Está en la página 1de 31

Strain controlled biaxial stretch: An experimental

characterization of natural rubber

by

Francesco Q. Pancheri
Luis Dorfmann

OCCAM Preprint Number 12/105


Strain controlled biaxial stretch:
An experimental characterization of natural rubber
Francesco Q. Pancheria , Luis Dorfmannb,c,∗
a
Department of Mechanical Engineering
Tufts University, Medford, MA 02155, USA
b
Department of Civil and Environmental Engineering
Tufts University, Medford, MA 02155, USA
c
OCCAM, Mathematical Institute
University of Oxford, Oxford OX1 3LB, UK

Abstract
In this paper we provide new experimental data showing the response of 40A
natural rubber in uniaxial, pure shear and biaxial tension. Real-time biaxial
strain control allows for independent and automatic variation of the velocity
of extension and retraction of each actuator to maintain the pre-selected
deformation rate within the gage area of the specimen. The remaining part
of the paper focuses on the Valanis-Landel hypothesis that is used to verify
and validate the consistency of the data. We use a three term Ogden model to
derive stress-stretch relations to validate the experimental data. The material
model parameters are determined using the primary loading path in uniaxial
and equibiaxial tension. Excellent agreement is found when the model is
used to predict the response in biaxial tension for different maximum in-plane
stretches. The application of the Valanis-Landel hypothesis also results in
excellent agreement with the theoretical prediction.

Keywords: Experimental characterization; Natural rubber; Finite deforma-


tions; Planar biaxial extension; Valanis-Landel hypothesis


Corresponding author
Email address: Luis.Dorfmann@tufts.edu (Luis Dorfmann)
URL: http://ase.tufts.edu/msml/ (Luis Dorfmann)

Preprint submitted to Rubber Chemistry and Technology November 28, 2012


1. Introduction
Rubber-like materials are widely used in industry to manufacture tires,
seals, belts, engine mounts and many other devices. To optimize the design of
these products for complex mechanical loading, in general, numerical meth-
ods are used in combination with accurate constitutive laws. The objective
of this paper is to summarize uniaxial and biaxial testing and to provide cor-
responding data of a natural rubber compound. The testing machine used
to characterize the material is equipped with a control mechanism to inde-
pendently adjust the movement of each of the four actuators. The control
mechanism includes real time strain control, which produces pure homoge-
neous deformation in the gage region by stretching the material specimen in
two mutually orthogonal directions to pre-selected target stretches. We ver-
ify and validate the consistency of the data, for primary loading, by applying
the Valanis and Landel (1967) hypothesis.
Planar biaxial tension has been used by many investigators to determine
the general form of the strain energy density function of nonlinear elastic
materials. Initial results, reported by Treloar (1948) and Rivlin and Saunders
(1951), used adjustable weights connected to the boundaries of thin sheet
specimens to control in-plane principal stretches. Similar investigations were
performed by Blatz and Ko (1962), Obata et al. (1970), Becker (1967), Jones
and Treloar (1975) and Kawabata et al. (1981). For an overview of early
devices, including geometric layouts, we suggest the paper by Arenz et al.
(1975). More recently, with growing interest in the mechanical behavior of
soft biological materials, the use of planar biaxial experiments has seen a
revival. For an excellent representation of the experimental and theoretical
aspects we refer to Sacks (2000) and Holzapfel and Ogden (2009).
Most constitutive laws are based on phenomenological or statistical the-
ories. The latter use statistical mechanics to capture the network evolution
of cross-linked long polymer chains and to predict the three dimensional re-
sponse of the material. Examples may be found in the works of Arruda and
Boyce (1993), Flory and Rehner (1943), James and Guth (1943), Treloar
(1946), Wang and Guth (1952), Drozdov and Dorfmann (2001) and Drozdov
and Dorfmann (2003). In contrast, phenomenological type models are based
on mathematical theories that capture the main observed behavior and fit the
experimental data with reasonable accuracy. For hyperelastic materials, in
particular, the mechanical response is given in terms of a strain-energy func-
tion defined on the space of deformation gradients. Formulations in terms of

2
strain invariants are proposed by, for example, Mooney (1940); Rivlin (1948);
Yeoh (1993); Gent (1996); Horgan and Murphy (2007) and Carroll (2011), in
terms of principal stretches by Valanis and Landel (1967), and Ogden (1972).
Frequently, nonlinear iterative methods are used to determine values of
the material model parameters. Numerous publications describing these
methods are available, however, most of them involve the use of a single
set of experimental data, typically a simple tension test. Only few publi-
cations discuss the importance of using multiple independent data sets to
determine the magnitude of the parameters and to ensure a stable response
over a wide range of deformations. For instance, Palmieri et al. (2009) and
Ogden et al. (2004) use a combination of simple tension and equibiaxial data,
while Twizell and Ogden (1983) and Stumpf and Marczak (2010) use simple
tension, equibiaxial tension and pure shear.
Unlike the protocol used in Jones and Treloar (1975), the procedure used
here focuses on periodic loading and unloading to preselected stretch lev-
els, the use of virgin samples at the start of each test, and the inclusion of
equibiaxial extension. Additionally, here, tests are performed in real time
while in Jones and Treloar (1975), Kawabata et al. (1981), Rivlin and Saun-
ders (1951) a point to point incremental stretch approach was used. In fact,
stretch was measured manually in the gage region at isochronal intervals, and
was a dependent variable to the applied displacement using a preconditioned
material sample.
The paper is organized as follows. In Section 2 we provide a short overview
of the basic equations describing the response of an incompressible nonlinear
elastic material. To validate the experimental data we specialize the theory
and provide stress-stretch relations for uniaxial and biaxial tension as well as
pure shear. The Ogden model, a formulation that can be used to apply the
Valanis-Landel hypothesis, is used to obtain explict expressions of the stress-
stretch relations in uniaxial and planar biaxial tension. Section 3 outlines
the testing protocol and summarizes experimental data. In Section 4 we
apply the Valanis-Landel hypothesis to verify and validate the consistency
of the data. A three term Ogden model is used to obtain numerical results,
which are compared to the experimental data. Values of the material model
parameters are determined using the primary loading path in uniaxial and
equibiaxial tension. Then, the same material model is used to predict the
response in biaxial tension and the results are compared to available data.
Excellent agreement is obtained in all cases.

