Professional Documents
Culture Documents
in cerebral aneurysms
A Thesis Submitted
in Partial Fulfillment of the Requirements
for
i
ii
CERTIFICATE
It is certified that the work contained in the thesis titled Effect of Coil Embolization on fluid
loading in cerebral aneurysms by Vishal Agrawal (Roll No. Y9227658) has been carried out
under my supervision and that this work has not been submitted elsewhere for a degree.
Dr. K. Muralidhar
Department of Mechanical Engineering
Indian Institute of Technology Kanpur
Kanpur-208016
June, 2014
iii
ACKNOWLEDGEMENTS
I would like to extend my gratitude towards my supervisors Dr. K. Muralidhar for giving me
the opportunity to work with him. He has been a constant source of inspiration and motivation
for me. He introduced me to the world of research, helped me in understanding the subject and
leading to conclusions through his constant cooperation, help and encouragement.
I would also like to thank Dr. M. K. Das for his invaluable suggestions and tips
throughout my work time and during teaching assistantship duties.
I am thankful to the Department of Mechanical Engineering at IIT Kanpur for providing
me with excellent technical facilities. I am thankful to the institute for the excellent infrastructure
and facilities provided to us.
I would like to thank my friends especially Nikhil and Nishchal for being there with me
in times of need and making my stay at IIT Kanpur memorable.
Last but not least, I would like to thank my parents, sister and brother for the
encouragement they have given me all my life.
Vishal Agrawal
ABSTRACT
The present work is concerned with the fluid dynamics aspects of pulsatile flow through a porous
medium and its application in biomedical systems. Coil embolization is a mildly invasive
endovascular method for treatment of a cerebral aneurysm (Kakalis, Mitsos et al. 2008). The
presence of a coil reduces fluid loading of the blood vessel and delays further deformation of the
walls. Its effectiveness depends on the coil porosity and permeability apart from the nature of
flow pulsations and its geometry.
In the present work, a three dimensional numerical study of pulsatile flow of blood
through an artery with saccular cerebral aneurysm and a patient-specific cerebral aneurysm is
reported. The flow is unsteady but is taken to be laminar and incompressible. The coil is treated
as homogeneous and isotropic porous medium.
For the case of saccular aneurysm, a comparative study has been carried out on
aneurysms with and without a coil insert considering blood as a non-Newtonian fluid. A shear
thinning characteristic model of viscosity is incorporated for the realistic model of blood. The
simulation is carried out for Reynolds numbers Re = 500 and 1500. Results show that the
velocity magnitude within the coil embolized aneurysm becomes negligible after coil insertion.
The wall shear stress within the aneurysm decreases to a great extent for both Reynolds numbers.
Pressure levels remain relatively unchanged. Overall, reduced wall loading with a coil stabilizes
the growth of the aneurysm and thus provides an advantage.
In the second case, coil embolization treatment of a patient-specific geometry under rest
and exercise condition is reported. The geometry is constructed from angiographic images using
vascular modeling tools. Simulations are carried out for rest and exercise condition. Exercise
condition is incorporated with a different physiological velocity waveform than that of rest
condition. A comparative study with and without coil insert with the fluid model being both,
Newtonian and non-Newtonian, is carried out for the rest and exercise condition. Results show
the stabilization of coil induced aneurysm with both rest and exercise condition The coils reduce
blood flow inside the aneurysm which can eventually lead to thrombosis formation and
eventually to aneurysm clotting.
v
TABLE OF CONTENTS
CERTIFICATE ............................................................................................................................... ii
ACKNOWLEDGEMENTS ........................................................................................................... iv
ABSTRACT .................................................................................................................................... v
6.1.2. Pulsatile flow in a patient-specific cerebral aneurysm under rest and exercise
condition ................................................................................................................................ 59
APPENDIX ................................................................................................................................... 61
A.2 PULSATILE FLOW THROUGH A TUBE FILLED WITH POROUS MEDIUM ...... 64
REFERENCES ............................................................................................................................. 68
LIST OF TABLES
Table 2.2: Flow Parameters for pulsatile flow within the aneurysm ............................................ 14
Table 2.3: Flow Parameters for pulsatile flow within the patient-specific aneurysm .................. 18
viii
LIST OF FIGURES
Figure 1.2: (a) Open surgical repair of AAA [http://www.vascularweb.org]. (b) Endovascular
stent graft treatment for AAA [http://www.vascularweb.org]. (c) Clipping treatment for cerebral
aneurysm [http://www.bafound.org]. (d) Coil embolization for cerebral aneurysm
[http://www.mayfieldclinic.com]. ................................................................................................... 4
Figure 2.1: Schematic diagram of the geometric model considered for analysis ......................... 12
Figure 2.2: Inlet average velocity waveform (Valencia, Morales et al. 2008). Points a and b
indicate the time instants at which the solution has been presented in Section 4.1. ..................... 13
Figure 2.3: Model of the patient-specific geometry considered for analysis. ............................... 15
Figure 2.4: Inlet mean velocity waveform for resting condition. Points a and b indicate the time
instants at which the solution has been presented in Chapter 5. ................................................... 16
Figure 2.5: Inlet mean velocity waveform for exercise condition (Bowker, Watton et al. 2010).
Points a and b indicate the time instants at which the solution has been presented in Chapter 5. 17
Figure 3.2: Grid description of patient-specific cerebral aneurysm. (i) Inlet surface mesh. (ii)
Aneurysm cross section. (iii) Surface mesh. ................................................................................. 25
Figure 3.3: Grid description of the aneurysm. (i) Front view, (ii) Crossection view from side
plane. ............................................................................................................................................. 26
Figure 4.1: Time-averaged distribution of absolute velocity (at mid-plane in front view), wall
pressure and wall shear stress for clear medium (first column), porous medium (second column)
and porous medium with non-Newtonian flow (third column) at Reynolds number Re=500 and
Womersley number Wo=7.82 ....................................................................................................... 28
ix
x
Figure 4.2: RMS deviation of absolute velocity (at mid-plane in front view), wall pressure and
wall shear stress from the time-averaged values of the flow variables for clear medium (first
column), porous medium (second column) and porous medium with non-Newtonian flow (third
column) at Reynolds number Re=500 and Womersley number Wo=7.82 ................................... 29
Figure 4.3: Time averaged distribution of absolute velocity (at mid-plane in front view), wall
pressure and wall shear stress for clear medium (first column), porous medium (second column)
and porous medium with non-Newtonian flow (third column) at Reynolds number Re=1500 and
Womersley number Wo=7.82 ....................................................................................................... 30
Figure 4.4: RMS deviation of absolute velocity (at mid-plane in front view), wall pressure and
wall shear stress from the time-averaged values of the flow variables for clear medium (first
column), porous medium (second column) and porous medium with non-Newtonian flow (third
column) at Reynolds number Re=1500 and Womersley number Wo=7.82 ................................. 31
Figure 4.5: Comparison of absolute velocity (at mid-plane in front view), wall pressure and wall
shear stress in clear media at selected phases of the cycle as shown in Figure 2.2 for Reynolds
number, Re=500 and Re=1500 and Womersley number, Wo=7.82.............................................. 32
Figure 4.6: Comparison of absolute velocity (at mid-plane in front view), wall pressure and wall
shear stress in porous media with non-Newtonian model at selected phases of the cycle as shown
in Figure 2.2 for Reynolds number, Re=500 and Re=1500 and Womersley number, Wo=7.82 .. 33
Figure 4.7: Stream traces at the mid-plane in front view at selected phases of the cycle as shown
in Figure 2.2 for Reynolds number, Re=500 and Re=1500 and Womersley number, Wo=7.82.
