.l. Ekmdda Vol. 23, No. 10, pp. 10314037, 1990.

oa21-92570/90
s3m+.OO
Printi In orat BrhiIl 0 t~perpmon-plo

PULSATILE TWO-DIMENSIONAL FLOW AND PLAQUE

FORMATION IN A CAROTID ARTERY BIFURCATION

M. NAZEMI,C. KLEINSTREUER*andJ.P. ARCHIEJR~

Department of Mechanical and Aerospace Engineering, North Carolina State University, Raleigh,
NC 27695-7910, U.S.A.

Abstract-Velocity and pressure fields, streamlines and wall shear stress distributions were numerically
obtained for two-dimensional, steady and pulsatile flow in a carotid artery segment. Distinct regions of
reverse flow near the bifurcation and wavy flow patterns in the branching channels were observed during
portions of the pulse. These phenomena disappear at the end of the systolic phase of the cardiac cycle. A
previously validated plaque formation model predicts that plaque sites and the local extent of
atherosclerotic lesions are similar for those present on human angiograms.

l.INTRODUCTION Numerical studies of steady and pulsatile laminar

flow in rigid two-dimensional and tubular bifurcations
Numerous coupled physical and biochemical factors have been carried out by several investigators. For
play an important role in atherogenesis and throm- example, Rindt et al. (1987), considering a rigid two-
bosis as documented and discussed in Texon (1980), dimensional model of a carotid artery bifurcation,
Pedley (1980), Fung (1984) and Strandness et al. observed high velocity gradients at the divider walls of
(1987). In particular, the influence of hemodynamics the daughter branches and large zones of reverse flow
and particle transport in the development of vascular near the non-divider walls of these branches. They
diseases is of major interest. Theoretical studies and in used a sinusoidal pulse with 20 fixed time increments
vivo observations have shown that atherosclerotic per pulse which is in the range 250 <Red < 770 where
lesions are much more likely to occur in the proximity Re, is the inlet Reynolds number. Perktold and Hilbert
of vessel branches and junctions. For example, (1986) simulated numerically pulsatile flow (Re, G 100)
Kleinstreuer et al. (1988a,b) developed a model for the in a two-dimensional carotid bifurcation model. Their
pseudo-transient formation of plaque at sites and work concentrated on reverse flow during the pulse
deposition rates detied by steady-state flow charac- period and the paths of single fluid particles in the flow
teristics in an aortic bifurcation. Their model predicts field of the carotid sinus. Clark et al. (1983) considered
the onset of atherosclerotic lesions in areas of low wall pulsatile flow in a symmetric conduit with a stenosis of
shear stresses. With local growth of particle depos- varying degrees of severity located slightly up- of
itions, plaque layers have been added in a stepwise the bifurcation point. Their finitedifferen& solution
fashion in regions of both high and low shear stresses of the Navier-Stokes equations indicated high shear
(cf. Nazemi et al., 1989). The validity of this model has stresses near the peak of the occlusion and rtverse flow
heen supported with physical evidence as shown in regions distal and proximal to the stenosis for this
DeBakey et al. (1985), Texon (1980) or Wylie and pulsatile flow system.
Ehrenfeld (1970), and in the particle trajectory ana- Zarins et al. (1983), using laser Doppler anemo-
lyses of Nazemi and Kleinstreuer (1989). Thus, a metry, determined the velocity profiles and shear
compromise between two somewhat conflicting hy- stresses of steady flow at Reynolds numbers Re,
pothetical correlations (cf. Caro et al., 1971; Fry, 1969, = 400, 800 and 1200. They compared these profiles
1973, p. 119) linking low and high wall shear stress with the location of in&al plaque of 12 adult human
levels and atheroma has been postulated. In this carotid bifurcations. They found the maximum inti-
paper, the plaque formation model is extended to mal thickness to occur in areas of low shear stress
pulsatile flow in a rigid carotid bifurcation. The work along the outer wall opposite to the flow divider and
is based on the two-dimensional analyses of steady the minimum thickness in areas of high shear stress
flow with plaque formation in (symmetric) aortic along the divider wall. Ku et al. (1985a,b), using a laser
artery bifurcations (Nazemi et al., 1989; Nazemi and Doppler anemometer and a pulsed Doppler ultra-
Kleinstreuer, 1989). sound, considered the hemodynamics of the normal
human carotid bifurcation both in vitro and in viw.
For the normal flow patterns in the proximal internal
carotid artery, they observed: unidirectional, helical,
Received in final form 3 April 1990. transient reversal, and low velocity regions. Using the
*Author to whom all correspondence should be ad-
pulsed DoppIer ultrasound, they also obtained a
dressed.
t Surgeon, Carolina Cardiovascular Surgical Associates, qualitative description of the complex flow velocity
P. A., Raleigh, NC, U.S.A. fields in the normal human carotid bifurcation, and by

