You are on page 1of 15

International Journal of Non-Linear Mechanics 47 (2012) 506–520

Contents lists available at SciVerse ScienceDirect

International Journal of Non-Linear Mechanics


journal homepage: www.elsevier.com/locate/nlm

Modeling and numerical simulation of blood flow using the theory of


interacting continua
Mehrdad Massoudi a,n, Jeongho Kim b, James F. Antaki b
a
U.S. Department of Energy, National Energy Technology Laboratory (NETL), P. O. Box 10940, Pittsburgh, PA 15236, USA
b
Department of Biomedical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA

a r t i c l e i n f o a b s t r a c t

Article history: In this paper we use a modified form of the mixture theory developed by Massoudi and Rajagopal to
Received 26 July 2011 study the blood flow in a simple geometry, namely flow between two plates. The blood is assumed to
Accepted 19 September 2011 behave as a two-component mixture comprised of plasma and red blood cells (RBCs). The plasma is
Available online 22 September 2011
assumed to behave as a viscous fluid whereas the RBCs are given a granular-like structure where the
Keywords: viscosity also depends on the shear-rate.
Mixture Theory Published by Elsevier Ltd.
Blood
Continuum mechanics
Shear-thinning
Two-phase flows

1. Introduction to be the flexibility and alignment of red blood cells (RBCs) at high
shear, and shear-thickening due to the aggregation of cells at low
The safety and efficiency of blood-wetted medical devices are shear (as shown in Fig. 21 of [29]). A further result of the multi-
closely tied to the physical and biological processes governing the component character of blood is the plasma-skimming phenom-
transport of blood. This has led to research in blood rheology, enon, whereby the hematocrit in branches of blood vessels with
trauma, and thrombosis. Despite these efforts, a fundamental size below 300 mm is reduced due to a phase separation and
understanding of device-induced blood trauma remains incom- deficit of RBCs near the wall of the parent vessel [19]. Considera-
plete. Due to a lack of predictive mathematical tools, the primary tion of the cellular component also involves cell–cell interaction
means to study the blood trauma is the costly method of [33] and cell–surface interaction [32]. The migration toward the
experimental trial-and-error. It is therefore necessary to develop centerline and the rotation of the RBCs are believed to increase
more accurate models for blood trauma and hemorheology, the platelet diffusivity and expel the platelet to the near wall
especially for applications in the blood-wetted devices. region, platelet margination [2].
It is known that in large vessels (whole) blood behaves as a Microscopic models of blood flow [14] that account for cell-
Navier–Stokes (Newtonian) fluid (see [28]; however, in a vessel scale lift and drag forces, collisions, deformation, etc. have the
whose characteristic dimension (diameter for example) is about potential to replicate some of the phenomena causing the non-
the same size as the characteristic size of blood cells, blood homogeneous distribution of RBCs and platelets. However, for
behaves as a non-linear fluid, exhibiting shear-thinning and stress most problems of practical value, it is prohibitive to consider the
relaxation. Thurston [82,83] pointed out the viscoelastic behavior three-dimensional dynamic interactions of individual blood
of blood while stating that the stress relaxation is more significant cells—which may amount to several thousand to millions. This
for cases where the shear rate is low. The micro-scale flow and has motivated the pursuit of meso-scale multi-phase (or multi-
deformation of blood have been studied for many years. Early component) models as a reasonable compromise between
in vitro investigations in rotational viscometers or small glass specificity and practicality. Motivated primarily by the plasma-
tubes revealed the characteristic rheological properties such as skimming phenomenon [19], investigators have developed four
the reduction in the blood apparent viscosity [20], Fahraeus effect classes of multi-phase models for blood: edge-core [79], averaging
[25] and Fahraeus–Lindqvist effect [26], revealing the manifested [42], immersed particle [37], and effective medium approach [70].
non-homogeneity of blood in microcirculation. Similarly, the In this paper we advocate using Mixture Theory or the theory
microscopic phenomenon responsible for shear-thinning is found of Interacting Continua, to propose a two-component model for
blood. The large numbers of articles published concerning multi-
component flows typically employ one of the two continuum
n
Corresponding author. theories developed to describe such situations: Mixture Theory
E-mail address: MASSOUDI@NETL.DOE.GOV (M. Massoudi). (or the theory of interacting continua) [73] or Averaging

0020-7462/$ - see front matter Published by Elsevier Ltd.


doi:10.1016/j.ijnonlinmec.2011.09.025
M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520 507

Nomenclature Subscripts

a acceleration vector 1 referring to the constituent 1 (-plasma)


b body force vector 2 referring to the constituent 2 (-Red Blood Cells)
D symmetric part of the velocity gradient m referring to the mixture
I identity tensor
L gradient of velocity vector Superscripts
p fluid pressure
T stress tensor T transpose
v velocity vector
W spin tensor
Other symbols
x position vector

div divergence operator


Greek letters
r gradient operator
tr trace of a tensor
lf second coefficient of viscosity  outer product
m first coefficient of viscosity  dot (scalar) product
r density
ro reference density
f volume fraction

Method(s) [38]. Both approaches are based on the underlying solve numerically the equations of motion for the fully developed
assumption that each component may be mathematically flow of such a mixture between two horizontal flat plates. Finally,
described as a continuum. The Averaging method directly modi- in Section 5, we present numerical solutions for a few cases.
fies the classical transport equations to account for the disconti-
nuities or ‘jump’ conditions at moving boundaries between the
components (cf. [5,24,31,43,42]). Although the two methods seem 2. A brief review of Mixture Theory
similar, the way they approach the formulation of constitutive
models are very different. In fact, as shown in Massoudi [54], Mixture Theory, or the Theory of Interacting Continua, traces
many of the interaction models used by researchers in the its origins to the work of Fick in 1855 (see [71]) and was first
Averaging community are not frame-indifferent, thus violating presented within the framework of continuum mechanics by
basic principles in physics. Other differences between the two Truesdell [85]. It is a means of generalizing the equations and
approaches are explained in Refs. [40,41,62,63]. principles of the mechanics of a single continuum to include any
In order to better understand atherosclerosis, Jung et al. in number of superimposed continua. More detailed information,
2008 [42] used the averaging approach to simulate a three- including an account of the historical development, is available in
component blood flow (RBC–WBC–plasma) in the right coronary the articles by Atkin and Craine [11,12], Bowen [17], Bedford and
artery using a commercial CFD package software, ANSYS Fluent Drumheller [15], and in the books by Truesdell [86], Samohyl [76],
[1], but there are several limitations in that study. For example, and Rajagopal and Tao [73]. Mixture Theory has also been used in
the employed drag model is valid only for solid spherical particles a variety of biomechanics applications [67,66,48,34,46,47,81,
or for fluid particles that are sufficiently small with low concen- 36,30,6,13,7,10,49,9,35].
tration, whereas RBCs in flows of interest are usually non-
spherical and not negligible in size. Further, the 3-dimensional 104
Eulerian–Eulerian code used is not appropriate for such dense Measured Data
concentration of RBCs (45% hematocrit) as studied. Also, Jung’s Nonlin Regression
definition of the relative blood mixture viscosity is not very clear. 103
Massoudi [58] has discussed this issue in more detail. Further-
more, many of constitutive models used in the averaging methods
in general and specifically in Jung et al. [43] are not frame-
Viscosity, cP

102
indifferent, for example, the virtual mass term and the shear lift
force violate the principle of frame-indifference (For more details
see [54]).
In Section 2 of this paper, we provide a brief review of Mixture 101
Theory, and then discuss certain issues in constitutive modeling
of blood. In the present formulation we assume blood to form a
mixture consisting of RBCs suspended in plasma, while ignoring 100
the platelets, the white blood cells (WBCs) and the proteins in the
sample. No biochemical effects or interconversion of mass are
considered in this model. The volume fraction (or the concentra- 10-1
tion of the RBCs) is treated as a field variable. We further assume 10-2 10-1 100 101 102 103
that the plasma behaves as a viscous fluid and the RBCs as an
Rate of Shear, 1/s
anisotropic non-linear density-gradient-type fluid (see [60]). In
Section 3, we discuss the constitutive modeling of the stress Fig. 1. Shear thinning model for viscosity of RBC phase, calibrated to experimental
tensors and the interaction forces. In Section 4, we study and data of [20] for 90% Ht.
508 M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520

