Vous êtes sur la page 1sur 13

Leslee W.

Brown
Engineering Manager
Mem. ASME
Gates Corporation,
2975 Waterview Drive,
Rochester Hills, MI 48309
e-mail: lwbrown@aol.com
Lorenzo M. Smith
Associate Professor
Department of Mechanical Engineering,
Oakland University,
118 Dodge Hall,
Rochester, MI 48309
e-mail: l8smith@oakland.edu
A Simple Transversely Isotropic
Hyperelastic Constitutive Model
Suitable for Finite Element
Analysis of Fiber Reinforced
Elastomers
A transversely isotropic ber reinforced elastomers hyperelasticity is characterized using
a series of constitutive tests (uniaxial tension, uniaxial compression, simple shear, and
constrained compression test). A suitable transversely isotropic hyperelastic invariant
based strain energy function is proposed and methods for determining the material co-
efcients are shown. This material model is implemented in a nite element analysis by
creating a user subroutine for a commercial nite element code and then used to analyze
the material tests. A useful set of constitutive material data for multiple modes of defor-
mation is given. The proposed strain energy function ts the experimental data reason-
ably well over the strain region of interest. Finite element analysis of the material tests
reveals further insight into the materials constitutive nature. The proposed strain energy
function is suitable for nite element use by the practicing engineer for small to moderate
strains. The necessary material coefcients can be determined from a few simple labo-
ratory tests. DOI: 10.1115/1.4003517
Keywords: hyperelasticity, elastomers, strain energy functions, transversely isotropic
nonlinear materials, rubber, nonlinear nite element analysis
1 Introduction
Combining an elastomeric matrix with a reinforcing material is
a valuable engineering solution to many design challenges. These
composites are often the key to success in many components such
as v-belts where they are used to withstand the wedging of the belt
into a sheave while also allowing the belt to remain exible in
bending. This desired combination of properties is accomplished
by creating an elastomeric composite, which has different stiff-
nesses in different directions.
The desired material properties can be obtained by mixing
small bers with an isotropic elastomer and milling the material to
align the bers. The direction of milling, or the machine direction,
can be referred to as the with-grain WG direction while nor-
mal to this direction can be referred to as the cross-grain XG
direction. Enough bers are oriented precisely in the with-grain
direction to substantially increase the stiffness of the material in
that direction. This process creates an elastomeric composite,
which can be best described as a transversely isotropic hyperelas-
tic material.
The degree of anisotropy can be quantied by the ratio of the
stress at a given elongation for a principal material direction to the
same stress in another principal material direction
R
a
=

