Annals of Biomedical Engineering, Vol. 28, pp. 373–385, 2000
00
Printed in the USA. All rights reserved.
Shear Augmented Dispersion of a Solute in a Casson Fluid Flowing
in a Conduit
R. K. DASH, G. JAYARAMAN, and K. N. MEHTA
Department of Mathematics, Indian Institute of Technology, Hauz Khas, New Delhi, 110016, India
(Received 21 July 1998; accepted 9 March 2000)
Abstract—The unsteady dispersion of a solute in a Casson
fluid flowing in a conduit 共pipe/channel兲 is studied using the
generalized dispersion model of Gill and Sankarasubramanian.
With this approach, the entire dispersion process is described
appropriately in terms of a simple diffusion process with the
effective diffusion coefficient as a function of time, in addition
to its dependence on the yield stress of the fluid. The results are
accurate up to a first approximation for small times, but veri-
fied with Sharp to be exact for large times. The model brings
out mainly the effect of yield stress, or equivalently, the plug
flow region on the overall dispersion process. It is found that
the rate of dispersion is reduced 共i.e., the effective diffusivity
decreases兲 due to the yield stress of the fluid, or equivalently,
the plug flow region in the conduit. Also, the effective diffu-
sivity increases with time, but eventually attains its steady state
value below a critical time 关0.48(a 2 /D m ) for dispersion in a
pipe and 0.55(a 2 /D m ) for dispersion in a channel—the critical
transient time for a Newtonian fluid—where ‘‘a’’ is the radius
of the pipe and D m is the molecular diffusivity兴. At steady
state, for dispersion in a pipe with the plug flow radius one
tenth of the radius of the pipe, the effective diffusivity is
reduced to about 0.78 times of the corresponding value for a
Newtonian fluid at equivalent flow rates; for dispersion in a
channel, the reduction factor is about 0.73 confirming the ear-
lier result of Sharp. Further, the location of the center of mass
of a passive species over a cross section is found to remain
unperturbed during the course of dispersion and for different
values of the plug flow parameter 共i.e., the yield stress of the
fluid兲. The study can be used as a starting first approximate
solution for studying the dispersion in the cardiovascular sys-
tem or blood oxygenators. © 2000 Biomedical Engineering
Society. 关S0090-6964共00兲01204-2兴
Keywords—Shear augmented dispersion, Generalized disper-
sion model, Casson fluid, Blood flow.
INTRODUCTION
The longitudinal dispersion of a solute in a solvent
flowing in a conduit 共pipe/channel兲 is a phenomenon of
wide application in chemical engineering, biomedical en-
Present address for R. K. Dash: Department of Mathematics, Texas
A&M University, College Station, TX 77843, electronic mail:
Ranjan.Dash@math.tamu.edu
Address correspondence to G. Jayaraman; electronic mail:
jgirija@cas.iitd.ernet.in
373
0090-6964/2000/28共4兲/373/13/$15.
Copyright © 2000 Biomedical Engineering Society
gineering, physiological fluid dynamics, and environmen-
tal sciences. The basic principle underlying the disper-
sion theory is the spreading of a passive species 共solute兲
in a flowing fluid 共solvent兲 due to the combined action of
molecular diffusion and nonuniform velocity distribution.
The first fundamental study on dispersion was that of
Taylor20 who showed that, if a solute is injected into a
solvent flowing steadily in a straight tube, the combined
action of the lateral molecular diffusion and the variation
of velocity over the cross section would cause the solute
ultimately to spread diffusively with the effective mo-
lecular diffusivity D eff given by D eff⫽a2wm 2
/48D m ,
where D m is the molecular diffusivity, w m is the mean
共average兲 axial velocity, and ‘‘a’’ is the radius of the
tube. The analysis showed that the spreading of the sol-
ute is symmetrical about a point moving with the aver-
age velocity w m of the fluid. Aris,3 using the method of
moments, showed that the effective molecular diffusivity
would be D eff⫽Dm ⫹a2wm 2
/48D m when the contribution
of axial molecular diffusion is also taken account. The
analysis showed that the Taylor’s dispersion theory is
valid for D effⰇDm .
The time development of dispersion has most com-
monly been studied by calculating the evolution of axial
moments of the solute concentration following its injec-
tion into the flow. Ananthakrishnan et al.2 obtained the
exact numerical solution for the complete convective dif-
fusion equation which takes into account both the radial
and axial molecular diffusion. Their results showed that
the Taylor–Aris’s dispersion theory gives a good de-
scription of the dispersion process if and only if the time
after injection of the solute exceeds about 0.5(a 2 /D m ).
The effect of inlet boundary conditions on the transient
approach to the asymptotic Taylor–Aris’s dispersion
theory was studied in a subsequent paper by Gill and
Ananthakrishnan10 and was validated by the experimen-
tal work of Reejhsingani et al.18
Gill9 generalized Taylor–Aris’s work by proposing a
series expansion about the mean concentration to de-
scribe the local concentration distribution. Gill and
Ananthakrishnan11 extended this theory to include the
374 DASH, JAYARAMAN, and MEHTA
effect of finite slug inputs on the dispersion process. In a
subsequent analysis, Gill and Sankarasubramanian12
showed that the method of series solution mentioned
earlier provides an exact solution for the unsteady con-
vective diffusion problem for laminar flow in a circular
tube if the coefficients in the dispersion model are ob-
tained as suitable functions of time. This model is widely
referred to as the generalized dispersion model. By trun-
cating the generalized dispersion model to two terms,
Gill and Sankarasubramanian12 showed that, for all time,
the mean concentration profile of the solute was symmet-
ric about a point moving with the average velocity of the
fluid. Their results also validated the finding of Anan-
thakrishnan et al.2 that the Taylor–Aris’s dispersion
theory is applicable for time exceeding 0.5(a 2 /D m ).
Most of the available literature on axial dispersion in
straight/curved pipes/channels having various engineer-
ing applications have been confined to Newtonian fluids
only. A detailed account of all these can be found in
Balasubramanian et al.4,5 and Jayaraman et al.13 But, the
developments in the area of polymer and biochemical
processings and biomedical engineering have contributed
to the ever-increasing interest in the study of dispersion
phenomena in non-Newtonian fluids. A brief review on
dispersion in power-law fluids can be found in Agarwal
and Jayaraman.1
The applicability of the Taylor–Aris’s theory of dis-
persion to the cardiovascular system is discussed in
Lighthill.15 In a recent paper, with a view to understand
the effect of non-Newtonian rheology of blood on the
dispersion process in the cardiovascular system, Sharp19
studied the shear-augmented dispersion in non-
Newtonian fluids 共Casson, Bingham plastic, and power-
law fluids兲 flowing in conduits 共pipes/channels兲 based on
the Taylor–Aris’s dispersion theory. It was shown that
the effective molecular diffusivity depends on the spe-
cific rheological parameter of the fluid 共i.e., the yield
stress for Bingham plastic and Casson fluids, and the
power-law index for power-law fluids兲. In fact, the axial
dispersion was found to reduce due to the non-
Newtonian rheology of the fluid. However, being based
on the Taylor–Aris’s dispersion theory, these results are
valid for time exceeding about 0.5(a 2 /D m ) which corre-
sponds to a very large time of dispersion in the actual
blood flow conditions. This limitation demands the
analysis to be undertaken by a dispersion model more
general than the Taylor–Aris’s model.
In this paper, the dispersion of a solute in a Casson
fluid flowing in a conduit 共pipe/channel兲 is studied using
the generalized dispersion model of Gill and
Sankarasubramanian.12 With this approach, the entire
dispersion process is described appropriately in terms of
a simple diffusion process with the apparent diffusion
coefficient as a function of time, in addition to its de-
pendence on the yield stress of the fluid. This extends
FIGURE 1. Schematic diagram of velocity profiles for New-
tonian and Casson fluids in a circular tube.
the validity of the results for time smaller than
0.5(a 2 /D m ). For larger time, the results approach asymp-
totically to the results of Sharp.19 The detailed math-
ematical analysis for dispersion in a pipe flow is given in
the Pipe Flow Analysis section, and the dispersion in a
channel flow is briefly presented in the Channel Flow
Analysis. The results on the effect of yield stress of the
fluid on the dispersion coefficient and overall dispersion
process is presented in the Results section. The discus-
sions and concluding remarks are given in the Discus-
sions section and the Conclusion section.
PIPE FLOW ANALYSIS
Governing Equations
Let us consider the dispersion of a bolus of solute
which is initially of z̄ s units in length and of uniform
concentration C 0 . For axisymmetric fully developed
steady laminar flow in a pipe 共the schematic diagram is
shown in Fig. 1兲, the unsteady convective diffusion equa-
tion which describes the local concentration C̄ of the
solute as a function of longitudinal 共axial兲 coordinate z̄,
transverse 共radial兲 coordinate r̄, and time t̄ can be writ-
ten in the form
冉 冊
⳵ C̄ ⳵ C̄ ⳵2
⫹w̄ ⫽D m L̄ 2 ⫹ 2 C̄, 共1兲
⳵ t̄ ⳵ z̄ ⳵ z̄
where
L̄ 2 ⫽
1 ⳵
冉 冊r̄
⳵
r̄ ⳵ r̄ ⳵ r̄
, 共2兲
where w̄ is the axial velocity of the fluid in the pipe and
D m is the coefficient of molecular diffusion 共molecular
diffusivity兲 which is assumed to be constant.
Initial and Boundary Conditions
For the slug input of solute length z̄ s under consider-
ation, the initial and boundary conditions are given by
 Shear Augmented Dispersion of a Solute in a Casson Fluid Flowing in a Conduit 375