First row represents the clear medium, second row is porous medium with Newtonian model,
third row being porous medium with non-Newtonian model. ...................................................... 34
Figure 4.8: Schematic model of the bulge region. Points A, B and C indicate the points at which
the pressure and wall shear stress variation .................................................................................. 35
Figure 4.9: Wall shear stress and pressure variation with time at selected points A, B and C in the
bulge region for clear media (first row), porous media (second row) and porous media with non-
Newtonian flow (third row) at Re = 500 (left column) and Re = 1500 (right column) ................ 36
xi
Figure 4.10: Wall Vorticity and wall shear rate contours for the time-averaged distribution at Re
= 500 for (i) Clear medium and (ii) porous medium. ................................................................... 37
Figure 4.11: Wall viscosity variation on the wall (first column) and in the mid-plane section
(second column) of the time-averaged distribution in porous media with non-Newtonian fluid
model for Re 500 and 1500. .......................................................................................................... 38
Figure 5.1: Planes across the geometry at which the velocity contours are shown. ..................... 39
Figure 5.2: Porous medium modeling of coil embolization. A, B, C and D are the points at which
the flow variables are discussed later............................................................................................ 40
Figure 5.3: Time-averaged distribution of the field variables in clear media with rest and exercise
condition: first row, absolute velocity at the planes shown in Figure 5.1; second row, wall
pressure and third row, wall shear stress. ..................................................................................... 42
Figure 5.4: RMS deviation of the field variables from their time-averaged values in clear media
with rest and exercise condition: first row, absolute velocity at the planes shown in Figure 5.1;
second row, wall pressure and third row, wall shear stress. ......................................................... 43
Figure 5.5: Time-averaged distribution of the field variables in porous media with Newtonian
model under rest and exercise condition: first row, absolute velocity at the planes shown in
Figure 5.1; second row, wall pressure and third row, wall shear stress. ....................................... 44
Figure 5.6: RMS deviation of the field variables from their time-averaged values in porous media
with Newtonian model under rest and exercise condition: first row, absolute velocity at the
planes shown in Figure 5.1; second row, wall pressure and third row, wall shear stress. ............ 45
Figure 5.7: Time-averaged distribution of the field variables in porous media with non-
Newtonian model under rest and exercise condition: first row, absolute velocity at the planes
shown in Figure 5.1; second row, wall pressure and third row, wall shear stress. ....................... 46
Figure 5.8: RMS deviation of the field variables from their time-averaged values in porous media
with non-Newtonian model under rest and exercise condition: first row, absolute velocity at the
planes shown in Figure 5.1; second row, wall pressure and third row, wall shear stress. ............ 47
xii
Figure 5.9: Instantaneous distribution of the field variables in clear media at selected phases of
the cycle (shown in Figure 2.4) with resting condition: first row, absolute velocity at the planes
shown in Figure 5.1; second row, wall pressure and third row, wall shear stress. ....................... 49
Figure 5.10: Instantaneous distribution of the field variables in porous media with Newtonian
model at selected phases of the cycle (shown in Figure 2.4) with resting condition: first row,
absolute velocity at the planes shown in Figure 5.1; second row, wall pressure and third row,
wall shear stress. ........................................................................................................................... 50
Figure 5.11: Instantaneous distribution of the field variables in porous media with non-
Newtonian model at selected phases of the cycle (shown in Figure 2.4) with resting condition:
first row, absolute velocity at the planes shown in Figure 5.1; second row, wall pressure and third
row, wall shear stress. ................................................................................................................... 51
Figure 5.12: Instantaneous distribution of the field variables in clear media at selected phases of
the cycle (shown in Figure 2.5) with exercise condition: first row, absolute velocity at the planes
shown in Figure 5.1; second row, wall pressure and third row, wall shear stress. ....................... 52
Figure 5.13: Instantaneous distribution of the field variables in porous media with Newtonian
model at selected phases of the cycle (shown in Figure 2.5) with exercise condition: first row,
absolute velocity at the planes shown in Figure 5.1; second row, wall pressure and third row,
wall shear stress. ........................................................................................................................... 53
Figure 5.14: Instantaneous distribution of the field variables in porous media with non-
Newtonian model at selected phases of the cycle (shown in Figure 2.5) with exercise condition:
first row, absolute velocity at the planes shown in Figure 5.1; second row, wall pressure and third
row, wall shear stress. ................................................................................................................... 54
Figure 5.15: Wall shear stress and pressure variation with time under resting condition at
selected points A, B, C and D in the bulge region for clear media (first row), porous media
(second row) and porous media with non-Newtonian flow (third row). ...................................... 55
xiii
Figure 5.16: Wall shear stress and pressure variation with time under exercise condition at
selected points A, B, C and D in the bulge region for clear media (first row), porous media
(second row) and porous media with non-Newtonian flow (third row). ...................................... 56
NOMENCLATURE
G ϕ Re G
n for clear medium; n for porous medium
B
iω iReω + ϕ / Da
n n
Cf inertia coefficient
D maximum diameter of bulge (m)
d diameter of the tube (m)
Da darcy number (K/d2)
f frequency of oscillation (Hz)
Ik modified Bessel function of kth order.
K permeability (m2)
L length of bulge (m)
m relaxation time (sec)
M1 coefficient of Darcy term
M2 coefficient of Forchheimer term
n harmonic
p intrinsic phase average pressure (Pa)
ρU ref d
Re reynolds number ( )
µ∞
t time (s)
T time period of oscillation (s)
u x-component of velocity (m/s)
ui tensor notation for velocity component (m/s)
Uref reference velocity (time-averaged velocity of mean velocity waveform) (m/s)
v y-component of velocity (m/s)
V ui + vj + wk , velocity vector (m/s)
V* dimensionless velocity vector V
Vt tangential velocity at a point on wall (m/s)
w z-component of velocity (m/s)
xiv
xv
ω
Wo womersley number ( 0.5d )
ν
WSS wall shear stress (N/m2)
xi direction tensor (x, y, z)
Greek symbols
ϕ porosity
Γ exchange coefficient
η normal direction to surface
µ dynamic viscosity (Pa-s)
ν kinematic viscosity (m2/s)
ρ density (kg/m3)
ω 2 , angular frequency of waveform (rad/s)
Suffixes
i denotes the ith direction
n harmonic
CHAPTER 1
INTRODUCTION AND LITERATURE REVIEW
The human body is a complex system that requires transport of materials such as air, water,
minerals and nutrients for survival and functioning. The fluid systems in the human body operate
at different scales. The cardiovascular system is the most important transport system which is
responsible for adequate blood circulation ranging from macro to nano and much smaller scales.
It consists of the heart, blood vessels (arteries, arterioles, venules, veins and capillaries),
lymphatic vessels, and the lungs. . The circulations associated with the cardiovascular system
may be considered under three subsystems. These are the (i) systemic circulation, (ii) pulmonary
circulation, and (iii) coronary circulation.
However, the diseases associated with the cardiovascular system are the leading causes of the
death worldwide. These diseases are mainly caused due to the wrong eating habits of the
individuals. The most common cardiovascular diseases are:
(i) Atherosclerosis: It is known as hardening of the arteries due to formation of plaque,
caused by the deposition of fats and cholesterol. Such deposition causes the arteries to
lose its elasticity, as the arterial wall thickens. It also makes the arteries narrow, thus
impairing the normal blood circulation (Figure 1.1).
(ii) Abdominal Aortic Aneurysm (AAA): Aneurysm is the abnormal localized dilation of
any vessel. AAA is a localized dilatation of the abdominal aorta exceeding the normal
diameter by more than 50 percent, and is the most common form of aortic aneurysm
(Figure 1.1).
(iii) Cerebral Aneurysm: A cerebral aneurysm is a cerebrovascular disorder in which
weakness in the wall of a cerebral artery or vein causes a localized dilation or ballooning
of the blood vessel. A common location of cerebral aneurysms is on the arteries at the
base of the brain, known as the Circle of Willis. Cerebral aneurysms can leak or rupture
causing symptoms from severe headache to stroke-like symptoms, or death (Figure 1.1).
1
2
There are several other common diseases of the cardiovascular system including
Hypertension, Varicose veins and Chronus venous insufficiency. In the present work we restrict
out analysis to cerebral aneurysm of different kinds.
1.1. DIAGNOSIS
The diagnosis of different cardiovascular diseases is performed using various methods:
(i) Radiography: In AAA cases, X-ray images show the calcium deposits on the wall of the
aneurysm. Plain radiographs are accurate in their diagnosis as many as 90% of their cases
(www.medicinenet.com). In case of suspected AAA, X-rays are followed by ultrasound
and CT scans for confirmation. Both walls of the aorta are to be clearly seen for the AAA
aneurysm to be conclusive.