Let X1 and X2 denote the positions of particles of S1 and S2 in where r10 is the pure density of the component 1, r20 is the pure
the reference configuration. The motion of the constituents is density of component 2, g is the volume fraction of the compo-
represented by the mappings nent 1, and f is the volume fraction of component 2. For a
saturated mixture g ¼ 1f. The mixture density, rm and the
x1 ¼ v1 ðX1 ,tÞ and x2 ¼ v2 ðX2 ,tÞ: ð1Þ mean velocity vm of the mixture are defined by

These motions are assumed to be one-to-one, continuous, and rm ¼ r1 þ r2 , rm vm ¼ r1 v1 þ r2 v2 : ð4a; bÞ


invertible. The kinematical quantities associated with these Assuming no interconversion of mass between the two con-
motions are: stituents, conservation of mass for the two components are
d1 v1 d2 v2 d2 v2 @v1 @v2 @r1 @r2
v1 ¼ ; v2 ¼ , a2 ¼ , L2 ¼ , L1 ¼ , þdivðr1 v1 Þ ¼ 0, þ divðr2 v2 Þ ¼ 0: ð5a; bÞ
dt dt dt @x1 @x2 @t @t
If T1 and T2 denote the partial stress tensors of the two
W1 ¼ 12 ðL1 LT1 Þ, W2 ¼ 12ðL2 LT2 Þ components, then the equations for the balance of linear momen-
tum are given by
D1 ¼ 12 ðL1 þ LT1 Þ, D2 ¼ 12ðL2 þLT2 Þ ð2Þ
d1 v1 d2 v2
r1 ¼ divT1 þ r1 b1 þ f I , r2 ¼ divT2 þ r2 b2 f I ð6a; bÞ
where v denotes velocity, a is the acceleration, L is the velocity dt dt
gradient, D denotes the symmetric part of the velocity gradient, where b represents the body force and fI represents the interac-
and W is the spin tensor. d1/dt denotes differentiation with tion forces (exchange of momentum) between the components.
respect to t, holding X1 fixed, and d2/dt denotes the same The balance of moment of momentum implies that
operation holding X2 fixed. Also, r1 and r2 are the bulk densities
of the mixture components given by T1 þ T2 ¼ TT1 þ TT2 ð7Þ
which implies that the total stress tensor is symmetric; however,
r1 ¼ gr10 , r2 ¼ fr20 , ð3Þ
the partial stresses need not be symmetric. With these equations
as the basis, in order to solve any two-component flow problem,
one needs to specify or derive the constitutive relations for the
interaction forces (see [54,55]) and the stress tensors. This is
called the ‘closure’ problem in the two-phase community. In the
next section, we discuss the constitutive relations used in
our study.
Fig. 2. Flow between parallel plates located at Y¼  1 and Y¼ 1.

3. Constitutive equations
Table 1

Re 0.1, 1, 100, 1000 Deriving constitutive relations for the stress tensors and the
N 0.1, 0.4, 0.7 interaction forces are among the outstanding issues of research in
C1 1, 10, 100
multicomponent flows. In general, the constitutive expressions
C2 10, 100, 1000
C3 1, 10, 100
for T1 and T2 depend on the kinematical quantities associated
G1 0.1, 1, 10 with both the constituents. However, it can be assumed that T1
B31 10, 50, 100 and T2 depend only on the kinematical quantities associated with
B32 100, 200, 500 the plasma (component 1) and the RBCs (component 2), respec-
k 0, 0.1 20, 60, 100, 1000
tively (sometimes called the principle of component (phase)
Q 1,3
separation, see Adkins [3,4]).

Q=3, N=0.4, C1=1, C2=10, C3=1, G1=0, k=20 Q=3, N=0.4, C1=1, C2=10, C3=1, G1=0, k=20
1 1

0.8 0.8

0.6 0.6
Non-Dimensional Location

Non-Dimensional Location

0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 Re = 0.1 -0.4


Re = 1
-0.6 -0.6 Re = 0.1
Re = 100
Re = 1
-0.8 -0.8 Re = 100
-1 -1
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Non-Dimensional Plasma Velocity Non-Dimensional RBC Velocity

Fig. 3. Effect of Reynolds number (Re) on the plasma velocity (left) and the RBC velocity (right).
M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520 509

Here we provide a brief description of a mixture theory assume that blood is a two-component mixture, composed of
formulation proposed by Massoudi and Rajagopal (M-R). In this the red blood cells (RBCs) suspended in a (platelet rich) plasma.
approach, we consider a mixture of an incompressible fluid In the following description, the plasma in the mixture will be
infused with solid particles, wherein the principles of mechanics represented by S1 and the RBCs by S2. We also assume that f
of granular materials are used to describe the behavior of represents the concentration of the RBCs, also commonly referred
particles. This model has been used and discussed in Massoudi to as hematocrit (Ht). Now, r1 and r2 are the bulk densities of the
[51–55,57–59], Johnson et al. [40,41], Rajagopal et al. [74,72] mixture components given by: r1 ¼ grP , r2 ¼ frRBC , where rP is
Massoudi et al. [63,61], Massoudi and Rao [64], and Ravindran the density of the plasma, rRBC is the density of the RBCs, f is the
et al. [75]. We need to mention that alternatively blood can be volume fraction of the RBCs, and g is the volume fraction of the
viewed as a suspension where it can be modeled using the plasma. Once the individual stress tensors are derived (or pro-
techniques of non-Newtonian fluid mechanics. We, however, posed), a mixture stress tensor can be defined asTm ¼ T1 þ T2 ,
where T1 ¼ ð1fÞTP and T2 ¼ TR , so that the mixture stress
Q=3, N=0.4, C1=1, C2=10, C3=1, G1=0, k=20 tensor reduces to that of plasma as f-0 and to that of a RBCs as
1 g-0. Note that T2 may also be written as T2 ¼ fT^ R , where T^ R may
Re = 0.1 be thought of as representing the stress tensor in a reference
0.8 Re = 1 configuration. Kesmarky et al. [44] show that plasma behaves like
0.6 Re = 100 a Newtonian fluid and for the range of shear rates studied there is
Non-Dimensional Location

no significant departure from linearity. Accordingly, the constitu-


0.4 tive relation for plasma is assumed to be
0.2 T1 ¼ ½pðr1 Þ þ l1 ðr1 ÞtrD1 Iþ 2m1 ðr1 ÞD1 ð8Þ
0 where p is the fluid pressure, I is the identity tensor, m1 is the
-0.2 viscosity and D1 is the symmetric part of the velocity gradient of
the plasma, and l1 is the second coefficient of viscosity. We
-0.4 assume that the mixture is saturated, i.e., g ¼ 1f; furthermore
we assume that
-0.6
pðr1 Þ ¼ pð1fÞ
-0.8
l1 ðr1 Þ ¼ lð1fÞ
-1
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 m1 ðr1 Þ ¼ mð1fÞ ð9Þ
Volume Fraction of RBCs
and thus Eq. (8) becomes
Fig. 4. Effect of Reynolds number (Re) on the volume fraction of RBCs.
T1 ¼ ½pð1fÞ þ lð1fÞtrD1 I þ 2mð1fÞD1 ð10Þ

where p is the pressure, mis the ‘pure’ viscosity, i.e., the viscosity
Table 2
before mixing, and l is the ‘pure’ second coefficient of viscosity.
Flow rates (Q) for plasma and RBC components.
The RBCs in general should be represented as an anisotropic
Reynolds number 0.1 1 100 non-linear density-gradient-type fluid [60]. In the current study,
Qplasma 2.0 2.2 2.5 we use a simplified version of their model which was also
QRBC 1.0 0.8 0.5 proposed by Massoudi and Rajagopal for granular materials.
Qtotal 3.0 3.0 3.0
Qslip 1.0 (33%) 1.4 (47%) 2.0 (66%)
According to this model, the Cauchy stress tensor T depends on
the volume fraction f, the gradient of f, and the symmetric part

Q=3, Re=1, C1=1, C2=10, C3=1, G1=0, k=0 Q=3, Re=1, C1=1, C2=10, C3=1, G1=0, k=0
1 1

0.8 0.8

0.6 0.6
Non-Dimensional Location

Non-Dimensional Location

0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 N = 0.1 -0.4


N = 0.4
-0.6 N = 0.7 -0.6 N = 0.1
N = 0.4
-0.8 -0.8
N = 0.7
-1 -1
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3
Non-Dimensional Plasma Velocity Non-Dimensional RBC Velocity