WG

XG
at a given elongation 1
This has been referred to as modular anisotropy 1.
While extensive work has been done on characterizing isotropic
elastomers, relatively very little work has been done in character-
izing anisotropic elastomers. The existing isotropic elastomer
models are insufcient to simulate the typical response of an an-
isotropic elastomer. What is needed is an efcient stable aniso-
tropic elastomer model and appropriate constitutive tests to char-
acterize the material.
In this effort, a series of tests was used to gain a constitutive
understanding of a particular anisotropic elastomer, a model con-
sistent with those test results was developed, the model imple-
mented in a commercial nite element code, and the response of
the material examined through nite element analysis FEA of
the tests.
2 Constitutive Tests
In 1944, Treloar provided a set of test data that became a cor-
nerstone of efforts to develop adequate hyperelasticity models for
isotropic elastomer studies 2,3.
Using data from more than one constitutive test is recom-
mended for an isotropic elastomer. The tests should cover the
types of deformation, the temperature, the strains, and the strain
rate expected in service 4. This practice prevents using an elas-
tomeric material model that is not well suited for the analysis 5.
Test data to t material models for use in commercial nite
element programs can be obtained from a combination of uniaxial
tension, uniaxial compression, biaxial tension, pure shear, and
simple shear tests 6,7. There is even a standard, BS 903-5, to
guide the selection of suitable tests for an isotropic elastomer. A
simple efcient method suitable for many industrial laboratories is
the combination of uniaxial tension, pure shear, and uniaxial com-
pression tests 8.
Very little has been written about recommended tests for char-
acterizing transversely isotropic and anisotropic elastomers de-
spite the great deal of testing that has been conducted. A lot of
testing has been done on biological materials, which are also an-
isotropic hyperelastic materials, but because of the complexity of
Contributed by the Materials Division of ASME for publication in the JOURNAL OF
ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received May 26, 2010; nal
manuscript received December 31, 2010; published online March 23, 2011. Assoc.
Editor: Hussein Zbib.
Journal of Engineering Materials and Technology APRIL 2011, Vol. 133 / 021021-1
Copyright 2011 by ASME
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
obtaining materials, creating samples, and maintaining proper test
conditions this is often limited to uniaxial tensile or uniaxial com-
pression 9,10.
A transversely isotropic lamina oriented ber embedded in a
matrix can be characterized by a combination of uniaxial tension
in the ber direction with-grain, uniaxial tension in a direction
normal to the ber direction cross-grain, and a uniaxial tension
test at 45 deg to the ber direction 11. This approach was used to
characterize a ber reinforced elastomer for medium tensile
strains 0150% 12. The behavior of this elastomer in shear,
compression, and combined deformation states was not examined
and it would have to be used with caution based on the experience
of isotropic elastomer models that t only to uniaxial tensile data.
The desirability of using both uniaxial tensile and biaxial ten-
sile data to obtain a suitable material model has been described
13.
The tensile properties of typical v-belt neoprene elastomers
were measured in both the with-grain and cross-grain directions
1. The elastomers were reinforced using various bers cotton,
polyester, and cellulose and milled. The only tensile data reported
were the stress at 10% and 20% strains in both the with-grain and
cross-grain directions. The reported anisotropy ranged from 2.0 to
7.8.
Similar tensile data have been reported for various milled short
ber reinforced elastomers similar to those used in v-belts, hoses,
and tires 12,14,15. Tensile stress-strain curves for a milled short
melamine ber reinforced ethylene propylene diene monomer
EPDM rubber were reported in both the cross- and with-grain
directions 16. The anisotropy in mechanical properties was
shown to be more prominent at low strains than at large strains.
The uniaxial tensile and compression stresses in the ber and
cross ber directions were reported for a nylon ber reinforced
rubber compound used in an industrial v-belt 17.
The uniaxial tension results in the cross- and with-grain and
biaxial tension data for a calendared natural rubber have been
reported by Diani et al. 13.
A summary of these previous efforts is provided in Table 1.
Clearly these efforts have focused on uniaxial tension tests and
neglected the equally important compression, shear, and volumet-
ric tests.
A set of constitutive test data, similar to that provided by Tre-
loar for isotropic elastomers, is needed for anisotropic elastomers.
These data would allow the selection and development of accurate
and stable material models suitable for nite element analysis and
other uses. By combining the lessons learned from characterizing
isotropic elastomers and linear composite materials, the materials
should be characterized by tests of multiple modes of deformation
and in multiple directions e.g., 0 deg, 90 deg, and 45 deg.
Test samples of a suitable anisotropic elastomer were obtained
and tested using four different tests: uniaxial tension, uniaxial
compression, simple shear, and a conned compression test. For
each test, samples with different material orientations were tested.
These orientations were 0 deg, 45 deg, and 90 deg to the with-
grain direction of the material for tensile and simple shear speci-
mens and 0 deg and 90 deg for the compression and pure shear
samples.
The tests were done at the same strain rate to minimize the
effects of viscoelasticity and at the same temperature. In each
case, three samples were tested and the mean stress for the group
of samples is reported.
2.1 Elastomer. Two elastomers were mixed, milled, and
formed into test specimens. The elastomer formulation was based
on that used in an earlier published study of anisotropic elas-
tomers used in v-belts 1. The formula is given in Table 2. The
rst elastomer is an isotropic master batch LB1, which was then
used to create the second elastomer LB2 by adding bers. This
second elastomer LB2 is the anisotropic elastomer examined in
this study; the rst elastomer LB1 is the isotropic elastomer
matrix.
2.2 Uniaxial Tensile Tests. The material was tested in
uniaxial tension, as shown in Fig. 1. Samples were created with
three orientations: cross-grain, with-grain, and 45 deg to the with-
grain direction.
The mean stress results for three samples are shown in Fig. 2. It
can clearly be seen that the anisotropic elastomer has become
oriented with a different response in each material orientation. The
anisotropic elastomer has a much stiffer response than that of the
isotropic elastomer matrix.
Table 1 Summary of previous measurements of anisotropic elastomers
Date and Authors UT UC SS PS
BT VC. Orientation 0 deg 90 deg 45 deg 0 deg 90 deg 0 deg 90 deg 45 deg 0 deg 90 deg
1981 Rogers
1987 Foldi
1988 Lee
1994 van der Pol
2002 Rajeev et al.
2004 Diani et al.
2008 Ishikawa
Notes: UTuniaxial tension, UCuniaxial compression, SSsimple shear, PSpure shear, BTbiaxial tension, and VCvolumetric compression.
Table 2 Elastomer formulations
Ingredient
Compound LB1
phr
Compound LB2
phr
Neoprene 100
Carbon black 44
Process oil 4
Antioxidant 3
Magnesium oxide 4
Stearic acid 2
Curatives
LB1 100
Cotton ber 0.01 cm 35
Fig. 1 Tensile test
021021-2 / Vol. 133, APRIL 2011 Transactions of the ASME
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
The ratio of anisotropy was calculated and the results are shown
in Fig. 3. Over the 020% strain range the ratio was approxi-
mately 4 with a maximum of 4.5. This compares well with the
4.04.6 modular anisotropy reported for the similar compound
1. The ratio generally decreases at a larger strain as was reported
in earlier measurements.
There is a curious drop and rise in the ratio of anisotropy
around 5% strain that results from a decrease in the tangent modu-
lus of the with-grain response. This might be attributable to varia-
tions in ber curvature.
2.3 Uniaxial Compression Test. For the v-belt compound, a
dominant mode of deformation is compression between two metal
surfaces. Therefore, this test is of particular interest for this
investigation.
The uniaxial compression test is a relatively easy test that can
be done in an industrial laboratory. The basic test is to compress a
sample of material between two platens while measuring the de-
ection and load, as shown in Fig. 4. Here samples of the aniso-
tropic material were created with two orientations: the with-grain
oriented vertically V and the with-grain oriented horizontally
H. The horizontally oriented samples were created by stacking
layers with varying orientation; this creates a unique sample with
a cross-grain orientation through its thickness, but with no den-
able with-grain orientation.
The compression test remains a controversial choice due to ef-
fects of friction and the resultant tendency of the sample to bulge,
or barrel, during testing 4. This can be minimized by lubricating
the sample and using samples with a ratio of loaded area to force
free area of 1 4,7,8. For this study the samples were lubricated
disks with a 0.6 shape factor.
The mean stress results for the LB2 compound and the refer-
ence isotropic elastomer LB1 are shown in Fig. 5. The ratio of
anisotropy was calculated and is shown in Fig. 6. The ber rein-
forcement clearly increased the compressive stiffness of the ma-
terial, but the degree of anisotropy is not as strong as in tension.
At about 10% strain, the cross-grain H response is stiffer than
for the with-grain V samples. The ratio of anisotropy falls below
1. This is attributed to two factors. First, the change in tangent
stiffness of the with-grain sample is attributed to an effect similar
to buckling of the ber reinforcement in a composite lamina 11.
Second, the bers in the cross-grain H response are expected to
be in a tensile state, which resists the compression of the sample.
The overall increase in stiffness compared with the isotropic elas-
tomer is attributed to an effective particulate reinforcement of the
compound.
2.4 Dual Lap Shear Tests. The response of a hyperelastic
material to shear deformation is an important distinguishing char-
acteristic. Hyperelastic materials can be described as materials
with very large resistance to volumetric changes but a relatively
soft resistance to shear. Both linear elasticity and simple hyper-
elasticity neo-Hookean and Mooney are based on the assump-
tion that the shear stress will be proportional to the shear strain
Fig. 2 Tensile test results
Fig. 3 Tensile anisotropy
Fig. 4 Compression test
Fig. 5 Compression test results
Journal of Engineering Materials and Technology APRIL 2011, Vol. 133 / 021021-3
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
18.
A simple test for a simple shear is the dual lap shear test 4.
Typically this test is done with samples bonded to rigid xtures,
which results in an appreciable bending. For this study a modied
version of the test was done where samples are gripped between
high friction surfaces and the amount of deection is restrained, as
shown in Fig. 7. Asmall amount of compression is used to prevent
slippage; this introduces a small compression into the sample that
is neglected. Very small deections of thin test samples yield very
large shear deformations; a deection of 1 mm for a 3 mm thick
sample yields approximately 0.33 shear strain.
Material samples for the anisotropic elastomer were created
with three orientations: with-grain, cross-grain, and with the with-
grain direction oriented 45 deg from the test direction. The mean
shear stress results are shown in Fig. 8. It is clear that the shear
response of the anisotropic rubber is stiffer than the reference
isotropic elastomer matrix material and that the material orienta-
tion does not affect the shear response. The stiffness of all three
material orientations is essentially the same.
2.5 Conned Compression Tests. While very few materials
are truly incompressible, many elastomers can be considered
nearly incompressible. Does adding a compressible ber to a
nearly incompressible elastomer yield a nearly incompressible
material?
A nearly incompressible material can be dened as a material
for which a much larger effort is required to change its volume
than is required to change its shape while preserving its volume
19. The compressibility of a material is quantied by the bulk
modulus. For an incompressible material the bulk modulus would
be innite; for a nearly incompressible material the ratio of the
shear modulus to the bulk modulus is considered. For linear iso-
tropic materials the bulk modulus K, the shear modulus G, and
Poissons ratio are related 20 as follows:
=
3K 2G
6K + 2G
2
The bulk modulus can be measured using a conned compres-
sion test where a sample is compressed in one direction and re-
strained from expanding in all other directions 4. This method
has been found to be easier than trying to measure Poissons ratio
and relatively unaffected by small gaps between the sample and
the test xture 20.
The samples for this test are similar to those used in the
uniaxial compression test. Samples were created with two orien-
tations: with-grain V and cross-grain H. The cross-grain
samples were made as earlier described.
The constrained volume compression test results are shown in
Fig. 9. The measured bulk modulus, the initial shear modulus, the
Fig. 6 Compression anisotropy
Fig. 7 Dual lap shear test
Fig. 8 Dual lap shear test results
Fig. 9 Constrained volume compression test results
021021-4 / Vol. 133, APRIL 2011 Transactions of the ASME
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
ratio of the two, and the calculated Poissons ratio are given in
Table 3. The bulk modulus was determined from a linear least-
squares t over the nal linear strain range. The shear modulus
was estimated based on the initial slope of the shear stress data.
The anisotropic elastomer LB2 can be considered nearly in-
compressible because of the high ratio of the bulk modulus to
shear modulus. A rule of thumb for nearly incompressibility is
having Poissons ratio greater than 0.49 21. There does not ap-
pear to be a signicant difference between samples with a differ-
ent material orientation. The addition of the ber did not change
the compressibility of the material signicantly.
3 Constitutive Model
The general approach of using a strain energy function as the
basis for a constitutive model of a transversely isotropic material
subjected to a large elastic deformation was pioneered by Ericksen
and Rivlin 22, Spencer 23, and Green and Adkins 24.
In this effort, we restrict our attention to constitutive models
based on the approach of characterizing the strain energy as a
function of the invariants of the deformations. Other approaches
have been tried including linear transversely isotropic material
models 25 and generalized strain energy functions 13,2628.
What is needed is a stable efcient constitutive model suitable
for implementation in the nite element method. The nature of
current iterative nonlinear nite element methods is best suited to
models that do not violate the Drucker stability criteria. The ef-
ciency of the model is determined by the number of parameters
and the required material tests to accurately describe the material.
A suitable strain energy function was constructed, the constitu-
tive stress-strain relationships were derived, and the parameters
determined from the material tests.
3.1 Strain Energy Function. The essence of elasticity is in
dening how a material resists applied forces and deformation.
Under load the material deforms, storing the work being applied
as internal elastic energy strain energy, and creating internal
stresses that react any applied loads. The constitutive relationship
between the deformation strain and the internal stresses allows
the denition of a strain energy function of the form 19,24,29
= C 3
where the deformation is described by the right CauchyGreen
tensor C, which is dened as
C = F
T
F 4
in terms of the deformation gradient F as follows:
FX =
xX
X
5
The deformed geometry is described by x and the initial geometry
is described by X. It is convenient to work with the deformation
gradient and the right CauchyGreen tensor in terms of their ei-
genvalues that are the principal stretches the ratio of the de-
formed length to the undeformed length in the three orthogonal
directions
1
,
2
, and
3

