1

Advanced Finite Element Analysis

Prof. R. KrishnaKumar
Department of Mechanical Engineering
Indian Institute of Technology, Madras

Lecture - 25

We were discussing the constitutive equations and we had discussed quite a bit of
both what is hyperelastic material, the philosophy behind it.

(Refer Slide Time: 1:01)


The next two classes, we will quickly summarise the equations which are necessary
for us to go forward to look at the finite element implementation of the hyperelastic
case. I told you right in the beginning that I may not be deriving all the equations,
may be this class I will derive a few of them; may be next class, I may not derive the
equations, I am going to give you the final versions of it, so that we will use that for
our finite element work later. Let us look at some specific things in hyperelasticity. In
fact, if you look at the materials that come under or that can be classified under
hyperelastic materials, they can be classified as of course, isotropic and anisotropic. I
am not going to cover the anisotropic part in this course, because that is again a huge
area. You need to know lot more of continuum mechanics in order to cover
anisotropic part of it. May be towards the end if there is time, I will indicate how we
approach the anisotropic problem, but the applications of anisotropic hyperelasticity is
2
not very high. Except biological materials, you really do not use anisotropic
hyperelasticity to that extent. In fact, I will just comment on that in a minute, let me
complete this, the other classification.

The other one is, of course, isotropic hyperelastic materials. The other type of
classification comes from whether the material is compressible or incompressible. So,
you will say that incompressible and compressible. Essentially what we are going to
do now is that for hyperelastic material which undergoes say, isotropic incompressible
we are going to put down some equations in terms of general size that is general strain
energy function. Then, we will go to specific strain energy function and state some of
them.

(Refer Slide Time: 3:35)


Essentially, what I am going to cover is isotropic, isotropic incompressible which has
the maximum amount of, I would say, applications, so, isotropic incompressible
hyperelasticity. For sake of completion also I am going to state the isotropic
compressible hyperelasticity. I am not going to the details, but I will also complete in
the next class what are called as isotropic compressible hyperelasticity. What I am not
going to cover, though I may state one or two sentences, is the anisotropic
incompressible hyperelasticity and anisotropic compressible hyperelasticity or in
other words, anisotropic hyperelasticity I am not going to deal with in this course.

3
Let us, just before we go further, because we have made this assumption, let us just
check what this anisotropy is, physically what it means and why we can, for many,
many problems, we can do away with anisotropy and use isotropic materials. For
many practical problems, for example, if you take say tyres, tyres is basically, I had
talked about hyperelasticity; most of them are incompressible hyperelastic materials
and of course, you can also classify another category of materials called isotropic
nearly incompressible materials also, but usually do not put down a separate
constitutive equation for it. You can also say that there are some materials which are
nearly incompressible, but many times we treat, because of the finite element
formulations we treat, incompressible material to be nearly incompressible and so on.
That we will discuss when we come to the finite element pattern after two classes, but
what I want to specify is that, say for example, if you take tyre, then there are
reinforcements running that makes the material composite or that makes in a gross
sense, the behaviour to be anisotropic.

In other words, here are reinforcements which can be identified say that, this is the
steel reinforcement which can be identified and then you look at its behaviour along
with the rubber matrix. If this is the case, then you do not write down separately or
you do not classify this as anisotropic on a micro scale. What you usually do is to
write down isotropic behaviour or isotropic constitutive equation for the matrix
material and then take that reinforcement separately and usually put special elements
which are called as reinforcement elements or re bar elements or in other words, you
treat them as two different materials, as two different materials and there are ways of
handling them in finite element analysis.

On the other hand, you have biological materials, where there are collagen fibres
tens of thousands of millions of them. If you take a piece, there will be so many of
them. You cannot one by one pull up and say that, yeah, this is the collagen fibre; this
is the fibre that is running and this is the other fibre. So, I will put one special element
for this fibre another for this fibre and so on. In other words, it is so very intermingled
at a scale, at a much lower scale, that it will not be very easy to separate them out and
treat them as a separate material and say that I have a base material which behaves in
an isotropic fashion and I on top of it I have material which behaves, which makes it
to behave anisotropic. I cannot say that take a piece; the whole piece behaves as an
4
anisotropic, as an anisotropic material. There is always a problem with that an;
anisotropic and when I say an isotropic, it looks as if it is anisotropic. So, I said an
anisotropic material.