(ii) CT scan: A CT scan accurately shows the dilation of the aorta and the involvement of the
major vessels in close proximity. The treatment method, surgical or endovascular, can be
decided by knowing the degree of calcification. CT scans are considered as 100%
accurate in diagnosis of AAA.
(iii) Angiography: This method is generally employed as a preoperative evaluation technique
in patients with atherosclerotic vascular disease. This method leads to better examination
of the length of the aneurysm as catheters that are used to make measurements in this
method follow the contours of the vessels.
(iv) Medical Ultrasonography: USG is a low cost, highly efficient tool for the diagnosis
where abdominal aorta is entirely seen.
(a)
(b)
(c)
graft is advanced to the aneurysm, and then opened, creating new walls in the blood
vessel through which blood flows (Figure 1.2).
(ii) Endovascular coil embolization: A catheter is inserted into the blood vessel and guided
to ane aneurysm position using the X-rays. Once it is done, tiny metal coils are inserted
inside the aneurysm. The coils cause the aneurysm to clot off and prevent rupture. The
risk in this technique is very low (Figure 1.2).
(a) (b)
(c) (d)
Figure 1.2: (a) Open surgical repair of AAA [http://www.vascularweb.org]. (b) Endovascular
stent graft treatment for AAA [http://www.vascularweb.org]. (c) Clipping treatment for cerebral
aneurysm [http://www.bafound.org]. (d) Coil embolization for cerebral aneurysm
[http://www.mayfieldclinic.com].
5
The primary purpose of this work is to determine the progression of aneurysm in a human artery
using the technique of coil embolization. The method is applied by modeling pulsatile flow
through porous media within the aneurysm region. In this chapter, the generalized model of flow
through a porous region and is mathematical modeling is described. Also, the rheological
properties of blood are described to consider blood as a Newtonian or non-Newtonian model.
Futhermore, the geometry and boundary conditions are elaborated.
Here Vf is the fluid volume and V is the total volume of solid and liquid phase.
8
9
ρUd
Re =
µ
Here U is the reference velocity (generally taken as peak average velocity attained within a
cycle) at the inflow plane.
The Womersley number is measured as the ratio of the oscillatory inertia force to the viscous and
is defined as:
ω 2π f
Wo = 0.5d = 0.5d
ν ν
where f is the frequency of the pulsatile waveform in units of Hz.
The meaning of each symbol is given in the nomenclature. For porous medium, V represents
volumetric phase averaged velocity and p represents intrinsic phase averaged pressure. The
continuity equations remains the same for both the media ( Equation (2.3) and (2.5)) whereas the
clear momentum equation does not contain the Darcy and Forchheimer terms (last two terms in
Equation (2.6)).
The governing equations can be presented in the following non-dimensional form:
Clear medium:
∇i V * = 0 (2.7)
10
∂V * 1 µ
+ ( V * i∇ ) V * = −∇p + ∇. [∇V * + ∇V *T ] (2.8)
∂t Re µ∞
Porous medium (Nield and Bejan 2006):
∇i V * = 0 (2.9)
ϕC
∂V * 1 * 1 µ ϕ f * *
+ ( V .∇ ) V = −ϕ∇p +
*
∇. [∇V + ∇V ] −
* * T *
V − V V (2.10)
∂t ϕ Re µ∞ Re Da Da
The meaning of each symbol is given in nomenclature. The terms used for normalization are
mentioned in the table below. For Newtonian fluid viscosity μ/ ∞ is considered unity.
For low shear rates γɺ < 200 sec-1 the variation of shear stress with strain rate is non-
linear. At higher shear rates this relationship become linear. Generally low shear rates is
observed in small vessels (capillaries and small arteries) and high shear rates are observed in
large vessels.
Several studies have been conducted over the years to study the nature of solid particles
in blood flow. Boryczko et al. (2003) modeled the dynamics of red blood cells and plasma using
a discrete particle approach. It is now understood that the solid phase constitutes almost 45% of
the total blood composition - red blood cells are predominant.
2.4.1. Realistic Model of Blood
A realistic blood model of blood should consider the prominent characteristics of blood, in
particular shear thinning viscosity, viscoelasticity and thixotropy. When blood flows through the
vessels the interaction between blood and solid walls (fluid structure interaction (FSI))
significantly affect hemodynamics. Therefore, FSI also needs to be incorporated to get a realistic
view of a cardiovascular system. However, blood is weakly viscoelastic and shear-thinning
behavior is the more prominent non-Newtonian characteristics of blood. Hence, in this study,
only shear-thinning property of blood is taken into account. A model, called as Carreau-Yashuda
model, is incorporated for the shear-thinning characteristics of blood (Leuprecht and Perktold
2001).
1
µ0 − µ∞ 1 2 ∂u ∂u
µ (γɺ ) = µ∞ + where γɺ = eij eij and eij = i + j (2.11)
a ∂x j ∂xi
(1 + mγɺ n ) 2
Here, µ o and µ ∞ represents the viscosity at zero and infinite shear rate respectively and m denotes
the relaxation time. The shear rate ( γɺ ) is obtained from the strain-rate tensor (eij) as defined
below. On calculation, the shear rate ( γɺ ) comes out to be:
2 2 2 2
∂u ∂v ∂w ∂u ∂v
γɺ = 2 + 2 + 2 + +
∂x ∂y ∂z ∂y ∂x
1
2 22 (2.12)
∂u ∂w ∂v ∂w
+ + + ∂z + ∂y
∂ z ∂ x
Figure 2.1:: Schematic diagram of the geometric model considered for analysis
Pulsatile flow of blood through the bulge is unsteady but is considered incompressible and
laminar. The porous medium fills the bulge but the legs carry clear fluid
fluid. The porous medium is
considered as homogeneous, isotropic and sat
saturated with fluid.
13
At the inflow plane, the Womersley profile (Womersley 1955), derived from a
physiological velocity waveform of internal carotid artery is specified as the boundary condition
for both clear and porous media (Valencia, Morales et al. 2008). The no slip condition is used at
the solid walls. Pressure is specified as constant at the outlet while for velocity, the gradient
outflow condition is prescribed. The time-dependent waveform for the inlet velocity profile is
shown in Figure 2.2.
Figure 2.2: Inlet average velocity waveform (Valencia, Morales et al. 2008). Points a and b
indicate the time instants at which the solution has been presented in Section 4.1.
∞ I (r A ) I (0.5 A ) iω t
n / 1 − 4
∑ Wavgn 1 −
w( z ) = 2Wavg 0 (1 − 4r 2 ) + 0 1 n e n (2.13)
n = −∞ I (0.5 A ) A I (0.5 A )
0 n n 0 n
n≠0
This is the Womersley profile (Appendix A.1). Here r is the non-dimensional radial coordinate
measured from the center point of the inlet. Harmonics Wavgn could be calculated from the
Table 2.2: Flow Parameters for pulsatile flow within the aneurysm
Study Parameters
f = 1.17 Hz
1. Aneurysm without coil inserts. Wo = 7.82
2. Aneurysm with coil and = 0.735
Comparison 1 Da = 1.55 × 10-4 (Based on
Newtonian fluid model.
(For Re 500) inlet diameter)
3. Aneurysm with coil and non-
Newtonian fluid model. Cf = 0.2
ρ = 1050 kg/m3
Carreau-Yashuda Model:
1. Aneurysm without coil inserts. µ o = 0.16 Pa-s
2. Aneurysm with coil and µ ∞ = 0.0035 Pa-s
Comparison 2
Newtonian fluid model. m = 8.2
(For Re 1500)
3. Aneurysm with coil and non- n = 0.64
Newtonian fluid model. a = 1.23
The simulation is carried out for two flow conditions: Rest and Exercise condition. A
comparative study with and without the coil inserts is carried out for both the conditions. The
aim of the study is to determine the effect of exercise condition as compared to the rest condition
within the aneurysm region and to find the effect of coil insertion under these conditions. The
coils are treated as homogeneous and isotropic porous media. A constant pressure condition is
specified at the outlet in this study for both the rest and exercise conditions. No slip condition is
used at the walls. At the inflow plane, different womersley profiles for rest and exercise
conditions, derived from separate physiological velocity waveforms of internal carotid artery, are
specified as the boundary condition.
