You are on page 1of 35

Comp. by: Prabhu Stage: Proof Chapter No.

: 14 Title Name: Grotberg


Date:23/2/21 Time:06:28:30 Page Number: 1

14 Constitutive Equations 3:
Viscoelastic Fluids

Introduction from the interaction of suspended particles and


entangled macromolecules, see Figure 10.4, that form
This is our final foray into constitutive equations for these complex biofluids like applesauce, mayonnaise,
biofluids. In Chapter 6 we discussed inviscid fluids blood and mucus.
whose stress tensor is comprised only of the pressure, In the present chapter, we extend further into
Tij ¼ pδij . Inserting this form into the Cauchy complex fluid behavior by allowing the fluid to have
momentum equation, Eq. (5.2.2), gave us Euler’s “memory.” What do we mean by memory? When a
equation, Eq. (6.1.5), and its integral along a stress is applied to an elastic solid it creates a
streamline, the Bernoulli equation, see Eqs. deformation, and when that stress is removed the solid
(6.2.7–6.2.9). Next we added viscous effects through returns to its original configuration. The elastic solid has
the extra-stress tensor, τij , to give the full stress tensor “memory” of its initial state. When a fluid has memory it
as Tij ¼ pδij þ τij . For a Newtonian fluid the extra- is said to be viscoelastic. So instead of the viscous effect
stress tensor is proportional to the strain rate tensor depending on the instantaneous strain rate, it can
through the constant viscosity, μ, as τij ¼ 2μEij . Then incorporate the history of the strain rate. To that end, the
in Chapter 10 we analyzed generalized Newtonian constitutive equation for a viscoelastic fluid is, itself, a
fluids whose constitutive relationship looks like that partial differential equation. These may include time and
of a Newtonian fluid, except the effective viscosity, spatial derivatives of the stress tensor and velocity field.
μeff , depends on the instantaneous value of the As a consequence, it is coupled to the Cauchy
pffiffiffiffiffiffiffiffiffiffiffiffi
invariant strain rate scalar γ_ ¼ 2Eij Eij , so that momentum equation, so the two partial differential
τij ¼ 2μeff ðγ_ ÞEij . The nomenclature describing equations must be solved simultaneously. As we saw
generalized Newtonian fluids includes shear-thinning with other non-Newtonian fluids, viscoelastic behavior
(pseudoplastic) and shear-thickening (dilatants) fluids also arises from the entangled macromolecules within
as well as fluids with a yield stress. We explored the fluid. They exhibit elastic properties like resistance to
various yielded models including power-law, Carreau, bending and stretching. It turns out that blood, mucus
Carreau–Yasuda, modified Cross and Quemada. For and synovial fluid are all viscoelastic to varying degrees,
the broad topic of fluids with a yield stress, Herschel– which can influence their function and vary with
Bulkley fluids, we examined yield-stress + power-law disease. A familiar personal experience is the increase of
and Casson models. The shear strain dependence of viscosity and elastic behavior of nasal mucus during an
the viscosity, as well as the yield behavior, comes upper respiratory infection.

...........................................................................................................................................................................................

Topics Covered

We begin the chapter by introducing basic concepts of viscoelastic materials and


the well-known Maxwell fluid model which consists of a spring and linear dashpot
in series. From this configuration, step inputs of displacement or force are
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:31 Page Number: 2

2 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

examined, revealing the concept of a stress-relaxation time, λ ¼ μ=k, where μ is


the dashpot viscosity and k is the spring constant. This is the time it takes for the
force to relax to a steady (lower) state following a step change in displacement.
Then we examine oscillatory shear forcing, with k replaced by the shear modulus,
G, displacement replaced by shear strain, γ, and force by shear stress, τ. Assuming
the forms τ ¼ τ0 eiωt , γ ¼ γ0 eiωt , the response is treated first as a viscoelastic
solid, τ0 ¼ G∗ γ0 , relating stress amplitude,τ0 , to strain amplitude, γ0 . The
00
complex shear modulus is G∗ ¼ G0 þ iG where the shear storage modulus, G0 ,
00
and the loss modulus, G , are functions of ω, G, μ and examined in detail. From
the analysis, an important dimensionless parameter arises, De ¼ ωλ, the Deborah
number. De<<1 indicates low frequency oscillations where the relaxation time is
short compared to the oscillation period and viscous effects dominate. For De>>1
the system is at high frequency with the oscillation period short compared to the
relaxation time and elastic effects dominate. Then the deformation is treated as a
viscoelastic fluid, τ0 ¼ μ∗ ðiωγ0 Þ, relating stress amplitude to the strain rate
00
amplitude, iωγ0 . The complex viscosity is μ∗ ¼ μ0  iμ , which is easily shown to
00 00
be related to μ0 ¼ G =ω and μ ¼ G0 =ω. We revisit Stokes’ second problem of
Section 8.2, the semi-infinite fluid bounded by a wall oscillating in its own plane,
and study the influence of De on the velocity profiles compared to the Newtonian
De ¼ 0 case. Similarly, we analyze oscillatory channel flow, as in Section 8.3.
Because many biofluids are mixtures of long chain polymers of differing lengths
and properties, a single relaxation time does not model the system well, so a
generalized Maxwell model is explored which consists of multiple Maxwell
elements in parallel, each with a different spring constant and viscosity. Trying to
apply the Maxwell model directly to a flow problem shows an important defect,
that it does not transform properly under a Galilean coordinate transformation. So
more sophisticated models are explored, including the upper convected Maxwell
00
and Oldroyd B. Data for G0 , G , and jμ∗ jas functions of oscillatory frequency,
ω, and μeff as a function of steady shear rate, γ, _ are discussed for blood, mucus
and synovial fluid. A very important relationship is the Cox–Merz rule (Cox and
Merz, 1958), which states that the steady effective viscosity is equal to the
magnitude of the complex viscosity, when γ_ ¼ ω, i.e. μeff ðγ_ Þ ¼ jμ∗ ðωÞjγ¼ω_ . We
show it holds fairly well for these biofluids. To gain more insight into synovial fluid
and joint space mechanics, we revisit the squeeze film model of Chapter 11, but
use an oscillatory force as one might experience in walking or running. For a
Maxwell fluid model, we examine how the displacement amplitude of the plates
depends on the parameters and its implications for joint mechanics. Then a unique
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:31 Page Number: 3

14.1 Viscoelastic Materials 3

property of viscoelastic fluids under a simple shearing flow is their ability to


generate both a shear stress in the direction of flow, as well as a normal stress
perpendicular to flow. These normal stresses account for fluid climbing a spinning
rod, like cake mix on a beater shaft. For completeness we show the equivalent
approach for viscoelastic fluids using memory functions, which turns out to be an
integral of the Maxwell model differential equations, and then a viscoelastic solid
model called the Kelvin–Voigt model.

14.1 Viscoelastic Materials


14.2 Maxwell Fluid Response to Step Inputs
14.3 Maxwell Fluid Response to Oscillatory Forcing
14.4 Complex Fluid Viscosity
14.5 Stokes’ Second Problem for a Maxwell Fluid
14.6 Generalized Maxwell Fluid
14.7 More Viscoelastic Models
14.8 Rheological Measurements of Biofluids: The Cox–Merz Rule
14.9 Oscillatory Squeeze Film: Synovial Fluid
14.10. Oscillatory Channel Flow
14.11 Normal stresses
14.12 The Equivalent Approach for a Fading Memory Fluid
14.13 Kelvin–Voigt Viscoelastic Solid: Biological Tissues

...........................................................................................................................................................................................

14.1 Viscoelastic Materials


W ∆U1
Shear deformation of a material is shown in
Figure 14.1 where the displacement ΔU1 results from
the shear stress τ. The shear strain is
γ ¼ ΔU1 =Δx2 ¼ ∂U1 =∂x2 . Viewed as an elastic solid,
∆X2
the constituive equation is
∂U1
τ ¼ Gγ ¼ G (14.1.1)
∂x2

where G is the elastic shear modulus and τ is the ∆X1


shear stress.
To model viscoelastic materials, we are interested in Figure 14.1 Shear deformation of a material.
combining this elastic term into the constitutive
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:32 Page Number: 4

4 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

equation for a fluid, where the fluid velocity and its (a)
gradients are the working dependent variables. So,
taking the time derivative of Eq. (14.1.1) gives us F
∂τ ∂u1
¼ Gγ_ ¼ G (14.1.2)
∂t ∂x2

where u1 ¼ U_ 1 is the velocity. This relationship is in


contrast to Newtonian viscous fluid behavior in which
∂u1
τ¼μ ¼ μγ_ (14.1.3)
∂x2

where μ is the constant “shear viscosity” or “dynamic


viscosity.” Note that the Greek letter η is sometimes
used for viscosity.
A viscoelastic fluid can be described by a (b)
combination of the ideas in Eqs. (14.1.2) and (14.1.3).
The simplest example is a linear combination of the k
effects, known as a Maxwell fluid
µ F
μ ∂τ
τþ ¼ μγ_ (14.1.4)
G ∂t Ud Us
In Eq. (14.1.4) we see that the ∂τ=∂t term is multiplied U
by the shear stress-relaxation time, λ ¼ μ=G. In the
limit of large G, the relaxation time is short and the Figure 14.2 Maxwell fluid model for (a) shear deformation of a
material behaves as a viscous fluid with the familiar material by force F is (b) a spring and dashpot in series.
balance between τand the viscous shear rate on the
right hand side as in Eq. (14.1.3). However, in the limit
of large relaxation time, ðμ=GÞ∂τ=∂t dominates the constant fluid viscosity, μ, the same quantity in both
left hand side of Eq. (14.1.4). Then the viscosity, μ, settings. The force, F, is representative of the shear
cancels as the behavior is dominated by elastic forces stress, τ.
resulting in Eq. (14.1.2). The combination of terms The force of the dashpot, Fd, is linearly related to its
captures the behavior in between these two velocity as the piston moves through the viscous fluid
limiting cases. filling the cylinder. The force of the spring, Fs, is
The Maxwell model is based on viscous (dashpot) linearly related to its change in length or displacement
and elastic (spring) components in series as shown in Fd ¼ μU_ d , Fs ¼ kUs (14.1.5)
Figure 14.2. So let us use terminology from that
setting as it is usually taught in an introductory The distance U is the sum of Us and Ud,
physics course common to a wide spectrum of student U ¼ Ud þ Us (14.1.6)
backgrounds. Along the way we will point out the
equivalence for shear deformation. The spring length so that U represents the total shear strain, γ ¼ γd þ γs .
is Us and the dashpot position is Ud. For shear The time derivative of Eq. (14.1.6) is
deformation these are representative of shear strain γs U_ ¼ U_ d þ U_ s (14.1.7)
and γd , respectively. The spring constant is k, which
represents the shear modulus, G, and the dashpot has Using Eq. (14.1.5) in Eq. (14.1.7) we get
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:32 Page Number: 5

14.2 Maxwell Fluid Response to Step Inputs 5

Fd F_ s So the final result is


U_ ¼ þ (14.1.8)
μ k
F ¼ kU0 eμ=k HðtÞ
t
(14.2.7)
We know that in the absence of a mass, the forces are
So the step change in strain gives us a step change in
all equal,
stress by the spring instantly lengthening. This is
Fd ¼ Fs ¼ F (14.1.9) followed by a decay in stress as the spring shortens
and pulls the dashpot through the fluid. The ratio
so Eq. (14.1.8) becomes
μ=k ¼ λ is the relaxation time constant for the
F F_ Maxwell model. In takes longer for the spring to
U_ ¼ þ (14.1.10)
μ k return to its unstressed length if the fluid is more
viscous or if the spring constant is smaller, i.e. less
which is equivalent to Eq. (14.1.4) for the Maxwell
stiff. This phenomenon is called “stress relaxation,”
fluid shear model.
and is shown in Figure 14.3(a, b)
We may also consider that there is a step change in
F with a resultant displacement U. Let
14.2 Maxwell Fluid Response to
Step Inputs F ¼ F0 HðtÞ (14.2.8)

Inserting Eq. (14.2.8) into Eq. (14.1.10) yields


Suppose we impose a step change in U, such that
 
UðtÞ ¼ U0 HðtÞ HðtÞ H_ ðtÞ
(14.2.1) U_ ¼ F0 þ (14.2.9)
μ k
where U0 is a constant and HðtÞ is the Heaviside
function, a unit step at t ¼ 0, which can be integrated as

0t<0  
HðtÞ ¼ (14.2.2) t 1
1 t0 U ¼ F0 þ þc (14.2.10)
μ k
Since U is the constant value U0 for t > 0,
The constant of integration, c, is solved from the
U_ ðt > 0Þ ¼ 0 (14.2.3) initial condition
then its time derivative is zero. So the form of Eq.
Fð0Þ F0
(14.1.10) becomes Uð0Þ ¼ ¼ (14.2.11)
k k
μ
F þ F_ ¼ 0 (14.2.4) which we saw in Eq. (14.2.6). This condition renders
k
c ¼ 0, and the solution for UðtÞ is
which is satisfied by
 
1 t
F ¼ F 0 e
t
μ=k HðtÞ (14.2.5) UðtÞ ¼ F0 þ HðtÞ (14.2.12)
k μ
We solve for the integration constant, F0, by imposing
The strain continues as a linear function of time once
the initial condition that assumes the initial
the constant force is applied at t ¼ 0, see Figure 14.4.
displacement is instantaneously provided by only the
This is called “creep” for a solid. For a fluid it is simply
spring, there being insufficient time for flow to
flow. The slope of the linear time function is F0 =μ so
develop in the dashpot
the creep is slower when the viscosity is larger or the
Fð0Þ ¼ kUð0Þ ) F0 ¼ kU0 (14.2.6) force is smaller.
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:32 Page Number: 6

6 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

(a)
F (b)

kU0
t<0

t=0
0 t
U
t1

U0

0 t

Figure 14.3 Stress relaxation for step change at t ¼ 0. (a) Force response to step change in displacement. (b) Time sequence of dashpot
and spring for stress relaxation showing spring shortening to equilibrium length pulling the piston through the viscous dashpot fluid.