Fig. 5. Effect of average volume fraction (N) on the plasma velocity (left) and the RBC velocity (right), without shear-thinning.
510 M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520

of the velocity gradient tensor D2 (see Massoudi [53] for details) compressible fluid, b1 and b4 are the material parameters con-
nected with the distribution of the RBCs, b3 is the viscosity, and b5
T2 ¼ ½bo þ b1 rfUrf þ b2 trD2 I þ b3 D2 þ b4 rf  r f þ b5 D22 ð11Þ
is similar to the cross-viscosity of a Reiner–Rivlin fluid. A distinct
where the dot designates the scalar product, ‘r’ the gradient feature of this model is its ability to predict the normal stress
operator, ‘tr’ the trace norm, and ‘’ the outer product of two differences which are often related to dilatancy effects. In this
vectors, wherer2 ¼ rs f, with rs being constant, and the b’sare paper we assume that the viscous effects (b3 ) predominate over
material properties, which in general are functions of the appro- the effects of the gradient of RBC volume fraction, the second
priate principal invariants of D2 . The volume fraction field f(x, t) coefficient of viscosity and the normal stresses. Therefore, it is
in this model represents the spatially dependent concentration of assumed as a first approximation that b1, b2, b4, and b5 are
RBCs, which can be likened to a hematocrit field variable. That is, negligible. Thus, the stress tensor for the RBCs reduces to the
even though the RBCs are acknowledged to be distinct bodies structure
with certain properties (shape, deformability, aggregability, etc.) T2 ¼ b0 I þ b3 D2 ð12Þ
this theory homogenizes the RBC phase by assuming that their
ensemble influence on the flow is a continuous function of where b0 and b3 are given by [60]
position [22]. The material properties b0 y b5, can be construed b0 ¼ pf ð13Þ
to have the following rheological interpretation: b0 is similar to
pressure in a compressible fluid and is to be given by an equation
b3 ðfÞ ¼ b30 ðf þ f2 Þ ð14Þ
of state, b2 corresponds to the second coefficient of viscosity in a

Q=3, Re=1, C1=1, C2=10, C3=1, G1=0, k=20 Q=3, Re=1, C1=1, C2=10, C3=1, G1=0, k=20
1 1

0.8 0.8

0.6 0.6
Non-Dimensional Location

Non-Dimensional Location

0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4
N = 0.1
-0.6 N = 0.4 -0.6
N = 0.7 N = 0.1
-0.8 N = 0.4
-0.8
N = 0.7
-1 -1
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Non-Dimensional Plasma Velocity Non-Dimensional RBC Velocity

Fig. 6. Effect of average volume fraction (N) on the plasma velocity (left) and the RBC velocity (right) with shear-thinning.

Q=3, Re=1, C1=1, C2=10, C3=1, G1=0, k=20 Q=3, Re=1, C1=1, C2=10, C3=1, G1=0, k=20
1 1

0.8 0.8

0.6 0.6
Non - Dimensional Location
Non-Dimensional Location

0.4 0.4

0.2 0.2

0 0

-0.2 -0.2
N = 0.1
-0.4 N = 0.4 -0.4
N = 0.7
-0.6 -0.6
N = 0.1
-0.8 -0.8 N = 0.4
N = 0.7
-1 -1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Volume Fraction of RBCs Volume Fraction of RBCs

Fig. 7. Effect of average volume fraction (N) on the volume fraction of RBCs without shear-thinning (left) and with shear-thinning (right).
M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520 511

The first equation, namely b0 ¼ pf, which is different from viscosity (for infinite shear rate), and k is a material parameter
that suggested by Massoudi and Rajagopal, in conjunction with describing the character of shear thinning. Thus
Eq. (9)a, i.e., pðr1 Þ ¼ pð1fÞ, imply that the total pressure is  
1 þlnð1 þ kg_ Þ 2
weighted (distributed) among the two components according to b3 ðfÞ ¼ m1 þðm0 m1 Þ ðf þ f Þ ð16Þ
1 þ kg_
the volume fraction. This is an accepted assumption in many two-
component theories, and unless there are clear experimental The parameters above are theoretically associated with a
observations pointing otherwise, this is reasonable. Shear-thin- homogeneous continuum comprised entirely of RBCs. This is
ning effects were incorporated by adopting a shear-dependent however infeasible to test experimentally; therefore, the best
viscosity for the RBC phase, proposed by Yeleswarapu et al. [88], practical approximation is to extrapolate from a very high con-
provided. in Eq. (15). The three characteristic constants mN, m0, centration of RBCs for which blood continues to flow. In the
and k are determined by a non-linear regression analysis. It was present case, the seminal experimental data of Chien et al. [20] for
indicated that this viscosity function has the least error over the a 90% Ht suspension of (canine) RBCs were used. (See Fig. 1.)
widest range of shear rates when compared to other popular We now turn our attention to the interaction forces which are
viscosity models. In addition, this viscosity model is able to a critical component of the Mixture Theory. Although these forces
represent the plateau viscosity at very low shear rates are difficult to measure explicitly, arguments from multiphase
flows [54,55] dictate that they are generally a function of the fluid
1 þ lnð1þ kg_ Þ
b30 ¼ m1 þ ðm0 m1 Þ ð15Þ pressure gradient, density gradients, relative velocity, relative
1 þ kg_
acceleration, magnitude of the rate of the deformation tensor of
where g_ ¼ ½2 trðD22 Þ1=2 is the proper norm (generalized shear rate), the fluid, spinning motion as well as the translation of particles
m0 is the viscosity under zero shear rate, mN is an asymptotic (Faxen’s force), tendency of the particles to move toward the

Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, G1=0 Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, G1=0
1 1

0.8 0.8

0.6 0.6
Non-Dimensional Location

Non-Dimensional Location

0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4
k = 0.1
-0.6 k = 100 -0.6 k = 0.1
k = 1000 k = 100
-0.8 -0.8
k = 1000
-1 -1
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Non-Dimensional Plasma Velocity Non-Dimensional RBC Velocity

Fig. 8. Effect of the material constant (k) on the plasma velocity (left) and the RBC velocity (right).

Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, G1=0 Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, G1=0
1 1

0.8 0.8

0.6 0.6
Non - Dimensional Location

Non - Dimensional Location

0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4 k = 0.1


-0.6 -0.6 k = 100
k = 0.1 k = 1000
k = 100
-0.8 -0.8
k = 1000
-1 -1
0 0.5 1 1.5 2 2.5 3 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
Non-Dimensional Mixture Velocity Volume Fraction of RBCs

Fig. 9. Effect of the material constant (k) on the mean velocity of the mixture and the volume fraction of RBCs.
512 M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520

region of higher velocity gradients (Magnus force), history of the approximation, we assume A1 to be constant. Müller’s (1968)
particle motion (Bassett force), temperature gradient and other work indicates that a term of the form A1gradf must be included
secondary effects. If the flow is steady, the virtual mass effects can in the interactions in order to get well-posed problems. Within
be neglected. In the absence of experimental data, various func- the framework of mixture theory developed here, we cannot
tions have been proposed by different researchers based on directly account for particle size, or different particle sizes, or
mathematical arguments, such as the generalization of physics particle shape, or surface roughness, etc. However, since the
of single-particles. Order-of-magnitude analysis suggests that, in coefficients in this equation are determined either by performing
most applications, the spin lift is negligible compared to the shear simple experiments on an assembly of known size and known
lift [39]. For the interaction forces we use a simplified version of a shape particles, or by extrapolating and extending the results of a
model proposed by Johnson et al. [40,41] where single particle, material or geometrical properties can enter in the
constitutive relation through these coefficients. Therefore, we
FI ¼ A1 grad f þ A2 fð1þ 6:55fÞðv2 v1 Þ
have, for example, (cf. Johnson et al. [39] for details)
þ A3 fð2 tr D1 Þ1=4 D1 ðv2 v1 Þ ð17Þ 1=2 1=2
9 mf 3ð6:46Þ rf mf
A2 ¼ , A3 ¼ ð18a; bÞ
where the terms on the right-hand side refer to diffusion, drag, 2 a2 4p a
and shear lift forces, respectively. Appropriate coefficients for where ‘a’ is the particle radius. The actual form for the force due
specific interaction forces (e.g. drag force, lift force, etc.) are to density gradients (see [69]) multiplied by A1should include the
derived from several sources. The drag coefficients and lift terms (a1gradr1 þ a2gradr2) where a1 and a2 are constants. If
coefficients were directly adapted from literature [39]. As a first we assume that the system is a saturated mixture with

Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, G1=0, k=20 Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, G1=0, k=20
1 1

0.8 0.8

0.6 0.6
Non - Dimensional Location

Non - Dimensional Location

0.4 0.4

0.2 0.2

0 0

-0.2 -0.2
B31 = 10
-0.4 -0.4
B31 = 50
B31 = 10
-0.6 -0.6 B31 = 100
B31 = 50
-0.8 B31 = 100 -0.8

-1 -1
0 0.5 1 1.5 2 2.5 3 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
Non-Dimensional Mixture Velocity Volume Fraction of RBCs

Fig. 10. Effect of B31 on the mixture velocity (left) and the volume fraction of RBCs (right).

Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, G1=0, k=20 Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, G1=0, k=20
1
B32=100 1
0.8 B32=100
B32=200 0.8
B32=200
0.6 B32=500
0.6
Non-Dimensional Location

B32=500
Non-Dimensional Location

0.4
0.4
0.2
0.2
0 0
-0.2 -0.2
-0.4 -0.4
-0.6 -0.6

-0.8 -0.8

-1 -1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
Non-Dimensional Mixture Velocity Volume Fraction of RBCs

Fig. 11. Effect of B32 on the mixture velocity (left) and the volume fraction of RBCs (right).
M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520 513

incompressible components, this expression simplifies to 4. Flow between two flat plates
A1gradf, where A1 ¼ a2  a1.
Finally, in our approach no couple stresses are allowed. Substituting Eqs. (10) and (17) in (6)a we obtain the dimen-
Nevertheless, in general, due to the higher order gradients of sionless forms of the two momentum equations in their expanded
the volume fraction, it is necessary to provide additional forms. These are, for the plasma (component 1)
boundary conditions (see Massoudi [57] for a discussion of  
@V1
boundary conditions). For most practical applications, these can ð1fÞrf þðgradV1 ÞV1 ¼ ðgradfÞPð1fÞgradP
@t
be satisfied by certain symmetry conditions; in certain cases  
þ L ðgradfÞdivV1 þ ð1fÞgradðdivV1 Þ þ ð1fÞðgradLÞdivV1
the values of the unknowns or their derivatives have to be
specified as surface conditions at the solid walls or at the free 2  
þ ðgradfÞD1 þ ð1fÞdivD1
surface. Re
In the next section we derive and present the dimensionless rf
þ ð1fÞb1 þ C 1 gradf þC 2 FðfÞðV2 V1 Þ
forms of the governing equations. To gain further insight into the Fr
nature and influence of the various terms in these equations, þ C 3 fð2trD21 Þ1=4 D1 ðV2 V1 Þ ð19Þ
especially the constitutive parameters, we will numerically solve
and for the RBCs (component 2)
the simplified equations for the fully developed flow of blood,  
modeled as a two-component mixture (the plasma and the RBCs), @V2
frs þ ðgradV2 ÞV2 ¼ ðgrad fÞPf grad P
between two long horizontal plates. @t

Q=3, Re=1, N=0.4, C1=1, C2=10, C3=0, k=20 Q=3, Re=1, N=0.4, C1=1, C2=10, C3=0, k=20
1 1

0.8 0.8

0.6 0.6
Non-Dimensional Location

Non-Dimensional Location

0.4 0.4

0.2 0.2
G1=0.1:Plasma
0 0
G1=0.1:RBCs
-0.2 G1=10:Plama -0.2

-0.4 G1=10:RBCs -0.4

-0.6 -0.6 G1=0.1


G1=1
-0.8 -0.8
G1=10
-1 -1
0 0.5 1 1.5 2 2.5 3 0.3 0.32 0.34 0.36 0.38 0.4 0.42 0.44 0.46 0.48 0.5
Non-Dimensional Velocity Volume Fraction of RBCs

Fig. 12. Effect of dimensionless gravity (G1) on the plasma and RBC velocity profiles (left) and the volume fraction of RBCs (right).

Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, k=20 Q=3, Re=1, N=0.4, C1=1, C2=10, C3=1, k=20
1 1
G1=0.1:Plasma G1=0.1
0.8 G1=0.1:RBCs 0.8
G1=1
0.6 G1=10:Plama 0.6 G1=10
Non-Dimensional Location

Non-Dimensional Location

G1=10:RBCs
0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 0.5 1 1.5 2 2.5 3 3.5 4 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
Non-Dimensional Velocity Volume Fraction of RBCs

Fig. 13. Effect of combination of dimensionless gravity and shear lift on plasma and RBC velocity profile (left) and the volume fraction of RBCs (right).
514 M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520

h i h i
2 2 2
þ B31 ðf þ f Þ þ B32 ðf þ f ÞP divD2 þ B31 gradðf þ f Þ D2 where H is some characteristic length, for example half the space
h i between the two flat plates or a tube radius, u0 is a characteristic
2 2
þ B32 gradðf þ f ÞP þðf þ f ÞgradP D2 velocity, r0 is a characteristic density, and x is the position vector
rs (the asterisks have been dropped for simplicity). Finally
þ fb2 C1 gradfC2 FðfÞðV2 V1 Þ
Fr 1þ lnð1 þ kg_ Þ
C3 fð2trD21 Þ1=4 D1 ðV2 V1 Þ
P¼ , g_ ¼ ½2 trðD22 Þ1=2 ð22a; bÞ
ð20Þ 1 þ kg_
where The following dimensionless numbers are identified:
v1 v2 x tu0 r0 u0 H lf u2
V1 ¼ ; V2 ¼ ; x ¼ ; t¼
n
Re ¼ ; L¼ ; Fr ¼ 0
u0 u0 H H mf r0 u0 H Hg
rf rs n b1 n b2
rnf ¼ ; rns ¼ ; b1 ¼ ; b2 ¼ m1 m0 m1
r0 r0 g g B31 ¼ ; B32 ¼ ;
r0 u0 H r0 u0 H
p n n
P¼ ; div ð:Þ ¼ H divð:Þ; grad ð:Þ ¼ H gradð:Þ
r0 u20 A1 A2 H A3 H1=2
h i C1 ¼ ; C2 ¼ ; C3 ¼ ; ð23Þ
r0 u20 r0 u0 r0 u0 1=2
Dn1 ¼ 12 grad V1 þðgrad V1 ÞT
n n

h i where Re is the Reynolds number for the plasma, L is related to


Dn2 ¼ 12 grad V2 þðgrad V2 ÞT
n n
ð21Þ
the second coefficient of viscosity of plasma, Fr is the Froude

Q=3, Re=1, N=0.4, C1=1, C2=10, G1=0.1, k=20 Q=3, Re=1, N=0.4, C1=1, C2=10, G1=0.1, k=20
1 1
C3=0:Plasma C3=0
0.8 0.8
C3=0:RBCs C3=1
0.6 C3=1:Plama 0.6
Non-Dimensional Location

Non-Dimensional Location

C3=1:RBCs
0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 0.5 1 1.5 2 2.5 3 3.5 4 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
Non-Dimensional Velocity Volume Fraction of RBCs

Fig. 14. Effect of lift coefficient (C3) on the plasma and RBC velocities (left) and the volume fraction of RBCs (right).

Q=3, Re=1, N=0.4, C1=1, C2=10, G1=10, k=20 Q=3, Re=1, N=0.4, C1=1, C2=10, G1=10, k=20
1 1
C3=0:Plasma C3=0
0.8 0.8
C3=0:RBCs C3=1
0.6 C3=1:Plama 0.6
Non-Dimensional Location

Non-Dimensional Location

C3=1:RBCs
0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 0.5 1 1.5 2 2.5 3 3.5 4 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
Non-Dimensional Velocity Volume Fraction of RBCs

Fig. 15. Effect of lift coefficient (C3) on the plasma and RBC velocities (left) and the volume fraction of RBCs (right).
M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520 515

Q=3, Re=1, N=0.4, C2=10, C3=1, G1=0, k=20 Q=3, Re=1, N=0.4, C2=10, C3=1, G1=0, k=20
1 1
C1=1 C1=1
0.8 0.8
C1=10 C1=10
0.6 C1=100 0.6 C1=100
Non-Dimensional Location

Non-Dimensional Location
0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Non-Dimensional Plasma Velocity Non-Dimensional RBC Velocity

Fig. 16. Effect of density gradient coefficient (C1) on the plasma velocity (left) and the RBC velocity (right).