1031
1032 M. NAZEMIet al.

comparing the two methods, they found similar pri- Nazemi et al., 1989). If the actual wall shear stress
mary and secondary flow features. Fukushima et al. distributions fall within a predetermined band-width,
(1988) observed helical motions of fluid particles inside i.e. T,~”< rw < z,.~, which may be the case after some
a human carotid bifurcation model under steady and initial plaque development, the model stops and no
pulsatile flow conditions. They also found that low- more plaque layers are added (cf. Nazemi, 1989).
ering the Reynolds number and the Womersley num- However, changes in wall or plaque surface curvature
ber, W= oRFlu (where o is the angular frequency of as well as changes in fluid flow inlet conditions may re-
the motion) weakens the intensity of the helical flow. start the plaque formation model. Different levels of
In summary, disturbed flow patterns of interest here, critically low or high wall shear stresses have been
such as recirculation zones and wavy flow patterns, tested, and other potentially important flow para-
were observed by Ku and Giddens (1983), Ku et al. meters such as the wall shear stress gradient have been
(1985a) and Fukushima et al. (1988), among others. used to study the impact of different system para-
When compared with previous publications, the meters that might atkct local plaque formation (cf.
present analysis considers a number of new features Nazemi, 1989). While the threshold of critical wall
such as a realistic input pulse, detailed velocity pro- shear stress levels influence the rate of local plaque
files, pressure distributions and streamlines, and, most growth, but not the lesion sites, the other flow para-
importantly, an advanced plaque formation model. meters tested did not reproduce the actual plaque
locations.
2. THEORY
2.2. Governing equations and boundary conditions
2.1. Modeling approach
Considering pulsatile laminar incompressible flow
In an attempt to simulate and analyze blood flow of a Newtonian fluid through a two-dimensional
through a realistic carotid bifurcation, and to study branching conduit, the fluid flow equations are for
the development and physical effects of atherosclerotic region R (cf. Fig. 1):
lesions, the Navier-Stokes equations are solved for
pulsatile laminar flow in a two-dimensional rigid au au
-+.-_=o
az ay
(1)
channel representing a carotid bifurcation (cf. Fig. 1).
This geometry is the same as the one used by Rindt
au ttau vau
et al. (1987) and Perktold and Hilbert (1986) and is
based on the average values of the dimensions meas- dt+dz+ay= ++v ($+$) (2)

ured in twenty-five post mortem carotid artery bifurc- au au au 1 ap a% a5

ations given by Bharadvaj et al. (1982).
Of particular interest are wall regions which, based
z+“z+v;j;
z---+v
P ay (-+-
a22 ay= )
. (3)

