See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/238589397

An investigation of pulsatile blood flow in a bifurcation artery using a grid-

free method

Conference Paper · December 2006

CITATIONS READS

55 2,520

3 authors:

Matt Sinnott Paul W. Cleary

The Commonwealth Scientific and Industrial Research Organisation The Commonwealth Scientific and Industrial Research Organisation
116 PUBLICATIONS 2,532 CITATIONS 405 PUBLICATIONS 14,719 CITATIONS

SEE PROFILE SEE PROFILE

Mahesh Prakash
The Commonwealth Scientific and Industrial Research Organisation
98 PUBLICATIONS 2,126 CITATIONS

SEE PROFILE

All content following this page was uploaded by Paul W. Cleary on 17 December 2013.

The user has requested enhancement of the downloaded file.

 Fifth International Conference on CFD in the Process Industries
CSIRO, Melbourne, Australia
13-15 December 2006

AN INVESTIGATION OF PULSATILE BLOOD FLOW IN A BIFURCATION ARTERY

USING A GRID-FREE METHOD
1 1 1
Matthew SINNOTT , Paul W. CLEARY and Mahesh PRAKASH
1
CSIRO Minerals, Clayton, Victoria 3169, AUSTRALIA

ABSTRACT was maximal. Conversely, plaque build-up was reduced

where WSS and flow velocity were highest. The flow
CFD modelling is a powerful, but largely under-utilised mechanism driving plaque growth is still uncertain and the
tool for biomedical applications. In particular, it has great rate of growth could depend on steady WSS; the shear
potential for helping us better understand the mechanisms stress gradient; or even an oscillating stress field.
responsible for cardiovascular disease such as
artherosclerosis and thrombi formation. Flow behaviour in Computational modelling of flow in diseased arteries
blood vessels has been shown to depend strongly on using realistic geometries derived from Magnetic
features of the local geometry such as branching, bending, Resonance Imaging (MRI) is gaining favour as a tool for
and regions of flow constriction. The simplest blood flow understanding and predicting cardiovascular disease (see
models only consider steady flow. However within the Botnar et al. 2000, Prosi et al. 2004 and Marshall et al.
circulatory system, the periodic nature of the cardiac cycle 2004). This is because in vivo measurements of the flow
induces a pulsatile, unsteady flow. This periodic pressure field in an artery can be costly and are only possible for
perturbation is expected to have significant implications arteries that are easily accessible. Such measurements also
for localised flow velocities and stress distributions. provide very limited detail about the flow. Numerical
models of arterial flow may provide researchers with a
We propose here a pulsatile flow model using a grid-free platform for testing how flow conditions can be modified
method, Smoothed Particle Hydrodynamics. This method in order to change or prevent disease.
is well suited to transient flows within geometries of
complex shape. The arterial geometry used here is a real Current CFD methods used for modelling of arterial flows
carotid bifurcation derived from MRI. Rigid walls and rely on computing on a structured grid. Smoothed Particle
Newtonian flow are assumed. In this paper we compare Hydrodynamics (SPH) is a fully transient, Lagrangian
pulsatile and steady flow for this geometry. CFD solver and does not use a grid. Instead computations
are done on SPH ‘particles’ that travel with the flow
INTRODUCTION carrying local state information with them. Boundary
Blood flow is responsible for nutrient and waste transport geometries of almost arbitrary complexity may be included
within the closed-loop, cardiovascular network. Typically such as artery walls. Since the boundaries also contain
flow is laminar in healthy arteries, but the presence of SPH particles, it is possible to make these walls elastic and
abnormal flow conditions can promote the development of compliant with the fluid stresses exerted on the walls. The
cardiovascular disease such as artherosclerosis and advantage of SPH is that no re-meshing of the domain is
thrombus formation (excessive clotting). required to model the wall deformations. Furthermore
SPH particles carry local state information and can thus
The systemic flow is characterised predominantly by its track fluid history easily. This, combined with the ability
pulsatile nature and the many levels of branching of the to modify particles according to a rule base, means that we
vascular network. The cyclic pumping of the heart can grow thrombic protrusions into the flow and have
periodically evacuates blood out of the ventricle and into them break off. We can then, in principle, follow this
the aorta generating a pressure pulse that propagates dangerous clot downstream into the cerebral vasculature.
downstream through the arterial system. The combination This is very hard in a traditional grid-based method and
of unsteady flow with certain features of geometry such as would enable investigation of critical medical problems
branch junctions, curved sections and flow constrictions well beyond the scope of traditional solvers. So the
can lead to complicated secondary flows incorporating motivation for using SPH here, lies in the future
flow separation, recirculation, stagnation and regions of deployment of the advantages described above.
high or low wall shear stress.
In this study, we examine the differences between steady
Local haemodynamics particularly in and around stenoses and pulsatile flow conditions for the flow field within a
are believed to play an important role in cardiovascular real, diseased, carotid artery bifurcation with rigid walls
disease. The likelihood of artherosclerotic plaque using the SPH method. The aim is to understand whether
deposition is increased in regions of disturbed flow, but flow pulsatility is important for the development of
the relationship between fluid stresses and plaque growth cardiovascular disease in the carotid arteries; and as a
is still not fully understood. Sawchuk et al (1994) and prelude to looking at deformable walls interacting with the
Zarinset et al. (1983) both found that plaque thickness pulsatile flow.
depended on shear stress. Low wall shear stress (WSS)
and flow velocity were observed where plaque thickness

