Hello, welcome to basics of finite element

analysis, this is the third day of the first

week. In our last lecture we had explained
two important concepts; one is using this
example of where we were trying to calculate
the area under a complicated curve. We had
explained what is meant by an interpolation
function or a shape function that is one concept
we had introduced, and the second thing we
had shown was that the error at least in context
of that particular example between the exact
value and the numerical value of the area
under a curve dependent on two parameters,
one was the size of the element of a you know
size of the element, so the larger the number
of elements the lesser was the size and then
the second way to reduce.
Error was to have a more accurate interpolation
function or shape function. What we are going
to do today is we will continue that discussion
but in context of a different example, a different
example.
So, so what we will do is consider the example
of heat conduction okay, so suppose I have
a hollow pipe
and this is the outside diameter, outside
circle, the inside circle is having a radius
of Ri and the outside circle has a radius
of Ro and at some radius it is R. So you have
a hollow pipe so what essentially you are
seeing is the cross section of the pipe, and
let us look at the boundary conditions.
The first boundary condition, so what is the
boundary condition, whatever are the physical
boundaries of the object, whatever conditions
physical conditions are present there that
is what is referred as boundary condition,
so the first boundary condition we are specifying
we know that is that on the inside surface
we have perfect insulation
at r =Ri okay, so if you have perfect insulation
there is no heat, no heat either going into
the is not crossing this boundary either in
outwards or inwards at r =Ri
Okay so mathematically I can express that
condition as K times R dT/dr is equal to 0
at r = Ri okay, so how does that come I mean
for this you have to refer to some heat conduction
equations, the focus in this class will not
be explaining how these conditions are coming,
but if we know the condition then how do we
solve the differential equation? The second
condition is that at r is equal to Ro temperature
is specified okay.
So Ro, so T so let us say temperature is designated
by letter T so that is T at Ro is equal to
T0, so these are the two boundary conditions,
there is no heat crossing the boundary on
the inner surface, on the outer surface temperature
is T0 okay. Now we had explained earlier that
if we have to solve n the other third thing
is and that is not a boundary condition that
there is heat being generated in, in this
whole volume of the thing.
So when you look at this problem what you
see is that the cross-section of this, of
the pipe is axi symmetric right, cross-section
is axi symmetric.
Which means there is no change in cross-section
if I move in the ? direction, the second thing
we notice that does not matter whether I am
at this position, this position, this position,
this position you know, the boundary conditions
are also axi symmetric, so whether ?=0 degrees
on the outside boundary temperature is T0
and for any value of ? temperature is T0 on
the outside boundary condition boundary for
any value of ? there is perfect insulation
on the inside boundary.
So cross section is axi symmetric, boundary
conditions are axi symmetric, and then I am
telling you because we know this that material
properties
are axi symmetric. An example of this would
be suppose you have a pipe and that pipe,
on that pipe I have put another pipe, material
properties are varying in the radial direction
but in the ? direction they are not varying,
the ? direction are not varying.
So this is one of those problems they are
material properties this case conductivity
K can vary in the radial direction but it
is not varying in the ? direction, so material
properties are also a thing
So if these three conditions are met where
the geometry is axi symmetric, boundary condition
is axi symmetric, and material properties
are axi symmetric okay, and fourth is loading
conditions are axi symmetric, so this is the
fourth condition. What does I mean by loading
condition, so it is in the volume of this
pipe in some locations heat is being generated
maybe heat is getting generated let us say
R=Rh but it is getting generated at in an
axi symmetric sort of way.
At a particular radius, one example could
be you have a wire, a conductor wire, copper
wire which is having several layers of insulation
right and current is passing through it, so
wire is a source of heat, that heat is being
generated in an axi symmetric way okay it
is getting generated in an axi symmetric way.
The properties of the conductivity of the
wire in the center it is that of copper, outside
there is one insulation, outside of that insulation
there may be another layer of insulation,
so there may be three or four layers.
So the material properties are changing in
radial direction but they are not changing
in ? direction, also the cross section of
the wire and all individual content is also
axi symmetric so everything is axi symmetric
in that.
Situation in case of wire and same thing is
here, so if all these four conditions are
met, cross-section, boundary conditions, material
properties and loading conditions, all of
them they have to, you have to meet. Then
we can use a governing equation which neglects
role of ? and z okay, so z is the direction
of axis and ? is the direction of angle, the
hoop angle so we can use a governing equation
which is which is also axi symmetric in nature
but if even one of these conditions was not
axi symmetric, suppose material was not axi
symmetric.
Then the governing equation we use cannot
be of one dimensional nature okay, so now
I am going to write down the governing equation
for this problem because that is the first
step that we have to figure out what is the
governing equation, so we will not derive
that equation in this course we will just
take it from some other book, so for axi symmetric.
Heat conduction problem steady state heat
conduction problem the governing equation
is 1/ r [d / dr ( k times r dT) over dr] is
equal to q(r) and I will explain that, so
here r is the radius so if I want to calculate
the value of temperature at radius R that
is what it means okay, k is thermal conductivity
and this thermal conductivity does not change
with ? direction but it can change in the
radial direction that is why.
It is in the brackets otherwise it would have
been a constant and it would have been out
of the brackets. And q is thermal load or
heat generation rate I will say, so this is
the governing equation and we have also.
We have already explained that these are the
boundary conditions. So this is the geometry,
these are the boundary conditions also heat
is getting generated somewhere in the body
in a axi symmetric way.