U(t) 14.3 Maxwell Fluid Response to


Oscillatory Forcing

Consider that the upper plate is oscillated by the shear


F0 stress, τ, as in Figure 14.5 so that
k
τ ¼ τ0 eiωt , γ ¼ γ0 eiωt (14.3.1)

0 t where the angular frequency is ω ¼ 2πf and f is the


oscillation frequency. The shear stress amplitude is τ0
while the shear strain amplitude is γ0 . Note that
F(t) eiωt ¼ cos ðωtÞ þ i sin ðωtÞ and both τ0 , γ0 can be
complex. So we seek the real part of the products
τ0 eiωt , γ0 eiωt in Eq. (14.3.1). Inserting these forms
into the Maxwell model of Eq. (14.1.4) we get
F0  
1 iω
þ τ0 ¼ iωγ0 (14.3.2)
μ G
0 t The limiting behavior of the system can be
Figure 14.4 Creep or flow for step change at t ¼ 0. appreciated in Eq. (14.3.2). For μ ! ∞ we have
Displacement response to step force input. τ0 ¼ Gγ0 , which is purely elastic. The resistance in the
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:33 Page Number: 7

14.3 Maxwell Fluid Response to Oscillatory Forcing 7

τ
We can define a dimensionless frequency using the
Deborah number,

γ De ¼ ωλ ¼
μω
G
μ=G relaxation time
¼ e
1=2πf observation time ði:e:oscillation periodÞ
(14.3.5)

where a system is dominated by viscosity for De << 1


and by elasticity for De >> 1. Rewriting Eqs. (14.3.4)
h gives us
00
G0 De2 G De
¼ , ¼ (14.3.6)
G 1 þ De2 G 1 þ De2

00
Note that G0 =G ¼ De ¼ ωλ or, equivalently,
00
Figure 14.5 Oscillatory shear forcing of a viscoelastic fluid. λ ¼ G0 =ωG .
Plots of these two components as functions of De
are shown in Figure 14.6(a, b) in linear and log–log
dashpot is very high so the dashpot strain, γd eUd ,
formats. G0 =G starts at zero and asymptotes to 1.
goes to zero. In that limit, all of the deformation is
Since this is the real part of the modulus, it relates the
relegated to the spring strain, γs eUs , and the system
displacement component to the in-phase force
reaches an elastic limit. As ω ! ∞ the balance is also
component and tells us about the elastic nature of the
the elastic limit τ0 ¼ Gγ0 . Since the dashpot strain is
system under oscillation. As De increases, the strain
γd e1=μω it cannot keep up with the faster forcing. For
rate amplitude of the dashpot decreases and the
G ! ∞ the stress becomes τ0 ¼ iωμγ0 , which is purely 00
system approaches an elastic limit. G =G also starts
a viscous fluid. The spring is rigid so its strain decreases
from zero but passes through a local maximum when
to zero and all of the strain is taken up by the dashpot.
De ¼ 1 where viscous and elastic effects are equal. At
Equation (14.3.2) can be rearranged to represent the 00
that value, both components are equal, G0 ¼ G ,
stress amplitude τ0 as a linear function of the strain 00
called the “crossover point.” Then G =G asymptotes to
amplitude, γ0 ,
! zero when De increases without bound as the system
∗ ∗ ðωμÞ2 G þ iωμG2 is dominated by its elasticity. For De < 1 we see that
τ0 ¼ G γ0 , G ¼ (14.3.3) 00
G2 þ ðωμÞ2 G > G0 and the system behaves more like a viscous
00
fluid. On the other hand, when De > 1, G < G0 and
where the real and imaginary parts of the complex the system behaves more like an elastic solid.
00
shear modulus G∗ ¼ G0 þ iG are readily found to be 00
The local maximum of G =G is derived by setting to
ðωμÞ2 G 00 ωμG2 zero its derivative with respect to De and solving for
G0 ¼ , G ¼ (14.3.4) the value of De ¼ Dem at the maximum
G2 þ ðωμÞ2 G2 þ ðωμÞ2
!
In Eq. (14.3.4), G0 is the shear storage modulus which 00  1 þ De2m  2De2m
d G 
 ¼  2 ¼ 0 ! Dem ¼ 1
is in-phase with the oscillation since De G  1 þ De2m
00 De¼Dem
ei0 ¼ cos ð0Þ þ i sin ð0Þ ¼ and G is the loss modulus
(14.3.7)
which is π=2 ¼ 90 out of phase with the oscillation
π
since ei 2 ¼ cos ðπ=2Þ þ i sin ðπ=2Þ ¼ i.
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:33 Page Number: 8

8 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

(a) (b)

1 100

10–1
0.8

10–2
0.6
10–3
0.4
10–4

0.2 10–5

0 10–6
0 2 4 6 8 10 0.1 1 10 100
De De

00
Figure 14.6 G0 =G and G =G vs De in (a) linear and (b) log–log plots.

The local maximum occurs at Dem ¼ 1. The crossover ω ¼ ωc identifies the relaxation time. There we know
00
value of De, call it Dec, is found by setting that G0 ðωc Þ ¼ G ðωc Þ and it occurs when
00
G0 =G ¼ G =G and solving for De ¼ Dec Dec ¼ ωc λ ¼ 1. Then the relaxation time is simply
00
λ ¼ 1=ωc ¼ λ ¼ 1=2πf c where fc is the crossover
G0 G De2c Dec
¼ ! ¼ ! Dec ¼ 1 (14.3.8) frequency. A second dimensionless group defined for
G G 1 þ De2c 1 þ De2c oscillatory viscoelastic flows is the Weissenberg
So Dem ¼ Dec ¼ 1 where the values of the scaled number, Wi ¼ ωγ0 λ, which differs from De by
00
moduli are G0 =G ¼ G =G ¼ 1=2. including the strain amplitude γ0 . Karl Weissenberg
When interpreting data of a Maxwell fluid where G0 (1893–1976) was an Austrian physicist who made
00
and G are plotted vs ω, the crossover frequency at many contributions to rheology and crystallography.

Box 14.1 The Deborah Number

The Deborah number was a name created by Markus Reiner from The Technion (Israel) who
also worked in the US with Eugene C. Bingham at Lafayette College (Easton, PA) in the early
days of rheology (Reiner, 1964). The reference is to the prophetess Deborah who was a leader
(Judge) of the Jewish people in the twelfth century BCE, see Figure 14.7. Her song after victory
in battle over the Philistines included the phrase in the Bible (Judges 5:5): “The mountains

6
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:34 Page Number: 9

14.3 Maxwell Fluid Response to Oscillatory Forcing 9

Figure 14.7 The prophetess Deborah singing in triumph over Jabin, by Gustave Doré (1865).

flowed before the Lord.” Reiner sees this as an opportunity to connect “rheology” to
“theology,” humored by his incoming surface mail often addressed to a theology department.
The idea is that only the Lord can view a mountain over a long enough time period to observe
its flow. So De is the relaxation time divided by the observation time.

00
Representing G* in complex polar form gives us shows us that the crossover point, where G0 ¼ G ,
occurs where tan δ ¼ 1 or δ ¼ 45 . Dividing jG∗ j by
G∗ ¼ jG∗ jeiδ ¼ jG∗ jðcos δ þ i sin δÞ (14.3.9)
G in Eq. (14.3.11) and using the forms of Eq. (14.3.6),
so that the dependence on De is found to be

G0 ¼ jG∗ j cos δ, G ¼ jG∗ j sin δ


00
(14.3.10) jG∗ j De 1
¼ 1=2
, tan δ ¼ (14.3.12)
00
k ð1 þ De Þ
2 De
0
The magnitude and angle are related to G and G as

1=2 00 The two properties of Eq. (14.3.12) are plotted in
∗ 02 00 2 G
jG j ¼ G þ G , tan δ ¼ 0 (14.3.11) Figure 14.8. As De increases from zero, the Maxwell
G
fluid behaves increasingly like an elastic solid, with
where jG∗ jis the dynamic modulus of elasticity and internal damping decreasing and the dynamic elastic
tan δ is the internal damping. Equation (14.3.11) modulus increases to equal G.
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:34 Page Number: 10

10 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

(a) (b)
1 100

0.8
10

0.6
1
0.4

0.1
0.2

0 0.01
0 2 4 6 8 10 0.1 1 10 100
De De

Figure 14.8 Dynamic elastic modulus scaled on G, jG∗ =Gj, and internal damping tan δ vs De as (a) linear and (b) log–log plots.

14.4 Complex Fluid Viscosity where we have included the relationships between the
components of μ∗ and G∗ ; μ0 is called the dynamic
00
An alternative view is to consider the material as a viscosity and μ is the out-of-phase viscosity. Note
00 00
fluid model, where the shear stress is linearly that μ =μ0 ¼ De ¼ ωλor, equivalently, λ ¼ μ =ωμ0 .
proportional to the strain rate. With this approach we The magnitude of Eq. (14.4.2) is
can rearrange Eq. (14.3.2) to keep iωγ0 ¼ γ_ as a jτ0 j ¼ jμ∗ jωγ0 (14.4.4)
group, and solve for τ0 ,
  where the magnitude of μ∗ is given by
1 iω 1
1=2
τ0 ¼ iωγ0 þ (14.4.1) jμ∗ j ¼ μ0 2 þ μ
00 2
μ G .
In dimensionless form the two components of the
Rearranging Eq. (14.4.1), and using the definition of
complex viscosity, μ0 =μ andμ00 =μ, and the magnitude
De ¼ μω=G yields
jμ∗ j=μ are plotted vs De in Figure 14.9 in linear and
ð1  iDeÞ log–log formats
τ0 ¼ μ ðiωγ0 Þ ¼ μ∗ ðiωγ0 Þ (14.4.2)
ð1 þ De2 Þ 00
μ0 1 μ De jμ∗ j 1
¼ 2 , ¼ 2, ¼
where the complex viscosity, μ∗ , is defined as μ 1 þ De μ 1 þ De μ ð1 þ De2 Þ
1=2
 00 
μ∗ ¼ μ0  iμ with real and imaginary components. (14.4.5)
According to Eq. (14.4.2) those components are
00
For increasing De the viscous behavior μ0 =μ
1 G 00 De G0 diminishes. As we have learned, whenever the
μ0 ¼ μ 2 ¼ , μ ¼μ ¼
ð1 þ De Þ ω ð1 þ De2 Þ ω frequency is high, i.e. the oscillation period is small
(14.4.3) compared to the relaxation time and the system is not
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:35 Page Number: 11

14.5 Stokes’ Second Problem for a Maxwell Fluid 11

(a) (b)

1 100

0.8
10-1

0.6
10-2
0.4

10-3
0.2

0 10-4
0 2 4 6 8 10 0.1 1 10 100
De De
m /m m /m |m|/m m /m m /m |m|/m

Figure 14.9 The real and imaginary components of the complex viscosity, μ0 =μ and μ00 =μ, and the magnitude jμ∗ =μj vs De displayed in
(a) linear and (b) log–log plots.

