Advanced Finite Element Analysis Prof. R. Krishnakumar Department of Mechanical Engineering Indian Institute of Technology, Madras Lecture - 11

Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

Advanced Finite Element Analysis

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

Lecture - 11

In the last class we had seen, looked at the mixed formulation.

(Refer Slide Time: 1:01)


We will just quickly summarise what all we did in the last class, so that we can follow
the derivations further. We started with the series of definitions and for example we
started with what is M, then what is I
d
and then we said that epsilon can be split into
two parts - the volumetric part and the deviatoric part; same way sigma can also be
split. But, one of the things which we emphasized is the fact that we are now going to
look at not the irreducible form; that means displacement alone, but we said we will
look at both - displacement, pressure and epsilon
v
as three of the variables that we
will be interested in and hence we are stating sigma, I mean this is the common thing
what you know. The only thing is that we defined delta sigma with that inverted hat to
be one, which you will get from delta epsilon through the constitutive equation. So,
this is the deviatoric part that is the pressure part, so that that together will define us
the sigma.
1

In other words, this is the fundamental definitions which all of you know and then we
wrote down the variational form.

(Refer Slide Time: 2:16)


The variational form is very straight forward and the first variational form is exactly
the same as that of the virtual work form which you had encountered in the previous
course. There is no difference between them, between this and what you had done
before and the only thing is that, to that we add two more. If you really look at it
closely what essentially we are doing is that what happens or what should be zero, we
are multiplying it by weight function delta epsilon
v
and delta p. These three are quite
easy to understand and then we combine these three things. In other words, we have
three equations; note that we have now three equations.









2

(Refer Slide Time: 3:02)


We can combine them together to write Hu-Washizu principle. In fact, you can come
the other way as well. You can define the Hu-Washizu function and from here,
recognising that we have displacement, p and sigma or rather sigma and epsilon
v
, you
can rewrite that in terms of p and epsilon
v
as well and then come down from Hu-
Washizu principle in the same fashion, as you do for the irreducible form, for
displacement form. Of course, this is written in terms of u, sigma and epsilon. You
can re write this in terms of p and epsilon
v
as well and so from here you can come to
this place or from here you can go to the other side, in a very similar fashion as we
had done before.

We will not worry about the Hu-Washizu principle right now, but what we are
interested in is, here it is much easier to understand from the formulation, from this
place. So we will start from here, we will come to Hu-Washizu principle more in an
elaborate fashion, when we talk about the large deformation case.






3

(Refer Slide Time: 4:11)


In order to convert this into a finite element formulation, what we need is things
which are again familiar to us and these are the things that we need and in other
words, what we have done is to now define the continuum displacement in terms of
the displacements of the nodes and look at these two quantities. Now, we have
pressures as well defined in a very analogous fashion to that of displacement and then
we have what are called as the shape functions for the pressure and the volumetric
term as well; that is N
p
and N
v
. Both of them defined these pressures, which we have
substituted or which can be useful for us to substitute it back into the variational form.

Having said that, having written these things I can, do write epsilon and delta epsilon
and sigma in terms which are familiar to us, now in terms of shape functions and the
variations, now variations defined with respect to the discretized form and our job is
quite simple. Look at the logic. It is very, very simple. Having defined this, we will go
and plug these equations back into my variational forms and then write down a series
or set of equations which I will use in order to solve for u, p and epsilon
v
. It may not
be necessary for us to solve all these things together and it may be at most instances
possible to eliminate say, p and epsilon
v
at an element level. We will see whether it is
possible to do that.

4

So, as the first step, what I do is to substitute all the discretized form, written in terms
of shape functions, into this mixed form and write down the corresponding equations
and let us see how they look like. I think we stopped at that point in the class and if
there are no questions, we will progress further. Let us see what happens to the first
equation. I told you that you can yourself substitute it; you can see what happens.

(Refer Slide Time: 6:09)


First equation can be written say, in terms of M; M u double dot is equal to f. Look at
that equation and look at this. It is very obvious that the first term which is the inertia
term goes as the M term. It is again a very familiar matrix, M matrix, which is the
mass matrix that you would have called in your earlier classes. So, the mass matrix
you substitute that. You can see that delta u transpose you can substitute in terms of
N. So, this is again N, so N transpose, row N transpose N is what the mass matrix is;
so you can define that N d omega.