In this case, if I have collagen fibres that are running and I take biological specimens,
I want to treat it using finite element analysis; I have to necessarily treat them as
anisotropic material. Now, I will also indicate one or two other things with anisotropic
materials, but the, but the major good thing is that, the major thing is that the
procedure that I am going to put forward now is exactly the same for anisotropy. J ust
that we do not have time enough, you know, time to, enough time to complete this
whole anisotropic material behaviour. It is only that, but the, but the procedure is
exactly the same. So, that way if you want to go back and read a paper, it will not be
anyway different. I will also indicate what they are in a minute.

(Refer Slide Time: 9:10)


One of the most important theorems that we are going to look at is what is called as
representation theorem; representation theorem or sometimes people call this as
representation theorem of invariance. What it means is very simple.




5
(Refer Slide Time: 9:39)


What it means is that, if you have a scalar value tensor function, nothing great, your
strain energy, scalar value, tensor function which means that it is a function of tensor
say C, such that it is not affected by rotations or in other words, if it is isotropic, if it is
isotropic, then this function can be written in terms of the invariance of the tensor
valued function. Though the proof is very complex, are quite involved, I am not going
to prove this, but it is very simple to understand. What it means is that for isotropic
materials, now how do I express when I, when I say C and how do I express this as
isotropic materials.

(Refer Slide Time: 10:38)

6
Remember, yesterday we wrote or last class we wrote that this, if I have to have this
to be an isotropic material, then this has to be written as or this is equal to F of Q
transpose. We already said that the strain energy function can also be written in terms
of C and E, because they are all functions of say, U and so on. Suppose I write this as
C, let us see in a minute, tell me how this should behave, this function should behave
if it were to be an isotropic material. What is C? No, no, what is C in terms of Yes,
yeah, F transpose F, so, just check how this should be. Note that F varies as or if I just
for a minute call that as F star, so, F is equal to F Q transpose. Right Cauchy tensor or
in other words that is F transpose F. So, this should be in terms of star. How it should
be, the stared co-ordinate? J ust substitute from here, you will get that, that has to be
Q
C
Q transpose, just verify that.

If you can have or if you have a strain energy function like this, with this equation,
then the representation theorem states that such an equation can be written in terms of
the invariance of these tensors. What are this invariance by the way? Yes, they are
different for different things.

(Refer Slide Time: 12:28)


Say I
1
is equal to the invariance of C and so, if you can, if you want to express this
say, in terms of lambdas which are the principle values or lambda squared, which are
the principle values of C, then you can express them as I
1
as lambda
1
squared plus
lambda
2
squared plus lambda
3
squared. Lambda squares are the Eigen values and then
7
I or I
2
rather can be expressed as lambda
1
squared lambda
2
squared lambda
2
squared
lambda
3
squared plus lambda
3
squared lambda
1
squared and 3 is equal to the
determinant of that. So, it is equal to .. of this diagonal matrix and that is equal to
lambda
1
squared lambda
2
squared lambda
3
squared. Some of the text books like Wood
expresses this as lambda
1
power 4 plus lambda
2
power 4 plus lambda
3
power 4. So, if
you see this in a different form, do not worry about it. That is another form of writing
the second invariant.

Yeah, it will not be different, because that can also be shown to be an invariant. You
have to know some invariant theory to say that it can also be written like this. No, no,
see, invariance, what are invariance? What are invariance? Yeah, this will not change
with the co-ordinates. So, you can also show that if you write this in terms of lambda
1

power 4 plus lambda
2
power 4 plus lambda
3
power 4, they also can be formed as an
invariant. J ust as a caution, I am just telling you, if you see different things do not
worry about it. That is another way of writing the second invariance. First and third
invariance are usually what I have written here, they are the same. So, that theorem
gives us a lot of advantage of writing down the stress strain relationship.

There are two ways in which you can write down the relationship. One is straight
away in terms of invariance. Of course, if I do not, if I do not want to write it in terms
of lambda
1
squared, I have to, I mean for, this is one way of writing it; simple to
write, but of course, if this is not given, you can also write in terms of C itself, I
1
I
2

and I
3
and how do I write that?











8
(Refer Slide Time: 15:17)


This is trace of C. This is one way of writing it. But, that is also equal to, I am going
to use that later, trace of C and this is equal to half of trace of C squared minus trace
of C, square of C that is I
2
and I
3
becomes determinant of C. These are the three ways
in which it can be written and now my procedure is very straight forward as to first,
so, what is my goal.

My goal is just to express, so, I have come one step down saying that this strain
energy function which is a function of C, can now be expressed in terms of this
invariance. You know that is the one step which I have come towards writing down
the relationship between deformation and the stress for an isotropic material. What is
the next step? Next step is just to substitute this into my expression, which I had given
you in the last class. What is the expression that we saw in the last class?









