You are on page 1of 32

26.8.

2021 Mooney-Rivlin

Search Continuum Mechanics Website


Mooney-Rivlin Models
Search

 home > material behavior > mooney-rivlin



Introduction
Mooney-Rivlin models are popular for modeling the large strain
nonlinear behavior of incompressible materials, i.e., rubber.
It's important to
understand that Mooney-Rivlin models do not
give any special insight into material behavior. They are merely curve-fits of various polynomials to
test data. The
numerical values of coefficients resulting from the curve-fits are entered into FEA programs for use in mechanical analyses. The
FEA program knows how stiff the rubber is based on the values of the coefficients.

Keep in mind that auto-summation of duplicated subscripts will not apply to the equations on this page. Hopefully, the reasons for this will be
obvious!

Strain Energy
Recall that we saw on the thermodynamics page
that the 2nd Piola-Kirchhoff stress is the partial derivative of the Helmholtz free energy with
respect to the elastic part of the Green strain
tensor (with a density thrown in).

PK2
∂Ψ
σ = ρo
el
∂E

The Helmholtz free energy contains thermal energy and mechanical strain energy. But in most every discussion of Mooney-Rivlin coefficients, the
thermal
part is neglected, leaving only the mechanical part, W . (Actually, W
is declared to represent ρo Ψ, not just Ψ). Second, since all of the
deformation of a hyperelastic material is elastic by definition,
it is sufficient to write E simply as E.
This gives

el

www.continuummechanics.org/mooneyrivlin.html ∂ 1/32
26.8.2021 Mooney-Rivlin

∂W
PK2
σ =
∂E

But there is a challenge with this general approach. It is the determination


of off-diagonal (shear) terms. As with the shear terms in Hooke's Law,
they
are not independent of the normal terms, but must be consistent with coordinate
transformations that transform normal components into
shears and vice-versa.
And as with Hooke's Law, the resolution is to define the material behavior
for the principal values and rely on coordinate
transformations to give
the appropriate corresponding behavior of the shear terms.

∂W
PK2
σ =
i
∂Ei

But alas, even this is not quite what is done in Mooney-Rivlin models. Instead, derivatives
are taken with respect to stretch ratios, λ, which are the
ratios of initial and final lengths in the principal directions, (LF /Lo ). So the stretch ratio is
"one plus engineering strain," λ = 1 + ϵEng , and
therefore λ − 1 = ϵEng = ΔL /Lo .

We can get there using the chain rule.

∂W ∂W ∂λi
PK2
σ = =
i
∂Ei ∂λi ∂Ei

We just need an equation relating Ei to λi . Recall that a principal Green strain equals

2
ΔLi 1 ΔLi
Ei = + ( )
Lo 2 Lo

And insert the relationship λi − 1 = ΔLi /Lo to get

1
2
Ei = (λi − 1) + (λi − 1)
2

www.continuummechanics.org/mooneyrivlin.html 2/32
26.8.2021 Mooney-Rivlin

So ∂Ei /∂λi = λi and therefore

∂λi 1
=
∂Ei λi

Substituting this into the equation containing σ PK2 and W gives

1 ∂W
PK2
σ =
i
λi ∂λi

This equation by itself is not helpful. But recall from the energetic conjugates page that
σi = (1 + ϵi )2 σiPK2 and therefore
σi 2
= λ σ
i
PK2
i
. So if we
simply multiply
both sides of the above equation by λ2i , we will have something much more useful.

∂W
2 PK2
σi = λ σ = λi
i i
∂λi

So each principal Cauchy stress can be related directly to the partial derivative of
strain energy through

∂W
σi = λ i
∂λi

In this case, the stretch ratio is in a principal strain direction, and will
give a principal Cauchy stress in that same principal direction.

