Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

PULSATILE PRESSUREAND FLOW IN ARTERIALSTENOSES

SIMULATEDIN A MATHEMATICAL MODEL

Sven 0ivind Wile and Lars Walbe

ABSTRACT

This paper presents the detailed pulsatile pressure and flow isobars at16 different instances in time,each 15 degrees apart
velocity patterns inside an axis symmetric stenosis model with during a cardiac cycle. The calculated velocity field shows that
75% constriction. The pressure and velocities have been calcu- a vortex isdeveloped atthe wall distal to the stenosis as the vel-
lated by solving the Navier-Stokes equations by the finite ocity decreases from the peak systolic value. At the site of the
element method, the velocity profile in a straight tube caused by vortex, a local pressure minimum is found due to the conversion
a pulsating driving pressure has been calculated first and then of pressure to kinetic energy. When the flow is reversed, the rev-
used as a boundary condition for the stenosis calculations. ersal occurs first along the wall, thus forcing the vortex toward the
The results of the mathematical simulations of the stenosis model the centre of the tube. As the reverse flow velocity increases, a
have been obtained in terms of velocib vectors, streamlines and vortex is also developed at the proximal site of the stenosis.

INTRODUCTION
The steady flow situation in stenosis models has
previously been studied for different Reynolds
numbers and different constrictions of the cross-
section arear. The steady state problem is easier
to solve numerically but does not adequately des-
cribe what is happening with flow and pressure
patterns around constrictions of blood vessels in the
arterial tree.
Measurement of velocity profiles in blood vessels
has been carried out by means of pulsed ultrasound
Doppler flowmeters 2. However, the space resolu-
tion of this instrument is poor, especially along the
wall.
“=O au_.
an an-
Experimental investigations which have been carried
out on physical models have mainly been concerned
with measuring flow velocities by laser Doppler
flowmeter3, visualization of the streamlines, either av_, g=o
by contrast dye4 or hydrogen bubbles’. The large an-
amount of data which has to be collected in order
to map velocity profiles by the laser Doppler flow-
meter, especially for pulsatile flow, has complicated
these measurements. The visualization of streamline
P=Pv)
patterns by contrast dye injection into the moving
fluid has been successful for steady flow. When pul-
satile flow is studied, the problem of dye mixing
in the fluid complicates these investigations. Visua-
lization of the streamlines by hydrogen bubbles has
been more successful as the bubbles are dissolved in
the fluid after a short time.
au
z=o
Viscous, time dependent fluid flow is adequately v=O
described by the Navier-Stokes equations and the
continuity equation. The method which has been Figure 1. The finite element model of the circular
applied to solve these equations is the finite element straight tube. The boundary conditions u = 0, v = 0
method6. In view of the limitations of the experi- correspond to the wall, au/an = 0, v = 0 correspond
mental methods, mathematical models play a useful to the axis of symmetry, au/an = 0, au/an = 0, p = 0
role in the analysis of flow and pressure patterns and correspond to the outlet and au/an = 0, au/&a = 0,
Institute of Informatics and Physiology, University of Oslo,
p = p(t) correspond to the inlet. p(t) is given in
Norway Figure 2a.

0141-5425/81/010017-08 $02.00
0 1981 IPC Business Press J. Biomed. Engng. 1981, Vol. 3, January 17
Pressure and flow in stenoses: S.0. Wile and 1. Wake

cm6

a04

0.02

a V I
3

Time (I)
I

Time ($1
Figure 2. (a) The pressure gradient, which drives the flow in a straight circular tube as a function of time.
(b) The resulting pulse velocity profile at diff erent sites in the tube as a function of time. The pulse profile
at y = 0 (0) is at the axis of the tube, the profile at y = 0.5 (0) shows the velocities at a distance equal to half
the radius from the centre axis and the profile at y = 0.8 (X) at 0.8 times the radius away from the axis.

Figure 3. The velocity profiles in a straight circular tube as a response to the pressure gradient in Figure 2a.
The velocity profiles are shown at l/24 s interval.
\,
4
Ia . . . . _ _^__ . . . _
Pressure and flow in stenoses: S.0. Wille and 1. Wallae

may give useful additional information, particularly


where the geometries are complicated.

METHOD

The time dependent, axis symmetric Navier-Stokes


equations and the continuity equation which des-
cribes viscous, incompressible flow are given by:

+u *+u-=o
3X ar
th
(1)

1.ap au av
1 -__ v +jh+a2v
P ar
- + at+
i F l ar r2 ax2 c
1

+u g+, $0 (2)

