Advanced Finite Element Analysis

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

Lecture 1

In this course, we are going to talk about basically non-linear finite element analysis.

(Refer Slide Time: 1:03)

In the last course we have done finite element analysis, which is basically applicable for
linear problems. The concepts that we developed in the previous course will no doubt be
very useful for us in this course, but we have to learn lot more in this course in order that
we can apply the concepts to non-linear problems. Non-linear problems are very
important especially in mechanical engineering, where you come across situations of
non-linearity because of contact, because of material behaviour, as well as because of
change in the displacement constraints and so on. Now, let us look at the type of
problems that we can do using finite element analysis and then we can understand what
the importance of non-linearity is, very clearly.

(Refer Slide Time: 2:24)

If you look at the broad class of problems that we can solve using FEA, you will agree
that it can be broadly classified into static and dynamic problems. Of course, we are
talking only about solid mechanics problems; we will come to other problems which can
be dealt with using finite element analysis such as heat transfer problems, fluid flow and
so on. In fact, today finite element is applied to problems of electromagnetics; things
which you might have thought not possible from the earlier course on linear finite
element analysis itself, because basically that course was targeted to solid mechanics and
to a certain extent heat transfer. But, remember that finite element analysis is a general
technique to solve partial differential equations and once you understand the type of
partial differential equations which govern a particular problem, it is possible to evolve or
to formulate a finite element analysis algorithm and implement that in order to solve this
kind of issues and problems. Though, in this course we will not be dealing with
electromagnetics or fluid flow and so on because that is a topic by itself, we will be
concentrating in this course on solids and solid mechanics problems, especially solid
mechanics problems which can be classified as static analysis cases.

Of course, many of the techniques that we are going to develop for static analysis are
applicable straight away to the dynamic problems and we may have to resort to certain

simple special means to solve them using dynamic analysis; may be towards the end of
the course, we will indicate how for example explicit finite element analysis works in the
dynamic case. But, basically the course will concentrate on non-linear static analysis
using or for application with an aim to apply them to solid mechanics problems. The
advantage here is again that once we see that the techniques are developed for static solid
mechanics analysis, it can be extended very easily not only to dynamic analysis, but
many of the techniques are common to other problems like say, heat transfer and so on,
as well. We need lot of, of course, background theory in order to understand the
application of the non-linear techniques to static problems. But, even before we go to that
level, let us see what the type of problems are that we can do in these two classifications.

(Refer Slide Time: 5:31)

Now if we look at static analysis, we can do of course an elastic problem or elastic

analysis or what I would call as inelastic analysis. In elastic analysis itself this is, this is
as far as material behaviour is concerned, we will come to other geometric issues later.

(Refer Slide Time: 6:10)

As far as the elastic analysis itself, you would see that this can be classified again into
linear elastic and non-linear elastic. We will define very carefully what this non-linear
elasticity means, later in the course. It is very important that we understand lot of
continuum mechanics in order to appreciate what this non-linearity means and what it
involves. So, in this course we are going to cover lot of continuum mechanics portions as
well, so that we will get the overall picture. What we have essentially done is static
elastic, linear elastic that is what the route is, which we have taken. Please note that, this
is not very separate or can be separated from say dynamic analysis. In other words, if you
want to understand the whole picture, you should realize that dynamic analysis for
example if I do an explicit analysis or an implicit analysis or for that matter an Eigen
value problem, whatever I do in dynamic analysis as far as the material behaviour is
concerned we can pluck from this route as well.

(Refer Slide Time: 7:45)

When we talk about inelasticity, of course, one of the most important things is plasticity
and viscoplasticity. We can add if you want, may be one more case which we call as
viscoelastic materials as well. Somewhere between the two, we can say that we have
what is called as viscoelastic materials. This is as far as material behaviour is concerned.
Of course, there are lot more models under say, for example non-linear elasticity. We will
not worry about that right now, but these things can again be classified from the point of
view of the type of deformation that the body would undergo. The body would undergo
either what we call as small deformation or finite or large deformation.