C̄ 共 0,r̄,z̄ 兲 ⫽C 0 , if 兩 z̄ 兩 ⭐z̄ s /2, 共3a兲 C 共 t,r,⬁ 兲 ⫽0, 共8c兲

C̄ 共 0,r̄,z̄ 兲 ⫽0, if 兩 z̄ 兩 ⬎z̄ s /2, 共3b兲 ⳵C ⳵C

共 t,0,z 兲 ⫽0⫽ 共 t,1,z 兲 . 共8d,e兲
⳵r ⳵r
C̄ 共 t̄ ,r̄,⬁ 兲 ⫽0, 共3c兲
Casson’s Constitutive Equation and Velocity
Distribution
⳵ C̄ ⳵ C̄
共 t̄ ,0,z̄ 兲 ⫽0⫽ 共 t̄ ,a,z̄ 兲 . 共3d,e兲
⳵ r̄ ⳵ r̄ In nondimensional form, the constitutive equation for
a Casson fluid relating the shear stress 共␶兲 and shear rate
Conditions 共3a兲 and 共3b兲 constitute the initial conditions, (dw/dr) is given by
共3c兲 constitutes the far downstream condition, and 共3d兲
and 共3e兲 constitute, respectively, the symmetry condition
at the center and the no-flux condition at the imperme-
able wall of the pipe; ‘‘a’’ is the radius of the pipe.
␶ 1/2⫽ ␶ 1/2
y ⫹ ⫺冉 冊 dw
dr
1/2
, if ␶⭓␶y 共9a兲

and
Nondimensionalization
Let us introduce the following nondimensional vari- dw
⫽0, if ␶⭐␶y , 共9b兲
ables: dr

C̄ w̄ r̄ D m z̄ D m t̄ where
C⫽ , w⫽ , r⫽ , z⫽ , t⫽ ,
C0 w0 a a 2w 0 a2
共4兲 ¯␶ ¯␶ y
␶⫽ and ␶ y⫽ . 共10a,b兲
␮ ⬁ w 0 /a ␮ ⬁ w 0 /a
where
¯␶ and ¯␶ y are the dimensional shear stress and yield
a 2 dp̄ stress, respectively. Blood possesses a yield stress of
w 0 ⫽⫺ 共5兲
4 ␮ ⬁ dz̄ 0.04 dyn/cm2 at a hematocrit of 40% 共see McDonald兲.16
␶ y is the dimensionless yield stress and is called the yield
is the characteristic velocity 共centerline velocity in a Poi- stress parameter. The earlier relations correspond to van-
seuille flow兲. ␮ ⬁ is the Newtonian viscosity 共viscosity at ishing of velocity gradient in the region where the shear
high rate of shear兲 of the fluid and dp̄/dz̄ is the applied stress ␶ is less than the yield stress ␶ y . This in turn,
pressure gradient along the axis of the pipe. implies a plug flow whenever ␶ ⭐ ␶ y .
With the earlier substitutions, the unsteady convective For axisymmetric fully developed steady laminar flow
diffusion Eq. 共1兲, in nondimensional form, becomes of a Casson fluid in a pipe 共see Fung8 or Sharp兲,19 the
velocity distribution, in nondimensional form, is given
⳵C
⳵t
⫹w
⳵C
⳵z 冉
1 ⳵2
⫽ L 2 ⫹ 2 2 C,
Pe ⳵ z 冊 共6兲
by

w⫽w ⫹ ⫽ 共 1⫺r 2 兲 ⫺ 38 r 1/2

p 共 1⫺r 兲 ⫹2r p 共 1⫺r 兲 ,
3/2

where
if r p ⭐r⭐1 共11a兲

L 2⫽
1 ⳵
r冉 冊
r ⳵r ⳵r
⳵
and Pe⫽
aw 0
Dm
共Péclét number兲. and
共7a,b兲
p ⫹2r p ⫺ 3 r p ,
w⫽w ⫺ ⫽w p ⫽1⫺ 38 r 1/2 1 2
if 0⭐r⭐r p ,
In dimensionless form, the initial and boundary condi- 共11b兲
tions 共3兲 are reduced to
where
C 共 0,r,z 兲 ⫽1, if 兩 z 兩 ⭐z s /2, 共8a兲
a¯␶ y ␶y
r p⫽ ⫽ 共12兲
C 共 0,r,z 兲 ⫽0, if 兩 z 兩 ⬎z s /2, 共8b兲 2 ␮ ⬁w 0 2
376 DASH, JAYARAMAN, and MEHTA

is the dimensionless radius of the plug flow region. The ⳵Cm ⳵Cm 1 ⳵ 2C m
subscripts ‘‘⫹’’ and ‘‘⫺’’ correspond to the values for ⫹ 共 w⫺w m 兲 ⫺ 2
⳵t ⳵ z 1 Pe ⳵ z 21
the shear flow (r p ⭐r⭐1) and plug flow (0⭐r⭐r p )