V
+&+&=I) (3)
7 ax Time ( s)

where u is velocity in x (longitudinal) direction, v is Figure 5. The pressure gradient across the model as
velocity in the r (radial) direction,p is pressure, p is function of time.
density of the fluid and P the kinematic viscosity.
ward scheme is more stable and convergence is
By applying the finite element method in solving
obtained in many cases where the trapezoidal rule
these equations, the pressure and velocities are app-
leads to a divergent solution. In some cases it can
roximated by polynomials defined on triangles, the
also be advantageous to obtain a solution by the
pressure with linear and the velocities with quadra-
Euler backward scheme and then increase the accu-
tic polynomials. The non-linear equations are solved
racy by increasing 0 and use the trapezoidal rule in
by the Newton-Raphson method, The time deriva-
the final iterations. In a previous paper, the pulsa-
tives of the velocities are replaced by a finite diffe-
tile flow in a circular straight tube and the flow in
rence time step scheme:
a suddenly blocked tube has been investigated by
a4 _ this method’.
at - f.
In the present work the backward difference scheme
is also tried, which is given by
By introducing the e-method, the time derivative
takes the form:

where the double bar indicates the value two time-


Tile bars denote the values at the previous time. For steps backwards. This scheme is claimed’ to give the
0 = ?4 the integration is the trapezoidal rule and for same accuracy as the trapezoidal rule and to have
8 = 0 the Euler backward scheme is obtained7. equal stability as the Euler backward scheme. How-
The advantage of the trapezoidal integration scheme ever, this method has been applied in the present
is that higher accuracy is obtained than with the work without success. It gave divergent solutions
Euler backward scheme. However, the Euler back- where the Euler backward scheme converged.

Figure 4. The finite element model of the stenosis axis symmetric tube. The boundary condition u = 0,
u = 0 correspond to the wall, au/an = 0, v = 0 correspond to the symmetry axis, au/an = 0, %/an = 0
correspond to the outlet and v = 0, u = u(t) correspond to the inlet. u(t) is given in Figure 3.

I 0:--,-I t____ ,001 x1-1 9 I__ _~ .A


Pressure and flow in stenoses: S.O. Wille and 1. Wallee

Figure 6. The pressure and flow patterns, l/24 s after the start of a cardiac cycle. Only results on the upper
half of a longitudinal section through the tube are shown. The upper panel shows the velocity vector field,
the one below shows the streamlines of flow and the lower show the pressure isobars. The bottom curve
shows the pressure variation along the axis of symmetry.

Pulsatile flow in a straight circular tube with a In Figure 2b the response of the pressure gradient,
‘physiological’ pressure gradient the time-velocity waveforms, are shown for diffe-
rent distances from the axis of symmetry. The
The finite element model of the straight circular
detaih of the veIocity profiles at 16 instances in the
tube with boundary conditions is shown in Figure 1.
cycle each l/24 s or 15 degrees apart are shown in
The boundary conditions are zero velocities along
the wall and zero normal derivatives of the veloci-
Figure 3.
ties at both the inlet and outlet. Along the axis of
symmetry, the normal velocity and the normal deri- Pulsatile flow through a stenosis
vative of the tangential velocity are zero. The flow
The finite element model of the stenosis with boun-
in the tube is driven by the pressure gradient shown
dary conditions is shown in Figure 4. The boundary
in Figure 2a. The shape of this pressure gradient is
conditions at the walI are zero velocities. At the
chosen to obtain a velocity waveform with charac-
outlet the normal derivatives of the velocities are
teristics similar to physiological velocity pulses.
zero and along the axis of symmetry the normal
The radius of the tube is 5 x 10m3m and the kine- velocity and the normal derivative of the tangen-
matic viscosity is 4 x 10e6 m* s-l. The maximum tial velocity are zero. At the inlet the tangential
mean velocity during the cycle is 4.7 x lO‘*m s-l, velocity is zero whilst the normal velocities are
thus giving a maximum Reynolds number of 120. those of the velocity profiles obtained for pulsatile
By taking the angular frequency as 2n, the flow in a straight tube. These velocity profiles are
Womersley cx becomes 6.27. used as boundary conditions at the inlet at each
Pressure and flow in stenoses: S.0. Wile and 1. Wallee

instant in time where the pressure and velocity stream of the stenosis is seen in Figure 10, where
patterns in the stenosis model are calculated. With the fluid lamina along the wall are moving upstream
the velocity profiles in Figure 3 as boundary con- through the stenosis and thus forcing the vortex
ditions at the inlet, the pressure gradient across the towards the axis. The reverse flow is increasing from
stenosis model which is obtained is shown in 9/24 s to 12/24 s. At 13/24 s the velocity vector plot
Figure 5. Figures 6-l 1 show the velocity vectors, shows that another vortex has developed u;rstream
streamlines and pressure patterns at selected instants of the stenosis. At 14/24 s (Figure 11) this vortex
in time. has increased and it is also apparent in the stream-
line plot. At 15/24 s and 16/24 s all velocities decay
In the early part of systole (Figure 6) the flow is and at 16/24 s there are only two vortices with low
accelerated and no separation is seen downstream velocities left, one upstream and one downstream
of the constriction. The pressure gradient is slightly of the stenosis. The calculations were not carried
greater at the stenosis than in the upstream and out for subsequent points in time because the nume-
downstream part of the model. Later in the cycle, rical methods produced divergent solutions.
at 3124 s, a vortex formation has developed down-
stream of the stenosis and the relative pressure drop
DISCUSSION
across the stenosis has increased. The size of the
vortex is increased during the following time steps, Pressure and velocity patterns have been calculated
4/24 s, 5/24 s, 6/24 s and 7/24 s, (Figure 7,8) and by the finite element method both in a circular
a local pressure minimum develops at the level of straight tube and in an axis symmetric constricted
the vortex. At 8/24 s (Figure 9), the direction of tube. The results for the straight tube are similar to
flow is reversed along the wall upstream of the those obtained analytically by Womersley’. The
stenosis. The influence of the reversed flow down- two situations cannot be compared exactly because

Figure 7. The pressure and flow patterns at 4124 s.