9
(Refer Slide Time: 16:45)


P say for example, if you look at S or if you look at P, then we had written that in
terms of dow psi by dow C or in other words, if I remember right, we had written that
as 2 into dow psi by say, dow C and note of course, one of the important thing which I
am not, I may not continue with that line again, this is in the, this is in the reference
configuration; this equation is written in the reference configuration. Since the Eigen
values happen to be the same whether I use C or whether I use b, this whole equation
can again be written in terms of b as well. That is where F is in the left of F transpose
that is in the spatial co-ordinate.

(Refer Slide Time: 17:32)

10

So, I can write this down say, as like that as well; either way I can write it, either in
terms of C or I can write that in terms of P as well.

(Refer Slide Time: 17:54)


So, my job becomes very simple, because, now that I know how to write this for S,
what I need is only that quantity. In fact, in fact you will see that what all we expand
after this is only algebra. Only one more concept is there, after this it is algebra. Let us
see. That is why I said I am not going to derive all the equations. Let us see how we
write this, this strain energy function psi with respect to, differential with respect to C.
Very simple, how do I write it? Chain rule or chain rule that is all; dow psi by dow I
a

dow I
a
dow C or in other words, dow psi by dow I
1
dow I
1
by dow C plus I
2
plus I
3
.
Let us keep this, because this is what will vary from one material to another material.
This function is one which is going to vary or in other words, this function here, what
we have put here, is going to vary from one material to another material or one type of
material to another type of material.

Let us keep that as it is, but let us now look at a general formulation or general
equation. From there we can quickly reduce it to different forms. In fact, that also I
am going to reduce in a minute. So, what is the general equation which everyone
writes? Simple, I need to know, what are the things I need to know? I need to know
dow I
1
by dow C, dow I
2
by dow C and dow I
3
by dow C.
11
(Refer Slide Time: 20:09)


If I know all these things, that is dow I
1
by dow C, I
2
and I can get first one expression
for S, no problem at all, substitute it here. Now if I know this function, I can get
straight away the relationship. Let us see what they are. Let us say what is dow I
1
by
dow C? Have a look at this. Anyway I am, I am going to write the results. May be I
will give you couple of minutes, at least the easier part of it. Let us check whether you
are able to do it. The first two, third one I will derive it. It is slightly more involved
and that is a very important relationship. So, I will derive that; that derivation is quite,
it is not very straight forward; so, I will just derive that part alone, but the rest of it, let
us see whether you are able to do. That is dow by dow C of det C, determinant of C, is
the only expression which requires some amount of manipulation.

Let us write down these two. So, dow I
1
by dow C is equal to what? Fantastic; so, that
is I, sorry, that is I, basically because trace of C that means C
11
plus C
22
plus C
33
that
is what you will get for trace of C. So, dow I
1
dow C of say AB is equal to delta
AB
. In
other words, what it means is dow I
1
by dow C
11
is delta 11 which is equal to 1 and so
on.





12
(Refer Slide Time: 22:13)


So, obviously it is an excellent answer that this is going to be I. Let us see how you
do. dow I
2
by dow C, have a look at this here and see whether you will be able to do
that. Chain rule, do not forget the chain rule. So, first term is what? 2 into trace of C
into, so, first term is half 2 into trace of C. Is it C? Then, dow of trace of C by dow C,
which is I minus, simple, no, dow of 2 C; dow of trace of C squared, which is equal to
2 C. So, that is the first part of it and now most important thing, how do I get and this
and this of course, can be written in a slightly different fashion. Trace of C I can be
written as I
1
. This is I
1
I minus 2 C or minus C. The most important thing is, hey is
that correct? One minute, trace 2 times trace of C, yes, into dow trace of, yeah, that is
correct. So, C squared 2 of 2 trace of C squared dow by dow C squared, no, yeah, C is
there; correct and lastly dow I
3
by dow C; there is dow by dow C of determinant C.
What is that expression?









13
(Refer Slide Time: 24:40)


How do we get this, dow of any quantity? So, dow of dow A say, determinant of A. It
is actually equal to, the result is determinant of A A inverse transpose. Now, to start
with, this is a very standard procedure, actually. Let us say determinant of A plus dA
can be written as determinant of A into I plus A inverse dA which can be written as
determinant of A into, determinant of AB is equal to determinant of A into
determinant of B; so, determinant of A into determinant of I plus A inverse dA. Now,
what is this? Let us take that second term determinant of, let us take determinant of I
plus A inverse dA. Does it ring a bell? This is very similar to the determinant which
we use for characteristic equation, as an Eigen value problem.
What is that? The determinant of say, if I want to write down the characteristic
equation for A, how do I write down? Determinant of A minus lambda I; yes,
compare that and this and say what is the lambda? Say, for example this can be
looked at as if lambda is equal to, if you want you can write that down as determinant
of A inverse dA minus of minus 1 I. Compare these two now. So this is the
characteristic equation for A inverse dA, small jugglery; you will see most of the
derivations are like this, because you want a result and a very nice result. So, usually
what you do is this kind of small mathematical jugglery; that is what you will be
doing, nothing else.