3
2. Basic equations
In this section we provide an overview of the basic equations describing
the nonlinear elastic response of rubber-like solids. These will be used to
validate uniaxial loading-unloading and planar biaxial experimental data of
thin sheets of natural rubber. The orientation of the reference system is such
that in all cases the loading directions are parallel to the Cartesian coordinate
axes.
It has been shown by Christensen and Hoeve (1970), Gee et al. (1950)
and Penn (1970) that in uniaxial tension the change in volume is between
0.01% − 0.02% depending on the type of rubber used. In view of these
results we restrict the theory applicable to incompressible materials only.
For a more general representation of the theory of nonlinear elasticity the
interested reader is referred to, for example, Ogden (1997) and Holzapfel
(2001).
To describe the deformation, we denote the stress-free reference configu-
ration of the body by Br and identify a generic material point by its position
vector X relative to an arbitrary chosen origin. Application of mechanical
forces deforms the body, so that the point X occupies the new position x in
the deformed configuration B. This deformation is completely described by
the deformation gradient tensor F defined by
F = Gradx, (1)
where Grad denotes the gradient operator with respect to X. The Cartesian
components of F are Fiα = ∂xi /∂Xα , where i, α = 1, 2, 3. Roman indices
are associated with the current configuration B and Greek indices with the
reference configuration Br . We also use the standard notation
dV
J = det F = > 0, (2)
dv
where dV and dv denote corresponding volume elements in Br and B, re-
spectively. Under the deformation of an incompressible material there is no
change in volume and the corresponding deformation is said to be isochoric
satisfying the internal constraint
J = det F ≡ 1. (3)
The deformation gradient can be decomposed according to the unique
polar decomposition
F = RU = VR, (4)

4
where R is a proper orthogonal tensor and U and V are positive definite and
symmetric and are known as the right and left stretch tensors. These can
be expressed in spectral form. The tensor U, for example, has the spectral
decomposition
X3
U= λi u(i) ⊗ u(i) , (5)
i=1

where the principal stretches λi are the eigenvalues of U, u(i) are the unit
eigenvectors of U, and ⊗ denotes the tensor product.
In hyperelastic materials, for a time independent deformation, the work
done by the body force and surface traction is stored entirely as elastic en-
ergy. The energy density is denoted W and represents the work done by
the stress in deforming a unit volume in Br to B. For homogeneous mate-
rials W depends only on the deformation gradient and we use the notation
W = W (F).
Ogden (1986, 1997), using the energy balance equation, shows that for a
homogeneous incompressible material the nominal stress tensor S is given by
∂W
S= − pF−1 , (6)
∂F
where p is an arbitrary scalar. In component form equation (6) reads

∂W −1
Sαi = − pFαi , (7)
∂Fiα
where we note the switch of indices in performing derivatives with respect to
a non-symmetric tensor.
Using Nanson’s formula Ogden (1997) gives the connection between the
symmetric second-order Cauchy stress σ and the nominal stress S as

σ = J −1 FS. (8)

For an incompressible material J ≡ 1 and equation (8) gives

∂W ∂W
σ=F − pI, σij = Fiα − pδij , (9)
∂F ∂Fjα

where I is the identity tensor and δij the Kronecker delta.

5
The notion of objectivity requires that the stored energy is unchanged
when the body is subject to an arbitrary rigid motion in the deformed con-
figuration, i.e. the energy W depends on F only through U and we write

W (F) = W (U). (10)

The energy function W can be further specialized depending on the mate-


rial symmetry group relative to the reference configuration Br . An important
symmetry is that of isotropy for which the energy function W depends only
on the principal invariants of U, equivalently C = U2 . These are denoted
I1 , I2 , I3 and are defined by
1 2
I1 − tr C2 , I3 = det C = J 2 ,

I1 = trC, I2 = (11)
2
where we recall that for incompressible materials I3 ≡ 1. In terms of principal
stretches we have

I1 = λ21 + λ22 + λ23 , I2 = λ21 λ22 + λ22 λ23 + λ23 λ21 , I3 = λ21 λ22 λ23 . (12)

Ogden (1997) and Valanis and Landel (1967), for example, have shown
that the energy of an unconstrained, isotropic and elastic material can equiv-
alently be given as function of I1 , I2 , I3 or in terms of the principal stretches
λ1 , λ2 , λ3 . A necessary and sufficient condition is that the energy is a sym-
metric function of the principal stretches, which we write as

W (λ1 , λ2 , λ3 ) = W (λ2 , λ1 , λ3 ) = W (λ1 , λ3 , λ2 ) . (13)

In terms of principal stretches equation (6) gives

∂W
si = − pλ−1
i i = 1, 2, 3, (14)
∂λi
where si are the eigenvalue of S. Similary, the specialization of (9) gives the
principal Cauchy stresses
∂W
σi = λi − p, i = 1, 2, 3, (no summation over i) . (15)
∂λi
Note that the transformation (8) between the principal components of S and
σ reduces to σi = λi si , with no summation over i.