Overly Simplified Uniaxial Tension Example

Let's take the above equation and propose a very simple function for the strain energy
www.continuummechanics.org/mooneyrivlin.html
W for a special case of uniaxial tension 3/32
26.8.2021 Mooney-Rivlin
Let's take the above equation and propose a very simple function for
the strain energy, W , for a special case of uniaxial tension.
Propose that

1
2
W = E(λ − 1)
2

Then the Cauchy stress is

1 2
∂ [ E(λ − 1) ]
2
σ = λ
∂λ

= λE(λ − 1)

Recall that σ Eng = σ/λ , so dividing through by λ gives

Eng
σ = E(λ − 1)

and λ = 1 + ϵ
Eng
, so λ − 1 = ϵ
Eng
.
And this leads to

Eng Eng
σ = Eϵ

That was fun because of its simplicity and the familiar result we obtained, but it is in fact not very useful. Why? Because we have
over-
simplified things so much that we have lost the ability to
incorporate Poisson Effects. So our proposed function for the
strain energy, W ,
in terms of λ could not be used
in, say, a general 3-D finite element program. We will see how to overcome these limitations in the
following sections.

www.continuummechanics.org/mooneyrivlin.html 4/32
26.8.2021 Mooney-Rivlin

Invariants
The Mooney-Rivlin class of models express the strain energy in terms of
invariants, which are in turn expressed as functions of stretch ratios.

The invariants are of the product of the deformation gradient with its transpose,
F ⋅ F . This result is symmetric because the product of any
T

matrix with its transpose is always symmetric. This makes it possible


to calculate principal values, which will become important shortly.
Recall
T T
from the polar decomposition page,
that F ⋅ F = V ⋅ V because

T T T T T
F ⋅ F = (V ⋅ R) ⋅ (V ⋅ R) = V ⋅ R ⋅ R ⋅ V = V ⋅ V

And that the principal values of V are

(LF /Lo )1 0 0
⎡ ⎤

VPr = ⎢ 0 (LF /Lo )2 0 ⎥

⎣ ⎦
0 0 (LF /Lo )3

Why V ⋅ R?

One might wonder why we are focusing on F ⋅ FT = V ⋅ VT


rather than FT ⋅ F = UT ⋅ U. This is because
V expresses the
deformations in the final rotated configuration, which is consistent with the Cauchy stress being based on a final deformed and rotated
configuration as well. In contrast, recall that U expresses deformations in the unrotated configuration.

Nevertheless, it should be noted that the invariants of


U and V are equal! This might make more sense when one
recalls that
V = R ⋅ U ⋅ R , so the two
differ only by a rigid body rotation. This makes it possible to also
express everything in terms of U rather
T

than V if desired, although a little more complexity may result.

The ratios of (LF /Lo )i are replaced by the single symbol, λi ,


called stretch ratios. (Note that λi = 1 + ϵi where ϵi is the ith principal
engineering strain.)
So the above tensor becomes.

λ 0 0
www.continuummechanics.org/mooneyrivlin.html ⎡ ⎤ 5/32
26.8.2021 Mooney-Rivlin
λ1 0 0
⎡ ⎤
VPr = ⎢ 0 λ2 0 ⎥
⎣ ⎦
0 0 λ3

Therefore, the principal values of V ⋅ V are

2
λ 0 0
⎡ 1 ⎤
T 2
(V ⋅ V )Pr = ⎢
0 λ 0 ⎥

⎢ 2 ⎥

⎣ 2 ⎦
0 0 λ
3

is symmetric, so it does
have three invariants. (Recall that invariants are independent of
coordinate transforms.)

T
V ⋅ V

2 2 2
I1 = λ + λ + λ
1 2 3

2 2 2 2 2 2
I2 = λ λ + λ λ + λ λ
1 2 2 3 1 3

2
VF
2 2 2 2
I3 = det(. . .) = λ λ λ = ( ) = J
1 2 3
Vo

Keep in mind that these equations for the invariants are special cases because they are of a symmetric
matrix in its principal orientation, so off-
diagonal terms can be neglected because they are zero. But this does not mean that they are incorrect in anyway.

VF
I3 merits a few additional comments. First, it is the square of the Jacobian, J ,
because λ1 λ2 λ3 = (
Vo
) , and therefore
I3 is the square of the
volume ratio. Also, for incompressible materials, I3
always equals 1, and therefore λ1 λ2 λ3 = 1 .

The incompressibility relationship is often used to simplify the terms of the invariants,
particularly I2 . For example, note that the first term in I2 ,

λ
2
λ
2
is equal to 1/λ2 This can also be applied to the two remaining terms to give the following set of equations
www.continuummechanics.org/mooneyrivlin.html 6/32
26.8.2021 Mooney-Rivlin
λ
1
λ
2
, is equal to 1/λ3 . This can also be applied to the two remaining terms to give the following set of equations.