1
MODEL DESCRIPTION SPH formally resolves all length scales of the flow above
the resolution length; much like a Large Eddy simulation.
Smoothed Particle Hydrodynamics
However, there is no formal turbulence modelling since
The SPH methodology (Monaghan, 1992, 1994, 2005) the present SPH formulation does not have a sub-grid
consists of converting the partial differential equations scale model.
encountered in fluid flow into algebraic equations. The
interpolated value of a function A at any position r can be Carotid Arteries
expressed using SPH smoothing as:
Ab
A(r ) = ∑ mb W (r − rb , h ) (1) ECA
b ρb
Where mb and rb are the mass and density of particle b and
the sum is over all particles b within a radius 2h of r. Here
W(r,h) is a C2 spline based interpolation or smoothing
kernel with radius 2h that approximates the shape of a
ICA
Gaussian function. The gradient of the function A is given
by differentiating the interpolation equation (1) to give:
∇A(r ) = ∑ mb b ∇W (r − rb , h )
A (2) Figure 1: 3D model of real MRI derived arterial geometry
b ρ b
of a diseased, carotid artery bifurcation. On the right, the
Using these interpolation formulae and suitable finite upper daughter branch is the ECA and the lower branch is
difference approximations for second order derivatives, the ICA feeding blood to the brain.
one is able to convert parabolic partial differential The carotid arteries are located at the sides of the neck and
equations into ordinary differential equations for the supply blood to the face and brain. The common carotid
motion of the particles and the rates of change of their artery (CCA) branches into the internal carotid artery
properties. (ICA), which feeds the brain, and the external artery
(ECA), which transports blood to the muscles of the face
From Monaghan (1992), the most suitable form of the and head.
SPH continuity equation is:
dρ a Injury due to stroke arises from disease in the ICA
= ∑ mb (v a − v b ) • ∇Wab (3)
manifesting itself as an artherosclerotic stenosed flow
dt b

where ?a is the density of particle a with velocity va and constriction. Complex flows immediately downstream
mb is the mass of particle b. We denote the position vector from a stenosis are believed to be responsible for stresses
from particle b to particle a by rab=ra-rb and let leading to further plaque formation; or even plaque
rupture which can lead to clotting at the stenosis.
Wab = W(rab,h) be the interpolation kernel with smoothing
Furthermore ischemic stroke is a result of clot lodging in a
length h evaluated for the distance |rab|. This form of the
blood vessel in the brain. It originates from places such as
continuity equation if Galilean invariant (since the
the carotid artery, aorta and heart.
positions and velocities appear only as differences), has
good numerical conservation properties and is not affected
Boundary conditions are required for the arterial flow
by free surfaces or density discontinuities.
simulations in this study. Figure 1 shows a real geometry
of a diseased carotid artery derived from MRI (Beare et
The momentum equation can be written as:
al., 2006). Raw surface geometry was extracted from a
 Pb Pa   blocky, coarse resolution (1mm) MRI dataset. Spline
 2 + 2   (4)
ρ ρ based 2D profiles were fitted to the MRI voxel data and a
= g − ∑ m b  b a  ∇ W
dv a
dt b
 ξ 4 µ a µ b v ab rab 
a ab solid geometry was constructed in the CAD package,
−  Solidworks, by lofting between the 2D sections. A surface
 ρ a ρ b (µ a + µ b ) rab + η 
2 2
was extracted which represents the inner endothelial
where Pa and µa are pressure and fluid viscosity of particle surface of the artery, and was then meshed with a
a and vab = va–vb. Here ? is a factor associated with the commercial meshing package at a resolution of 0.5 mm.
viscous term (Cleary 1996), ? is a small parameter used to
smooth out the singularity at rab=0 and g is the gravity An inflow velocity boundary condition was applied at the
vector. entrance of the CCA (to the left of Figure 1). For the
steady case, a velocity of 0.15 m/s was chosen based on
Since the SPH method used here is a quasi-compressible clinical measurements for this artery. Pulsatile conditions
one needs to use an equation of state, giving the will be discussed in the next section. Outflow boundary
relationship between particle density and fluid pressure. A conditions were also required to prevent the artery from
suitable one is: draining and a constant pressure of 10 kPa was enforced at
 ρ  γ  the exit points of the ICA and ECA arteries (to the right of
(5)
P = P0   − 1 Figure 1). Simulations commenced and the arteries were
ρ
 0   filled from the inflow until the arteries were fully
where P0 is the magnitude of the pressure and ?0 is the pressurised as determined by measuring the pressures at
reference density. For water or blood we use ? = 7. This the outflows.
pressure is then used in the SPH momentum equation (3)
to give the particle motion.

