A Numerical Study On The Hemodynamic Characteristics Within Secondary Curvature Vascular Geometric Model

A Numerical Study on the Hemodynamic
Characteristics within Secondary Curvature

Vascular Geometric Model

Nguyen Minh Tuan

A Numerical Study on the Hemodynamic
Characteristics within Secondary Curvature
Vascular Geometric Model

:

2012 6

The Department of Mechanical and Automotive Engineering
Nguyen Minh Tuan
Nguyen Minh Tuan

()
()
()

2012 6

i

ACKNOWLEDGMENT
This research would not have been possible without the support of many people. First and
foremost, I would like to express my gratitude to my supervisor, Professor Sang-Wook Lee for
who was abundantly helpful and offered invaluable assistance, support and guidance. His
willingness to motivate me contributed tremendously to my thesis. I can not say thank you
enough for his tremendous support and help. Besides, special thanks also to all my lab mates
who willingly helped me and made me feel more at home. This stay has been part of my life that
I will fondly remember for the rest of my life.
Last but certainly not least, I would like to express my deep gratitude with all my heart to my
parents. Even from thousands of miles away, your endless support and encouragement have
strengthened me to complete this long journey.

ii

1. ABSTRACT
In this study, numerical simulations on laminar flow in secondary wavy shaped tubes were
conducted for mean Reynolds number of 250 and 350, which is in the range of physiological
flow-rate and investigated flow structures and particle trajectories both in steady and periodic
inflow conditions. For extensive comparisons, various wave lengths and amplitudes of sinusoidal
and helical function for geometry of tube models were employed. The results showed that small
amplitude secondary curvature has significant influence on the nature of flow patterns and
particle mixing mechanism. This implies that characterizing accurate geometry is essential in
accurate predicting of in vivo hemodynamics and may motivate further study on any possibility
of reflection of secondary flow on vascular remodeling and pathophysiology.
KeywordsSecondary curvature, Sinusoidal wavy tubes, Hemodynamics, finite element
technique
iii

TABLE OF CONTENTS

ACKNOWLEDGEMENT ..........................................................................................................i
ABSTRACT...................................................................................................................................ii
TABLE OF CONTENTS.............................................................................................................iii
LIST OF FIGURES...................................................................................................................v

1. INTRODUCTION.....................................................................................................................1
1.1 Historical works.............................................................................................................1
1.2 Thesis organizing..........................................................................................................2
2. METHOD.................................................................................................................................3
2.1 Geometry.................................................................................................................5
2.2 Mesh generation................................................................................................... 6
2.3 Boundary condition....................................................................................................6
2.3.1 Steady analysis..............................................................................................6
2.3.2 Unsteady analysis.....................................................................................7
2.3.2.1 Womersley solution.........................................................................7
2.3.2.2 Sinusoidal flow-rate waveform.......................................................9
2.3.2.3 Canonical flow-rate waveform.....................................................10
2.4 Numerical method....................................................................................................10
2.4.1 Reynolds number..........................................................................................11
2.4.2 Womersley number.......................................................................................11
2.4.3 Numerical scheme.........................................................................................13
3. RESULTS AND DISCUSSION.............................................................................................16
iv

3.1 Mesh dependent analysis............................................................................................16
3.2 Steady analysis...........................................................................................17
3.3 Unsteady analysis Sinusoidal flow-rate waveform..................................................20
3.4 Unsteady analysis Canonical flow-rate waveform..................................................22
3.4.1 Unsteady analysis.....................................................................................22
3.4.1.1 Axial velocity.........................................................................22
3.4.1.2 Axial vorticity...........................................................................23
3.4.1.3 In-plane velocity.........................................................................25
3.4.1.4 Wall shear stress.........................................................................26
3.4.2 Effect of varying Womersley number..........................................................29
3.4.2.1 Womersley number 4............................................................30
3.4.2.2 Womersley number 2, 2.5, 3.0, 3.5.........................................40
3.5 Helical tube.................................................................................................................42
3.5.1 Effect of geometry on periodic flow............................................................42
3.5.2 Effect of Womersley number on the flow....................................................53
3.6 Conclusion and future study........................................................................................55
4. REFERENCES.........................................................................................................................57
viii

2. LIST OF FIGURES

Figure 2.1 (a) Cerebral blood vessel by color-coded DSA, (b) Pelvic-iliac artery by DSA ........... 3
Figure 2.2 Geometry of sinusoidal wavy tube ................................................................................ 4
Figure 2.3 Sinusoidal tube for CFD simulation with various amplitude of curvature .................... 4
Figure 2.4 Helical geometries with different amplitude of curvature ............................................. 5
Figure 2.5 (a) Planar faces and node located at middle sides of quadratic tetrahedron element, (b)
Curved side and face quadratic tetrahedron ................................................................. 6
Figure 2.6 Nondimensional flowrate as inflow boundary condition for unsteady simulations. ..... 9
Figure 2.7 Canonical flow-rate waveform derived from in vivo measurement in common carotid
as inflow boundary condition ..................................................................................... 10
Figure 3.1, (a) 300,000 quadratic tetrahedron elements, (b) 600,000 quadratic tetrahedron
elements ..................................................................................................................... 16
Figure 3.2 Grid quality independence test for steady flow in sinusoidal tube of A = 0.45D, (a)
axial velocity, (b) axial vorticity, (c) in-plane velocity .............................................. 17
Figure 3.3 Axial velocity (the 1
st
row), axial vorticity (the 2
nd
row) and in-plane velocity (the 3
rd

row) contours at x = 25.5D for sinusoidal wavy tube with different amplitude in
steady analysis. (a) A = 0.1D, (b) A = 0.3D, (c) A = 0.5D, (d) A = 0.1D, (e) A = 0.3D,
(f) A = 0.5D, (g) A = 0.1D, (h) A = 0.3D, (i) A = 0.5D ............................................ 17
Figure 3.4 Radial offset of peak axial velocity varying with sine radius in Cartesian coordinate
system......................................................................................................................... 19
Figure 3.5 Circulation I of dominant vortex varying with sine radius in Cartesian coordinate
system......................................................................................................................... 20
st
nd
rd
row) contours at x = 25.5D at the different time points (phases) for sinusoidal inflow
condition in case of A = 0.1D. ................................................................................... 21
ix

Figure 3.7 Axial velocity contour for several curvature tubes: A = 0.1D, A = 0.35D, A= 0.5D at
several cross sections A-A, B-B, C-C, at peak of systolic, deceleration and diastolic
phase........................................................................................................................... 23
Figure 3.8 Axial vorticity contour for several curvature tubes: A = 0.1D, A = 0.35D, A= 0.5D at
several cross sections A-A, B-B, C-C, at peak of systolic, deceleration and diastolic
phase........................................................................................................................... 24
Figure 3.9 In-plane velocity contour with streamtrace for several curvature tubes: A = 0.1D, A =
0.35D, A= 0.5D at several cross sections A-A, B-B, C-C, at peak of systolic,
deceleration and diastolic phase. ................................................................................ 25
Figure 3.10 Wall shear stress distribution at three different time point for different degree of
curvature..................................................................................................................... 26
Figure 3.11 Time varying wall shear stress for sinusoidal tube with amplitude curvature of 0.1D
at several positions: x = 45R (A), x = 48R (B), x= 51R (C) ..................................... 27
Figure 3.14 Axial velocity (left), axial vorticiy (middle) and in-plane velocity profile (right) for
three time points at three sections in case of A=0.5D ................................................ 30
Figure 3.15 Comparison in axial vorticity profile when A = 0.3D between 85 . 2 = o and 4 = o
.................................................................................................................................... 31
three time points at three sections in case of A = 0.2D .............................................. 31
three time points at three sections in case of A =0.5D ............................................... 33
Figure 3.20 Degree of mixing k according to sinusoidal amplitude curvature, at 4 = o ............ 36
x

Figure 3.21 Particle mixing of sinusoidal tube with amplitude of curvature from 0.1D to 0.5D
with interval of 0.05D at each end of cycle, at deceleration time t = 0.35, at 4 o = . 38
with interval of 0.05D at each end of cycle, at deceleration time t = 0.35, 2.85 o = .
.................................................................................................................................... 39
Figure 3.23 Degree of mixing according to sinusoidal amplitude curvature, 2.85 o = ............ 40
three time points at three sections in case of A =0.3D, 2 o = . .................................. 40
Figure 3.25 At two time points: systolic (t
1
) and deceleration phase (t
2
), considering effect of
varying Womersley number from 2 to 3.5 with interval of 0.5 on distribution of axial
velocity, vorticity, and in-plane velocity of sinusoidal tube with amplitude of
curvature of A = 0.5D. .............................................................................................. 42
Figure 3.26 Inplane velocity transformation of helical tube with alplitude of curvature of A =
0.1D at three time points: systolic (t1), deceleration phase (t2), diastolic (t3), at three
locations. (a) Axial velocity; (b) V
Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-
plane
; (f) Vorticity. ....................................................................................................... 45
Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-
plane
; (f) Vorticity. ....................................................................................................... 46
Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-
plane
; (f) Vorticity. ....................................................................................................... 47
Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-
plane
; (f) Vorticity. ....................................................................................................... 48
Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-
plane
; (f) Vorticity. ....................................................................................................... 49
xi