冋冉 冊
⬁
regions, respectively. In the plug flow region, the fluid ⳵fj ⳵ jCm ⳵ j⫹1 C m
particles do not move themselves, but are merely carried ⫹ 兺
j⫽1 ⳵t
⫺L 2 f j
⳵ z 1j
⫹ 共 w⫺w 兲
m jf
⳵ z 1j⫹1
along by the fluid particles in the adjacent shear flow
region as a solid body with a constant velocity w p —the
plug flow velocity. The mean velocity of the fluid in
dimensionless form is given by
⫺
1
Pe 2
f j
⳵ j⫹2 C m
⳵ z 1j⫹2
⫹ f j
⳵ j⫹1 C m
册
⳵ t ⳵ z 1j
⫽0. 共17兲

Now if it is assumed that the process of distributing C m

is diffusive in nature right from the time ‘‘zero,’’ then as
w m ⫽ 21 共 1⫺ 167 r 1/2
p ⫹ 3 r p ⫺ 21 r p 兲 .
4 1 4
共13兲
in Gill and Sankarasubramanian,12 the generalized dis-
persion model for C m can be written as
To analyze the dispersion of a solute in a Casson fluid ⬁
⳵Cm ⳵ iC m
兺
flowing in a pipe, the unsteady convective diffusion Eq.
⫽ K i共 t 兲 共18兲
共6兲 has to be solved for the local concentration C subject ⳵t i⫽1 ⳵ z i1
to conditions 共8兲 with the axial velocity w given by Eq.
共11兲. with dispersion coefficients K i as suitable functions of
time t. The first two terms on the right-hand side of Eq.
共18兲 describe the transport of C m in axial direction z 1
Method of Solution through convection and diffusion, respectively, and
Let us consider the convection across a plane which therefore, the coefficients K 1 and K 2 are termed as the
moves with the average velocity w m of the fluid. For longitudinal convection and diffusion coefficients for
this, we need to define a new coordinate system (r,z 1 ,t) C m . The Taylor–Aris’ dispersion model for a Newtonian
with the new axial coordinate z 1 given by fluid corresponds to the value zero for K 1 and 1/Pe 2
⫹1/192 for K 2 . So, if the higher order terms in Eq. 共18兲
are neglected, then it reduces to a simple diffusion equa-
z 1 ⫽z⫺w m t. 共14兲 tion for C m with diffusion coefficient as K 2 . Since K 2
describes the entire dispersion process as a simple diffu-
sion process in axial direction z 1 , it is termed as the
effective longitudinal diffusivity. Now, substituting Eq.
The solution of Eqs. 共6兲 and 共8兲, formulated as a series
共18兲 in Eq. 共17兲 and rearranging the terms, we get
expansion in ⳵ j C m / ⳵ z 1j , as in Gill and
Sankarasubramanian,12 is given by
冋 ⳵f1
⳵t
⫺L 2 f 1 ⫹ 共 w⫺w m 兲 ⫹K 1 共 t 兲
⳵Cm
⳵z1 册
冋
⬁
⳵ jCm
C⫽C m ⫹ 兺 f j 共 t,r 兲
⳵ z 1j
, 共15兲 ⫹
⳵f2
⫺L 2 f 2 ⫹ 共 w⫺w m 兲 f 1 ⫹K 1 共 t 兲 f 1 ⫹K 2 共 t 兲
j⫽1 ⳵t

where ⫺ 册
1 ⳵ 2C m
2 ⫹
Pe ⳵ z 1
2
j⫽1
⳵ f j⫹2
⳵t 兺
⬁

冋
⫺L 2 f j⫹2 ⫹ 共 w⫺w m 兲 f j⫹1

C m ⫽2 冕 1
Crdr 共16兲
1
⫺ 2 f j⫹
Pe i⫽1
j⫹1

兺
K i 共 t 兲 f j⫹2⫺i ⫹K j⫹2 共 t 兲
⳵ j⫹2 C m
⳵ z 1j⫹2
册
⫽0.
0
共19兲

If Eq. 共19兲 is satisfied, then the coefficients of

is the mean 共average兲 concentration over a cross section.
⳵ j C m / ⳵ z 1j , j⫽1,2,..., must vanish. This leads to an infi-
On transforming the unsteady convective diffusion Eq.
nite set of differential equations given by
共6兲 into the moving coordinate system (r,z 1 ,t), where z 1
is defined by Eq. 共14兲, and substituting Eq. 共15兲 into the
transformed unsteady convective diffusion equation, we ⳵f1
⫺L 2 f 1 ⫹ 共 w⫺w m 兲 ⫹K 1 共 t 兲 ⫽0, 共20a兲
obtain ⳵t
 Shear Augmented Dispersion of a Solute in a Casson Fluid Flowing in a Conduit 377

⳵f2 1 f 1 共 t,r 兲 ⫽ f 1s 共 r 兲 ⫹ f 1t 共 t,r 兲 . 共26兲

⫺L 2 f 2 ⫹ 关共 w⫺w m 兲 ⫹K 1 共 t 兲兴 f 1 ⫹K 2 共 t 兲 ⫺ 2 ⫽0,
⳵t Pe
共20b兲 where f 1s (r) is the large time solution which corre-
sponds to Taylor–Aris’s dispersion theory and f 1t is the
⳵ f j⫹2 transient part which describes the time-dependent nature
⫺L 2 f j⫹2 ⫹ 关共 w⫺w m 兲 ⫹K 1 共 t 兲兴 f j⫹1 of the dispersion process corresponding to a Casson
⳵t
model. Substituting Eq. 共26兲 in Eq. 共20a兲 and conditions

冋 册
j⫹2
1 共21兲, and noting that K 1 ⫽0 关by Eq. 共23兲兴, we obtain
⫹ K 2共 t 兲 ⫺ 2 f j ⫹
Pe i⫽3
兺
K i 共 t 兲 f j⫹2⫺i ⫽0,
L 2 f 1s ⫺ 共 w⫺w m 兲 ⫽0, 共27兲
for j⫽1,2,..., with f 0 ⫽1. 共20c兲

⳵ f 1t
The initial and boundary conditions for f j ’s can be ob- ⫽L 2 f 1t , 共28兲
tained from Eqs. 共8兲 and 共15兲. Since C m can be chosen to ⳵t
satisfy the initial condition on C, we have
with
f j 共 0,r 兲 ⫽0, 共21a兲
d f 1s d f 1s
共 r⫽0 兲 ⫽0⫽ 共 r⫽1 兲 , 共29a,b兲
⳵fj ⳵fj dr dr
共 t,0兲 ⫽0⫽ 共 t,1兲 . 共21b,c兲
⳵r ⳵r
⳵ f 1t ⳵ f 1t
Again, from Eqs. 共15兲 and 共16兲, we have the solvability 共 t,0兲 ⫽0⫽ 共 t,1兲 , 共30a,b兲
⳵r ⳵r
condition as

f 1t 共 0,r 兲 ⫽⫺ f 1s 共 r 兲 . 共30c兲
冕 0
1
f j rdr⫽0. 共22兲
The solvability condition 共4.22兲 takes the form
Multiplying Eq. 共20a兲 by r and integrating from 0 to 1
with the help of solvability condition 共22兲, we get 冕 0
1
f 1t rdr⫽⫺ 冕 0
1
f 1s rdr⫽0. 共31兲