6
2.1. Application to planar biaxial deformations
To validate the experimental results shown in Section 3.3 we apply the
theory to the case of biaxial loading with the principal axes of deformation
coinciding with the Cartesian coordinate directions. Then, the corresponding
deformation gradient F is diagonal with components λ1 , λ2 , λ3 . The stress-
deformation relation in terms of the extension ratios, rather than in terms
of the strain invariants, is most suitable for the determination of the energy
function, which is the formulation we consider in the following (Valanis and
Landel, 1967; Ogden, 1997).
Using the incompressibility condition (3) it is possible to express the
stretch in the third direction as λ3 = λ−1 −1
1 λ2 . It follows that only two of the
principal stretches are independent and the material response can therefore
be defined by a symmetric function of the stretches λ1 and λ2 as

Ŵ (λ1 , λ2 ) = W λ1 , λ2 , λ−1 −1
1 λ2 , (16)
which replaces equation (13). Using equation (16) in (15) gives the principal
stress differences
∂ Ŵ ∂ Ŵ
σ1 − σ3 = λ1 , σ2 − σ3 = λ2 , (17)
∂λ1 ∂λ2
where we note that the scalar p has been eliminated. For an incompressible
material it is always possible to superimpose a hydrostatic stress without
producing strain and equations (17) specialize to
∂ Ŵ ∂ Ŵ
σ1 = λ1 , σ2 = λ2 . (18)
∂λ1 ∂λ2
2.1.1. Simple tension
Simple tension corresponds to an elongation in one direction accompanied
by free contraction in the other two. For an extension in the 1-direction, for
example, we have
λ1 = λ, λ2 = λ3 = λ−1/2 , (19)
and the energy (16) is then a function of the independent stretch λ
W̄ (λ) = Ŵ (λ, λ−1/2 ), (20)
which gives the corresponding principal stresses as
dW̄(λ)
σ1 = σ = λ , σ2 = σ3 ≡ 0. (21)

7
2.1.2. Pure shear
Pure shear deformation occurs when a thin rectangular sheet is stretched
in the 1-direction, free to contract in the 2-direction and no deformation in
the third direction. The corresponding principal stretches are

λ1 = λ, λ2 = λ−1 , λ3 = 1, (22)

and we define the energy

W̃ (λ) = Ŵ (λ, λ−1 ), (23)

which leads to
dW̃(λ)
σ1 = σ = λ , σ2 ≡ 0, (24)

and σ3 as given by (17)2 .

2.1.3. Equibiaxial tension


Equations (16) and (17) can be specialized to equibiaxial tension, which
in terms of the principal stretches is given by

λ1 = λ2 = λ. (25)

The corresponding energy has the form


¯ (λ) = Ŵ (λ, λ),
W̄ (26)

which gives the prinicipal components of the Cauchy stress


¯ (λ)
1 dW̄
σ1 = σ2 = σ = λ , σ3 ≡ 0. (27)
2 dλ
2.2. The Valanis-Landel hypothesis
Valanis and Landel (1967), using a large number of experimental data,
postulated that the energy density W can be expressed equivalently in the
form
W (λ1 , λ2 , λ3 ) = w(λ1 ) + w(λ2 ) + w(λ3 ). (28)
The function w is scalar valued and, due to the symmetry condition (13),
has the same form for each of the principal stretches λ1 , λ2 and λ3 . Equation
(28) simplifies an experimental program to determine the form of the energy

8
function and is known as the Valanis-Landel hypothesis. For an incompress-
ible material the three principal stretches are not independent and (28) is
replaced by

Ŵ (λ1 , λ2 ) = W λ1 , λ2 , λ−1 −1
1 λ2 = w(λ1 ) + w(λ2 ) + w̄(λ1 λ2 ), (29)
where we note that the argument of the function w̄ is λ1 λ2 , i.e. w̄(λ1 λ2 ) =
w(λ−1 −1
1 λ2 ) = w(λ3 ). Equation (18), using the formulation (29) for Ŵ , gives

σ1 = λ1 w′ (λ1 ) + λ1 λ2 w̄′ (λ1 λ2 ), (30)


σ2 = λ2 w′ (λ2 ) + λ1 λ2 w̄′ (λ1 λ2 ), (31)
where, for convenience of notation, we used the superscript ′ to indicate
differentiation with respect of the argument of the function concerned.

2.2.1. Simple tension


As in Section 2.1.1, for simple tension we have λ1 = λ and λ2 = λ3 =
λ −1/2
. The energy (29) is then replaced by
W̄ (λ) = Ŵ λ, λ−1/2 = w(λ) + w λ−1/2 + w̄ λ1/2 ,
  

= w(λ) + 2w λ−1/2 , (32)
and the principal stress components, making use of (21), are

σ1 = σ = λw′ (λ) − λ−1/2 w′ λ−1/2 , σ2 = σ3 = 0. (33)
2.2.2. Pure shear
The principal stretches of a pure shear deformation for an incompressible
material are defined in Setion 2.1.2 as λ1 = λ, λ2 = λ−1 with λ3 = 1. This
suggests to replace the energy formulation (29) by
 
W̃ (λ) = Ŵ λ, λ−1 = w λ) + w(λ−1 . (34)
From equation (24) it follows that
σ1 = σ = λw′ (λ) − λ−1 w′ (λ−1 ), σ2 = 0, (35)
where again the superscript ′ indicates differentitation with respect of the
argument of the function w. From (17)2 we find the component σ3 as
σ3 = −λ−1 w′ (λ−1 ) − w̄′ (λ1 λ2 ), (36)
evaluated at λ1 = λ and λ2 = λ−1 .

9
2.2.3. Equibiaxial tension
In Section 2.1.3 we defined the principal stretches of equibiaxial tension
as λ1 = λ2 = λ with λ3 = λ−2 . This allows the energy formulation (29) to
be replaced by
¯ (λ) = Ŵ (λ, λ) = 2w(λ) + w̄ λ2  ,
W̄ (37)
and from equation (27) the principal stress components are

σ1 = σ2 = σ = λw′ (λ) + λ2 w̄′ (λ2 ), σ3 = 0, (38)

where the superscript ′


on w̄ indicates differentitation with respect of the
argument λ2 .