allowed to “relax” as the flow direction changes too and only one component of velocity,
quickly. For De ! 0 we see that μ0 =μ ! 1, which is u1 ðx2 ; tÞ ¼ uðy; tÞ. With the stress tensor
purely viscous behavior as elastic effects disappear. Tij ¼ pδij þ τij , the i ¼ 1, j ¼ 2 component is all that
00
There is a local maximum at De ¼ 1 for μ =μ and remains in the Cauchy momentum equation, Eq.
00
μ0 =μ ¼ μ =μ ¼ 1=2 there. When evaluating data (5.2.2). That component is
through the Maxwell model viscosities, the frequency
∂u ∂τ12
at the maximum, call it ωm , gives us the relaxation ρ ¼ (14.5.1)
∂t ∂y
time λ ¼ 1=ωm . We see that ωm ¼ ωc , the crossover
00
frequency for G0 and G in Figure 14.6. The magnitude We will see in Section 14.7 that the Maxwell model
of the complex viscosity, jμ∗ j=μ, decreases extended into continuum mechanics suffers from
monotonically with De, indicating that the Maxwell important deficiencies because, in general, it is not
fluid is shear thinning. invariant to coordinate transformations. However, for
this simple flow those issues disappear. The only
component is
14.5 Stokes’ Second Problem for a ∂τ12 ∂u
Maxwell Fluid τ12 þ λ ¼μ (14.5.2)
∂t ∂y

In Section 8.2 we studied Stokes’ second problem for a Again, as G ! ∞, λ ! 0 and the balance in Eq.
wall oscillating in its own plane with a Newtonian (14.5.2) is a Newtonian viscous fluid τ12 ¼ μ∂u=∂y,
viscous fluid above it. Let’s repeat this analysis for a where we are reminded that this “u” is a velocity. For
viscoelastic fluid. Recall there is no pressure gradient μ ! ∞ the time constant, λ ¼ μ=G ! ∞, the balance
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:35 Page Number: 12

12 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

approaches ∂τ12 =∂t e G∂u=∂y, or when integrated where β2 ¼ α2 ðDe þ iÞ. Solutions of Eq. (14.5.10)
τ12 e G∂U=∂y ¼ Gγ, which is the elastic shear limit. are of the form
The kinematic boundary conditions are the wall ^ ðYÞ ¼ c1 eβ Y þ c2 eβ Y
U (14.5.11)
velocity at y ¼ 0, and zero flow far from the wall
where the complex number β ¼ βr þ i βi has the real
uðy ¼ 0Þ ¼ u0 cos ωt, uðy ! ∞Þ ¼ 0 (14.5.3)
and imaginary components
where we can consider the velocity amplitude to be pffiffiffi
2α  1=2
1=2
linearly related to the frequency u0 ¼ aω. βr ¼ De þ 1 þ De2 >0
2
For scales, let pffiffiffi (14.5.12)
2α  
1=2
2 1=2
u y τ12 βi ¼  De þ 1 þ De <0
U¼ , T ¼ ωt, Y ¼ , σ ¼ (14.5.4) 2
ωa a G
The boundary conditions of Eq. (14.5.7) become
Insert Eqs. (14.5.4) into Eqs. (14.5.1) and (14.5.2) to
find the Cauchy momentum equation U ^ ðY ¼ 0Þ ¼ 1
^ ðY ! ∞Þ ¼ 0 ) c1 ¼ 0 U ) c2 ¼ 1
∂U ∂σ (14.5.13)
α2 De ¼ (14.5.5) ^ ðYÞ is
∂T ∂Y The solution for U

where α2 ¼ ωa2 =νand α is the Womersley parameter ^ ðYÞ ¼ eβ Y


U (14.5.14)
while De is the Deborah number, De ¼ ωλ. The
^ ðYÞ into Eq. (14.5.9) for
Inserting the solution for U
Maxwell constitutive equation becomes
^ ðYÞ yields
σ
∂σ ∂U
σ þ De ¼ De (14.5.6) βDeð1 þ iDeÞ βY
∂T ∂Y ^ ¼
σ e (14.5.15)
ð1 þ De2 Þ
Now insert Eqs. (14.5.4) into the boundary conditions,
Eqs. (14.5.3), to find and the total solution is
   
UðY ¼ 0Þ ¼ cos T ¼ ℜ eiT UðY; TÞ ¼ ℜ eβ Y eiT
UðY ! ∞Þ ¼ 0
(14.5.7)  
βDeð1 þ iDeÞ βY iT
σðY; TÞ ¼ ℜ  e e
In Eq. (14.5.7), ℜ indicates the real part. Try a ð1 þ De2 Þ
separation of variables approach using the sinusoidal (14.5.16)
time dependency from the wall, We can simplify the velocity solution by combining
 
UðY; TÞ ¼ ℜ U ^ ðYÞ eiT the exponential terms
  (14.5.8)  
σðY; TÞ ¼ ℜ σ ^ ðYÞ eiT UðY; TÞ ¼ ℜ eβ Y eiT ¼ eβr Y cos ðT  ðβi YÞÞ
Inserting Eqs. (14.5.8) into Eq. (14.5.5) and Eq. (14.5.6) (14.5.17)
we arrive at coupled ODEs for σ ^
^ , and U, Since the wall velocity is cos T, the phase lag for the
σ
d^ ^
dU fluid velocity in Eq. (14.5.17) is βi Y > 0, similar to
^ ¼
iα2 DeU ^ ð1  iDeÞ ¼ De
σ (14.5.9) what we found for the Newtonian system in Section
dY dY
8.2, Eq. (8.2.17), except now the effects of
Substituting the derivative of the second equation of viscoelasticity influence the motion.
Eqs. (14.5.9) into the first leads to a second-order ODE Results of the model are shown in Figure 14.10 for
^
for U, three values of De ¼ 0:1, 1, 10 . As De increases the
^
d2 U fluid becomes more elastic and the amplitude of the
^ ¼0
 β2 U (14.5.10)
dY2 velocity increases, while the spatial frequency in the
Y-direction increases and the rate of decay decreases.
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:35 Page Number: 13

14.6 Generalized Maxwell Fluid 13

α = 1.0, T = π/2
10

8
k3
6 m3 F3
Y
4
k2
m2 F2
2

–0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 k1


U m1 F1
De = 0.1 De = 1 De = 10

Figure 14.10 Stokes’ second problem for a Maxwell fluid for


De ¼ 0:1, 1, 10 .
U
14.6 Generalized Maxwell Fluid
Figure 14.11 Generalized Maxwell model with multiple
Maxwell elements.
Biofluids are often polydisperse, that is, the long chain
molecules which make up the viscoelastic fluid have
where k(t) is called the stress-relaxation modulus and
more than one type of molecule and more than one
ki ¼ Fi =U0 . For a continuous distribution of
length. Under these circumstances, the stress
relaxation times the summation in Eq. (14.6.3) can be
relaxation, which comes from the untangling of these
replaced by the integral
molecules, is going to have more than one relaxation
time. The stress relaxation of this generalized system ð∞
kðtÞ ¼ kðλ Þeλ dλ
t

is in response to a step change in U which is (14.6.4)


experienced by all of the spring–dashpot components, 0

see Figure 14.11. U ¼ U0 HðtÞ as in Eq. (14.2.1) and the Let us consider a two-element generalized Maxwell
individual forces are fluid. The constitutive equations are given by
Fi ¼ ki U0 e
λt
(14.6.1) μ
i
_
τ1 þ λ 1 τ_ 1 ¼ μ1 γ, λ1 ¼ 1
G1 (14.6.5)
where λ i ¼ μi =ki is the time constant for the Maxwell μ
_ λ2 ¼ 2
τ2 þ λ 2 τ_ 2 ¼ μ2 γ,
element. The sum of the forces is G2

X
n X
n
λt
and we know that the total shear stress is the sum of
F¼ Fi ¼ U0 ki e i (14.6.2) the two Maxwell elements
i¼1 i¼1
τ ¼ τ1 þ τ2 (14.6.6)
or rearranged as
Xn We can resolve Eqs. (14.6.5) into a single equation,
F t
kðtÞ ¼ ¼ ki e λi (14.6.3) which is readily done by using the Laplace transform,
U0 i¼1 F(s), defined as
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:36 Page Number: 14

14 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

ð∞ and we assume τð0Þ, τ_ ð0Þ, γð0Þ, γ_ ð0Þ are all zero. The
FðsÞ ¼ f ðtÞest dt (14.6.7) last of Eqs. (14.6.9) can be rearrange to yield
0
1 þ ðλ 1 þ λ 2 Þs þ λ 1 λ 2 s2 Τ

The transform of a function f(t) is F(s), while ¼ ðμ1 þ μ2 Þs þ ðμ1 λ 2 þ μ2 λ 1 Þs2 Γ (14.6.11)
transforms of its time derivatives are
Then the inverse Laplace transform of Eq. (14.6.11)
ð∞ gives us a single, second-order differential equation
df
ðtÞest dt ¼ sFðsÞ  f ð0Þ for this generalized Maxwell fluid
dt
_ €ðμ1 λ 2 þμ2 λ 1 Þ€
0
(14.6.8) τþ ðλ 1 þλ 2 Þτ_ þλ 1 λ 2 τ ¼ ðμ1 þμ2 Þγþ γ
ð∞ 2
d f (14.6.12)
ðtÞest dt ¼ s2 FðsÞ  sf ð0Þ  f_ ð0Þ
dt2
0 Equation (14.6.12) has the steady solution
τ ¼ ðμ1 þ μ2 Þγ_ so that the effective viscosity is the
The Laplace transforms of Eqs. (14.6.5) and (14.6.6) sum of the two viscous elements
are
μeff ¼ τ=γ_ ¼ μ1 þ μ2 (14.6.13)
ð1 þ λ 1 sÞΤ1 ¼ μ1 sΓ
The general form of Eq. (14.6.12) is
ð1 þ λ 2 sÞΤ2 ¼ μ2 sΓ
(14.6.9) τ þ p1 τ_ þ p2 τ ¼ q1 γ€_ þ q2 γ€ (14.6.14)
μ1 sΓ μ2 sΓ
Τ ¼ Τ1 þ Τ2 ¼ þ
ð1 þ λ 1 sÞ ð1 þ λ 2 sÞ For n Maxwell elements a similar approach yields the
form of the differential equation as
where
X
n
dj Xn
dj
ð∞ pj j
τ¼ qj j γ (14.6.15)
dt dt
ΤðsÞ ¼ τðtÞest dt j¼0 j¼0

0
(14.6.10) where q0 ¼ 0 while pj and qj depend on the
ð∞ viscoelastic parameters of the elements. Equation
ΓðsÞ ¼ γðtÞest dt (14.6.15) forms the basis of the generalized
0 Maxwell model.

Worked Example 14.1 Viscoelastic Behavior of a Two-Element Generalized


Maxwell Fluid

To find the relationship between shear stress and shear strain amplitudes for oscillatory
forcing, insert Eqs. (14.3.1) into Eq. (14.6.12), to give

1 þ ðλ 1 þ λ 2 Þiω  λ 1 λ 2 ω2 τ0 ¼ ðμ1 þ μ2 Þiω  ðμ1 λ 2 þ μ2 λ 1 Þω2 γ0 (14.6.16)

Rearranging Eq. (14.6.16) leaves us with the familiar form of a complex shear modulus

6
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:36 Page Number: 15

14.6 Generalized Maxwell Fluid 15

10

0.1

0.01

0.001

0.01 0.1 1 10 100


De

00 00
Figure 14.12 G0 =G1 and G =G1 vs De for a two-element Maxwell fluid model with β ¼ 1 and ϕ ¼ 0:1, 1, 10. G =G1 curves
exhibit a local maximum.