(Refer Slide Time: 8:55)

Please note that the small deformations and finite or large deformations, may be first
thing is that you may be worried what this finite deformation is. You may worry or you
may be worried or you may be wondering small deformation is finite, why is that I put
this finite and large in one single bracket? Note that this is finite versus infinitesimal. So,
small deformation if I have to call, they are called as infinitesimal deformation. It is not
that it is finite deformation and infinite deformation, it is infinitesimal; that means very
small deformations versus finite or large deformation.

These two which define the kinematics and motion and deformation have important
implications as far as definition of strain or deformation and stress is concerned. The way
we have done so far the analysis for linear elastic case, was one which would address the
small deformation problem or infinitesimal deformation problems, where we are not
worried about very rigorous definition of stress and strain. In order to give a small
background even before we proceed and do lot more mathematics, we should understand
from our earlier classes that for example the area of cross section of a piece, say for
example, a bar is important for the definition of stress as well as the length.

(Refer Slide Time: 11:02)

Probably all of you would remember that when you did, say for example, metal forming
problems you would have defined two types of stresses, which you would have called as
engineering stress or nominal stress and true stress. Do you remember what the difference
between the two is? Yeah, the difference is that when I define the stress to be the load per
unit area, it is important that the area is properly addressed; this concept of area is
properly addressed. If the deformations are small or in other words if the area change of
this body is small, then we need not bother or we need not worry whether the initial area
say let me call that as say A0 and the final area after deformations have taken place after
the body has come to equilibrium, because of the application of load are they the same or
different. That is the question.

If the deformations are small, the initial area and the final area say A will not be very
different. We can approximate A0 to be the same as A. So, here we can either put A or
A0, so, that is fine. But, if the deformations are large the body is going to say for
example, you are testing rubber piece. Take a bar of rubber piece and then extend it. Then
definitely you will know, you would have seen it that the area or the cross sectional area
changes drastically; you would have seen that. In fact, the area may reduce 20, 30, 40%,

may be even more, much more, in which case what is the area that you put would decide
the value of the stress.

So, we distinguish between two stress quantities by saying that if you put the current area
of cross section we call that as true stress and when you put or substitute the original area
you call this as nominal or engineering stress. In the same fashion it is important to
realize that this length also plays a role because you would have defined strain in your
earlier classes to be, how did you define?

(Refer Slide Time: 14:19)

For example for this simple bar, change in length by original length very well known
formula, so that we call that as delta l by L and it is important to realize that here again
we have a very similar issue whether you put the original length or the current length,
current length which changes with time. Again, we get into two different strain measures
which you would call as engineering strain or true strain.

Of course, this will result in a measure of log to the base e, natural log and so on. So, you
realize that this is going to have a major role play the measure that we use for defining

strains. So, both the stress measure and the strain measure they not only tell you or
measure the deformation, but their definition itself depend upon the deformation.

(Refer Slide Time: 15:47)

So, you have to really look at these two guys here, whether the deformation is going to be
small or infinitesimal or the deformation is going to be finite or large. You now have an
idea, what we mean by small deformation and what we mean by finite or large
deformation? We have to study really what are, these are okay; I mean these are fine, for
unidirectional case.

(Refer slide Time: 16:15)

Yes; please note, yes, the question is what is the relationship between stress and strain? Is
it that they are functions of time and so on? There are two different types of behaviour,
we will come to that in a minute, whether they are functions of time. But, please note that
we have not yet come into the region of stress strain relationship. We are only talking
about stress measure and strain measure. The relationship between stress and strain are
defined through what are called as, Youngs modulus is only a very specific case, but
they are defined through what are called as constitutive relationships, constitutive
equations which would define that is constitutive relationships or equations which in a
very strict term gives us the behaviour of the body or tells us how the body behaves when
the body is subjected to external loads or forces or influences.