2.3. The Ogden Model


There are many strain energy formulation that satisfy the Valanis-Landel
hypothesis (28). In what follows we restrict attention to the model proposed
by Ogden (1972)
M
X µn
W = (λα1 n + λα2 n + λα3 n − 3), (39)
n=1
αn

where µn and αn are material constants to be determined experimentally and


M is a positive integer. These constants must fulfill the requirement
M
X
µn αn = 2µ, (40)
n=1

where µ is the shear modulus of the material in the reference configuration.


With reference to equation (28) we find that
M
X µn
w(λi ) = (λαi n − 1), i ∈ {1, 2, 3}. (41)
n=1
αn

The Ogden model (39), for an incompressible material, can be written in


terms of the two inpendent stretches λ1 , λ2 as
M
X µn
Ŵ (λ1 , λ2 ) = (λα1 n + λα2 n + λ1−αn λ2−αn − 3). (42)
n=1
αn

10
From equation (29) we have
M
X µn  
w̄(λ1 λ2 ) = (λ1 λ2 )−αn − 1 , (43)
α
n=1 n

with w(λ1 ) and w(λ2 ) unchanged as given by (41).


Using equations (30) and (31) gives the expressions for the in-plane stress
components σ1 and σ2 as
M
X M
X
σ1 = µn (λα1 n − λ1−αn λ2−αn ), σ2 = µn (λα2 n − λ1−αn λ2−αn ). (44)
n=1 n=1

For simple tension, equation (42) simplifies to


M
X µn
W̄ (λ) = Ŵ (λ, λ −1/2
)= (λαn + 2λ−αn /2 − 3), (45)
n=1
αn
which gives the nonzero stress component
M
X
σ= µn (λαn − λ−αn /2 ). (46)
n=1

For pure shear, equation (42) can be written as


M
X µn
W̃ (λ) = Ŵ (λ, λ ) = −1
(λαn + λ−αn − 2), (47)
n=1
αn
with the stress components in the 1- and 3-directions given by
M
X M
X
αn −αn
σ1 = µn (λ −λ ), σ3 = µn (1 − λ−αn ), (48)
n=1 n=1

and with σ2 = 0.
For equibiaxial tension, we have
M
¯ (λ) = Ŵ (λ, λ) =
X µn
W̄ (2λαn + λ−2αn − 3), (49)
n=1
αn
with the stress component in the direction of elongation given by
M
X
σ= µn (λαn − λ−2αn ). (50)
n=1

11
3. Experimental methods and results
Samples of natural rubber are used to perform a series of uniaxial and
planar biaxial loading-unloading tests. We provide a summary of the main
features of the testing equipment and an overview of the experimental pro-
tocol. A detailed description of specimen shape and preparation is given by
Dorfmann and Pancheri (2012).

3.1. Equipment
All tests are performed on a custom built Zwick/Roell planar biaxial
testing machine. The machine is outfit with four diametrically opposed
independently-controlled linear actuators, each of 2 kN capacity, travel reso-
lution of 0.1 µm and a dedicated load cell with a resolution of ±0.003 N. A
video extensometer equipped with a charge-coupled device (CCD) continu-
ously monitors the deformation within the gage area of the specimen at 50
frames per second and determines both longitudinal and transverse strains in
separate input channels. Physical gage marks are attached to the sample to
identify the gage region and are used by the video extensometer for tracking
and controlling deformation, see Figure 1.
The arms of the cruciform shaped specimen are securely held in place by
screw-type grips connected to individual load cells. All recorded data, po-
sition of gage marks, actuator drives, load cell readings, stretches and time,
are captured at a sampling rate of 15 Hz. Experiments are performed at
isothermal and quasi-static conditions with ambient temperature of 22±1◦ C
and with constant strain rate of 0.005 s−1 . Control deviations are ±1% from
the specified target stretches. The ability to perform tests using real-time
strain control allows for independent and automatic variation of the veloc-
ity of extension and retraction of each actuator to maintain the prescribed
deformation rate within each pair of gage marks. Therefore, the technol-
ogy corrects for the small sample-to-sample difference in the positioning of
gage marks and also compensates for the differences in specimen arm lengths
caused by the relative position of the sample with respect to the machine’s
grips. This produces a pure homogeneous deformation within the gage re-
gion regardless of what occurs outside of it. To the best of the authors’
knowledge, this is the first time a full strain control protocol has been used
to characterize natural rubber in planar biaxial tension.

12
digital
gage marks

gage physical
region gage marks

Figure 1: Geometric layout of material specimens. The digital gage marks follow the
location of the physical markers by using the provided contrast. Each of the gage marks
is tracked independently and used to adjust actuator movements to control pre-selected
values of stretch and strain rate.

3.2. Testing protocol


Pressure rolling used to manufacture thin rubber sheets, in general, in-
duces a preferred direction in the material. To compensate for this bias the
same loading routine is performed twice, using virgin specimens, with the co-
ordinate directions interchanged with respect to the calendering process. All
results in this section are based on the averaged values of two corresponding
data sets.
The position of the physical gage marks defines the reference configuration
Br when a 1.5 N preload is detected by each of the four load cells. The applied
preload ensures the proper progression of permanent set and the prevention
of out of plane buckling upon completion of each unloading phase.
The first characterization test consists of five uniaxial loading-unloading
cycles with maximum extension of λ = 2.5. The cruciform shaped specimen,
shown in Figure 1, is stretched in one direction using one pair of actuators
and is free to contract in the other two. Next in the characterization se-
quence, using a virgin sample, we perform five cycles of planar equibiaxial
tension. Starting from the reference configuration Br two pairs of actuators
simultaneously stretch the specimen in two mutually orthogonal directions
to the same target stretch.
To generalize in-plane biaxial tension it is convenient to define a Cartesian

13
coordinate system with the 1- and 2-directions coinciding with the in-plane
principal axes of deformation. Planar biaxial characterization consists of
first stretching the specimen to a pre-selected value λ2 contracting freely in
−1/2
the other two directions, i.e. λ1 = λ3 ≈ λ2 . Next, starting from this
configuration, with λ2 held constant, the sample is subjected to five loading-
unloading cycles in the 1-direction with maximum stretch λ1 = 2.5. During
each cycle unloading is completed and reloading initiated when the initial
preload of 1.5 N is reached.
Following these specifications and using virgin specimens we perform four
planar biaxial tension tests. For the first test, known as pure shear, the spec-
imen is subjected to five loading-unloading cycles with maximum extension
λ1 = 2.5 and no deformation in the 2-direction (λ2 = 1). Additional biaxial
characterization tests are performed by changing the value of the maximum
stretch in the 2-direction to λ2 = 1.5, λ2 = 2.0 and λ2 = 2.25. With these
values held constant, each specimen is subjected to five loading-unloading
cycles in the 1-direction with maximum stretch λ1 = 2.5.