Figure 3.31 Particle mixing of helical tubes with amplitude of curvature of 0.1D, 0.2D, 0.3D,
0.4D, 0.5D with interval at end of the first, the second, the third and the fifth cycle,
at deceleration time t = 0.35. ...................................................................................... 52
Figure 3.32 Degree of mixing k according to sinusoidal amplitude curvature ............................ 53
Figure 3.33 Comparison flow of properties in helical pipe with alplitude of curvature of A =
0.5D for different Womersley number at three time points: systolic (t1), deceleration
phase (t2), diastolic (t3). (a) Axial velocity; (b) V
Cartesian
; (c) V
Cartesian
V
translation
; (d)
V
Rotation
; (e)V
in-plane
; (f) Vorticity. ............................................................................... 54
1

1. 1. INTRODUCTION
1.1 Historical works
Cardiovascular diseases relate to all kinds of disease that harmfully influence on heart
activity and blood vessel such as stroke, heart attack, high blood pressure According to report
from the annual American Heart Association about mortality rate data, only in 2007 (VeroniqueL.
2010), more than 2200 Americans killed by cardiovascular disease every day. Although quality
of health care incessantly increases over year to deal with severe lesions, about 795 000 people
still have to cope with new or recurrent stroke.
There is much evidence suggesting that the development of cardiovascular diseases such as
atherosclerosis and aneurysm are influenced by local hemodynamic environment. In particular,
the wall shear stress, which is known to be one of the most important hemodynamic factors, has
been demonstrated to be strongly correlated with biological responses.
In the majority of vascular structure, vessel of interest appears relatively straight, the wall
shear stress were estimated based on an assumption of fully developed symmetric velocity
profile. However, sinusoidal wavy shaped vessels rather than straight tubes can normally be
observed in in vivo vascular system. In study about common carotid artery (FordM. 2008), Ford
et al. has presented that the secondary curvature (minor wiggle) in common carotid artery is not
exception, but more common by characterizing the shape and degree of skewing of velocity
profile from cine phase contrast magnetic resonance images. When investigating factors
influencing blood flow in right coronary artery (MyersJ. 2011), Myers et al. also demonstrated
the presence and significance of small amplitude out-of-plane curvature in human right coronary
artery. This small amplitude sinusoidal wavy geometry may have some clinical implication on
the thrombosis and atherosclerosis by the mechanism of secondary flow mixing. In-plane mixing
features in helical tubes were investigated study of Cookson et al. (CooksonA. 2009) and they
demonstrated that the discrepancy between flow field and the particle trajectories would be
explained by considering a combination of translational and rotational reference frame. This
study also showed that secondary curvature in helical geometries were proposed to induce
significant effect on flow structures and mixing effectiveness, then reduce activated platelets that
prone to narrow or occlude arteries outright and cause atherosclerosis.
2

According to the study (Caro 1969) in Nature, it was suggested that local fluid mechanics of
blood have controlling effects on atherosclerosis. After that many studies demonstrated clearly
that blood flow dynamics were important factors in the development, diagnosis and treatment of
vascular disease. In a continuing attempt to achieve the best computational fluid dynamic
analysis, enormous affairs were carried out to get more realistic models, and also apply
appropriate numerical methodology. Moreover, with advancement of in vivo magnetic resonance
imagining angiogram (MRI), subject specific image-based CFD studies have been increasingly
conducted.
Varying in geometry of conduit or changes in vessel shape obviously affects blood flow. In
the present study, in order to better understand flow structures and hemodynamic features in
normal vasculature, sinusoidal geometries and then helical geometries were considered rather
than complicated realistic human vascular. CFD simulations were conducted with constant pitch
of six tube diameters 6D and various curvatures by relatively small amplitudes in steady and
unsteady analysis in which periodic and pulsatile inflow conditions based on sinusoidal flow-rate
waveform and carotid flow-rate wave form.
1.2 Thesis organizing
In chapter 1, some historical works relating to current study and the purpose as well as the
necessity of our work are introduced. Then in chapter 2, geometry and mesh generation for
sinusoidal tube and helical tube are carried out. Chapter 2 also focuses on different types of
boundary condition which are imposed at inlet and numerical scheme for solving the problems of
interest. Chapter 3 is comparison and discussion on the results including results of flow in
different degree curvature sinusoidal tube in steady analysis. The unsteady study to examine
behavior of periodic flow with sine flow-rate wave form and carotid flow-rate wave form for
sinusoidal tubes and helical tubes are also mentioned in chapter 3. The effect of Womersley
number on flow field, and particle mixing behavior are also important parts. Further, the same
unsteady analysis are also considered for helical tube with different amplitude of curvature and
discussed deeply in chapter 3.
3

2. 2. METHOD
2.1 Geometry
Hoi (ABME 2009), Myers (ABME 2001), Kohnston (Phys Med Biol 2007), Caro (Proc R
Soc Lond 1996) and Manbachi (PM 2011) by in vivo studies have shown that small or mild
vascular curvature obviously exist as significant part of vessel as illustrating in figure 2.1.
However, in many studies about blood flow in vessel, blood vessel in which there was presence
of small amplitude curvature usually was considered as straight pipe for simplification reason.
On other hand, in an investigation about thrombotic in helical tube, Huijbregts et al. (Surg. 2009)
has proposed that presence of small or mild vascular curvature can become a possible low risk of
thrombotic occlusion in helical tube. Hence, including curvature even with small degree in
geometry as considering the flow property in a pipe is a significant requirement. In current study,
two kinds of secondary curvature geometries were investigated: sinusoidal and helical tube.

Figure 2.1 (a) Cerebral blood vessel by color-coded DSA, (b) Pelvic-iliac artery by DSA
Sinusoidal curve formulation for the centerline of tube geometry in three dimensional space
of Cartesian coordinate can be described as in following equation.
Asin(t/c). = y
t x =

(2.1)
4

Figure 2.2 Geometry of sinusoidal wavy tube
where the wave length or pitch of the curve is

and Ais the amplitude of the sine
curve. Figure 2.3 shows all the sinusoidal geometries that were used for simulation with various
amplitudes of curvature varying from 0.1D to 0.5D with interval of 0.05D.

Figure 2.3 Sinusoidal tube for CFD simulation with various amplitude of curvature
Centerline of helix tube in 3-D coordinate can be described by following equation
Acos(t/c) z
Asin(t/c) = y
t x
=
=
(2.2)
5

Only five geometries with the range of various amplitude of curvature from A = 0.1D to A =
0.5D with interval of 0.1D were used to investigate the flow properties in helical tube as shown
in Figure 2.4.

Figure 2.4 Helical geometries with different amplitude of curvature
2.2 Mesh generation
Ten-node quadratic tetrahedron as shown in figure 2.5 was used to generate the mesh for
simulation instead of four-node linear tetrahedron element. With second order accuracy, as a
result, quadratic tetrahedron has a significant role in more accurate stress analysis in structures
and also fluid mechanics. The side nodes have not to locate exactly at the midpoints of sides.
Likewise, six nodes that define element face either do not necessarily lie on a plane, but can not
vary too much from it. This constraint not only ensures positive Jacobian determinant, but also
forms a flexible quadratic tetrahedron with curved sides and faces as in figure 2.5 (a).
6

Figure 2.5 (a) Planar faces and node located at middle sides of quadratic tetrahedron element, (b)
Curved side and face quadratic tetrahedron
ICEM-CFD, a commercial mesh generation code, was used to generate quadratic tetrahedron
mesh with smaller element near the wall, and gradually increasing size of elements as coming to
core to guarantee the high quality of the mesh.
2.3 Boundary condition
2.3.1 Steady analysis
Flow was assumed as Newtonian fluid which is proposed to be a reasonable approximation
for flow in large arteries (Friedman 1993). Constant density and temporal independence for flow
properties also were assumed for unsteady analysis. No slip boundary condition was applied for
the wall that indicated zero velocity on the wall boundary. At the other boundary interface,
velocity will be calculated based on value of their neighbor nodes by setting zero gradient
velocity condition. Pressure was assumed to be zero at outlet and zero gradient pressure was put
at inlet and also on the wall of the pipe. Initial value for both pressure and velocity field were set
to be zero. A Reynolds number of 250 based on mean flow-rate
m
Q and tube diameter, which is
in the range of physiological condition, and Womersley number of 2.85 was employed for steady
flow study for all cases of nine geometries of different curvature tubes.
7

2.3.2 Unsteady analysis
2.3.2.1 Womersley solution
Steady flow in a straight pipe analysis ignoring influence of time on flow pattern has a fully
developed characteristic parabolic axial velocity profile. However, considering the effect of time
when investigating flow in a specific conduit is much more complicated. In an attempt to reduce
the CPU time for achieving fully developed flow and ensure accuracy for numerical solutions,
analytical solution to the Navier-Stokes known as Womersley solution (WomersleyR.J. 1955)
with several simplified assumptions was imposed at the conduit inlet. Considering such pulsatile
viscous flow in a straight pipe induced by periodical flow-rate and given pressure gradient in
circumferential and axial coordinate, some representative characteristic fluid properties were
evidently yielded. At a sufficient distance from inlet, velocity is merely a function of radius and
pressure only depends on axial position. A circular pipe of length l , radius R, filled with a
viscous liquid of density and viscosity . Momentum equation at a position r from the axis of
the pipe can be given as (2.3).
)
`
c
c
+
c
c
+
c
c
=
c
c
r
v
r r
v
z
p
t
v
z z z
1 1
2
2
v

