Materials and Design 41 (2012) 410–420

Contents lists available at SciVerse ScienceDirect

Materials and Design

journal homepage: www.elsevier.com/locate/matdes

Technical Report

Deployment of a self-expanding stent inside an artery: A finite element analysis

M. Azaouzi ⇑, A. Makradi, S. Belouettar
Advanced Material and Structure Department, Centre de Recherche Public Henri Tudor, 66 rue de Luxembourg, L-4002 Esch-sur-Alzette, Luxembourg

a r t i c l e i n f o a b s t r a c t

Article history: The superelastic, shape memory, biocompatibility, and fatigue properties of Nitinol, a nickel–titanium
Received 22 March 2012 alloy, have made the material attractive for medical devices such as self-expanding cardiovascular stents
Accepted 8 May 2012 (tubular mesh-like structures). Self-expanding stents are used in medical surgery to restore blood flow in
Available online 18 May 2012
a diseased artery segment (narrowing of the blood vessel due to plaque build-up) and keep the artery
open after angioplasty. Finite element analyses (FEAs) of Nitinol devices such as stents reduces testing
and time-to-market by allowing the designer to simulate the stent delivery inside the vessel. In this paper
a FEA has been performed to simulate the deployment and the pulsatile loading of a self-expanding Niti-
nol stent inside an artery. The results of the FEA have been used to assess the impact of the stent on the
artery and to assess the influence of the artery on the deformation field within the stent. Such analyses
will provide valuable information about the stent design and struts dimensions that can be optimized in
order to maximize the effectiveness of the stent during the deployment process. This work was con-
ducted using AbaqusÓ/standard finite element code in conjunction with a user material subroutine
(umat/Nitinol).
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction intravascular device can result in undesirable clinical events such

Self-expanding stents are being used extensively to treat occlu-
sions in endovascular arterial lumens, such as narrowing of the
blood vessels due to cholesterol plaque build-up. Self-expanding
stents are mesh-like tube structures made of a nickel–titanium al-
loy (Nitinol). The superelastic and shape memory properties of
Nitinol, along with its biocompatibility and fatigue properties have
made the material attractive for medical applications. The large
elastic strains possible in Nitinol reduce the risk of damage to
the stent both during delivery into the body and due to accidents
while in operation [1]. Stent designs vary in complexity from cylin-
drical braided wire meshes (Fig. 1) to laser-cut metal tubes. Nitinol
stents can be manufactured from thin tubes into which a pattern is
electromachined. After a sequence of operations, the stent is
mounted on a catheter and inserted into the blood vessel to coun-
teract the effects associated with arterial diseases. After being re-
leased by a delivery system (usually a system of catheter tubes)
as shown in Fig. 1, the stent self-expands and exerts a radial force
on the blood vessel to keep it open. Once delivered, the response to
blood pressure (cyclic pulsating load) determines the fatigue life of
the stent. Self-expanding stents have proven to reduce the extent
of arterial recoil and restenosis as compared to balloon angioplasty
procedures and provide a less invasive alternative in the treatment
of endovascular disease [2]. However, poor performance of the
⇑ Corresponding author. Tel.: +352 42 59 91 1; fax: +352 42 59 91 555.
E-mail address: mohamed.azaouzi@tudor.lu (M. Azaouzi).
as thrombosis and intimal hyperplasia [3]. In order to minimize
these adverse events, it is important to obtain a thorough under-
standing of the factors that contribute to an adverse outcome dur-
ing the stent deployment process. Such device-related factors
could include the stent design and material as well as the deploy-
ment procedure.
The mechanical behavior of Nitinol stent has been studied by
several researchers. Thériault et al. [4] developed a numerical solu-
tion for the progressive expansion of Nitinol stent covered by thin
polyethylene layer, allowing smooth and gradual contact between
the stent and the artery’s wall by creep effect. Wu et al. [5] studied
the biomechanical properties of Nitinol carotid stents and their
interactions with carotid arteries by building a finite element mod-
el, the authors demonstrated that the stent design has an effect on
the carotid vessel geometry. Kleinstreuer et al. [6] conducted a FEA
in order to study the effects of crimping, deployment, and cyclic
pressure loading on stent–graft fatigue life. The authors analyzed
the radial force and wall compliances of the vessel. Kim et al. [7]
developed a mechanical model for designing and studying the
structural behavior of a self-expandable stents fabricated using
braiding technology. Later, the same authors investigated the fea-
sibility of developing a temperature-responsive braided stent using
shape memory polyurethane through finite element analysis [8].
Similarly, Reese et al. [9] studied a new kind of stents, so-called
shape memory polymer stents.
Although an extensive work has been done to evaluate the
structural behavior of self-expanding stents, little attention has

0261-3069/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.matdes.2012.05.019
 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420 411