K 1 共 t 兲 ⫽⫺2 冕 1

0
共 w⫺w m 兲 rdr⫽0. 共23兲 The solution for f 1s , obtained from Eq. 共27兲 and condi-
tions 共29兲 and 共31兲, is given by
Applying the same procedure to Eqs. 共20b兲 and 共20c兲,
we obtain f 1s 共 r 兲 ⫽ f 1s ⫹ 共 r 兲

1
K 2 共 t 兲 ⫽ 2 ⫺2
Pe
冕
0
1
f 1 wrdr 共24兲
⫽⫺
115r 4p
7056
⫺
r 4p
84
ln 冉冊冉r
rp
⫹ 1
8 ⫺
8r 1/2
21
p
⫹
rp
⫹
3 168
r 4p 2
r冊
⫺ 29 r p r 3 ⫹ 147 r p r ⫺ 161 r 4 ⫹CI,
32 1/2 7/2
if r p ⭐r⭐1,
and 共32a兲

K j⫹2 共 t 兲 ⫽⫺2 冕 0
1
f j⫹1 wrdr, j⫽1,2,... . 共25兲
f 1s 共 r 兲 ⫽ f 1s ⫺ 共 r 兲 ⫽ 冉 1
8 ⫺
8r 1/2
p
⫹
r p r 2p
⫺ ⫹
r 4p
r2 冊
21 3 12 168
Solution for f 1 : ⫹CI, if 0⭐r⭐r p , 共32b兲

The function f 1 is the most important coefficient of

where
the series in Eq. 共15兲. It gives a measure of deviation of
the local concentration C from the mean concentration
C m . The solution to the nonhomogeneous parabolic par- r 6p 13r 4p
r 4p 7r p 60r 1/2
p 1
CI⫽ ⫹ ⫺ lnr p ⫺ ⫹ ⫺ .
tial differential Eq. 共20a兲 and conditions 共21兲 can be 1320 1764 84 90 539 24
written in the form 共33兲
378 DASH, JAYARAMAN, and MEHTA

The steady-state solution f 1s described by Eqs. 共32a,b兲 are likely to decrease further in magnitude due to the
and 共33兲 is also obtained in Sharp.19 The solution for f 1t yield stress ␶ y of the fluid.
is obtained as an eigen value problem from Eq. 共28兲 and
conditions 共30兲 and 共31兲 which is given by
Solution for Mean Concentration C m
⬁

兺
2
f 1t ⫽ A n e ⫺␭ n t J 0 共 ␭ n r 兲 , 共34兲 Neglecting K 3 (t) and other higher order coefficients,
n⫽1 the generalized dispersion model 共18兲 reduces to

where ⳵Cm ⳵ 2C m
⫽K 2 共 t 兲 . 共37兲
⳵t ⳵ z 21
冕 1

0
J 0 共 ␭ n r 兲 f 1s 共 r 兲 rdr
The initial and boundary conditions for C m are given by
A n ⫽⫺
冕 1

0
关 J 0 共 ␭ n r 兲兴 2 rdr
C m 共 0,z 1 兲 ⫽1, if 兩 z 1 兩 ⭐z s /2, 共38a兲

2冕 1

0
J 0 共 ␭ n r 兲 f 1s 共 r 兲 rdr C m 共 0,z 1 兲 ⫽0, if 兩 z 1 兩 ⬎z s /2, 共38b兲
⫽⫺ . 共35兲
关 J 0 共 ␭ n 兲兴 2 C m 共 t,⬁ 兲 ⫽0. 共38c兲

Here J 0 and J 1 denote Bessel’s functions of first kind of The solution of Eq. 共37兲 with the help of conditions 共38兲
order zero and one, respectively, and the eigen values is given by
␭ n ’s are the roots of the equation J 1 (r)⫽0. We evaluate

冋 冉 冊 冉 冊册
the integrals in Eq. 共35兲 by numerical integration.
z ⫺z 1
1
2 s z ⫹z 1
1
2 s
C m⫽ 1
erf ⫹erf , 共39兲
2 冑␰ 2 冑␰
2
Calculation of Dispersion Coefficient K 2 (t)
The most important coefficient in the generalized dis-
where
persion model 共18兲 is K 2 (t). As seen in Eq. 共24兲, it

冕
depends on the function f 1 (t,r). Substituting the expres-
t
sion for f 1 , given by Eqs. 共26兲 and 共32兲–共35兲, in Eq. ␰⫽ K 2 共 t 兲 dt. 共40兲
共24兲 and simplifying we get 0

1 Equation 共39兲 suggests that, for all time t, the mean

K 2共 t 兲 ⫽ ⫹ 1 关 1⫺ 1155 r p ⫹ 558
5888 1/2
56595 r p ⫺ 715 r p ⫹ 45 r p
368 6144 3/2 128 2
concentration C m is symmetric about a point moving
Pe 2 192
with the average velocity w m of the fluid. By using the
⫺ 218 r 4p ⫹ 3773 r p ⫺ 45
3840 9/2
r p ⫹ 558 r 6p ⫺ 1155
32 5
r p ⫹ 165
512 13/2 64 7
rp expression for C m and f 1 in Eq. 共15兲 and neglecting the

冕 1
higher order terms, we can obtain the local concentration
⫺ 22943
295 r p ⫺ 147 r p ln r p ⫹ 1155 r p 兴 ⫺2
8 8 8 8 10
f 1t wrdr. C to first approximation as
0

共36兲
Once K 2 (t) is known, f 2 (t,r) can be obtained from Eq.
共20b兲 in a manner similar to that for f 1 (t,r). Substitution
of expression for f 2 in Eq. 共25兲 will give K 3 (t). This
way, we can find f 3 (t,r), K 4 (t), f 4 (t,r), K 5 (t),... . and
so on recursively from Eqs. 共20c兲 and 共25兲. Since the
expressions for f 1 (t,r) and K 2 (t) are complicated in
nature, it is very difficult to evaluate f 2 (t,r),K 3 (t),... .,
and so on. But, for dispersion in a Newtonian fluid
which correspond to r p ⫽ ␶ y ⫽0, it was shown 共see Gill
and Sankarasubramanian兲12 that, K 3 (t→⬁)⫽⫺1/230 40
and the magnitude of higher order coefficients decrease
further. We have not evaluated these coefficients which
⳵Cm
C 共 r,z 1 ,t 兲 ⫽C m 共 z 1 ,t 兲 ⫹ f 1 共 r,t 兲 共 z ,t 兲 . 共41兲
⳵z1 1
CHANNEL FLOW ANALYSIS
We use the Cartesian coordinate system (x̄,z̄), where
x̄ denotes the transverse coordinate and z̄ denotes the
axial coordinate, to describe the dispersion of a solute in
a Casson fluid flowing in a channel. The flow is assumed
to be axisymmetric, fully developed, steady, and laminar.
Proceeding on the similar lines as in the case of pipe
flow, we can write the unsteady convective diffusion
equation for the dispersion of the solute in the channel
flow, in nondimensional form, as
 Shear Augmented Dispersion of a Solute in a Casson Fluid Flowing in a Conduit 379