Rest Condition:
The boundary condition specified at the outlet and wall is same for both the rest and exercise
conditions. However, different physiological velocity waveforms are applied at the inlet plane.
The boundary condition at the inlet is given by:
16
∞ I (r A ) I (0.5 A ) iω t
n / 1 − 4
∑ Wavgn 1 −
w( z ) = 2Wavg 0 (1 − 4r 2 ) + 0 1 n e n (2.15)
n = −∞ I (0.5 A ) A I (0.5 A )
0 n n 0 n
n≠0
The derivation of the above equation can be seen in Appendix A.1. Harmonics Wavgn could be
calculated from the physiological waveform (Figure 2.4) (Bowker, Watton et al. 2010):
∞ iω t
wavg (t ) = ∑ Wavgn e n (2.16)
n = −∞
In mathematical terms:
Figure 2.4: Inlet mean velocity waveform for resting condition. Points a and b indicate the time
instants at which the solution has been presented in Chapter 5.
17
Exercise condition:
The boundary condition at the inlet is specified by the same equation as Eq. (2.15). However, the
physiological velocity waveform (Figure 2.5) differs from the rest condition waveform. In
mathematical terms, the waveform is specified as:
Figure 2.5: Inlet mean velocity waveform for exercise condition (Bowker, Watton et al. 2010).
Points a and b indicate the time instants at which the solution has been presented in Chapter 5.
A comparative study is done between the rest and exercise condition and to analyze the
effect of coil embolization under both the conditions. The coils are treated as homogeneous and
isotropic porous media. The porous medium properties have been taken from Mitsos et al.
(2008).
18
Table 2.3: Flow Parameters for pulsatile flow within the patient-specific aneurysm
Study Parameters
In this chapter, numerical technique, grid generation and code validation are presented. The
governing transport equations are given in Section 2.3.
3.1. DISCRETIZATION
The governing equations ((2.7), (2.8), (2.9) and (2.10)) are solved by using finite volume method
(FVM) on unstructured tetrahedral meshes. Date (2005) has proposed a method for solution of
transport equations on unstructured meshes with cell centered collocated variables approach. The
same approach is followed for simulating pulsatile flow in a clear medium. However, for porous
medium the method specified by Date (2005) is extended to incorporate Darcy and Forchheimer
terms. A method of solution for the transport equations on unstructured meshes with cell-
centered collocated variables approach was discussed earlier by Gohil et al. (2012). This
approach has been extended to incorporate Darcy and Forchheimer terms in the governing
equation as well as the non-Newtonian nature of blood. Equations (2.7)-(2.10) can be rewritten
in the following conservative tensorial form:
∂ ( B) ∂q
j
+ +M B+M V B =S
∂t ∂x 1 2 B
j (3.1)
∂ ( B)
⇒ + div ( q ) + M B + M V B = S
∂t 1 2 B
1 ∂B
where V = u1i + u2 j + u3k , q = q1i + q2 j + q3k and q = u B − Γ B
j ϕ j eff ∂x
i
Here B refers to any flow field variables (e.g. velocity, vorticity). Meanings of rest of the
B
symbols are given in the nomenclature. The expressions of Γ
eff
, SB, M1 and M2 are listed in
Table 3.1. The approach specified by Date (2005) belongs to family of SIMPLE algorithms.
However the concept of pressure smoothing is added by Date (2005). By pressure smoothing the
numerical algorithm could provide reasonably accurate results with very coarse grids. On an
unstructured mesh, it is convenient to employ collocated variables so that scalar and vector
19
20
variables are defined at identical locations. Convective terms are discretized by a hybrid upwind
scheme that combines first and second order schemes. Geometry invariant features of the
tetrahedral element are used so that the calculation of gradients at cell faces is simplified using
nodal quantities of a particular variable. Nodal quantities in turn are calculated as a weighted
average of the surrounding cell-centered values. Diffusion terms are discretized using a second
order central-difference scheme.
Continuity equation
1 0 0 0 0
(porous and clear medium)
Momentum equation (clear 1 µ ∂p
− 0 0
medium) Re µ∞ ∂xi
Momentum equation 1 µ ∂p ϕ ϕ
−ϕ
(porous medium) Re µ∞ ∂xi Re Da Da
The tetrahedral control volume is shown in Figure 3.1. The final form of the discretized
momentum equation (where B refers to velocity) is:
4 ρ∆V 4
AP u P + ∑ AEK u E ,k = S ∆V + u 0p + ∑ Dk (3.2)
k =1 ∆t k =1
Here, p – suffix, signifies control volume cell, k − suffix, signifies neighboring cells
u P − velocity component (either x or y or z)
∆V − control volume
∆V 4
AP = − ∑ AE + M1∆V + M 2 VP ∆V , coefficient of control volume cell
∆t k =1 k
1
AEk = − d − (1 − f ck ) Cck , coefficient of neighboring cells
ϕ ck
th
d − diffusion coefficient for k face
ck
f ck − weighting factor
α A2fk
AEk = u , coefficient of neighboring cells
APcki
D p ' − contains the gradient of pressure correction terms. These terms are neglected
k
(vi) Evaluate residuals from the imbalance equation for all velocity components according to:
0.5
4
2
Rϕ = ∑ AP u + ∑ AE u − S
all nodes P K E ,k total
k =1
Here, S is the total source term, which is the sum of all the term in right hand side of
total
Equation (3.2).
(vii) The mass residual is evaluated as:
0.5
4
2
Rm = ∑ AP p ' − ∑ AE p '
all nodes P
k =1
K E ,k
(viii) Check the convergence criteria based on Rϕ and Rm .
(ix) If convergence criterion is not satisfied, then set pPl +1 = pPl and u Pl +1 = u Pl and go back to
step (ii).
(x) If convergence criterion is satisfied go to next time step and set the solution of the present
time step as the initial solution of next step.
23
parallel construct are executed in parallel by each thread in the team until a work sharing
construct is encountered. The PARALLEL DO or DO directive is such a work sharing construct
which distributes the workload of a DO loop among the members of the current team. An
implied synchronization occurs at the end of the DO loop unless an END DO NOWAIT is
specified. Data sharing of variables is specified at the start of parallel or work sharing constructs
using the SHARED and PRIVATE clauses. In addition, reduction operations (such as
summation) can be specified by the REDUCTION clause. Upon completion of the parallel
construct, the threads in the team synchronize and only the master thread continues execution.
The concept of orphan directives in OpenMP allows user to specify control or synchronization
from anywhere inside the parallel region, not just from the lexically contained region.
3.4. GRID GENERATION
ICEM-CFD tool (a commercial software) is used to generate high quality unstructured
tetrahedral meshes within the geometry. A very high quality mesh is used for both clear and
porous medium (85 % of the tetrahedrons have quality greater than 0.9 and minimum quality of
0.65 is ensured).
25
(i) (ii)
(iii)
Figure 3.2: Grid description of patient-specific cerebral aneurysm. (i) Inlet surface mesh. (ii)
Aneurysm cross section. (iii) Surface mesh.
26
3.4.2. Sphere-shaped
shaped Saccular Aneurysm
The mesh used for pulsatile through through a sphe
sphere-shaped
shaped aneurysm is shown in
Figure 3.3.. The grid is quite refined with over 1.1 x 106 elements. It contains a total of 1,92,640
nodes to form a tetrahedral mesh where each cell is highly regular with a near
near-unity
unity aspect ratio.
Code validation of the FVM solver is further discussed by Sikarwar et al. (2013)).
(i)
(ii)
Figure 3.3:: Grid description of the aneurysm. (i) Front view
view, (ii) Crossection view from side
plane.
CHAPTER 4
PULSATILE FLOW IN A SACCULAR CEREBRAL ANEURYSM
In this chapter the simulation results of flow in a tube with sphere-shaped aneurysm at the
bending location are reported.
Velocity
bulge is a zone of separation and hence spatially uniform pressure. This trend continues with the
bulge carrying the coil, uniform pressure arising from slow unidirectional flow through the
model. Figure 4.1 also shows that the increase in overall resistance to flow in the presence of a
coil is small, mainly because flow has bypassed the bulge region.