number, N is a prescribed number which is an average measure of Q=3, Re=1, N=0.4, C2=10, C3=1, G1=0, k=20
the amount of particles in the system, C1 is related to the 1
coefficient for the forces due to density gradients, C2 is the drag C1=1
0.8
coefficient, C3 is the lift coefficient, G1 is the gravity coefficient C1=10
(a measure of the importance of gravity), B31 and B32 are related 0.6 C1=100
Non-Dimensional Location

to the viscous effects of the RBCs (similar to the Reynolds


0.4
number) and k is a parameter related to shear-thinning effects
of the RBCs, and & accounts for the shear-thinning effect. Our 0.2
intention is to perform a parametric study where different values
of the dimensionless numbers are used. 0
Let us now consider the pressure driven flow of a mixture
-0.2
modeled by Eqs. (10), (12)–(17) between two horizontal long flat
plates, where X is the direction of the flow, and the plates are -0.4
located at Y¼  1 and Y ¼1 (see Fig. 2). If the flow is steady and
-0.6
laminar, the velocity profiles and the volume fraction can be
assumed to have the form -0.8
V1 ¼ VðYÞex
-1
V2 ¼ UðYÞex 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
f ¼ fðYÞ ð24Þ Volume Fraction of RBCs
The equations for balance of mass are automatically satisfied. Fig. 17. Effect of density gradient coefficient (C1) on the volume fraction of RBCs.
Gravity is assumed to be the only body force present. The non-
dimensional momentum equations are greatly simplified by these ku0 Hg_ rp rRBC
assumptions and they can be written in component form as k¼ ,G ¼ ,G1 ¼ and G2 ¼ ð29Þ
H u0 Fr Fr
 
0 dP where G is the dimensionless shear rate, and G1 and G2 are
ð1fÞV 00 f V 0  Reð1fÞ þ C 2 ReFðfÞðUVÞ ¼ 0 ð25Þ
dX dimensionless gravity coefficients. For blood with 50% hematocrit,
   the plasma density is 1025 g/ml and the RBC density is 1125 g/ml
2 1 dP 02 0 where the density ratio of plasma to RBC is 0.9111. Thus, G1/G2 is
ðf þ f Þ B31 þ B32 P þ B32 U U 00 þ ð1 þ2fÞf ðB31 þB32 PÞU 0
G dG 0.9111 and only G1 is considered as a free parameter. Overall, we
  have four non-linear coupled ordinary differential equations
dP
C 2 FðfÞðUVÞf ¼0 ð26Þ (ODEs), which need to be solved numerically.
dX
Based on Sugii et al. [80] who used micro PIV technique to
  
0 dP 1=2 measure velocity distributions of plasma and red blood cells, we
ðP þ C 1 Þf þ þG2 þ C 3 9V 0 9 ðUVÞV 0 f ¼ 0 ð27Þ select a matrix of values for the remaining parameters, summar-
dY
ized in Table 1.
dP Where Q is the dimensionless volumetric flow rate of the
¼ G1 ð1fÞG2 f ð28Þ
dY mixture, defined as
Z 1 Z 1
where
Q¼ V m dY ¼ ½ð1fÞV þ fUdY ð30aÞ
FðfÞ ¼ 1þ 6:55f 1 1

1 þ lnð1þ kGÞ It can be seen that for the reduced forms of the equations, we

1 þ kG need three boundary conditions for V, two conditions for U, and
516 M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520

two conditions for the volume fraction f. We use the adherence corresponding concentration profiles are plotted in Fig. 4. It can
boundary conditions on both constituents at each plate be seen that increasing Re causes a reduction of centerline
velocity for plasma and RBCs, exhibited by blunting of the
Uð1Þ ¼ Uð1Þ ¼ Vð1Þ ¼ Vð1Þ ¼ 0 ð30bÞ
profiles. A similar pattern is also reported in [40] for a mixture
when the mixture is neutrally buoyant i.e., rp ¼ rRBC ¼ r, then of granular particles and a fluid, however only the fluid velocity
Q m ¼ rQ . The appropriate boundary conditions for f are the value was blunter and the granular velocity remained parabolic. The
at a plate and a prescribed average volume fraction defined inclusion of shear-thinning viscosity of the RBC component of the
through present model (Eq. (16)) apparently causes this velocity to also be
Z 1 blunted. The degree of blunting is however disproportionate
N¼ fdY ð31Þ thereby resulting in a greater slip velocity indicated in Table 2
1
which shows the flow rates for each component. It is observed
These conditions may be specified in two ways. Symmetric that increasing the Reynolds number results in an increase in the
solutions may be obtained by specifying that f(1) or f(  1) and flow rate of plasma, but a decrease in that of RBC.
giving a value for N. Solutions that are in general not symmetric Fig. 4 shows the volume fraction of RBCs (f), for different
may be obtained if N and a value for f on one boundary are given. values of Re where average volume fraction (N) is set to 0.4, which
This condition is expressed as is calculated using Eq. (31). Increasing Re causes a non-uniform
distribution of the volume fraction, as observed experimentally
f-Y as Y-1 ð32Þ
by Lih [50].
where Y is a constant. Then f(1) is determined from the solution
of the equation. When gravity is included in the equations, the
symmetric boundary condition, f(1) ¼ f(  1) is not physically Q=3, Re=1, N=0.4, C1=1, C3=1, G1=0, k=20
0
possible since f cannot equal zero anywhere in the domain. In 1
fact, f(  1) and N are specified such that fð1Þ 4 fð1Þ. Numerical C2=10
0.8
values for f on the boundary could be obtained from experiments C2=100
such as those performed by Segre and Silberberg [77,78]. 0.6 C2=1000
Non-Dimensional Location
Eqs. (25)–(28) are highly non-linear ordinary differential
equations, which are solved using a collocation code [8], which 0.4
was found to be suitable to be more robust and more efficient
0.2
than the other methods such as multiple shooting and finite
difference codes [40]. The program implements finite element 0
collocation methods based on piecewise polynomials for the
spatial discrimination techniques. In the next section we discuss -0.2
our approach to the parametric study for the various dimension- -0.4
less numbers.
-0.6

-0.8
5. Numerical results and discussion
-1
5.1. Effect of Reynolds number (Re) 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
Volume Fraction of RBCs
The effects of the (plasma) Reynolds number (Re) on the
velocity profiles of both constituents are shown in Fig. 3. The Fig. 19. Effect of drag coefficient (C2) on the volume fraction of RBCs.

Q=3, Re=1, N=0.4, C1=1, C3=1, G1=0, k=20 Q=3, Re=1, N=0.4, C1=1, C3=1, G1=0, k=20
1 1
C2=10
0.8 0.8
C2=100
0.6 0.6 C2=1000
Non-Dimensional Location

Non-Dimensional Location

0.4 0.4

0.2 0.2

0 0
C2=10
-0.2 -0.2
C2=100
-0.4 C2=1000 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Non-Dimensional Plasma Velocity Non-Dimensional RBC Velocity

Fig. 18. Effect of drag coefficient (C2) on the plasma velocity (left) and the RBC velocity (right).
M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520 517

Q=3, Re=1, N=0.4, C1=10, C2=100, G1=0, k=20 Q=3, Re=1, N=0.4, C1=10, C2=100, G1=0, k=20
1 1
C3=1 C3=1
0.8 0.8
C3=10 C3=10
0.6 C3=100 0.6 C3=100
Non-Dimensional Location

Non-Dimensional Location
0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Non-Dimensional Plasma Velocity Non-Dimensional RBC Velocity

Fig. 20. Effect of lift coefficient (C3) on the plasma velocity (left) and the RBC velocity (right).