3.3. Experiment results


Figures 2 to 4 summarize experimental data of the tests described in
Section 3.2. In all graphs the Cauchy stress is given in units of MPa and
reported as a function of stretch.
The left and right plots in Figure 2 depict the stress versus stretch of
natural rubber in uniaxial and equibiaxial tension. As expected, the stress
for equibiaxial tension, which is shown on the right, is higher for the same
amount of stretch. The data clearly show that the effects of stress softening
and residual strain, even though present, are not a major concern for an un-
filled compound. These results can be compared to the behavior of a particle
reinforced rubber where significant stress softening and large residual strain
occur, see corresponding data by Dorfmann and Ogden (2003), Dorfmann
and Ogden (2004) and Dorfmann and Pancheri (2012).
Figures 3 and 4 show the stress-stretch responses of natural rubber during
pure shear and planar biaxial tension tests. The results shown in each of
the four columns represent the material response during one of these tests.
Specifically, the left column in Figure 3 depicts the behavior in pure shear.
The right column in Figure 3 and the columns in Figure 4 correspond to
planar biaxial tension with maximum stretches in the 2-direction respectively
of λ2 = 1.5, λ2 = 2.0 and λ2 = 2.25.

14
6 6

4 4
σ σ

2 2

0 0
1 1.5 2 2.5 1 1.5 2 2.5
λ λ
Figure 2: Graphs on the left and right show five loading-unloading cycles of natural rubber
in uniaxial and equibiaxial tension, respectively. Data are shown as Cauchy stress versus
stretch with stress values given in MPa.

The graphs on top in Figures 3 and 4 depict the response during initial
uniaxial tension in the 2-direction to a pre-selected value of λ2 . With λ2 held
constant, the specimen is preconditioned in the 1-direction with maximum
extension λ1 = 2.5 (graphs below). Bottom plots show σ2 as a function of λ1
with λ2 held constant.

3.4. Verification and validation of data


Data verification and validation is performed by applying the Valanis-
Landel hypothesis to evaluate whether or not they are generated correctly.
We recall that data shown in Figures 3 and 4 are generated by first stretching
the specimens in the 2-direction to a pre-selected value of λ2 contracting freely
−1/2
in the 1- and 3-directions such that λ1 = λ3 = λ2 . Next, with λ2 held at
a constant value, the specimens are subjected to loading-unloading cycles in
the 1-direction.
In the following we focus on the material response during primary loading.
Using (30) and (31) we find that

σ1 − σ2 = λ1 w′ (λ1 ) − λ2 w′ (λ2 ), (51)


−1/2
which can be applied when λ1 ≥ λ2 with λ2 ≥ 1. We also recall that
the superscript indicates differentiation with respect to the argument of the
function. For planar biaxial tension, with λ2 constant, the right term on the

15
6 6

4 4
σ2 σ2

2 2

0 0
1 1.5 2 2.5 1 1.5 2 2.5
λ2 λ2

6 6

4 4
σ1 σ1

2 2

0 0
1 1.5 2 2.5 1 1.5 2 2.5
λ1 λ1

6 6

4 4
σ2 σ2

2 2

0 0
1 1.5 2 2.5 1 1.5 2 2.5
λ1 λ1
Figure 3: Left and right columns show the stress-stretch responses, respectively, in pure
shear with λ2 = 1 and planar biaxial tension. For the latter, the specimen is first stretched
in the 2-direction to λ2 = 1.5 and free to contract in the other two. Then, with λ2 held
constant, the material is subjected to five loading-unloading cycles in the 1-direction with
maximum stretch λ1 = 2.5. 16
6 6