10

(Refer slide Time: 17:11)

The stress strain relationships are given by constitutive equations, which actually result in
the body being classified as linear elastic, non-linear elastic, viscoelastic, plastic,
viscoplastic and so on. Please also note that your time factor comes into picture in the
constitutive relations as well. For example, plasticity basically deals with rate
independent that means the time is not a factor and viscoplasticity deals with time
dependent behaviour, when the material becomes plastic. We will see in a minute where
these are applied, but first thing I want to distinguish, I want you to distinguish is the fact
that material behaviour is one aspect and the deformations and the ensuing definitions of
stress and strain are another aspect. Is that clear? So they are two different aspects.

For example, I can have small deformation; I can have plasticity along with it. It is not
that when material becomes plastic, you need to all the time do only a finite or large
deformation. I can do plasticity with small deformation. Is that clear? So, they are two
different things. Yes, may be when I do non-linear elastic say, for example rubber, the
deformations are usually large. So, I may not be in a position to do small deformation, I
may have to do only large deformation; but nevertheless, they are independent. In fact, in
other words, what we mean to say is the stress measures that we use, stress measures in

11

the sense that the definition of stress, one of these guys here, the stress measures that I
use, are to a great extent independent of the type of material behaviour. Is that clear?

I am going to give special names to that nominal stress, when I go to multiaxial

situations. I would call it as say, first . of your .. stress. I am going to define them
very carefully later, but once I define them from this point of view, from deformation
point of view, I will use the same stress measure whether I am looking at a problem here
or a problem here. So, they are two different things, because one we would look at the
kinematics and the other is the material behaviour or the constitutive relationships
between stress and strain. That is the next thing. We will talk about that later. These are
the material behaviours and those are the kinematic relationships. Note that these two can
also be combined with an implicit or an explicit dynamic analysis as well. In other words,
I can do a dynamic analysis, an explicit analysis, plasticity, finite deformations.

(Refer slide Time: 21:02)

I can go, I can join them together; I can join them together. In fact, if you want to do
crash worthiness of a vehicle for example, you would do an explicit plastic analysis with
finite or large deformation. There, they can come together or I can do, if I want an
implicit plastic small deformation. I can, I can pluck one of them from here, one of them

12

from here, one of them from here and then combine them. The most important decision in
finite element analysis, especially if you are doing practical finite element modeling is to
pluck the proper leaf from here, from here and here. You have to know whether you are
going to do a static analysis or a dynamic analysis, you are going to do whether an elastic
or an inelastic case, whether you are going to do small deformation or large deformation
and so on. Then only, a picture would be complete.

For example, many of the metal forming problems, you are looking at metal forming
problems, many of the problems can be considered as a static problem; can be considered
as a static problem, which is basically inelastic which can be looked at as, in other words
as plasticity being important or the constitutive equations being governed by the plastic
type of equations or plasticity type of equations along with large deformations. So if I
want to do a metal forming problem, I will pluck static, I will pluck plastic, I will pluck
finite deformations. Is that clear? So, this is possible that we do.

Now, we can extend this further by putting lot more models. But before we do that, let us
realize or it is very important to realize that non-linearity also comes into picture through
what is called as contact; through contact.

(Refer Slide Time: 23:34)

13

I am sure all of you know that contact is very important say, for metal forming problems,
which we have discussed; most of the time the contact between the punch and the
material, material and the dye really gives that shape for the body. So, contact introduces
non-linearity through an inequality relationship. So, again contact is a separate entity. It is
not that contact is there only for contact with large deformation, with plasticity alone. We
can pluck another leaf from contact. In fact contact can be classified in a very broad scale
as say frictionless or with friction. We will go into more details later, just say for example
frictionless and friction depending upon what is important. Most of the time, we would
look at friction to be important.

