CHINESE JOURNAL OF PHYSICS VOL. 42, NO.

4-I AUGUST 2004

Numerical Study of Pulsating Flow Through a Tapered Artery with Stenosis

Guo-Tao Liu, Xian-Ju Wang, Bao-Quan Ai, and Liang-Gang Liu

Department of Physics, ZhongShan University, GuangZhou 510275, China

(Received November 26, 2003)
Numerical simulations of pulsating blood flow through models of stenotic and tapered
arteries have been performed to investigate the distributions of the wall shear stress. In this
paper, a mathematical model of a rigid tapered artery with axially symmetric stenosis is
established. The incompressible Navier-Stokes equations are solved numerically by using the
finite difference method. The effects of tapering and stenosis are considered. The results
show that stenosis disturbs the flow field at the vicinity of the stenosis, especially at the
throat and downstream, and easily leads to the formation of a flow separation region in the
post-stenotic region.

PACS numbers: 87.19.Uv, 87.19.Tt, 83.85.Pt, 47.60.+i

I. INTRODUCTION

Arteriosclerosis is a common disease which severely influences human health. Early

arteriosclerotic lesions are not randomly distributed throughout the arterial tree; they usu-
ally tend to form and grow at certain locations, such as the distal to abdominal aorta,
coronary arteries, and carotid bifurcations. It has been found that the initiation and local-
ization of arteriosclerosis is closely related to local hemodynamic factors (such as wall shear
stress, etc.). Although there remains some uncertainty with regard to the exact hemody-
namic factors responsible for the initiation of arteriosclerosis, it has been demonstrated that
the development of arteriosclerosis is strongly related to the characteristics of the blood flow
in the arteries [1–5].
Recently theoretical, experimental, and numerical investigations [6–12] have been
performed to study the blood flow in arteries. The complex geometry of arteries (viz.
bending, bifurcation, stenosis, etc.) is also an important factor, which obviously affects
the local hemodynamics. Arteriosclerosis is characterized by localized arterial narrowing
(or stenosis). Stenosis may alter the local hemodynamics, at the same time the altered
local hemodynamics can cause the re-narrowing of blood vessel (restenosis). Some stud-
ies, for example Saas A. Ahmed [13] and M. Siouffi et al. [14, 15], have reported their
experimental work performed in rigid stenoses. Stergiopulos et al. [16] have measured
pulsatile flow in flexible stenoses. Numerical studies of pulsatile flow through axially sym-
metric smooth rigid stenoses, with blood as an incompressible Newtonian fluid, have been
reported [17–19]. Furthermore, Tu and Deville [20] and Ishikawa and coworkers [21] have
taken the non-Newtonian properties of blood into consideration; Stroud et al. [22] have
emphasized the influence of stenosis morphology; Andersson et al. [23] have taken stenosis

http://PSROC.phys.ntu.edu.tw/cjp 401 c 2004 THE PHYSICAL SOCIETY

OF THE REPUBLIC OF CHINA
402 NUMERICAL STUDY OF PULSATING FLOW . . . VOL. 42

FIG. 1: The geometry of the stenotic and tapered artery.

as surface irregular stenosis; Moayeri and Zendehbudi [24] have considered the effects of
elastic property of the wall.
There is no question that tapering is a significant aspect of mammalian arterial sys-
tem. Thus, in this paper, we are interested in pulsating flow through a tapered artery with
stenosis. A mathematical model of tapered arteries having an axially symmetric stenosis is
established. The blood (taken as a Newtonian fluid) flowing through several stenotic and
tapered rigid models is investigated and numerical calculations are carried out using a high
order difference scheme. Since the development of arteriosclerosis in arteries reduces the
elastic property of the arterial wall, the assumption of a rigid wall may be reasonable. Con-
sidering the wall shear stress distribution is an important diagnostic factor for examining
the flow characteristics of the blood through the arteries; here we present the obtained wall
shear stress distributions.

II. GOVERNING EQUATIONS

The tapered vessel segment having an axially symmetric stenosis is mathematically

modeled as a rigid tube with a circular section. Let (r, θ, z) be the coordinates where
the z-axis is taken along the axis of the artery while r and θ are along the radial and the
circumferential directions, respectively. The geometry of the artery (see Fig. 1) is modeled
mathematically as