You can, if you want for further analysis you can, either remove it or you can keep it
as it is. If you are going to do a static problem, then you can remove that particular
term. If you are going to do a dynamic problem, then you can keep that straight away
there. So, next term is the P term, which is the internal force term. Remember that the
internal force term comes from here, the second term here. Remember that S u is
replaced in terms of b u, so you will, b u hat; you will get actually delta u transpose
5

hat B transpose sigma. So, B transpose sigma is what was coming into the picture as
the internal forces. Remember that that is what we used. So, that is the second term
that I have. So p is equal to integral B transpose sigma d omega. So, that talks about
our first equation.

Now, let us look at the second equation. Let me write this down, the second equation.

(Refer Slide Time: 8:05)


What is the second equation? How do I get it? I substitute the discretized form into
my second equation. Look at these two terms; they are very, very easy to understand
again. So, you substitute for delta epsilon
v
and p from the previous case and let me
write this down as P
p
minus C say, p hat is equal to zero. Can we repeat M? Yes; M is
what? M, I had already defined 1 1 1 0 0 0. I mean just I had, why I had M? M I had
put here, basically because it is easy to write it down.

Yeah, mass; this capital M, what is capital M? Capital M is nothing but the mass
matrix. If I write it down as it is, it is consistent mass matrix, very familiar term in the
structural dynamics. If you are, as I told you if you are, not interested you can remove
that and you will get back your original equation. So, let us see what or how you get
term? term comes from this first term and what is that term will be? So, delta
6

epsilon
v
, you are going to substitute from there; obviously, from this term here and so
substitute that and see how we can write down that term.

(Refer Slide Time: 9:40)


Let us see, what you get? , substitute it here; so, let us see what you get? Simple;
integral omega delta epsilon
v
. If I want to write, this is a number; but, if you want to
write it in a matrix notation, you can write it as transpose into this in a matrix
notation, because it is easier to follow matrix notation. So, you can convert this and
write it, write this down in matrix notation. You can write it down as delta epsilon
v

transpose, the other terms there. So, you can say that 1 by 3, I will keep that out; this 1
by 3 here I will keep that out. So, delta epsilon
v
I am going to substitute in terms of
N
v
. So, N
v
transpose is my first term there. Then of course, I have my M transpose
term; M transpose term, then sigma term there, d omega is my first term.








7

(Refer Slide Time: 10:51)


Then, my second term comes from this term here, again delta epsilon
v
transpose. My
second term is what I call as C. So, it is obvious that now, it is very simple; now I
substitute this in terms of N
p
p hat and this again, the same way delta epsilon
v
hat N
v

transpose. So, together I get N
v
transpose N
p
d omega, p being taken outside.

(Refer Slide Time: 11:24)


So, C is equal to omega into N
v
transpose N
p
d omega. I mean this is what we did
also in last class. I hope things are clear.
8

One of points I want to state is that once you, it is so simple; finite element
formulation is extremely simple. Once you have this kind of formulation, look at what
we are doing? Very, very mechanical; I defined discretized form. Till now, till this
point, it is not finite element per say. Hu-Washizu principle comes into variational
forms; you know, they are not finite element. I enter finite element once I write these
things. So, once I know the physics, once I know the formulation, I just, what I have
to do is only take that out, substitute it into my variational form. In fact, if you want
you can branch off from here, you can do some other technique; you can solve using
some other technique as well. But we are, once we are into finite element we just
substitute it into that expression. Is that clear? So, we go to, yes, any question?

Delta epsilon
v
transpose, yes. Yes, let me write that down clearly. If you have any
doubts we will, so, you write this down. So, this is the procedure. The doubt is how
do I get this? Still not very clear; it is very simple.

(Refer Slide Time: 13:04)


You see that I can write this down into delta v transpose N
v
transpose into one third
of m sigma minus p, instead of p I am substituting it N
p
p hat, d omega is equal to
zero. Delta epsilon
v
transpose is that which belongs to the nodes, because actually I
should put B ., what we got from discretization. So, this is similar to delta u hat
transpose that you would have had when we defined B transpose db as K. Take that
9

out; take that out, so, I can write this down as delta epsilon
v
transpose omega,
integrating throughout omega, N
v
transpose, so, one third m N
v
transpose, sorry, N
v

transpose m, N
v
transpose m sigma d omega. That is the, that is my first term, because
this delta v hat is the same for both the terms, minus the second term N
v
transpose N
p

into p hat. So, the first term is what I go or I call this as because, this being common,
this has to be satisfied at every point; so, this being common, we can take that out. So,
minus C
p
hat and C is what is the second term N
v
transpose N
p
p hat. So, that is what
you get here in this term.