00

τ0 ¼ G∗ γ0 ¼ G0 þ iG γ0 (14.6.17)
00 00
where G∗ ¼ G0 þ iG , and G0 is the shear storage modulus while G is the loss modulus given
by
½ðμ1 λ 2 þ μ2 λ 1 Þð1  λ 1 λ 2 ω2 Þω2 þ ðμ1 þ μ2 Þðλ 1 þ λ 2 Þω2 
G0 ¼
ð1  λ 1 λ 2 ω2 Þ2 þ ðλ 1 þ λ 2 Þ2 ω2
(14.6.18)
ðμ þ μ2 Þð1  λ 1 λ 2 ω2 Þω þ ðμ1 λ 2 þ μ2 λ 1 Þðλ 1 þ λ 2 Þω3
00
G ¼ 1
ð1  λ 1 λ 2 ω2 Þ2 þ ðλ 1 þ λ 2 Þ2 ω2

If we define the following dimensionless parameters for the ratios of the viscosities,
β ¼ μ2 =μ1 , and time constants, ϕ ¼ λ 2 =λ 1 , and let the Deborah number be De ¼ ωλ1 then
Eqs. (14.6.18) can be simplified to the dimensionless forms
G0 ð1 þ βϕ þ ϕðϕ þ βÞDe2 ÞDe2
¼  
G1 ð1 þ De2 Þ 1 þ ϕ2 De2
    (14.6.19)
00
G 1 þ β þ β þ ϕ2 De2 De
¼  
G1 ð1 þ De2 Þ 1 þ ϕ2 De2

6
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:37 Page Number: 16

16 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

00
Sample plots of Eqs (14.6.19) are shown in Figure 14.12. G0 =G1 and G =G1 are plotted vs De
for β ¼ 1 and ϕ ¼ 0:1, 1, 10. In all cases G0 =G1 asymptotes to a constant as De >> 1 while
00
G =G1 has a local maximum. The behavior of G0 =G1 can be seen in Eq. (14.6.19) where
 0
G ðϕðϕ þ βÞDe4 Þ ϕðϕ þ βÞ β
e  2 4 ¼ ¼1þ (14.6.20)
G1 De>>1 ϕ De ϕ 2
ϕ

which explains why the ϕ ¼ 10 curve is shifted downward while the ϕ ¼ 0:1 curve is shifted
00
upward compared to ϕ ¼ 1, for De >> 1. Note that the De value for the maximum of G =G1 ,
call it Dem, shifts considerably to the right for ϕ ¼ 0:1 as does the crossover point.

14.7 More Viscoelastic Models We should be cautious about Eq. (14.7.4) because it
is not invariant to rigid body coordinate
The Cauchy momentum equation is the momentum transformations. The difficulty rests with the time
conservation we studied in Chapter 5, Eq. (5.2.2), derivative. Recall that we examined the Galilean
  invariance of the Navier–Stokes equations in Section
∂ui ∂ui ∂Tij
ρ þ uk ¼ þ bi (14.7.1) 11.1, Eqs. (11.1.18)–(11.1.21). The prime coordinate
∂t ∂xk ∂xj
system travels in the +x-direction at speed
where the stress tensor is the sum of the isotropic U compared to the unprimed system, Figure 11.4, and
pressure contribution multiplying the Kronecker delta, the relationships for the variables in the two systems
or identity tensor,pδij , and the extra, or viscous, are x0 ¼ x  Ut, y ¼ y0 , u0 ¼ u  U, v0 ¼ v. From
stress tensor, τij , is the chain rule we found that the time derivatives in
the two systems are related by ∂=∂t ¼ ∂=∂t0  U∂=∂x0 .
Tij ¼ pδij þ τij (14.7.2)
Applied to the acceleration term, which is the material
 
Substituting Eq. (14.7.2) into Eq. (14.7.1) leads us to derivative of the velocity, Du=Dt ¼ ∂u=∂t þ u u, we
  showed that it is invariant, here for the x-component
∂ui ∂ui ∂p ∂τij
ρ þ uk ¼ þ (14.7.3)  0   
∂t ∂xk ∂xi ∂xj ∂u ∂u ∂u ∂u ∂u0 ∂u0 0 ∂u
0
þu þv ¼ U 0 þ U 0 þu 0
∂t ∂x ∂y ∂t0 ∂x ∂x ∂x
The constitutive equation for the Maxwell fluid is the ∂u 0
þ v0 0
equivalent of Eq. (14.1.10) for each component of the ∂y
stress tensor ∂u0 ∂u0 ∂u0
¼ 0 þ u0 0 þ v 0 0
  ∂t ∂x ∂y
∂τij ∂ui ∂uj
τij þ λ ¼ 2μEij ¼ μ þ (14.7.4) (14.7.5)
∂t ∂xj ∂xi
The key point is that the convective acceleration term
where λ ¼ μ=G is the time constant. Equations
produces U∂u0 =∂x0 , which cancels the U∂u0 =∂x0 term
(14.7.3) and (14.7.4) are coupled PDEs. Recall that for
arising from the time derivative. The full Maxwell
a Newtonian fluid, G ! ∞, λ ! 0 and we can
  model of Eq. (14.7.4) does not have a convective
substitute τij ¼ μ ∂ui =∂xj þ ∂uj =∂xi into Eq. (14.7.3)
acceleration term to cancel the time derivative term.
to obtain the Navier–Stokes equation. For λ > 0we
So Eq. (14.7.4) is not Galilean invariant. To deal with
have the coupled system, however.
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:37 Page Number: 17

14.8 Rheological Measurements of Biofluids: The Cox–Merz Rule 17

this difficulty, a number of models for viscoelastic for a wide range of Ω. For viscoelastic fluids, similar
fluids have been developed which include convective devices are driven in an oscillatory mode at frequency
terms to accompany the time derivative on the left ω ¼ 2πf rad=s, so that the fluid is forced to cyclically
hand side of Eq. (14.7.4) so that the new model is reverse direction while the torque is measured. A wide
invariant. In fact, often they use the range of ω is also explored, such as several decades
material derivative. like 0:01  ω  100 rad=s. The force (torque) data
One example is the “upper convected Maxwell are separated into the component in-phase with the
model” or UCM, oscillation, G0 the shear storage modulus, and the
rij
component 90 degrees out-of-phase with the
τij þ λ τ ¼ 2μEij (14.7.6) 00
oscillation, G the loss modulus. Both of these are
00
related to the complex viscosity μ∗ ¼ μ0  iμ shown
where the “upper convected time derivative” is
in Eq.(14.4.3). In this mode, these devices are
defined by
measuring more than viscosity, so are called
rij ∂τij ∂τij ∂ui ∂uj “rheometers.” Additional technologies for measuring
τ ¼ þ uk  τkj  τik (14.7.7)
∂t ∂xk ∂xk ∂xk rheology of materials utilize smaller scale approaches
Note that the first two terms on the right hand side of where magnetic beads with diameters less than ~
Eq. (14.7.7) form the material derivative of τij , i.e. 5 microns are placed into the material, i.e. on or inside
Dτij =Dt ¼ ∂τij =∂t þ uk ∂τij =∂xk , which we know is a cell (Bausch et al., 1998). Then an oscillatory
Galilean invariant, which guarantees the same for this magnetic fields is applied and the measured forces
00

model. Another popular choice is the Oldroyd and motion are decoupled to yield G0 and G .
B model, which is an extension of the UCM In the steady flow viscometer the effective viscosity
 was measured as μeff ðγ_ Þ ¼ τ=γ.
_ Can this viscosity be
rij rij
τij þ λ 1 τ ¼ 2μ Eij þ λ 2 E Þ (14.7.8) related to the complex viscosity, μ∗ ðωÞ, of an
oscillatory rheometer? The short answer is yes, but not
where λ 1 is the relaxation time and λ 2 is the retardation always. In a fundamental 1958 paper by Cox and
time. The material derivatives of both τij and Eij are now Merz, the authors proposed that the relationship is
involved. Still more models start with the generalized μeff ðγ_ Þ ¼ jμ∗ ðωÞjγ¼ω
_ (14.8.1)
Maxwell approach, but replace the linear dashpots with
Equation (14.8.1) says that μeff ðγ_ Þis equal to the
nonlinear fluids such as power-law fluids, and linear
magnitude of the complex viscosity jμ∗ ðωÞj for the
springs with nonlinear springs (Monsia, 2011).
_ –1) equal to the oscillation frequency
strain rate γ(s
ω ¼ 2πf (rad/s), i.e. γ_ ¼ ω. This equivalence in Eq.
(14.8.1) is known as the Cox–Merz rule (Cox and
14.8 Rheological Measurements of
Merz, 1958) and holds well for numerous polymeric
Biofluids: The Cox–Merz Rule
systems. Since its publication, many viscoelastic
materials have been tested in steady and oscillatory
Measuring Viscoelastic Behavior of Fluids
viscometers to see if the left hand side really equals the
In Section 10.8 we discussed measurement of right hand side. It does not always hold, especially as the
viscosity for generalized Newtonian fluids with material is increasingly complicated in microstructure
examples of a concentric-cylinder viscometer like foods (Bistany and Kokini, 1983; Vernon-Carter
(Figure 10.22), and a cone-plate viscometer et al., 2016). For example, salad dressing emulsions do
(Figure 10.23). For that presentation, the viscometers not follow the Cox–Merz rule (Riscardo et al., 2005). The
were applying steady, unidirectional flow at equivalence is often restricted to low frequency
rotational frequency Ω. The fluid response is explored oscillations on the right hand side. However, for blood
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:38 Page Number: 18

18 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

(Beissinger and Williams, 1985), mucus and synovial 100


fluid it is fairly accurate.
We computed the right hand side of Eq. (14.8.1) for
the Maxwell model already in Eq. (14.4.5)
10
jμ∗ ðωÞj 1
¼ (14.8.2)
μ ð1 þ De2 Þ
1=2

Of course the left hand side of Eq. (14.8.1) is simply 1


μeff ¼ μ for a Maxwell fluid, i.e. from Eq. (14.1.3)
μeff ¼ τ=γ_ ¼ μ. So according to Eq. (14.8.2), the Cox–
Merz rule is only accurate for the Maxwell unit when 0.1
De2 <<1, i.e. low frequency. A Taylor series of
jμ∗ ðωÞj=μ in Eq. (14.8.2) in this limit is
jμ∗ ðωÞj  1=2 0.01
¼ 1 þ De2
μ   0.1 1 10 100
1   De
¼ 1  De2 þ O De4 De <<1
2
2 |m*|/m 1/De
(14.8.3)
Figure 14.13 Maxwell fluid jμ∗ j=μ (solid line) and 1=De vs
where we see that jμ∗ ðωÞj=μe1 for De2 <<1. The De (boxes).
other asymptotic limit of Eq. (14.8.2) is for De>>1,
which yields
From Eq. (14.8.2) jμ∗ j=μ vs De is shown in

jμ ðωÞj 1 Figure 14.13 in a log–log plot along with 1=De, which
e De2 >>1 (14.8.4)
μ De is the De>>1 solution.

Worked Example 14.2


00
For the two-element Maxwell model we found G0 =G1 and G =G1 in Eq. (14.6.19). Then to
calculate jμ∗ j yields
h  00 2 i1=2
ðG0 Þ þ G
2

jμ∗ j ¼ (14.8.5)
ω
and scaling it on μ1 leads to
h  00 2 i1=2
ðG0 =G1 Þ þ G =G1
2
jμ∗ j
¼ (14.8.6)
μ1 De

6
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:38 Page Number: 19

14.8 Rheological Measurements of Biofluids: The Cox–Merz Rule 19

10

|m*|/m1 0.5

0.1

0.05

0.01 0.1 1 10 100


De
b = 10.0, j = 1 b = 10.0, j = 10
b = 10.0, j = 0.1

Figure 14.14 jμ∗ j=μ1 vs De for β ¼ 10, ϕ ¼ 0:1, 1, 10.

A plot of jμ∗ j=μ1 from Eq. (14.8.6) is shown in Figure 14.14 for β ¼ 10 and ϕ ¼ 0:1, 1:0, 10:0,
the same parameter set used in Figure 14.12. A Taylor series expansion of jμ∗ j=μ1 for
De<<1 is
 
jμ∗ j=μ1 ¼ ð1 þ βÞ þ O De2 De<<1 (14.8.7)

The asymptote for De ! 0 is jμ∗ j=μ1 ¼ 1 þ β, which only involves the viscosities of the
Maxwell elements. Dimensionally this limit is jμ∗ j ¼ μ1 ð1 þ βÞ ¼ μ1 þ μ2 , the sum of the two
viscosities, which recovers the steady solution in Eq. (14.6.13) as we expect when the
frequency, ω, approaches zero.

00
Blood plot. G is larger than G0 indicating what we all know
from experience – that blood is more of a viscous
Figure 14.15 shows measurements of blood rheology
liquid than an elastic solid. The elasticity derives from
using both oscillatory and steady shear rates
interaction of the red blood cells, so is dependent on
(Tomaiuolo, 2014; D’Apolito et al., 2015; Tomaiuolo
hematocrit among other parameters. Figure 14.15(b)
et al., 2016). In Figure 14.15(a) the storage modulus,
00
shows the steady viscosity ηðγ_ Þ and the magnitude of
G0 , and loss modulus, G , are plotted versus the the complex viscosity jη∗ ðωÞj plotted on the same
angular frequency, ω (in rad/s). As ω increases both axis where γ_ ¼ ω, demonstrating that the Cox–Merz
moduli increase as a power law ~ω0:7 with a slope of rule holds well for this shear-thinning fluid.
~0.7 for the straight part of the data in this log–log
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:38 Page Number: 20

20 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

00
Figure 14.15 Blood rheology for Hct = 45%. (a) Storage modulus, G0 , and loss modulus, G , as functions of angular frequency, ω (rad/
s). The gray dashed straight lines represent a power-law fitting of the experimental data (R2 = 0.999) with slope ~ 0.7. (b) Comparison
of steady shear viscosity η and magnitude of complex viscosity jη∗ j illustrating the Cox–Merz rule. Here η is our μ.