4 4
σ2 σ2

2 2

0 0
1 1.5 2 2.5 1 1.5 2 2.5
λ2 λ2

6 6

4 4
σ1 σ1

2 2

0 0
1 1.5 2 2.5 1 1.5 2 2.5
λ1 λ1

6 6

4 4
σ2 σ2

2 2

0 0
1 1.5 2 2.5 1 1.5 2 2.5
λ1 λ1
Figure 4: Left and right columns show the stress-stretch responses in planar biaxial tension.
The specimens are first stretched in the 2-direction to λ2 = 2.0 (left column) and λ2 = 2.5
(right column) and free to contract in the other two. Then, with λ2 held constant, the
material is subjected to five loading-unloading cycles in the 1-direction with maximum
stretch λ1 = 2.5. 17
right hand side is constant. It follows that graphs showing the stress differ-
ence, for different values of λ2 , are of identical shape and can be superposed
using a vertical shift (Jones and Treloar, 1975; Vangerko and Treloar, 1978;
Ogden, 1986).
The use of the incompressibility condition (3) allows to extend the ex-
−1/2
perimental data set to the region λ1 ≤ λ2 . In particular, for each set
of recorded values λ1 and λ2 we calculate λ3 = λ−1 −1
1 λ2 , which results in
−1/2
λ3 ≤ λ2 . An equivalent extension of the range of measured data has been
used by Jones and Treloar (1975) and Ogden (1986). From equation (17)2 ,
using (29), we find that
σ3 − σ2 = −λ1 λ2 w̄′ (λ1 λ2 ) − λ2 w′ (λ2 ), (52)
where λ2 is again constant for each biaxial test and therefore λ2 w′ (λ2 ) is con-
stant as well. The data are combined with the measured values by renaming
−1/2
λ3 as λ1 and are shown in Figure 5. In particular, for values of λ1 ≥ λ2 the
stress difference σ1 − σ2 is given by (51) and represents the measured data.
−1/2
For λ1 ≤ λ2 the stress −σ2 is obtained from (52) with σ3 = 0. The point
−1/2
λ1 = λ2 , for each of the tests performed, is indicated by an arrow. We note
that the values of the experimental data for σ1 and σ2 , shown in Figure 5,
coincide when the stretches in the 1- and 2-directions are equal and therefore
the stress difference σ1 − σ2 = 0. For reinforced rubber, for example, this is
not possible since stress relaxation occurs in the direction of constant held
stretch. Therefore, not surprisingly, the Valanis-Landel hypothesis is only
satisfied for nonlinear elastic materials with negligible stress relaxation.
The right term on the right hand side of equations (51) and (52) is con-
stant for constant λ2 . This implies that data in Figure 5 can be shifted
vertically by a theoretically determined amount to overlap, for example, with
data representing pure shear. The vertical shift that transposes biaxial ten-
sion data to the pure shear curve is obtained from equation (51) and has the
form
λ2 w′ (λ2 ) − w′ (1), (53)
where λ2 = 1.5, 2.0 and 2.25. In Section 4 we determine the corresponding
numerical values and verify the accuracy of the data.

4. Numerical results
The material response during primary loading in uniaxial and equibiaxial
tension, shown in Figure 2, is used to determine the values of the material

18
λ2 = 1
4
λ2 = 1.5

2 λ2 = 2.0
λ2 = 2.25

0
σ1 − σ2 ❄
(or −σ2 )
−2 ❄

−4

−6
0 0.5 1 1.5 2 2.5
λ1
Figure 5: Primary loading in planar biaxial tension with λ2 held constant. For each
−1/2 −1/2
test the location λ1 = λ2 is indicated by an arrow. For λ1 ≥ λ2 the graphs depict
measured data as stress difference σ1 −σ2 versus λ1 . Use of the incompressibility condition,
−1/2
interchanging λ3 with λ1 , extends range to λ1 ≤ λ2 , where −σ2 is shown as a function
of λ1 . Stress values are given in MPa.

model parameters of a three term Ogden model (39). The nonlinear iterative
method for parameter estimation proposed by Buzzi-Ferraris and Manenti
(2009) is used for calculating the constants µn and αn , n = 1, 2, 3. These val-
ues are summarized in Table 1 and, using equation (40), give the magnitude
of the shear modulus µ = 0.7609 MPa. Figure 6 shows the correspond-
ing numerical results for loading up to λ = 2.75 and compares them to the
experimental data during primary loading taken from Figure 2.
With the material parameters given in Table 1 the model is now used to
determine the in-plane Cauchy stress components σ1 and σ2 in pure shear
and biaxial tension with stretches in the 2-direction, λ2 = 1.5, λ2 = 2.0 and
λ2 = 2.25, held constant. For each value of λ2 , the stretch in the 1-direction is

Table 1: Material model parameters using data from primary loading in uniaxial and
equibiaxial tension. The values of µ1 , µ2 and µ3 are given in MPa.
Material model parameters, Ogden M = 3
µ1 α1 µ2 α2 µ3 α3
0.86253 1.63718 0.01063 5.28573 -0.02097 -2.47885

19
8 8

6 6

σ σ
4 4

2 2

0 0
1 1.5 2 2.5 1 1.5 2 2.5
λ λ
Figure 6: Left and right figures show numerical results and experimental data during
primary loading in uniaxial and equibiaxial tension, respectively. Numerical results, using
a three term Ogden model, are shown by solid lines. Stress values are given in MPa.

increased from λ1 = 0.5 to a maximum value of 2.75. Figure 7 compares the


numerical results to the experimental data during primary loading, which are
taken from Figures 3 and 4. The numerical response of the material, shown
by the solid line, is in excellent agreement with the experimental data.
Using the material parameters in Table 1 we now determine the values of
the vertical shift (53) for λ2 = 1.5, 2 and 2.25 that are needed to transpose
the corresponding biaxial tension data. These are listed in Table 2 and the
quality of the overlap is displayed in Figure 8. All data after vertical transla-
tion fall within 3% of the pure shear curve, which is remarkable considering
small variations in sample geometry and material properties. The solid line
in Figure 8 represents the stress difference σ1 − σ2 in pure shear using a
three term Ogden model. The results again show excellent agreement with
experimental data.

Table 2: Vertical shift factor (53) for λ2 = 1.5, 2 and 2.25. Values are given in MPa.
Biaxial tension λ2 = 1.5 λ2 = 2.0 λ2 = 2.25
Vertical shift 0.9059 2.2417 3.1714

20
8 8

λ2 = 1 λ2 = 1.5 σ1
6 6
σ1
4 4

σ σ σ2
2 2
σ2
0 0

−2 −2

0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5


λ1 λ1
8 8

λ2 = 2 λ2 = 2.25
6 6

σ2
4 4
σ2
σ σ
2 2
σ1 σ1
0 0

−2 −2

0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5


λ1 λ1
Figure 7: Solids lines show the numerical prediction in pure shear and biaxial tension and
compares results to experimental data. Values of material model parameters of the 3 term
Ogden model have been obtained using data in uniaxial and equibiaxial tension. Stress
values are given in MPa.

