Introduction to Finite Element Method

Dr. R. Krishnakumar
Department of Mechanical Engineering
Indian Institute of Technology, Madras
Lecture - 20
Before we go further, let us summarize what all we did in the last class about
isoparametric formulation. Remember that we talked about isoparametric formulation
as a very general methodology for a variety of elements; the elements may be plane
stress, plane strain, solid elements and so on. What we are going to do is to quickly
summarize all the steps that were involved in the last class and then go ahead to
evaluate the stiffness matrix as well as the right hand side that is the load. If you
remember that about two or three classes or even before, we had derived the stiffness
matrix terms. We said that the stiffness matrix can be written in terms of B transpose
DB dv.

(Refer Slide Time: 1:55)

We started, in the last class actually we started, there and we introduced the concept
of the natural coordinates and how mapping could be done. Now we transform this
equation into an equation of this form, where psi and eta if you remember, we
introduced it as natural coordinates. Please note that this J we are talking about, is a
Jacobian about which we will see in a minute and that this symbol indicates that it is a

determinant of Jacobian. There should not be any confusion. It is not a matrix and it is
just a symbol for determinant of the Jacobian. We wrote down Jacobian as dow x by
dow psi and so on and then we went ahead to derive the strain matrix. Have a look at
the strain matrix. The strain matrix has u comma x u comma y v comma x and v
comma y. In order to transform this into a strain matrix, we introduced a 1000 0001
0110 and so on

We defined the Jacobian such that u comma x and u comma y can be related to u
comma psi and u comma eta through a Jacobian. Just have a look at that and it will be
clear that we introduced the Jacobian to, I think there is a small mistake.

(Refer Slide Time: 3:50)

I will just get back here dow x by dow psi dow y by dow psi and sorry about that; I
think dow x dow eta and dow y by dow eta. From this, we derive that u comma x and
u comma y could be related to u comma psi and u comma eta through an inverse of
Jacobian. We did that in the last class and called that as gamma s.

(Refer Slide Time: 4:07)

Ultimately, when we stopped in the last class, we said that u comma x u comma y v
comma x and v comma y could be related to u comma psi u comma eta v comma psi
and v comma eta through a gamma matrix. So, we call this as a gamma matrix. What
is the whole idea of doing this? The whole idea of doing this is to replace the
dependent of x in the B matrix to the dependent of psi and eta. In other words B, a
function of x and y should be replaced in terms of psi and eta. That is what we are
trying to do. We have come to a stage now, where I can write out u comma x u
comma y v comma x and v comma y in terms of my psi and eta. What is that I am
going to do here? At the next stage, I am going to plug that into this particular matrix.
You can now see this. So, u comma x, this matrix will be replaced by this gamma
matrix into u comma psi u comma eta matrix; this matrix, this vector. How do I
replace this?

(Refer Slide Time: 5:18)

This is now replaced in terms of the Ns, because remember u is equal to sigma Ni ui
and so on. u comma psi u comma eta and so on will be replaced by means of Ns and
Ds.

(Refer Slide Time: 5:35)

This goes here, so, you will see that let us call this matrix as DN, part of it we called it
as DN, because we are completing it let us call that as DN. Then you would see that
this is gamma; gamma into DN into D is equal to this vector, which goes here. That
complete thing multiplied by this 1s and zeroes would now give me B.
4

(Refer Slide Time: 6:05)

In other words, epsilon is equal to the 1s and zeroes matrix multiplied by this gamma
matrix, what we just now saw, multiplied by this DN matrix multiplied by d. Those are
the three matrices that we would have in order to define d. We know that epsilon is
expressed as B into D. So, obviously B now becomes, all these terms put together, all
these terms put together now becomes B. Please note the difference between the DN,
which I have defined here and the same DN, name which I have given in the previous
class.

(Refer Slide Time: 7:05)

May be if you want to avoid confusion you can call this as DN total, because the
difference is that it has four rows and previous one had two rows. May be, if you are
bit confused about it, you can give another name to it, but any way DN is nothing but
dow Ns by dow psi and dow eta; that is the matrix.

Having done this, what is the next step? Just substitute this B into my first equation
here, first equation here, B transpose DB. What is D? D is nothing but the stress strain
matrix; you know about that already. For plane stress, we know what it is? So, we
substitute this B into this equation. We know all the terms here.