(Refer Slide Time: 14:59)


This is, as I said, I deliberately did that because that is the standard way of writing
that equation. Now, let us come back to the third equation.









10

(Refer Slide Time: 15:22)


Third equation is delta p; I think I will just remove that and write that out again;
integral delta p, into what is that we wrote? m transpose S u; m transpose S u minus
epsilon
v
. Strictly speaking, they should be the same and they should go to zero. So,
you can view it, in a slightly crude fashion we can say that this is zero; zero multiplied
by some quantity is equal to zero. Now, you can look at it, in order to understand it,
you can look at it in that fashion as well.

There are, we are going to get three equations I know the first equation; that is the
first equation and I know the second equation and let us now develop the third
equation. What is the third equation? Again, the procedure is exactly the same.
Substitute for these quantities from the discretized quantity and what do you get? Let
us see what you will get as the first term.








11

(Refer Slide Time: 16:41)


Delta p, again same way delta p transpose I have to put, so, let me call this as, you
will see that I will get C transpose out of it; C transpose or minus in fact, epsilon, for
the second term epsilon hat v minus E u hat is equal to zero. This is the second term
that you will get here from this term; it is very simple. Delta p, now, I am going to
replace it with delta p hat. This will be replaced by transpose, so delta p hat transpose
N
p
transpose. If you look at the second term, let us look at the second term first. I will
get N
p
transpose N
v
N
v
into epsilon
v
hat. So, N
p
transpose N
v
d omega is nothing but
C transpose and so I get the first term to be minus C transpose epsilon
v
hat.

Look at the second term; the second term is here. Substitute in terms of S u, in terms
of B, so that E will be, have a look at this term now; have a look at this term here.
Then, you can write down E to be N
p
transpose, that is the first term that comes here,
N
p
transpose. That N
p
transpose comes here; then m transpose that is the m transpose
comes here. Then, B; S u is written in terms of B u hat. So, B d omega is what you
will get as the three things. So, you get that. That is the third equation and that is the
definition of E. Let me complete the picture by writing the second equation as well
here which we just removed. We will write the second equation minus C p hat is
equal to zero. So, that is the first equation, the second equation, this is the third
equation. Remember what we wrote for C? C is N transpose, N
v
transpose N
p
d
12

omega is what we wrote. These are the three equations that we have and we have to
solve these three equations.

Now, look at them; let us see how we can solve them, solve these three equations,
because it is, by the way, what is f? How do I get f? External forces and that comes
from this, obviously. I mean, though it is standard, we know that delta u transpose I
substitute; so, N transpose B, then N transpose t d omega. Now, you can solve these
equations straight away. For example, look at that equation. It is so nice; it is very
simple to solve them.

I can write p hat, p hat to be, I can bring this guy to the other side. So, I can write p
hat in terms of C inverse; I can write p hat and so, the first thing I can do is to solve
for, straight away from these two equations without using the other guy can solve for
this equation. Let us see how I am going to do that?

(Refer Slide Time: 20:24)


So, p hat, what is p hat from here? So, C inverse P
p
. I mean of course, they are matrix;
just for clarity, I have written them and epsilon
v
hat can be obtained from the second
equation; from the second equation. Let us see, but there is going to be a small issue
there; that is equal to C inverse E u hat; C inverse E, so, C inverse transpose E u hat
which can be slightly re written as say W u hat, where W is C inverse transpose E. Of
13

course, when N
v
is equal to N
p
, which is the case most of the times we take, then C
becomes C transpose. So, this C inverse transpose can be written as C inverse. Is that
clear? Yeah; no, there is a minus here. No, this should be plus here, first term;
because, the first term was a plus, so we get, that i a plus there; where W is that
quantity. Is it clear?

So, now substituting for epsilon, let us now do a small jugglery; substituting for
epsilon, I can write epsilon to be I
d
from the, look at this expression here. I can write
epsilon to be I
d
into B u hat. Look at what I am going to do. It is very interesting, so,
plus one third m N
v
epsilon
v
hat. Now, for epsilon
v
hat, I am going to substitute in
terms of u hat, W u hat. So, I can write this down as W into, sorry, u hat. Look at this
expression now. Does it strike anything, does it?

Look at that expression; this expression. What we have done is epsilon
v
we have
written in terms of u hat, then re substituted back into epsilon. Look at this expression
here. Absolutely; so, there is a relationship. Looks like this relationship is very similar
to my old relationship of B. The B is now, looks as if the B is now modified. The B
looks like, B has two terms in it; B has two terms in it - one term, which is the
dilatational term and the other is the deviatoric term. B has split this; I mean it looks
as if now B is split.