2
Pulsatile Flow Profile RESULTS
Following each expulsion of blood from the heart and into Two cases have been simulated here for the same
the aorta, a pressure wave propagates downstream through geometry artery: a steady flow, and a pulsatile flow.
the arterial system initiating local changes in pressure. Passage of flow through the bifurcation
These transients can significantly alter the flow field To aid visualisation, SPH fluid particles were tagged
through the arteries. The flow field may also depend on periodically at the inflow to define a moving coherent
wall elasticity and blood rheology, but these are often band of fluid that could be tracked as it flowed through the
regarded as secondary effects relative to flow pulsatility. artery. This band is shown in Figure 3 for the steady case
For the purposes of this study we have assumed the at three different times, as it moves through the
boundaries are rigid and that blood behaves as a bifurcation. The SPH particles are shown here (and in the
Newtonian fluid. following figures) by a surface mesh surrounding them.

Figure 2: Pulsatile flow profile approximation used as a

boundary condition for SPH simulation of pulsatile flow.

The physiological profile of a real pressure pulse at the

entrance of the CCA depends on the cyclic pumping of the
heart, the distensibility of the walls, and upstream and
downstream pressures. During the diastolic phase of the
heart when flow from the ventricle to the aorta is shut off,
a non-zero flow through the arterial system is maintained
as the elastic walls of the aorta contract.

We include pulsatile flow in the SPH simulation by

imposing a time-varying velocity boundary condition at
the entrance of the artery. As a first approximation to a
physiological pulse, we use a sinusoidal profile. The shape
of this pulse is given in Figure 2. The period used here is
0.5 s corresponding to a rapid heartbeat of 120 beats per
minute (or 2 Hz), and the duration of the pulse is much
shorter (just 0.125 s). The peak velocity is 0.5 m/s and the Figure 3: A band of fluid coloured blue is shown passing
minimum is 0.1 m/s. A steady flow trailing the pulse through the bifurcation at three different times.
represents the cut-off in supply from the heart during
diastole. In a real pulse, there can be a degree of backflow Blood flows from the CCA up to the junction, before
during this time giving a small, negative pressure. splitting into two separate flows into the daughter
branches. Both the ICA and ECA branches are observed to
The Womersley number is a dimensionless parameter have flow constrictions (stenoses) just past the junction.
(similar to the Reynolds number) used specifically to
characterise pulsatile flow within an artery. It is the ratio The blue fluid band can be seen entering the branching
of inertial forces relative to viscous forces. junction at 1.6 s. An earlier band is visible in the top
2πfρ (6) artery to the right. There is an asymmetry already in the
α =R
µ band, with fluid moving forward into the carotid sinus in
where R is the radius of the artery, f is the pulsatile the bottom branch. The sinus is a characteristic enlarged
frequency, ? =1000 kg/m3 has been used for fluid density, region of the ICA just past the junction.
and µ=0.003 Pa s is the dynamic viscosity. For the pulse
described above in the common carotid artery of Figure 1 By 1.65 s, the flow is markedly split with the bulk of the
where the arterial diameter is 0.011 m, we obtain a fluid preferentially directed into the sinus. The narrow
Womersley number of 11, which is typical for this size artery and mild flow constriction in the top artery,
artery. For large a, the flow becomes momentum combined with the curved geometry leading up to the
dominated. branching point, show only a small volume of fluid

3
entering the ECA. In the bottom ICA branch, fluid is ICA stenosis. Dark blue corresponds to 0.1 m/s (and
accelerating into the stenosis. greater) directed upstream, and red is 2.0 m/s (and greater)
downstream.