(2.3)
With pressure gradient varies with time, considering pressure gradient as
t i
Ae
dz
dp
e
=

(2.4)
Subsequently, fully velocity field can be performed in this form with magnitude component
varying with radius
t i t i
z
e y f e
R
r
f v
e e
) ( ) ( = =

(2.5)
Substitute (2.4) and (2.5) into (2.3), original momentum equation becomes following
ordinary differential equation
)
`
c
c
+
c
c
+ =
t i t i t i t i
e
y
f
R y
e
y
f
R
e
A
e i y f
e e e e
v
e
2 2
2
2
1 1 1
) )( (

(2.6)
Simplify and rearrange (2.6) give us
8

v v
e
2 2
2
2
) (
1 AR R i
y f
y
f
y y
f
=
c
c
+
c
c

(2.7)
The quantity
v
e
2
R
in (2.7) is a non-dimensional parameter known as Womersley number.
We shall rewrite (2.7) in term of Womersley number
v
e
o
2
R
= and y i r o
2
3
= .
v
o o
o
o
2
2 2
2
3
2
3
2
2
2
2
3
) ( ) ( ) (
AR
r if
r
f
i
r
i
r
f
i =
c
c
+
c
c

(2.8)
Simplify (2.8) leads to a single, complex, ordinary differential equation
e
o
i
A
r f
r
f
i
r
r
f
= +
c
c
+
c
c
) ( ) (
1
2
2
3
2
2
(2.9)
Equation (2.9) is in form of a standard Bessel equation of order zero. The general solution for
the above equation is
e
o o
i
A
y i Y B y i J B f
o o
+ = ) ( ) (
2
3
2
2
3
1
(2.10)
This solution must be bounded for 0 = r and satisfies no slip condition at wall of rigid tube
0 ) , ( = t R v
z
. After applying these two boundary conditions, longitudinal velocity becomes
t i
o
o t i
e
i J
y i J
i
A
fe t r
e e
o
o
e
(
(
(
= = 1
) (
) (
) , ( v
2
3
2
3
z

(2.11)
Then, shear rate can be achieve by taking derivative of axial velocity to radius
t i
o
R r
z
e
i J
i J A R i
r
v
t
e
o
o
o
(
(
(
=
c
c
=
=
) (
) (
) (
2
3
2
3
1
2
2
1

(2.12)
Flow-rate can be calculated by integrating Womersley velocity profile in (2.11) over cross
section
9

t i
o
t i
o
R
z
e
i J i
i J
i
A R
e
R
i J i
i J R
i
A
rdr v t Q
e
e
o o
o
e
t
o o
o
e
t
t
(
(
(
=
(
(
(
= =
}
) (
) ( 2
1
2
) (
) ( 2
2 ) (
2
3
2
3
2
3
1
2
2
2
3
2
3
2
3
1
2
0

(2.13)
Flow-rate function expressed in (2.13) was implemented as inflow boundary condition at a
specific time step.

2.3.2.2. Sinusoidal flow-rate wave form
For unsteady simulations, two different types flow-rate waveform normalized by mean flow-
rate at inlet were imposed as inflow condition to investigate the response of flow field. The two
kinds of flow-rate wave form were shown in figure 2.6 and figure 2.7. First one is sinusoidal
flow-rate waveform with the same mean value as the steady case, in another word, the same
value of Reynolds number with steady simulation was used in this case. The second one is
carotid flow-rate waveform that was derived from in vivo measurement in common carotid. In
case of this pulsatile flow, Reynolds number of 350 that matches representative physiological
conditions in common arteries was used instead of 250 as in the other cases. For all cases, fully
developed velocity derived from Womersley flow in rigid pipe was imposed as inflow velocity
profile.

Figure 2.6 Nondimensional flow-rate as inflow boundary condition for unsteady simulations.
10

2.3.2.3. Canonical flow-rate wave form

Figure 2.7 Canonical flow-rate waveform derived from in vivo measurement in common carotid
as inflow boundary condition
2.4. Numerical method
To investigate in vivo flow environment computationally, a necessary starting point is to
recover a geometrically similar definition. However in addition to geometric similarity, we need
to assure dynamic similarity that is the dynamics of the model system must match the real
dynamics. It is not necessary that all the governing parameter values in the model match those in
the real case, but only that the relevant dimensionless similarity parameters of the flow agree.
Considering the non-dimensionlized governing equation including momentum equation and
continuity equation of Newtonian, incompressible flow in steady and unsteady analysis as in
equation (2.14) and unsteady case as in equation (2.15), nondimensional parameters such as
Reynolds number and Womersley naturally appear in these equations.
0 .
Re . Re
2
= V
V + V = V
u
u p u u

(2.14)
0 .
Re . Re
2 2
= V
V + V = V +
c
c
u
u p u u
t
u
o

(2.15)
In curved pipes important parameters are the Reynolds numbers, whilst for unsteady flows
the Womersley number and reduced velocity are also relevant. From now on, we start to consider
the physical interpretation of these parameters to complement their mathematical role in
establishing invariant group transformations.
11

2.4.1 Reynolds number
Reynolds number frequently refers to dynamic similitude analysis in which dynamic
similitude often implies both geometric similarity and kinematic similarity in testing of scaled
models. Thus, Reynolds number has an important role in constructing of a scale model as well as
determines what conditions it should be tested. Reynolds number is a dimensionless parameter
and specifies for a viscous flow with characteristic properties of density, mean velocity pass over
or inside an object with characteristic length or characteristic dimension. With this definition,
Reynolds number is given by
force viscous
force inertial
D
U
V UD
D
= = =
2
Re

(2.16)
The suffix D in formulation indicates that diameter in characteristic length when considering
in a vessel. As mention in (2.15), Reynolds number parameter comes up naturally after non-
dimensional analysis Navier-Stoke equation for a viscous incompressible flow. However, in
another point of view, Reynolds number is interpreted as ratio of inertial forces to viscous forces.
Consequently this leads to a significant role of Reynolds number as factor which identifies
different flow regimes. At a low Reynolds number, where viscous force is dominant, a laminar
flow occurs leading a smooth constant fluid motion. Contradictorily at high Reynolds number,
where inertial force therefore dominates over viscous force, turbulent flow arises and tends to
produce chaotic vortices and unstable flow.
2.4.2 Womersley number
When considering unsteady flows, a commonly used flow parameter is Womersley number
which was defined as
T
D
W
o
v
t 2
2
=
(2.17)
where
v = and T is normally taken as the fundamental period of oscillatory flow. The

problem of interest to (WomersleyJ.R 1957) was oscillatory in a straight pipe. Physically we can
12

interpret the Womersley number as the ratio of pipe diameter to the laminar boundary layer
growth over the pulse period T
T time in growth length Boundary
Diameter
T
D
T
D
W
o
= =
v
v
t 2
2

(2.18)
A dimensional argument commonly used in laminar boundary growth over flat plates is that
the boundary layer growth is proportional to T v and hence the interpretation of this parameter.
The choice of appropriate dimensionless numbers is typically motivated by identifying
parameters which collapse non-dimensionalised data into identifiable regimes. The relevance of
the Womersley number is linked to the exact solution of a Newtonian fluid in a straight circular
pipe subject to a periodic pressure difference. According to Womersley solution which was
found in section 2.3.2.1, the fully developed solution when all components of the flow do not
vary with streamwise distance, is periodic with only the axial velocity component. For a given
Womersley number, the axial velocity component depends only on the ratio of the radial
coordinate over the vessel diameter, r/D, and on t/T and is given in a different from with
equation (2.11)
(
(
(
(
(
(
(
(
= =
T
t
i
o
o
t i
e
i J
D
r
i J
i
AT
fe t r
t
e
o
o
t
2
2
3
2
3
z
1
) (
) 2 (
2
Re ) , ( v

(2.19)
Here Re is the real part of a complex number, i indicates the imaginary unit and J
0
is the
Bessel function of order zero.
As mention in (DoorlyD. 2009), one of the potential limitations of the Womersley number is
that it is related to physical scales within a given cross section of the pipe and not to any axial
length scales. Therefore, in problems involving variation of the flow in the streamwise direction
and hence axial length scales, such as flow within double bends or stenosis, the role of the
longitudinal geometry is not represented. Consequently meaningful ways to present the data may
not easily be identified.
13

2.4.3 Numerical scheme
Numerical simulations were carried out by employing a well-validated, in-house CFD solver
(Ethier 1999) (Ethier 2000) applying quadratic tetrahedral elements based finite element method.
Mesh dependence tests were firstly conducted and demonstrated that approximately 300,000
elements was sufficient to ensure mesh independence solutions. All simulations were run for
sufficient time period in cardiac cycles to ensure fully developed flow and damp out the initial
transients.
The governing equations are spatially discretized using a Galerkin finite element approach in
conjunction with P2-P1 tetrahedral Taylor-Hood elements having 10 velocity nodes and 4
pressure nodes (figure 2.5). The elemental velocity shape function are listed below in local
coordinates ) , , ( , q .
) 1 2 (
1
= |