⳵C
⳵t
⫹w
⳵C
⳵z 冉
1 ⳵2
⫽ L 2 ⫹ 2 2 C,
Pe ⳵ z 冊 共42兲 C m⫽ 冕 0
1
Cdx. 共48兲

where The change in the definition of C m and the operator L 2

will necessitate some minor modifications in the solv-
⳵2 aw 0 ability condition 共22兲 and the expressions for K j , j
L 2⫽ and Pe⫽ 共 Péclét number兲 . ⫽1,2,..., given by Eqs. 共23兲–共25兲, which are straight
⳵x2 Dm
共43a,b兲 forward. Therefore, we do not go into the details again
and directly write down the solutions for f 1s , f 1t , and
K 2 later. The steady-state solution f 1s described later by
The initial and boundary conditions are as in Eq. 共8兲
Eqs. 共49a,b兲 and 共50兲 is also obtained in Sharp.19
with x replacing r. Here, a is the half of the channel
width, and w 0 is the characteristic velocity 共centerline
velocity in a plane Poiseuille flow兲 given by
f 1s 共 x 兲 ⫽ f 1s ⫹ 共 x 兲 ⫽
x 4p
x 3p
⫺ x⫹
84 15
冉 1
6 ⫺
8x 1/2
15
p
⫹ ⫹
2 30
冊
x p x 3p 2
x
a 2 dp̄
w 0 ⫽⫺ . 共44兲 x p 3 32x 1/2
p
2 ␮ ⬁ dz̄ ⫺ x ⫹ x 7/2⫺ 121 x 4 ⫺CI,
3 105

For axisymmetric fully developed steady laminar flow of if x p ⭐x⭐1, 共49a兲

a Casson fluid in a channel 共see Fung8 or Sharp兲,19 the
velocity distribution, in dimensionless form, is given by

w⫽w ⫹ ⫽ 共 1⫺x 兲 ⫺ x
f 1s ⫽ f 1s ⫺ ⫽ 冉 1
6 ⫺
8x 1/2
p
15
⫹
x p x 2p x 3p 2
2
⫺ ⫹
6 30
x ⫺CI, 冊
p 兲 ⫹2x p 共 1⫺x 兲 ,
8 1/2
2
3 p 共 1⫺x 3/2
if 0⭐x⭐x p , 共49b兲
if x p ⭐x⭐1 共45a兲

where
and

104x 1/2
x p x 3p x 4p x 5p
w⫽w ⫺ ⫽w p ⫽1⫺ 38 x 1/2
p ⫹2x p ⫺ 3 x p ,
1 2
if 0⭐x⭐x p , CI⫽ 7
180 ⫺
p
⫹ ⫺ ⫹ ⫺ . 共50兲
945 12 45 84 540
共45b兲

where The transient solution f 1t is given by

a¯␶ y ␶y ⬁
x p⫽ ⫽ 共46兲
兺
2
2 ␮ ⬁w 0 2 f 1t ⫽ A n e ⫺␭ n t cos共 ␭ n x 兲 共51兲
n⫽1

denotes the half of the dimensionless thickness of plug

with coefficients A n ’s and eigenvalues ␭ n ’s given by
flow region in the channel. The mean velocity, in dimen-
sionless form, is given by

w m ⫽ 共 1⫺ x ⫹ x ⫺ x 兲 .
2 12 1/2 3
2 p
1 3
共47兲
冕 0
1
cos共 ␭ n x 兲 f 1s 共 x 兲 dx
A n ⫽⫺
3 5 p 10 p

To analyze the dispersion of a solute in a Casson fluid

冕 1

0
cos2 共 ␭ n x 兲 dx

冕
flowing in a channel, the unsteady convective diffusion
1
Eq. 共42兲 has to be solved for the local concentration C ⫽⫺2 cos共 ␭ n x 兲 f 1s 共 x 兲 dx, 共52a兲
subject to conditions 共8兲 with the axial velocity w given 0
by Eq. 共45兲. We proceed to find the solution for the
problem in a coordinate system moving with the average
␭ n ⫽n ␲ , n⫽1,2,3,... . 共52b兲
velocity w m of the fluid, given by Eq. 共47兲, in a manner
similar to that for pipe flow analysis. In the present case,
the mean concentration C m is defined as The dispersion coefficient K 2 is given by
380 DASH, JAYARAMAN, and MEHTA

1
K 2共 t 兲 ⫽ ⫹ 8 关 1⫺ 1064
195 x p ⫹ 240 x p ⫺ 11 x p ⫹ 16 x p
1/2 2737 119 3/2 63 2
Pe 2 945
⫺ 49
80 x p ⫹ 15 x p ⫺ 16 x p ⫹ 48 x p ⫺ 15 x p ⫹ 80 x p
3 26 7/2 21 4 7 5 7 11/2 49 6

⫺ 2280 x p ⫹ 240
449 7 7
x 8p 兴 ⫺ 冕
0
1
f 1t wdx. 共53兲

RESULTS

The time-dependent nature of the dispersion coeffi-

cient K 2 is illustrated in Figs. 2 and 3. The plot of
192(K 2 ⫺1/Pe 2 ) verses time t for dispersion in a pipe
with different values of r p 共plug flow radius兲 is shown in
Fig. 2共a兲, where as, the plot of (945/8)(K 2 ⫺1/Pe 2 ) ver-
sus time t for dispersion in a channel with different
values of x p 共half thickness of the plug flow region兲 is
shown in Fig. 2共b兲. The dotted lines in these figures
describe the dispersion coefficient for a Newtonian fluid
共i.e., r p ⫽0 or x p ⫽0; Gill and Sankarasubramanian兲.12
The plot of relative effective diffusivity 关i.e., (K 2
⫺1/Pe 2 )/(K N2 ⫺1/Pe 2 ), where the superscript ‘‘N’’ rep-
resents the value corresponding to a Newtonian fluid兴 for
dispersion in a pipe and a channel with different values
of r p or x p is shown in Figs. 3共a兲 and 3共b兲, respectively.
It is observed from Fig. 2 that the dispersion coeffi-
cient K 2 decreases considerably with the increase in the
value of r p or x p . Also, it increases with the increase in
time t, but becomes essentially a constant as time takes
larger values. The time t required to reach the steady
state is seen to depend on the plug flow parameter r p or FIGURE 2. Illustration of time-dependent nature of Disper-
x p 共i.e., on the yield stress ␶ y of the fluid兲 which is sion Coefficient „ K 2 …; „a… Plot of 192„ k 2 À1Õ Pe 2 … vs time for
dispersion in a pipe with different values of r p ; „b… plot of
shown in Table 1. For dispersion in a Newtonian fluid
„945Õ8…„ K 2 À1Õ Pe 2 … vs time dispersion in a channel with dif-
flowing in a pipe 共i.e., for r p ⫽0兲, the time t necessary to ferent values of x p .
reach the steady state is seen to be about 0.48, whereas,
for dispersion in a Newtonian fluid flowing in a channel
共i.e., for x p ⫽0兲, this critical value is about 0.55. For dispersion due to the yield stress of the fluid at constant
larger values of r p or x p , this critical value is less than flow rates. Here, the superscripts ‘‘P’’ and ‘‘C’’ refer to
the value corresponding to r p ⫽0⫽x p . The steady-state the values for a pipe and a channel, respectively.
values of K 2 correspond to the result of Sharp.19 It is The variation of 192(K 2P ⫺1/Pe 2 ), 2w mP and 48(K 2P
observed from Figs. 3共a兲 and 3共b兲 that the relative effec- ⫺1/Pe 2 )/(w mP ) 2 with r p 共plug flow radius兲 for steady-
tive diffusivity is almost independent of time t and de- state dispersion in a pipe is shown in Table 2, whereas,
pends only on the values of r p or x p 共or equivalently, on the variation of (945/8)(K C2 ⫺1/Pe 2 ), 3w m C
/2 and
the yield stress ␶ y of the fluid兲. (105/2)(K C2 ⫺1/Pe 2 )/(w m ) with x p 共half thickness of
C 2