2 2 2
I1 = λ + λ + λ
1 2 3

1 1 1
I2 = + +
2 2 2
λ λ λ
1 2 3

2
I3 = (λ1 λ2 λ3 ) = 1

Finally, it is worth noting that the invariants do not equal zero in the
undeformed state, i.e., when the strains are zero. Keep in mind that λ = 1
when ϵ = 0. Therefore, the undeformed state corresponds to
I1 = 3, I2 = 3, and I3 = 1. Also keep in mind that I3 will
always equal 1 for an
incompressible material. So it cannot reflect any deformation state since its value never changes. This leaves I1 and I2
to do all the work.

Mooney-Rivlin Models
The Mooney-Rivlin class of models expresses the mechanical strain energy as a sum of the invariants as follows.

i j 2
W = ∑ ∑ Cij (I1 − 3) (I2 − 3) + D (J − 1)

i j

Note that the series is not a function of I3 because it is a constant value, 1.


The constants, Cij and D, will be determined by curve-fitting
measured
stress-strain curves to the derivative of the equation. The number of terms in the expansion is determined by the application's accuracy
requirements.

As an example, the first few terms of the series are

2 2
W = C10 (I1 − 3) + C01 (I2 − 3) + C11 (I1 − 3) (I2 − 3) + C20 (I1 − 3) + +D (J − 1)
www.continuummechanics.org/mooneyrivlin.html 7/32
26.8.2021 Mooney-Rivlin
W = C10 (I1 − 3) + C01 (I2 − 3) + C11 (I1 − 3) (I2 − 3) + C20 (I1 − 3) + . . . +D (J − 1)

Incompressibility

The final D(J − 1)2 term is unique. Despite all the talk of incompressibility, this term relies on the material model actually being ever-
so-slightly compressible so that J varies slightly from 1. We will see that it reflects the finite
bulk modulus of a nearly incompressible
material. It also functions somewhat (though not exactly) like hydrostatic stress.

Each principal Cauchy stress is related to the derivative of the above equation with respect to the corresponding λ. For example, the 1st principal
Cauchy stress corresponds to
derivatives of W with respect to the first stretch ratio, λ1 .

∂W ∂W ∂I1 ∂W ∂I2 ∂W ∂J
σ1 = λ 1 = λ1 ( + + )
∂λ1 ∂I1 ∂λ1 ∂I2 ∂λ1 ∂J ∂λ1

The derivatives of the strain energy with respect to the invariants, and J , are

∂W ∂W
= C10 + C11 (I2 − 3) + 2C20 (I1 − 3) + . . . = C01 + C11 (I1 − 3) + . . .
∂I1 ∂I2

∂W
= 2 D (J − 1)
∂J

And the derivatives of the invariants, and J , with respect to λ1 are

∂I1 ∂I2 2 ∂J
= 2λ1 = − = λ2 λ3
3
∂λ1 ∂λ1 λ ∂λ1
1

www.continuummechanics.org/mooneyrivlin.html 8/32
26.8.2021 Mooney-Rivlin

All of these terms can be combined to give polynomials relating stretch ratios
to principal stresses, with coefficients such as C10 , C01 , C11 , and
C20
that are determined from curve-fitting these equations to experimental data.

The Simplest Case - Treloar's Law

The simplest case of all the Mooney-Rivlin type models is the so-called
Treloar's Law. In this case, only C10 is used, and all other
constants are ignored, as if they are zero. (Note that it is always
necessary to use D.) This produces

2
W = C10 (I1 − 1) + D(J − 1)

Taking derivatives with respect to λ1 to obtain σ1 gives

∂W ∂I1 ∂W ∂J
σ1 = λ 1 ( + )
∂I1 ∂λ1 ∂J ∂λ1

= λ1 [C10 (2λ1 ) + 2 D (J − 1)(λ2 λ3 )]

2
= 2 C10 λ + 2 D (J − 1)(λ1 λ2 λ3 )
1

Once again, the D term requires some explanation. First, since J = VF /Vo ,
then (J − 1) = ΔV /Vo = ϵVol . So we will replace
(J − 1)
with the volumetric strain. Second, the product (λ1 λ2 λ3 )
is also J or VF /Vo . We will assume that this value is so close to 1

for an incompressible material that it can be dropped.