on their actual shear stress levels, may be susceptible The boundary conditions are (cf. Fig. 1):
sites for atherosclerotic lesions. The existence of a (4a) At the inlet of the common carotid (z=O,
correlation between low and/or oscillating wall shear O<y<R,): u is prescribed, e.g. u=3/21,(1-(y/R,)=),
stress levels and the onset of atherosclerotic lesions is and v=O where II, is the averaged inlet velocity.
pretty much established. Additional (local) plaque (4b) At the outlet of the internal and the external
build-up due to critically low and high wall shear stress carotid: p=O, &if&?=0 and o’=O.
levels may lead to severely stenosed arteries, including (4c) Along the walls, i.e. on aR: u = v = 0 and t?=
total lumen occlusion over time. Thus, in examining o”=o.
the transient wall stress distributions during the car- The initial conditions are as follows (cf. Fig. 2). All
diac cycle and comparing them to a critically low internal nodes have the initial value of zero for u and v,
reference shear, susceptible sites can be identified and but at the inlet C,=vReJD, varies according to the
the plaque formation scheme can be activated (cf. pulse shown in Fig. 2 (cf. Bharadvaj et al., 1982).

u=v=Oen JR
v t=o
b.6 0.1 0.4 0.4 0.8 1.0 1.2

Time, t (WC)
Fig. 1. System schematics with coordinate system (graph is
not to scale). Fig. 2. Input pulse for common carotid artery.
 Plaque formation in a carotid artery bifurcation 1033

Calculations were carried out for two consecutive N Re/50) for the steady flow case and slightly less for
pulses after all of the start-up effects had been purged, each time step in the pulsatile flow case. On the
and the results at the end of the second pulse were used average, 100 time steps were required for integration
as initial condition for the simulation. throughout each pulse with the majority of the time
steps required during the deceleration phase of the
pulse, i.e. 0.24 s < t < 0.44 s (cf. Fig. 2).
3. SOLUTIONMETHOD In order to obtain the appropriate initial condi-
tions, the solution for the onset Reynolds number flow
Equations (l)-(3) subject to the boundary condi- (Rei =200) is obtained by starting with the initial
tions @a)-(4~) and the initial condition described condition of zero values assigned to the dependent
above,’ including moving boundaries representing variables, and then gradually increasing the axial
plaque formation, are solved using the Galerkin finite velocity to the inlet Reynolds number of Re,=200.
element method (cf. Cuvelier et al., 1986; Engleman, Using thii solution as the initial condition for the
1982). The irregular flow domain is disc&xxi with pulse, integration can be performed for a number of
quadrilateral elements with nine nodes allowing bi- cycles until the difference of the results between the
quadratic shape functions for the velocity profiles and two consecutive cycles is negligible. The accuracy of
a constant interpolation function for the pressure. A the present code was tested by comparing transient
preprocessor generates a mesh of variable density axial velocity profiles (cf. Fig. 3) with measurements
based on the desiied channel configuration and the conducted by Rindt et al. (1987) using the same
anticipated region of steep gradients and recirculating geometry and the same input pulse.
flow. Approximately 115 elements in the horizontal or The wall shear stresses, the pressure distributions
z-direction and 25 elements in the vertical or y- and the velocity profiles are of particular interest,
direction were used for the simulation. The mesh especially in the vicinity of the channel junction which
spacings ranged from less than 0.2 mm near the apex contains areas most susceptible to atherosclerotic
point, to nearly 2 mm near the inlet of the common lesions. In order to test the opposing hypothetical
carotid and to a maximum of 3-4 mm near the outlets correlations between critical wall shear stresses and
of internal and external carotid arteries which were plaque formation, as proposed by Fry (1969,1973) and
extended to match (4b). A maximum total of 5850 Caro et al. (1971), a lower and an upper threshold
nodes have been necessary for an accurate simulation. shear stress, i.e. r,*,, =0.5z, and t,,,_= 1.57,, are de-
Using the mesh point coordinates and the element f&d, where rp is the wall shear stress for the equival-
connectivity, the preprocessor assembles the system ent Poiseuille flow. Neither the magnitude of the
matrix by integrating the shape functions in the reference shear stress, TV,nor the coefficients 0.5 and
Galerkin weighted residual formulation of the gover- 1.5 are that significant for the stress-induced depos-
ning equations (l)-(3). The penalty function formula- ition analysis; they simply relate to the rate of particle
tion (penalty parameter, E= 1 x 10-s) has been used to deposition. However, a correlation between simulated
eliminate the pressure degree of freedom, thereby plaque growth rates and actual developments of
reducing the size of the system matrix of the unknowns atherosclerotic lesions is subject to future statistical
(the pressure is recovered after the solution for inand u analyses.
has been obtained). Quasi-Newton’s method with Sites for which T~<‘T,,,~,, =0.5r, are identified as
Broyden’s update is used to linearize the resulting areas for the onset of plaque formation. Then a mass
algebraic system based on the initial values of u and u. layer with a thickness proportional to AT= 1r_., -
After obtaining the optimal re-numbering of the nodes T,,,*~, mxJ is incorporated. The mesh is locally re-
to minimize computational requirements, the pro- designed to incorporate the new ‘solid wall’, and the
cessor uses Gaussian elimination without pivoting to
solve the profiled linearized matrix at every given time
step. The new values of Y and u are compared with the
previous set, until, iteratively, convergence is reached.
A second order accurate implicit time integrator
using a variable time step scheme based on the control
of the local time truncation error (cf. Gresho et al.,
1980) is used since, due to steep gradients in the pulse
(cf. Fig. 2), a small time increment may be required to
track the transient flow behavior at a particular time
interval whereas, at other times, a much larger time
increment may be appropriate. The time steps ranged
from AtminN 1 x lo-’ s for the decelerating phase to
AL N 5 x 10e2 s for the accelerating phase. The ex-
ecution time required on an IBM 3090 without using Fig. 3. Comparison between measured velocity profiles
the parallel processors ranged between 1 and 2 min for (Rindt et al., 1987; their time sampling point S2) and pre-
each continuation step (No. of continuation steps dicted results.
1034 M. NAZEMIet al.