(Refer Slide Time: 7:52)

In other words, we can write down that k as phi say psi comma eta t J d psi sorry d
eta. We can write down k like that. Our next step, our next step is to evaluate this
particular integral. This is a very happy situation, because the integral was surface
integral, volume integral which was bit difficult for you to derive, especially when the
shape is quite bad. What is that we have done? We have replaced that by means of psi
eta coordinate system and the coordinate varies now only from minus 1 to plus 1.
What is the advantage now? The advantage is that we can resort to what is called as
numerical integration. We are going to see what numerical integration means. I hope
things are clear now. We are now in a position to calculate stiffness matrix.

(Refer Slide Time: 9:25)

Please note again, a reminder that whatever be the shape, whatever be the
quadrilateral, we had just derived it for quadrilateral, whatever be the quadrilateral,
even whether the quadrilateral is regular or whether it is sorry, irregular, whatever be
the shape, the procedure is the same, because we map all of them into a lovely square
in the psi eta coordinate system. The procedure becomes quite well established and it
is very easy now to write down computer program. Let us look at quickly summarize
what we are going to do with respect to this numerical integration. What is numerical
integration?

(Refer Slide Time: 10:12)

Let us look at an equation of the form, say integral minus 1 to plus 1 phi d psi. We can
evaluate this by different schemes, numerical integration schemes. What are these
schemes? These schemes are called as Gauss quadrature scheme and then Simpson's
rule, for example is one of the numerical integration schemes and so on. Usually in
finite element analysis, it is the Gauss quadrature rule that is extensively followed.
There are other rules like Newton-Cotes and so on. Thus Gauss quadrature rule is
what is usually followed.

What does this rule say? The rule states that if you evaluate phi, what is phi? Phi is
now a function of psi. If you now evaluate phi, which is a function of psi at certain
designated positions and multiply it by weight and sum them up and sum them up
then that result whatever you have, would be very accurate depending upon how
many positions of this designated places you include. Let us look at the simplest one
dimensional case. Let us say that we are trying to evaluate that kind of curve. Let us
say that we trying to integrate this between minus 1 to plus 1.

In a one point quadrature rule, the position at which we are going to evaluate this is
given by that position where psi is equal to zero. Obviously, this is our I and that is
the psi; sorry that is the phi and that is the psi. The designated position here is psi is
equal to zero and what is it that you get here and yeah, that is nothing but phi at psi is
equal to zero. Let us call this as say phi is equal to zero. If I now I ask what is the
integral value of this curve by evaluating that function phi at a position called psi is
equal to zero, your immediate answer would be that look that is only the area under
the curve and so you can write this down as 2 phi zero. In other words what does it
mean? It means that this two is the, what is called as the weight function. This is one
point quadrature rule. You can have more than one points as the quadrature rule and
let us see what the other rules say.

Now, let us look at the two-point quadrature rule.

(Refer Slide Time: 13:36)

Two-point quadrature rule says that you evaluate the same function at two positions,
two positions; no more at this centre, no more at this position, I said two positions.
These two positions are given by plus or minus 1 by root 3. So, you evaluate this
function at two positions. What do you mean by evaluate the function? Phi is a
function of psi. What is that you do? You substitute the value of psi to be plus 1 by
root 3 and then again minus 1 by root 3. So, you will get at two positions, the value of
phi. Let us call the positions as say phi1 and phi2.
The numerical integration rules states that this integral can be written as W1phi1 plus
W2phi2, where W1 and W2 are the weights and they happen to be 1 in this particular
problem or in this particular rule. So, this is called as two point quadrature rule. You
have also three point quadrature rules.

(Refer Slide Time: 15:13)

The three points are spread at three different locations and again these locations are
given in terms of psi and eta. The locations happen to be zero, plus and minus root of
0.6 plus and minus of root of 0.6.

(Refer Slide Time: 15:42)

In other words, the three point integration rule states that, one point here, one point
here and one point here. These are the three points at which the value of phi is
evaluated and the value of phi at psi is equal to zero plus root of 0.6 minus root of 0.6.
What are the weights here? The weight happens to be 5 by 9 for this position, 8 by 9
10

this position and of course 5 by 9 for this position. In other words, now if I use three
point quadrature rule, then this, say this integral say I, now becomes W1 which is say
5 by 9 into phi1. Phi1 is the value of phi at minus root of 0.6 plus 8 by 9 into phi2. Phi2
is the value of phi at psi is equal to zero plus 5 by 9 into 5 by 9 into phi evaluated at
root of 0.6. W1 phi1 plus W2 phi2 plus W3 phi3 is what we get as a three point
quadrature rule. You can keep on extending. In fact there are tables that are available.
I am not going into the details of it; you can have four point quadrature rule and so on.
I am not going to go into the details of it. Nevertheless, it is important that the
accuracy of it is well understood in the sense that whenever you have root of 0.6 and
you do not want to calculate it or you want to calculate and put it, you do not want the
computer to calculate it, then this root of 0.6 should be given to very large number of
significant digits, because that will affect our results.