An alert reader will notice an apparent contradiction between the dropping of J


and the retaining of (J − 1) as ϵVol . It is trully
suspicious
at first glance. The justification is that (J − 1) represents the difference of two relative large numbers to give a small result
that is nevertheless critical to the build-up of hydrostatic stresses. This difference may be small compared to 1, but it is not small when
compared to a reference volumetric strain of zero. In contrast, once the small value is obtained, the effect of multiplying it by J again is
www.continuummechanics.org/mooneyrivlin.html 9/32
26.8.2021 Mooney-Rivlin

negligible.

So the equation can be written as

2
σ1 = 2 C10 λ + 2 D ϵVol
1

Likewise, the equations for the other two principal Cauchy stresses are

2
σ2 = 2 C10 λ + 2 D ϵVol
2

2
σ3 = 2 C10 λ + 2 D ϵVol
3

The constants are determined by curve-fitting the equations to laboratory


test data. They can then be input into finite element programs
to describe the material behavior.

Note that 2D serves as the bulk modulus, K , in the equations.


Therefore, the proper value for D is K/2.

Popular Coefficients
It is popular to choose the following coefficients: C10 , C01 ,
and C20 . These three lead to the following equation(s) to be
curve-fit to experimental
data.

3
σ1 = λ1 {[C10 + 2 C20 (I1 − 3)][2λ1 ] + [C01 ][−2/λ ]} + 2 D ϵV l
www.continuummechanics.org/mooneyrivlin.html 10/32
26.8.2021 Mooney-Rivlin
σ1 = λ1 {[C10 + 2 C20 (I1 3)][2λ1 ] + [C01 ][ 2/λ ]} + 2 D ϵVol
1

1
2 2 2 2 2
= 2 C10 λ − 2 C01 + 4 C20 λ (λ + λ + λ − 3) + 2 D ϵVol
1 1 1 2 3
2
λ
1

Likewise, the equations for σ2 and σ3 are

1
2 2 2 2 2
σ2 = 2 C10 λ − 2 C01 + 4 C20 λ (λ + λ + λ − 3) + 2 D ϵVol
2 2 2 1 2 3
λ
2

1
2 2 2 2 2
σ3 = 2 C10 λ − 2 C01 + 4 C20 λ (λ + λ + λ − 3) + 2 D ϵVol
3 2 3 1 2 3
λ
3

Don't forget that the stresses here are Cauchy stresses, not engineering stresses.

Keep in mind that there is no physical insight built into these equations! They
are merely a collection of polynomials that happen to curve this way
and that,
in ways that are similar to the stress-strain curves one gets
from actual laboratory tests on rubber test specimens. The equations also
possess one additional key critical property. As will be seen shortly,
they predict higher principal stresses for equibiaxial tension than plane strain
tension, and then higher again against uniaxial tension.

Specific Deformation Modes


Recall the three main deformation modes covered on the
special strain topics page.
These constitute the two extreme deformation modes, along
with the plane strain tension mode midway between the two.
Note that plane strain tension has the same principal values
as the pure shear case
when strains are small. For this reason, it is often compared to shear deformations.

Developing Mooney-Rivlin equations for these special cases appears to be a very popular thing to do. There are a couple reasons for this.
First,
the deformation modes are (relatively) easily accomplished in lab tests, at least the uniaxial tension case is. The second
reason is that the
ti b i l t d t li i t th b lk
www.continuummechanics.org/mooneyrivlin.html d l hi h k lif littl i Thi ill b d t t d t 11/32
26.8.2021 Mooney-Rivlin
equations can be manipulated to eliminate
the bulk modulus, which makes life a little easier.
This will be demonstrated next.

The key to eliminating the bulk modulus for these three special cases is to recognize that
σ3 = 0 in each. Therefore, one can subtract two
equations from
each other to accomplish σ1 − σ3 as follows.

1 1
2 2 2 2 2 2 2
σ1 − σ3 = 2 C10 (λ − λ ) + 2 C01 ( − ) + 4 C20 (λ − λ )(λ + λ + λ − 3)
1 3 2 2 1 3 1 2 3
λ λ
3 1

But σ3 = 0 , so

1 1
2 2 2 2 2 2 2
www.continuummechanics.org/mooneyrivlin.html 2C (λ λ ) 2C ( ) 4C (λ λ )(λ λ λ 3) 12/32
26.8.2021 Mooney-Rivlin
2 2 2 2 2 2 2
σ1 = 2 C10 (λ − λ ) + 2 C01 ( − ) + 4 C20 (λ − λ )(λ + λ + λ − 3)
1 3 2 2 1 3 1 2 3
λ λ
3 1

An alternative popular form is obtained by dividing through by λ1


to obtain engineering stress. It is this form that is often used to
determine the
Mooney-Rivlin coefficients by fitting the equation(s) to experimental data.