Now, having done that what is my next step? Write this down in terms of the
polynomial expression that you usually do. How do you write this down in a
14
polynomial expression? Yeah, no, no, usual, yeah, in terms of invariance, usual
expression that you write down will be say, lambda Q minus I
1
lambda squared plus I
2

lambda plus I
3
is equal to zero. This is how you would write it. If instead of that, you
just write this down as, so, let us see how you will write it down.

(Refer Slide Time: 28:20)


Substitute for lambda and then, simple, you will get 1 plus the first invariant of A
inverse dA; I am just changing the notation slightly, because I
2
, if I write two here you
will get confused. So, I am just saying that it is two A inverse of dA plus three A
inverse of dA, where I have substituted into my expression of lambda Q minus I
1

lambda squared plus, no; yeah, minus 1. So, lambda, everything will be minus, so,
lambda Q when I substitute, that becomes minus. So, everything will be minus, so, we
can write that down as something like this. This is the characteristic equation which
you can write it down.

Now, what is this? This is 1 plus trace of A inverse of dA plus the second and the
third invariance have higher order terms of dA, squared terms of dA and so, I am
going to, because this is, this determinant is very close to dA, so, I am going to
remove that. I am saying, I am going to say that these are order of dA.



15
(Refer Slide Time: 30:19)


Substituting this into this expression here, you can write that down as determinant of
A into this expression 1 plus or in other words, you can say determinant of A plus
determinant of A trace of A inverse dA and we will stop with that, because other
terms or d squared terms we can neglect or in other words, we are linearizing it. In
other words, it brings out another equation. Of course, this determinant of A trace of
A inverse dA can also be written as, because alpha of determinant of A is equal to
determinant of alpha A, so, we can, we can write this down as trace of that is, sorry,
alpha of trace of A is equal to trace of alpha A. You can say determinant, this will
become, determinant A A inverse dA.

Now, that is the first thing. Since I have normalised it, I can also express the
determinant of A plus dA in terms of the Taylor series expansion. Let us see how we
do the Taylor series expansion. Now, what I am going to do is simple. Let us see how
many of you do it. I will give you two minutes, write this down in Taylor series
expansion determinant of A plus dA, compare the second term of the Taylor series
expansion with this and you get the answer. That is all. So, the procedure is very
simple. Then, the next step is to write this down in terms of Taylor series expansion.
Write it down.



16
(Refer Slide Time: 32:30)


Suppose I have phi of A plus dA, how do I write that down? Suppose I have some
function phi is equal to phi of A plus dow phi by dow phi by dow A dA plus higher
order terms. Here, instead of this phi what is that you have?

(Refer Slide Time: 33:05)


Determinant of C or here, sorry, determinant of A; so, this will be det A dA. Now, let
us see, compare this and that. Yeah, dow of, simple; this becomes, actually this
becomes determinant of A. Compare this expression with this expression what we
have here. The second expression is what we are interested in. So, dow of dow A
17
determinant of A is equal to trace of what? Determinant of A A inverse dA. No, no; I
have not yet come to that. That is, no, no, dA, right; yeah, sorry, this I left out. J ust I
have added that, but what is this? This is nothing but I want to write it as a double
contraction or double dot product. So, this becomes determinant of A A inverse
transpose, noting that determinant of A is equal to A transpose dA. Yeah, this is
simple; we had done that yesterday. What is A double dot B? This is trace of A
transpose B is equal to B double dot A trace of B transpose A and so on. Yesterday
we, we had done that in the last class and so, comparing the left hand side and the
right hand side, very simple, if you, if you have any difficulties, it is quite straight
forward. You say, that B you can write it in terms of indicial notation. A
ij
B
ji
or A
ij
B
ij
,
then you can go on look at the trace of this. You will see that they are the same.

Comparing the left hand side and the right hand side, you will see that dow by dow A
of determinant of A which we are interested in is determinant of A A inverse
transpose. That is the expression we have been looking at. Substitute that or in other
words, A is replaced by C and that is what you get. So, dow of dow C determinant of
C is equal to determinant of C C inverse and determinant of C is nothing but I
3
, so, I
3

C inverse. I have removed many of those things in order derive this. Go and substitute
that back. Now, go back and substitute all these things into my expression for S and
tell me what that expression is.