It is to be noted here that the quantities 192(K 2P
⫺1/Pe 2 ) and (945/8)(K C2 ⫺1/Pe 2 ) give the measures of
reductions in dispersion due to the yield stress of the
fluid at a constant pressure gradient. However, in flow of
a Casson fluid in a pipe or a channel, the flow rate is
reduced due to the yield stress of the fluid at a constant
pressure gradient. The quantities 2w mP and 3w m C
/2, de-
fined by Eqs. 共13兲 and 共47兲, give the reductions in flow
rate in a pipe and a channel, respectively. Therefore, the
quantities 48(K 2P ⫺1/Pe 2 )/(w mP ) 2 and (105/2)(K C2
⫺1/Pe 2 )/(w m C 2
) describe the corresponding reductions in
the plug flow region兲 for steady-state dispersion in a
channel is shown in Table 3. It is seen that the effective
diffusivity at a constant flow rate for dispersion in a
channel is smaller than that of the corresponding value in
a pipe. As r p or x p approaches the value unity, the
effective diffusivity approaches the value zero. In this
case, the flow becomes more and more plug-like and the
area over which the shear may augment the dispersion
disappears. Sharp19 has tabulated the possible values of
the plug flow parameters r p or x p for different vessels in
the cardiovascular system. These values are of O(10⫺1 ).
 Shear Augmented Dispersion of a Solute in a Casson Fluid Flowing in a Conduit 381

TABLE 2. Variation of 192„ K 2P À1ÕPe 2 …, 2 w m

P
and 48„ K 2P
À1ÕPe …Õ„ w m … with r p „plug flow radius… for steady-state
2 P 2

dispersion in a pipe.

rp 192( K 2P ⫺1/Pe 2 ) P
2wm 48( K 2P ⫺1/Pe 2 )/( w m
P 2
)

0.00 1.000 1.000 1.000

0.05 0.264 0.556 0.857
0.10 0.131 0.411 0.779
0.15 0.070 0.315 0.709
0.20 0.038 0.244 0.644
0.25 0.021 0.190 0.581
0.30 0.011 0.148 0.520
0.35 0.006 0.113 0.461
0.4 0.003 0.086 0.404

f 1 approaches its steady state value f 1s 共dotted lines in

FIGURE 3. Plot of relative effective diffusivity „ K 2
À1Õ Pe 2 …Õ K 2N À1Õ Pe 2 … vs time t for dispersion „a… in a pipe
and „b… in a channel for different values of r p or x p .
At steady state, for dispersion in a pipe with r p ⫽0.1, the
effective diffusivity is reduced to approximately 0.78
times of the corresponding value for a Newtonian fluid at
equivalent flow rates; for dispersion in a channel with
x p ⫽0.1, this reduction factor is about 0.73 共these results
match with the findings of Sharp兲.19
The time evolution of the dispersion function f 1 for
dispersion in a pipe and a channel with different values
of the plug flow parameters r p and x p is described
through Figs. 4 and 5, respectively. It gives a measure of
deviation in the local concentration C from the mean
concentration C m . At t⫽0, f 1 ⫽0 uniformly over the
cross section of the pipe/channel. But as time t increases,
Figs. 4 and 5兲. For r p ⫽0⫽x p , which corresponds to the
dispersion in a Newtonian fluid 共see Gill and
Sankarasubramanian兲,12 the magnitude of f 1 共and hence,
f 1s 兲 is maximum. But it decreases as the value of r p or
x p increases. It is depicted from these figures that the
dispersion function f 1 共and hence, f 1s 兲 passes through a
common point for all time t and for all values of the plug
flow parameters r p and x p . At this point, the functions
f 1 and f 1s vanish, and the local concentration C of the
solute becomes equal to the mean concentration C m .
Therefore, this point can be regarded as the center of
mass of the solute over a cross section of the conduit.
The location of the center of mass of the solute is inde-
pendent of time t and plug flow parameters r p and x p .
For dispersion in a pipe, the center of mass is located
closer to the wall. But, for dispersion in a channel, its
location is midway of the channel width. It is further
seen that the role of f 1 共and hence, f 1s 兲 is to reduce the
concentration C near the center of the conduit and to
enhance it near the wall.
DISCUSSIONS
The objective of the present investigation is to study
the unsteady dispersion of a solute in a Casson fluid
flowing in a conduit 共pipe/channel兲. The study may be
relevant in understanding many physiological processes
which involve injecting a quantity of solute into the
blood stream and measuring its concentration at some
downstream point 关e.g., 共i兲 dye dilution technique to
measure cardiac output, and 共ii兲 dispersion of drugs/
nutrients in the circulatory system兴. This analysis can

TABLE 1. The critical transient time to reach the steady-state dispersion in a pipe and a channel
for different values of the plug flow parameters r p or x p .

r p⶿ x p 0.00 0.02 0.04 0.06 0.08 0.10 0.15 0.20 0.25

Pipe 0.48 0.35 0.32 0.30 0.27 0.25 0.24 0.23 0.22
Channel 0.55 0.47 0.42 0.38 0.35 0.32 0.30 0.27 0.24
382 DASH, JAYARAMAN, and MEHTA

also be applied to artificial blood handling devices such TABLE 3. Variation of „945Õ8…„ K 2C À1ÕPe 2 …, 3 w m C
Õ2 and
as blood oxygenators. Since some devices involve flow „105Õ2…„ K 2 À1ÕPe …Õ„ w m … with x p „half thickness of plug flow
C 2 C 2

region… for steady-state dispersion in a channel.