The question now is what is the quadrature rule that you are going to use or you
should use? Before we come to that, let us now extend our concept to 2D aspect,
because my K matrix, which I had and which I transformed is now in 2D; may be we
have to extend to it 3D as well. Let us look at 2D, because if you understand 2D, how
to do 2D, then 3D extension is quite simple. In other words, what is that we are going
to look at?

(Refer Slide Time: 18:31)

11

Instead of what i had instead so Now, K is minus 1 to plus 1, minus 1 to plus 1, some
phi t J d psi d eta and phi is a function of psi comma eta. What is phi, psi comma eta is
always confusion for students. Phi psi comma eta is nothing but B transpose DB
calculated at psi and eta.

(Refer Slide Time: 19:00)

Let us say that that is the element after the transformation into the natural coordinate
system and so on. What is the rule that we are going to use? Say for example, we use
a 2 by 2 rule. We use two integration point for one-dimensional case. Now, we are
going to use 2 by 2 rule, which means that I will have totally 4 points. How do I get
four points? By extending that psi plus or minus 1 by root 3 and eta plus or minus 1
by root 3. In other words that, this line, this line and this line are such that, this is
equal to plus 1 by root 3 minus 1 by root 3 and minus 1 by root 3 and so on. In other
words, the coordinates of this is minus 1 by root 3, minus 1 by root 3, 1 by root 3
minus 1 by root 3, 1 by root 3 1 by root 3 and so on, so, I get 4 points. Then how do I
now evaluate this stiffness matrix? That is evaluated as two sigmas; two sigmas ij
both of them vary from 1 to 2. Phi calculated at these positions, four positions, let me
call at phi i j w i. Two sigmas, let me call Wi Wj and phi calculated at psii etaj, i and j
i and j varies from 1 to 2.

I will have now have 4 points or in other words, W1 into W1 phi into, phi calculated at
not into, phi calculated at psi1 eta1, then W1 W2 phi calculated at psi1 eta2; what is psi1
12

eta1 and so on. These are points; that is psi1 eta1 which is minus 1 by root 3, minus 1
by root 3. What is psi1 eta2? That is this point and so on. You calculated these four
points, multiplied by the weights then or rather this point and so on. You get or in
other words, to calculate the psis at all the 4 points and the weights happen to be for
this problem 1 and so Wi into Wj into phis and these four points give you the value of
K. In other words, what is that we do? We calculate K at each of these points and then
sum them up. That is what we mean by stiffness matrix. Is that clear? May be you can
have a look at this whole thing again, what all we did, so this stiffness matrix
calculation becomes quite nice and easy.

There are two other things that we have to notice or we have to do. That is quite
involved. Let us see what are the other things that we have to be involved? I hope that
I have followed all the symbols what we did in the last class. Now, what are the other
things that we have to do? We have to calculate the right hand side. In the right hand
side, we have basically two types of loads, apart from the concentrated load; the body
force and the surface force. Body force is not a problem, because the procedure is
exactly similar, but surface forces creates certain troubles. What is the trouble that we
are going to have with the surface forces?

Let us look at this element.

(Refer Slide Time: 23:38)

13

Let us say that that is the psi eta coordinate system. Let us say that this element sits on
the boundary of a component, say for example, that is the boundary of the component
and so on. In that component or in that element, we have defined a surface force or
surface traction, pressure whatever we call it. Let us say that we have given a pressure
of say phix and phiy in this particular surface. What is it that I have to do? I have to
now calculate the right hand side.

(Refer Slide Time: 24:41)

What is the right hand side? If you remember, the right hand side is N transpose phi
ds at s. In other words, we have to evaluate a surface integral which is N transpose phi
ds. This phi is given as phix and phiy. We know what is N? We have been through it
for quite some time and hence we know what N is. The whole idea now, is to
calculate what N transpose phi ds is. How are we going to calculate? Because, we
have already we are into what are called as natural coordinate system, we have
already transformed the element into a natural coordinate system and hence my job is
to now transform this equation as well to the natural coordinate system.