For example, I am going to do analysis of a tyre. Then what would I do? I would pluck
depending upon what I want. I want to know what would be the pressure distribution or
foot print loading of the tyre depending upon whether I want to do that or I want to find
out how the tyre behaves during cornering or during breaking. Depending upon the
problem which I want to do, I will either pluck a static or dynamic load. Then, I will go
ahead and see what the type of behaviour is. I know that it is an elastic behaviour and I
know that it is non-linear elastic. So, I would pluck this or this, then this, this and I know
that rubber is subjected to very high strains. So, I would pluck this and I know that there
is a contact between the tyre and the rim and the tyre and the road; so, I would go and
pluck contact. So, the problem is complete when I model using a contact algorithm, when
I model the problem using finite deformation, when I model the problem using inelastic
material behaviour, model the problem using static or dynamic analysis. So, the route is
complete. Any question?

Yeah, so, if there are no questions, let us look at contact in slightly more detailed fashion.
Of course, as I told you contact introduces the non-linearity through what is called as
inequality constraints. We are going to have, you know, separate lesson on contact, but I
just want to classify again contact to be in a slightly more broader sense.

14

(Refer slide Time: 26:58)

The contact can be between a deforming body and what is considered to be rigid body.
Please note that any body can be modeled as a rigid body. That means that we make an
assumption that the deformations are very, very small, so small, that the body is
considered to be rigid for all practical purposes. The tyre example which I said and I said
that there is a contact between the rim and the tyre. Immediately that would prompt you
to tell me that the rim can be modeled as a rigid body when compared to the tyre which
would deform to very great extent. Note that this concept of which is deforming and
which is rigid requires an engineering, a good engineering sense and understanding of the
problem.

For example, I have a very, very thin sheet of steel, extremely thin sheet of steel, which is
deformed by a thick rubber block, thick rubber block. Then, which would be rigid and
which would be deformable? Say, we are talking about 1 mm thin sheet, which is
deformed by a rubber block like this or this duster. So, obviously that thin sheet of steel is
going to deform much more than this block, rubber block. So you can, for all practical
purposes consider that block to be rigid and that sheet to be deformable. Yes, that is a
good question. Can we consider both of them to be deformable? Yes. That is the next
thing or next classification that both of them are deformable. Yes? Yes, how do you

15

decide in other words, the question is how do you decide whether a body is rigid or
deformable?

As I told you, the first thing is that there is an engineering judgement that is involved. I
will come to that answer in a minute, but before that let me tell you what you should not
do. It is not that you compare say the stiffness, E values of these two and take a decision
based on E; which is deformable and which is, because most of you would hear that when
E is very large, the body is very rigid. True, absolutely true; but please note that, the
stiffness of the body depends not only on the material property called E or the Youngs
modulus, but also on the geometry. So, both of them go together to define the stiffness.

What is the type of engineering judgement we are talking about? Suppose, let us look at
two gears mating. You want to analyze the contact stresses of two gears that are mating,
whatever be the type of gears. Now, can you classify this under a deforming body and
rigid body under this category? Both are deforming, you cannot say gear 1 is not
deforming and gear 2 is deforming and so on. You have another set of contact conditions
called deformable-deformable contact or deformable body, two deformable bodies
contacting.

(Refer Slide Time: 30:55)

16

We usually call this as deformable and deformable contact; loosely defined like that or
you can say, in more correct sense contact between two deformable bodies. The third
category is what is called as self contact. Body can deform and come into contact with
itself. Suppose I have, say for example I have a very thin sheet and I apply load. It can
buckle and when you look at it in the side view, it may start touching. If I start applying
load like that, it may start touching or during impact, say you are testing crash worthiness
of a car; may be the hood of the car may crumble or may collapse and may start
contacting itself at different places. So, that is what we call as self contact. It is very
important to realize that there are many packages which do not support self contact for
static analysis, there are reasons to it. So these are the three categories. You have to be
very careful what you choose.

Why is that we are classifying it like this, because the ease of doing analysis depends also
on your proper choice of the type of I would say, modeling that you do. In other words,
all components in this world are 3D; all components in this world deform. If you look at
it from that philosophy, then you have to really go for a deformable-deformable contact,
all the time 3D analysis and so on. Then it would not be possible for us to solve very
complicated problems. So, we have to make some simplifying assumptions in such a
fashion that it does not affect the results. We have to be as close to reality as possible, at
the same time make good engineering model out of it or a mathematical model out of it.