Nomenclature

q (kg/m3) material density rEL (N/mm2) end of transformation loading

e strain T0 (°C) reference temperature
r (N/mm2) stress (dr/dT)U (N/mm2 °C1) (dr/dT) unloading
r (N/mm2) Von-Mises equivalent stress rSU (N/mm2) start of transformation unloading
P (N/mm2) pressure stress rEU (N/mm2) end of transformation unloading
T (°C) temperature rSCL (N/mm2) start of transformation stress during loading in
S reduced polynomial strain energy density function compression, as a positive value
I1 ; I2 ; I3 invariants of the Cauchy–Green tensor Ur, Uh, Uz (mm) translational degree of freedom in the cylindrical
k1, k2, k3 principal stretches coordinate system
J total volume ratio URr, URh, URz (mm) rotational degree of freedom in the cylindri-
F transformation potential cal coordinate system
f fraction of martensite Dartery (mm) internal diameter of the artery
EA (N/mm2) austenite elasticity dartery (mm) internal diameter of the artery after expansion
mA austenite Poisson’s ratio Dstrut (mm) external diameter of the strut
EM (N/mm2) martensite elasticity dstrut (mm) external diameter of the strut after deployment
mM martensite Poisson’s ratio DD(mm) Dstrut  dstrut
eEL elastic strain L, W, H (mm) and N length, width, thickness and number of strut
eL transformation strain TAf (°C) austenite finish temperature of the Nitinol alloy
(dr/dT)L (N/mm2 °C1) (dr/dT) loading
rSL (N/mm2) start of transformation loading

been focused on characterizing the structural interaction of these 2. Stent design

stents with the arterial wall. The present FEA considers both the
deployment and pulsatile loading of the stent inside the artery.
One of the key issues in medical implants is the device lifetime
or, in engineering terms, fatigue life. The present FEA provides
quantitative measures of stress and strain that could be used to
make fatigue estimates for these devices. This allows the optimiza-
tion of designs and allows prediction of the device’s life. A simpli-
fied model of a Nitinol stent is considered in order to perform a FEA
of the stent deployment as well as the pulsatile loading inside the
artery. In Section 2 stent CAD geometry is presented. Then, in Sec-
tion 3 the adopted modeling methodology is discussed. After that,
some numerical results are illustrated in Section 4, the numerical
results give an idea about the stent design effect on the Chronic
Outward Force (COF, defined as the force exerted by the stent to
keep the artery open) as well as the interaction between the stent
and the vascular tissue after deployment. Section 5 discuss the
Nitinol stent behavior during cyclic pulsatile loading (blood pres-
sure) obtained by numerical simulation. Before the conclusion, a
general discussion is presented in Section 6. The quantitative FEA
conducted in this work could be used as a basis for the fatigue life
prediction of Nitinol stents.
Numerous Nitinol stents design are used in clinical practice to-
day, some are constructed by welding thin wires together, whilst
others are made by removing material from thin-walled tubes by
laser cutting. In any case, all Nitinol stents contain stress-concen-
tration features (corners, holes, joints) at which the stress is locally
high, and it is at these locations that failure may potentially occur.
Generally, Nitinol stents are composed of a series of V-shaped
struts in circumferential rows which are connected by bridges
(Fig. 2). Mechanical and geometric features that influence the Niti-
nol stent performance has been studied in previous work [10]. It
has been demonstrated that the Chronic Outward Force is a param-
eter of interest in the design of an endovascular stent. The COF
determine the amount of lumen expansion and can be calculated
by considering the geometry and motion of a unit-cell ‘‘Strut V’’
[10] (Fig. 2). The COF depends on the stent design such as the thick-
ness (H), width (W) and length (L) of each strut as well as the num-
ber (N) of struts along the circumferential direction. In this study,
as the pattern repeats itself, only a part of the stent with appropri-
ate symmetry has been considered as shown in Fig. 2. The stent
geometry used in this paper was taken from Ref. [5]. The external
diameter of the stent Dstent is 10 mm and the dimensions of the
strut are as follow: L = 2.25 mm, W = 0.15 mm and H = 0.2 mm.

3. Finite element modeling

3.1. Nitinol behavior modeling

The key characteristic of Nitinol is its superelastic material

Fig. 1. Self-expanding Nitinol stent.
behavior [11], making it an extremely flexible metal alloy that
can undergo very large deformations without losing the ability to
recover its original shape upon unloading (Fig. 3). At rest, the mate-
rial presents itself in an austenite phase (1). When loaded mechan-
ically beyond a certain stress rSL and at a temperature greater than
TAf, the austenite phase transforms into a martensite phase (2). The
transformation produces a substantial amount of strain, which on
unloading is reversible ((4) and (5)) because the martensite is not
stable for T > TAf Since the transformation strains are large (of the
order of 6%) compared to elastic strains in typical metals, the
412 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420

Fig. 2. Self-expanding Nitinol stent (a) and ring and unit-cell model (b).

Fig. 3. Uniaxial mechanical behavior of the Nitinol [12].