(2.20)
, | 4
2
=

(2.21)
) 1 2 (
3
= , , |

(2.22)
q, | 4
4
=

(2.23)
) 1 2 (
5
= q q |

(2.24)
q | 4
6
=

(2.25)
) 1 ( 4
7
, q | =

(2.26)
) 1 ( 4
8
, q , | =

(2.27)
) 1 ( 4
9
, q q | =

(2.28)
) 1 )( 2 2 2 1 (
10
, q , q | =

(2.29)
The elemental pressure shape function are given by
=
1

(2.30)
, =
2

(2.31)
q =
3

(2.32)
14

, q =1
4

(2.33)
Using the shape functions, the velocity components
i
u and the pressure are written as
=
=
N
j
j j i i
u u
1
,
|

(2.34)
=
=
p
N
j
j j i i
p p
1
,

(2.35)
Where there are N velocity degrees of freedom for each coordinate direction and N
p
pressure
degrees of freedom. In the above equation,
j i
u
,
refers to the j
th
component of the vector of nodal
velocities{ }
j i
u
,
.
The mapping between physical space and reference coordinate is given by a standard
isoparametric mapping on each element.
=
=
10
1
,
j
j j i i
x x |
(2.36)
where
i
x is three dimensions in Cartesian coordinate. The Jacobian matrix for the coordinate
transformation ) , , ( z y x to ) , , ( , q is therefore given by
=
c
c
=
c
c
=
10
1
,
,
k i
k
k i
i
j
j i
x
x
J

(2.37)
where ) , , ( , q =
i
.
After spatial discretization by Galerkin method, the governing equations (2.15) can be
written in semi-discrete form as
0 } ]{ [
} { ] [ Re } ]{ [ } ]{ [
} {
] [
3
1
1
=
= + + +
= i
i i
i
T
i i i
i
u L
terms boundary p L u S u N
dt
u d
M

(2.38)
15

where } {
i
u is the vector of nodal velocity values for
i
u , } {p is the vector of nodal pressure
values, and entries in the matrices ] [ ], [ ], [ ], [
i
L S N M are given by
}
O
O = d M
j j l k
| | o
2
,
] [

(2.39)
}
= =
O
O
c
c
=
3
1
10
1
, ,
Re ] [
j m j
l
k m m j R l k
d
x
u N
|
| |

(2.40)
}
O
O V V = d S
l k l k
| | . ] [
,

(2.41)
}
O
O
c
c
= d
x
L
j
l
m l k i
|
,
] [

(2.42)
The boundary terms are given by
}
I
O
c
c
+ = d
n
u
x n p terms boundary
i
i k R i
) . ( Re |

(2.43)
where Iis the boundary of O and
i
x
is the unit vector in the
i
x direction. In this numerical
method, contribution of boundary terms is assumed to be zero, which corresponds to either
traction-free or symmetric conditions on all non Dirichlet boundary nodes. Use of boundary
conditions other than the above would require generalizing the treatment of the boundary terms.
In practice, we absorb the factor
l
Re into the pressure vector, and solve for the product
l
p Re } { . Thus, from now on, we will take } {p to be the actual pressure vector times
l
Re , and
drop the factor of
l
Re scaling the pressure term.

16

3. 3. RESULTS AND DISCUSSION
3.1 Mesh dependent analysis
Mesh dependence analysis is necessary before continuing further. A successful mesh analysis
leads to appropriate number of the mesh which saves the memory, CPU time but still ensures
correct computing results. Hence, two different density mesh of 300,000 elements and 600,000
elements were generated for sinusoidal tube with amplitude curvature of A =0.45D by ICEM-
CFD mesh generator software. These two cases then were calculate for steady case, and their
results of axial velocity, axial vorticity and in-plane velocity were compared as shown in figure
3.2. The mild difference between results with 300,000 meshes and 600,000 meshes indicate that
increasing the number of elements more than 300,000 does not affect strongly flow property.
Hence, the mesh of 300,000 elements will be used for the rest of this study.

Figure 3.1, (a) 300,000 quadratic tetrahedron elements, (b) 600,000 quadratic tetrahedron
elements
17

Figure 3.2 Grid quality independence test for steady flow in sinusoidal tube of A = 0.45D, (a)
axial velocity, (b) axial vorticity, (c) in-plane velocity
3.2 Steady analysis

st
nd
rd

row) contours at x = 25.5D for sinusoidal wavy tube with different amplitude in steady analysis.
(a) A = 0.1D, (b) A = 0.3D, (c) A = 0.5D, (d) A = 0.1D, (e) A = 0.3D, (f) A = 0.5D, (g) A = 0.1D,
(h) A = 0.3D, (i) A = 0.5D
18

Axial velocity for three different degree of curvature was extracted at 25.5D x = in
Cartesian coordinate frame as in Figure 3.3 a-c. In overall view, when curvature degree is small
in case of A = 0.1D, axial velocity profile is similar to the one of Poiseuille flow. However, as
the sinusoidal amplitude curvature increases, peak of axial velocity tends to move downward
closer to the tube wall and forms the characteristic crescent shape stream wise velocity profile
that can be commonly observed in typical Dean flow inside a curve pipe.
Figure 3.3 (d)-(f) shows the corresponding axial vorticity contours. Behavior of vorticity in
considering cross sections is a combination of symmetric positive and negative vorticity with
opposite rotation direction. The strength of the vorticity becomes stronger as increasing
curvature of the tube from R = 0.1D to R = 0.5D. Looking at vorticity pattern more deeply,
except two strong vorticities that dominant the vorticity field, there are two small vorticities in
the core and two weak vorticities nearest the wall. The development and decay of these kinds of
vorticity will affect strongly on behavior of velocity field, especially the behavior of flow in a
specific section and consequently influence in flow mixing. In steady analysis, since neglecting
the effect of time varying, periodic variation of vorticity pattern can not be observed. However,
this problem will be dealt with in later part of this study when including sinusoidal and carotid
waveform flow-rate at inlet of conduit.
The planar cross-section velocity magnitude profiles overlaided by streamtraces for the in-
plane reference were shown in figure 3.3 (g)-(i). The in-plane velocity can be calculated as

inplane Cartesian Translation
V V V = (3.1)
where the translation velocity can be achieved by taking the derivative of the geometric equation
of sinusoidal wavy centerline. Since particle trajectories correspond to the stream-trace pattern,
considerable particle mixing can be imagined by the simple sinusoidal wavy tube with relatively
small amplitude.
19

Figure 3.4 Radial offset of peak axial velocity varying with sine radius in Cartesian coordinate
system.
Figure 3.4 shows radical distance of the peak of axial velocity from the center line of
sinusoidal tube varying with amplitude curvature. According to the data, for the smallest
amplitude curvature tube A = 0.1D, since the effect of curve and bend shape on flow field are
relatively weak, axial velocity profile is relatively resemble with one of Poiseuille flow, hence
distance between peak of axial velocity to center is small. In a more curve tube, radical offset of
peak axial velocity obviously increased rapidly until curvature of tube reach value of A = 0.4D.
Above this value, in case of high curvature tube with amplitude of A = 0.45D or with largest
curvature tube A = 0.5D, the rate is much smaller and seem keep relatively unchanged when
increasing amplitude of curvature. Or we can interpret results in a different way, in case of
amplitude of curvature is fairly small, increasing curvature of the tube will cause location of
axial peak velocity move rapidly from center toward the wall. However, peak of axial velocity
20

tends to stop moving as amplitude of curvature becomes so high. Clearly, there is a limit in
motion of peak of axial velocity under the effect of the wall.
Velocity field then was used to achieve vorticity distribution as shown in figure 3.3, and then
vorticity in turn was integrated over the area to obtain the value of circulation
I
. Variation of
circulation versus amplitude of curvature of the tube was calculated and shown in Figure 3.5.
From the data results, circulation started from a small value in small curvature tube then increase
linearly with more curvature. The rate of change of circulation is nearly unchanged when
amplitude A of curvature increases.