5. Concluding Remarks
A custom built planar biaxial testing machine is used to characterize
a natural rubber compound. The machine is equipped with four control
mechanisms to independently adjust the movement of each of the actua-
tors stretching simultaneously a cruciform shaped specimen in two mutually
orthogonal directions to pre-selected target stretches. For example, during
pure shear, the specimen is stretched in one direction to the prescribed tar-

21
6

2
λ1 w′ (λ1 ) − w′ (1)
0

−2

−4
0 0.5 1 1.5 2 2.5
λ1
Figure 8: Vertical shift of results shown in Figure 5 to overlap with data in pure shear.
Accuracy of overlap is used to verify and validate consistency of data. After translation,
all points fall within 3% of the pure shear curve. Solid line represent the numerical fit
using a three term Ogden model. Stress values are given in MPa.

get while the orthogonal direction is maintained at the reference stretch of 1


with an accuracy of ±1%.
Uniaxial and planar biaxial tension data of natural rubber are summa-
rized. As expected, the data show that stress softening and residual strain,
even though present, are not a major concern. Figure 7 shows that values of
the principal in-plane stress components are equal for the same amount of
stretch even though λ1 is changing and λ2 is held constant, which indicates
that time dependent effects are negligible. This is not the case for a filled
compound where different rates of stress relaxation would occur in the two
directions.
The consistency of the data is verified by applying the Valanis-Landel hy-
pothesis, whereby the in-plane stress difference is shifted by a theoretically
determined amount to overlap, for example, with the curve representing the
material response in pure shear. Comparing the vertical shift to the experi-
mental data in pure shear, for corresponding stretches, validates the consis-
tency of the data. The maximum difference, for any values of λ1 and λ2 is
found to be 3%, which is remarkable considering small variations in sample
geometry and material properties.
The magnitudes of the material model parameters of a three term Og-
den model are determined using primary loading in uniaxial and equibiaxial

22
tension. The same material model is then used to predict the response in
biaxial tension, where the constant held stretch assumes values of 1.5, 2 and
2.5. The numerical results are compared to experimental data and excellent
agreement is obtained in all cases.

Acknowledgements
This publication was based on work supported in part by Award No KUK-
C1-013-04, made by King Abdullah University of Science and Technology
(KAUST).

References
Arenz, R. J., Landel, R. F., Tsuge, K., 1975. Miniature load-cell instrumenta-
tion for finite-deformation biaxial testing of elastomers. Exp. Mech. 15 (3),
114–120.
Arruda, E. M., Boyce, M. C., 1993. A 3-dimensional constitutive model for
the large stretch behavior of rubber elastic materials. J. Mech. Phys. Solids
41 (2), 389–412.
Becker, G. W., 1967. On phenomenological description of nonlinear deforma-
tion behavior of rubberlike high polymers. J. Polym. Sci. Pol. Sym. (16PC),
2893–2903.
Blatz, P. J., Ko, W. L., 1962. Application of finite elastic theory to the
deformation of rubbery materials. T. Soc. Rheol. 6, 223–251.
Buzzi-Ferraris, G., Manenti, F., 2009. Kinetic models analysis. Chem. Eng.
Sci. 64 (5), 1061–1074.
Carroll, M. M., 2011. A strain energy function for vulcanized rubbers. J.
Elasticity 103 (2), 173–187.
Christensen, R. G., Hoeve, C. A. J., 1970. Comparison between theoreti-
cal and experimental values of the volume changes accompanying rubber
extension. J. Polym. Sci. A1 8 (6), 1503–1512.
Dorfmann, A., Ogden, R. W., 2003. A pseudo-elastic model for loading,
partial unloading and reloading of particle-reinforced rubber. Int. J. Solids
Stuct. 40 (11), 2699–2714.

23
Dorfmann, A., Ogden, R. W., 2004. A constitutive model for the Mullins
effect with permanent set in particle-reinforced rubber. Int. J. Solids Stuct.
41 (7), 1855–1878.

Dorfmann, A., Pancheri, F. Q., 2012. A constitutive model for the Mullins
effect with changes in material symmetry. Int. J. Nonlinear Mech. 47 (8),
874–887.

Drozdov, A. D., Dorfmann, A., 2001. Stress-strain relations in finite vis-


coelastoplasticity of rigid-rod networks: applications to the Mullins effect.
Continuum Mech. Therm. 13 (3), 183–205.

Drozdov, A. D., Dorfmann, A., 2003. A micro-mechanical model for the


response of filled elastomers at finite strains. Int. J. Plasticity 19 (7), 1037–
1067.

Flory, P. J., Rehner, J., 1943. Statistical mechanics of cross-linked polymer


networks. J. Chem. Phys. 11 (11), 512–526.

Gee, G., Stern, J., Treloar, L. R. G., 1950. Volume changes in the stretching
of vulcanized natural rubber. T. Faraday Soc. 46 (12), 1101–1106.

Gent, A. N., 1996. A new constitutive relation for rubber. Rubber Chem.
Technol. 69 (1), 59–61.

Holzapfel, G., 2001. Nonlinear Solid Mechanics: A Continuum Approach for


Engineering. John Wiley & Sons, Chichester.

Holzapfel, G. A., Ogden, R. W., 2009. On planar biaxial tests for anisotropic
nonlinearly elastic solids. A continuum mechanical framework. Math.
Mech. Solids 14 (5), 474–489.

Horgan, C. O., Murphy, J. G., 2007. Limiting chain extensibility constitutive


models of Valanis-Landel type. J. Elasticity 86 (2), 101–111.

James, H. M., Guth, E., 1943. Theory of elastic properties of rubber. J.


Chem. Phys. 11 (10), 455–481.

Jones, D. F., Treloar, L. R. G., 1975. The properties of rubber in pure-


homogeneous strain. J. Phys. D. Appl. Phys. 8 (11), 1285–1304.

24
Kawabata, S., Matsuda, M., Tei, K., Kawai, H., 1981. Experimental sur-
vey of the strain energy density function of isoprene rubber vulcanizate.
Macromolecules 14 (1), 154–162.

Mooney, M., 1940. A theory of large elastic deformtion. J. Appl. Phys. 11 (9),
582–592.

Obata, Y., Kawabata, S., Hiromichi, K., 1970. Mechanical properties of natu-
ral rubber vulcanizates in finite deformation. J. Polym. Sci. A2 8, 903–919.