material is said to be superelastic. When the direct and inverse temperature effects. The material parameters used in the FEA were
transformations take place, the stress remains mainly constant taken from Ref. [12] (see Table 1).
over a large range of deformation respectively at the upper and A user material subroutine (UMAT/Nitinol) implemented in
lower plateau. This unusual behavior above TAf is called pseudo- Abaqus FE code [12] following the model proposed by Auricchio
elasticity. Further loading beyond superelastic limit reveals plastic et al. [13] was used for the present stent simulations. It should
behavior within the martensite phase (3). If a reverse loading is ap- be noted that the superelastic model cannot describe the plasticity
plied (for example, compression instead of tension), a similar of the martensite. The model is based on the loading–unloading
behavior is observed (6), with the exception that the stress levels irreversibility and the concept of generalized plasticity and physi-
required to produce the transformations are higher, while the cal principles [14]. Under the small strain assumption, the theory
transformation strain is lower. The material data required to cali- decomposes strain e into two parts, a purely linear elastic compo-
brate the Abaqus material model can be obtained from the uniaxial nent eEL and a transformation component eL:
behavior in terms of loading, unloading, reverse loading, and
e ¼ eEL þ eL ð1Þ
The austenite to twinned martensite transformation is driven
Table 1 by the resolution of shear forces, and it takes place within a range
Parameters of the Nitinol/UMAT material subroutine.
of stress levels that are characteristic of the material.
Nitinol/UMAT parameters
@F
q 6450 Kg/m3 eL ¼ aDf ; FS 6 F 6 FF ð2Þ
@r
EA 68,000 MPa
mA 0.3 where f is the fraction of martensite phase, a is a material constant
EM 36,000 MPa
which can be calibrated from uniaxial tensile test, and F is a trans-
mM 0.3
eL 0.04
formation potential. The same is true for the reverse transformation
(dr/dT)L 6.5 but at different stress levels. The intensity of the transformation fol-
rSL 520 MPa lows a stress potential law:
rEL 635 MPa
T0 37 °C
Df ¼ f ðr; fÞDF ð3Þ
(dr/dT)U 6.5
Any change in stress direction produces a reorientation of the
rSU 180 MPa
martensite with negligible additional effort. Changes in tempera-
rEU 26 MPa
780 MPa
ture produce a shift in the stress levels at which the transforma-
rSCL
tions take place. This shift is linear in temperature. Because there
 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420 413

is a volume increase associated with the transformation, it requires Table 2

less stress to produce the transformation in tension and more in Material properties of the artery.

compression. This is modeled with a linear Drucker–Prager ap- Coefficient C1 C2 C3 C4 C5 C6

proach for the transformation potential: Value 0.00652 0.0489 0.00926 0.76 0.43 0.0869

F¼r
  p tan b þ CT ð4Þ

with r being the Mises equivalent stress, p the pressure stress, T the
temperature, b and C are material constants.
A rule of mixtures is used to implement the change in linear
elasticity from the austenite phase into the martensite phase. In
addition to the usual stresses and strains, the Nitinol/UMAT rou-
tine tracks variables specific to the model. These variables include
the distribution of the fraction of martensite, transformation
strains, and equivalent stresses and strains. These are points in
the uniaxial tensile curve into which material points that have a
three-dimensional state, specific fraction of martensite, and a
specific loading/unloading history are mapped. The material data
required by the model are obtained from straightforward observa-
tions of uniaxial tests in terms of loading, unloading, reverse load-
ing, and temperature effects. The calibration consists of 14 values
as listed below in Table 1.
The mechanical behavior of the Nitinol alloy is illustrated in
Fig. 4 Upon loading, stress first increases linearly with strain up

using a homogeneous, isotropic and hyper-elastic constitutive
to about 1% (austenite phase). After a first yield point rSL , more
strain can be accumulated with only a little increase of the stress.
The end of this

almost plateau (loading plateau) is reached at about
6% strain rEL . After that, there is another linear increase of stress
with strain (martensite phase). During the unloading, the material
remains fully martensite till the stress decreases rapidly to the

S ¼ C 1 ðI1  3Þ þ C 2 ðI1  3Þ2 þ C 3 ðI1  3Þ3 þ C 4 ðI1  3Þ4
reverse transformation critical stress rSU . At the lower plateau
(unloading plateau), the strain is recovered with only little


decrease of stress. When the applied stress is lower than rEU ,
the stress-induced martensite is completely transformed to be aus-
tenite if no plastic strain occurs during the tensile loading. The last
portion of the deforming strain is finally recovered in a linear fash-
ion again.
Fig. 4 shows the cyclic stress–strain curves at two different tem-
peratures, the stresses increase when the temperature is increased
from 20 °C to 37 °C. The unusual feature of Nitinol stents is their
temperature dependent stiffness. Stents with a transition temper-
ature of 30 °C feel quite weak when squeezed or crushed at room
or lower temperature. In contrast, they feel much stiffer when
squeezed at temperatures above 30 °C. The material behavior mod-
el has been experimentally validated on the basis of uniaxial ten-
sile test as shown in Fig. 4. The implemented FE model closely
matches experimental measurements obtained from the uniaxial
test, where successively large strain loads are applied to the Nitinol
alloy at two different temperatures [15]. Since the superelastic
model cannot describe the plasticity of the martensite phase, there
are some deviations between numerical and experimental results
during unloading.
For the Nitinol superelastic material model, a small positive
slope was provided for the transformation between the austenite
and martensite phases, even though in reality the transformation
zone slope is close to zero or sometimes negative. Due to the differ-
ence between the loading and unloading curves, convergence diffi-
culties can arise at the junction of the start or the end of the
transformation zone if a large increment is used. To avoid this
problem during FEA, a maximum increment size of 0.25 was
imposed.
3.2. Artery behavior modeling
The mechanical behavior of the coronary artery was modeled
model based on the work of Holzapfel et al. [16]. The material
parameters used in (Eq. (5)) are listed in Table 2. The constitutive
law is based on a reduced polynomial strain energy density func-
tion, S, of sixth order:
þ C 5 ðI1  3Þ5 þ C 6 ðI1  3Þ6 ð5Þ
Here I1 is the first invariant of the Cauchy–Green tensor
I1 ¼ k21 þ k22 þ k23 ; ki ¼ J 1=3 ki ð6Þ
where ki are the principal stretches and J is the total volume ratio.
3.3. FEA procedure
The manufacturing process of Nitinol stents starts from a thin
tube in which a pattern is micro-machined. The finite element
model is built from this machined tube. Both the deployment
and pulsating load of the stent are considered (Fig. 5). At first,
the stent is expanded to its nominal dimensions, typically at a
diameter much larger than the original tube diameter. The stent
is then annealed to provide its new unloaded configuration. It is
then crimped from the outside (Fig. 5a) and inserted into the deliv-
ery system (catheter tube). Once inside the blood vessel, the stent

Fig. 4. Experimental validation of the Nitinol FE model [15].