Pressure and flow in stenoses: S.O. Wile and L. Wallee

Figure 8. The pressure and flow patterns at 6124 s.

Womersley used a slightly different pressure gra- The velocity profiles used as boundary conditions
dient. The results show that the reversal of flow in for the stenosis at the inlet are the corresponding
the diastolic phase of the cycle first occurs at the velocity profiles in a long straight tube. However,
vessel wall and that the peak pressure occurs before the feedback from the stenosis may influence the
the peak velocity, that is: the pressure leads the velocity profiles at the inlet. The correct velocity
velocity. profiles at the inlet are probably slightly different
from those in a straight tube. More correct velocity
In the model of the arterial stenosis, a vortex forma- profiles are not known, either from experimental or
tion has been shown to appear downstream of the theoretical work, and the main patterns of the
constriction in the systolic and upstream of the results are believed not to be affected by small
constriction in the diastolic part of the cycle. As the deviations from our assumptions.
flow is reversed in diastole, the vortex is forced
With the present method, it is not possible to obtain
towards the centre of the vessel whilst the reversed
convergent solutions towards the end of the cardiac
velocities occur in the area close to the wall. At the
level of the vortex a low pressure zone is developed. cycle. The fact that the downstream vortex grows
out through the outlet of the tube, thus violating
the outlet boundary conditions, may be the reason
In the present simulation, the walls of the stenosis
for the divergent solution.
model are assumed to be stiff. This is not the case
in actual blood vessels, vhich have visco-elastic pro- In the last part of the cardiac cycle, where it was
perties with varying diameter during the cardiac not possible to obtain convergent solutions, we
cycle. However, the relative change in diameter is in believe that the velocities will decay and that the
the order of 10%” and it is therefore believed that effects of small movements in late diastole will have
the size of the error introduced by having a fixed negligible influence on the next cycle. At least, if
diameter is small. we use the last obtained so!ution (16/24 s) as the
Pressure and flow in stenoses: 50. Wile ond 1. Wallae

Figure 9. The pressure and jlow patterns at 8/24 s.

:-====-

Figure 10. The pressure and flow patterns at IO/24 s.


Pressure and Now in stenoses: S.O. Wile and 1. Walloe

Figure 11. The pressure and flo w patterns at 14124 s.

solution at the end of diastole, the effects of the 2 Peronneau, P., Hinglais, J., Xhaard, M., DeIouche, P.
two vorticies disappeared after two time steps. and Phillipo, J. The effects of curvature and stenosis on
pulsatile flow in vivo and in vitro. in Cardiovascular
Azuma and Fukushima’ investigated the velocity Application of Ultrasound, (Ed. Renemann, R.S.),
patterns downstream of a stenosis in a glass model. North Holland Publishing Co, Amsterdam, 1974, 203-
Their constriction ratio of 0.6, was slightly more 215.
serious than the one used in this study. The flow 3 Durst, F., Melling, A. and Whitelaw, J.H. Principles and
was driven by a sinusoidal piston pump and the practice of laser-doppler anemometry, Academic Press,
sinusoidal oscillation was superimposed on a steady London, 1976.
flow component. Their results are difficult to com- 4 Roach, M.R. Poststenotic Dilatation in Arteries in
pare with those obtained in the present work as Cardiovascular fluid dynamics, (Ed. Bergel.), 1972
their Reynolds number of 1210, is several times 111-139
5 Azuma, T. and Fukushima, T. Flow patterns in stenotic
higher than used here. Azuma and Fukushima
blood vessel models, Biorheology, 1976, 13, 337-35.5.
showed that the ffow was turbulent downstream of 6 Zienkiewicz, O.C. The Finite Element Methods in
the stenosis. Engineering Science, McGraw Hill, London, 1971.
7 Lambert, J.D. Computational Methods in Ordinary
ACKNOWLEDGEMENT Differential Equations, John Wiley 8c Sons, London,
1973.
The authors are grateful to Olav Dahl for valuable 8 Wille, SO. A finite element formulation of the time
discussions of the numerica methods. dependent Navier-Stokes equations. Comput. Meth. Appl.
Mech. Engxg, in press
9 Womersley, J.R. Methods for calculation of velocity,
REFERENCES
rate and flow and viscous drag in arteries when the pres-
1 Wille, SO. Pressure and flow in arterial stenoses simu- sure gradient is known, J. Physiol., 1955, 127, 553-563.
lated in mathematical models. Appl. Math. Modelling, 10 McDonald, D.A. Blood Flow in Arteries, The Camelot
in press. Press Ltd., 1974.

24 J. Biomed. Ennnn 1981 Vol. 3 lnntlnrv

You might also like