So the governing equation for this problem
is this, and our aim is to find T as a function
of r okay, so what we are interested in finding
out is how does T change as I move from Ri
to R0 from inside radius to outside radius,
that is what my aim is okay, so.
If you try to solve this problem using some
standard way in an exact form, in an exact
form you will not be able to solve that problem,
so what we will do is.
We can solve this problem however using finite
element method, so now what I will do is I
will just conduct first few three, four steps
not in detail but at a slightly more detailed
level compared to the example which we saw
earlier where we were having, we were trying
to calculate the area under a curve, so the
first
Thing we do is, so what I said that we are
trying to.
We are trying to calculate how is temperature,
see here r=Ri and here r =RO here r=Ri how
is temperature, so temperature varying, it
could vary this readily in the radial direction,
so this is so I am trying to make a plot,
x axis of the plot represents radius, y axis
represents temperature, so I am trying to
get a plot of this nature, this is radius,
this is temperature as a function of r and
the domain is Ri to Ro, inner radius to outer
radius okay.
My domain is I am trying to compute the value
of temperature for all values of r between
Ri and Ro .
So my domain is this, this is Ri this is RO
okay, this is my axis for radius and this
is my domain okay, so the first step we do
is we break this Ri to Ro this is my domain,
so I break domain into sub-domains, sub-domains
I, so what, what I am doing, I will break
it up into let us say n elements, so this
is first boundary of first element, boundary
of second element, boundary of third element
and so on and so forth and this is the boundary
of nth element.
So this is my first element, second element,
third element, and so on and so forth and
I get my nth element here, is this clear,
this is the first step. Second step, so like
in case of area when we are trying to compute
between the first boundary of the end point
of the first element and the end point of
and the second end point of the second first
element we assumed how that function is changing
and we could make lots of choices for those
functions, in first case we said that that
function was constant, in second case we said
that function was linear, in third case we
said function was quadratic.
We make a, we have to make similar choices
here okay, so what we do is
We develop those relations, so we say that
temperature in the first element T1 is first
element at radius r is equal to okay T1 ?1,
and I will explain all this so just be patient
and there is a + T2 ?2 r excuse me+T13, what
have I written, so before I explain what all
this means thus so I have a bunch of T’s
and a bunch of ?’s okay. The superscript
of T designates the element number. The superscript
of T and also the superscript of ? it designates
the element number; the superscript designates
the element number because this is the terminology
we will follow consistently through the course.
So this is important to we have to get it
right okay, so we are trying in this equation
to figure out how is temperature changing
from the first point of from the first between
the first node of element number 1 to the
second node of element number 1 right, and
that is why we have written
T1 so this expression is it tells us how is
temperature changing over the first element
that is all it does, second thing these ? they
also have a superscript and subscript. The
superscript again for these ?’s designates
that element number and thus and they also
have a subscript, these ?’s are interpolation
functions
and they are not any significantly different
than the functions which we had considered
when we were computing the area under the
curve you can assume that it is a constant
function you can assume.
That it is a linear function, you can assume
it is a quadratic function so this i could
be constant and actually we will when we will
do computing since greater after four five
lectures we will actually use only these simple
functions, we will not use some very complicated
functions, so these are interpolation functions
or shape functions understood, and I can have
n different functions, it is there is no rule
which says that only one function will help
me explain how the shape is changing okay.
So I can have a constant function and I can
also add that constant function to a linear
function and I can also add that to a quadratic
function and so on and so forth right, so
I can have n number of functions.
I can have n number of functions which help
us understand how the temperature is changing
over the length of the element not the overall
cylinder, we are right now only at the first
element understood, so these are interpolation
functions which help us understand how temperature
is changing over the first element. They can
be ? functions what is the amplitude of a
? function, one right?
These T’s are the amplitudes of these functions,
these T’s are their amplitudes and there,
we will interpret these T’s in some other
way also, so I can compress this thing as
T1r is nothing but T1 instead of 1 I say j
?j r ? 1 1 and j is equal to 1 2 small n,
small is thus the number of functions which
we are considering to capture the variation
of that function on an element okay, big N
is the number of elements, this is big N.
Capital N is the number of elements, small
n is the number of functions which we are
using on each specific element understood
okay. Similarly T2r and y ever do I have bracket
in parentheses because these size they are
not constants, they are functions and they
are dependent on r right, similarity T2r is
j =1 to n is Tj ?j so you have to be patient
while writing these because there are a lot
of subscripts and superscripts and the superscript
will be 2.
And for the nth that is big Nth element TNr
it will be ? Tj and N ?j and I am summing
up over 1 to n is this clear, right this function
is valid from x0 to x0 + h1 let us say the
h1 is the length of the first element, so
let us say this length is h1 this h2 this
h3 and so on so forth this is hN begin, so
the first function is valid after that the
element does not exist so we do not have to
worry about that function.
So the first function will be computed only
from x is equal to x0 to x so I am sorry it
should not be x0 it is Ri to Ri +h1, so it
is valid from Ri which is the internal radius
to Ri +h1 which is the length of the first
element. This function is valid from Ri +h1
to Ri + h1 + h2 is this clear, and so on and
so forth. Similarly we can find this nthT
will be valid from Ro either you add up h1
+ h2 + you know hn -1 or from the other side
it will be Ro-hn to Ro does not matter okay,
so for the
Eth element so
This is T1 this T2r this T3r suppose I have
to compute the value of T for eth element.
Then Ter = Tj ?jr e,e n okay, so I
think we will close at this stage today and
we will continue this discussion tomorrow.