Or in other words, my first job is to recognize that the phase in which I have given the
load happens to be what? Eta is equal to minus 1 in the natural coordinate system. In
other words, symbolically writing it, if I write this as the natural coordinate system,
no, sorry; psi eta eta is equal to minus 1 and hence eta is equal to minus 1. That is the
phase in which I have given, I have given the load or in other words here is the phase
14

in which I have mentioned phix and phiy. Now, how am I going to evaluate this? In
other words, my equation now actually becomes, I am sorry, evaluating it as d psi;
this implies I should not say equivalent. But there are lots of things I have to do in
order to get to this.

What are the things that I have left out? I have left out quite a few things. They are
implied; no, they are not equal, because I have left out certain things deliberately.
What are the things that I have left out? I cannot straight away convert it like this.
Though actually I want to have equation of this sort, I cannot straight away do it. First
of all, let me convert this surface into line. Why is that we have to convert the surface
into line, because we are now talking only about a two dimensional situation, a plane
stress example.

(Refer Slide Time: 27:34)

My first step is not to write like this, though ultimately I want to do it. First step is say
that look integrate it say along the length of the element. Let me call this as zero to l N
transpose phi dl. That is my first step. What is N transpose phi dl? How do I convert
it? In other words, I can do that as minus 1 1 N transpose phi J into d psi. That is
exactly how I do it. What is J? Determinant of Jacobian or the Jacobian; in this case,
what is J and how do I calculate N? Usually these are questions which are confusing
to the students, the answer is very simple.

15

How do I calculate N now, because you may wonder N had psi and eta. How do I
calculate N now for one dimensional case, from my previous Ns which I have? How
do I calculate that? The answer is very simple. In the previous Ns, what are the Ns
you had? 1 by 4 into 1 minus psi into 1 minus eta plus that is for N1; N2 you had 1 by
4 into 1 plus psi 1 minus eta and so on. In those Ns just substitute eta is equal to
minus 1. In other words, what we mean is this N or N transpose is calculated at eta
equal to minus 1. So, that problem is solved.

When I substitute eta is equal to minus 1 by N, purely become a function of psi. No

problem, so, straight away I can calculate. Phi is a value, phix and phiy that is in other
words, the x and y direction they are given. That is not an issue, because let us for the
moment assume that they are constants or else we may have to do something else. Let
us not confuse the issue. Let us say that they are constants that are acting in the x and
the y direction or else we have to interpolate them as well. Let us not worry about that
or in other words phix is constant through out this phase, phiy is constant through out
that phase. They are just written there in terms of numbers.

The first term is known. Second term is a vector, it has two entries or two rows, one
column and next is J. Once I do that, once I calculate the J, my calculation of this
integral becomes very simple, because I can use numerical integration. Question is,
now what is that J? Let us take a pause and see how to calculate that J matrix? Yeah,
so have a look at this and see for a minute as to how we can now calculate J? In other
words, we have to very carefully define what J is, the Jacobian. After all what is that
we have done here in this equation? We have replaced dl by J d psi. So, what should
be J? J should be dl by d psi.

We have to find out now dl and write it in terms of d psi. That is all we have to do.
Once we do that, in other words, once we calculate dl, write down dl, then it is very
simple for us to calculate J. What is dl or how do you calculate dl, is the question.

16

(Refer Slide Time: 32:00)

dl can be written as, let us look at that figure. Have a look at this figure and see how
we can write down. Have a look that figure and see how we can write down dl. We
can write down dl in terms of dx by d psi whole squared or dow x by dow psi whole
squared plus dow y by dow psi whole squared whole power half

d psi. How do I

get dow x by dow psi and dow y by dow psi? That is very simple because, I know that
this can be written in terms of Ns.

When I now substitute dow x by dow psi and dow y by dow psi, where x is equal to
Ni xi and then simplify this, I get that dl is equal to l by 2 d psi. I am not going to do
that exercise; that is very straight forward, just substitute it and you will see that dl
and d psi are related by l by 2 and now what is this l by 2? l by 2 is nothing but the
Jacobian. With that result, we can write down our previous expression for N transpose
phi J d psi as minus 1 to plus 1 N transpose, phi remains the same; now, J becomes l
by 2 and we get d psi.