Rheological literature often uses η as viscosity, as auditory and visual systems. One of its major
opposed to our μ. They are the same here. constituents is mucin, which is produced in
Using a Maxwell model, the stress-relaxation time, specialized cells called goblet cells that can secrete
λ, was measured to be ~0.03 s over the frequency onto the epithelial surface. Mucin is a large and bushy
range 1–3 Hz for normal hematocrits (Long et al., glycoprotein molecule as shown in Figure 14.16.
2005). Adult and pediatric blood had similar behavior. Typically it forms long chain oligomers of repeated
Over a wider range of frequencies, however, λ did not mucin molecules which can aggregate.
remain constant indicating a more complicated In Figure 14.17 shows a compilation of steady
behavior than a single Maxwell unit. For human mucus viscosity measurements. The slopes in the log–
physiology fetal heart rates are typically 150 bpm log plots of both are in the range –1 to –0.5 with an
(2.5 Hz) while the adult range is 60–100 bpm average of –0.86 for the steady shearing (Yeates et al.,
00
(1–1.66 Hz). These values place the range of the 1997). Oscillatory values of μ0 eG =ω are also
Deborah number, De ¼ 2πf λ, as approximately presented in (Lai et al., 2009) but not jη∗ j which we
0:2  De  0:5. However, higher frequencies can would need to test the Cox–Merz rule. The horizontal
occur in blood processing equipment, for example, line is water viscosity, 10–3 Pas, so the range of
extending the upper range of De. Within the animal mucus viscosities range from tears near water to
kingdom, mouse heart rates are in the range of 10,000–100,000 times water for chronic bronchitis
500 bpm (~8 Hz), so larger De values are possible. which, in addition to mucus, has inflammatory cells
and debris added to the mix.
Figure 14.18 shows rheological data from
Mucus
oscillatory forcing of mucus (Hill et al., 2014). In
00
Mucus is a lubricating and protective secretion which Figure 14.18(c, d) G0 and G are plotted on log–log
coats mucous membranes. In humans that includes axes vs oscillation frequency, f, (Hz) for five different
the epithelial cells of the gastrointestinal, mucus solids concentrations: 1.5, 2, 3, 4, 5 wt%. For
genitourinary and respiratory tracts as well as the Figure 14.18(c) with 1.5, 2.0 and 3.0 wt%, we see that
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:39 Page Number: 21

14.8 Rheological Measurements of Biofluids: The Cox–Merz Rule 21

complex viscosity magnitude jη∗ ðωÞj vs frequency f


(Hz). Figure 14.18(b) plots the slope of a power-law
curve fit to the data in Figure 14.18(a). Assuming a
form jη∗ ðωÞjeωs ¼ ð2πÞs f s such that
O-Linked Glycans
log ðjη∗ ðωÞjÞ ¼ log ðð2πÞs Þ  s log ðf Þ (14.8.8)

and s is the slope to the data in Figure 14.18(a). The 5


wt%, for example, has s ¼ 0:77 in Figure 14.18(a),
which compares well to the steady shear viscosity of
Figure 14.17 values where m ¼ 0:86. So the Cox–
Merz rule appears well intact for mucus in this range
of parameter values. Recall the power-law fluid model
of Section 10.3 and Eq. (10.3.2) where
Core Protein μeff ¼ τ=γ_ ¼ mγ_ n1 . There we presented additional
mucus data, Figure 10.19 reviewed in (Velez-Cordero
and Lauga, 2013) to show that n ¼ 0:2 gives a similar
slope of –0.8 compared to –0.77 of Figure 14.18(a)
and –0.86 of Figure 14.17.
Figure 14.16 Mucin glycoprotein molecular structure showing a We expect a spectrum of relaxation times for
core protein with many glycan branches. mucus; however, investigators look for the dominant
value. In Hwang et al. (1969) the crossover frequency
103 yielded λe100 s for lung mucus while Gilboa and
Mucus Viscosity vs Shear Rate
102 Silberberg (1976) find λe30 s. Cervical mucus
A B
101 relaxation times are in the range 1–10 seconds
Viscosity (Pa-s)

m=–0.86
(Eliezer, 1974;Litt et al., 1976;Wolf et al., 1977; Tam
100
E et al., 1980). Note that typical driving frequencies of
10–1 D
C respiratory cilia are f ¼ 10Hz and sperm swimming
10–2 by the beating flagella at f ¼ 20  50 Hz. So
10–3
F De ¼ 2πfλeOð103 Þ for these situations and the fluid
water viscosity

10–4
behaves elastically. However, for gravity driven flows
10–1 100 101 102 103 104 in the lung a characteristic time scale is Ts ¼ μ=ρgh,
Shear Rate (1/s) where h is the mucus film thickness. Let
Figure 14.17 Steady viscosity of various types of human mucus. h ¼ 10 microns103 cm, g ¼ 980 cm= s2 and density
Steady shear viscosity as a function of shear rate for (a, b) ρ ¼ 1:0 g=cm3 . Choose the viscosity to be
chronic bronchitis mucus circle (Puchelle et al., 1983) and μ ¼ 10 g=cms ¼ 1 Pas, which is a typical value in
square (Dulfano et al., 1971); (c) nonovulatory cervical mucus Figure 14.17. The result is Ts  10s, which is close to
(Lai et al., 2007); (d) normal gastric mucus (Curt and Pringle,
the relaxation time.
1969); (e) duodenal ulcer mucus (Curt and Pringle, 1969); and (f )
tears (Tiffany, 1991). Thin dashed lines indicate the typical range
of viscosity values for human mucus suggested by (Cone, 1999).
The thin solid line represents the viscosity of water (103 Pas). Synovial Fluid
00
G0 < G and the mucus behaves more like a viscous We introduced synovial fluid in Chapter 11 when
fluid than an elastic solid. However, for Figure 14.18 studying the lubrication theory of a squeeze film, see
00
(d) with 4 and 5 wt%, G0 > G and the mucus is more Figure 11.28. There we treated it as either a Newtonian
elastic than viscous. Figure 14.18(a) shows the fluid (Section 11.7), or a power-law fluid (Section
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:39 Page Number: 22

22 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

Figure 14.18 Viscoelastic properties


of mucus. (a) Complex viscosity
magnitude jη∗ ðωÞj vs frequency f
(Hz). (b) The slope of a power-law
curve fit, jη∗ ðωÞjeωs to the data in
00
(a). G0 and G vs f (Hz) for (c) mucus
solids concentrations of 1.5, 2, 3 %wt
and (d) mucus solids concentrations
of 3, 4, 5 %wt.

(a)
102 (b)
TKA, 76 y.o. M
101
101 G′ 88 y.o. F
Viscosity (Pa-s)

Moduli (Pa)

100 100 G″ 88 y.o. F


Revision (W), 61 y.o. F
TKA, 70 y.o. F
Revision (N), 75 y.o. F
G′ 72 y.o. M
10–1 10–1 G″ 72 y.o. M
TKA, 83 y.o. F

10–2
Revision (N), 48 y.o. F 10–2
10–1 100 101
10–3 Frequency (1/s)
–3 10–2 10–1 100 101 102 103 104
10
Shear Rate (1/s)

Figure 14.19 (a) Viscosity (Pas) vs steady shear rate (s-1) for samples of synovial fluid in patients undergoing total knee arthroplasty
00
(TKA) and revision TKA. (b) G0 and G vs frequency f (Hz) for two sample patients with one showing crossover.

11.8). Now we explore its viscoelastic behavior. Recall water viscosity e103 Pa  s, which forms the
Figure 14.19(a) shows the viscosity (Pas) vs steady horizontal axis, so the range is 10 to 10,000 times
00
shear rate (s–1) for samples of synovial fluid in water viscosity. Figure 14.19(b) shows G0 and G vs
patients undergoing total knee arthroplasty (TKA) and frequency f (Hz) for two sample patients, one showing
revision TKA. All exhibit shear thinning and most also crossover at 0.87 Hz (for a 88-year-old female) and
have a low frequency plateau which we learned is a the other no crossover (a 72-year-old male). Walking
feature of the Carreau and Cross models, Section 10.7. and running frequencies are indicated. The
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:40 Page Number: 23

14.8 Rheological Measurements of Biofluids: The Cox–Merz Rule 23

O OH
Hyaluronic acid Structural Formula of O 100 : h; : |h*|
Hyaluronic acid: O
HA (Hyaluronan) OH 20 mg/mL
VECTOR OBJECTS HO O
EPS 10 OH
OH 15 mg/mL
HN
10

h, |h*| (Pa.s)
Molecular Formula of O 10 mg/mL
Hyaluronic acid:

(C14H21NO11)n
1 5 mg/mL

0.1
N Nitrogen
C Carbon
O Oxygen n 0.01
H Hydrogen 0.001 0.01 0.1 1 10 100 1000
w(rad/s), g(s-1)
Figure 14.20 Hyaluronic acid is long chains of this disaccharide
molecule where n can be 25,000. Figure 14.21 Cox–Merz plot of HA-1 solutions at 25  Cshowing
experimental data (symbols) for steady viscosity, η, and the
magnitude of the complex viscosity jη∗ j. Lines are predictive
models. Note the strong evidence of the Cox–Merz rule.
experimental results are consistent with the well-
known fact that synovial fluid not only acts as a (Giesekus, 1982), which are additional, more complex,
viscous liquid in low frequency regions, responding to constitutive models for viscoelastic fluids.
slowly moving joints, but also acts as an elastic Synovial fluid relaxation times vary depending on
behavior in high frequency regions, responding to the state of disease for the joint and its fluids. Schurz
rapidly moving joints. and Ribitsch (1987) found a range 40–100 s for
Synovial fluid is composed primarily of hyaluronic normal, 8–20 s for degenerative and 0.02–1.0 s for
acid (HA) molecules. HA is a disaccharide polymer inflammatory conditions. They inferred a relaxation
and can form long chains, see Figure 14.20, where n time as the reciprocal of the steady shear rate which
can be 25,000, for example. corresponded to the onset of shear thinning. Other
Figure 14.21 shows experimental data (symbols) for investigators used the crossover frequency
the steady viscosity, η, and the magnitude of the λ ¼ 1=ωc ¼ 1=2πf c and found a normal range of
complex viscosity, jη∗ j, vs ω ¼ γ_ for HA solutions. 0.65–5.82 s (Safari et al., 1990). Still another study
Note their overlap as strong evidence of the Cox–Merz showed ~10 s (Lai et al., 1977). In terms of HA alone,
rule holding for this fluid. The lines are for the an average relaxation time constant can be derived
predictive models Phan-Thien–Tanner (solid line) from the multimode approaches in Yu et al. (2014)
(Thien and Tanner, 1977) and Giesekus (dashed line) and a value of λ ¼ 3:31 s is reported there.

Box 14.2 Lifting, Walking and Running

Athletics exposes the joints, and the synovial fluid in them, to a wide range of frequencies. For
example, basketball and running, see Figure 14.22(a, b), push the upper and lower extremities
to frequencies in the range of 2 Hz, while slow-lift weight training can be at 0.1 Hz, see
Figure 14.22(c).