So, again we can pluck say for example I am going to do my old problem, metal forming.
Then, I can do what is called as a deforming body and a rigid body to be the contact
algorithm or the choice of contact is now between rigid body. Which is going to be rigid?
The dye. For example, sheet metal I am doing; I can use this dye to be or the punch to be
the rigid body and that sheet, steel which is deforming is going to be my deforming body.
But please note one thing very carefully that, these are or these define type of analysis.
But, these are also linked to the elements. They are again independent of this kind of
analysis. You know the type of elements.

17

(Refer Slide Time: 34:35)

Probably you would have studied about or yes, we did that on plane stress or plane strain
or beam, axi-symmetric, shell, whatever it is. Again, one more leaf you have to pluck
from that list. So, I can do for example, a shell analysis; a shell analysis which is or I can
choose an element called shell. I can do an inelastic, in other words, a plastic analysis or
elastoplastic analysis for sheet metal forming, with finite deformation along with contact
for deforming and rigid body. So, for sheet metal forming I do that or membrane
elements. Many times we use membrane elements, where the stiffness in the bending
directions is low. So we choose membrane and then go through this whole thing or I can
choose 3D; 3D element from here, pick that up, then I can go through this whole route.

So, the type of elements, of course, mimics the structural behavior. Yes, this selection,
this selection depends upon or in other words how do I select an element? That selection
depends upon the type of structural behaviour. I know that the structure is going to
behave in a three dimensional fashion. I know that this is a very long bridge. So, I know
that the behaviour is going to be plane strain or I know that this is going to be an
extremely thin sheet subjected to in plane loading; no, it is going to be plane stress and so
on. So, that is different; the structural behaviour is different, the structural behaviour.

18

What are the things you have to consider? The structural behaviour, the material
behaviour, the type of deformation and whether you have contact or not. The structural
behaviour you can include also whether it is static or dynamic. So, when you have a
problem in hand, you look at all these things. When you have a problem in hand and
choose say one of the elements which is independent of the further choices that you are
going to make, when you choose say a static analysis, elastic analysis, linear elastic
analysis, small deformation with no contact, you are doing a simple linear elastic problem
and all the concepts that we did in the last course are valid for that.

But, look at that, choices. If you remove that and choose any other option from here, any
other option, you move away from this first category and choose any other option here
then, you cannot do a linear analysis. Then, you have to do a non-linear analysis. It is
unfortunate that there is a general tendency to believe that it is possible to reduce most of
the problems to linear case. Please understand that especially in mechanical engineering,
you do not use a component independent of others or in other words most of the
components are in assembly, they are in contact. So, contact is very important, because
many times we may not be able to judge how forces are transmitted from one component
to another component.

We are also interested to find out what would be the contact stress between the
components, because many times failure originates at these contact locations. All of you
know about this, say for example in gears. So, contact is a very important thing in
mechanical engineering. You may, you may be surprised to look at the results when you
remove contact and make some very simplifying assumptions. No doubt, I said that
simplifying assumptions have to be made, but again you have to know the consequence
of such an assumption. It is not that I am making a simplifying assumption, I will leave
this out and do the problem as a simple case, because I have, I have only this kind of
package or is easier to do and so on, may not fetch you good results. That is the first thing
that if it is a linear elastic case, you will be in that first ring of things. Is that clear?

19

This is only one part of the story that I can pluck here, do this and so on. But, we may not
be or it may not be enough or we may not be satisfied doing only a solid mechanics
problem. Many problems warrant what we call as coupled problem as well or coupled
analysis as well; many problems warrant a coupled analysis.