2 2
λ 1 1 λ
Eng 3 3
2 2 2
σ = 2 C10 (λ1 − ) + 2 C01 ( − ) + 4 C20 (λ1 − ) (λ + λ + λ − 3)
1 2 3 1 2 3
λ1 λ1 λ λ λ1
3 1

Uniaxial Tension

Recall that for uniaxial tension in the 1 direction, incompressibility requires

1
λ1 = λ λ2 = λ3 =
√λ

Substituting these into the equation for engineering stress gives

Eng 1 1 1 2
2
σ = 2C10 (λ − ) + 2C01 (1 − ) + 4 C20 (λ − ) (λ + − 3)
Uniaxial 2 3 2
λ λ λ λ

Plane Strain Tension (Shear)

For this case

1
λ1 = λ λ2 = 1 λ3 =
www.continuummechanics.org/mooneyrivlin.html 13/32
26.8.2021 Mooney-Rivlin
λ1 λ λ2 1 λ3
λ

And the equation for stress is

Eng 1 1 1 1
2
σ = 2 C10 (λ − ) + 2 C01 (λ − ) + 4 C20 (λ − ) (λ + − 2)
Plane 3 3 3 2
λ λ λ λ
Strain

Tension

Equibiaxial Tension

For this case

1
λ1 = λ2 = λ λ3 =
2
λ

And the stress is

Eng 1 1 1 1
3 2
σ = 2C10 (λ − ) + 2C01 (λ − ) + 4 C20 (λ − ) (2λ + − 3)
Equibiaxial 5 3 5 4
λ λ λ λ

Uniaxial Tension

This shows the basic curves for uniaxial tension when each coefficient is
set to 1.

www.continuummechanics.org/mooneyrivlin.html 14/32
26.8.2021 Mooney-Rivlin

Stress-Strain Curves

For the case where C10 = 1 and all other constants are zero...

www.continuummechanics.org/mooneyrivlin.html 15/32
26.8.2021 Mooney-Rivlin

For the case where C01 = 1 and all other constants are zero...

www.continuummechanics.org/mooneyrivlin.html 16/32
26.8.2021 Mooney-Rivlin

For the case where C20 = 1 and all other constants are zero...

www.continuummechanics.org/mooneyrivlin.html 17/32
26.8.2021 Mooney-Rivlin

MA10 and MA100


MA10 and MA100 are secant moduli at 10% and 100% engineering strain under uniaxial tension conditions. λ = 1.10 for MA10 and
λ = 2.00
for MA100. The way to compute MA10
from Mooney-Rivlin coefficients is to first calculate the stress
at λ = 1.10 and then divide that stress by
ϵ = 0.10. The equation for uniaxial tension is

1 1 1 2
Eng 2
σ = 2 C10 (λ − ) + 2 C01 (1 − ) + 4 C20 (λ − ) (λ + − 3)
2 3 2
λ λ λ λ

Insert λ = 1.10 into the equation to get

Eng
σ = 0.547 C10 + 0.497 C01 + 0.031C20
10%

www.continuummechanics.org/mooneyrivlin.html 18/32
26.8.2021 Mooney-Rivlin

And divide by 0.10 to get MA10.

MA10 = 5.47 C10 + 4.97 C01 + 0.31 C20

And MA100 is

MA100 = 3.5 C10 + 1.75 C01 + 14 C20

For reference, the slope at zero strain is the following

dσ ∣
Eϵ=0 = ∣ = 6 (C10 + C01 )
dλ ∣λ=1

The C20 term has zero slope at ϵ = 0 .

Experimental Data

www.continuummechanics.org/mooneyrivlin.html 19/32
26.8.2021 Mooney-Rivlin

www.continuummechanics.org/mooneyrivlin.html 20/32
26.8.2021 Mooney-Rivlin

www.continuummechanics.org/mooneyrivlin.html 21/32
26.8.2021 Mooney-Rivlin

www.continuummechanics.org/mooneyrivlin.html 22/32
26.8.2021 Mooney-Rivlin

Found on a Forum...
Problem with parameters in Mooney-Rivlin model

Hello,

I want to use Mooney-Rivlin parametrs from the paper publish in the internet - they received parameters :

www.continuummechanics.org/mooneyrivlin.html 23/32
26.8.2021 Mooney-Rivlin

C10: 2.7507e-2
C01: 1.3077e-1
C11: 6.8027e-2
D: 4.8694e-2

My problem is :

Authors of these paper didn't mention what units of stress and strain they used to obtained the C10, C01, C11, D parameters. Is the
units of input data of stress-strain important?