Figure 3.5 Circulation I of dominant vortex varying with sine radius in Cartesian coordinate
system.
3.3 Unsteady analysis Sinusoidal flow-rate waveform
In figure 3.6, axial velocity, axial vorticity and in-plane velocity for the case A = 0.1D were
also calculated at the same axial position x = 25.5D at four different time points of a cardiac
cycle in the same manner as the steady case. Overall flow patterns are similar as those of
corresponding steady cases. On the other hand, it can be recognized the presence of unusual axial
velocity pattern at the middle of decelerating phase, which showed high velocity zone stretched
21

laterally. This type of velocity profile was reported in (WakeA. 2005) from magnetic resonance
flow measurement in common carotid artery. Although the flow-rate is same at the time points of
t = 0.5 and t = 1.0, which are the middle of decelerating and accelerating phase respectively, it
was found that different velocity profiles are produced. The influence of modest secondary
curvature also generates non-uniform distribution of hemodynamics parameters such as wall
shear stress, oscillating shear index and particle residence time which was suggested linked to
vascular biology.

st
nd
rd
row) contours at x = 25.5D at the different time points (phases) for sinusoidal inflow condition in
case of A = 0.1D.
Axial velocity, axial vorticity and velocity in-plane results also show a rapid development of
the characteristic crescent-shaped axial velocity profile and the growth of streamwise vorticity
are both evident. The results also show that the growth of streamwise vorticity in the core is
counteracted by the generation of oppositely signed streamwise vorticity near the wall along the
pipe while temporal variety mostly affects on the strength of vorticity without roughly changing
22

the pattern. The vortex pair in core shows a roughly annihilation at t = 0.75 where flow-rate gets
its minimum value.
3.4 Unsteady analysis Canonical flow-rate waveform
3.4.1 Womersley number 85 . 2 = o
3.4.1.1 Axial velocity
In this part, the flow is driven by a non-dimensional canonical pulsatile flow as shown in
figure 2.7 which has been normalized by the value of mean flow-rate at the inlet. The results of
axial velocity distribution are shown in figure 3.7. Axial velocity distribution were obtained in
several sections A-A, B-B, C-C at locations corresponding to x = 22.5D, x = 24D, x = 25.5D and
at different time point in cardiac cycle: peak of systolic t
1
when flow-rate reaches its maximum
value, deceleration phase t
2
when flow-rate decrease in magnitude and finally the at time point t
3

peak of diastolic when flow-rate gets its minimum value.
As expecting, when curvature of sinusoidal tube is small, in case of curvature radius equals
0.1D, axial velocity profile resembles a temporal flow in straight pipe. However, as curvature of
the pipe increases, as in case of curvature radius equals 0.35D or 0.5D, velocity profile rapidly
change to crescent shape which is typical velocity profile shape in curve pipe. At peak of systolic
and diastolic phase, symmetrical velocity with axial velocity peak tending to skew upward or
downward the wall side according to slice position is evidently shown in case of high curvature
tube R = 0.35D and R = 0.5D. In the other hand, under the effect of varying time flow-rate, flow
field in high curvature pipe in deceleration phase becomes unstable, the flow seem to separate
from the wall, hence result in a asymmetric axial velocity profile.
23

Figure 3.7 Axial velocity contour for several curvature tubes: A = 0.1D, A = 0.35D, A= 0.5D at
several cross sections A-A, B-B, C-C, at peak of systolic, deceleration and diastolic phase.
3.4.1.2 Axial vorticity
A similar reaction of axial vorticity distribution was obtained in case of smallest curvature R
= 0.1D, flow in this case act like a flow in straight pipe due to relatively small in curvature.
Obviously, at peak systolic, magnitude of vorticity is more considerable when comparing with
the other time points in cardiac cycle. When time varies from systolic to diastolic, all available
vortex pairs in slice becomes weaker. In case of A = 0.1D, considering periodic influence on the
behavior of vorticity at an interest section such as A-A, two vortices in the core becomes weaken
with time and only a very weak vortex pair remnant at diastolic. Similar behavior was obtained
with two stronger vortexes near the wall. Magnitude of this vortex pair decreases as changing in
time but still dominants at diastolic.
Investigating the vorticity change along the longitudinal direction in case of A = 0.1D at the
time t
1
shows that the growth of axial vorticity pair in the core is weakened by the developing of
two contrariwise vorticity near the wall. The two vorticities in the core becomes weaker in next
section B-B and nearly dissipates in section C-C. The vorticity pair generated near the wall rolls
24

up, detaches from the wall and becomes stronger and forms a new dominant vorticity in cross
section. For preventing this new detached vorticity, nearest the wall, a new pair vorticity is
generated and has the same sign with the primary vorticity pair in the core.

Figure 3.8 Axial vorticity contour for several curvature tubes: A = 0.1D, A = 0.35D, A= 0.5D at
several cross sections A-A, B-B, C-C, at peak of systolic, deceleration and diastolic phase.
In deceleration phase, an unstable occurrence in flow field might cause rotation in vorticity
pattern as shown in figure 3.8 in row 2
nd
of high curvature case R = 0.35D and R =0.5D. In this
stage, the primary vorticity in core rotates and can be subsumed by the vorticity near the wall
whereas the vorticity near the wall can absorb the same sign single vorticity core or is separated
to form a new small vorticity. This process was represented clearly in figure 3.8 at section B-B
of R = 0.5D in deceleration phase. Referring to the dominant vorticity distribution in plane of
high curvature tube, division vorticity pattern into four parts by order of vorticities indicates that
there are four main dominant vortices occurring in cross section. In case of less curve pipe R =
0.1D, there are only two main vortices observed in cross section even at time of systolic when
velocity is relatively high.
25

3.4.1.3 Inplane velocity

Figure 3.9 In-plane velocity contour with streamtrace for several curvature tubes: A = 0.1D, A =
0.35D, A= 0.5D at several cross sections A-A, B-B, C-C, at peak of systolic, deceleration and
diastolic phase.
The planar cross section velocity magnitude profiles overlaided by streamtraces pattern for
pulsatile flow were shown in figure 3.9. The results were obtained in the same manner with the
one in steady case and at the same position and three different time points as the previous work
for axial velocity and axial velocity distribution. In-plane velocity and streamtrace results
revealed the behavior of fluid particles in a plane after subtracting translation motion. As
recording merely the motion in plane, in-plane velocity closely relates to the in-plane mixing or
stirring that was proposed to influence on preventing activated platelets from residing near the
wall and severely accumulating according to study of (CooksonA. 2009). Velocity in-plane in
figure 3.9 also shows that increasing tube curvature leads to a higher velocity field in-plane, and
there are four vortices with stronger intensity dominant section instead of only two vortex in case
of R = 0.1D. Rotation in flow one more time was obtained by the effect of pulsatile flow at
deceleration phase when curvature of tube is relative high.
26

3.4.1.4 Wall Shear Stress

Figure 3.10 Wall shear stress distribution at three different time point for different degree of
curvature
Wall shear stress (WSS) distribution was calculated for all case of varying amplitude of
curvature sinusoidal tube and shown in figure 3.10 with three selected geometries of smallest
curvature tube A = 0.1D, more curvature tube A = 0.35D and the most curve tube A = 0.5D.
Time dependency also took into account in WSS pattern along the pipe where t
1
is peak systolic,
t
2
is one time point in deceleration phase and t
3
is diastolic phase. More detail analysis of WSS
was carried out by considering WSS results at several positions for different degree of curvature.
Figure 3.11, figure 3.12 and figure 3.12 displayed time varying WSS at three positions A, B, and
C with WSS value of straight pipe with the same boundary condition. In figure 3.11, when
curvature of the pipe is small A = 0.1D, that flow field relatively resemble with flow in straight
pipe consequently yielded similar results between four WSS lines. There is minor different
between the pattern between them.
27

at several positions: x = 45R (A), x = 48R (B), x= 51R (C)
However, as curvature of the tube becomes relatively high A =0.35D or in case of the highest
curvature A = 0.5D, flow becomes unstable in deceleration phase at the current Womersley
number of 2.85 as mention in above discussion. Hence, WSS lines at A, B and C position differ
from WSS line of straight pipe both magnitude and pattern shape either. Difference is more
clearly observed at deceleration phase where there is pulse in shape of WSS line. The pulse
become noticeable when increasing A, especially in case of A =0.5D, the peak of pulsatile WSS
gets over than five.
28

29

3.4.2 Effect of varying Womersley number
As mentioned in the previous section, hemodynamics results of flow field in sinusoidal tubes
in case of Womersley number 2.85 o = revealed an unstable occurrence in deceleration phase
with existence of the rotation in axial velocity profile, vorticity pattern, and also in in-plane
velocity profile in spite of symmetrical sinusoidal geometry. We next consider the effect of
varying Womersley number on the flow field by examining a pulsatile flow under similar
conditions but at different frequency which corresponds to a new Womersley number. Several
Womersley numbers were used in this part ranging from 2 o = to 4 o = with interval of 0.5 in
which all the case of geometries with different curvature degree were re-simulated in case of
Womersley number of 4, for the rest smaller Womersley numbers, just few geometries were
considered.
30

3.4.2.1 Womersley number 4 o =

three time points at three sections in case of A=0.5D
Figure 3.14, figure 3.15, figure 3.16, figure 3.17 and figure 3.18 show the results of axial
velocity, axial vorticity and velocity in-plane of sinusoidal tube with amplitude of curvature A of
0.2D, 0.3D, 0.4D, 0.5D respectively. Similar results responding by variation of time and change
in space as in the case of Womersley number of 2.85 was obtained such as, how axial velocity
changes its shape and magnitude, the way vorticity develops, the annihilation of vorticity in the
core by generating new vortex near the wall, and inplane behavior of flow. There is no
significant difference in developing tendency of the flow throughout domain between the case of
85 . 2 = o and 4 = o except the followings.
The first difference is that in case of smaller Womersley number 85 . 2 = o where smaller
frequency pulsatile flow appears, magnitude of axial velocity is smaller than in case of 4 = o .
However, vorticity profiles showed contrary results. Stronger vorticiy in core as well as near the
wall was found in all the cases of 85 . 2 = o . Even at section C-C when amplitude of curvature A
= 0.3D at time t
2
as shown in figure 3.15, while two vorticities in the core which became weaker
after passing through section B-B totally disappeared in case of 4 = o , these two vorticities still
remained as 85 . 2 = o though with small intensity.
The second and the most significant difference is the flow is not unstable anymore as in case
of 4 = o . Hence, symmetrical fluid properties profiles were obtained for all geometry at
31

deceleration phase for high curvature tube and flow pattern was also more easily observed in
specific cross section without rotation.

