phase material

1. Introduction
Recently, Brady et al. [1] computed the bulk viscosity (also
known as the second or volume viscosity) of a dilute colloidal
dispersion to O(φ 2 ) in the volume fraction φ of the (rigid spherical) suspended particles. This calculation requires determination of the dispersion microstructure, which reflects a balance
between an imposed uniform expansion flow and Brownian diffusion of the particles. The O(φ 2 ) contribution to the bulk viscosity arises from two-particle interactions, whereas the O(φ)
contribution is due to the disturbance flow generated by a single particle in the expansion flow. We call the O(φ) contribution
the ‘Einstein correction,’ in homage to Einstein [2] who computed the O(φ) contribution to the shear viscosity of a dilute
colloidal dispersion of rigid spheres. In this Note, we compute
the Einstein correction to the bulk viscosity for a dispersion of
n-dimensional hyperspheres. The correction for n = 2 represents the main outcome of this work; it gives the bulk viscosity
for a dilute dispersion of two-dimensional rigid cylinders—
a result of practical significance. Furthermore, the general ndimensional problem provides a useful exercise for students of
transport processes and fluid dynamics, by introducing basic
concepts in suspension mechanics, while, at the same time, offering a glimpse into an active research area.

E-mail address: [email protected].
0021-9797/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcis.2006.07.076

2. Analysis
We begin with the constitutive equation for the stress σ in a
n-dimensional compressible Newtonian fluid (see, e.g., Batchelor [3]), viz.


2
σ = −pth I + 2ηe + κ − η (∇ · u)I ,
(1)
n
where pth is the thermodynamic pressure (as defined by the
fluid’s equation of state), e = 12 [∇u + (∇u)† ] is the rate of
strain tensor with u the fluid velocity, η is the shear viscosity,
κ is the bulk viscosity, and I is the identity tensor. In zeroReynolds-number flow the velocity field can be decomposed
into an uniform expansion everywhere in the fluid and a disturbance flow created by any immersed particles, which satisfies
the usual incompressible Stokes equations; that is,
1
u = er + us ,
n
where r is the position vector, e is the expansion rate, ∇ · u = e,
and ∇ · us = 0. In turn, the fluid stress σ can be split into a contribution due to the uniform expansion flow, σ e = (κe − pth )I ,
and a disturbance stress σ s = −p s I + 2ηes , with ∇ · σ s = 0.
Here, p s is the dynamical pressure field of the incompressible
Stokes flow.
Consider a force- and torque-free n-dimensional hypersphere of radius a immersed in the uniform expansion flow.
Exploiting the linearity of the Stokes equations, and noting the

A.S. Khair / Journal of Colloid and Interface Science 302 (2006) 702–703

fact that the only vector present is r, one immediately concludes
that the (harmonic) disturbance pressure p s is identically zero.
Thus, the disturbance flow us is also a (vector) harmonic function and is given by
 
1 a n
us = − e
r,
n r
where r = |r|. Note, in one dimension (n = 1) the disturbance
flow is spatially constant, as required by the incompressibility
condition, ∇ · us = 0.
To calculate the effective bulk viscosity of a dilute suspension of hyperspheres we form the volume average of the Cauchy
stress tensor σ (Brady et al. [1]) to obtain


2
¯
σ  = −pth f I + 2ηe + κ − η ∇ · uI + NS,
(2)
n
where . . . denotes an average over the entire dispersion (particles plus fluid), . . .f is an average over the fluid phase only,
N¯ is the particle number density, and the average extra
particle
stress is a number average defined by S = (1/N) N
α=1 S α ,
where the contribution from particle α is given by

1
Sα =
(3)
(rσ · n + σ · nr) dSα ,
2
∂Sα

with n the unit normal pointing out of the particle. (Note,
Eq. (3) is applicable to rigid particles only; a more general
expression for the extra particle stress—valid for drops and
bubbles in addition to rigid particles—is given by Eq. (3.1) of
Brady et al. [1].)
The effective bulk viscosity, κeff , relates the deviation of the
trace of the average stress from its equilibrium (e = 0) value to
the average rate of expansion e, namely,
 eq 
1
σii  − σii ,
(4)
n
where the summation convention is applied to repeated indices.
The trace of the average stress is


¯ ii .
σii  = n κe − pth f + NS
(5)
κeff e =

The trace of average extra particle stress is calculated as

Sii  = ri σij nj dS



2(n − 1)
dS
ηea
= (κe − pth )a +
n

703




2π n/2
2(n − 1)
= (κe − pth )a +
ηea a n−1
,
n
Γ (n/2)

(6)

where Γ (z) is the Gamma function. Thus, from (5) we have
σii  = n(κe − pth ) + 2(n − 1)ηeφ,

(7)

where we have used e = e(1 − φ) and pth f = pth (1 − φ),
and the volume fraction of hyperspheres is defined by
φ=

a n 2π n/2 ¯
N.
n Γ (n/2)

From (4), the effective bulk viscosity is found to be


2(n − 1)
1
κeff = κ +
ηφ
n
1−φ
2(n − 1)
∼κ +
ηφ as φ → 0.
n

(8)

(9)

The case n = 3 gives a correction of 43 ηφ, as reported by Brady
et al. [1]. And for n = 2—two-dimensional rigid cylinders—the
correction is ηφ. In one dimension, n = 1, the spatially uniform disturbance flow does not generate any viscous stresses in
the fluid; consequently, the correction is zero. Interestingly, as
n → ∞ the correction approaches a limiting value of 2ηφ; in
contrast to the Einstein correction for the shear viscosity, which
grows as nηφ for large n (Brady [4]). A limiting value exists
for the bulk viscosity correction for the following reason. The
(constant) expansion rate e is (equally) distributed over an increasing number of spatial dimensions; hence, the expansion
rate per dimension is O(1/n). On the other hand, the extra particle stress scales as O(n), for large n (recall, the particle stress
is proportional to the disturbance velocity gradient, which increases with increasing n owing to the more rapid decay of the
disturbance velocity with distance). Thus, from (4), their product gives a κeff that is independent of n as n → ∞.
Acknowledgment
The author thanks Dr. John F. Brady for fruitful discussions
and critical reading of the manuscript.
References
[1] J.F. Brady, A.S. Khair, M. Swaroop, J. Fluid Mech. 554 (2006) 109.
[2] A. Einstein, Ann. Phys. 19 (1906) 289.
[3] G.K. Batchelor, An Introduction to Fluid Dynamics, Cambridge Univ.
Press, Cambridge, 1973.
[4] J.F. Brady, Int. J. Multiphase Flow 10 (1984) 113.

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close