21 Comments Continuum Mechanics 🔒 Disqus' Privacy Policy 


1 Login

 Recommend t Tweet f Share Sort by Oldest

Join the discussion…

LOG IN WITH
OR SIGN UP WITH DISQUS ?

Name

Shreeram S • 3 years ago • edited


Hi Bob,

Thank you so much for the coherently written amazing content.

I think you should probably stop doing whatever you are doing right now and look for a teacher's (Professor's) job.

Never mind if you are already one. :)

Cheers,

Shree
△ ▽ • Reply • Share ›

www.continuummechanics.org/mooneyrivlin.html 24/32
26.8.2021 Mooney-Rivlin

BobMcG Moderator > Shreeram S • 3 years ago


LOL. Thanks. I moonlight as a professor.
△ ▽ • Reply • Share ›

Shreeram S > BobMcG • 3 years ago


Haha.. I guessed so. :)

Thank you once again and looking forward to your other websites :)
△ ▽ • Reply • Share ›

MatteC • 3 years ago


Hi,

Thank you for doing a great job at explaining the concept. You mentioned that the V matrix is easier to use than the U matrix from the polar
decomposition to solve for the invariants. What additional considerations would be required if the U matrix is being used. I am thinking of using
Singular Value Decomposition to solve for the U matrix, and then using the main diagonal for the invariants. Would this be a correct method?

I was also wondering what the relation was from the strain energy density to the Young's modulus, or how exactly the Mooney-Rivlin is used to
create a stress strain relationship? If used in FEA then an elements Young's modulus could be used to either update stiffness matrix, or find how
hard one element is to deform versus an another.

Thanks again,

Chris
△ ▽ • Reply • Share ›

BobMcG Moderator > MatteC • 3 years ago

Hi Chris. Thanks for your interest. Concerning using SVD to get U and main diagonal components for the invariants.... I can't be certain
that's correct. And in fact, I think it is not. One reason I say that is because I don't see how to get U from an SVD. If it can, then great.
Recommend making up a numerical example and trying it out to confirm one way or another. (I'm way too busy to do so myself at the
moment.)

As for relationship between strain energy density (SED) and Young's Modulus (E), it is simple for materials that behave like Hooke's Law,
but doesn't exist for materials that don't. Recall that SED is the integral of sigma * d_epsilon, so it's the area under a stress-strain curve. If
material follows Hooke's Law, then it would be SED = 0.5 * sigma * epsilon (SUMMED OVER EACH COMPONENT OF THE TENSORS!).
But since sigma = E * epsilon, this can be subbed into SED equation to get 0.5 * E * epsilon^2 and/or 0.5 * sigma^2 / E. But if material does
not follow Hooke's Law then about all you can do is go back to the principle that it is the area under a stress-strain curve (Note that E is
www.continuummechanics.org/mooneyrivlin.html 25/32
26.8.2021 Mooney-Rivlin
not follow Hooke s Law, then about all you can do is go back to the principle that it is the area under a stress strain curve. (Note that E is
not in a Mooney-Rivlin model, at least not explicitly.)
△ ▽ • Reply • Share ›

MatteC > BobMcG • 3 years ago


Thanks for the quick response, very informative.

My idea for the SVD came from a few sources, but here is a link where it is outlined: http://buzzard.ups.edu/cour...

I will try creating a mathematical example for verification just to be certain.

Since hyperelastic materials don't follow Hook's law as you mentioned, would it then be fair to assume that the ∑∑Cij(I1−3)i(I2−3)j
represents a tensile modulus at a given deformation? In this sense it can be a measure of relative hardness(effort required to
deform) of two hyperelastic elements for FEA.
△ ▽ • Reply • Share ›

Julio Fernández • 3 years ago


Hi Bob,

Thank you for this great website! It is amazing!

I have doubts regarding the calculation of Strain Energy Density (SED) for the particular case of volumetric compression. The problem is that the
result of SED from the hyperleastic function is different to the result of SED calculated by integrating the product of Cauchy stress by the increment
of true strain.