new Z, is computed via a repeat solution of the 4. RIESULTS

governing equations. Further mass increases may
occur in a transient stepwise fashion at each sampling 4.1. Base case study
point N-S6 of the cardiac cycle (Fig. 2). Sites with For our base case study, we focused on six time
?V>LX- - l.Sz, are subjected to subsequent deposits sampling points H-S6 during the cardiac cycle (cf.
which are determined by a first-order polynomial Fig. 2), to show velocity vector profiles in the vicinity
function in terms of 2,. The maximum deposition at of the bifurcation point and two representative
each iteration is 7 % of the local channel height streamline pictures (Figs 4 and S), respectively. Qual-
which occurs for r,=O in case of low, and for itative comparisons can be made between these graphs
7,>2.fk, in case of high, shear stress area. Thus, and experimental observations as discussed pre-
plaque originates only at sites experiencing actual viously. Both our numerical predictions and published
shear stresses below a pre-set rmin, and further plaque flow visualization studies shov the temporal appear-
build-up occurs whenever t,tUp, > z,,. Plaque forma- ance of up to three recirculation zones and the wavy
tion may be terminated when the actual wall shear flow patterns, especially during the diastolic and early
stress falls between the lower and upper bounds, i.e. systolic phases of the cardiac cycle. Specifically, the
z,in < L < 7max at all times, for a given geometry and recirculation regions completely disappear at t
prescribed input/output pulses (cf. Kleinstreuer et al., = 0.24 s (cf. sampling point S3 in Fig. 2) because of the
1989). Otherwise, like in real life, a vessel may become dominant time acceleration term, a/& in equations (2)
90-100 % occluded (cf. Texon, 1980). The maximum and (3). During the deceleration phase, i.e. 0.24 s < c
inlet Reynolds number of lumen constrictions were <O&l s, the recirculation regions move and grow
selected such that local turbulence was avoided at all which causes the velocity profiles to shift first towards
times. A the divider walls and then towards the outer walls,

’ S6

Fig. 4. Time sequence of velocity profiles near apex shown at six sampling points, SlS6, as depicted in
Fig. 2.
 Plaque formation in a carotid artery bifurcation 1035

2 @

I! -1*

B-88
0 2 4 6 8 18

Axial IQcaliQQ, 1 (cm)

Fig. 6(a). Shear stress along upper wall of internal carotid

artery at Re,=400 under accelerating, decelerating and
steady flow conditions.