F =

1
0 0
0
2
0
0 0
3

6
C =

1
2
0 0
0
2
2
0
0 0
3
2

7
The change in volume of the material can be calculated using
the Jacobian J
J = F =

C 8
For materials that are transversely isotropic the strain energy
function can be dened as a function of the three invariants of the
right CauchyGreen tensor and two additional quasi-invariants
based on the orientation of the material a
o
2224
= I
1
, I
2
, I
3
, I
4
, I
5

I
1
= trC
I
2
=
1
2
trC
2
trC
2

I
3
= C
I
4
= a
o
Ca
o
=
f
2
I
5
= a
o
C
2
a
0
=
f
4
9
A subset of all functions that meet the general denition is
those functions that split the strain energy into two components:
The rst is the energy stored in the deformation of the matrix
material and the second is that energy stored in the reinforcement
=
iso
I
1
, I
2
, I
3
+
f
I
4
, I
5
10
If the matrix is incompressible, then I
3
becomes 1 and
iso
becomes a function of I
1
and I
2
.
Based on the results of the simple shear tests where there is no
effect of material orientation the uncoupling of the ber and ma-
trix strain energy functions seems appropriate.
For the contribution of the ber, several models with various
degrees of success have been proposed. These models are summa-
rized in Table 4 where they are ranked by the number of material
constants required efciency and the order of the ber strain
energy function indicated. These models combine existing hyper-
elastic models with different reinforcing functions.
The difculty in using the existing strain energy functions to
describe the with-grain compressive and tensile nature of the sub-
ject elastomer LB2 is shown in Fig. 10. Additionally, the func-
tion used by Ishikawa et al. shows an undesirable elastic instabil-
ity elastic stress decreases with increasing strain. A better strain
energy function is needed.
3.1.1 Strain Energy of the Elastomer Matrix. For the matrix
contribution the MooneyRivlin or the neo-Hookean elastomer
model is well suited for small to moderate strains as expected for
the following material 3,18:
Table 3 Material compressibility
Material and
orientation
Bulk modulus
MPa
Initial shear
modulus MPa
Ratio of bulk
to shear
modulus
Calculated
Poissons ratio
LB2-0 V 900 5.3 170 0.497
LB2-90 H 900 5.1 176 0.497
LB1-0 V 890 3.2 278 0.498
LB1-90 H 870 3.3 264 0.498
Journal of Engineering Materials and Technology APRIL 2011, Vol. 133 / 021021-5
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms

iso
I
1
, I
2
= C
1
I
1
3 + C
2
I
2
3 11
3.1.2 Strain Energy of the Fiber Reinforcement. Many of the
earlier strain energy functions have been used to describe the re-
sponse of biological materials, which have a characteristic highly
nonlinear response over very large strains 30. They use an ex-
ponential function for the ber strain energy contribution. This
seems very different from the response of a reinforcing ber under
much smaller strains.
The anisotropic elastomer LB2 has been reinforced with small
bers that have an initial curvature. Even after milling, curvature
is expected. Consider the nite element of a curved ber, similar
to that used in the elastomer LB2, as shown in Fig. 11. The ex-
ternal work required to extend and compress this ber model is
shown in Fig. 12.
Relatively little strain energy is stored in compression of the
ber as the degree of curvature increases. As the ber straightens,
the strain energy becomes proportional to I
4
. As shown in Fig. 12,
this response differs substantially from the characteristic response
of the types of reinforcing functions used previously. The follow-
ing strain energy function ts the bers response more accu-
rately:

f
I
4
, I
5
= C
4
I
4
1/2
1
2
C
4
=
C
4t
ifI
4
1
C
4c
ifI
4
1
12
3.1.3 New Strain Energy Function. Combining the two con-
stituent strain energy functions, given in Eqs. 11 and 12, yields
a suitable strain energy function for small to moderate strains of a
transversely isotropic hyperelastic material
= C
1
I
1
3 + C
2
I
2
3 + C
4
I
4
1/2
1
2
13
This strain energy function meets two essential criteria for a
strain energy function. First, if there is no deformation I
1
=I
2
=3
and I
4
=1 there is no stored elastic energy and the strain energy
function must be zero. Second, the strain energy should always be
Table 4 Transversely isotropic invariant form strain energy functions
Strain energy functions Rank Order
Standard reinforcing model 35:
=C
1
I
1
3 +C
2
I
4
1
2
2 4
Fiber reinforced cardiac muscle 36:
=C
1
expC
2
I
1
3
2
+C
3
I
1
3I
4
1 +C
4
I
4
1
2
1
4 E
Caterpillar muscle 38:
= C
1
I
1
3 + C
2
1 C
3
tanh
I
4
1
C
4

expC
5
I
4
1
2
1
5 E
Fiber reinforced rubber 17:
=C
10
I
1
3 +C
01
I
2
3 +C
42
I
4
1
2
+C
43
I
4
1
3
C
42
and C
43
have different values for tension and compression.
6 6
Human medial collateral ligament 33:
=C
1
I
1
3 +C
2
I
2
3 +C
3
expI
4
1 I
4

More complex function C


3
, C
4
, C
5
, and C
6
was used to capture ber under three states.
8 E
Cord-rubber composite 37:
=V
m
C
1
I
1
3 +C
2
I
2
3 +V
C
C
3
expI
4
1 I
4
+C
5
I
1
3 +C
6
I
1
3
2
9 E
Fig. 10 Comparison of models and measured results for WG
uniaxial stress Fig. 11 Finite element model of short ber
021021-6 / Vol. 133, APRIL 2011 Transactions of the ASME
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
positive under deformation. The latter is satised if the signs of all
the constants are positive. Additionally, this strain energy function
could be combined with other reinforcing functions particularly
those using I
4
based polynomials.
3.2 Stress Tensor. In this paper we will consider the second
PiolaKirchoff PK stress and the transformation of this stress
into the rst PiolaKirchoff stress related traction force vector
force measured per undeformed unit surface area. This choice
allows the comparisons of the measured engineering stress and
strain to those calculated in this section. The second Piola
Kirchoff stress can be determined from the strain energy function
19
S = 2

C
= 2

i=1
5

I
i
I
i
C
14
The partials of the invariants with respect to the right Cauchy
Green tensor are 19
I
1
C
= I 15
I
2
C
= I
1
I C 16
I
3
C
= I
3
C
1
= C
1
for an incompressible material 17
I
4
C
= a
o
a
o
18
I
5
C
= a
o
Ca
o
+ a
o
C a
o
19
The partials of the proposed strain energy function with respect
to the invariants are

I
1
=
1
= C
1
20

I
2
=
2
= C
2
21

I
3
=
3
= 0 22

I
4
=
4
= C
4
1 I
4
1/2
23

I
5
=
5
= 0 24
Combining Eqs. 1424 yields the stress function for the pro-
posed strain energy function
S = 2C
1
+ C
2
I
1
I C + C
4
1 I
4
1/2
a
o
a
o
pC
1
25
The second PiolaKirchoff stress can be transformed into the
rst PiolaKirchoff stress 19
P = FS 26
3.3 Uniaxial Tension and Compression
3.3.1 With-Grain Uniaxial Stress. In this case the material is
stretched along its oriented direction a
o
, as illustrated in Fig. 13.
In the isotropic plane normal to this orientation, the material
contracts equally in both directions. Assuming the material is in-
compressible yields the following deformation gradient:
F
WGU
=

0 0
0
1/2
0
0 0
1/2

, a
o

1
0
0

27
Using Eqs. 2527, the nominal traction force required to
stretch the material will be
T
WGU
= 2C
1

2
+ 2C
2
1
3
+ 2C
4
1 28
3.3.2 Cross-Grain Uniaxial Stress. The material is stretched in
a direction perpendicular to the material orientation, as illustrated
in Fig. 13. Because of the difference in stiffness in the remaining
two principal directions, the stretches in each of these directions
are not the same 13. Assuming incompressibility the third stretch
can be expressed in terms of the rst two as follows:
F
XGU
=

1
0 0
0
2
0
0 0
1
1

2
1

, a
o
=

0
1
0

29
Noting that two of the three principal stresses are zero those in
the contracting directions, Eqs. 25 and 29 yield three equa-
tions with three unknowns S
11
, p, and
2
. Using Eq. 26 gives
the following equations for the traction force:
T
XGU
= 2C
1