Q=3, Re=1, N=0.4, C1=10, C2=100, G1=0, k=20 understand the physical behavior of dilute particles in the field of
1 biomedical engineering research [45].
C3=1
0.8 C3=10
5.3. Effects of k, B31, and B32
0.6 C3=100
Non-Dimensional Location

0.4 Fig. 8 shows the velocity profiles for values of k ranging from
0.1 to 1000. The centerline velocity of RBCs is found to increase
0.2 with increasing k, gradually becoming parabolic as k approaches
0 1000. A similar pattern for the mean velocity of the mixture
shown in Fig. 9, is also found in Yeleswarapu [87]. The non-
-0.2 dimensional mean velocity can be calculated using the following
equation:
-0.4
V m ¼ ð1fÞV þ fU ð33Þ
-0.6
where Vm is the non-dimensional mixture velocity, V is the non-
-0.8 dimensional plasma velocity, U is the non-dimensional RBC
-1 velocity, and f is the volume fraction of the RBC component. As
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 k approaches infinity, the viscosity depends on B31, which implies
that the fluid behaves as a Newtonian fluid. Thus, it is observed in
Volume Fraction of RBCs
Figs. 8 and 9 that both plasma and RBC velocities become more
Fig. 21. Effect of lift coefficient (C3) on the volume fraction of RBCs. parabolic. Conversely, increasing k represents greater shear-thin-
ning, leading to increased bluntness of the velocity profile.
Fig. 9 illustrates the influence of k upon the concentration profile
5.2. Effect of average volume fraction (N) of RBCs (the F-L effect.) As k approaches 0.1, there is approximately
a 12% depletion of RBCs near the wall and a 7% excess at the
The parameter N (related to Hct, hematocrit) is found to affect centerline. When k is much greater than 100, the variation from
the plasma and the RBC velocity profiles differently, depending on core to wall decreases below 2%, and becomes nearly uniform.
the inclusion of shear thinning. In the absence of the shear- Fig. 10 shows the effect of B31. In this range, i.e., from 10 to
thinning effects (k ¼0), the plasma velocity profile is blunter 100, there is negligible influence on the velocity and the volume
while the RBC velocity profile is more parabolic. (See Fig. 5.) By fraction distribution. By contrast, a 6 times change in B32 causes a
contrast, when shear-thinning is included, increasing N results in far more prominent effect on the bluntness of the velocity
a decrease in maximum velocity and further blunting of plasma distribution and the near-wall depletion of RBCs (see Fig. 11). A
and the RBC velocity. (See Fig. 6.) similar pattern is found in Yeleswarapu [87].
Fig. 7 demonstrates that the effect of increasing the overall
volume fraction of RBCs (hematocrit N) upon the concentration 5.4. Effect of the gravitational parameter (G1)
profile is much more pronounced above 0.1. Below this value, the
concentration profile is exponential, which implies that blood can Additional simulations were performed in which the gravita-
be regarded as a dilute suspension where RBC particle-particle tional parameter G1 was varied from 0.1 to 10. Fig. 12 shows
interactions and particle-plasma interactions could be ignored the velocity profiles and the volume fraction for two values of the
[65]. Thus, the plasma component affects the RBC velocity but not dimensionless gravity coefficient G1 without lift force where the
vice versa; this type of flow is called a one-way coupling. This dimensionless lift coefficient (C3) is set to zero. Increasing G1 is
assumption has been widely used in numerical simulations to observed to cause a non-symmetric velocity profile for both the
518 M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520

Q=3, Re=1, N=0.4, C1=10, C2=100, G1=1, k=20 Q=3, Re=1, N=0.4, C1=10, C2=100, G1=1, k=20
1 1
C3=1 C3=1
0.8 0.8
C3=10 C3=10
0.6 C3=100 0.6 C3=100
Non-Dimensional Location

Non-Dimensional Location
0.4 0.4

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Non-Dimensional Plasma Velocity Non-Dimensional RBC Velocity

Fig. 22. Effect of lift coefficient (C3) on the plasma velocity (left) and the RBC velocity (right).

plasma and the RBCs; thus f( 1)a f(1). This observation is also Q=3, Re=1, N=0.4, C1=10, C2=100, G1=1, k=20
reported in Johnson et al. [40] who found that the particle velocity 1
became skewed towards the lower plate. The volume fraction C3=1
0.8
distribution is affected in a similar fashion. At the upper plate C3=10
(Y¼1) it is seen that f increases as G1 increases, indicating a 0.6 C3=100
Non-Dimensional Location

translocation of RBCs in the direction of the gravity force. This effect


0.4
is unlikely to be observable at Reynolds 41, as the magnitude of the
gravitational force is one tenth or less of the drag force 0.2
It is interesting to note that the addition of the (shear) lift force
creates the opposite effect to that of gravity. (See Fig. 13.) Under 0
these conditions, the RBC velocity is skewed away from the lower -0.2
plate towards the upper plate. However, the RBC volume fraction
was skewed towards the bottom wall (in the direction of gravity) -0.4
as observed in the previous case (without the lift force). (See
-0.6
Fig. 12.) Experimental observation [25] shows that RBCs migrate
towards the center and collect there (symmetrically). This indi- -0.8
cates that the magnitude of gravity is much smaller than that of
lift force to cause the migration of the RBCs. -1
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
When the lift force is more dominant than the gravity, that is
for larger values of C2, the velocity profile becomes more para- Volume Fraction of RBCs
bolic. It is also observed in Fig. 14 that the lift force has a Fig. 23. Effect of lift coefficient (C3) on the volume fraction of RBCs.
significant role in causing a decrease between the plasma velocity
and the RBCs velocity. The RBCs distribution could vary from a
uniform distribution to a symmetrical one due to the dominant RBCs to be more parabolic, and the RBCs distribution to be more
effect of the lift force when compared to gravity. However, as G1 uniform. (See Figs. 18 and 19.) Unlike the case for increasing the
increases, the RBCs are distributed more non-uniformly (Fig. 15). diffusion term, increasing the drag coefficient (C2) results in an
This observation is also found in Johnson et al. [41]. increase in the slip velocity.

5.5. Effect of the diffusion coefficient (C1) 5.7. Effect of the lift coefficient (C3)

Fig. 16 indicates that as the value of the density gradient A parametric study of the lift coefficient C3 in the range of
coefficient C1 becomes larger, both plasma and RBC velocity profiles 1–100 was performed and it is observed to minimally influence the
become more parabolic and the centerline slip velocity between the velocity profiles of RBCs and plasma, but dramatically affects the
two constituents increases. In addition, as C1 increases the RBCs RBCs distribution. Increasing C3 by a factor of 10 (from 1 to 10 and
volume fraction becomes more uniform, effectively counteracting 10 to 100) greatly accentuated the near-wall depletion of RBCs,
the influence of lift and drag (see Fig. 17). Based on the numerical causing them to collect toward the core (see Figs. 20 and 21).
observations it can be implied that the magnitude of the diffusion Experimental and numerical observations demonstrate that RBCs
force could be less than that of the lift force. migrate towards the tube axis [16,32,90,89]. In Fig. 20, it can be seen
that as C3 increases, there is an, albeit small, reduction in the plasma
5.6. Effect of the drag coefficient (C2) velocity in regions of higher RBCs concentration and an increase in
the plasma velocity in the regions of lower RBCs concentration. That
Similar to the effect of diffusion, the influence of the drag is, the RBCs move slower in the regions of higher concentration and
coefficient is to cause the velocity profiles of the plasma and the faster in the regions of lower concentration.
M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520 519