. 1 4 6 6
Axial location, L (cm)

Fig. 5. Streamlines in carotid bifurcation at time-sampling Fig. 6(b). Shear stress along lower wall of external carotid
points Sl and SS. artery at Re,=400 under accelerating, decelerating and
steady flow conditions.

setting up temporarily wavy flow patterns especially in enlargement is not sufficient to overcome the domi-
the internal carotid artery segment. These experi- nant time acceleration term. The corresponding pres-
mentally observed two-dimensional wavy flow pat- sure profiles are shown in Fig. 7(a)-(c). It is apparent
terns (Fig 5) have now been numerically reproduced that high pressure levels are needed to accelerate the
for the 5rst time. flow when compared with the moderate levels needed
Figure 6(a) and (b) shows the wall shear stress for maintaining steady flow and negative pressure
distributions along both non-divider walls for a rep- levels for decelerating flow. Although the pressure at
resentative inlet Reynolds number of Re,=400, i.e. the inlet is lower than at the outlet during a portion of
steady flow as well as the appropriate points in time the pulse, it is important to note that the average flow
during the systolic phase (Rel =400 up) and the dia- in the carotid artery is never reversed, i.e. the mean
stolic phase (Re+OOdown) of the pulse (Fig 2). flow rate is always greater than zero.
During steady flow, a moderately sixed recirculation
region exists in the carotid sinus, and low shear 4.2. Wall shear stress induced deposition
stresses are present near the bifurcation along the Based on the temporal values of the local wall shear
lower and upper non-divider wall [Fig 6(a),(b)]. Near stress relative to the pre-set critical r,-levels, as out-
bifurcation, the conduit area increases which results in lined in the previous section, the plaque formation
an adverse pressure gradient causing the flow near the mode1 predicts the onset of atherosclerotic lesions at
walls to be retarded, and in the case of strong adverse first only at two sites near the bifurcation, i.e. inside
gradien& causing the 5ow to separate from the wall In the carotid sinus on the upper wall and on the lower
the case of decelerating flow, i.e. ~/&CO, the adverse non-divider wall. Then, further plaque growth is pre-
pressure gradient and the flow deceleration eflkctjoin dicted at the same areas, and a third plaque site is
forces to cause an even bigger recimulation region in appearing at the upper divider wall near the end of the
the carotid sinus than in the steady flow case, and to carotid sinus because of the changing geometry and
create a new recirculation region near the lower non- wall shear stress distributions (Fig 8). After six con-
divider wall. For accelerating flow, i.e. a/&>O, the secutive stepwise additions of plaque in areas pre-
mcirculation region is completely swept away since dicted by the model, a comparison has been made
any potential adverse pressure gradient due to area (cf. Fig. 8) with human angiograms and autopsy
1036 M. NAZEMIet al.

6 Layers ,

Fig. 7(a). Pressure profiles at Re,=400 under steady flow 6 Layers _

conditions.
Fig. 8. Comparisons between stepwise plaque-layer forma-
tion and angiograms.