(Refer Slide Time: 24:17)

14

Let me define N
v
W as say, B
v
, B
v
and write that down in a matrix form, so that
epsilon is now written as I
d
one third m B and B
v
. Yes, B B
v
; yes of course, u hat is
common to this. So, I can write u hat ultimately. You can immediately foresee what I
am going to get. Note that this is very similar to my splitting B into B deviatoric B
dilatational in our B bar method, but we came through an entirely different route, a
variational route; much more meaningful mathematically, but still arrive at the same
result.

Let us now proceed and see what would be my tangent stiffness matrix. You can
foresee what we are going to get, the procedure is again standard; there is no
difference in the procedure. Let us see what we are going to do. What we are going to
do is to take my first equation. As I said, let me remove this from further
considerations, we have already considered these two guys here. If you want to add,
of course, the second term that is m u double dot term, you can keep adding it; it is
not going to change any of the things that I am going to do, you can keep adding it.
But, only thing I am going to do is to now take this B transpose sigma term and then
see what we can do about it. That is equal to, remember what we wrote for sigma, I
d

sigma hat; look at this term here.

(Refer Slide Time: 26:41)


15

I am just keeping that there; please look at the term there and then, substitute that term
here instead of this sigma, so that I will get B transpose, can you read that out?

(Refer Slide Time: 26:48)


I
d
sigma plus m N
p
into p hat. But what is p hat? We just now derived what is p hat. C
inverse ,so, I can write that down as C inverse .I hope it is clear that the two terms I
d

sigma this inverted hat m N
p
C inverse d omega is what is my term. Now, what I am
going to do is to take this and then do a small jugglery; nothing else. If you want to do
it, you can do it or else you can do it in any other fashion you want. But just to make
things easier to implement, we are going to go through this step.

I hope you are not lost out in what we are doing. The two equations we have taken.
Essentially I am able to, from the second equation I am able to get p. Third equation, I
am able to eliminate epsilon
v
in favour of u hat; re substitute it back. So, all my
equations, ultimately when I solve it, all my equations are in terms of nodal
displacement only. But, only thing is that the K
T
matrix what I am going to get is not
my original B transpose dt B, but that is going to change. That is all, nothing else. So,
the second term, the two terms to this, let me write that down.



16

(Refer Slide Time: 28:34)


Omega B transpose I
d
sigma hat plus B transpose m N
p
C inverse d omega; so that
would be the two terms. Now this term, the second term I am going to re write it a bit
differently, because my, remember what my is; has an integral term in it. So, I am
going to write this down in a slightly different fashion and I am going to write that as
N
p
d omega C inverse and then substitute for W from my previous expression and
then write this down as B transpose I
d
. We will verify that what I am writing, straight
forward substitution. I do not have all the equations here, but you can verify that what
you have; plus, yeah, integral I think one third will come into picture because of
definition of W and so on.

So, W transpose N
v
transpose m transpose N, yeah, N transpose, not P, but sigma hat
d omega. I am substituting back all the equations which I know, which I d and so, I
am going to write that down something like that.







17

(Refer Slide Time: 30:45)


Now, again I am doing some juggleries here. Re substituting it, all expressions, so
that I will write this down as B transpose I
d
plus one third B
v
transpose m sigma hat
or m transpose rather m transpose sigma hat d omega is what I will have. So, this is
what we actually arrived at as B transpose; so, let me call this as my new B and so
that, I will call this as say B bar, a new B which is B bar.

Now the next step; this is p. The next step is my calculation of K
T
; my calculation of
K
T
would lead to my introducing D
T
here; omega B bar transpose, that is, this is
rather B bar transpose, not B bar. So, d sigma hat by d epsilon into d epsilon by du hat
which from here would again give rise to another B or B bar, so that I can write this
down as B bar transpose, sorry, D
T
; D
T
is T sigma hat by d epsilon which is or delta
epsilon delta sigma hat by delta epsilon, that is what we introduced right in the
beginning, into B bar into d omega.

So, this is, the step is standard. Again, p is what? We remember that this is what we
did for the Newton-Raphson scheme. If you remember, we had B transpose sigma
instead of this and we defined this as B transpose d sigma by d epsilon into d epsilon
by d u. Remember that we had this as D
T
and remember that we had here B. So, B
transpose D
T
B is what we defined. In this case, it so happens that that B is not there,
we have B bar. B bar is the very thing that we had looked at; modified B in our earlier
18

approach. So, that is the K
T
now you are going to use to solve the equations, the
Newton-Raphson scheme.