1
1

2
2
+ 2C
2

1
1

2
2

1
3
+ 2C
4

1
1

1
1

2
2

30
C
1
+ C
4

1
2
+ C
2

1
4

2
4
C
4

1
2

2
3
C
2

1
2
+ C
1
= 0 31
While a closed form solution for
2
was found, it is more prac-
tical to use a numerical method to solve Eq. 31 for a given
1
.
Fig. 12 Comparison of reinforcing functions with predicted re-
sults for short ber
Fig. 13 Uniaxial deformation
Journal of Engineering Materials and Technology APRIL 2011, Vol. 133 / 021021-7
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
3.4 Simple Shear. The case of simple shear is shown in Fig.
14.
3.4.1 With-Grain Simple Shear. The shear motion is parallel
to the material orientation. The deformation is intentionally isoch-
oric. The resultant deformation gradient is
F
WGS
=

1 0
0 1 0
0 0 1

, a
o
=

1
0
0

32
Using Eqs. 25, 26, and 32, the nominal traction force is
T
WGS
= 2C
1
+ C
2
33
3.4.2 Cross-Grain Simple Shear. The deformation is similar to
the with-grain simple shear. The difference is that the material
orientation is normal to the plane of shearing. In this case the
deformation gradient and material orientation are
F
XGS
=

1 0
0 1 0
0 0 1

, a
o
=

0
0
1

34
Using Eqs. 25, 26, and 34, the nominal traction force is
T
XGS
= 2C
1
+ C
2
35
It is interesting to note that the orientation of the material does
not affect the shear response. Both T
WGS
and T
XGS
are the same.
3.5 Plane Strain (Pure Shear). In his earlier work, Treloar
2 used the pure shear test to test the validity of the emerging
theory of what is now known as the neo-Hookean elastomer
model. The test was done by stretching a short wide sample 5
75 mm
2
. This establishes a state of plane strain in the sample
by constraining the width. The third direction, through the thick-
ness, contracts. It is also known as a constrained tension state
31. The deformation in the pure shear test is considered to be
equivalent to that in a simple shear test 7,30. The principal
stretch can be related to the simple shear strain 29 as
follows:
=
1

36
This state of plane strain is often referred to in the hyperelas-
ticity literature as pure shear. The case is illustrated in Fig. 15.
3.5.1 With-Grain Plane Strain. In this case the material is
stretched in the orientation direction. Assuming the material is
incompressible results in the contraction in the remaining free
direction to be the inverse of the applied stretch. The deformation
gradient and the material orientation are
F
WGPS
=

0 0
0 1 0
0 0
1

, a
o
=

1
0
0

37
Using Eqs. 25, 26, and 37, the nominal traction force is
T
WGPS
= 2C
1
+ C
2

3
+ 2C
4
1 38
3.5.2 Cross-Grain Plane Strain. In this case the material is
restrained from contracting in the orientation direction and
stretched in a direction perpendicular to the orientation direction.
Assuming the material is incompressible results in the contraction
in the remaining free direction to be the inverse of the applied
stretch. The deformation gradient and the material orientation are
F
XGPS
=

0 0
0 1 0
0 0
1

, a
o
=

0
1
0

39
Using Eqs. 25, 26, and 39, the nominal traction force is
T
XGPS
= 2C
1
+ C
2

3
40
The material orientation had a much larger effect on the pure
shear test than on the simple shear test. This difference suggests
that the two modes of shear deformation are not equivalent as they
are for an isotropic elastomer. This is attributed to the difference
in the stiffening effect between shear and extension predicted by
composites micromechanics theory for short bers in a matrix
11 where the reinforcement is not as effective in increasing the
shear modulus of the composite.
3.6 Determining Model Coefcients. As was earlier stated,
if the material moduli C
1
, C
2
, C
4c
, and C
4t
are all positive, then
the strain energy function is positive for all deformations. This
also results in stable stress-strain relationships. If the material
moduli are positive, then stress always increases with increasing
strain tangent stiffness is always positive. This satises the
Drucker stability criteria for hyperelastic materials.
The material moduli C
1
, C
2
, C
4c
, and C
4t
can be determined by
minimizing the differences between the measured force per unit
undeformed area and the calculated nominal traction force for the
same measured strain. This problem is well suited to a least-
squares method to minimize the difference.
This was done using the commercial spreadsheet program Mi-
crosoft Excel. The relationships in Secs. 3.13.6 were pro-
grammed as user functions, and the built in solver was used to
minimize the difference between calculated and measured values.
The resulting t of the model to the test data is shown in Fig.
16. The data were tted over the range 10% because of the
limits of the with-grain tensile data. The t did not include the
cross-grain compression samples because of their unique
construction.
The resulting t is quite good for all modes of deformation in
the range of the t and in many cases outside of the t. The most
signicant difference would seem to be the calculated shear re-
sponse. The difference in the with-grain tensile response is be-
lieved to be related to failure of the material. The difference in the
with-grain tensile compression response is interesting as there is a
marked change in the response that occurs after 10% compres-
sion; this will be examined in more detail in Sec. 4. The actual
Fig. 14 Simple shear deformation
Fig. 15 Pure shear plane strain deformation
021021-8 / Vol. 133, APRIL 2011 Transactions of the ASME
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
cross-grain compression response is stiffer than calculated; this is
believed to be the effect of the unique sample construction, which
will also be examined in Sec. 4.
The nature of the constitutive relationships suggests an alterna-
tive method of tting the data than spreading the error over all
the results implicit in the above method. This method focuses on
determining the material constants from specic tests. Similar al-
ternative methods have been used 17,32. The following methods
could be used:
1. determine the shear modulus 2C
1
+2C
2
from simple shear
T
WGS
or cross-grain pure shear T
XGPS
tests
2. determine C
4t
from the with-grain tensile response T
WGU