R − m(z + d) − h cos φ (1 + cos πz ) |z| ≤ z ,


0 0
R(z) = 2 z0 (1)
R0 − m(z + d) |z| > z0 ,


where R(z) denotes the radius of the tapered arterial segment in the constricted region,
R0 is the constant radius of the straight artery in the non-stenotic region, φ is the angle
of tapering, h cos φ is the height of the stenosis for tapered arteries, z0 is the half-length of
the stenosis, m(= tan φ) represents the slope of the tapered vessel.
If we consider the blood flow in the stenotic and tapered arterial segment to be
pulsating, axisymmetric, two-dimensional and fully developed, the Navier-Stokes equations
VOL. 42 GUO-TAO LIU, XIAN-JU WANG et al. 403

and the equation of continuity which govern the blood flow are written in the cylindrical
coordinates as
∂u ∂u ∂u 1 ∂p µ ∂ 2 u 1 ∂u ∂ 2 u
+u +v =− + ( 2 + + 2), (2)
∂t ∂z ∂r ρ ∂z ρ ∂r r ∂r ∂z

∂v ∂v ∂v 1 ∂p µ ∂ 2 v 1 ∂v ∂ 2 v v
+u +v =− + ( 2+ + 2 − 2), (3)
∂t ∂z ∂r ρ ∂r ρ ∂r r ∂r ∂z r

∂u 1 ∂(rv)
+ = 0, (4)
∂z r ∂r
where u and v are the axial and the radial velocity components, respectively, p is the
pressure, ρ is the density of blood, and µ represents the viscosity of blood.
We introduce the following dimensionless variables:
r z u v t
r∗ = , z∗ = , u∗ = , v∗ = , t∗ = , (5)
R0 R0 u0 u0 T
where u0 is the average velocity in a period over the inlet section, T is the pulsating period
of the blood flow.
For convenience in solving the governing equations, we adopt the stream function-
vorticity form of the equations. This form has a major advantage over the primitive variable
form in two-dimensional incompressible flow, as it satisfies the equation of continuity au-
tomatically. So we introduce the dimensionless stream function Ψ∗ and the dimensionless
vorticity Ω∗ by
1 ∂Ψ∗ 1 ∂Ψ∗ ∂v ∗ ∂u∗
u∗ = , v∗ = − , Ω∗ = − ∗. (6)
r ∗ ∂r ∗ r ∗ ∂z ∗ ∂z ∗ ∂r
Eliminating p from equations (2) and (3) and using equations (5) and (6), we get the
following equations for the stream function and vorticity (after omitting the asterisks in all
the dimensionless variables)

∂ 2 Ψ ∂ 2 Ψ 1 ∂Ψ
+ − = −rΩ , (7)
∂z 2 ∂r 2 r ∂r

1 ∂ 2 Ω 1 ∂Ω ∂ 2 Ω
 
∂Ω 1 ∂Ψ ∂Ω 1 ∂Ψ ∂Ω Ω Ω ∂Ψ
St + − = 2
+ + 2
− 2 − 2 , (8)
∂t r ∂r ∂z r ∂z ∂r Re ∂r r ∂r ∂z r r ∂z
where Re is the Reynolds number, Re = ρR0 u0 /µ, and St is the Strouhal number, St =
R0 /(T u0 ).
The boundary conditions associated with the problem are given by:
at the inlet cross section, the flow is assumed to be fully developed,
∂Ψ ∂Ω
= = 0, (9)
∂z ∂z
404 NUMERICAL STUDY OF PULSATING FLOW . . . VOL. 42

FIG. 2: The nonuniform grid.

at the outlet cross section, we assume the flow field has no change,

∂2Ψ ∂2Ω
= = 0, (10)
∂z 2 ∂z 2
along the z-axis, owing to the axisymmetry of the flow,

Ψ = Ω = 0, (11)

at the upper wall,

Ψ = 0.25(1 + sin 2πt) . (12)

Only the upper-half flow in the stenotic and tapered tube is computed, because an
axisymmetric flow is assumed. The governing equations, subject to the boundary condi-
tions, are solved by the finite difference method. The numerical results given below are
obtained using a non-uniform grid (see Fig. 2), where a fine grid is employed near the wall
and the stenosis and a coarser one for the other parts of the tube. The calculation region
is −5 ≤ z ≤ 5, where z has been non-dimensionalized. The resulting grid has 200 × 35 grid
points.

III. NUMERICAL RESULTS AND ANALYSIS

During the computational procedure, we used the characteristic parameters: u0 = 50

cm/s, R0 = 0.3 cm, ρ = 1.056 g/cm3 , µ = 0.04 dyn·s/cm2 , and T = 0.4 s. So we have
Re = 396, and St = 0.015. For convenience of comparison we have adopted five models
(see Table I), in which there are four (M 1, M 3–M 5) with the same stenosis length. The
models M 3–M 5 have the same taper angle (φ > 0), and the angle of tapering of model M 1
equals zero.
Numerical results were obtained by using the models and the boundary conditions as
mentioned above. The results have been presented in order to estimate the contributions
of the different physical parameters to the present phenomena. The choice of φ = 0◦
corresponds to a non-tapered artery, while the values of φ causing the changes in the
VOL. 42 GUO-TAO LIU, XIAN-JU WANG et al. 405

TABLE I: The geometrical models of the blood vessels.