Sometimes what people do is to combine these terms here into D
T
and define a new
D
T
hat. There is nothing very different from what we are doing; just D
T
is slightly
rearranged. So, here you have, in this formulation, please note that there was a
question on cross terms. In this formulation, you have all the cross terms intact. It is a
very rigorous formulation. In this formulation, you will have all the cross terms intact.
This whole thing is here and then this would be written like this. So, you will actually
have four terms, all the cross terms, which will go into the implementation of the
finite element analysis.

Now, coming back to where we started from, we started with elastoplastic analysis.
What we have essentially done is to write down a small strain formulation for an
incompressible or merely incompressible material. That is, we have diversified or we
digressed from our elastoplastic analysis. We wrote down a small strain formulation, a
mixed formulation for the analysis of nearly incompressible material and what I want
to state is that, now if you go and substitute at this D
T
your elastoplastic D
T
, then this
formulation can be directly used for an elastoplastic analysis. If you are not interested
in elastoplastic, but if you are interested in any other material, still in the realm of
small deformation and if you are interested that means that mu is equal to say 0.49,
495 and so on, but the deformations are not very large, then you can use the same
formulation; no difference, same formulation, but a corresponding D
T
can be used.

This is what I wanted to tell you that most of these algorithms are plug and play
algorithms. It is not that I will repeat this, if I have to do non-linear elasticity; the
procedure is the same. So, for solving a small deformation problem, this is the
procedure. That is all, but I will add one more to this; one more to this. What is it that
I will add? I will add stress update algorithm, if I have to solve plasticity problem.
Now, the comment about these elements is that we should understand that most of the
non-linear analysis is done with, not with higher order elements. It is not, it is not
usually recommended to do due to various reasons, theoretical reasons. Right now, we
19

are not in a position to see that, but usually most often we use only lower order
elements.

(Refer Slide Time: 36:51)


Say, for example one of the elements which you may use here is a say, quadrilateral
element in which case you will define N in the same fashion as you had done for your
earlier classes, but you would define N
p
, since Np .. the dow Np. Look at this B.
What enters here is only the gradient that is dow N
u
or the first differential of N is
what appears there, but the first differential of p what we had defined, there is no
corresponding p for N
p
; there is no B
p
term. That o enter that place, so, we do not
have problem of defining p in the same fashion as that of N. So, usual method, usually
what is done is that you will use the .. and also note what is B
v
term? B
v
term is
what we define here. This is the B
v
term. It does not come from m
v
term, it comes
from this B
v
term.

So, it is usual practice to define the u in this kind of element to be that at the nodes
and define p and epsilon
v
to be a constant inside an element. In other words, it is
usual practice to distinguish between the degrees of freedom u and the other two
degrees of freedom and you please note that it is not all the time that you treat u, p and
epsilon
v
to be as if they are degrees of freedom at the node. There are certain places
where you will put degree of freedom at the node as well, p as a degree of freedom,
20

but it is not the usual practice to define at the node, degree of freedom u, p epsilon
v
. It
is a very important point, because that is where there is usually confusion. It is not
that here I have u, p and epsilon
v
; no, that is not what we are talking about. p and
epsilon v are defined such that say, for example for this element, N
p
and N
v
are
constants or unity. That is as if it is defined for this element, so, it is discontinuous
across the next element and so on. Same way, you can extend this concept to a solid
element as well. Is that clear?

We will may be later in the course, we will define or rederive equations for
tetrahedron elements. But, what I want to again point out is that, this is a general
formulation and can be say, can be used or information can be extracted from this to
derive formulations or derive K
T
for plane stress or plane strain or axisymmetric and
so on; 3D solid and so on. So, the corresponding terms are now replaced when we go
to say, plane strain and axisymmetric. Of course plane stress, there are certain issues;
I may not have time to cover every element formulation; every element how it looks
like it may not possible for us to cover in this course. The whole idea is to give you a
background. If I recommend Zinkevichs volume II, if you look at say, Zinkevichs
volume II, you will see how this particular element or this procedure is now used for
deriving plane stress element. There are some changes; there are going to be some
changes, because of the way in which we operate for plane stress, but the procedure is
the same; the concept behind it is the same. We are not going to specialise everything,
because we have lot more to cover; so that we are not going to do. But as an exercise,
I would like you to have a look at the plane stress formulation in Zinkevichs book.