Figure 4: 3D clipped section of bifurcation for steady

Figure 6: 3D clipped section of the bifurcation for steady
flow case showing the distribution of stream-wise
flow case showing the distribution of pressure. Dark blue
velocities. Dark blue corresponds to 0.1 m/s (and greater)
corresponds to 10 kPa (and lower), and red is 20 kPa (and
directed upstream, and red is 2.0 m/s (and greater)
greater).
downstream.

Figure 7: 3D clipped section of the bifurcation at 3

Figure 5: 3D clipped section of the bifurcation at 3
different times for the pulsed case showing the distribution
different times for the pulsed case showing the distribution
of stream-wise velocities as the pulse travels through the

4
of pressure. Dark blue corresponds to 10 kPa (and lower), The velocity and pressure histories in the ICA stenosis and
and red is 20 kPa (and greater). near the inflow are reported here for the pulsatile flow
case only. The velocity profiles near the inflow and at the
stenosis are shown in Figure 8. The data is given for one
period of the pulse. The velocity profiles are almost
The flow through the ICA stenosis is well underway by
identical in shape, but the peak amplitude of the pulse in
1.67 s with an equal amount of fluid before and after the
the stenosis is about 3.5 times greater than at the inflow
flow constriction. Significant drag on the fluid from the
and the minimum velocity is about 50% greater.
sinus walls is observed, while the fluid in the central
region flows rapidly through the stenosis.

Steady Flow vs Pulsatile Flow

The steady flow, coloured by stream-wise velocity and
pressure respectively, is shown in Figures 4 and 6.
Following the ICA branch, flow enters an enlarged section
called the carotid sinus. Approaching the sinus the flow
converges, increasing the pressure on the upstream side of
the stenosis. Some backflow is observed on the walls of
the sinus (see Figure 4) in trying to push too much volume
through the stenosis. To maintain a constant volumetric
flow rate through either the ECA or ICA stenosis, the flow
necessarily accelerates as a narrow jet in the centre of the
stenosis. These high velocities (up to 0.6 m/s for steady
flow in the ICA constriction) generate a low pressure
region within the stenosis as can be seen in Figure 6. For
stenoses with near complete occlusion, a sufficient drop in
pressure can result in closing the artery.

There is a drop in pressure visible across the ICA (see

Figure 6) with more green fluid lying to the left of the
stenosis and more blue to the right. This drives the high
velocities through the stenosis. In the upper ECA branch,
pressures look to be equalised across the stenosis with the
downstream outflow pressure comparable to the pressure
upstream of the stenosis. Limited flow occurs through this
narrow artery and flow velocities appear to be very low
downstream where the artery widens. Less blood supply is
required for the face than for the brain; and hence the
narrower artery. Since velocities in the ECA stenosis are Figure 8: Velocity time series as measured for the
smaller than in the ICA, and the volume of fluid is also pulsatile flow case near the inflow [top] and inside the
less, any effects from secondary flows downstream from ICA stenosis [bottom].
the stenosis will be limited. Thus there is typically less
medical concern for disease build-up in this region. Pressure profiles are given in Figure 9. The pressure
profile near the inflow corresponds to the shape of the
We now consider the pulsatile case shown in Figures 5 velocity profile but with a small, constant offset in phase
and 7 coloured by velocity and pressure respectively. At for all pulses. The pressure leads the velocity by 0.04 s
1.5 s, following the last pulse, the pressure has partially (about 30 degrees).
equalised across the lower ICA stenosis and the velocity is
correspondingly reduced. As the next pulse reaches the Pressures within the stenosis are related to the inflow
stenosis (Figure 7 middle) pressures rise upstream, pressure but differ in profile. Near the inflow, the pulse
maximising the pressure difference across the stenosis. peak pressure is reported as 23 kPa, and following the
The flow accelerates (to around 2.5 m/s) through the pulse the mean background pressure is 16 kPa. Reduced
stenosis until the downstream side achieves a pressure pressures of 17 kPa (peak) and 12 kPa (background) are
equivalent to that for the pressure drop for the steady case found inside the stenosis as supported by the 3D flow
(Figure 7 bottom). The velocity then falls. As the pulse pictures. The artery walls induce a drag in the flow, which
passes through the stenosis, the upstream pressure is creates a drop in the baseline pressure in the stenosis of
relaxed and flow velocity decreases further until the about 25% relative to the inflow pressure. A non-trivial
downstream side once again pressurises. variation in pressure is observed in the stenosis for the
duration of the pulse and appears to follow the velocity
Sensor Readings gradient of the pulse. The pressure appears to be high
Pressure and velocity data is difficult to interpret while the velocity is increasing, falls when the peak
quantitatively from 3D flow pictures and so we make use velocity is reached, and remains constant at a lower
of virtual pressure sensors. A sensor was positioned in the pressure than the baseline pressure while flow through the
stenosis of the ICA and measured time series data such as stenosis decelerates. Once the pulse has passed pressures
velocities and pressures in the local flow field. then return to the baseline level.