414 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420

Fig. 5. The artery expands after stent deployment.

Fig. 6. Boundaries conditions in the cylindrical coordinates system (unit-cell model).

self-expands out of its containment by removing progressively the
catheter tube. After expanding, the stent exerts a Radial Force (RF)
on the blood vessel (Fig. 5b).
Numerical simulation and FEA was conducted using Abaqus/im-
plicit Finite Element (FE) code. In this simplified Nitinol stent sim-
ulation, only a fraction of the stent is modeled and the expansion is
constrained by the artery when the tool is released. As the FEA is
highly nonlinear (nonlinear geometry and material behavior),
unsymmetric matrix storage has been turned on in Abaqus/implicit
in order to aid convergence of the FEA since the stent surface is
highly curved (radius of curvature = 0.1 mm) which will give rise
to unsymmetric terms in the stiffness matrix.
To solve for static equilibrium, the static procedure in Abaqus/
Standard was used for stent analysis (Newton–Raphson algo-
rithm). Instabilities are one of the main convergence difficulties
for stent analysis using a static procedure. Sources of instabilities
include local buckling, slip and stick for frictional contact and, most
frequently, loss of contact between different components. In real-
ity, in these situations, the stored strain energy will turn into ki-
netic energy. In this analysis, static stabilization coefficient was
used carefully to dissipate the local dynamic effect.
3.4. Boundaries conditions and FE mesh
Nitinol stents are commonly used in various vascular beds,
including the carotid, femoral, renal, and iliac arteries. Each of
these anatomic locations, and even the individual lesions, presents
different mechanical challenges. The loading conditions corre-
sponding to the clinical stent deployment should be carefully con-
sidered when modeling and designing the Nitinol stent. The
oversizing of the stent to the arterial wall and the worst-case
strains with respect to fatigue life should be considered in the
 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420
FEA. While more complete modeling of loading conditions is one
way to obtain a more accurate estimate of the fatigue life of a stent,
frequently these loading conditions are not definitively known,
especially for relatively complex stenting indications.
Since the pattern repeats itself symmetrically, only specific part
of the stent (Unit-cell model) with appropriate symmetry bound-
ary conditions was considered (Fig. 6). The tool used to crimp the
stent ‘‘strut’’ is considered rigid and cylindrical. A radial displace-
ment Ur = + 3.75 mm was imposed on the external surface of the
cylinder (Fig. 6b) for the crimping simulation. Then, the same ra-
dial displacement is imposed on the cylinder (Ur = 3.75 mm) in
the opposite direction for the self-expansion simulation. A
smooth-step amplitude curve was defined to ramp up the radial
displacement from zero at the beginning of the crimping and ramp
the displacement down to zero at the end of the step.
Two symmetrical (axial and circumferential) boundaries condi-
tions were used to constrain the translational (U) and rotational
(UR) degree of freedom of the Unit-cell model (Fig. 6c and d). The
contact interactions between the cylinder and the strut as well as
the artery and strut were modeled with the penalty method and a
pure master–slave formulation for both contact interactions. The
penalty method was used because it prevents the chattering behav-
ior that might occur due to the differences in mass densities between
the stent and the artery materials during strut–artery contact. A Cou-
lomb friction coefficient of 0.1 was used to model the contact inter-
action. A self contact interaction was activated to prevent the mesh
overlapping of the strut at the end of crimping. Once, the numerical
simulation of expansion is finished, a pulsating load corresponding
to the diastolic/systolic cyclic blood pressure (Fig. 7) is exerted on
the inner surface of the strut and artery. Both, stent and artery are
meshed with solid linear hexahedron Finite Element (FE) in order
to capture the local peak stress and strain. The FE mesh size of the
strut and artery are respectively 0.025  0.025  0.05 and
0.1  0.2  0.15. This FE is the best candidate for the FEA of stent
models because it provides good balance between accuracy and effi-
ciency. As stents are loaded dominantly in bending, the fully-inte-
grated linear hexahedron element with incompatible mode C3D8I
has been used. This FE does not suffer from hourglassing and has
integration points closer to the surface. The only concern while using
the C3D8I element is that care must be taken to make the initial ele-
ment shape as close to a perfect cube as possible.
4. FEA of the stent deployment
4.1. Strut–artery interaction
As mentioned previously, the initial diameter of the stent is
10 mm. The stent is crimped up to a diameter smaller than the
Fig. 7. Pulsating load in kPa (one cycle).
415
internal diameter of the artery (dstrut = 2.5 mm). The unusual elastic
hysteresis of Nitinol allows the continuing opening force of the
stent acting on the vessel wall (COF), to remain very low even
through large deflections and oversizing of the stent. Meanwhile
the forces generated by the stent to resist compression (Radial
Resistance Force, RRF), increase rapidly with deflection until the
plateau stress is reached (Fig. 3). The amount of the RRF reached
at the end of crimping is 6 N. Fig. 8 shows the evolution of the Ra-
dial Force (RF) as a function of the stent diameter. After crimping,
the strut is allowed to expand progressively by its own stored
strain energy inside the artery. The strut expands until its external
diameter matches the initial internal diameter of the artery. When
the contact between the strut and artery is activated, the strut ex-
erts a RF and the artery expands. When the contact is activated, the
RF value becomes 4.85N and 4.3N respectively for an artery diam-
eter of 6 mm and 8 mm.
The expanded diameter of the artery depends on the RF and the
stent oversizing (defined as the difference between the initial stent