(Refer Slide Time: 36:27)


18
So, that is the fundamental expression. That is why I am taking time to derive it. Go
back and substitute, you have all those things before; go back and substitute and say
what that S should be. Remember that we had written down this as 2 into, sorry, this
is how we had written it down. So, this will now become, let me write down, let me
substitute for all that things which we did; 2 into dow psi by dow I
1
plus I
1
dow psi by
dow I
2
into I
1
sorry, into I minus dow psi by dow I
2
C. What I am, what I am
essentially doing is to re substitute back what all I did plus I
3
dow psi by dow I
3
C
inverse is my final expression. This is the relationship. Once I give this, you will be
able to find out S. You can also write this in terms of sigma, please write that down.
So, in terms of sigma, sigma is equal to J inverse F S F transpose J F F, I mean, J
inverse F S F transpose. Substitute that back, you can write the expression for sigma
to be 2 J inverse multiplied by or pre multiplied by F post multiplied by F transpose,
you will get that to be I
3
; do that. I am not, as I told you, I am not going to derive all
these things. Dow psi by dow I
3
I, it is a very important expression, dow psi by dow I
1

plus I
1
dow psi by dow I
2
into, if you want to express this in terms of b straight away
instead of C it is more appropriate, so, you can write this as b minus dow psi by dow
I
2
b square.

Note that, note this carefully that we have got this expression from here, which means
that the psis, this free energies are still in terms of C, they are still in terms of C.
Though we have written here b, they are in terms of C. We can also write down sigma
if psi happens to be written as, the free energy function happens to be written in terms
of b as well; you can, you can write that down also.











19
(Refer Slide Time: 40:32)


If that is the case, in fact you can write down sigma to be 2 J inverse dow b b. One of
the beauty of this is that you will see that this expression commute; same way this
expression that you can verify this that also commutes. You can verify this, please
verify it; so, that also commutes. We have now grand expressions for S and sigma.
Now, all other expression that we are going to use actually comes from this. They are
simplifications of this expression. What is the assumption that we have made? What
we have made is that the material is isotropic, material is isotropic. Having made that
assumption, we get this. From here, you can follow through for incompressibility,
compressibility and so on. Now, it is usually customary to write down not this free
energy function, not in terms of I
1
, but in terms of lambda
1
, lambda
2
and lambda
3
.











20
(Refer Slide Time: 42:00)


You can write down in fact this in terms of lambda
1
, lambda
2
and lambda
3
. That is the
usual way of writing down this expression. I am not going to derive this and again it is
a big derivation. We can write down P
a
, the principle stress of or in other words, a
corresponding Piola-Kirchhoff stress, a varies from 1 to 3, to be dow psi by dow
lambda
a
, often used expression. S
a
, which is the second Piola-Kirchhoff stress is
written as 1 by lambda
a
dow psi by dow lambda
a
and sigma J inverse lambda
a
. These
are the expressions in terms of the principle stretch values. In other words, when I
express these strain energy functions in terms of principle stretch values, which I did
yesterday, for example, for Ogden model what we did was to express these things.
Remember that sigma sigma of alpha
p
by mu
p
lambda power alpha
p
and so on; you
know we had expressed that in terms of an expression.

So, lambda
a
lambda
1
, lambda
2
and lambda
3
and if you remember that we had used for
Neo-Hookean and for the Yoehs model in terms of Is, I
1
and I
2
; remember that we
had used that as well. Now note that, one more small thing which you can, which you
can see that when F is equal to 1, that in other words, it is necessary that when there is
no deformation, lambdas are what? 11.




21
(Refer Slide Time: 44:19)


So, I
1
becomes 3, I
2
becomes, sorry, yeah, I
2
becomes 3 and I
3
becomes 1. With this
background, let us now move over to what we call as the incompressible
hyperelasticity, incompressible hyperelasticity.

(Refer Slide Time: 45:05)


What do we mean by incompressible hyperelasticity? What is the condition that I
have to Very good; so, determinant, not zero, so, we said that it cannot be, yeah,
we had put down important conditions that determinant of J is equal to 1. We have to
also install or implement that condition to the or in other words, I should have that to
22
consist of two terms, two terms to take care of the incompressibility as well and the
second term would be in terms of p into J minus 1. We will talk more about that in the
next class. We are now moving to incompressible hyperelasticity, which will have
two functions. Fine, we will stop here. It is a nice time, good time to stop, because we
will start incompressible hyperelasticity in the next class. Is there any question on
what we have done, we will answer that and then close this. Please revise whatever
we have done. We will stop here and will continue in the next class.