between parallel plates and membranes, the study of dis-
persion in channel flows has practical importance. xp (945/8)( K 2C ⫺1/Pe 2 ) C
3wm /2 (105/2)( K 2C ⫺1/Pe 2 )/( w m
C 2
)
The blood flow in the cardiovascular system is char-
0.00 1.000 1.000 1.000
acterized by many fluid dynamical complexities such as 0.05 0.239 0.538 0.824
pulsatile flow, curvature, branching, and elastic proper- 0.10 0.112 0.391 0.733
ties of the arterial wall 共see Lighthill兲.14 Although the 0.15 0.057 0.295 0.654
transport of any passive species in blood stream is af- 0.2 0.030 0.226 0.583
0.25 0.016 0.173 0.517
fected by these phenomena as well as by reactions 共both
0.30 0.008 0.133 0.455
heterogeneous and homogeneous兲 and the multiphase 0.35 0.004 0.101 0.397
character of the fluid concerned, the knowledge of modi- 0.40 0.002 0.076 0.342
fication to the classical Taylor–Aris’s and Gill–
Sankarasubramanian’s dispersion theory arising due to
non-Newtonian rheology of blood is still very useful. An
account of the former is given recently in Sharp.19 In this
paper, we present an account of the latter. blood flow conditions. But, the typical time of dispersion
The applicability of the Taylor–Aris’s fundamental in measure cardiovascular system is usually small. For
theory of dispersion for analyzing the dispersion process example, for oxygen transport in aorta, the typical time
in the cardiovascular system is well discussed by in nondimensional term is of the order of 10⫺5 共see
Lighthill15 in his paper on the initial development of Caro兲,6 which is very small compared to the value 0.5
diffusion in Poiseuille flow. However, it has now been 共the lower bound of the time of validity of Taylor–Aris’s
well estimated by experimental as well as computational analysis in nondimensional terms兲. Also, the condition of
work 共for example, see Ananthakrishnan et al.,2 Gill and applicability of the Taylor–Aris’s analysis to the disper-
Sankarasubramanian,12 and Reejhsingani et al.兲18 that the sion process in the cardiovascular system makes the
Taylor–Aris’s analysis gives a good description of the analysis valid for very small flow rates. It is these limi-
dispersion process if and only if the time after injection tations which requires the analysis to be undertaken by a
of the solute exceeds about 0.5(a 2 /D m ) which corre- dispersion model more general than the Taylor–Aris’s
sponds to a very large time of dispersion in the actual model.

FIGURE 4. Distribution of dispersion

function „ f … at different values of time „ t …
for dispersion in a pipe with „a… r p Ä0, „b…
r p Ä0.05, „c… r p Ä0.1, and „d… r p Ä0.2.
 Shear Augmented Dispersion of a Solute in a Casson Fluid Flowing in a Conduit
The generalized dispersion model of Gill and
Sankarasubramanian12 is employed to analyze the un-
steady dispersion process. This way, the entire dispersion
process is described appropriately in terms of a simple
diffusion process with the effective diffusion 共dispersion兲
coefficient K 2 as a function of time t. This extends the
validity of the results for time smaller than 0.5(a 2 /D m ).
In real terms, this can correspond to relatively higher
flow rates, say corresponding to blood flow in small
arteries and large arterioles where the Casson fluid model
for blood is very appropriate. For large time t, this analy-
sis recovers the results of Sharp19 which were based on
the Taylor–Aris’s dispersion theory. The dispersion co-
efficient K 2 depends on the non-Newtonian rheological
parameter ␶ y 共yield stress兲 of the fluid through its depen-
dence on the plug flow parameter r p or x p . For the case
␶ y ⫽0⫽r p ⫽x p , which corresponds to a Newtonian fluid,
the results of this study agree with those of Gill and
Sankarasubramanian.12
Since the dispersion coefficients K 2 ,K 3 ,K 4 ,... in the
generalized dispersion model are obtained as suitable
functions of time t with the higher order coefficients
K 3 ,K 4 ,... becoming negligibly small at larger values of
t, this model can be thought of describing the whole
dispersion process to a first approximation for small time
(t⬍0.5), but exact for large time (t⬎0.5). To obtain an
exact description of the overall dispersion process start-
ing from time t⫽0⫹, one needs to consider sufficient
383
FIGURE 5. Distribution of dispersion
function „ f … at different values of time „ t …
for dispersion in a channel with „a… x p
Ä0, „b… x p Ä0.05, „c… x p Ä0.1, and „d…
x p Ä0.2.
number of terms in the generalized dispersion model
given by Eqs. 共15兲 and 共18兲 as in Yu.21–23
It is to be noted here that the present analysis deals
with the study of unsteady dispersion process when the
flow is steady. If the flow is considered to be unsteady
共pulsatile兲, the dispersion process will certainly change
due to the change in the width of the plug flow region
during a cycle of oscillation. In our previous paper 共Dash
et al.兲,7 where we have studied the pulsatile flow in an
annulus using Casson fluid model, we have discussed in
detail the change of yield plane locations, and hence, the
width of plug flow region, during a cycle of oscillation
due to pulsatility. Once the plug flow region changes
with time, it will definitely have an effect on mass trans-
fer characteristics, and hence, the dispersion phenomena.
Thus, the results are likely to be changed due to the
oscillatory character of the flow, but that study will be
reported in a subsequent paper.
It is well known that one of the factors which influ-
ences the dispersion of solutes in blood flow is flexing of
red blood cells, their migration to the core etc. 共See Patel
and Sirs兲.17 The importance of the Casson fluid model
for blood will be clearer when we include the uptake by
the walls 共work under preparation兲. The cell free layer
near the wall will have to be included for a description
of the dispersive phenomena.
384 DASH, JAYARAMAN, and MEHTA

CONCLUSION Cm mean concentration of the solute defined by Eq.