diameter and the initial artery diameter, Dstrut  Dartery). The artery
expansion is 42% and 23.75% in the case of an artery of 6 mm and
8 mm respectively. Fig. 9 shows the equivalent transformation
strain (eL) of the strut’s crimped (a) and expanded (b, c) configura-
tions. The maximum strain is located in the curved zone of the
strut. It was observed also that eL decreases when the stent oversiz-
ing decreases (Fig. 9b and c), which means more strain is recovered
after expansion of the strut. Fig. 10 shows the contours plots of the
circumferential stresses generated in the arterial wall upon stent
deployment. While there is a base level of stress generated by
the bulk expansion of the artery, the stresses that arise near the
contact interface dominate the interaction. Within the contact re-
gions, local stress concentrations arise where the stent exhibits
the largest radial deformation. The circumferential stress magni-
tudes obtained by numerical simulation are in good agreement
with those of experimental data published in [17]. It was also no-
ticed that the amount of stress in the wall artery depends on the
stent oversizing (Fig. 10).
4.2. Effect of the strut dimensions
As mentioned before, the strut design is crucial in the treatment
of stenosed artery [18]. The stiffness and flexibility are two main
characteristics of the stent. Stiffness is necessary to maintain the
blood vessel open and at the same time the stent should be as flex-
ible as possible during its delivery into the specified location of the
stenosed artery. Those two characteristics are dependent on the
strut design and dimensions [10]. In order to study the effect of
the dimensions and number of struts on the RF and artery
Fig. 8. Radial force (N) of the stent.
416 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420
Fig. 9. Equivalent transformation strain eL of the strut, (a) end of crimping, (b) end of expansion Dartery = 6 mm, (c) end of expansion Dartery = 8 mm.
Fig. 10. Circumferential stress (MPa) in the artery after stent deployment, (a) Dartery = 6 mm and (b) Dartery = 8 mm.
expansion, four geometrical parameters: L, W, H and N have been
varied as illustrated in Table 3.
The variation of the radial force and artery expansion as a func-
tion of the strut dimensions and number is illustrated in Fig. 11. It
can be noticed that the effect of L, W and H is remarkable on the RF.
A small variation of W and H generates an important variation of
the artery expansion. However the number N of strut along the cir-
cumferential direction has not a significant effect. An increase in L
or decrease in W and H results in a decrease of the strut stiffness
and thus lead to a decreased artery expansion. On the other hand,
an increase in M results in a moderate decrease of the artery
expansion. Fig. 11 shows a visible dependency of the artery expan-
sion on the stent oversizing and dimensions. An oversizing of
2 mm generates less variation of the artery expansion in compari-
son with a larger over-sizing of 4 mm.
Table 4 shows the amount of residual strain of the strut at the
end of expansion which is obtained for different strut dimensions
(A, B, C and D). It can be seen that the difference between Dstrut
and Dartery (which has been defined previously as the stent oversiz-
ing) has an influence on the amount of strain recovery. From
Table 4 it could be noticed that in the case of an artery diameter
Dartery of 6 mm more strain was recovered as L, H and W increase
(A, B and C) and in contrary less strain is recovered when N in-
crease (D). In the other hand, the amount of strain recovery in-
creases when W, H and N increase (B, C and D), but it decreases
if L increase (A). The effect of strut dimensions on the strain recov-
ery is considerable and it depends at the same time on the strut
oversizing. Strut length, width and thickness have the greatest
Table 3
Different strut dimensions.
Reference model A B C D
L (mm) 2.25 L/2 2L L L
W (mm) 0.15 W 2W/3 4W/3 W W
H (mm) 0.2 H H H/2 2H
N 12 N N N 2N/3
influence on strain for given conditions of oversizing and number
of struts. It has been demonstrated that the amount of strain is
an important factor in the fatigue life prediction of Nitinol stent,
which is loaded during in vivo service by a cyclic blood pressure
due to the heart beating. Pelton et al. [19] demonstrated from
experimental data that there is an increased fatigue life at mean
strains above about 1.5% compared with the zero-mean conditions.
The author speculated that this result could be due to strain
accommodation from stress-induced martensitic transformation.
From Table 4 it could be expected that the fatigue life of Nitinol
stent is directly related to the oversizing amount, dimensions
and number of struts. This result indicates that for an accurate fa-
tigue life prediction it is crucial to take into account the stent over-
sizing as well as the interaction between the stent and artery.
5. FEA of the pulsating load
5.1. Evolution of the stress and strut diameter
After deployment of the stent, a cyclic pulsating load (Fig. 7)
was applied to the inner surface of the strut. The strut responses
corresponding to an artery of 6 mm and 8 mm are illustrated in
Fig. 12. The equivalent stress req of the strut increases with the
stent over-sizing. During progressive expansion of the strut, req in-
creases up to 350 MPa in the case of an artery diameter of 6 mm.
But it remains constant in the case of an artery diameter of 8 mm.
Fig. 13 shows the behavior of the strut which occurs during the
transitory step following the deployment inside the artery. The
stent response to the cyclic pulsating load depends on the stent
over-sizing as well as the stent design. In the case of an important
over-sizing (Dstrut  Dartery = 4 mm), the stent continue its progres-
sive expansion until stabilization. In the case of an artery diameter
L of 6 mm, more elastic strain has been recovered after stabilization
of the strut response to 125 cyclic pulsating loads. In the other
H
hand, the strain amount remains almost constant over 125 cycles
4N/3
for an artery diameter of 8 mm.
 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420 417

Fig. 11. Effect of the strut dimensions on the radial force and artery expansion.