Figure 3.15 Comparison in axial vorticity profile when A = 0.3D between 85 . 2 = o and 4 = o

three time points at three sections in case of A = 0.2D
32


Increasing curvature of the tube indicates vorticity with higher intensity. As show in figure
3.19 for A = 0.5D, two very strong vortexes in the core at section A-A catches a good agreement
with corresponding streamtrace pattern in velocity in-plane results. There are still four vortices in
this section, but two vortices in core are stronger and closer to the center while the others are
nearly attached to the wall.
33

three time points at three sections in case of A =0.5D
In an early study about flow in helical pipe, (YamamotoK. 2002) by using numerical
calculation of fluid particle trajectory and also experiment method has visualized mixing
behavior caused by helical tube. However, in this study, in-plane mixing merely was considered
qualitatively for achieving a numerical pattern of fluid particle trajectories at specific cross
section with an agreement visualization of secondary flow by experiment. Quantification mixing
behavior was not mentioned on this study. In a later study about chaotic micro mixers particle
tracking method for mixing analysis of (KangT. 2004), after getting position of all color labeling
particle in cross section of interest, a concept with formulation called entropy mixing measure
was used to quantify mixing degree. Next study of (CooksonA. 2009), information entropy of
mixing was used again to display quantification of mixing performance induced by various
curvature helical geometries. Unfortunately, in these studies, only steady analysis was considered
when examine the flow field as well as particle mixing. In current study, the same analyses were
also considered, however, included with time varying effect by changing boundary condition
with time. Hence, mixing behavior will be obtained in a similar manner with previous study
through three basic steps.
The first step is to analysis the flow field in geometry, sinusoidal tube and helical tube with
different amplitude of curvature as in our study, to obtain the periodic velocity field in the whole
domain. The second step is to obtain particle position in desired cross section. In this step, an
initial distribution of particles was generated at the inlet of fluid domain. Particles were labeled
34

by coloring according to their species. In this case, we used two species of particle. Hence
60,000 initial particles consisted of red and blue particles were seeded uniformly in such a way a
relatively equal number between red and blue particles was achieved. Distribution of seeding
particles was illustrated at section A-A, the first column, in figure 3.21 where blue particle was
generated near the wall and the core particle had read color. In order to trace the position
particles throughout fluid domain, advection equation was numerically integrated through the
velocity field. Advection equation respecting to time was shown as.
) , ( t x u
dt
dx
=
(3.2)
where x is the vector position of particle in Cartesian coordinate, u is velocity which is a function
of space and time. Generally, a numerical integration scheme of Runge-Kutta of order 4
integration was used to obtain position of particle inside the fluid. However, for integrating
equation (3.2) information of velocity at any position (x,y,z) was required. Thus, an interpolation
the velocity data at nodal points given by solution of finite element method was carried out to
find out the velocity at desired location. After collecting all information about coordinate and
corresponding color, particle trajectory results or distribution of colored particles in cross section
was yielded by considering the slice has very small thickness. Obviously, the indeed slice where
mixing is examined has a zeros thickness. However, in a limited of choosing time interval as
taking the results from equation (3.2), particle might pass or not reach to the interest cross
section. Thus, a slice with small thickness is a substantial assumption. The final step is
quantification mixing behavior by measuring information of entropy. In a general meaning,
entropy can be interpreted as a tendency of a process or an expression of disorder of randomness.
Thus, non-uniformity or an increasing disorder leads to an increasing in entropy value.
Formulation for calculating mixing entropy was introduced by (KangT. 2004) as in equation
(3.3).

= =
(
=
c s
N
i
N
k
k i k i i
n n w S
1 1
, ,
) log (
(3.3)
In the definition of entropy mixing S shown in equation (3.3), i is the index of the cell, N
c
is
the total number of the cell, in current study N
c
is 10,000 cells in case of sinusoidal tube, w
i
is
the weighting factor defined in such way that w
i
is zero if there is no particle in the cell or there
35

are only particles with single color inside the cell. For the cells in which there are both red and
blue particles, w
i
is set to one. N
s
is the number of species and equals 2 in the current study. n
i,k

is the particle number fraction of the k
th
species in the i
th
cell. An interpretation of equation (3.3)
is that entropy mixing in a cross section is a summation of all entropy information of all the cell
components in which only cells having more than two species inside indeed contribute to total
entropy mixing. Since the entropy itself does not have an adequately physical meaning, an
alternative property, degree of mixingk , is introduced.
0 max
0
S S
S S
= k
(3.4)
where S
0
is minimum mixing entropy at the inlet, S
max
is uniform mixing state where maximum
entropy mixing can be obtained. From equation (3.4), an increasing in entropy mixing or a better
mixing performance indicates an increasing of k . Equation (3.4) also reveals that k equals one
indicating an ideal perfect mixing where red and blue particle totally mix together and k equals
zeros is the case of there is no mixing occur. It should be noticed that mixing entropy will
decrease as increasing number of the box. As number of the box goes to infinity, entropy
consequently becomes zeros. Thus, entropy only exists meaningfully at appropriate number of
boxes and comparing entropy values between cases has a physical meaning if the same number
of counting box was considered.
36

Figure 3.20 Degree of mixing k according to sinusoidal amplitude curvature, at 4 = o
Figure 3.21 shows distribution of colored particles at initial section A-A and at five
subsequent- pitch-length sections for various sinusoidal tube with curvature ranging from A =
0.1D to A = 0.5D with interval of 0.05D. In the case of A =0.1D, because of the small curvature
of the tube, axial flow profile shown in figure 3.14 resembles to flow pattern in a straight pipe
with peak of velocity skews from center line toward the wall in systolic and diastolic. Therefore,
mixing occurred weakly in this case with only a small amount of distortion. Only a small amount
of blue particles near the wall came into core even at downstream location x = 30D at section F-F.
More curve tube induced a strong mixing behavior; more unstructured distribution pattern
between two kinds of species. In case of A = 0.3D, after only three pitch length, most of blue
particle quickly moved to the core and relatively symmetrically located there. Comparison
mixing at downstream location x = 30D, section F-F, of all cases, it is clear that mixing nearly do
not occur in small A case and becomes stronger as increasing A. The strongest mixing in which
blue particles strongly mingled red particles was found is case of A = 0.5D.
37

Due to symmetrical geometry of sinusoidal tube, corresponding symmetrical color particle
distribution was also obtained in case of Womersley number of 4 in which flow field was
revealed steadily even with high curvature amplitude. Moreover, the pattern of blue particles also
gave a good agreement with the results of in-plane velocity profile. In case of A = 0.1D,
according to figure 3.14, vorticity pattern shows that there are merely two vortices dominant in
in-plane frame for the whole periodic cycle. However, in more curve tube, two smaller vortices
were found in deceleration phase. As considering figure 3.21 where particle mixing was obtained
at a time in deceleration phase, the same conclusion also was obtained. The first line of A = 0.1D
obviously yielded there was only two vortex along the tube. While at section C-C with more
curvature tube such as A = 0.35D, it easily realized that besides two dominant vortexes near the
wall, more two smaller vortexes in the core was obviously observed.
Figure 3.20 shows the normalized mixing entropy varying with amplitude of curvature. The
results clearly shows that k increases or mixture becomes stronger after each pitch length.
Magnitude of k depends on position of cross section along the tube and increases linearly
against changing in amplitude of curvature. At the first pitch length, k increases with a small rate
while this rate becomes rapidly considerable at cross section downstream far from the inlet.
We next consider the transient behavior of the unstable flow in case of Womersley number of
2.85. We already discussed this interesting case in part above. When Womersley number is small,
flow become unstable with presence of small scale vortices and deform in flow characteristics
pattern in high amplitude curvature sinusoidal tube. Considering blue particle pattern in particle
trajectory study shown in figure 3.22, at section E-E in case of A = 0.35D, blue particles displays
a structure with two vortices in core and two smaller vortices near the wall with a rotation and
deform in shape. It indicates a transient phenomenon due to separation in flow even though
geometry is totally symmetric and Reynolds number with average of 350 and maximum of 800
that is much smaller from the critical turbulent value. Non-dimensional increasing of entropy
mixing in figure 3.23 shows a decrease of k

in sections of high curvature tube in case of A =
0.35 D to A = 0.45D from which unstable flow occurs. This rate then tend to increase suddenly
as A = 0.5D, however the value of k is generally still smaller than the one of less curve tube A =
0.3D. In another word, the graph of k proposes a negative effect of unstable flow on mixing
behavior by reducing disorder in downstream flow far from inlet.
38

with interval of 0.05D at each end of cycle, at deceleration time t = 0.35, at 4 o = .
39

with interval of 0.05D at each end of cycle, at deceleration time t = 0.35, 2.85 o = .
40