6
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:40 Page Number: 24

24 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

Figure 14.22 Running and jumping are typically in the frequency range of 2 Hz: (a) John Grotberg, Grinnell College Basketball
and (b) Anna Grotberg, Castle Triathlon Series. (c) Slow-lift weight training is more like 0.1 Hz, for example bicep curls.
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:40 Page Number: 25

14.9 Oscillatory Squeeze Film: Synovial Fluid 25

14.9 Oscillatory Squeeze Film: For simplicity we choose the constitutive equation
Synovial Fluid for the Maxwell fluid, which reduces to the τ12 -
component
To obtain more experience with a viscoelastic fluid, ∂τ12 ∂u
consider a squeeze film as shown in Figure 14.23. The τ12 þ β ¼μ (14.9.4)
∂t ∂y
plates are of length L and average distance h0 apart.
The walls oscillate toward and away from one another The boundary conditions are no-slip and no-
at frequency, f, or angular frequency ω ¼ 2πf. The penetration at the plates, which is equivalently
upper wall position is h ¼ h0 ð1 þ δ sin ω tÞ, while the symmetry at the midline and no-slip and no-
lower wall is 180 out of phase. This makes the penetration at the upper wall
centerline of the film, y ¼ 0, a line of symmetry for ∂u
the flow where ∂u=∂y ¼ 0 and v ¼ 0. This system is ðy ¼ 0Þ ¼ 0, vðy ¼ 0Þ ¼ 0
∂y (14.9.5)
similar to the squeeze film we studied in the context dh
uðy ¼ hðtÞÞ ¼ 0, vðy ¼ hðtÞÞ ¼
of lubrication theory, Section 11.7, except now the dt
driving force is oscillatory The continuity equation is
 
f ¼ f 0 cos ðωtÞ ¼ ℜ f 0 eiωt (14.9.1) ∂u ∂v
þ ¼0 (14.9.6)
∂x ∂y
In the lubrication limit of ε ¼ h0 =L<<1 and
negligible inertia, α2 ¼ ωh20 =ν<<1, the Cauchy Our approach is to separate variables according to the
momentum equation for this system simplifies to oscillatory driving force
∂τ12 ∂p ^ ðx; yÞeiωt v ¼ v^ ðx; yÞeiωt p ¼ p
u¼u ^ ðxÞeiωt (14.9.7)
¼
∂y ∂x
∂p (14.9.2)
¼0 d^
p ∂^
u
∂y ð1 þ iDeÞ y¼μ (14.9.8)
dx ∂y
Since ∂p=∂y ¼ 0, we know that p ¼ pðx; tÞ and Eq. which can be directly integrated as
(14.9.2) can be integrated directly
p y2
d^
∂p ^ ðx; yÞ ¼ ð1 þ iDeÞ
u þ c1 ðxÞ (14.9.9)
τ12 ¼ y (14.9.3) dx 2μ
∂x
The remaining kinematic boundary condition on u ^ is
where the constant of integration has been chosen to
no-slip at y ¼ hðtÞ. Of course, u ^ cannot depend on t,
be zero so that there is zero shear at y = 0.
so to satisfy no-slip we let the wall position be
hðtÞ ¼ h0 ð1 þ δeiωt Þ and assume small amplitude
oscillations so that jδj<<1. Then we can apply the
y no-slip condition at y ¼ h0 , u^ ðy ¼ h0 Þ ¼ 0, as the
first term of a Taylor series in δ, which solves for c1 ðxÞ
h(t) in Eq. (14.9.9). The result is
pa pa
x
 
p y2  h20
d^
L ^ ðx; yÞ ¼ ð1 þ iDeÞ
u (14.9.10)
dx 2μ
Figure 14.23 Oscillatory squeeze film.
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:41 Page Number: 26

26 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

We can derive v^ from the continuity equation, 10


ð
∂^
u 8
v^ ¼  dy þ c2 ðxÞ
∂x  
^ y3 =3  h20 y
d2 p 6
¼ ð1 þ iDeÞ 2 þ c2 ðxÞ (14.9.11)
dx 2μ
4
From symmetry we have no cross-flow along the
centerline, v^ ðy ¼ 0Þ ¼ 0, which determines |d|
c2 ðxÞ ¼ 0. Finally, the vertical fluid velocity at the
upper plate, vðy ¼ hðtÞÞ, must equal the plate
2
velocity, dh=dt ¼ iωh0 δeiωt . Again using a Taylor
series this condition can be applied at y ¼ h0 so that
v^ ðy ¼ h0 Þ ¼ iωh0 δ, giving
^ h30
d2 p
ð1 þ iDeÞ ¼ iωh0 δ (14.9.12) 0.1 0.5 1 5 10 50 100
dx2 3μ
De
Equation (14.9.12) can be solved for the pressure as e = 0.1, F0 = 1.0
3μωδðDe þ iÞ x2 Figure 14.24 jδj vs De for ε ¼ 0:1, F0 ¼ 1.
^¼
p þ c3 x þ c4 (14.9.13)
h20 ð1 þ De2 Þ 2
ε2 F0  1=2
Imposing symmetry in x at x ¼ 0 on the pressure is jδj ¼ 1 þ De2 (14.9.17)
De
p =dx ¼ 0, which yields c3 ¼ 0. Then let the average
d^
pressure at x ¼ L be zero. That solves for c4 and the To ensure jδj<<1 in Eq. (14.9.17) the range of De is
1=2
result is limited by the relationship ð1 þ De2 Þ =De<<1=ε2 F0 .
Figure 14.24 shows the result that as the synovial
3μωδðDe þ iÞ ðL2  x2 Þ fluid becomes more elastic, with De increasing, the

p (14.9.14)
h20 ð1 þ De2 Þ 2 amplitude of the motion is reduced. This is of great
The force balance on the upper plate is benefit to the knee joint, for example, so that the
surfaces are kept further apart during walking (2 Hz)
ðL
μωδL3 Lz ðDe þ iÞ and running (4 Hz), which protects them from contact
^ dx ¼
f 0 ¼ Lz p (14.9.15)
h20 ð1 þ De2 Þ damage. For a relaxation time of λ ¼ 5 s those
0
frequencies correspond to De ¼ ð2πÞð2Þð5Þ ¼ 62:8
Equation (14.9.15) can be rearranged by defining a and De ¼ ð2πÞð4Þð5Þ ¼ 125:6.
dimensionless force F0 ¼ f 0 =GLLz and recalling that
ε ¼ h0 =L, so that 14.10 Oscillatory Channel Flow
DeðDe þ iÞδ
F0 ¼ (14.9.16)
ε2 ð1 þ De2 Þ In Section 8.3 we analyzed oscillatory flow in a
channel for Newtonian fluids, see Figure 8.8. For a
We can assume that F0 is the real force magnitude Maxwell fluid the governing equations are
input and our goal is to understand how the
viscoelasticity of the synovial fluid modifies the ∂u ∂p ∂τ12
ρ ¼ þ
displacement amplitude jδj, which can be obtained by ∂t ∂x ∂y
∂u ∂τ12 (14.10.1)
rearranging Eq. (14.9.16) to solve for δ and taking the μ ¼λ þ τ12
∂y ∂t
magnitude of the result
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:41 Page Number: 27

14.11 Normal Stresses 27

 
where ∂p=∂x ¼ Bp ei ωt ¼ Bp cos ðωtÞ. Use the ð1 þ iDeÞ cosh ðβYÞ
^
U¼ 1 (14.10.7)
following scales to make Eqs. (14.10.1) dimensionless: β2 cosh ðβÞ
U ¼ u=Us , σ ¼ τ12 =σs , T ¼ ωt, Y ¼ y=b, where the
^ is solved, σ
Now that U ^ becomes
velocity and shear stress scales are
Us ¼ Bp b2 =μ, σs ¼ Bp b. The resulting system is ^ ¼ ð1þiDe
σ 1 ^
Þ dU
ðβYÞ
(14.10.8)
∂U ∂σ dY ¼  β1 sinh
cosh ðβÞ
α 2
¼ eiT þ
∂T ∂Y (14.10.2)
∂U ∂σ Examples of viscoelastic effects on U are shown in
¼ De þ σ
∂Y ∂T Figure 14.25 at time T ¼ π=4 for α ¼ 0:1, 1, 5, 10 and
with boundary conditions of no-slip at the upper wall, De ¼ 0, 0:5, 1, 2. Figure 14.25(a, b) exhibit the phase
UðY ¼ 1Þ ¼ 0, and symmetry at the midline and amplitude changes due to increasing De from
∂U=∂YjY¼0 ¼ 0. The two dimensionless parameters Newtonian, De ¼ 0, to larger elastic contributions. As
are the Deborah number, De ¼ ωλ, and the α becomes larger, α>>1, U exhibits more oscillatory
pffiffiffiffiffiffiffiffiffi
Womersley parameter, α ¼ b ω=ν. To solve the layers Figure 14.25(c, d). This is what we saw in
coupled PDE system in Eqs. (14.10.2), separate Figure 14.10 for the oscillatory wall.
variables by defining
 
UðY; TÞ ¼ ℜ U ^ ðYÞeiT , σðY; TÞ ¼ ℜ σ ^ ðYÞeiT and
substitute 14.11 Normal Stresses
^ ¼ 1 þ d^
iα2 U σ
(14.10.3) Figure 14.26 shows the Weissenberg effect, which is
dY ^
dU
dY¼ð1þiDeÞ^
σ also called “rod climbing.” For the Newtonian fluid,
the rotating rod creates a centrifugal force that throws
Taking the Y-derivative of second equation and
^ the fluid radially outward, lowering the fluid level at
inserting into the first leads to a single ODE for U,
the rod and raising it at the container wall. By
^
d2 U contrast, the viscoelastic fluid develops normal
2^
(14.10.4)
dY  β U ¼ ð1 þ iDeÞ
2 stresses which are stronger than the centrifugal force.
These normal stresses push the fluid radially inward,
where β2 ¼ iα2 ð1 þ iDeÞ. Taking the complex square
pffiffiffiffiffiffiffiffiffiffiffiffiffi forcing it to the climb the rod. The normal stresses
root shows that β ¼ α i  De is create tension along the circular streamlines which act
pffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

α 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi like a hoop stress. Rod climbing is a common
β¼ De2 þ 1  De experience when using an electric mixer for cake
2 pffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
α 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi batter, as well as frosting.
þi De2 þ 1 þ De (14.10.5)
2 Two examples of simple shear flows are shown in
The general solution form for Eq. (14.10.4) is the sum Figure 14.27 with velocity field u ¼ ðu1 ðx2 Þ, 0, 0Þ
of homogeneous and particular contributions _ 2 and γ_ ¼ V0 =h is the uniform
where u1 ðx2 Þ ¼ γx
shear strain rate. On the left is a Newtonian fluid
^ ¼ A cosh ðβYÞ þ B sinh ðβYÞ þ ð1 þ iDeÞ
U where the fluid exerts a shear stress on the upper
β2 moving wall felt as a drag force. On the right is a
(14.10.6) viscoelastic fluid which also exerts a drag force, but in
Imposing the two boundary conditions solves for addition it generates a thrust force perpendicular to
A and B in Eq. (14.10.6), the final form being the moving wall and streamlines. This feature of
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:41 Page Number: 28

28 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

(a) (b)
α = 1.0, T = π/4 α = 1.0, T = π/4
1 1

Y 0.5 Y 0.5

0
–0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
U U
–0.5 –0.5

–1 –1
De = 0 De = 0.5 T = De = 1 De = 2 De = 0 De = 0.5 T = De = 1 De = 2

(c) (d)
α = 5, T = π/4 α = 1.0, T = π/4
1 1

Y 0.5 Y 0.5

0 0
0.01 0.02 0.03 0.04 0.05 0.002 0.004 0.006 0.008 0.010 0.012
U U
–0.5 –0.5

–1 –1
De = 0 De = 0.5 T = De = 1 De = 2 De = 0 De = 0.5 T = De = 1 De = 2

Figure 14.25 U(Y) for T ¼ π=4 comparing De ¼ 0, 0:5, 1:0, 2:0 for (a) α ¼ 0:1, (b)α ¼ 1, (c)α ¼ 5 and (d) α ¼ 10.

viscoelastic fluids derives from the tangled chains of The average normal stress in terms of the stress
its components. The design of a cone-plate viscometer tensor, Tij , is Tm ¼ 13 ðT11 þ T22 þ T33 Þ. It was our
(rheometer), see Figure 10.23, can include a measure choice to relate this average to the pressure as
of vertical thrust to derive the normal stress in Tm ¼ p, consistent with the definition
addition to the torque which gives the shear stress. Tij ¼ Tm δij þ τij , which separates the stress tensor into
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:42 Page Number: 29

14.12 The Equivalent Approach for a Fading Memory Fluid 29

X2 V0 X2 V0
drag drag
h h thrust
g
Newtonian X1 Viscoelastic X1

Figure 14.27 Simple shearing of Newtonian and viscoelastic


fluids showing drag force in both, but also a normal thrust force
in the viscoelastic fluid.