5.2. Fatigue life analysis
In this paragraph, the fatigue life analysis of the stent is limited
to a preliminary scoping assessment. In general, both stress based
and strain based approaches provide insight into the mechanisms
of fatigue behavior to predict lifetimes and the appropriate choice
often depends on the specifics of the specimens and test environ-
ment. Regarding the material properties of Nitinol and the small
cross-section of the stent, it is recommended to conduct fatigue life
analysis with strain based approach. From the FEA conducted in
this paper, it has been demonstrated that the equivalent transfor-
mation strain eL (strain recovery) is related to the amount of
418 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420

Table 4
Effect of the strut dimensions on the strain recovery.

Dartery A B C D
L/2 2L 2W/3 4W/3 2H/3 4H/3 2N/3 4N/3
eL (%) 6 mm 3.025 1.5 3.382 2.675 3.612 2.433 3.57 4.19
8 mm 0.036 0.23 0.376 0.032 0.306 0.03 0.083 0.055

Fig. 12. History of the maximum Von-Mises stress req in the strut during pulsating load.

Fig. 13. History of the stent expansion during pulsating load.

over-sizing of the stent and its strut dimensions (Table 4). The fa-
tigue lifetime of Nitinol depend on eL [19]. As a consequence, the
fatigue lifetime of this medical device will depend on the stent
over-sizing as well as the strut dimensions.
In order to show the stent behavior regarding the cyclic loading,
it is possible also to consider the stress based approach. The result
illustrated in Fig. 14 was obtained using AbaqusÓ finite element
code. The post-processing of the result was performed using a py-
thonÓ script [20]. The stress amplitude and average stress was cal-
culated after stabilization over the strut. The stabilization of the
strut response to the pulsating load occurs after 125 cyclic pulsat-
ing loads. Each point in this figure represents the stress which was
calculated at the centroid of the hexahedron finite element. The
chattering of stresses results from the variation of the equivalent
transformation strain obtained over the strut (Fig. 9b and c). From
the two diagram illustrated in Fig. 14, it can be noticed that the
stress amplitude Dr which is determined as a function of the aver-
age stress rmean is considerably depending on the stent over-sizing.
These results are in concordance with the Nitinol behavior that re-
flects an elastic response at a large amount of stress, and then a
superelastic response at almost constant stress (see plateau stress
shown in Fig. 3). Fig. 14 shows also a relatively dependence of the
stress amplitude on the stent oversizing.
6. Discussion
Nowadays, more than 100 types of stents are available on the
market [21,22]. Stents can be categorized according to their expan-
sion mechanism, geometrical shape, and manufacturing process
[21]. In terms of expansion mechanism, stent can be classified as
balloon-expandable where the angioplasty balloon is employed
to both open the blocked artery and expand by plastic deformation
the stent, or as self-expanding where as the stent is pushed out of
the catheter it immediately opens out to support the already di-
lated lumen. Balloon-expandable stents have to be overexpanded
 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420
Fig. 14. Stress amplitude over the strut, (a) Dartery = 8 mm and (b) Dartery = 6 mm.
to achieve a certain diameter due to the elastic springback after
deflation. The springback is highly undesirable feature and the
overexpansion may damage the vessel and cause restenosis. On
the other hand, self-expanding stent will continue to gently push
outward against the artery wall after deployment and follow artery
movements. A classification and comparison of the self-expanding
stents design and properties can be found in [18,22]. The design
requirements for a stent include: flexibility, trackability, biocom-
patibility, low surface area (Scaffolding), minimal foreshortening,
high radial strength (kink resistance), radiopacity, and fatigue
resistance [18,23].
The stent design is a major factor which determines its reliabil-
ity during insertion into the blocked artery and throughout the
long term in vivo service. For example, the stent performance could
be improved by increasing strut thickness (which increases radio-
visibility, radial stiffness and arterial wall support). Nevertheless,
excessive strut thickness may impart more vascular injury, cause
more intimal hyperplasia, and engender a higher risk for restenosis
than thinner strut. A comparative study regarding the relationship
between strut dimensions and arterial wall reaction as well as the
impact of stent design on clinical outcome are discussed in [24]. In
clinical practice, the sent design and material is very important for
the lesion that is going to be treated. Balloon-expandable stent are
used especially for coronary arteries, though self-expanding stents
are more appropriate for peripheral arteries such as carotid and
femoral artery where ballon-expandable stents could become
crushed and compromise blood flow.
In the last decade, self-expanding stents comprise the vast
majority of the endovascular stent market where the SMART Niti-
nol stent (manufactured by CordisÓ Corporation) is widely used. A
survey of a popular self-expanding stents design can be found in
[21,23]. The success of the SMART stent design was largely due
to its very fine mesh structure that offered exceptional contouring,
flexibility, and apposition characteristics. However, as the stent is
subjected to a long-term cyclic pulsatile (typically 4  107 cycles/
year) and musculoskeletal loading, fatigue fracture may occur.