Non-linear distribution of the volume fraction causes an should approach zero near the wall due to the depletion of the
increase in the plasma velocity near the walls where the RBCs RBCs. The other possibility is to consider a modified lift force or to
volume fraction decreases and a decrease in the plasma velocity add another force due to the wall effect also called a lubricant
where the RBCs concentration increases (Figs. 22 and 23). effect. This might cause infinite force acting on the RBCs when
they approach the wall, thus resulting in the RBC depletion near
wall. From the current study, we conclude that the drag and lift
6. Concluding remarks forces are the most important ones for hemodynamics applica-
tions. Thus, the next step is to focus on improving or identifying
The constitutive equation used in our study for the stress better forms for the interaction forces [23,21,27]. For the future
tensor of the RBCs is assumed to be isotropic, but in reality the studies, we intend to consider the complete model shown in
RBCs are anisotropic and deformable. The difficulty is the orienta- Eq. (11).
tion or the alignment of the non-spherical cells [20,90]. This could
be addressed by using an anisotropic representation for the RBCs,
as proposed by Massoudi and Antaki [60], in which the stress Acknowledgment
tensor is a function of the symmetric part of velocity gradient, D2,
and the orientation vector, n. This was not employed here as it This project was supported by NIH R01 HL089456-01.
would have introduced yet additional governing conservation
equations such as the balance of angular momentum, additional
boundary conditions, and associated constitutive parameters References
[56,60]. It is nevertheless a worthy topic of future study.
According to Muller [68], a force of the type C1grad j, must be [1] ANSYS FLUENT 12.0 User Guide.
included in the interaction forces for the problem of multi- [2] P. Aarts, S.A.T. Vandenbroek, et al., Blood-platelets are concentrated near the
wall and red blood-cells, in the center in flowing blood, Arteriosclerosis 8 (6)
component flows to be well-posed. We also observe the impor- (1988) 819–824.
tant role of the diffusion term in the numerical simulations. [3] J.E. Adkins, Non-linear diffusion, 1. diffusion and flow of mixtures of fluids,
Especially when the lift force is included, the diffusion term is Philosophical Transactions Royal of Society London A 255 (1963) 607–633.
[4] J.E. Adkins, Non-linear diffusion, 2. constitutive equations for mixtures of
critically important to make the simulations stable. A similar
isotropic fluids, Philosophical Transactions of Royal Society London A 255
numerical observation is also found in Johnson et al. [40,41] in (1963) 635–648.
which the gradient of the volume fraction of the granular [5] T.B. Anderson, R. Jackson, A fluid mechanical description of fluidized beds,
Industrial & Engineering Chemistry Fundamentals 6 (4) (1967) 527.
component, similar to the diffusion was found necessary for
[6] R.P. Araujo, D.L.S. Mcelwain, A mixture theory for the genesis of residual
stability of the simulations. We also notice another important stresses in growing tissues I, Siam Journal on Applied Mathematics 65 (2005)
relationship between C1, C2, and C3 from the sensitivity studies 1261.
performed here. If C3/C2 is greater than C1, the numerical simula- [7] R.P. Araujo, D.L.S. McElwain, A mixture theory for the genesis of residual
stresses in growing tissues II: solutions to the biphasic equations for a
tions are unstable. In general, it is found that C1 should be an multicell spheroid, Siam Journal on Applied Mathematics 66 (2) (2006)
order of magnitude greater than C3/C2 to assure numerical 447–467.
stability. This can be used to provide a limiting value of the [8] U. Ascher, J. Christiansen, et al., Collocation software for boundary-value
ODEs, ACM Transactions on Mathematical Software 7 (2) (1981) 209–222.
diffusion coefficient for future 3-dimensional simulations. [9] G.A. Ateshian, On the theory of reactive mixtures for modeling biological
There have been no investigations pointing to the form of A1 growth, Biomechanics and Modeling in Mechanobiology 6 (6) (2007)
(or C1) in Eq. (17). The remaining coefficients have been studied to 423–445.
[10] G.A. Ateshian, M. Likhitpanichicul, et al., A mixture theory analysis for
a limited extent for the general two-component flows. Thus, the passive transport in osmotic loading of cells, Journal of Biomechanics 39
forms given above are ad-hoc and they are perhaps valid under (3) (2006) 464–475.
strict conditions. The forms given to the coefficient of the [11] R.J. Atkin, R.E. Craine, Continuum theories of mixtures: applications, Journal
of the Institute of Mathematics and Its Applications 17 (1976) 153–207.
interaction forces are either generalization of a single particle [12] R.J. Atkin, R.E. Craine, Continuum theories of mixtures: basic theory and
result or due to some other limiting conditions. historical development, Quarterly Journal of Mechanics and Applied Mathe-
Furthermore, many of the dimensionless numbers defined in matics 29 (1976) 209–244.
[13] N.K. Axtell, M. Park, et al., Micromorphic fluid in an elastic porous body:
Eq. (22) would not be correct or appropriate for non-spherical
blood flow in tissues with microcirculation, International Journal for Multi-
particles such as fibers or blood cells. Despite the assumptions scale Computational Engineering 3 (1) (2005) 71–83.
involved, these expressions in many ways effectively describe [14] P. Bagchi, Mesoscale simulation of blood flow in small vessels, Biophysical
how the interaction forces vary with the system parameters. Journal 92 (6) (2007) 1858–1877.
[15] A. Bedford, D.S. Drumheller, Recent advances: theories of immiscible and
Another limitations of the constitutive equations for the RBC structured mixtures, International Journal of Engineering Science 21 (1983)
component is the lack of explicit consideration of RBCs aggrega- 863–960.
tion [20] and elasticity [82–84]. Instead, these phenomena have [16] E.H. Bloch, A quantitative study of hemodynamics in living microvascular
system, American Journal of Anatomy 110 (2) (1962) 125.
been incorporated into a shear thinning viscosity function, which [17] R.M. Bowen (Ed.), Theory of mixtures. Continuum Physics, Academic,
in turn is based on experiments in whole blood. New York, 1976.
When blood flows through a micro-scale tube, it is observed [18] G. Bugliarello, C. Kapur, et al., The profile viscosity and other characteristics
of blood flow in a non-uniform shear field, in: Proceedings of the fourth
that under certain conditions the red blood cells (RBCs) migrate interantional rheology conference.
away from the wall of the tube and concentrate toward the center [19] R.T. Carr, L.L. Wickham, Plasma skimming in serial microvascular bifurca-
of the tube, thus causing a cell-free layer near the wall, called the tions, Microvascular Research 40 (2) (1990) 179–190.
[20] S. Chien, S. Usami, et al., Effects of hematocrit and plasma proteins on human
plasma-skimming layer. The other observed phenomenon is the blood rheology at low shear rates, Journal of Applied Physiology 21 (1) (1966) 81.
reduction of the apparent viscosity near the wall. Both experi- [21] R. Clift, J.R. Grace, et al., Bubbles, drops, and particles, Dover Publications,
mental observations were reported first by Fahraeus and Lindq- 2005.
[22] I.F. Collins, Elastic/plastic models for soils and sands, International Journal of
vist [18,25,26]. One of the limitations in our study is that the RBC
Mechanical Sciences 47 (2005) 493–508.
cell free layer is not exhibited. No theoretical studies have been [23] C. Crowe, M. Sommerfeld, et al., Multiphase flows with droplets and particles,
able to describe this phenomenon. One possible solution could be CRC Press, 1997.
to consider a modified form of the viscosity model proposed by [24] D.A. Drew, L.A. Segel, Averaged equations for 2-phase flows, Studies in
Applied Mathematics 50 (3) (1971) 205.
Yeleswarapu, which could be a function of RBC concentration as [25] R. Fahraeus, The suspension stability of the blood, Physiological Reviews 9 (2)
well as the shear rate. As seen experimentally, the RBC viscosity (1929) 241–274.
520 M. Massoudi et al. / International Journal of Non-Linear Mechanics 47 (2012) 506–520