or the with-grain planar shear test T


WGPS

3. determine C
4c
from the with-grain compression response
T
WGU

4. determine a value of C
2
that improves the larger strain re-
sponse of the tensile and compression data; this is not nec-
essary if C
1
alone results in a good t
These methods were also used and the comparison of this t
with the experimental data is shown in Fig. 17.
While this second t has a larger error than the rst t, it high-
lights the strengths and weaknesses of the material model. The t
with the simple shear results is improved while the with-grain
tensile and compression ts remain good. A remarkable difference
is that the cross-grain compression result is softer than the test
data this difference is not considered in either t because of the
unique construction of these samples. The cross-grain tensile re-
sponse is softer and this will be investigated further in Sec. 4.
4 Implementation Within the Finite Element Method
The new invariant based strain energy hyperelastic model was
implemented within the nite element method using the user sub-
routine hypela2 available in the commercial nite element code
MSC.MARC. Finite element models of the test specimens were cre-
ated and used to investigate in more detail the response of the
material including the samples where the material was oriented at
45 deg and for the unique construction of the cross-grain compres-
sion samples.
4.1 Nearly Incompressible Form. In the analysis of nearly
incompressible materials the nite element method can be dif-
cult. These difculties are described by Weiss et al. 33. A
method that improves this is to decouple the deviatoric and dila-
tional responses of the material and consequently to separate the
strain energy function into two components. First the deformation
gradient and the right CauchyGreen tensor can be decoupled 19
as follows:
F = F
vol
F, F
vol
= J
1/3
1, F = J
1/3
F
F
vol
= J, F = 1 41
C = F
T
F = J
2/3
C, C = 1 42
Using the modied deformation gradient F and the invariants
of the modied right CauchyGreen tensor C, the strain energy
function can be split into volumetric and isochoric parts 19
=
vol
J +
iso
I
1
, I
2
, I
4
, I
5
43
The second PiolaKirchoff stress can now be calculated as 19
S = JpC
1
+ J
2/3
P:S
P = I
1
3
C
1
C
S =

1
1 +

2
C +

4
a
o
a
o
+

5
a
o
Ca
o
+ a
o
C a
o

1
= 2

iso
I
1
+ I
1

iso
I
2

2
= 2

iso
I
2

4
= 2

iso
I
4

5
= 2

iso
I
5
44
Using the incompressible form of the strain energy function in
Eq. 13 as
iso
, and the bulk modulus measured in Sec. 2.4, the
second PiolaKirchoff stress can then be calculated from Eq. 44
using the following:

1
= 2C
1
+ I
1
C
2
45

2
= 2C
2
46

4
= 2C
4
1 I
4
1/2
47

5
= 0 48
p = KJ 1 49
4.2 Elasticity Tensor. The methods to solve nonlinear hyper-
elasticity problems depend heavily on iterative incremental tech-
niques that are based on linearizing the material stiffness at a
Fig. 16 Minimized difference material model t
Fig. 17 Deterministic material model t
Journal of Engineering Materials and Technology APRIL 2011, Vol. 133 / 021021-9
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
given state. Within the hypela2 subroutine, MSC.MARC requires the
output of the tangent material stiffness matrix. The rate of conver-
gence, or even obtaining convergence, depends on this stiffness
34.
The tangent material stiffness matrix for a hyperelastic material
in 3 dimensions can be visualized as being dened as the change
in 6 stresses to an incremental change in the 6 strains and is
comprised of 36 separate components relating the change in each
stress to a change in each strain. Given the constitutive relation-
ship between stress and strain and the hyperelastic strain energy
function, the change in stress is related to the change in strain
energy resulting from any change in deformation. This is more
concisely expressed using the fourth order elasticity tensor C
19 as follows:
C = 4

2

C C
50
The calculation of the 21 independent components of the elas-
ticity tensor using a continuum mechanics approach can be done
using the following equations 19:
C = C
vol
+ C
iso
C
vol
= Jp C
1
C
1
2pC
1
C
1
C
iso
= P:C:P
T
+
2
3
trJ
2/3
SP

2
3
C
1
S
iso
+ S
iso
C
1

C
1
C
1

abcd

1
2
C
ac
1
C
bd
1
+ C
ad
1
C
bc
1
51
For the strain energy function used here, Eq. 13, and using the
Kronecker delta
ij
the C tensor would be
C = J
4/3