Pelton et al. [19] investigated the in vivo fatigue resistance and
durability of Nitinol SMART stent, and demonstrated that there is
a significant effect of the equivalent transformation strain eL on
the fatigue life of stents. Based on the fundamental strain-life ap-
proach, prediction of a stent’s fatigue performance requires a pre-
cise analysis of the fatigue strains of a stent in vivo and a profound
understanding of Nitinol’s fatigue behavior under the similar
deformation patterns. Several researchers investigated the effects
of mean and alternating strain on the fatigue behavior of super-
elastic Nitinol stent and observed that the oscillating strain has a
greater effect on fatigue life than the mean strain [25]. In the other
hand, the FEA performed herein is based on the stress-life ap-
proach which demonstrated the dependency of the alternating
stress (Dr) and mean stress (rmean) on the stent oversizing.
419
Because of the superelastic behavior of Nitinol alloy, it is recom-
mended to use strain-life approach for the fatigue prediction
[19], but we believe that it is relatively acceptable to use stress-life
approach when we take into account the in vivo interaction be-
tween the stent and artery. It is obvious that the numerical results
could be investigated deeply for the fatigue life estimation by tak-
ing into account the in vivo interactions between the stent, the
cholesterol’s plaque and artery.
In self-expanding Nitinol stents, the COF and RRF are two
important factors. As it has been discussed previously, Nitinol
stents exert a low chronic outward force against the wall vessel
but resist deformation with a much higher force. The radial
strength (RRF) or crush recoverability is an important feature for
stents in superficial vessels (carotid or femoral artery) that could
be deformed through outside forces. An increase or decrease of
the radial strength depends on the stent design and strut
dimensions. Experimental analysis of the radial strength and flex-
ibility relevant to the performance of 14 stents can be found in
[26].
The mechanical properties of self-expanding Nitinol stent re-
ported in the past include chronic outward force (COF), radial resis-
tance force (RRF), flexibility, shortening ratio, radiopacity, and
trackability [23]. The first 3 properties are related to device behav-
ior after stent deployment, and the last 3 are important factors
during the insertion process. We believe that the first 3 properties,
namely, COF, RRF and flexibility have an effect on the clinical per-
formances such as patency and complications. Among these prop-
erties, the radial stiffness or strength is obviously the most
important because in self-expanding stent, radial force maintains
and expands the luminal patency at the stricture once the stent
is deployed. The FEA presented herein focused on the radial force
which reflects the stent strength. This FEA is very useful in terms
of the understanding of stent behavior during its deployment,
the interaction with blood vessel and the relationship between
the stent design and the COF as well as the RRF. The FEA was con-
ducted using a pattern similar to the Nitinol SMART stent design
[20] in order to demonstrate the effect of strut dimensions on
the strain recovery as well as the radial strength. However, due
to the complexity of the stent deployment process as well as
in vivo pulsating load, it is necessary to conduct a relatively sophis-
ticated FEA.
The fatigue resistance is related to the amount of the residual
strain obtained after deployment, an improvement of the self-
expanding stent reliability in terms of fatigue resistance, may be
performed by combining numerical simulations with optimization
strategies. Accordingly, the stent behavior and its interaction with
the artery during deployment could be refined by optimizing the
radial strength (COF and RRF). It should be also mentioned that
the transition temperature TAf of the stent can be adjusted to a
certain extent during material processing. This gives the designer
420 M. Azaouzi et al. / Materials and Design 41 (2012) 410–420
another option to increase or decrease the radial strength of the
stent without changing the design or dimensions of the strut [27].
7. Conclusion
Modeling and simulation of the stent–artery interaction were
presented. Nonlinear FEA was used to determine stresses and
strains of a stent subcomponent subjected to oversizing and pulse
pressures. The pulsating load simulation was used to dictate the
baseline for fatigue prediction. A finite element methodology has
been developed to study the nature of the stent–artery interaction.
This methodology can be used to determine the impact of various
stent designs attributes on the vessel wall, as well as the impact
that the tissue has on the behavior of the stent. One of the key is-
sues in medical implants is device lifetime or, in engineering terms,
fatigue life. The FEA presented in this paper provides quantitative
measures of stress and strain that could be used to make fatigue
estimates for these devices. This allows the optimization of designs
and allows prediction of the device’s life. It has been demonstrated
that the Unit-cell model could be used for a simplified and quanti-
tative FEA of the Nitinol stent deployment and pulsating load. For a
given stent design, there is a remarkable influence of the strut
dimensions (L, W and T) on the stent strength as well as the artery
expansion. Whoever, the number N of struts along the circumfer-
ential direction has not a significant effect on the stent and artery
behavior. The strain recovery of the stent depends on the differ-
ence between the artery diameter and the initial stent diameter