5
 be important for understanding the local flow stresses
leading to the development of cardiovascular disease.

ACKNOWLEDGEMENTS
We gratefully acknowledge the support of Prof. Reutens’
group from the Department of Medicine at Monash
University in supplying the clinical geometry used in this
study.

REFERENCES
BEARE, R., PHAN, T. and REUTENS, D., (2005),
Department of Medicine, Monash University, Private
Communication.
BOTNAR, R., RAPPITSCH, G., SCHEIDEGGER,
M.B., LIEPSCH, D., PERKTOLD, K. and BOESIGER,
P., (2000), “Hemodynamics in the carotid artery
bifurcation: a comparison between numerical simulations
and in vitro MRI measurements”, J. Biomech., 33, 137-
144.
CLEARY, P., (1996), “New implementation of
viscosity: tests with Couette flows”, SPH Technical Note
8, CSIRO DMS, Technical Report DMS – C 96/32.
MARSHALL, I., ZHAO, S., HOSKINS, P.,
PAPATHANASOPOULOU, P. and YUN XU, X., (2004),
“MRI and CFD studies of pulsatile flow in healthy and
stenosed carotid bifurcation models”, J. Biomech., 37,
679-687.
MONAGHAN, J. J., (1992), “Smoothed Particle
Figure 9: Pressure time series as measured for the
Hydrodynamics”, Ann. Rev. Astron. Astrophys., 30, 543-
pulsatile flow case near the inflow [top] and inside the
574.
ICA stenosis [bottom].
MONAGHAN, J. J., (1994), “Simulating free surface
flows with SPH”, J. Comp. Phys., 110, 399-406.
CONCLUSION MONAGHAN, J. J., (2005), “Smoothed Particle
Hydrodynamics”, Rep. Prog. Phys., 68, 1703-1759.
We have studied the differences between steady and
PROSI, M., PERKTOLD, K., DING, Z. and
pulsatile flow behaviour in a stenosed carotid artery
FRIEDMAN, M.H., (2004), “Influence of curvature
bifurcation using SPH simulations with a model pulse
dynamics on pulsatile coronary artery flow in a realistic
profile. Velocities and pressures have been analysed for
bifurcation model”, J. Biomech., 37, 1767-1775.
the two cases using 3D flow pictures.
SAWCHUK, A.P., UNTHANK, J.L., DAVIS, T.E,
and DALSING, M.C., (1994), “A prospective, in vivo
Steady flow appears to be preferentially directed into the
study of the relationship between blood flow
internal carotid artery supplying blood to the brain. The
haemodynamics and artherosclerosis in a hyperlipidemic
external artery is narrow and only transports a relatively
small volume for the same assumed downstream pressure swine model”, J. Vasc. Surgery, 19, 58-64.
as the internal artery. The stenosis in the internal carotid WOOTTON, D.M. and KU, D.N., (1999), “Fluid
artery produces high velocities and low pressures with a mechanics of vascular systems, diseases, and thrombosis”,
well defined jet travelling from the centre of the sinus and Ann. Rev. Biomed. Eng., 1, 299-329.
through the centre of the stenosed region. Some drag is ZARINS, C.K., GIDDENS, D.P. and
observed for blood adjacent to the sinus walls and perhaps BHARADAVAJ, B.K., (1983), “Carotid bifurcation
even some backflow along the walls. artherosclerosis: quantification of plaque localization with
flow velocity profiles and wall shear stress”, Circ.
Pulsatile flow is responsible for an oscillation in pressure Research, 53, 502-514.
and velocity across the ICA stenosis. There is interplay
between rising pressures at the stenosis and pressure
equalisation across the stenosis that drives this oscillation.
Velocities in the stenosis are very high for the pulsatile
case and pressures are mildly reduced. The peak velocity
for the pulsatile case is about 3.5 times that for the steady
flow case but the peak pressure is reduced by 25%
compared to the steady case.

The transient flow field from the pulsatile case has thus
been shown to differ markedly from the steady case. The
changes in pressure and velocity across the stenosis will

View publication stats