Figure 4.2 presents RMS variation of total velocity, wall pressure, and wall shear stress
after removing the time-averaged
averaged value at every cell. The trends here are similar to Figure 4.1 in
29
Velocity
the sense that (a) flow in the bulge significantly reduces in the presence of a coil, (b) coil reduces
the wall shear stress, and (c) the difference between Newtonian and non
non-Newtonian
Newtonian treatments is
small. The changes in RMS wall pressure are however quite sm
small.
all. Consequently, wall pressure
is mostly unaffected by the presence of the coil in the bulge. This result arises from the small
absolute velocity prevailing in the bulge with and without the coils in place.
Figure 4.3 represents the time
time-averaged
averaged values for a higher Reynolds number, i.e. Re =
30
Velocity
1500. On increasing the Reynolds number from 500 to 1500 with the same flow conditions, the
embolization technique still works effectively. The wall loading factors increase to very high
values, without any coil insertion, in the bulge region. The extent of the embolization method can
be gaged by the decrement seen in the wall loading factors after coil insertion. The velocity
within the bulge region comes down to zero order values, as is the case observed with the
th wall
shear stress. However, the pressure values remain the same within the bulge region. The velocity
and wall shear stress decreases to 1/10th of their values in the porous medium as compared to the
31
Velocity
clear medium. The aneurysm region is fully stabilized after coil insertion in both the cases with
Reynolds number, Re = 500 and 150
1500.
Calculations were repeated at a Re of 1500 for the geometry of interest (see Figure 4.4
for RMS data). The trends seen at Re=500 were realized here as well, except that the velocity
magnitude in the bulge was higher; so was the dimensionless pressure magnitude. With the
insertionn of a coil, velocity and shear stress values fell by nearly a factor of ten. The pressure
values stayed nearly constant. Hence, the wall loading of a clear bulge increased with Re but
32
as well for a clear medium. With the placement of a coil, circulating velocities in the bulge
reduce, wall shear reduces, while pressure levels remain relatively unaffected. Instantaneous
Instantaneo
flow patterns confirm that the coil creates a bypass that stabilizes wall loading of the bulge.
Thus, the instantaneous variation in the flow properties from phase ‘a’ to ‘b’ is analogous to a
change in Reynolds number in the time
time-averaged and RMS data.
The stream traces of the flow field within the aneurysm are depicted at selected phases of
the cycle in Figure 4.7.. The first and second column shows the stream traces
trace for Reynolds
number 500, while the other two are for Reynolds number 1500. The flow is observed to be
recirculated within the aneurysm in the case of clear medium for both the Reynolds number
thereby increasing the pressure and stress on the walls. Howev
However,
er, there is no recirculation pattern
Re=500 Re=1500
Velocity
after coil insertion and the flow is redirected towards the main artery from the aneurysm region.
After coil insertion, the streamlines tend to redirect towards the main tube and the density of
stream traces within the aneurysm also decreases.
Apart from overall trends in the bulge, it is of interest to know if the loading is particularly high
at selected locations. For this purpose, instantaneous wall shear and pressure are provided in
Figure 4.9 at points A, B, and C of the bulge (shown in Figure 4.8). Reynolds numbers of 500
and 1500 are shown. Wall shear stresses at points A, B, and C are distinctly different for a bulge
with clear fluid. In the presence of a coil, the flow pattern is regularized and the differences are
smaller. Wall shear stresses increase substantially with Reynolds number in a
Figure 4.7: Stream traces at the mid-plane in front view at selected phases of the cycle as shown
in Figure 2.2 for Reynolds number, Re=500 and Re=1500 and Womersley number, Wo=7.82.
First row represents the clear medium, second row is porous medium with Newtonian model,
third row being porous medium with non-Newtonian model.
35
clear bulge but the increase is smaller in the bulge with a coil. For a clear bulge, point A shows
the largest wall shear and is expected to be the point of highest wall loading. Point B shows
lower wall shear over a cycle and is also a point of interest. These differences are broadly
submerged when a coil is used since the flow in the bulge is homogenized.
The pressure variation, plotted in the Figure 4.9, shows that the three points A, B and C
are largely indistinguishable in terms of pressure and consistently reveal the effect of Re. Thus,
pressure is spatially uniform in the bulge at all time instants. Pressure increases with Re and is
slightly lowered when the coil is placed within the aneurysm. A non-Newtonian fluid treatment
shows a slight augmentation in pressure because of the shear-thinning behavior of the fluid. As
shown in Figure 4.9, the presence of a porous medium inside the bulge softens the flow pulsation
and prohibits the inflow pulsation to reach the wall. A comparison between Figure 4.9 (second
and third row) shows that Newtonian fluid suppresses flow pulsation more efficiently than the
shear-thinning fluid. Suppression of flow pulsation indicates the dominance of diffusive
momentum-transport promoted by the presence of the porous medium. The shear thinning
rheology of blood, on the other hand, reduces viscosity in the regions of high velocity gradients
promoting the convective and Darcian transport mechanisms. Dominance of such transport
mechanisms preserves the inflow pulsation, as shown in Figure 4.9 (third row). The shear
thinning behavior of blood dominates the near-wall region of the blood vessel, including points
A, B, C. These points, therefore, experience the rheology-induced low viscosity effect and
transport the inflow pulsation up to the wall, Figure 4.9 (third row).
Figure 4.8: Schematic model of the bulge region. Points A, B and C indicate the points at which
the pressure and wall shear stress variation
36
Figure 4.9: Wall shear stress and pressure variation with time at selected points A, B and
a C in the
bulge region for clear media (first row), porous media (second row) and porous media with non-
non
Newtonian flow (third row) at Re = 500 (left column) and Re = 1500 (right column)
37
Figure 4.10 depicts the comparison between wall vorticity and wall shear rate (Eq, (2.12))
calculated for the time-averaged distribution in clear and porous media. The difference obtained
in of considerable value. The latter one is used for calculating the wall shear stress on the
aneurysm walls. Figure 4.11 shows the variation of normalized wall viscosity with reference to
the Carreau-Yashuda model used for the shear-thinning characteristics of fluid. The viscosity is
(i) (i)
(ii) (ii)
Figure 4.10: Wall Vorticity and wall shear rate contours for the time-averaged distribution at Re
= 500 for (i) Clear medium and (ii) porous medium.
38
normalized with the reference viscosity given in Table 2.1. The contours show that the variation
in viscosity with respect to Carreau-Yashuda model decreases with increase in Reynolds number.
The difference decreases on moving towards the wall region from the middle of the aneurysm.
This shear-thinning characteristic can be seen near the wall region where the shear stress is
critical near the walls while the viscosity decreases moving towards the walls.
Re = 500
Re = 1500
Figure 4.11: Wall viscosity variation on the wall (first column) and in the mid-plane section
(second column) of the time-averaged distribution in porous media with non-Newtonian fluid
model for Re 500 and 1500.
CHAPTER 5
PULSATILE FLOW IN A PATIENT
PATIENT-SPECIFIC
SPECIFIC CEREBRAL
ANEURYSM
plane A plane B
Figure 5.1: Planes across the geometry at which the velocity contours are shown.
39
40
Continuity of velocity is imposed at the clear and porous medium interface for coil
insertion within the aneurysm (Figure
Figure 5.2).
). The points A, B, C and D indicate the points at which
the wall pressure and wall shear stress are discussed later in this section.
Figure 5.2: Porous medium modeling of coil embolization. A, B, C and D are the points at which
the flow variables are discussed later
later.
The results are presented for the rest and exercise flow conditions of blood as mentioned in
Figure 2.4 and Figure 2.5.. A constant pressure condition is specified at the outlet for these
simulation conditions.
Figure 5.3 and Figure 55.4 show the time-averaged
averaged and RMS deviation contours of
velocity, wall pressure and wall shear stress in clear medium. The first two columns represent the
contours for the rest condition while the other two columns represent the contours generated for
the exercise condition. Without any coil insertion significant flow can be seen inside the
aneurysm for both the conditions. The magnitude of the field variables are slightly escalated for
the flow under exercise conditions than the rest conditions. The ve
velocity
locity difference is not visible
after averaging the values for three cycles but the difference can be seen at selective instants of
time. The difference in wall pressure for the rest and exercise conditions is clearly visible in the
contours. This shows that
hat the blood flows faster during exercise than the rest condition and the
pressure and wall shear stress magnitudes exerted on the walls are slightly increased in case of
exercise, thus increasing the chances of rupture.