Having done that, we will get back to our elastoplastic analysis and close that part.
Before we go to the continuum mechanics aspects, let us now look at what the things
are. We will motivate the cases now and then write down the equations and other
things in the next class. Is it clear, this formulation is clear?





21

(Refer Slide Time: 41:22)


Now, we will now look at the stress update algorithm. What we are going to do is to
follow a very similar procedure as we did for one dimensional case. Stress update
algorithms are broadly classified into what are called as explicit algorithms, people
call this as forward Euler method; sub incrementation, explicit with sub
incrementation, we will define what sub incrementation is as we go along and then
implicit algorithms. All of you by now know that all explicit schemes have
conditional stability or in other words you cannot have a large delta epsilon, then the
algorithm is not stable. We know that intuitively, because we know explicit finite
element that is used in structural dynamics has some conditions attached to it called
Courant condition. So, same way here also you have some problems with explicit
finite element or explicit rather scheme. So, you may wonder why not we go to
implicit scheme. There are very interesting things that would happen if the type of
constitutive equations are different from that of Mises constitutive equations. So it is
important that we understand explicit schemes - explicit with sub integration or sub
incrementation, rather which is used extensively and we will also look at in a broad
sense what this implicit scheme is as well.

Now, let us define what this is. I have, I have already done, lot of times I have stated
what explicit is and what is implicit in this case.

22

(Refer Slide Time: 43:20)


As we go along we will see that as well. That is the say, a part of the yield surface; my
whole idea is that we are say, at this point. Remember where we are. We are in the,
that is the big loop, the incremental loop we are solving. We are inside the iteration
loop. Now I know how to form tangent stiffness matrix; now I know how to solve for
u, I know or I have to learn only how to do stress update. So, we are right here, stress
update; we are going out and looking at the issues that are involved in stress update.
So, at the end of this, you will know all aspects of small deformation elastoplastic
analysis.

We will complete the whole show there and on the way you also learn how to do
small deformation elastic analysis also. Though it has not much meaning to it,
puritans will not agree with you when you say I am going to do small deformation
elastic analysis with the or in other words, if I say that I will do using Hookes law, I
will make nu is equal to 0.49 and do certain analysis, people may not agree with you.
But, for many practical problems, this will be a good approach where nu is equal to
0.45 or 0.49 rather, you can use this formulation.

So, let us quickly look at what we mean by the stress update algorithm. The issues are
very similar to what we did in the previous class.

23

(Refer Slide Time: 45:04)


So, what I have to do is to now take this point, this point, the stress point which is say
inside the yield curve or the yield surface, which means that we are at the elastic
region, to the plastic region. So, I may have to go from this point to this point and as I
move from these two points, what I also have to do is to update the yield surface. So,
two things I am doing; one is to update my stress as well as update the yield surface.
So, what is the issue? As I update the stress and as I update the yield surface, it should
be in such a fashion that ultimately I will lie on the updated yield surface; that is the
issue. That is the only thing that poses a lot of problems.

I will state in a nut shell what we are going to do and then expand this in the next
class. I am going to follow our elastic predictor plastic corrector approach. So, from
here I am going to predict in an elastic fashion. I am going to go out of the yield
surface and reach a point which I will call as sigma elastic.

Obviously, you are going to pounce on me .you are going to say that it is not going
to lie outside it; yes, I know that. Then, I will correct it, I will come back and lie on, I
will come back to a point which is corrected point, which I will call as sigma T. As I
correct it and come back, I will also update the yield surface by calculating epsilon
p

or delta epsilon
p
. From here my yield surface will go out. So, that will be my new
yield surface. When I do that explicitly, I will find that obviously, I may not lie on the
24

yield surface. Then, I will correct this trial sigma by pulling it back in a simple
fashion to the yield surface, the updated yield surface and I will get what is called as
the sigma corrected.

Now in order to do this, I realize that I need one important factor. What is it? That R,
which I did in the one dimensional case; how did I calculate R? I know where I the
yield point. In the same fashion, I have to now find out as I move from this point to
this point where I will contact the yield surface. That point I need to know. So, that
point is called as the sigma say, contact; that point where I contact the yield surface is
called sigma
C
. Note again the concept is very similar. So, this is what I need not
correct and this is the part that is the part which I have to correct. How I am going to
do that and how I am going to ultimately calculate sigma CR by means of an explicit
scheme is what will be the focus of next lecture. So, we will stop here. Is there any
question? We will stop here and we will continue in the next class.



25

You might also like