(namely stent over-sizing). Thus, the stent oversizing has an
impact on the fatigue life of Nitinol stents. Complexity of stent de-
signs and of the actions taken to deploy them requires sophistica-
tion in the finite element model. Generally, advanced modeling
capabilities such as contact with both rigid tools and other flexible
bodies, large deformations, localized buckling and simulation of
complex loading sequences are needed. Therefore, a refined finite
element model has to be generated. In a future work, the stent
deployment will be fully modeled inside a curved artery that is
made of 3 layers and containing a cholesterol plaque. The FEA that
will be performed on this curved artery will allow the simulta-
neous study of stent flexibility and strength. Optimization of the
stent design has the advantage to improve the stent behavior dur-
ing deployment. Thus, the combination between FEA and optimi-
zation techniques such as Response Surface Method, Genetic
Algorithm or other techniques will be investigated in a future
work. The Unit-cell model will be used as a basis for the stent de-
sign optimization. Finally, some practical validations should be
performed in the future which involves testing the FEA model of
the stent deployment history to ensure that the model accurately
represents the actual history of a stent deployed in the body.
Acknowledgment
This work has been supported by the National Research Fund-
ing of Luxembourg (FNR) via the project DeStenEE, C09/MS/09.
References
[1] Morgan NB. Medical shape memory alloy applications – the market and its
products. Mater Sci Eng, A 2004;378:19–23.
[2] Duerig TW, Wholey M. A comparison of balloon and self-expanding stents. Min
Invas Ther Allied Technol 2002;11(4):173.
[3] Haroek J, Zoucas E, Stenram U, Cwikiel W. Insertion of self-expandable nitinol
stents without previous balloon angioplasty reduces restenosis compared with
PTA prior to stenting. Cardiovasc Intervent Radiol 2002;5:430.
[4] Thériault P, Terriault P, Brailovski V, Gallo R. Finite element modeling of a
progressively expanding shape memory stent. J Biomech 2006;39:2837–44.
[5] Wu W, Qi M, Liu XP, Yang DZ, Wang WQ. Delivery and release of nitinol stent
in carotid artery and their interactions: a finite element analysis. J Biomech
2007;40:3037–40.
[6] Kleinstreuer C, Li Z, Basciano CA, Seelecke S, Farber MA. Computational
mechanics of Nitinol stent grafts. J Biomech 2008;41:2370–8.
[7] Kim JH, Kang TJ, Yu WR. Mechanical modeling of self-expandable stent
fabricated using braiding technology. J Biomech 2008;41:3202–12.
[8] Kim JH, Kang TJ, Yu WR, Kim JH, Kang TJ, Yu WR. Simulation of mechanical
behavior of temperature-responsive braided stents made of shape memory
polyurethanes. J Biomech 2010;43:632–43.
[9] Reese S, Böl M, Christ D. Finite element-based multi-phase modelling of shape
memory polymer stents. Comput Methods Appl Mech Eng 2010;199:1276–86.
[10] Duerig T, Tolomeo D, Wholey M. An overview of superelastic stent design. Min
Invas Ther Allied Technol 2000;9(3/4):235–46.
[11] Book GB. Applications of titanium–nickel shape memory alloys. Mater Des
1983;4(4):835–40.
[12] Abaqus Manual, Version 6.11. Dassault systems; May 2011. <http://
www.simulia.com>.
[13] Auricchio F, Taylor RL, Lubliner J. Shape memory alloys: macromodelling and
numerical simulations of the superelastic behavior. Comput Methods Appl
Mech Eng 1997;146:281–312.
[14] Lubliner J, Auricchio F. Generalized plasticity and shape memory alloys. Int J
solids Struct 1996;33:991–1003.
[15] Rebelo N, Walker N, Foadian H. Simulation of implantable Nitinol stents.
Abaqus User’s Conference; 2001.
[16] Holzapfel GA, Sommer G, Gasser CT, Regitnig P. Determination of layer-specific
mechanical properties of human coronary arteries with nonatherosclerotic
intimal thickening and related constitutive modeling. Am J Physiol, Heart Circ
Physiol 2005;289:2048–58.
[17] Holzapfel GA, Sommer G, Regitnig P. Anisotropic mechanical properties of
tissue components in human atherosclerotic plaques. J Biomech Eng
2004;126:657–65.
[18] Bonsignore C. A decade of evolution in stent design. In: Pelton AR, Duerig TW,
editors. SMST 2003: international conference on shape memory and
superelastic technologies. California, Pacific Grove: International
organization on SMST; 2004. p. 519–28.
[19] Pelton AR, Schroeder V, Mitchell MR, Gong XY, Barney M, Robertson SW.
Fatigue and durability of Nitinol stents. J Mech Behav Biomed Mater
2008:153–64.
[20] Python, Version 2.7; July 2010. <http://www.python.org>.
[21] Stoeckel D, Bonsignore C, Duda S. A survey of stent designs. Min Invas Ther
Allied Technol 2002;11(4):137–47.
[22] Stoeckel D, Pelton A, Duerig T. Self-expanding nitinol stents: material and
design considerations. Eur Radiol 2004;14:292–301.
[23] Lally C, Kelly DJ, Prendergast PJ. Stents. Wiley Encyclopedia of biomedical
engineering. Jhon Wiley and Sons; 2006.
[24] Sangiorgi G, Melzi G, Agostoni P, Cola C, Clementi F, Romitelli P, et al.
Engineering aspects of stents design and their translation into clinical practice.
Ann Int Super SANITA 2007;43(1):89–100.
[25] Pelton AR, Gong XY, Duerig T. Fatigue testing of diamond-shaped specimens.
In: Pelton AR, Duerig TW, editors. SMST 2003: international conference on
shape memory and superelastic technologies. Pacific Grove, CA: International
Organization on SMST; 2003. p. 293–302.
[26] Isayama et al. Measurement of radial and axial forces of biliary self-
expandable metallic stents. Am Soc Gastr Endosc 2009;70(1):37–44.
[27] Pelton A, DiCello J, Miyazaki S. Optimization of processing and properties of
medical grade nitinol wire. Min Invas Ther Allied Technol 2000;9(1):107–18.