The deviation shown for the exerc
exercise
ise condition is also higher for wall pressure and wall
shear stress. This show the fluctuations in pressure and wall shear stress during the exercise
41
condition are of higher order than in rest condition. The fluctuation observed in wall pressure is
almost as of the same order as its time-averaged value.
After considering the flow in clear media, the effectiveness of the coil embolization
technique can be gaged from the contours mentioned in Figure 5.5 and Figure 5.6. The contours
show the results obtained after inserting coils inside the aneurysm region, with Newtonian fluid
model. The bulge region acts as porous medium after coil insertion. In case of clear bulge,
certain fluid velocity is attained within the region. After placing a coil in the region, the flows
seems to be bypassing the bulge region, thereby reducing the velocities by a factor of five,
suppressing the flow in the bulge region. Almost similar results are obtained with the use of non-
Newtonian model (Figure 5.7 and Figure 5.8). The pressure contours are .however, mostly
unchanged with or without any coil insertion. It can be said that the bulge acts as a zone of
separation for pressure and so spatially uniform pressure is obtained. The slow flow through the
bulge with coils is unidirectional and thus helps in arising a uniform pressure in the bulge region.
The wall shear stress in the bulge region reduces by a factor of three and also nullifies any
deviation in the region. The figures thus far show that the coils inside the aneurysm are effective
during both the rest and exercise condition.
The trends generated in case of Figure 5.7 and Figure 5.8 are similar to the trends in
Newtonian model. The similarity in trends can be judged in the following way (a) flow is
suppressed inside the aneurysm after coil insertion (b) the wall shear stress is reduced
significantly in the region (c) the wall pressure is mostly unaffected by the presence of coil inside
the bulge region, which arises due to the small velocity prevailing in the bulge region after coil
insertion.
Figure 5.9 to Figure 5.14 present instantaneous flow data at selective phases ‘a’ and ‘b’
defined in Figure 2.4 and Figure 2.5. The phases ‘a’ and ‘b’ correspond to the peak phase and
one of the prominent minimums. Figure 5.9 and Figure 5.12 are for clear medium, 5.10 and 5.13
are for porous medium with Newtonian model, and 5.11 and 5.14 include porous medium (coil in
the bulge region) with the fluid being non-Newtonian. The peak phases for rest and exercise
condition are at different time instants as shown in their waveforms and compared accordingly.
Figures 5.9 to 5.11 show the contours under rest condition while Figures 5.12 to 5.14 under
exercise condition.
42
Figure 5.3: Time-averaged distribution of the field variables in clear media with rest and exercise
condition: first row, absolute velocity at the planes shown in Figure 5.1; second row, wall
pressure and third row, wall shear stress.
43
Figure 5.4: RMS deviation of the field variables from their time-averaged values in clear media
with rest and exercise condition: first row, absolute velocity at the planes shown in Figure 5.1;
second row, wall pressure and third row, wall shear stress.
44
Figure 5.5: Time-averaged distribution of the field variables in porous media with Newtonian
model under rest and exercise condition: first row, absolute velocity at the planes shown in
Figure 5.1; second row, wall pressure and third row, wall shear stress.
45
Figure 5.6: RMS deviation of the field variables from their time-averaged values in porous media
with Newtonian model under rest and exercise condition: first row, absolute velocity at the
planes shown in Figure 5.1; second row, wall pressure and third row, wall shear stress.
46
Figure 5.7: Time-averaged distribution of the field variables in porous media with non-
Newtonian model under rest and exercise condition: first row, absolute velocity at the planes
shown in Figure 5.1; second row, wall pressure and third row, wall shear stress.
47
Figure 5.8: RMS deviation of the field variables from their time-averaged values in porous media
with non-Newtonian model under rest and exercise condition: first row, absolute velocity at the
planes shown in Figure 5.1; second row, wall pressure and third row, wall shear stress.
48
The flow is stronger in phase ‘a’ than phase ‘b’ and the velocity, pressure and wall shear
stress magnitudes are also greater in phase ‘a’ for simulation results in clear medium. Pressure
within the aneurysm at phase ‘a’ is approximately three times higher than that of phase ‘b’. After
the placement of a coil in Figure 5.10, the velocities reduce upto seven times inside the bulge
region also reducing the wall shear stress while the pressure levels remain unaffected as observed
in the previous contours. The instantaneous variation in the flow properties from phase ‘a’ to ‘b’
defines the range of flow magnitudes for which the coil embolization treatment stabilizes the
aneurysm region, or in other words the instantaneous variation in the flow properties from phase
‘a’ to ‘b’ is analogous to a change in Reynolds number in the time-averaged and RMS data.
Figure 5.11 shows the results obtained with the non-Newtonian fluid model. The trends obtained
are quite similar to those in Figure 5.10.
Figure 5.12, 5.13 and 5.14 show the flow being simulated in exercise condition at
selective instants of time. The contours obtained depict the same trend observed under rest
condition. The magnitudes of field variables are slightly heightened in case of clear bulge but
stabilized after the placement of coils inside the bulge region.
It is interesting to know if there are certain locations in the aneurysm region which are
more prone to rupture than the rest of the aneurysm. The chances of rupture starting from these
locations are very high compared to the other parts of the aneurysm. Analyzing the pressure and
shear stress on the walls of these region is more important. Figure 5.15 and 5.16 depict the
variations of wall shear stress and wall pressure at certain locations A, B, C and D within the
aneurysm region (Figure 5.2) for one physiological waveform cycle. Figure 5.15 shows the
variation at locations A, B, C and D under rest condition while Figure 5.16 shows the same under
exercise condition.
In Figure 5.15, the pressure variation is nearly same for all the locations after coil
insertion. However, in clear bulge the pressure at location ‘C’ is slightly more than the rest of the
locations, but after placing the coils, the pressure distribution seems spatially uniform at all time
instants irrespective of the fluid being Newtonian or non-Newtonian. There is no major change
in magnitude of the pressure region after coil insertion. The pressure variation at different
locations is hardly distinguishable after coil insertion. A non-Newtonian fluid treatment shows a
slight augmentation in pressure because of the shear-thinning behavior of the fluid.
49
Rest condition
Phase a Phase b
Figure 5.9: Instantaneous distribution of the field variables in clear media at selected phases of
the cycle (shown in Figure 2.4) with resting condition: first row, absolute velocity at the planes
shown in Figure 5.1; second row, wall pressure and third row, wall shear stress.
50
Rest condition
Phase a Phase b
Figure 5.10: Instantaneous distribution of the field variables in porous media with Newtonian
model at selected phases of the cycle (shown in Figure 2.4) with resting condition: first row,
absolute velocity at the planes shown in Figure 5.1; second row, wall pressure and third row,
wall shear stress.
51
Rest condition
Phase a Phase b
Figure 5.11: Instantaneous distribution of the field variables in porous media with non-
Newtonian model at selected phases of the cycle (shown in Figure 2.4) with resting condition:
first row, absolute velocity at the planes shown in Figure 5.1; second row, wall pressure and third
row, wall shear stress.
52
Exercise condition
Phase a Phase b
Figure 5.12: Instantaneous distribution of the field variables in clear media at selected phases of
the cycle (shown in Figure 2.5) with exercise condition: first row, absolute velocity at the planes
shown in Figure 5.1; second row, wall pressure and third row, wall shear stress.
53
Exercise condition
Phase a Phase b
Figure 5.13: Instantaneous distribution of the field variables in porous media with Newtonian
model at selected phases of the cycle (shown in Figure 2.5) with exercise condition: first row,
absolute velocity at the planes shown in Figure 5.1; second row, wall pressure and third row,
wall shear stress.
54
Exercise condition
Phase a Phase b
Figure 5.14: Instantaneous distribution of the field variables in porous media with non-
Newtonian model at selected phases of the cycle (shown in Figure 2.5) with exercise condition:
first row, absolute velocity at the planes shown in Figure 5.1; second row, wall pressure and third
row, wall shear stress.