Figure 3.23 Degree of mixing according to sinusoidal amplitude curvature, 2.85 o =
3.4.2.1 Womersley number

three time points at three sections in case of A =0.3D, 2 o = .
41

As mention above, small Womersley number 85 . 2 = o induced an unstable in flow pattern
in deceleration phase for high curvature tube. In order to deeply examine this interesting
phenomenon, larger Womersley number 4 = o was applied to calculated flow field and yielded
that there was no rotation in plane happened and flow was stable for all range of curvature. In
this part, a series of simulation with Womersley number from 2 = o to 5 . 3 = o , interval of 0.5
was carried out to find out the effect of varying Womersley number on stability of the flow.
Figure 3.25 shows a summary of axial velocity, vorticity and in-plane velocity at systolic and
deceleration phase in three different sections as changingo . The most disturb flow was observed
in case of 5 . 2 = o . In deceleration phase of this case, typically crescent shape of axial velocity
was roughly deformed in section B-B and C-C, in section A-A, high velocity region was skew
and shift to asymmetrical shape. Vorticity profile and velocity in-plane overlaid with streamtrace
visually displayed this unstable phenomenon. In case of higher Womersley number 3 = o , and
5 . 3 = o , flow still remain unstable although degree of asymmetry of velocity in-plane is
relatively weaker than the case of 5 . 2 = o . Axial velocity in deceleration phase in these cases
still kept the twisted shape in section C-C. In a logical thought, decreasing Womersley number
will lead to more unstable flow in deceleration phase. However, contradictory was found in case
of 2 = o . Increasing intensity of vorticity as reducing o was obtained again and hence different
vorticity pattern and velocity in-plane occurred in peak of systolic. At section A-A and C-C at
time t
1
, there were merely one vortex pair dominant in spite of two dominant vortex pair in other
cases. Two vortices near the wall kept developing but did not penetrate to the core yet. A similar
behavior was also obtained, streamtrace pattern of in-plane velocity at t
1
indicated only two
vortex appeared at section C-C.
42

Figure 3.25 At two time points: systolic (t
1
) and deceleration phase (t
2
), considering effect of
varying Womersley number from 2 to 3.5 with interval of 0.5 on distribution of axial velocity,
vorticity, and in-plane velocity of sinusoidal tube with amplitude of curvature of A = 0.5D.
3.5 Helical tube
3.5.1 Effect of geometry on periodic flow
The incompressible viscous flow through helical pipe has been investigated in many studies.
Most of them considered helical tube as a circular cross section translated by a helical centerline
as in study of (Cookson 2009) and (Yamamoto 2000). Hence, only 2-D problem was deal with
by solving Navier-Stoker equations which were transform into helical domain. In the current
study, flow in helical tube was investigated but for a true 3-D mesh. Some fluid properties such
as axial velocity, vorticity and in-plane velocity were calculated and compared between different
curvature helical tubes. The process to achieve in-plane velocity is similar with sinusoidal case,
except the effect of torsion caused by characteristic helical shape itself. This effect leads to an
extra term will be subtract from formulation of sinusoidal in-plane velocity in equation (3.1).
This term is rotation velocity that causes the flow profile rotate itself along the tube. Hence, a
new form of equation (3.1) for helical tube becomes
43

(3.5)
where translation velocity can be derived by taking derivative of helical centerline equation with
respect to time in a similar way with sinusoidal tube but translation in z direction was taken into
account.
) cos(
c
x
c
R V
y
x
=
(3.6)
) sin(
c
x
c
R V
z
x
=
(3.7)
Total velocity in Cartesian coordinate was shown in second column from figure 3.26 to
figure 3.30 and subtraction of it from translation velocity was shown in the third column (c). The
velocity in column (c) of different curvature tube is confined to the cross section and has a
similar pattern behavior. Structure of this velocity generally contains one large vortex and
another counter vortex near the wall. The intensity of counter vortex can be comparable with the
other vortex in case of A = 0.1D due to small torsion and small curvature of the pipe. As
increasing amplitude A, curvature consequently increases and leads to a smaller in counter
vortex and transport its core near to the wall. Rotation velocity at a point can be achieved from
the following equation
r V
rotation
e = (3.8)
where r is the distance from location of a point to tube center and e is rotation rate which
depends on axial velocity and pitch length c.
c
V
x
= e
(3.9)
The fourth column (d) shows the pattern induced by rotation velocity in several cross
sections overlaid by streamtraces. We can figure out that although the highest velocity occurs
around the location of peak of axial velocity, as comparison with column (a), the core of rotation
still at the center of the tube. The fifth column (e) displays the in-plane velocity that totally
confined in a plane after subtracting rotation velocity from the value of column (d). A quick view
over all the figures plotted below indicates a typical characteristic of helical tube. Flow in helical
tube is self-similar and rotates itself along the length of tube. This property is demonstrated not
only in the axial velocity profile but also all the other fluid properties. For instant, considering
44

flow motion at one time point such as peak of systolic (t
1
) through sequence of sections A-A, B-
B, C-C which locate at 1/2, 3/4 and 4/4 of the pitch, flow obviously is self-similar with
anticlockwise rotation.
A brief summery for helical tube flow can be draw out by dividing property shape of helical
tube into two: curvature and torsion. Curvature creates a flow with two vortices as discussed
above in sinusoidal tube and torsion induces a rotation in flow that distorts the vortex structure
and consequently strengthens and leads to one vortex prevail over the other. Axial vorticity
profile in column (f) from figure 3.26 to figure 3.30 clearly indicates that there is a single vortex
in the flow.
45

0.1D at three time points: systolic (t1), deceleration phase (t2), diastolic (t3), at three locations. (a)
Axial velocity; (b) V
Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-plane
; (f) Vorticity.
46

Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-plane
; (f) Vorticity.
47

Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-plane
; (f) Vorticity.
48

Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-plane
; (f) Vorticity.
49

Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-plane
; (f) Vorticity.
50

In the case of smallest curvature of the tube A = 0.1D, similar with characteristic of flow in
sinusoidal tube, flow in helical tube with A= 0.1D resembles a flow in straight pipe that shown in
figure 3.26. Period Axial velocity at three different locations has a radical structure with slight
shift from the core center. It indicates that the flow mostly transports follow the center line of
helical tube with minor effect of curvature. The relatively radial axial velocity leads to a uniform
rotation velocity distribution after multiplying with radial position. Hence, after including the
two contributions of translation velocity and rotation velocity into total velocity in Cartesian
coordinate by a subtraction, merely rotated was found as in-plane motion as shown in figure 3.26
(e). The only rotating in-plane results in a corresponding mixing behavior, almost no mixing
occurs in the case of A =0.1D. Figure 3.31 shows particle tracking at interest section at the end
of each pitch also confirm this conclusion. Along the tube length, blue particles near the wall at
initial condition, only rotate and gradually move into the core. The boundary between red and
blue particles still separate, degree of mixing still mild even at section E-E at the end of the
fourth pitch.
In the case of more cure pipe, A = 0.3 D as in figure 3.28, in-plane velocity profile with
streamtrace overlaid yields a clearly two vortices dominating in section instead of only one in
case of A = 0.1D. The others cases also show the similar results. The first vortex was caused near
the location where vorticity gets its maximum value. Besides, due to the large radial shift of the
peak of axial velocity from the center toward the wall and its value is much larger than in case of
A =0.1D, the resulting second vortex becomes stronger and dominates in the other quadrant in
in-plane frame. In this way, with constant torsion (pitch length is constant), magnitude of helical
curvature apparently influences on the size as well as the strength of these two vortices in cross
section. Merely mild change in vorticity structure is evident as increasing curvature of the tube.
Although the intensity of vorticity becomes stronger in more cure pipe, only one larger vortex
occurs with mostly unchanged size of strong vorticity region.
Time varying does not clearly influence on the structure of vorticity except making it weaker
or stronger according to boundary flow-rate value at specific time. Change in space only rotates
the profile with a self-similar structure. Hence, curvature of helical tube can be suggested as the
main factor that decides the size of two vortices in the cross section. The fifth column (e) from
figure 3.26 to figure 3.30 performs a gradually stronger and bigger in size of the first vortex
which is created an incensement in vorticity magnitude caused by more curve pipe. Starting with
51