τ11  τ22 N1 τ33  τ22 N2


ψ1 ¼ ¼ 2 , ψ2 ¼ ¼ 2
γ_ 2 γ_ γ_ 2 γ_
Newtonian Viscoelastic (14.11.4)

Figure 14.26 Rod climbing, the Weissenberg effect of In Figure 14.27 it is ψ1 < 0 from the relative size of
viscoelastic fluids. the normal stress in the vertical direction,τ22 . For the
Newtonian fluid in Figure 14.27, τ11 ¼ μ∂u1 =∂x1 ¼
an isotropic part, Tm δij , and deviatoric part, τij . The 0, τ22 ¼ μ∂u2 =∂x2 ¼ 0, τ33 ¼ μ∂u3 =∂x3 ¼ 0 so
trace of Tij is Tkk ¼ T11 þ T22 þ T33 where the repeated ψ1 ¼ ψ2 ¼ 0 as we expect. Normal stresses are
index implies the summation potentially important in synovial fluid flow, where
1 shearing can create thrust to keep joint surfaces apart
Tkk ¼ ðT11 þ T22 þ T33 Þδkk þ τkk
3 and reduce wear.
¼ ðT11 þ T22 þ T33 Þ þ τkk (14.11.1)

Noting that δkk ¼ 1 þ 1 þ 1 ¼ 3, subtract Tkk from


both sides of Eq. (14.11.1), and the result is the trace of 14.12 The Equivalent Approach for a
the extra stress tensor equals zero, Fading Memory Fluid
τkk ¼ τ11 þ τ22 þ τ33 ¼ 0 (14.11.2)
Another approach to the Maxwell model is to multiply
Because τij is symmetric, i.e. τij ¼ τji , its nine Eq. (14.7.4) by ð1=λ Þet=λ to obtain
components reduce to six independent components.  
1 t ∂τij
In addition, Eq. (14.11.2) further reduces the extra e τij þ λ
λ ¼ 2μEij (14.12.1)
λ ∂t
stress tensor to five independent components, the
three off-diagonal shear stresses and two normal and note that the left hand side simplifies to
∂ t

stresses or their linear combinations, for example.


μ t
With regard to viscoelastic fluids, it is standard to eλ τij ¼ 2 eλ Eij (14.12.2)
∂t λ
represent the normal stresses in terms of the first and
second normal stress differences, N1 , N2 , respectively Multiplying through by dt and integrating from all
previous times to the present time, t, via the dummy
N1 ¼ τ11  τ22 , N2 ¼ τ33  τ22 (14.11.3)
variable t0
where N1 and N2 emphasize that it is not the absolute ðt ðt
t0
μ t0
values of normal stresses that affect flow, but rather it 0
d e τij ðt Þ ¼ 2
λ e λ Eij ðt0 Þ dt0 (14.12.3)
is their differences. They are usually represented as the λ
∞ ∞
first and second normal stress coefficients, ψ1 , ψ2 ,
respectively yields the result
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:43 Page Number: 30

30 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

t ðt The first term on the right hand side is zero for the
t0  μ t0
e λ τij ðt0 Þ ¼2 e λ Eij ðt0 Þ dt0 (14.12.4) lower limit for t0 ¼ ∞ assuming that γij ðt0 ¼ ∞Þ is
∞ λ
∞ finite. It is also zero for the upper limit when t0 ¼ t
since γij ðt0 ¼ tÞ because the deformation is measured
We assume that the stress is finite at t0 ¼ ∞ so only
relative to the configuration at time t. The result is
the upper time limit, t0 ¼ t, contributes to the left
hand side, and rearranging gives us ðt  
μ ðtt0 Þ
τij ðtÞ ¼ 2  2e λ γij ðt0 Þ dt0 (14.12.9)
ðt λ
λt μ t0 ∞
τij ðtÞ ¼ 2e e λ Eij ðt0 Þ dt0 (14.12.5)
λ
∞ which now has the appearance of Hooke’s law for a
solid. The function in brackets is called the memory
Since t can be treated as a constant within the t0
function of this viscoelastic material.
integral, we can move terms to get
ðt  
μ ðtt0 Þ
τij ðtÞ ¼ 2 e λ Eij ðt0 Þ dt0 (14.12.6)
λ 14.13 Kelvin–Voigt Viscoelastic Solid:
∞
Biological Tissues
Equation (14.12.6) says that the current value of the
stress at time t depends on the history of the strain A simple model for a viscoelastic solid also involves a
rate at previous values at times t0 . The weighting term spring and dashpot, but instead of being arranged in
G exp ððt  t0 Þ=λ Þ is called the relaxation modulus, series like the Maxwell fluid, they are in parallel as
which incorporates the memory of the deformation shown in the Kelvin–Voigt model of Figure 14.28. The
rate. The further back in time, the greater is the total force is the sum of the two contributions
positive difference t  t0 , but its effect is less and less
F ¼ kU þ μU_ (14.13.1)
important due to the negative sign in the exponential,
i.e. it is an exponentially fading memory. Viscoelastic The corresponding equation for shear deformation is
fluids are often called memory fluids. An additional
τ ¼ Gγ þ μγ_ (14.13.2)
view is to let the rate of strain tensor, Eij, be defined
directly as the time derivative of the strain tensor, γij , For an applied constant stress F0, the solution for the
so that Eij ¼ γ_ ij , displacement U in Eq. (14.13.1) is
ðt   F0

1  eμ=k
t
μ ðtt0 Þ U¼ (14.13.3)
τij ðtÞ ¼ 2 e λ γ_ ij ðt0 Þ dt0 (14.12.7) k
λ
∞
which is the creep of a solid that has a relaxation
Integrating the right hand side of Eq. (14.12.6) by behavior with decay time constant λ ¼ μ=k, resulting
parts gives us in the steady state displacement F0/k. For an applied
constant displacement, U0, there is only the elastic
ðt  
μ ðtt0 Þ component of force,
τij ðtÞ ¼ 2 e λ γ_ ij ðt0 Þ dt0
λ
∞ F ¼ kU0 (14.13.4)
  t
μ ðtt0 Þ 
¼ 2 e λ γij  For shear oscillations of a Kelvin–Voigt model we use
λ ∞
ðt   the approach in Eq. (14.3.1), withτ ¼ τ0 eiωt , γ ¼ γ0 eiωt
μ ðtt0 Þ
2  2 e λ γij ðt0 Þ dt0 (14.12.8) inserted into Eq. (14.13.2) as
λ
∞ τ0 ¼ ðG þ iωμÞγ0 (14.13.5)
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:43 Page Number: 31

Summary 31

F
k

U
Figure 14.28 The Kelvin–Voigt model of a viscoelastic solid.

From Eq. (14.13.5) we see the complex shear modulus of a for G ! 0 or De>>1. These are the opposite limits
00
Kelvin–Voigt solid is G∗ ¼ G0 þ iG where for the Maxwell fluid where fluid behavior occurs
00
G0 ¼ G, G ¼ μω so that their dimensionless versions are for De<<1 and elastic behavior for De>>1.
00
Extending this to a generalized Maxwell model,
G0 G μω
¼ 1, ¼ ¼ De (14.13.6) with n Maxwell elements in parallel with the single
G G G elastic spring, is called the Wiechert model or
The Kelvin–Voigt model behaves as an elastic solid Maxwell–Wiechert model.
in the limit μ ! 0 or De<<1, and as a viscous fluid

Summary

Viscoelastic fluids cover a wide variety of biological In addition, the Cox–Merz rule proved to hold up well
situations and their interesting behaviors for these biofluids so that the steady shear viscosity
characterized by complex shear modulus, components could be equated to the magnitude of the complex
00
G0 and G , and complex viscosity, components μ0 and viscosity, μeff ðγ_ Þ ¼ jμ∗ ðωÞjγ¼ω
_ .
00
μ . In this chapter we have dealt with mucus, blood The important concepts of stress relaxation and a
and synovial fluid as primary examples, noting how relaxation time were discussed and tested by forcing
the elasticity and viscosity vary with the parameters. familiar systems over a range of frequencies: Stokes’
Mucus, in particular, is wide ranging in nature second problem, the squeeze film of a model joint
including respiratory, cervical and gastrointestinal, to with synovial fluid, and oscillatory channel flow. The
which we can add the slime of creatures like slugs Deborah number, De ¼ ωλ, told us when the system
which assists their locomotion. Key features included was at low frequency, viscous fluid dominated, and
identifying any crossover point on the frequency axis, high frequency, elastic dominated. While the Maxwell
00
ω ¼ ωc , where G0 ðωc Þ ¼ G ðωc Þ, and the system model was illuminating, it had only one relaxation
behavior is changing from viscous dominated G0 < G
00 time. Generalized Maxwell approaches, with multiple
00
for ω < ωc to elastic dominated G0 > G for ω > ωc . Maxwell units, on the other hand gave us more
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:43 Page Number: 32

32 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

options, and we explored a two-unit system to see more sophisticated models like the upper convected
how multiple relaxation times can influence the Maxwell and Oldroyd B. Finally, we discussed the
overall system behavior. During a more rigorous unique property of viscoelastic fluids to generate
examination of the Maxwell model we were able to normal stresses from shear flow, which leads to the
draw upon our discussions in Chapter 11 on Galilean Weissenberg effect.
invariance to demonstrate its flaws, and the need for

Problems

14.1 σ ^
^ ð1 þ iDeÞ ¼ dU=dη. Plot the dimensionless lower
wall shear Τ0 ¼ ℜðσ ^ ðη ¼ 0Þ eiT Þ and the upper
wall shear Τ1 ¼ ℜðσ ^ ðη ¼ 1Þ eiT Þvs T for one cycle
with De ¼ 0, 2, 5. Let α ¼ 1. Discuss your results.
e. Repeat part (d) with α ¼ 5 and compare
your results.

14.2 Try a simplistic cough model by considering an


applied shear stress τðt Þ ¼ τ0 et=T 0 to the surface of a
single fluid layer as in Figure 8.24, but with a
viscoelastic fluid. Ignoring inertia and assuming
In Section 14.5 we formulated Stokes’ second problem De<<1, the Cauchy momentum equation for
for a Maxwell Fluid, see Eq. (14.5.17) and conservation of momentum is ∂τ12 =∂y ¼ 0, which
Figure 14.10. Now we will modify that problem by implies that τ12 ¼ τ12 ðt Þ. The stress boundary
imposing a second wall at y ¼ L parallel to the condition is τ12 ðt Þ ¼ τðt Þ aty ¼ h and, therefore, for
oscillating wall which is aty ¼ 0, see figure. This all values of y.
second wall is stationary so the fluid velocity is zero
a. Insert for τ12 ðt Þ into Eq. (14.5.2) and then integrate
there. This is a model of flow in the joint space since
with respect to y to find u(y, t). Use the no-slip
synovial fluid has viscoelastic properties. The
  boundary condition that uðy ¼ 0Þ ¼ 0.
dimensionless velocity, Uðη; TÞ ¼ ℜ U ^ ðηÞeiT , has the
b. Scale your result in part (a) using the
same form as Eq. (14.5.8), except that now η ¼ y=L.
^ ðηÞ is the same as Eq. (14.5.11). dimensionless variables U ¼ u=ðτ0 b=μÞ,
The form of U
Y ¼ y=h¸T ¼ t=T0 and define De ¼ λ=T0 . Plot U vs
a. The boundary conditions to determine c1 and c2 are Y at times T ¼ 0, 1, 2, 3 for De ¼ 0:1.
now U ^ ðη ¼ 0Þ ¼ 1, U
^ ðη ¼ 1Þ ¼ 0where, due to the c. Plot UðY; T ¼ 1Þ for De ¼ 0, 0:25, 0:5. What is
scaling, the upper stationary wall is at η ¼ 1 . What the effect of De on the velocity profile?
are the values of c1 and c2 in terms of β? d. Now calculate the dimensional flow rate
b. LetT ¼ 0, α ¼ 1 and plot the velocity profile U ðηÞ Ðb
qðtÞ ¼ Lz 0 u dy for a unit length Lz in the z-
for De ¼ 0, 2, 5. Be sure you only plot where there direction. Using the scales in part (b), define the
is fluid.  
dimensionless flow rate as Q ¼ q= τ0 h2 Lz =μ and
c. Repeat part (b) for α ¼ 5 and compare your results. plot Q over the time range 0  T  5 for
d. We found in our analysis that the shear stress is De ¼ 0, 0:25, 0:5. What is the effect of
given by σðη; TÞ ¼ ℜðσ ^ ðηÞeiT Þand that viscoelasticity?
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:44 Page Number: 33