55
Figure 5.15: Wall shear stress and pressure variation with time under resting condition at
selected points A, B, C and D in the bulge region for clear media (first row), porous media
(second row) and porous media with non-Newtonian flow (third row).
56
Figure 5.16: Wall shear stress and pressure variation with time under exercise condition at
selected points A, B, C and D in the bulge region for clear media (first row), porous media
(second row) and porous media with non-Newtonian flow (third row).
57
In case of exercise condition also, the pressure variation is quite similar to what we
observed in rest condition plots. The cycle time is reduced in exercise condition which enables
the pressure to achieve peak values much faster than in rest condition. The magnitudes remain
unaffected and decreases slightly after coil insertion. The increased frequency in the exercise
condition increases the chances of aneurysm getting ruptured.
Wall shear stress at location C (Figure 5.15) is much different than the other locations A,
B, and D. The magnitude of wall shear stress at ‘C’ is two times the value at other regions. In the
presence of a coil, the flow variation is regularized and the differences are of very small order.
With exercise condition (Figure 5.16), the wall shear stress increase substantially in a clear bulge
than the rest condition. The magnitudes of wall shear stress increase with the increase in flow
magnitudes. The variation at ‘C’ is distinctively different than the other locations with both the
rest and exercise conditions. However, the difference is very much small after placing the coils.
As observed in the figures, Point ‘C’ shows the largest wall shear in the two figures followed by
point ‘B’, then point ‘A’ and the lowest wall shear is observed at point ‘D’. These differences are
broadly submerged after placing the coils and a uniform distribution nature is seen in the
aneurysm since the flow is also homogenized in the region. The increase in frequency due to
exercise condition is not of much concern in terms of wall shear stress as the stress is reduced to
zero order values.
The presence of coils inside the bulge region softens the flow pulsation inside the
aneurysm and restricts these pulsations to reach the wall of the aneurysm. From Figure 5.16
(second and third row), it is evident that Newtonian fluid is much more efficient in suppressing
flow than the non-Newtonian fluid. The shear-thinning rheology of blood reduces viscosity in the
regions of high velocity gradients in favor of Darcian transport mechanisms.
CHAPTER 6
CONCLUSIONS AND FUTURE WORK
The work done in this thesis is summarized in this chapter and possible future extension of the
work is also suggested.
In the present study coil embolization modeling of saccular cerebral aneurysm is reported. The
coils are modeled as porous medium and represented using Forchheimer-Brinkman extended
Darcy model. A realistic velocity waveform is applied on the inflow plane. The unsteady form of
the governing equations is solved in three dimensions by a finite volume method. Results
obtained in the present work show that the bulge is a region of flow recirculation. In the presence
of a coil, the bulge is bypassed and flow intensity in the bulge is greatly reduced. Consequently,
wall shear stress is also diminished. The pressure distribution in the bulge remains spatially
uniform with and without the coil. The effect of treating blood as non-Newtonian is to increase
wall shear and pressure levels, though the differences with respect to the Newtonian model are
only marginal. The particulate nature of blood is not expected to alter the results obtained since
the suspended RBCs are much smaller than the pore dimensions of the coil.
The sizes of the blood cells in the solid phase vary from 5 µm to 20 µm in case of RBC to
WBC, respectively. The dynamic clustering of red blood cells leads up to a dimension of 20-30
µm. The aneurysm region in the model used is quite large, with a diameter of 30 mm. The
packing density by volume for the coils placed will be 26.5 % (porosity 0.735). The pore length
scale for this volume can now be estimated as 10 mm. Since the pore-scale is much larger than
the blood particles, one can make an approximation that the coil will not selectively block the
passage of particulates. The possibility of the coil porosity diminishing with time is also
minimal.
58
59
6.1.2. Pulsatile flow in a patient-specific cerebral aneurysm under rest and exercise
condition
need to be taken into account. The governing equations of blood and artery wall are solved
simultaneously. The complexity of the problem leads to a large computational effort. In this
direction, the article of Takizawa et al. (2011) could be referred.
One can also continue to examine the shear stress and pressure levels in the solid arterial
walls. The solutions of the present study can be taken as the initial condition for this work. The
governing equations would require a complex numerical solver for the solid mechanics
equations. It will help in understanding the impact of fluid loading on the walls more
extensively.
APPENDIX
DERIVATION OF WOMERSLEY ANALYTICAL SOLUTION
∂w ∂p 1 ∂ 2 w 1 ∂w
= −ϕ + + (C.4)
∂t ∂z Re ∂r 2 r ∂r
As the flow is oscillatory we could represent the velocity and pressure gradient in following
form:
∞ iω t
w= ∑ W ( r )e n
n
n = −∞ (C.5)
dp ∞ iω t ∞
dz
=− ∑ G e n = ∑
n = −∞
n
n = −∞
p
z n ( )
Here, ωn = nω . Now substituting Equation (C.5) in Equation (C.4) and equating both sides for
each harmonic will give following equation:
1 d 2Wn 1 dWn
iωnWn = Gn + + (C.6)
Re dr 2 r dr
d 2Wn dWn
r2 +r + r 2 ( −i Re ωnWn + Re Gn ) = 0 (C.7)
dr 2 dr
W (r ) = W (1 − 4r 2 )
0 c0
G (C.9)
W = 0
c 0 16 Re
Now for n ≠ 0 consider,
−i Re ωnWn + Re Gn = U n (C.10)
61
62
d 2U n dU n
r2 +r + r 2 ( −i Re ωn ) U n = 0 (C.11)
dr 2 dr
Now consider
r 2 ( −i Re ω n ) = x 2 (C.12)
2 d 2U n dU n
x +x + x 2U n = 0 (C.13)
dx 2 dx
This is a Bessel’s differential equation, and the solution would involve Bessel functions of zeroth
order and complex arguments. Thus
U n = C1 J 0 ( x) + C2Y0 ( x) (C.14)
Now substituting Equation (C.10) and (C.12) in (C.14):
1
Wn = − (
C J (r −i Re ωn ) + C2Y0 (r −i Re ωn ) − Re Gn
i Re ωn 1 0
) (C.15)
Re Gn
Applying boundary condition (C.8) it can be shown that: C1 = and C2 = 0
J 0 (0.5 −i Re ωn )
πW ∞ W I (0.5 A ) iω t
1 − 1 1 n e n
Q (t ) =
2
c0 + 2π
∑
n=−∞
cn
8 2 A I (0.5 A )
(C.18)
n≠ 0 1 − 1 n 0 n
I (0.5 A )
0 n
Now average velocity is represented in the following form:
∞ iω t
wavg (t ) = ∑ Wavgn e n (C.19)
n = −∞
Again
π ∞ iω t
Q (t ) = π (0.5)2 wavg (t ) = ∑
4 n =−∞
Wavgn e n (C.20)
8W I (0.5 A )
Wavgn = cn 1− 1 1 n for n ≠ 0 (C.21)
8 2 A I (0.5 A )
1 − 1 n 0 n
I (0.5 A )
0 n
W
Wavg 0 = c0 for n = 0 (C.22)
2
With the help of Equation (C.21), (C.22) and (C.19) the average velocity is calculated.
1
W = B 1−
cn n I (R A )
0 n
ρω G
A =i n, B = n
n µ n i ρω
n
In dimensional form Equation (C.21) is as:
2W I (R A )
Wavgn = cn 1 − 1 1 n (C.24)
2 2 A I (R A )
1 − 1 n 0 n
I (R A )
0 n
where, R is the radius of tube.