M1 M2 M3 M4 M5
h/R0 1/3 0 1/4 1/3 1/2
z0 /R0 1 0 1 1 1
φ 0◦ 0.5◦ 0.5◦ 0.5◦ 0.5◦

results of relevance can help in estimating the effect of arterial tapering on the wall shear
stress. The consideration of h = 0 represents an artery without constriction in its lumen
and the result can be compared with the corresponding results for h 6= 0.
Figs. 3 (a)-(d) show the results for the wall shear stress distribution along the vessel
wall with φ = 0.5 at t/T = 0, 0.25, 0.5, and 0.75. When the flow begins to accelerate (at
t/T = 0), the wall shear stress increases to a maximum negative value at the throat of the
stenosis, and the maximum negative value is bigger with increasing h. It is observed that
the wall shear stress changes in direction in the downstream vicinity of the throat, that is,
the corresponding values change to low positive values. This implies that a flow separation
region (the separation point is when τwall = 0; the reattachment point is when τwall = 0
again) existed instantaneously in these three models (M 3–M 5), however the phenomenon
doesn’t appear in model M 2 (we regard M 2 as a normal case). At the peak flow time (i.e.
t/T = 0.25), the maximum negative value reaches its peak value and the flow separation
region grows. The peak value in model M 5 may be over 18 times bigger than that in the
normal case. If the critical value of the wall shear stress that may cause damage to the
endothelial layer of the vessel wall is about 344 dyne/cm2 [25], the wall shear stress for the
model M5 will likely tear the endothelial layer, allowing the development of arteriosclerosis,
and finally form a thrombus. At time t/T = 0.5, the maximum negative value decreases and
the flow separation region continues growing. The wall shear stress of model M 2 is almost
equal to zero. Until time t/T = 0.75 (zero net flow), the wall shear stress is oscillating and
positive everywhere in the computational region near the wall. As the blood flow is further
reduced until reaching zero net flow, the backward flow gradually occupies the whole blood
vessel. As the blood flow increases the positive wall shear stress decreases step by step and
then becomes negative. Finally the flow finishes a cycle period. It is noticed that the wall
shear stress upstream of the stenosis isn’t disturbed at these times, whether the stenosis
exists or not.
In order to estimate the effect of tapering on the wall shear stress, we give the results
for the distributions of the wall shear stress through models M1 and M4 at t/T = 0.5 (see
Fig. 4). Fig. 4 shows that the variation of the wall shear stress is almost the same except
at the throat of the stenosis. Due to the tapering, the maximal negative value is bigger
near the throat of the stenosis. We also compare the results at other times finding that the
situation is the same as that at t/T = 0.5.
Fig. 5 illustrates the results for the distribution of the wall shear stress at the throat
of the stenosis; it is well known that at this location the values of the wall shear stress are
the highest. With the value of h increasing, the oscillation amplitude of the wall shear stress
406 NUMERICAL STUDY OF PULSATING FLOW . . . VOL. 42

(a) t/T =0 (b) t/T = 0.25

(c) t/T =0.5 (d) t/T = 0.75

FIG. 3: Distributions of the wall shear stress along the vessel wall for different h at different times
(φ = 0.5◦ ).

increases. At about t/T = 0.25, the oscillating value reaches a maximum; the value for the
model M 5 (h = 1/2) is over 18 times more than that for M 2 (h = 0). The temporal values
of the wall shear stress for these models are compared. Based on model M 2, from t/T = 0
to about 0.6, the differences of the corresponding values gradually increase (the bigger the
value of h, the bigger the difference value), reaching a maximum at about t/T = 0.25, and
then diminishing. At a certain time, about t/T = 0.7, the curves converge to a point; later
the curves diverge.

IV. SUMMARY AND CONCLUSION

Wall shear stress is an important factor in the study of blood flow. Accurate predic-
tions of the distribution of the wall shear stress are particularly useful for the understanding
of the effect of blood flow on endothelial cells. In this paper, numerical studies have been
VOL. 42 GUO-TAO LIU, XIAN-JU WANG et al. 407

FIG. 4: Distributions of the wall shear stress along the vessel wall for models M 1 and M 4 at
t/T = 0.5.