5. DE3CUSSlON

Laminar steady and pulsatile flow inside a two-

dimensional, rigid carotid bifurcation model has been
simulated. The velocity profiles, pressure fields, and
wall shear stress distributions were obtained for differ-
ent sampling points throughout a cardiac cycle. For
comparison purposes, steady, decelerating and accel-
erating flows at an inlet Reynolds number of 400 were
investigated. The transient plaque formation model
based on the local levels of wall shear stress has been
used to predict the sites most susceptible to the onset
and growth of plaque. Results of the fluid flow analysis
Fig. 7(b). Pressure profiles at Re, = 400 under accelerating
flow conditions.
and the plaque formation model have been success-
fully compared with measured velocity profiles
(cf. Rindt et al., 1987) and angiographic findings (cf.
DeBakey et al., 1985; Texon, 1980; Wylie and Ehren-
feld, 1970).
The following conclusions can be drawn:
(i) The steady flow assumption, even if the simu-
iation is performed at different mean inlet Reynolds
numbers, does not give an accurate picture of the
actual flow behavior.
(ii) Decelerating flow causes the recirculation re-
gions to increase in size and extent; multiple zones of
recirculation may exist in the sinus for this phase of
pulsatile flow. Wavy flow patterns occur in both the
internal and the external carotid artery (Ku et al.,
1985a and Fukushima et al., 1988).
(iii) Three areas are susceptible to atherosclerotic
lesions: (1) the upper non-divider wall located in the
Fig. 7(c). Pressure profiles at Re,=400 under decelerating carotid sinus; (2) the upper divider wall near the end of
flow conditions. the carotid sinus; and (3) the lower non-divider wall
near bifurcation. Plaque formation predictions at
these sites are in good agreement with angiographic
and autopsy results. Critically low and high wall shear
results. The predicted and observed sites and forms of stress levels may contribute to the onset and develop-
plaque growth agree rather well. An increase of sam- ment of atherosclerotic lesions (cf. Fry, 1969, 1973;
pling points S,, i= 1, . . . , n= 12, resulted in different Car0 et al., 1971).
plaque growth rates but the characteristic sites stayed Based upon the results achieved with the validated
the same (cf. Nazemi, 1989). plaque formation model, it can be postulated that
 Plaque formation in a carotid artery bifurcation 1037