∂w ∂p 1 ∂ 2 w 1 ∂w ϕ
= −ϕ + + − w (C.25)
∂t ∂z Re ∂r 2 r ∂r Re Da
Similar to the Womersley solution, consider:
∞ iω t
w= ∑ W ( r )e n
n
n = −∞ (C.26)
dp ∞ iω t ∞
dz
=− ∑ G e n = ∑
n = −∞
n
n = −∞
p
z n ( )
Now substituting Equation (C.26) in Equation (C.25) and equating both sides for each harmonic
will give following equation:
d 2Wn
+ r 2 − i Re ωn +
dWn ϕ
r2 +r Wn + ϕ Re Gn = 0 (C.27)
dr 2 dr Da
The boundary conditions are:
Wn = finite at r = 0 and Wn = 0 at r = 0.5 (C.28)
Now consider
65
− i Re ωn +
ϕ
Wn + ϕ Re Gn = U n (C.29)
Da
Therefore, Equation (C.27) is written as:
d 2U n ϕ
+ r 2 − i Re ωn +
dU n
r2 +r Un = 0 (C.30)
dr 2 dr Da
Now consider
ϕ 2
r 2 − i Re ωn + =x (C.31)
Da
Therefore Equation (C.30) could be written as:
d 2U n dU n
x2 +x + x 2U n = 0 (C.32)
dx 2 dx
This is a Bessel’s differential equation, and the solution would involve Bessel functions of zeroth
order and complex arguments. Thus
U n = C1 J 0 ( x) + C2Y0 ( x) (C.33)
Substituting Equation (C.29) and (C.31) in (C.33):
1
Wn = −
An 1 0
(
C J (ir An ) + C2Y0 (ir An ) − ϕ Re Gn ) (C.34)
ϕ
where, A = iReω + .
n n Da
ϕ Re Gn
Applying boundary condition (C.28) we get: C1 = and C2 = 0 .
J 0 (0.5i An )
∞ W I (0.5 A ) iω t
Q(t ) = 2π ∑ cn 1 − 1 1 n e n (C.37)
n =−∞ 8 2 A I (0.5 A )
1 − 1 n 0 n
I (0.5 A )
0 n
Now average velocity is represented in the following form:
∞ iω t
wavg (t ) = ∑ Wavgn e n (C.38)
n = −∞
Again
π ∞ iω t
Q(t ) = π (0.5)2 wavg (t ) = ∑
4 n=−∞
Wavgn e n (C.39)
Therefore,
8W I (0.5 A )
Wavgn = cn 1− 1 1 n (C.40)
8 2 A I (0.5 A )
1 − 1 n 0 n
I (0.5 A )
0 n
With help of Equation (C.38) and (C.40) the average velocity is calculated.
I (r A )
0 n 1
W (r ) = W 1− / 1−
n cn I ( R A ) I ( R A )
0 n 0 n
1
W = B 1− (C.41)
cn n I (R A )
0 n
ρω ϕG
A =i n+ϕ , B = n
n µ K n ϕµ
i ρω +
n K
In dimensional form Equation (C.21) is written as:
2W I (R A )
Wavgn = cn 1 − 1 1 n (C.42)
2 2 A I (R A )
1 − 1 n 0 n
I (R A )
0 n
where, R is the radius of tube.
REFERENCES
Boryczko, K., W. Dzwinel and D. A. Yuen (2003). "Dynamical clustering of red blood cells in
capillary vessels." Journal of molecular modeling 9(1): 16-33.
Bowker, T., P. Watton, P. Summers, J. Byrne and Y. Ventikos (2010). "Rest versus exercise
hemodynamics for middle cerebral artery aneurysms: a computational study." American Journal
of Neuroradiology 31(2): 317-323.
Byun, H. S. and K. Rhee (2004). "CFD modeling of blood flow following coil embolization of
aneurysms." Medical Engineering and Physics 26: 755-761.
Canton, G., D. I. Levy and J. C. Lasheras (2005). "Changes in the intra-aneurysmal pressure due
to hydrocoil embolization." AJNR American Journal of Neuroradiology 26: 904-907.
Dash, R. K., K. N. Meheta and G. Jayaraman (1996). "Casson fluid flow in a pipe filled with
homogenious porous medium." International Journal of Engineering Science 34(10): 1145-1156.
Egelhoff, C., R. Budwig, D. Elger, T. Khraishi and K. Johansen (1999). "Model studies of the
flow in abdominal aortic aneurysms during resting and exercise conditions." Journal of
biomechanics 32(12): 1319-1329.
El-Shahed, M. (2003). "Pulsatile flow of blood through a stenosed porous medium under
periodic body acceleration." Journal of Applied Mathematics and Computation 138(2-3): 479 –
488.
Gobin, Y. P., J. L. Counord, P. Flaud and J. Duffaux (1994). "In vitro study of haemodynamics
in a giant saccular aneurysm model: influence of flow dynamics in the parent vessel and effects
of coil embolisation." Neuroradiology 36(7): 530-536.
Groden, C., J. Laudan, S. Gatchell and H. Zeumer (2001). "Three-dimensional pulsatile flow
simulation before and after endovascular coil embolization of a terminal cerebral aneurysm."
Journal of Cerebral Blood Flow & Metabolism 21: 1464-1471.
Haji-Sheikh, A. and K. Vafai (2004). "Analysis of flow and heat transfer in porous media
imbedded inside various-shaped ducts." International Journal of Heat and Mass Transfer 47:
1889–1905.
68
69
Ide, K. and N. H. Secher (2000). "Cerebral blood flow and metabolism during exercise."
Progress in neurobiology 61(4): 397-414.
Kaviany, M. (1999). Principles of Heat Transfer in Porous Media. New York, Springer.
Khanafer, K. M., P. Gadhoke, R. Berguer and L. B. Joseph (2006). "Modeling pulsatile flow in
aortic aneurysms: Effect of non-Newtonian properties of blood." Biorheology, IOS Press 43:
661-679.
Liou, T. M., S. M. Liou and K. L. Chu (2004). "Intra-aneurysmal flow with helix and mesh stent
placement across side-wall aneurysm pore of a straight parent vessel." ASME Journal of
Biomechanical Engineering 126: 36-43.
Morosuk, T. V. (2005). "Entropy generation in conduits filled with porous medium totally and
partially." International Journal of Heat and Mass Transfer 48: 2548–2560.
Nield, D. A. and A. Bejan (2006). Convection in porous media, Springer-Verlag, New York
Ogulu, A. and E. Amos (2007). "Modeling pulsatile blood flow within a homogeneous porous
bed in the presence of a uniform magnetic field and time-dependent suction." International
Communication in heat and mass transfer, doi:10.1016/j.icheatmasstran sfer.2007.05.012.
Sazonov, I., S. Y. Yeo, R. L. Bevan, X. Xie, R. van Loon and P. Nithiarasu (2011). "Modelling
pipeline for subject‐specific arterial blood flow—A review." International Journal for Numerical
Methods in Biomedical Engineering 27(12): 1868-1910.
Sheard, G. J. (2009). "Flow dynamics and wall shear-stress variation in a fusiform aneurism."
Journal of Engg. Mathematics 64: 379-390.
70
Sikarwar, B. S., S. Khandekar and K. Muralidhar (2013). "Simulation of flow and heat transfer
in a liquid drop sliding underneath a hydrophobic surface." International Journal of Heat and
Mass Transfer 57(2): 786-811.
Vafai, K. (2011). Porous Media: Applications in Biological Systems and Biotechnology. Boca
Raton, FL 33487-2742, CRC Press, Taylor and Frnacis Group.
Vafai, K. and S. Kim (1989). "Forced convection in a channel filled with porous medium: an
exact solution." ASME Journal of Heat Transfer 111(4): 1103–1106.
Valencia, A., H. Morales, R. Rivera, E. Bravo and M. Galvez (2008). "Blood flow dynamics in
patient-specific cerebral aneurysm models: The relationship between wall shear stress and
aneurysm area index." Medical Engineering & Physics 30(3): 329-340.
Weir, B. (1987). Aneurysms affecting the nervous system, Williams and Williams.
Womersley, J. R. (1955). "Velocity Profiles of Oscillating Arterial Flow, with Some Calculation
of the Viscous Drag and the Reynolds Number." Journal of Physiology 128: 629-640.
Yu, S. C. M. and J. B. Zhao (1999). "A steady flow analysis on the stented and non-syented
sidewall aneurysm models." Medical Engineering and Physics 21: 133-141.
Zhang, X., Z. Yau, Y. Zhang and S. Xu (2007). "Experimental and computational studies on the
flow field in aortic aneurysms associated with deployment of AAA stent-graft." Acta Mechanica
Sinica 23: 495-501.