Ogden, R. W., 1972. Large deformation isotropic elasticity - On the correla-


tion of theory and experiment for incompressible rubberlike solids. P. Roy.
Soc. Lond. A Mat. 326 (1567), 565–584.

Ogden, R. W., 1986. Recent advances in the phenomenological theory of


rubber elasticity. Rubber Chem. Technol. 59 (3), 361–383.

Ogden, R. W., 1997. Non-linear Elastic Deformations. Dover Publications,


New York.

Ogden, R. W., Saccomandi, G., Sgura, I., 2004. Fitting hyperelastic models
to experimental data. Comput. Mech. 34 (6), 484–502.

Palmieri, G., Sasso, M., Chiappini, G., Amodio, D., 2009. Mullins effect
characterization of elastomers by multi-axial cyclic tests and optical ex-
perimental methods. Mech. Mater. 41 (9), 1059–1067.

Penn, R. W., 1970. Volume changes accompanying the extension of rubber.


J. Rheol. 14 (4), 509–517.

Rivlin, R. S., 1948. Large elastic deformations of isotropic materials. 4. Fur-


ther developments of the general theory. Philos. T. R. Soc. A 241 (835),
379–397.

Rivlin, R. S., Saunders, D. W., 1951. Large elastic deformations of isotropic


materials. 7. Experiments on the deformation of rubber. Philos. T. R. Soc.
A 243 (865), 251–288.

Sacks, M. S., 2000. Biaxial mechanical evaluation of planar biological mate-


rials. J. Elasticity 61 (1-3), 199–246.

25
Stumpf, F. T., Marczak, R. J., 2010. Optimization of constitutive parame-
ters for hyperelastic models satisfying the Baker-Ericksen inequalities. In:
Dvorkin, E., Goldschmit, M., Storti, M. (Eds.), Asociación Argentina de
Mecánica Computacional. Vol. XXIX of Mecánica Computacional. Buenos
Aires, Argentina, pp. 2901–2916.

Treloar, L. R. G., 1946. The elasticity of a network of long-chain molecules.


T. Faraday Soc. 42 (1-2), 83–94.

Treloar, L. R. G., 1948. Stresses and birefringence in rubber subjected to


general homogeneous strain. P. Phys. Soc. Lond. 60 (338), 135–144.

Twizell, E. H., Ogden, R. W., 1983. Non-linear optimization of the material


constants in Ogden stress-deformation function for incompressible isotropic
elastic-materials. J. Aust. Math. Soc. B 24, 424–434.

Valanis, K. C., Landel, R. F., 1967. Strain-energy function of a hyper-elastic


material in terms of the extension ratios. J. Appl. Phys. 38 (7), 2997–3002.

Vangerko, H., Treloar, L. R. G., 1978. The inflation and extension of rubber
tube for biaxial strain studies. J. Phys. D Appl. Phys. 11 (14), 1969–1978.

Wang, M. C., Guth, E., 1952. Statistical theory of networks of non-Gaussian


flexible chains. J. Chem. Phys. 20 (7), 1144–1157.

Yeoh, O. H., 1993. Some forms of the strain energy function for rubber.
Rubber Chem. Technol. 66 (5), 754–771.

26
RECENT REPORTS
12/82 A new pathway for the re-equilibration of micellar surfactant solu- Griffiths
tions Breward
Colegate
Dellar
Howell
Bain

12/83 Object-Oriented Paradigms for Modelling Vascular Tumour


Growth: a Case Study Connor
Cooper
Byrne
Maini
McKeever

12/84 Chaste: an open source C++ library for computational physiology Mirams
and biology Arthurs
Bernabeu
Bordas
Cooper
Corrias
Davit
Dunn
Fletcher
Harvey
Marsh
Osborne
Pathmanathan
Pitt-Francis
Southern
Zemzemi
Gavaghan

12/85 A two-pressure model for slightly compressible single phase flow Schlackow
in bi-structured porous media Marguerat
Proudfoot
Bähler
Erban
Gullerova

12/86 Boolean modelling reveals new regulatory connections between Lovrics


transcription factors orchestrating the development of the ventral Gao
spinal cord Juhász
Bock
Byrne
Dinnyés
Kovács

12/87 Asymptotic solutions of glass temperature profiles during steady Taroni


optical fibre drawing Breward
Cummings
Griffiths

12/88 The kinetics of surfactant desorption at the airsolution interface Morgan


Breward
Griffiths
Howell
Penfold
Thomas
12/96 Mathematical Biomedicine and Modeling Avascular Tumor Byrne
Growth

12/97 Inference of the genetic network regulating lateral root initiation Muraro
in Arabidopsis thaliana Voß
Wilson
Bennett
Byrne
De Smet
Hodgman
King

12/98 Axisymmetric bifurcations of thick spherical shells under inflation deBotton


and compression Bustamante
Dorfmann

12/99 Calculus from the past: Multiple Delay Systems arising in Cancer Wake
Cell Modelling Byrne

12/100 Nonlocal models of electrical propagation in cardiac tissue: elec- Bueno-Orovio


trotonic effects and the modulated dispersion of repolarization Kay
Grau
Rodriguez
Burrage

12/101 Microfluidic Immunomagnetic Multi-Target Sorting A Model for Tsai


Controlling Deflection of Paramagnetic Beads Griffiths
Stone

12/102 A New Lattice Boltzmann Equation to Simulate Density-Driven Allen


Convection of Carbon Dioxide Reis
Sun

12/103 Control and optimization of solute transport in a porous tube Griffiths


Howell
Shipley

12/104 Air-cushioning in impact problems Moore


Ockendon
Oliver

Copies of these, and any other OCCAM reports can be obtained from:
Oxford Centre for Collaborative Applied Mathematics
Mathematical Institute
24 - 29 St Giles’
Oxford
OX1 3LB
England
www.maths.ox.ac.uk/occam

También podría gustarte