What is l by 2? What is l? l is the length of that element, length of this particular edge
we are talking about. So that is the l and hence this expression is useful to determine
re, or due to pressure. Please note that phi is actually a vector, phix and phiy.
Correspondingly this is also a vector and it has two entries. I need not tell you what N
has? We can very easily find out, because I have already given you a clue that this N

17

has to be calculated with eta is equal to minus 1. In other words, what is essentially or
what is the crux of our approach?

Our approach has been that we not only calculated the stiffness matrix by
transforming it to a natural coordinate system and then resorting to what we call as the
numerical integration.

(Refer Slide Time: 35:00)

We also calculated or we can also calculate the other res or the rs or the right hand
side, to be more precise or in other words the load due to the surface traction or
pressure can also be calculated by means of this natural coordinate system by
transforming the equation. What is this equation, how did we get? Please note this
equation we got it from our virtual work principle. By transforming the equation
which we got from the virtual work principle on to say the natural systems or natural
coordinate system, where psi varies from minus 1 to plus 1. That in a nut shell, shows
that all the aspects whether it is the stiffness matrix or whether it is the force terms, all
these things can be calculated in a very consistent fashion. There is always a tendency
to replace this phi by means of dividing that or multiplying this by total length and
then dividing it by 2 and putting it at 1 and 2.

One of the things we should notice, there are two things I should not say one; two
things that you have to notice. One is that, you should always have a look at how phi
18

is defined. Please note that in this particular definition we have assumed that phi is
defined per unit length.

(Refer Slide Time: 36:30)

If phi is not defined per unit length, then you will have to now look at t, thickness as
well so that, you convert it into unit length. Why is that we are emphasizing on unit
length is because we are now in a plane. So, it is important that you look at how phi is
defined. If phi is defined anywhere else, then your units on the left hand side and the
right hand side should be consistent. Say for example, we have defined a t in the left
hand side for K. Then you should also have a t, if it so happens that you have defined
phi as well as t. In that case, you have to multiply this by t as well. Please note another

This is a very simple thing that people make a mistake; please note how phi is
defined, so that the units what has been given there is consistent and thickness is taken
care of properly by the units that are given for phi. That is a very important thing and
the other one is that this is only an example for eta is equal to minus 1. It so happens
that the same element may be in the other end, at the other side of the boundary and
so that we may have to apply pressure not at eta is equal to minus 1, but say psi is
equal to 1 and so on. You have to be properly accounting for the surface where we
give this pressure. What is the surface that we are interested in, where is that we are
giving pressure and accordingly you have to do this derivation.

19

Thirdly if the pressure does not happen to be constant, if it varies, say varies, from
one end to the other, then you have to introduce linear interpolation even for phis as
well. That makes the problem slightly more complicated, but not difficult. It is very
straightforward. Only thing is that in that case, you have to now introduce
interpolation for the phis as well. There is always a tendency to lump these loads. We
will see later in the course that lumping may be very dangerous; it may not be
consistent with the result we have got.

In fact these loads which we have calculated by such a very rigorous procedure of
integration are called as consistent and nodal forces. They are consistent, they are
consistent with our other definitions. So, they are called consistent nodal forces. You
have to be very careful not to miss out this and just to divide say for example length is
equal some phi and if the load say phiy acting is say 5 per unit length, so you cannot
multiply 5 into 5 which is 25 and then divide it by 2 is 12.5, 12.5 and so on. Yes, it
will work for linear elements, but for other higher order elements this kind of
procedure, will not work. Even when you write a program, even if it works for some
case, it does not mean that you should not do this integration. You should go ahead
and do this kind of integration and if you want to write a code, then this is what
should be implemented and that is very important.

A very detailed account of how to write the code for this has already been given by
Owen; I think Hinton and Owen, in a book called Finite Element Programming,
where you will see that a very detailed exposition of how this code can be written is
given. Anyway, for our course it may not be right now necessary to go into the details
of how programming is done, but nevertheless, it is nice to know that these things are
important.

What is the second issue? We have now looked at what is called as a quadrilateral
element. What are the other types of elements which are very popular and which are
important?

20

(Refer Slide Time: 40:53)

Triangular elements; triangular elements become very important. Why is that

triangular elements are important? Because, most of the mesh generation codes or
mesh generation packages have triangular or tetrahedron . three dimensional
counter part as a mesh generation facility. Today there are lot of mesh generators,
which can do quadrilateral elements. So, there is no problem, but hardly there are any
mesh generators which can do what we call as a hexahedron element and hence it
becomes important that we know how to handle, how to handle triangular elements.