FIG. 5: Distributions of the wall shear stress at the throat of the stenosis.

carried out on pulsating flow through various types of stenotic and tapered vessels with the
main following assumptions: laminar flow, rigid walls, and Newtonian fluid. Both spatial
and temporal distributions of the wall shear stress have been obtained, and some of the
results have been compared. It has been proved that the height of the stenosis is a more
important factor influencing blood flow than tapering.
Comparison of the distributions of the wall shear stress at some instantaneous times
reveals that the height of stenosis has a strong influence on the wall shear stress at the throat
and downstream of the throat, and on the length of the flow separation region. Since high
wall shear stress not only may damage the vessel wall and cause intimal thickening, but
408 NUMERICAL STUDY OF PULSATING FLOW . . . VOL. 42

may also activate platelets, cause platelet aggregation, and finally result in the formation
of a thrombus, the wall shear stress at the throat deserves to be noticed. Considering that
the flow separation region may prolong the residence time of the particles in the blood,
the region may allow the phenomenon of lipid deposition and cell adhesion. Since low and
oscillating shear stress has been related to the initiation and localization of arteriosclerosis,
hence the post-stenotic zone could be prone to the propagation of the disease. However the
wall shear stress upstream of the stenosis isn’t disturbed at any time, whether the stenosis
exists or not.
Comparison of the time variation of the wall shear stress at the throat of the stenosis
indicates that the waveform of the wall shear stress is almost the same as that of the blood
flow at the inlet cross section. Due to the tapering, the oscillation amplitude of the wall
shear stress is bigger than that of a non-tapered artery.
A comparison of the effect of tapering on the wall shear stress reveals that the tapering
of the artery does not change the flow pattern, but only changes the values. The peak value
of the wall shear stress for the tapered artery is bigger than that without tapering.

Acknowledgments

This project was supported by the National Natural Science Foundation of China
under Grant Nos. 10275099 and 10175096.

References

[1] A. Gnasso et al., Stroke 28, 993 (1997).

[2] Y. Jiang et al., Stroke 31, 2319 (2000).
[3] L. Kornet et al., Arteriosclerosis, Thrombosis, and Vascular Biology 19, 2933 (1999).
[4] J. R. Buchanan, C. Kleinstreuer, S. Hyun, and G. A. Truskey, J. Biomech. 36, 1185 (2003).
[5] R. Botnar, G. Rappitsch et al., J. Biomech. 33, 137 (2000).
[6] G.-t Liu, X.-j. Wang, B.-q. Ai, and L.-g. Liu, Chin. J. Med. Phy. 20, 168 (2003).
[7] S. Cavalcanti, J. Miomech. 28, 387 (1995).
[8] H. Huang, V. J. Modi, and B. R. Seymour, Int. J. Engng Sci. 33, 815 (1995).
[9] J. M. Reese and D. S. Thompson, J. Biomech. 31, 1051 (1998).
[10] V. G. Hart and J. Shi, Int. J. Engng Sci. 33, 1121 (1995).
[11] P. Segers and P. Verdonck, J. Biomech. 33, 299 (2000).
[12] I. Bakirtas and N. Antar, Int. J. Engng Sci. 41, 1163 (2003).
[13] S. A. Ahmed, Experimental Thermal and Fluid Science 17, 309 (1998).
[14] M. Siouffi, V. Deplano, and R. Pelissier, J. Biomech. 31, 11 (1998).
[15] V. Deplano and M. Siouffi, J. Biomech. 32, 1081 (1999).
[16] N. Stegiopulos, M. Spiridon, F. Pythoud, and J. J. Meister, J. Biomench. 29, 31 (1996).
[17] Q. Long, X. Y. Xu, K. V. Ramnarine, and P. Hoskins, J. Biomech. 34, 1229 (2001).
[18] G. R. Zendehbudi, M. S. Moayeri, J. Biomech. 32, 959 (1999).
[19] C. Bertolotti and V. Deplano, J. Biomech. 33, 1011 (2111).
[20] C. Tu and M. Deville, J. Biomech. 29, 899 (1996).
VOL. 42 GUO-TAO LIU, XIAN-JU WANG et al. 409

[21] T. Ishikawa, L. F. R. Guimaraes, S. Oshima, and R. Yamane, Fluid Dynamics Research 22,
251 (1998).
[22] J. S. Stroud, S. A. Berger, and D. Saloner, J. Biomech. 33, 443 (2000).
[23] H. I. Andersson, R. Halden, and T. Glomsaker, J. Biomech. 33, 1257 (2000).
[24] M. S. Moayeri and G. R. Zendehbudi, J. Biomech. 36, 525 (2003).
[25] Z.-l. Tao, Bio-fluid mechanics, (Science Press, Beijing, 1984), in Chinese.