Let me show you a simple example:

Material: Neo-hookean, C10 = 1 MPa, D = 500 MPa.

Engineering strain = -0.3 (in each of the three directions).

J = epsilon_vol = (1 - 0.3)^3 = 0.343

W = D * (J - 1)^2 = 215.8245 MPa

true_strain = ln(1 + epsilon) = ln(1 - 0.3)

true_strain = -0.356675

Cauchy_stress = 2*D * (J - 1) = 2*500 * (0.343 - 1)

Cauchy_stress = -657 MPa

see more

www.continuummechanics.org/mooneyrivlin.html 26/32
26.8.2021 Mooney-Rivlin

△ ▽ • Reply • Share ›

BobMcG Moderator > Julio Fernández • 3 years ago

Thanks for your interest in the website, Julio. And no, I don't see why you got different results, at least based on an initial quick glance.
Certainly seems like both values should be the same. But evidently, we're both overlooking something simple.
△ ▽ • Reply • Share ›

Julio Fernández > BobMcG • 3 years ago


Thanks for your reply, Bob. I have tried to validate the problem with FEA by means of a single element of unit lenght.

Cauchy stress and true strain agree perfectly between FEA and theoretical calculations.

True strain = -0.3567

Cauchy stress = -657 MPa

However, again, the problem is the calculation of Strain Energy Density (SED):

SED_FEA = 629.2 MPa

SED = D * (J-1)^2 = 215.8245 MPa

SED_trapezoidal_integration = 413.025 MPa

Below, you can see the comparison between different calculations of SED. I have used 1000 increments for the trapezoidal
integration as well as for the FEA. Why do you think FEA and theoretical calculations are so different if cauchy stress and true strain
are exactly the same? This is crazy!

Any idea from your side would be great and very valuable for me. Thanks a lot in advance.

Kind regards,

see more

△ ▽ • Reply • Share ›

Julio Fernández > Julio Fernández • 3 years ago


Hi again Bob,

just an advance regarding my doubts. The FEA results agree perfectly well with the theory, the only difference is that the
theory provides the SED over undeformed volume (V0) and FEA provides the SED over deformed volume (Vf). Thus, both
www.continuummechanics.org/mooneyrivlin.html 27/32
26.8.2021 Mooney-Rivlin
theory provides the SED over undeformed volume (V0) and FEA provides the SED over deformed volume (Vf). Thus, both
results are related each other by J.

SED_theory * V0 = SED_FEA * Vf

SED_theory = SED_FEA * J

215.82 = 629.2 * 0.343

On the other hand, the calculation of SED by integrating Cauchy stress and True strain remains unsolved. I have realized
that the results of the integration agree perfectly well with the theory if I use nominal_stress and nominal_strain instead of
Cauchy_stress and true_strain. However, this only works for the particular case of volumetric compression (deviatoric part of
SED = 0). Do you know why this happens?

Am I wrong integrating Cauchy stress and true strain?

Thank you very much!

Kind regards,

Julio
△ ▽ • Reply • Share ›

BobMcG Moderator > Julio Fernández • 3 years ago

Hi Julio,

Good catch there regarding theory * V0 = FEA * Vf. I think that makes sense assuming FEA is always using current values
and parameters. However, I'm with you on the question of using Cauchy stress and true strain. That must be an acceptable
combo to use because they are energetically conjugate. Sorry I'm not more help.
△ ▽ • Reply • Share ›

Julio Fernández > BobMcG • 3 years ago


Thanks Bob!

An additional point, pictures below show the proof that, in volumetric compression, the SED corresponds to the integration of
Nominal Stress and Nominal Strain. For me, it does not make any sense. As far as I know, energetic conjugates are
Cauchy_stress-True_strain or 2nd Piola-Kirchhoff Stress and Green-Lagrange Strain.

Moreover, this proof only works for this particular case of volumetric compression. When deviatoric component is involved in
the strain tensor, the integration of Nominal Stress and Nominal Strain does not work.

This is a mess!
www.continuummechanics.org/mooneyrivlin.html 28/32
26.8.2021 Mooney-Rivlin
This is a mess!

If I find something new, I will let you know.