[26] R. Fahraeus, T. Lindqvist, The viscosity of the blood in narrow capillary tubes, [58] M. Massoudi, A note on the meaning of mixture viscosity using the classical
American Journal of Physiology 96 (3) (1931) 562–568. continuum theories of mixtures, International Journal of Engineering Science
[27] L.S. Fan, Principles of Gas-Solid Flows, Cambridge University Press, 2005. 46 (7) (2008) 677–689.
[28] Y.C. Fung, Biomechanics: Mechanical Properties of Living Tissues, Springer, [59] M. Massoudi, A Mixture Theory formulation for hydraulic or pneumatic
1993. transport of solid particles, International Journal of Engineering Science 48
[29] G.P. Galdi, R. Rannacher, et al., Hemodynamical Flows: Modeling, Analysis (11) (2010) 1440–1461.
and Simulation, Birkhäuser Basel, 2008. [60] M. Massoudi, J.F. Antaki, An anisotropic constitutive equation for the stress
[30] K. Garikipati, E.M. Arruda, et al., A continuum treatment of growth in tensor of blood based on Mixture Theory, Mathematical Problems in
biological tissues: the coupling of mass transport and mechanics, Journal of Engineering (2008) Article ID 579172, 31 pages.
the Mechanics and Physics of Solids 52 (2004) 1595–1625. [61] M. Massoudi, G. Johnson, On the flow of a fluid-particle mixture between two
[31] D. Gidaspow, Multiphase Flow and Fluidization: Continuum and Kinetic rotating cylinders, using the theory of interacting continua, International
Theory Descriptions, Academic Press, 1994. Journal of Non-Linear Mechanics 35 (6) (2000) 1045–1058.
[32] H.L. Goldsmith, Red cell motions and wall interactions in tube flow, Federa- [62] M. Massoudi, K.R. Rajagopal, et al., Remarks on the modeling of fluidized
tion Proceedings 30 (5) (1971) 1578. systems, AIChE Journal 38 (3) (1992) 471–472.
[33] H.L. Goldsmith, K. Takamura, et al., Shear-induced collisions between [63] M. Massoudi, K.R. Rajagopal, et al., On the fully developed flow of a dense
human-blood cells, Annals of the New York Academy of Sciences 416 particulate mixture in a pipe, Powder Technology 104 (3) (1999) 258–268.
(1983) 299–318 (DEC). [64] M. Massoudi, C.L. Rao, Vertical flow of a multiphase mixture in a channel,
[34] W.Y. Gu, W.M. Lai, et al., A mixture theory for charged-hydrated soft tissues Mathematical Problems in Engineering 6 (6) (2001) 505–526.
[65] S. Middleman, Trasport phenomena in the cardiovascular system, Wiley-
containing multi-electrolytes: passive transport and swelling behaviors,
Interscience, 1972.
Journal of Biomechanical Engineering–Transaction of the ASME 120 (2)
[66] V.C. Mow, M.H. Holmes, et al., Fluid transport and mechanical properties of
(1998) 169–180.
articular cartilage: a review, Journal of Biomechanics (1984) 377–394.
[35] J.D. Humphrey, S. DeLange, An Introduction to Biomechanics: Solids and
[67] V.C. Mow, S.C. Kuei, et al., Biphasic creep and stress relaxation of articular
Fluids, Analysis and Design, Springer, 2004.
cartilage in compression? Theory and experiments, Journal of Biomechanical
[36] J.D. Humphrey, K.R. Rajagopal, A constrained mixture model for growth and
Engineering 102 (1) (1980) 73–84.
remodeling of soft tissues, Mathematical Models & Methods in Applied [68] I. Muller, A thermodynamic theory of mixtures of fluids, Archive for Rational
Sciences (2002) 12407–12430. Mechanics and Analysis 28 (1) (1968) 1.
[37] S. Hyun, C. Kleinstreuer, et al., Hemodynamics analyses of arterial expansions [69] I. Muller, The influecne of density gradients on forces in a mixture, Journal of
with implications to thrombosis and restenosis, Medical Engineering & Non-Equilibrium Thermodynamics 2 (1977) 133–138.
Physics 22 (1) (2000) 13–27. [70] R. Pal, Rheology of concentrated suspensions of deformable elastic particles
[38] M. Ishii, Thermo-Fluid Dynamic Theory of Two-Phase Flow, Eyrolles, 1975. such as human erythrocytes, Journal of Biomechanics 36 (7) (2003) 981–989.
[39] G. Johnson, M. Massoudi, et al. A Review of Interaction Mechanisms in Fluid- [71] K.R. Rajagopal, On a hierarchy of approximate models for flows of incom-
Solid Flows, DOE/PETC/TR-90/9, 1990. pressible fluids through porous solids, Mathematical Models and Methods in
[40] G. Johnson, M. Massoudi, et al., Flow of a fluid infused with solid particles Applied Science 17 (2007) 215–252.
through a pipe, International Journal of Engineering Science 29 (6) (1991) [72] K.R. Rajagopal, M. Massoudi, et al., Flow of granualar-materials between
649–661. rotating disks, Mechanics Research Communications 21 (6) (1994) 629–634.
[41] G. Johnson, M. Massoudi, et al., Flow of a fluid solid mixture between flat [73] K.R. Rajagopal, L. Tao, Mechanics of Mixtures, World Scientific Publishing,
plates, Chemical Engineering Science 46 (7) (1991) 1713–1723. Singapore, 1995.
[42] J. Jung, A. Hassanein, Three-phase CFD analytical modeling of blood flow, [74] K.R. Rajagopal, W. Troy, et al., Existence of solutions to the equations
Medical Engineering & Physics 30 (1) (2008) 91–103. governing the flow of granular materials, European Journal of Mechanics
[43] J.H. Jung, A. Hassanein, et al., Hemodynamic computation using multiphase B-Fluids 11 (3) (1992) 265–276.
flow dynamics in a right coronary artery, Annals of Biomedical Engineering [75] P. Ravindran, N.K. Anand, et al., Steady free surface flow of a fluid-solid
34 (3) (2006) 393–407. mixture down an inclined plane, Particulate Science and Technology 22 (3)
[44] G. Kesmarky, P. Kenyeres, et al., Plasma viscosity: a forgotten variable, (2004) 253–273.
Clinical Hemorheology and Microcirculation 39 (1–4) (2008) 243–246. [76] I. Samohyl, Thermomechanics of Irreversible Processes in Fluid Mixtures,
[45] C. Kleinstreuer, Biofluid Dynamics: Principles and Selected Applications, CRC Tuebner, Leipzig, 1987.
Press, 2006. [77] G. Segre, A. Silberberg, Behaviour of macroscopic rigid spheres in Poiseuille
[46] S.M. Klisch, J.C. Lotz, A special theory of biphasic mixtures and experimental flow Part 1. Determination of local concentration by statistical analysis of
results for human annulus fibrosus tested in confined compression, ASME particle passages through crossed light beams, Journal of fluid mechanics 14
Journal of Biomechanical Engineering 122 (2000) 180–188. (1962) 115–135.
[47] S. Kuhn, W. Hauger, A theory of the adaptive growth of biological materials, [78] G. Segre, A. Silberberg, Behaviour of macroscopic rigid spheres in Poiseuille
Archive of Applied Mechanics 70 (1–3) (2000) 183–192. flow Part 2. Experimental results and interpretation, Journal of Fluid
[48] W.M. Lai, J.S. Hou, et al., A triphasic theory for the swelling and deformation mechanics 14 (1962) 136–157.
behaviors of articular-cartilage, Journal of Biomechanical Engineering— [79] M. Sharan, A.S. Popel, A two-phase model for flow of blood in narrow tubes with
Transaction of the ASME 113 (3) (1991) 245–258. increased effective viscosity near the wall, Biorheology 38 (5–6) (2001) 415–428.
[49] G. Lemon, J.R. King, et al., Mathematical modelling of engineered tissue [80] Y. Sugii, R. Okuda, et al., Velocity measurement of both red blood cells and
growth using a multiphase porous flow mixture theory, Journal of Mathe- plasma of in vitro blood flow using high-speed micro PIV technique,
Measurement Science & Technology 16 (5) (2005) 1126–1130.
matical Biology 52 (2006) 571–594.
[81] L. Tao, J.D. Humphrey, et al., A mixture theory for heat-induced alterations in
[50] M.M. Lih, A mathmatical model for axial migration of suspended particles in
hydration and mechanical properties in soft tissues, International Journal of
tube flow, Bulletin of Mathematical Biophysics 31 (1) (1969).
Engineering Science 39 (2001) 1535–1556.
[51] Massoudi, M. Application of Mixture Theory to Fluidized Beds. Ph.D. Dis-
[82] G.B. Thurston, Viscoelasticity of human blood, Biophysical Journal 12 (9)
sertation, Unviersity of Pittsburgh, 1986.
(1972) 1205.
[52] M. Massoudi, Stability analysis of fluidized beds, International Journal of
[83] G.B. Thurston, Frequency and shear rate dependence of viscoelasticity of
Engineering Science 26 (7) (1988) 765–769.
human blood, Biorheology 10 (3) (1973) 375–381.
[53] M. Massoudi, On the flow of granular materials with variable material
[84] G.B. Thurston, The elastic yield stress of human blood, Biomedical Sciences
properties, International Journal of Non-Linear Mechanics 36 (1) (2001) Instrumentation (1993) 87–93.
25–37. [85] C. Truesdell, Sulle basi della thermomeccanica, Rendiconti Lincei 8 (22)
[54] M. Massoudi, On the importance of material frame-indifference and lift (1957) 33–38 158–166.
forces in multiphase flows, Chemical Engineering Science 57 (17) (2002) [86] C. Truesdell, Rational Thermodynamics, Springer-Verlag, 1984.
3687–3701. [87] K.K. Yeleswarapu, Evaluation of continuum models for characterizing the con-
[55] M. Massoudi, Constitutive relations for the interaction force in mmulticom- stitutive behavior of blood, Ph.D. dissertation, University of Pittsbugh, 1994.
ponent particulate flows, Internatinal Journal of Non-Linear Mechanics 38 [88] K.K. Yeleswarapu, M.V. Kameneva, et al., The flow of blood in tubes: theory
(2003) 316–336. and experiment, Mechanics Research Communications (1998) 25.
[56] M. Massoudi, An anisotropic constitutive relation for the stress tensor of a [89] J.F. Zhang, P.C. Johnson, et al., Effects of erythrocyte deformability and
rod-like (fibrous-type) granular material, Mathematical Problems in Engi- aggregation on the cell free layer and apparent viscosity of microscopic
neering 6 (2005) 679–702. blood flows, Microvascular Research 77 (3) (2009) 265–272.
[57] M. Massoudi, Boundary conditions in mixture theory and in CFD applications [90] R. Zhao, J.N. Marhefka, et al., Micro-flow visualization of red blood cell-
of higher order models, Computers & Mathematics with Applications 53 (2) enhanced platelet concentration at sudden expansion, Annals of Biomedical
(2007) 156–167. Engineering 36 (7) (2008) 1130–1141.

You might also like