1
1 1 +
4
S +
7
a
o
a
o
a
o
a
o

S =
1
2
I + I
I
ijkl

ik

jl
I
ijkl
=
il

jk
52
The coefcients
1
,
4
, and
7
in Eq. 52 are 19

1
= 4

iso
I
1
I
1
+ I
1

iso
I
1
I
2
+

iso
I
2
+ I
1
2

2

iso
I
2
I
2

= 4C
2
53

4
= 4

iso
I
2
= 4C
2
54

7
= 4

2

iso
I
4
I
4
= 2C
4
I
4
3/2
55
4.3 Finite Element Analysis of Uniaxial Tension Tests. A
nite element model of the tensile specimen used in Sec. 2.2 was
created and solutions obtained using the subroutine. The second
set of material coefcients, the deterministic t, was used.
The nite element model and the strain energy contours are
shown in Figs. 18 and 19. The engineering stress was calculated
from the reaction forces and plotted against the measured results
in Fig. 20. The analysis was done for several material orientations
0 deg, 15 deg, 30 deg, 45 deg, and 90 deg
The calculated stresses for the with-grain samples are very
close to the measured stresses. The calculated stresses for the
cross-grain and 45 deg orientation were less than those measured
in the test.
Suspecting that this difference could be attributed to a distribu-
tion of ber orientation rather than the single orientation modeled,
a distributed ber orientation nite element model was con-
structed. The orientation was assumed to be a normal distribution
of orientation between the limits of 90 deg from the with-grain
direction. The distribution was divided into three sections and the
thickness of each section was proportional to the cumulative dis-
tribution for the section. This model is shown in Fig. 21.
The resulting calculated stress and strain are compared with the
measured values in Fig. 22. While there is a considerable im-
provement, particularly at strains less than 10%, there are still
differences. Using a more accurate denition of the distributed
orientations of the actual elastomer combined with more laminas
in the model might improve the comparison further.
It is also apparent that there is a point or a strain where the
Fig. 18 Tensile test nite element model
Fig. 19 Tensile FEA second PK stress contours MPa
Fig. 20 Tensile FEA results
021021-10 / Vol. 133, APRIL 2011 Transactions of the ASME
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
materials with-grain stress separates from the predicted response.
Rather than stiffening as occurs with biotissues, there is a soften-
ing of the material. This is attributed to a failure in the ber/matrix
interface and is believed to represent the beginning of an inelastic,
or damaged, response. This effect is also expected for cross-grain
and other material orientations.
4.4 Uniaxial Compression. Finite element models of the
compression tests were created and solutions obtained using the
subroutine. Again, the second set of material coefcients was
used. The cross-grain compression model was created using a
quasi-isotropic laminate 0 deg/45 deg/45 deg/90 deg to simu-
late the unique construction of the cross-grain H samples created
by randomly orienting layers of elastomer in the compression
mold.
The compression test nite element model is shown in Fig. 23.
Models were created for each of the two orientations tested plus
another cross-grain sample without the varying orientation. The
engineering stress was calculated from the reaction forces and
plotted against the measured results in Fig. 24.
The deformed shape and stress contours of the with-grain com-
pression sample V are shown in Fig. 25. The calculated stresses
are very close to the measured results through 15% compression.
It is interesting to note that at near 25% compression the stiffness
dramatically changes and the nite element stiffness matrix be-
comes nonpositive denite. Forcing the solution conrms the
change, and is indicative of a buckling of the specimen. These
results are very consistent with the change in stiffness measured
during the actual test.
The laminated structure, the stress contours, and the deformed
shape of the H compression sample are shown in Fig. 26. The
nite element results for the cross-grain compression samples are
closer to the measured results than that predicted by Eqs. 30 and
31. This is due to the sample being stiffened by the additional
laminate constraints that act to keep the radial deformation
uniform.
The deformed shape and stress contours for a cross-grain com-
pression sample are shown in Fig. 27. This model demonstrates
the difference in deformation of the with-grain direction from the
cross-grain response the ovalization of the model. The calculated
stresses compare well with those calculated using Eqs. 30 and
31.
4.5 Simple Shear Finite Element Model. Finite element
models of the simple shear test were created and solutions ob-
tained for each of the three material orientations tested. The nite
Fig. 21 Distributed ber orientation nite element model
Fig. 22 Distributed ber FEA results
Fig. 23 Compression nite element model
Fig. 24 Compression FEA results
Fig. 25 V Compression FEA second PK stress contours MPa
Journal of Engineering Materials and Technology APRIL 2011, Vol. 133 / 021021-11
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
element model undeformed and deformed is shown in Fig. 28.
The calculated engineering stress is compared with the measured
engineering stress in Fig. 29.
As expected the shear response matches very closely the ex-
perimental results; the shear response was the foundation of the
deterministic t used. The separation that is evident starting with
about 10% shear stress may be the result of either a nonlinear
shear response or the beginning of slipping of the samples within
the test xture. A known characteristic of the two term Mooney
Rivlin model is its inherent linear shear response.
The deformed shape and stress contours for each of the three
orientations are shown in Fig. 30. It is interesting to note that
despite the similarity in the overall stress results the surface con-
tours vary. However, if we look at the stress contours in the inte-
rior of the models, we nd very similar stress contours, as shown
in Fig. 31.
5 Conclusions
The combination of uniaxial tension, compression, simple
shear, and volumetric compression tests provides a very thorough
description of the response of a transversely isotropic elastomer to
various modes of deformation. These data ll gaps in previously
published data for similar materials. The data should provide a
reasonable basis to test the accuracy of proposed models to a
wider range of deformation types.
The strain energy function proposed here, and implemented in
the nite element model, is well suited for use by the practicing
engineer in analyzing parts comprised of this type of ber rein-
forced elastomer over a small to moderate strain range 030%. It
is efcient, requiring only three material moduli in its simplest
form, and inherently stable as long as positive moduli are used.
These material constants can be determined from a few simple
laboratory tests e.g., tension, compression, and simple shear.
Fig. 26 H Compression FEA second PK stress contours MPa
and laminated structure
Fig. 27 XG Compression second PK stress contours MPa
Fig. 28 Simple shear nite element model
Fig. 29 Simple shear FEA results
Fig. 30 Simple shear FEA second PK stress contours MPa
Fig. 31 Simple Shear FEA internal second PK stress contours
MPa
021021-12 / Vol. 133, APRIL 2011 Transactions of the ASME
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms
If the material has a signicant ber orientation distribution,
then a laminated approach is recommended. This should improve
the accuracy of the model for cross-grain tensile states. Alterna-
tively, using a minimized overall error t will also improve the
general accuracy of the model.
As indicated by the successful use of the model within the
context of a laminated structure to model the unique structure of