a very small vortex such as in case of A =0.1D in section A-A at deceleration phase (t
2
) or in
case of A =0.2D at all sections of interest at systolic (t
1
), this vortex keep increasing in size and
can be considerable with the size of the first vortex in case of A =0.3. In more curve tube A =
0.4D and A =0.5D, due to roughly stronger vorticity, the corresponding vortex starts being
stronger than the first vortex even at peak of systolic (t
1
) at which peak of axial velocity gets its
max value.
The influence of deceleration phase is only considerable in case of A =0.2D. Instead of
expecting small vortex that induced by vorticity structure, a larger one was found. The size of
this vortex somehow can be compared with the first vortex that can be obtained if increasing
curvature to A =0.3D.
Figure 3.31 illustrates clearly transition of particle mixing from single vortex to double
vortex and the growth of mixing behavior along the tube as varying A. Particle trajectories are
calculated in a similar manner with sinusoidal tube study. About 60,000 particles were seeded at
section A-A and tracked in domain throughout the tube length. Mixture behavior was captured at
desired sections in a sequence of pitch at the same time point in deceleration phase t = 0.35.
Considering disturbance of blue pattern inside red pattern, we can generally figure out that
mixing occurs increasingly stronger as increase curvature of the tube and also increases after
flow passes one pitch length. However, mixing reducing in case of modest maximum curvature
A = 0.5D was apparently found at section E-E.
A quantity analysis of degree of mixing based on entropy information was plotted in figure
3.32. In this calculation, 3,025 cells were used to obtain entropy mixing by using equation (3.3).
According to this plot, k rapidly increases from A = 0.1D until A =0.3D. However, beyond this
range, the rate tends to be smaller and results in mixing decrease at the third and the fourth pitch
length. An explanation for this plot can be suggest that as the beginning when curvature of the
helical tube relative small, increasing in curvature leads to a growth in strength of vorticity and
hence an increasing in the second vortex strength. The appearance of one more vortex that
induces two vortices dominate in in-plane frame in spite of only one in case of A =0.1D results
in increasing mixing by lengthening interface between blue and red particle, and transporting
more blue particles into the core. As the amplitude curvature of helical tube becomes higher,
increase in the second vortex leads to a decrease of the first vortex which is caused by axial
velocity. This leads to a smaller rate of k was performed after a quick enhancement. The more
52

curve tube in case of maximum value A = 0.5D, comparison between value of column (f) in
figure 3.29 and figure 3.30 shows reduce of vorticity strength due to increased arc length. The
similar pattern of vorticity but weaker strength is easily found in deceleration phase in case of A
= 0.4D and A = 0.5D. In these cases, increase in length of arc curve leads to a smaller of
centrifugal force and then weaker vortices. In this manner, k decrease after second pitch length
in high curve pipe.

Figure 3.31 Particle mixing of helical tubes with amplitude of curvature of 0.1D, 0.2D, 0.3D,
0.4D, 0.5D with interval at end of the first, the second, the third and the fifth cycle, at
deceleration time t = 0.35.
53

Figure 3.32 Degree of mixing k according to sinusoidal amplitude curvature
3.5.2 Effect of Womersley number on the flow
In order to investigate unstable phenomenon occurring in the flow in helical tube, several
value of Womersley number has been imposed. Indeed, decrease from 4 = o to 2 = o and
85 . 2 = o leads to different response of flow behavior. In a general view, this behavior tends is
similar with the remark we got as investigating flow in sinusoidal tube. At the same cross section
(B-B), axial velocity and the other velocity from column (b) to (d) in figure 3.33 becomes
stronger as increasing Womersley number. High axial velocity region in case 2 = o and 85 . 2 = o
has similar pattern of crescent structure whereas in case 4 = o , it is deformed lightly at systolic
(t
1
). Vorticity pattern also is strongly influenced by Womersley number. In relatively high curve
tube figure 3.33 (f), the dominant vorticity is increasingly stronger in low Womersley number.
The developing of intensity of this vorticity itself affects the behavior of in-plane velocity. In
peak of systolic as smallest modest Womersley number 2 = o , in-plane velocity overlaid by
streamtraces shows a totally difference from the others case. There are two more vortices occur
in the region near the peak of vorticity while a single vorticity still dominates in larger
Womersley number cases.
54

Figure 3.33 Comparison flow of properties in helical pipe with alplitude of curvature of A =
0.5D for different Womersley number at three time points: systolic (t1), deceleration phase (t2),
diastolic (t3). (a) Axial velocity; (b) V
Cartesian
; (c) V
Cartesian
V
translation
; (d) V
Rotation
; (e)V
in-plane
; (f)
Vorticity.
55

3.6 Conclusion and future study
Although the amplitude of secondary curvature of constant radius pipe is relatively small, the
study reveals that it has significant influence on the nature of flow patterns and particle mixing
mechanism. This implies that characterizing accurate geometry is essential in accurate predicting
of in vivo hemodynamics and may motivate further study on any possibility of reflection of
secondary flow on vascular remodeling and pathophysiology. Varying in the Womersley number
analysis also yields a strong effect of this factor on the stability of the flow and also the in-plane
behavior. Deceleration phase in pulsatile flow at small Womersley number causes destabilizing
effect on high curvature wavy tube. In-plane frame flow directly results a correlation with a
scalar particle mixing that was believed as a resistance factor for the activated platelets from
residing near the wall and ultimately accumulating. Difference in shape of geometry also plays a
significant role on flow field. Sinusoidal tube characterizes by symmetric flow with counter
rotating vortices whereas helical tube flow is a combination of these symmetric vortices and
rotation of geometry torsion that form a single vortex dominant throughout the tube. Additional
curvature on geometries is not always results in an increase of particle mixing in helical tube.
Too high curve tubes increase the arc length that leads to a decrease in centrifugal force and
consequently weaker vortices. This means the corresponding resultant mixing behavior will
decrease in the range in which amplitude of curvature is too high.
In this study, a good analysis mixing behavior against the change in amplitude curvature
quantitatively by using the entropy information was well performed. Besides, the influence of
Womersley number on particle mixing was shown in sinusoidal tube. Un-stability of flow-field
occurring in sinusoidal tube and disturb helical flow with more vortices in in-plane coordinate
under effect of low Womersley number is also investigated.
Although we successfully obtained dependence of mixing effectiveness on the curvature in
unsteady analysis, the work just stop at one time point in vascular periodic (t=0.35). Considering
mixing at more time point is also a significant requirement. In recent study of (Coppola 2008)
showed that even at a high Reynolds number of Re = 600, flow in helical tube still attached to
the wall bend. In current study, a unique Reynolds number of Re = 350 has been used for
unsteady analysis with carotid flow-rate wave form. This is also one of a limitation of this study.
An investigation of flow field in higher range of Reynolds number is needed in further study. In
a continuing attempt to apply more realistic, the effect of wall structure on the pulsatile flow also
56

will be include. Repeating all analysis consists of fluid structure interaction surely yield
interesting results. The current CFD code using for this study is well validated for laminar flow
for a long period time. However, a transient phenomenon occurring in sinusoidal tube at low
Womersley number even at low Reynolds number requires a further study in this problem. Thus,
higher order spectral scheme is planned to use to minimize any possibility of numerical
dissipation.
57

4. REFERENCES
Caro, C. G. et al. "Arterial Wall Shear and Distribution of Early Atheroma in Man." Nature 223
(1969): 1159-1161.
Cookso, et al. "Mixing through stirring of steady flow in small amplitude helical tubes." Ann.
Biomed. Eng. Vol. 37 (2009): 710-720.
Cookson, A. N. et al. "Mixing through stirring of steady flow in small amplitude helical tubes."
Ann. Biomed. Eng 37 (2009): 710-721.
Coppola, G. and Caro, C. "Oxygen mass tranfer in a model three-dimensional artery." J. R. Soc.
Interface 5(26) (2008): 1067-1075.
Doorly, D. and Sherwin, S. "Geometry and flow." Cardiovascular Mathematics 2 (2009): 177-
210.
Ethier, C. et al. "Comparision between computational hemodynamics, photochromic dye flow
visualization and magnetic resonance velocimetry"." WIT Press, 1999: 131-183.
Ethier, C. et al. "Steady flow separation patterns in a 45 degree junction." Journal of Fluid
Mechanics 411 (2000): 1-38.
Ford, M. et al. "Is flow in the common carotid artery fully developed?" Physiol. Meas 29 (2008):
1335-1349.
Friedman, M. H. "Arteriosclerosis research using vascular flow models: from 2-d branches to
compliant replicas." J. Biomech. Eng. 115(4B) (1993): 595-601.
Kang, T. G. and Kwon, T. H. "Colored particle tracking method for mixing analysis of chaotic
micromixers." J. Micromech. Microeng. 14 (2004): 891.
Myers, J. et al. "Factors influencing blood flow patterins in the human right coronary artery."
Ann. Biomed. Eng. 29 (2011): 109-120.
Veronique, L. R. et al. "Heart Disease and Stroke Statistics - 2011 Update." American Heart
Association, April 2010.
Wake, A. "Modeling fluid mechanics in individual human carotid arteries." PhD thesis, 2005.
Womersley, J. R. "Method for The Calculation of Velocity, Rate of Flow and Visous Drag in
Arteries when The Pressure Gradient Is Known." J. Physiol 127 (1955): 533-563.
Womersley, J.R. "Oscillatory Flow in Arteries: the Constrained Elastic Tube as a Model of
Arterial Flow and Pulse Transmission." Phys. Med. Biol. 2 (1957): 178.
58

Yamamoto, K. et al. "Flow through a rotating helical pipe with circular cross-section." Int. J. of
Heat and Fluid Flow 21 (2000): 213-220.
Yamamoto, K. et al. "Visualization of the flow in a helical pipe." Fluid Dynamics Research 30
(2002): 251 - 267.

A Numerical Study On The Hemodynamic Characteristics Within Secondary Curvature Vascular Geometric Model

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Numerical Study On The Hemodynamic Characteristics Within Secondary Curvature Vascular Geometric Model

Uploaded by

Copyright:

Available Formats

A Numerical Study on the Hemodynamic

Characteristics within Secondary Curvature

v = and T is normally taken as the fundamental period of oscillatory flow. The

You might also like