Thanks again,

Julio

see more

△ ▽ • Reply • Share ›

Emre Baris Yildiz > Julio Fernández • 2 years ago


Hello Mr. Fernández,

In your Cauchy Stress calculations, you shouldn't take J as 1. Mr. McGinty has pointed out to eleminate the J in the Cauchy
Stress calculations if the material is incompressible. However, in your example you have negative strains in all principal
directions meaning you have a high contraction and J<<1. I have calculated the same example in MATLAB using discrete
integration and if you do not omit the J term you would get the same Strain Energy Density from the W calculation. Also, the
first strain energy term C10*(I1−1) also contributes to strain energy even tough it is much smaller than D*(J-1)^2

Sigma(Cauchy)_i = 2*C10*λ_i^2+2*D*(J-1)*J

W=integral(sigma(Cauchy)*dEpsilon(True)) from Epsilon_true=0 to Epsilon_true_final

Use of Incompressible assumptions to compressible problems are problematic.

An also a correction for your solution to your problem:

At the start Mr. McGinty derived the "Cauchy Stress" with the assumption of incompressibility. Remember Piola Kirchoff
Stress is sigma_PK2= J * F^-1 *sigma(cauchy) *F^-T where F is the deformation gradient. In your problem F = V where F
has the stretches along the diagonals. It means that sigma(cauchy)_i = lambda_i^2 * sigma_PK2 * J which means we have
to multiply with J to get Cauchy Stress! What he has calculated is the Kirchoff stress therefore your equations hold if we take
see more

www.continuummechanics.org/mooneyrivlin.html 29/32
26.8.2021 Mooney-Rivlin

△ ▽ • Reply • Share ›

Shreeram S • 2 years ago


Hi Bob,

I have a small doubt when it comes to your compressible form of M-R model (also in Treloar's law). As you have included a term with J (assuming
that the model is compressible), should we not be using the invariants of the 'Unimodular Right Cauchy Green deformation tensor'? i.e. using
invariants of C_overbar instead of C, where C_overbar = (F_overbar^T x F_overbar).

Would be happy to really get your insights on this.

Thank you in advance!

Shree
△ ▽ • Reply • Share ›

BobMcG Moderator > Shreeram S • 2 years ago

Shree, unfortunately, I have no idea. I really don't know.


△ ▽ • Reply • Share ›

Dr. B R Gupta • 2 years ago


Thanks Bob for a delightful way to explain the basics of Mooney-Rivlin model. I was trying to understand it from different publication but could nod
not find such an easy to understand way of presentation.

Thanks again

Dr. B R Gupta brg123@hotmail.com


△ ▽ • Reply • Share ›

BobMcG Moderator > Dr. B R Gupta • 2 years ago

Thanks!
△ ▽ • Reply • Share ›

Antonio Andrés Umaña Rodríguez • 6 months ago


Hi Bob!

Thanks for all the content!

Could you tell me if you know which constants should I use for Viton? (FKM)
www.continuummechanics.org/mooneyrivlin.html 30/32
26.8.2021 Mooney-Rivlin
Could you tell me if you know which constants should I use for Viton? (FKM).

Thank you in advance!

Antonio
△ ▽ • Reply • Share ›

BobMcG Moderator > Antonio Andrés Umaña Rodríguez • 6 months ago

Hi Antonio. Thanks for the post. I have never even heard of Viton! The answer depends on the shape of the stress-strain curve and how
accurately do you want to fit it.
△ ▽ • Reply • Share ›

Antonio Andrés Umaña Rodríguez > BobMcG • 6 months ago


Hi Bob.

Thank You Material Behavior Chapter - $4.99 Entire Website - $12.95

Thank you for visiting this webpage. Feel For $4.99, you receive two formatted For $12.95, you receive two formatted
free to email me if you have questions.
PDFs
(the first for 8.5" x 11" pages,
the PDFs
(the first for 8.5" x 11" pages,
the

second for tablets) of the materials second for tablets) of the entire
website.

Also, please consider visiting an chapter.


advertiser above.
Doing so helps to cover
Click here
to see a sample page in each
website hosting fees. Click here
to see a sample page in each of the two formats.

of the two formats.



Email address to receive PDFs
Bob McGinty

bmcginty@gmail.com Email address to receive PDFs


  Copyright © 2012 by Bob McGinty

www.continuummechanics.org/mooneyrivlin.html 31/32
26.8.2021 Mooney-Rivlin

Table of Contents
Metal Plasticity Dynamic Material Properties

www.continuummechanics.org/mooneyrivlin.html 32/32

You might also like