共16兲 or 共48兲
The present analysis brings out some interesting re-
C0 initial concentration of the solute
sults on the shear augmented dispersion in flows with the
D eff effective molecular diffusivity
fluid modeled as a Casson fluid. It is found that the
Dm molecular diffusivity
effective diffusion coefficient is modified due to the non-
fj jth dispersion function ( j⫽0,1,2,...)
Newtonian rheology 共i.e., the yield stress兲 of the fluid. In
f 1s steady part of f 1
fact, the effect of yield stress is seen to reduce the rate of
f 1s⫹ f 1s in the shear flow region r p ⭐r⭐1 given by
dispersion of passive species in the flow. The effective
Eq. 共32a兲 or 共49a兲
diffusivity increases with time, but eventually attains its
f 1s⫺ f 1s in the plug flow region 0⭐r⭐r p given by
steady-state value below a critical time 关0.48(a 2 /D m ) for
Eq. 共32b兲 or 共49b兲
dispersion in a pipe and 0.55(a 2 /D m ) for dispersion in a
f 1t transient part of f 1 given by Eq. 共34兲 or 共51兲
channel—the critical transient time for a Newtonian
fluid兴. At steady state, for dispersion in a pipe with plug J0 Bessel function of first kind of order zero
flow radius one tenth of the radius of pipe, the effective J1 Bessel function of first kind of order one
longitudinal diffusivity is reduced to about 0.78 times of Ki ith dispersion coefficient (I⫽1,2,...)
the corresponding value for a Newtonian fluid at equiva- K C2 K 2 corresponding to a channel flow given by
lent flow rate; for dispersion in a channel, this reduction Eq. 共53兲
factor is about 0.73 共these results match with the findings K 2P K 2 corresponding to a pipe flow given by Eq.
of Sharp兲.19 Further, the location of the center of mass of 共36兲
a passive species over a cross section is found to remain K N2 K 2 corresponding to a Newtonian fluid
unperturbed during the course of dispersion and for dif- L differential operator defined by Eq. 共7a兲 or 共43a兲
ferent values of the plug flow parameter 共i.e., the yield Pe Péclét number defined by Eq. 共7b兲 or 共43b兲
stress of the fluid兲. r radial coordinate
Although the present analysis brings out the effect of rp radius of the plug flow region in a pipe
non-Newtonian rheology on the dispersion of passive t time
species in a Casson fluid flowing in a conduit, the model w axial velocity
has to be further refined for application to blood flow. w⫹ w in the shear flow region r p ⭐r⭐1 given by
Since blood is a concentrated suspension of formed ele- Eq. 共11a兲 or 共45a兲
ments 共mostly red cells兲 in plasma, the deformation and w⫺ w in the plug flow region 0⭐r⭐r p given by
aggregation of red cells will definitely affect the lateral Eq. 共11b兲 or 共45b兲
mixing of any solute introduced into the blood stream, wp w in the plug flow region 0⭐r⭐r p given by
and therefore, will modify the rate of axial dispersion. Eq. 共11b兲 or 共45b兲
Also, the transport of some solutes through cells and wm mean axial velocity defined by given Eq. 共13兲 or
convection of the solute contained in the cell due to cell 共47兲
motion may also affect the dispersion process. Finally, w mP w m in a pipe flow given by Eq. 共13兲
the facilitated transport of some solutes, such as oxygen wmC
w m in a channel flow given by Eq. 共47兲
combining with hemoglobin to form oxyhemoglobin, can w0 centerline velocity in a Poiseuille flow defined
also lead to further difficulties in understanding the dis- by Eq. 共5兲 or 共44兲
persion process. x transverse coordinate
xp half width of the plug flow region in a channel
z axial coordinate
ACKNOWLEDGEMENTS z1 axial coordinate moving with the mean velocity
of the fluid defined by Eq. 共14兲
This work was supported by a grant from the Depart-
zs initial length of the bolus of solute
ment of Science and Technology, Government of India
共DST/MS/007/93, dated 19.9.1995兲. The authors are
grateful to the referees for their useful comments and
suggestions.
Greek Symbols
⳵ partial derivatives
NOMENCLATURE ␰ variable defined by Eq. 共40兲
␮⬁ Newtonian viscosity 共viscosity of blood at
Alphabetical Symbols higher rates of shear兲 of the fluid 共blood兲
a radius of the pipe or half width of the channel ␶ shear stress defined by Eq. 共10a兲
C concentration of the solute ␶y yield stress parameter defined by Eq. 共10b兲
 Shear Augmented Dispersion of a Solute in a Casson Fluid Flowing in a Conduit 385

Note added in proof: overbar ‘‘¯ ’’ has been used to
represent the dimensional quantities.
REFERENCES
1 316:341–350, 1970.
Agarwal, S., and G. Jayaraman. Numerical simulation of dis-
persion in the flow of power-law fluids in curved tubes. Appl.
Math. Modeling 18:504–512, 1994.
2 ing walls. Q. J. Mech. Appl. Math. 51:577–598, 1998.
Ananthakrishnan, V., W. N. Gill, and A. J. Barduhn. Laminar
dispersion in capillaries: Part I. Mathematical analysis.
AIChE J. 11:1063–1072, 1965.
3 15
Aris, R. On the dispersion of a solute in a fluid flowing
through a tube. Proc. R. Soc. London, Ser. A 235:67–77,
1956.
4
Balasubramanian, V., G. Jayaraman, and S. R. K. Iyengar.
Effect of Secondary flows on contaminant dispersion with
weak boundary absorption. Appl. Math. Modeling 21:275–
285, 1997.
5 18
Balasubramanian, V., G. Jayaraman, and S. R. K. Iyengar.
Boundary retention effects upon contaminant dispersion in
secondary flows. Z. Angew. Math. Mech. 79:1–18, 1999.
6
Caro, C. G. The dispersion of indicator flowing through sim-
plified models of the circulation and its relevance to velocity
profiles in blood vessels. J. Physiol. (London) 185:501–519,
1966.
7
Dash, R. K., G. Jayaraman, and K. N. Mehta. Estimation of
increased flow resistance in a narrow catheterised artery - a
theoretical model. J. Biomech. 29:917–930, 1996.
8
Fung, Y. C. Biomechanics: Mechanical Properties of Living
Tissues. Springer: New York, 1981.
9 22
Gill, W. N. A note on the solution of transient dispersion
problems. Proc. R. Soc. London, Ser. A 298:335–339, 1967.
10
Gill, W. N., and V. Ananthakrishnan. Laminar dispersion in
capillaries: Part II. Effect of inlet boundary conditions and
turner type of system capacitance. AIChE J. 12:906–915,
1966.
11
Gill, W. N., and V. Ananthakrishnan. Laminar dispersion in
capillaries: Part IV. The slug stimulus. AIChE J. 13:801–807,
1967.
12
Gill, W. N., and R. Sankarasubramanian. Exact analysis of
unsteady convective diffusion. Proc. R. Soc. London, Ser. A
13
Jayaraman, G., A. Goyal, and T. J. Pedley. Dispersion of
solute in a fluid flowing through a curved tube with absorb-
14
Lighthill, M. J. Physiological fluid dynamics: a survey. J.
Fluid Mech. 52:475–497, 1972.
Lighthill, M. J. Initial development of diffusion in Poiseuille
flow. J. Inst. Math. Appl. 2:97–108, 1966.
16
McDonald, D. A. Blood Flows in Arteries. Edward Arnold:
London, 1974.
17
Patel, I. C., and J. A. Sirs. Dispersion of solutes during blood
flow through curved tubes. Med. Biol. Eng. Comput. 21:113–
118, 1983.
Reejhsingani, N. S., W. N. Gill, and A. J. Barduhn. Laminar
dispersion in capillaries: Part III. Experiments in horizontal
tubes including observations on natural convection effects.
AIChE J. 12:916–923, 1966.
19
Sharp, M. K. Shear-augmented dispersion in non-Newtonian
fluids. Ann. Biomed. Eng. 21:407–415, 1993.
20
Taylor, G. I. Dispersion of soluble matter in solvent flowing
slowly through a tube. Proc. R. Soc. London, Ser. A
219:186–203, 1953.
21
Yu, J. S. Dispersion in laminar flow through tubes by simul-
taneous diffusion and convection. J. Appl. Mech. 48:217–
223, 1981.
Yu, J. S. On laminar dispersion for flow through round tubes.
J. Appl. Mech. 46:750–756, 1979.
23
Yu, J. S. An approximate analysis of laminar dispersion in
circular tubes. J. Appl. Mech. 43:537–542, 1977.