Problems 33

e. The dimensional volume that moves during the parameters, De ¼ λ=T0 , β ¼ μm =μs to go along
Ð 5T
cough is v ¼ 0 0 q dt where you can ignore terms with δ. Show Us ðY; TÞ, Um ðY; TÞ and their
multiplied by e5 . What is your answer? What respective ranges of Y. Why is there no
makes the volume greater? How does dimensionless parameter involvingτ0 ?
viscoelasticity, with De<<1, affect it? d. Plot the dimensionless velocities, Us , Um vs Y in
their respective regions. Let
β ¼ 10, δ ¼ 1=2, De ¼ 0:25 and T ¼ 0, 1, 2, 3.
14.3
Describe your results.
e. To explore the effect of viscoelasticity, plot Us , Um
for β ¼ 10, δ ¼ 1=2, T ¼ 1 and De ¼ 0, 0:25, 0:5.
How does viscoelasticity affect the velocities?
f. Dehydration can reduce the serous layer thickness.
To simulate this situation for comparison, set the
serous layer to be 0  y  h=4 and the mucus layer
h=4  y  3h=4 so the mucus layer thickness
remains h=2. Plot Us ð0  Y  1=4Þ and
Consider a two-layer airway liquid lining as shown in
Um ð1=4  Y  3=4Þ, and let β ¼ 1 0, T ¼ 1 and
the figure. The lower layer adjacent to the wall,
De ¼ 0, 0:25, 0:5. Display on the same plot from
0  y  δh, is serous and Newtonian. The upper layer
part (e). How does a reduction of the serous layer
in contact with the air, δh  y  h, is viscoelastic
thickness influence the velocities? Is the De effect
mucus with viscosity μm and Maxwell stress-
modified?
relaxation time λ. Neglecting inertia for thin layers
g. Let the dimensionless mucus flow rate for equal
and assuming De<<1, the governing equations are
layer thicknesses be defined as
Ð1
∂2 us Qm 2 ðTÞ ¼ 1=2 Um dY and the dehydrated flow rate
μs ¼ 0 0  y  δh serous Ð 3=4
∂2 y as Qm 4 ðTÞ ¼ 1=4 Um dY. Plot them on the same
∂um ∂τ12
μm ¼ τ12 þ λ δh  y  h mucus graph for β ¼ 5, De ¼ 0, 0:25, 0:5 over the time
∂y ∂t
range 0  T  5. What are your conclusions?
and the shear stress for all y is the applied cough
shear, τc ¼ τ0 et=T0 .
14.4 For pulsatile flow of a Maxwell fluid the
a. Integrate the serous equation twice and call the applied pressure gradient has steady and time-
integration functions as ðtÞ, bs ðtÞ. Apply the no- periodic contributions, ∂p=∂x ¼ Bs þ Bp cos ðωtÞ. The
slip condition us ðy ¼ 0Þ ¼ 0 and the stress velocity solution in dimensionless form is the sum of
condition, μs ∂us =∂y ¼ τc ðtÞ, to find as ðtÞ, bs ðtÞ. the steady parabolic profile from Eq. (6.5.14)and the
What is your solution for us ðy; tÞ? oscillatory profile from Section 14.10
b. Insert τ12 ¼ τc into the mucus equation and
UðY; TÞ ¼ Us ðYÞ þ Up ðY; TÞ
integrate once to find um ðy; tÞ and let your 1  
integration function be am ðtÞ. Solve for am ðtÞ by ¼ 1  Y2 þ f Re U ^ ðYÞ eiT
2
equating the mucus and serous velocities at
y ¼ δh. What is your solution for um ðy; tÞ? f ¼ Bp =Bs is the ratio of the periodic to steady pressure
c. Use the following scales to make your velocity gradient magnitudes. The dimensionless variables are
 
solutions in parts (a) and (b) dimensionless Us ¼ U ¼ u= Bs b2 =μ , T ¼ ωt, Y ¼ y=band U ^ ðYÞis given
us =ðτ0 h=μs Þ, Um ¼ um =ðτ0 h=μs Þ, Y ¼ y=h, in Eq. (14.10.7) .
T ¼ t=T0 . Define two more dimensionless
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:45 Page Number: 34

34 Chapter 14 Constitutive Equations 3: Viscoelastic Fluids

a. Plot UðY; T ¼ π=2Þ for α ¼ 5, f ¼ 20 and large and small De and how these change with
De ¼ 0, 0:5, 1, 2. What is the effect of De on the the parameters.
profile? Which value of De is representative of f. What are the crossover values Dec and maximum
blood in human physiology? points Dem for part (d). Does Dec ¼ Dem for the
b. Repeat part (a) at time T ¼ π. Is there any region of same parameter set?
reverse flow? If so, does De modify it?
c. Repeat part (a) for α ¼ 1, f ¼ 0:2. Compare
14.6 In the notes for the oscillatory squeeze film of a
your results.
viscoelastic fluid we derived the x-velocity as the real
d. Repeat part (b) for α ¼ 1, f ¼ 0:2. Compare
part of u ¼ u^ ðx; yÞeiωt and the pressure as the real
your results.
part of p ¼ p^ ðxÞeiωt , see Eq. (14.9.10) and Eq.
e. The dimensionless upper wall shear is
(14.9.14). Rearranging Eq. (14.9.16) the displacement
τw ¼ τsw þ τpw where τsw ¼ dUs =dYjY¼1 and
 amplitude is δ ¼ F0 ε2 ðDe  iÞ=De
τpw ¼ f Re σ ^ ðY ¼ 1Þ eiT where σ^ ðYÞis given in
Eq. (14.10.8). Plot τw for α ¼ 5, f ¼ 20 and a. Make the velocity dimensionless with the scales
De ¼ 0, 0:5, 1, 2. How does viscoelasticity affect U ¼ u=ωL, T ¼ ωt, Y ¼ y=h0 , X ¼ x=L. What is
wall shear? Why? ^ ðX; YÞeiT ?
UðX; Y; TÞ ¼ U
Ð1
f. The dimensionless flow rate is Q ¼ 2 U dY. Plot b. Plot UðX ¼ 0:5; Y, T ¼ π=8Þ for De ¼ 0:1, 1, 10
Q for α ¼ 5, f ¼ 20 and De ¼ 0, 0:5, 1,0 2. How does over the interval 1  Y  1 and discuss your
viscoelasticity affect Q? Why? results. LetF0 ¼ 1, ε ¼ 0:1.
g. Repeat part (f ) for α ¼ 5, f ¼ 20. Compare c. Animate the three plots in part (b) on the same
your results. graph. Describe what younsee. o
Ð1
d. Plot the flow rate Q ¼ Re eiT 1 U ^ dY at X ¼ 0:5
for De ¼ 0:1, 1, 10. How does De influence the
14.5
flow rate?
a. For the two-element Maxwell model the
e. Find the dimensionless stream function, ψ, from
dimensionless equations, Eqs. (14.6.19) describe
00 your solution of U in part (a) assuming
the dependence of G0 =G1 and G =G1 as functions
ψðY ¼ 0Þ ¼ 0. Plot the streamlines for T ¼ 0with
of the parameters β ¼ μ2 =μ1 andϕ ¼ λ 2 =λ 1 .
00 F0 ¼ 1, ε ¼ 0:1 and De ¼ 1. Describe your results.
Calculate the values of De ¼ Dem where G =G1 is a
maximum for β ¼ 1 and ϕ ¼ 0:1, 1:0, 10:0.
b. Then calculate the crossover values of De ¼ Dec 14.7 In ventilatory modalities like high frequency
00
where G0 =G1 ¼ G =G1 for the same oscillatory ventilation the liquid lining of the lung is
parameter choices. exposed to a time-periodic shear stress,
c. For the single component Maxwell model, we τ ¼ τ0 cos ðωtÞ, well beyond normal breathing rates,
found Dec ¼ Dem ¼ 1 at both the crossover and the e.g. 3–10 Hz. For a thin viscoelastic film occupying
maximum, Eqs. (14.3.7) and (14.3.8). Is Dem ¼ Dec 0  y  h, with a wall at y ¼ 0 and the air–liquid
for these parameter choices in a two-component interface at y ¼ h, the dimensionless governing
Maxwell system? equations are shown in Eq. (1)
00
d. Plot G0 =G1 and G =G1 vs 0:01  De  100 for ∂σ
β ¼ 10 and ϕ ¼ 0:1, 1:0, 10:0. ¼ 0
∂Y (1)
e. Describe your graph in part (d) and the behavior of ∂σ ∂U
σ þ De ¼
the different parameter choices like the presence of ∂T ∂Y
maxima, crossovers and asymptotic behavior for
Comp. by: Prabhu Stage: Proof Chapter No.: 14 Title Name: Grotberg
Date:23/2/21 Time:06:28:47 Page Number: 35

Image Credits 35

nÐ o
1
d σ̂ f. The flow rate is Q ¼ Re 0 U dY eiT . Plot it for
¼ 0
dY (2) one cycle with α ¼ 0:1 and De ¼ 0, 0:5, 1, 1:5, 2.
̂ d Û What happens to the magnitude and phase of the
σð1 þ iDeÞ ¼
dY maximal Q?
where De ¼ ωλ. We have ignored inertia since
pffiffiffiffiffiffiffiffiffi
α ¼ h ω=ν<<1 due to h ~ 10–20 μm. The scaled
variables are T ¼ ωt, Y ¼ y=h , U ¼ u=ðτ0 h=μÞ, Image Credits
σ ¼ τ12 =τ0 . The boundary conditions are
UðY ¼ 0Þ ¼ 0 and σðY ¼ 1Þ ¼ eiT . Let the solutions
Figure 14.7 Source: photos.com/Getty
be separated into Y and T dependent terms, Figure 14.15a and b Source: author original adapted from
 
U ¼ Re U ^ ðYÞeiT , σ ¼ Re σ ^ ðYÞeiT , and insert into Tomaiuolo, G., A. Carciati, et al. (2016). "Blood linear
Eq. (1). The result is Eq. (2). Eliminating σ^ yields an viscoelasticity by small amplitude oscillatory flow."
^ Rheol Acta 55: 485–495, with gratitutude to
ODE for U
G. Tomaiuolo.
^
d2 U Figure 14.17 Source: author original adapted from Fig 4B of
(3) Lai SK, Wang Y-Y, Wirtz D, Hanes J. Micro- and
dY ¼ 0
2
macrorheology of mucus. Advanced Drug Delivery
Reviews. 2009;61(2):86–100, with gratitude to J. Hanes.
Figure 14.18 Source: Fig 14.18abcd is Fig 5 from Hill DB,
a. Solutions to Eq. (3) are of the form U^ ¼ A þ BY. Vasquez PA, Mellnik J, McKinley SA, Vose A, Mu F, et al.
Solve for A and B using the boundary conditions A biophysical basis for mucus solids concentration as a

^ ðY ¼ 0Þ ¼ 0, σ ^ candidate biomarker for airways disease. PLoS One.
U ^ ðY ¼ 1Þ ¼ ð1þiDe
1
Þ dY 
dU
¼ 1. 2014;9(2).
Y¼1
b. With your solution for U ^ in part (a), substitute into Figure 14.19a Source: author original adapted from Fig 1 of
the second of Eqs. (2) to solve for σ^. Mazzucco D, McKinley G, Scott RD, Spector M. Rheology
of joint fluid in total knee arthroplasty patients. Journal
c. For λ ¼ 30 s and 0:25 Hz  f  3 Hz, show that
of Orthopaedic Research. 2002;20(6):1157–63. With
the range of De is approximately 50  De  500. gratitude to M. Spector.
Plot UðY; TÞ at times T ¼ 0, π=2, π, 3π=2for Figure 14.19b Source: author original adapted from Fig 3 of
De ¼ 0, i.e. Newtonian. Is there symmetry to this Mazzucco D, McKinley G, Scott RD, Spector M. Rheology
flow with respect to time? Why? of joint fluid in total knee arthroplasty patients. Journal
of Orthopaedic Research. 2002;20(6):1157–63. With
d. Repeat part (c) for De ¼ 50. How does your result
gratitude to M. Spector.
differ from part (c)? What about the symmetry? Figure 14.20 Source: Iv_design/Getty
Recall that within the coefficient ð1 þ iDeÞ the i- Figure 14.21 Source: author original adapted from Fig 7 of
π
term is i ¼ ei 2 ¼ cos ðπ=2Þ þ i sin ðπ=2Þ. Yu FY, Zhang F, Luan T, Zhang ZN, Zhang HB.
e. To see the effect of viscoelasticity on the phase of Rheological studies of hyaluronan solutions based on the
the velocity, plot UðY; T ¼ 3π=2Þ for scaling law and constitutive models. Polymer. 2014;55
(1):295–301. With gratitude to HB Zhang.
De ¼ 0, 1, 5, 10, 50. What has happened to the
Figure 14.22a Source: courtesy of Grinnell College
phase and magnitude of the profile? Relate your Sports Information.
answer to Figure 14.9(b), where the behavior of Figure 14.22b Source: courtesy of peachysnaps.
jμ∗ =μj is shown as De>>1. Figure 14.22c Source: gawrav/Getty

You might also like