(Refer Slide Time: 40:20)

What is a coupled analysis? For example, let us look at a welding problem. Of course, we
are interested in welding problem. What is that we are interested in? Temperature
distribution and we are also interested in many places, stress or deformation, because the
deformations that result from this temperature distribution becomes very important; in
fact, may be much more important than just the temperature distribution. So, we do a
coupled problem where one field values affect another. On the other hand, look at say for
example machining, a machining problem. I want to analyze a machining problem. Now,
what is the situation here? In the first case, it was the heat or the temperature which had
an effect on the deformation, basically through co-efficient of expansion and so on. But,
in the second problem, the second example which I am telling you is a machining
problem and what is the problem here?

20

Yeah, deformation; agreed, large deformation. Yes; how do you get the heat? All of us
know that when you machine it, after you machine it you touch it, it is going to be very
hot. So, what has happened there? Heat produced due to so many effects; one of the
major effects is the conversion of plastic work into heat; conversion of plastic work into
heat. In fact, 90% of the plastic work that is done on the body is converted into heat. Is
that clear? So, here again there is a coupled problem. But, in both the cases the coupling
is one way. It is a one way coupling. I am interested in temperature, because I want to do
stresses. But, in the first problem my stress effects may not be very important from the
point of view of increasing the temperature. In other words, the temperature in that
welding problem is totally controlled by my weld rod, its movement, how much current I
am giving or in other words how much energy I am inputting and so on.

In the second problem, I am interested in the temperature, basically from the other end. I
have plastic work and the temperature is the result of this plastic work. But, I am not very
much worried whether the temperature would increase my stresses further and so on.
These are, I would say, I would classify them as loosely coupled problems, where one has
an effect on the other. But, there are problems in which you can say that one has an effect
on the other and vice versa. Then, the strategies have to be very, very different. This may
involve say, for example rolling tyres with viscoelasticity; it may involve temperature, as
well as the stresses and so on.

The next category of problems are coupled analysis problems, where again non-linearity
may play a great role. It is not that non-linearity is present only in the deformation, but
non-linearity may also be present in temperature. How? In the thermal problem. How can
non-linearity be present in the thermal problem? Yeah, so the properties or in other words
the thermal properties may also be a function of temperature; can also be a function of
temperature. This is again, this may again come under region of non-linear analysis. Of
course, one of the major, one of the most important problems today which has impact on
so many fields is what is called as fluid structure interaction problems. Here we saw the
effect of thermal loads acting on the body and giving rise to stresses. On the other hand

21

we can have a situation, where fluid flow has an effect on the body and would result in
increase in stresses.

(Refer Slide Time: 45:35)

In other words, fluid structure interaction, I will write that here, fluid structure interaction
also becomes a problem. When we develop finite element analysis, we have to take into
account so many things. But, this list does not stop here. Apart from all these things,
certain conditions may have to be met.

22

(Refer Slide Time: 46:32)

These conditions are broadly classified as constraint conditions or constraints. What are
they? The body because of its material behaviour may not only be, say for example non-
linear elastic, but also may become incompressible; may become incompressible, which
means that there will not be any volume change. Not that every body, every non-linear
elastic body is incompressible, but in many situations you may have to deal with certain
what we call as internal constraints; constraints with your material behaviour that the
body becomes incompressible. In the plastic case also the body becomes incompressible.
In other words, plastic part of the deformation results in incompressibility. All of you
would have known that the Poissons ratio approaches 0.5 when body goes into the
plastic region and the result is due to the definition or due to the constraint called
incompressibility. Is that clear?

Now, the challenge is how are we going to solve set of these problems? How are we
going to develop algorithms to solve them? What is the strategy which normally is
adopted in non-linear analysis? The first issue; second issue, of course, is how am I going
to define stress, strain, how am I going to characterize deformation? How am I going to
say that it is a finite deformation, large deformation or finite or large deformation, small
deformation, how am I going to characterize them? That is the next major issue. So, these

23

challenges are the ones which make this whole study of non-linear finite element analysis
very interesting. As I told you, in this course we are not going to take independent cases
and develop. We may do that for certain cases, but we are going to, you are going to see
from next class onwards that we are going to develop some general strategies, this plug
and play we are going to do that even for algorithms. How we are going to do that we
will see it in the next class.

24