the cross-grain test specimens used here, this simple model can be
used effectively to simulate the effects of more complex lami-
nates.
A plane strain constitutive test for this type of elastomer would
be very useful. This will require a very capable clamping system
to handle the relatively larger samples with this type of substantial
reinforcement. The force required to stretch a 200 mm wide rein-
forced rubber is much greater than expected for a typical isotropic
elastomer.
Further work on incorporating the effects of ber distribution
should improve the material model. This is not an inherent prob-
lem with the transversely isotropic model but rather requires the
extension of this simple model to incorporate layers of a dis-
tributed orientation.
This study did not consider strain rate effects viscoelasticity
or repeated loading effects Mullins effect on the material behav-
ior. Future research on incorporating these effects would be useful
3840.
Acknowledgment
We would like to thank the Gates Corporation for supporting
this research, the staff of the Advanced Materials Development
Group at the Gates Material Center in Columbia, MO, for mixing
and preparing the test samples, and Randy Roller who performed
the tests at the Gates Technical Center in Rochester Hills, MI.
References
1 Rogers, J. W., 1981, The Use of Fibers in V-Belt Compounds, Rubber World
March, pp. 2731.
2 Treloar, L. R. G., 1944, Stress-Strain Data for Vulcanised Rubber Under
Various Types of Deformation, Trans. Faraday Soc., 40, pp. 5970.
3 Marckmann, G., and Verron, E., 2006, Comparison of Hyperelastic Models
for Rubber-Like Materials, Rubber Chem. Technol., 79, pp. 835858.
4 Brown, R., 2006, Physical Testing of Rubber, Springer ScienceMedia Inc.,
New York.
5 Peeters, F. J. H., and Kussner, M., 1999, Material Law Selection in the Finite
Element Simulation of Rubber-Like Materials and Its Practical Application in
the Industrial Design Process, Constitutive Models for Rubber, A. Dorfmann
and A. Muhr, eds., Balkema, Rotterdam, pp. 2936.
6 Finney, R. H., and Kumar, A., 1988, Development of Material Constants for
Nonlinear Finite-Element Analysis, Rubber Chem. Technol., 61, pp. 879
891.
7 Charlton, D. J., Yang, J., and The, K. K., 1994, A Review of Methods to
Characterize Rubber Elastic Behavior for Use in Finite Element Analysis,
Rubber Chem. Technol., 673, pp. 481503.
8 Kim, W.-D., Kim, W.-S., Woo, C.-S., and Lee, H.-J., 2004, Some Consider-
ations on Mechanical Testing Methods of Rubber Materials Using Nonlinear
Finite Element Analysis, Polym. Int., 53, pp. 850856.
9 Holzapfel, G. A., 2005, Similarities Between Soft Biological Tissues and
Rubberlike Materials, Constitutive Models for Rubber IV, P.-E. Austrell and
L. Kari, eds., Taylor & Francis, London, pp. 607617.
10 Krouskop, T. A., Wheeler, T. M., Kallel, F., Garra, B. S., and Hall, T., 1998,
Elastic Moduli of Breast and Prostate Tissues Under Compression, Ultrason.
Imaging, 20, pp. 260274.
11 Jones, R. M., 1975, Mechanics of Composite Materials, Scripta Book Com-
pany, Washington, D.C.
12 Lee, M. C. H., 1988, The Mechanical Properties and Fractural Morphology of
Unidirectional Short-Fiber Reinforced Polychloroprene Composites, J.
Polym. Eng., 834, pp. 257282.
13 Diani, J., Brieu, M., Vacherand, J.-M., and Rezgui, A., 2004, Directional
Model for Isotropic and Anisotropic Hyperelastic Rubber-Like Materials,
Mech. Mater., 36, pp. 313321.
14 van der Pol, J. F., 1994, Short Para Aramid Fiber Reinforcement, Rubber
World June, pp. 3237.
15 Foldi, A. P., 1987, Reinforcement of Rubber Compounds With Short, Indi-
vidual Fibers, Rubber World May, pp. 1926.
16 Rajeev, R. S., Bhowmick, A. K., De, S. K., Kao, G. J. P., and Bandyopadhyay,
S., 2002, New Composites Based on Short Melamine Fiber Reinforced
EPDM Rubber, Polym. Compos., 234, pp. 574591.
17 Ishikawa, S., Tokuda, A., and Kotera, H., 2008, Numerical Simulation for
Fiber Reinforced Rubber, Journal of Computational Science and Technology,
24, pp. 587596.
18 Mooney, M., 1940, A Theory of Large Elastic Deformation, J. Appl. Phys.,
11, pp. 582592.
19 Holzapfel, G. A., 2006, Nonlinear Solid Mechanics: A Continuum Approach
for Engineering, Wiley, Chichester.
20 Peng, S. H., Shimbori, T., and Naderi, A., 1994, Measurement of Elastomers
Bulk Modulus by Means of a Conned Compression Test, Rubber Chem.
Technol., 67, pp. 871879.
21 MARC Analysis Research Corporation, 1996, Nonlinear Finite Element
Analysis of Elastomers, MARCAnalysis Research Corporation, Palo Alto, CA.
22 Ericksen, J. L., and Rivlin, R. S., 1954, Large Elastic Deformations of Ho-
mogeneous Anisotropic Materials, Journal of Rational Mechanics and Analy-
sis, 3, pp. 281301.
23 Spencer, A. J. M., 1984, Continuum Theory of the Mechanics of Fiber-
Reinforced Composites, Springer, New York.
24 Green, A. E., and Adkins, J. E., 1970, Large Elastic Deformations, Oxford
University Press, Belfast, UK.
25 Moghe, S. R., 1974, Short Fiber Reinforcement of Elastomers, Rubber
Chem. Technol., 47, pp. 10741081.
26 Itskov, M., and Aksel, N., 2004, A Class of Orthotropic and Transversely
Isotropic Hyperelastic Constitutive Models Based on a Polyconvex Strain En-
ergy Function, Int. J. Solids Struct., 41, pp. 38333848.
27 Itskov, M., Aksel, N., and Ehret, A., 2003, Constitutive Modelling of Calen-
daring Induced Anisotropy in Rubber Sheets, Constitutive Models for Rubber
III, J. J. C. Buseld and A. H. Muhr, eds., Sets & Zeitlinger, Lisse, The
Netherlands, pp. 401404.
28 Shariff, M. H. B. M., 2008, Transversely Isotropic Strain Energy With Physi-
cal Invariants, Constitutive Models for Rubber V, Taylor & Francis, London,
pp. 6772.
29 Ogden, R. W., 1997, Non-Linear Elastic Deformations, Dover, Mineola, NY.
30 Fung, Y. C., 1994, A First Course in Continuum Mechanics for Physical and
Biological Engineers and Scientists, Prentice-Hall, Englewood Cliffs, NJ.
31 Gent, A. N., 1992, Elasticity, Engineering With Rubber How to Design
Rubber Components, A. N. Gent, ed., Carl Hanser, Munich, Germany, pp.
3366.
32 Peng, X. Q., Guo, Z. Y., and Moran, B., 2006, An Anisotropic Hyperelastic
Constitutive Model With Fiber Matrix Shear Interaction for the Human Annu-
lus Fibrosis, ASME J. Appl. Mech., 73, pp. 815824.
33 Weiss, J. A., Maker, B. N., and Govindjee, S., 1996, Finite Element Imple-
mentation of Incompressible, Transversely Isotropic Hyperelasticity, Comput.
Methods Appl. Mech. Eng., 135, pp. 107128.
34 MSC Software, 2008, MARC, Vol. D, User Subroutines and Special Routines.
35 Qiu, G. Y., and Pence, T. J., 1997, Loss of Ellipticity in Plane Deformation of
a Simple Directionally Reinforced Incompressible Nonlinearly Elastic Solid,
J. Elast., 49, pp. 3163.
36 Ishikawa, S., and Kotera, H., 2005, Constitutive Equations for Fiber Rein-
forced Hyperelasticity, Constitutive Models for Rubber IV, P.-E. Austrell and
L. Kari, eds., Taylor & Francis, London, pp. 619625.
37 Zhang, F., Fan, Z., Du, X., and Kuang, Z., 2004, Study on Constitutive Model
and Failure Criterion of Cord-Rubber Composite, J. Elastomers Plast., 36,
pp. 351362.
38 Dorfmann, A. L., Woods, W. A., Jr., and Trimmer, B. A., 2008, Muscle
Performance in a Soft-Bodied Terrestrial Crawler: Constitutive Modeling of
Strain-Rate Dependency, J. R. Soc., Interface, 5, pp. 349362.
39 Pea, E., and Doblar, M., 2009, An Anisotropic Pseudo-Elastic Approach for
Modeling Mullins Effect in Fibrous Biological Materials, Mech. Res. Com-
mun., 36, pp. 784790.
40 Horgan, C. O., Ogden, R. W., and Saccomandi, G., 2004, A Theory of Stress
Softening of Elastomers Based on Finite Chain Extensibility, Proc. R. Soc.
London, Ser. A, 460, pp. 17371754.
Journal of Engineering Materials and Technology APRIL 2011, Vol. 133 / 021021-13
Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 11/25/2013 Terms of Use: http://asme.org/terms

Vous aimerez peut-être aussi