Basically there are two ways in which triangular elements stiffness matrix are
determined. But, by the way, we already noted one thing important about triangular
element. If you remember that, we said that triangular elements are constant strain
triangular elements or in other words we said that the strain happens to be a constant
in a triangle. We illustrated this with a very simple example. We said that, what is that
we said? We said that let us say that we call these three nodes as 1 2 and 3. What did
we say? We said displacement can be a0 plus a1x and a2y can be written like this.
We said, if we want to calculate dow u by dow x, what is that we get? a1. What does it
mean? It means that, when we express the displacement like that as an interpolated
value, then dow u by dow x becomes a constant. It does not matter whether we
replace these as by the degree of freedoms at the three nodes 1 2 and 3; does not
matter, because the basis is only a linear interpolation formula like that and hence
21

dow u by dow x becomes a1 and so on. So, the strain within that element becomes
constant.

What is the trouble with this? We have already seen this with an example. We saw
that because of this, we have to have lot of elements, especially when there is bending
that is involved. There are lots of elements that we have to use in order that the
bending exercise or when an element or when the component bends, bending is
properly taken care of. I am reemphasizing it, basically because I have found lot of
mistakes during modeling. Why is that there are lot of mistakes? Because it is very
easy to do what we call as free meshing. Take a three dimensional component, mesh it
with a tetrahedron and just go ahead and finish the problem. But, if you do not do
proper mesh sensitivity, as we did on the other day, then you are going to be in
trouble.

With all these drawbacks, still we have to learn about triangular element and it is three
dimensional generalization which is the tetrahedron element. We will concentrate
only the triangular element, because what all we are talking about in the triangular
element is very similar to the tetrahedron elements. There are two procedures, the two
ways in which you can calculate the stiffness matrix. One, procedure number one, is
to collapse two nodes in a quadrilateral element. What do we mean by that?

(Refer Slide Time: 44:58)

22

Let us say that I have an element like this, simple element, just illustrate how we
collapse. Let me call this as 1 2 3 and 4, anticlockwise sets. Fine; now I can develop a
triangular element following exactly the same procedure, whatever I have done so far,
exactly the same procedure, by collapsing 1 and 2 and which results in an element
something like this. Whatever be the shape, it does not matter. What do I do? I
collapse x1 to x2.
What do you mean by collapsing? I simply mean that I give the coordinates of 1 and 2
to be the same, x1 is equal to x2, y1 is equal to y2. Where do I do that? I do that in my
expression for x and y, first step.

(Refer Slide Time: 46:09)

What is that I mean by collapsing? By substituting that x1 is equal to x2, y1 is equal to

y2 in these two equations; is that all? No, because we are using isoparametric
formulations and hence we have similar values for u and v as well; for u and v as
well. So, u and v are again written and this collapsing exercise is done. So, I can say
that it is 1 comma 2 3 and 4. The collapsing exercise is done there as well, so, u1 is
now the same as u2. In other words, when I do that what essentially I get as degrees of
freedom? Simple; u2 v2 u3 v3 and u4 v4. These are the degrees of freedom that I get.
One way of calculating the stiffness matrix is by doing this collapsing business. This
is very simple and I will do that in the next class by taking an example, simple

23

example and following all the steps that I have done in this class and the previous
class.

(Refer Slide Time: 47:28)

The other way of doing it, there is another way of doing it, is by following what is
called as area coordinate systems. The area coordinate system is some sort of a
generalization or an extension of the natural coordinate system, which we have in
place right now. So, the second way of doing it is by area coordinate system.
Whatever we have been talking about so far has been for linear element. But that is
not sufficient. I said that the interpolation values can be of a higher order polynomial.
It can be quadratic, can be cubic and so on. Whatever we have been taking about,
whether it is the quadrilateral element or whether it is a triangular element, it is
possible to raise the polynomial level or polynomial degree, from a linear exercise to
quadratic, cubic and so on.

Our exercise, next exercise is to complete the triangular element by collapsing and get
into this area coordinates and then extend this whole exercise to a higher order
element which is not very difficult. I am not going into each of the element. For
example, continuum elements of 3D, I am not going to the details, because it is
exactly similar. I leave that out as an exercise. We will meet in the next class and see
how we tackle these triangular elements.

24