transient two-dimensional rigid conduit flow bifurcation with stenosis and surgical reconstruction.
model@ is an important first stop in the simulation of Paper presented at the World Congress on Medical Phys-
ics and Biomedical Engineering, San Antonio, TX, August
hemodynamic factors relevant in atherogenesis. How-
1988 (Edited by Clark, J. W. et al.). Phys. Med. Biol. 33,
ever, it would be of interest to study secondary effects Suppl. I.
in pulsatile three-dimensional flow in complex arteries Kleinstreuer, C., Naxemi, M. and Archie, J. P. Jr (1988b)
with viscoelastic wall response in order to co&m the Analysis of atherosclerotic plaque formation in artery
present 6ndings and to study further the causative bifurcations. Paper presented at the 10th Annual Intema-
tional Conference, IEEE Engineering Society in Medicine
relationships between physico-biochemical factors and Biology, New Orleans, LA, November 1988 (Edited by
and atherosclerosis including local plaque build-up. Harris, G. and Walker, C.), Vol. 10, Part 2/4. IEEE, New
However, the present plaque formation model pre- York, NY.
dicts plaque sites in branching arteries under pulsa- Kleinstreuer, C., Naxemi, M. and Archie, J. P. Jr (1989) CAD
applications in surgical reconstruction and replacement of
tile flow conditions quite accurately, and hence it may
stenosed artery segments. Paper presented at the 3rd Joint
be already useful as a predictive tool for improved ASCE/ASME Mechanics Conference, San Diego, CA, July
surgical reconstructions, say, after carotid endarterect- 1989 (Edited by Torxilli, P. A. and Friedman, M. H.), AMD
omy, and for optimal designs of end-to-side anaato- Vol. 98, New York, NY.
moses. Ku, D. N. and Giddens, D. P. (1983) Pulsatile flow in a model
carotid bifurcation. Arteriosclerosis 3, 31-39.
Ku, D. N., Giddens, D. P., Phillips, D. 3. and Strandness, D.
E. Jr (1985a) Hemodynamics of the normal human carotid
REFERENCES
bifurcation: in vitro and in vivo studies. Ultrasound Med.
Biol. 11, 13-26.
Bharadvaj, B. K., Mabon, R. F. and Giddens, D. P. (1982) Ku, D. N., Giddens, D. P., Zarins, C. K. and Glagov, S.
Steady flow in a model of the human carotid bifurcation. (1985b) Pulsatile flow and atherosclerosis in the human
Part I-flow visualixation. J. Biomechanics lS,349-362. carotid bifurcation. Arteriosclerosis 5. 293-302.
Caro, C. G., F&Gerald, J. M. and Schroter, R. C. (1971) Nammi, M. (1989) Hemodynamics and particle deposition in
Atheroma and armrial wall shear observations, correlation artery bifurcations with implications to atherogenesis and
and proposal of a shear dependent mass transfer mech- surgical reconstruction. Ph.D. thesis, MAE Department,
anism for atherogenesis. Proc. R. 80~. 177,109-159. NC State University, Raleigh, NC.
Clark, M. E., Robertson, J. M. and Cheng, L. C. (1983) Naxemi, M. and Kleinstreuer, C. (1989) Analysis of particle
Stenosis severity effects for unbalanced simple-pulsatile trajectories in aortic artery bifurcations with stenosis. J.
bifurcation flow. J. Biomechanics 16, 895-906. biomech. Engng 111, 311-315.
Cuvelier, C., Sagal, A. and Stecnhoven, A. A. (1986) Finite Nszemi, M., Kleinstreuer, C., Archie, J. P. Jr and Sorrell, F.
Element Methods and Nauier-Strokes Equations. Reidel, Y. (1989) Fluid flow and plaque formation in a two-
Dordrecht. dimensional aortic bifurcations. J. biomech. Engrr8 111,
DeBakey, M. E., Lawrie, G. M. and Glaesser, D. H. (1985) 316-324.
Patterns of atherosclerosis and their surgical significance. Pedley, T. J. (1980) The Fluid Mechanics of Large Blood
Ann. Surg. 201,115-131. Vessels. Cambridge University Press, NY.
E&man, M. S. (1982) FIDAP-a fluid dynamics analysis Perktold, K. and Hilbert, D. (1986) Numerical simulation of
package. A& Engng Sojtw. 4,163-166. pulsatile flow in a carotid bifurcation. J. biomed. Engng 8,
Fry, D. L. (1969) Certain histological and chemical responses 193499.
of the vascular interthce to acutely induced meohanical Rindt, F. N., Vosse, F. N., Steer&oven, A. A., Janssen, J. DC
stress in the aorta of the dog. Circ. Res. 24,93-125. and Reneman, R. S. (1987) A numerical and experimental
Fry, D. L. (1973) Responses of the arterial wall to certain analysis of the flow field in a two-dimensional model of the
physical factors. In Atherogenesis: Initiating Factors (Edi- human carotid artery bifurcation. J. Biomechanics Xl,
ted by Porter, R. and Knight, J.), pp. 93-105. Associated 499-509.
Scientific Publishers, Amsterdam. Strandem, D. E., Didisheim, P.,.Clowes, A. A. and Watson,
Fukushima, T., Homma, T., Harakawa, K., Sakata, N. and J. T. (eds) (1987) Vascular Diseases: Current Research and
Axuma, T. (1988) Vortex generation in pulsatile flow Clinical Applications. Grune and Stratton, Orlando, FL.
through arterial biication models including the human Texan, M. (1980) Hemodynamics Basis of Atherosclerosis.
carotid artery. J. biomech. Engng 110, 166471. Hemisphere, Washington, DC.
Fung, Y. C. (1984) Biodynamics~S&ingcr, New York. Wylie, J. E. and Ehrenfeld, W. K. (1970) Extracranial Occlu-
Gresho. P. M.. Lee. R. L. and Sani. R. L. (19801On the time sive Cerebrovascular Disease Diagnosis and Management.
dcp&nt &&on of the incompressible NavierStokes W. B. Saunders, Philadelphia, PA.
equations in two and three dimensions. In Recent Aduan- &tins, C. K., Giddens, D.-P., Bharadvaj, B. K., Sottiurai,
ces in Numerical Methods in Fluids, pp. 83-92. Pineridge V. S. and Mabon. R. F. (1983) Carotid bifurcation athero-
Press, Swansea, U.K. sclerosis, quantitative c&relation of plaque localiration
Kleinstreuer. C., Naxemi, M. and Archie, J. P. Jr (1988a) with flow velocity profiles and wall shear stress. Circ. Res.
Numerical analysis of fluid-particle dynamics in artery 53, 502-514.