SOLUTIONS OF SELECTED PROBLEMS OF EXERCISE 4

ANS 4.7

d2y dy
Differential equation: 5  7y  0
dx 2
dx
Galerkin method:
Let w be an arbitrary weight function:

 w(
d2y dy
h
 5  7 y )dx=0
0
dx 2
dx
Integrating by parts:

 dx+5 w dx+7  wydx=0

dy dw dy dy
h h h h
w
dx 0 0 dx dx 0
dx 0

 dx-5 w dx-7  wydx=w

dw dy dy dy
h h h h

0
dx dx 0
dx 0
dx 0

Aproximating y and w by shape fun.:

y   N   y ne  w   wne   N 

 x
1  h 
Where N    h=element length
x 
 h 

On putting these values, we get:

 dy 
 dx 
0  w   N, x   N, x  dx  y   50  w   N   N, x  dx  y   7 0  w   N   N  dx  y    w   dy 
ne  0
h h h
 ne
 ne
 ne
 ne
 ne
 ne
 
 
 dx h 

 dy 
 dx 
   N , x   N , x  dx  5  N   N , x  dx  7   N   N  dx   y   
 h h h
 ne  0

0 0 0   dy 
 dx h 
……………since nodal values of weights are arbitrary

1
  dy 

  1 1 7  2 1  5  1 1   y1   dx 0 
On integrating : 2           
  1 1  12 1 2  2  1 1   y2   dy 
 dx h 
……… see 4.31&4.32 from lecture notes(8-10 )

 dy 
 
1  40 61   y1   dx 0 
     …………………element equation
12  1 20   y2   dy 
 dx h 

For 2 elements:

 dy 
 
 40 61 0   y1   dx 0 
1     
 1 20 61   y2   0 
12   
 0 1 20  y3  dy 
 
 dx 1 
Using boundary conditions
y=0 at x=0
y’=0 at x=1

1  20 61   y2  0 
  
12  1 20   y3  0 

y2  0
y3  0
For given differential equation and boundary conditions we will get zero nodal
displacements, no matter how many elements we take.

Exact solution:
3
y  e ( A cos  x  B sin  x) where   2.5 ,  
2

On putting boundary conditions, we get A=0 B=0

i.e. y=0

2
ANS 4.8

Solution:

Since the given equation is in variational form. Therefore we can directly proceed, before
that we have to first convert 10 u(1) into a function by using the property of deriac delta
function.
Therefore we can rewrite the above equation as:

I   5( ) dx  10  uδ(1  x )dx
du 2
1 1

0
dx 0
Now putting
u  u ne   N 

We will get

I   5 u ne   N , x   N , x  u ne  dx  10 u ne   N  δ(1  x )dx

1 1

0 0

1  x / h 
Putting N    & solving, we get
x / h 

 u   10 
 1 1 ne 1  x / h  ne
I  5 / h u ne    u 
 1 1   x / h 1

I
Now finding =0
 u ne 

 u   10   =0
I  1 1 ne 0 
 10 / h 
 u   1 1  1 
ne

 u   10   -----------------------------(1)
 1 1 ne 0 
10 / h 
 1 1  1 

1. Now discretizing the whole domain into 1 element:

Putting h=1,

3
We have,
 1 1 u1  0 
10      10  
 1 1  u2  1 

Also we have, u1 =0,

Therefore, on solving above equation we get

u2 =1

2. Now discretizing the whole domain into 2 elements:

We already know that the assembled matrix will be

 1 1 0  u1  0 
    
10 / h  1 2 1 u2   0 
 0 1 1  u  10 
  3  

Putting h=0.5, & u1 =0 we will get,

u2 =0.5 & u3 =1

3. Now discretizing the whole domain into 3 elements:

For three elements the resulting assembled matrix will be

 1 1 0 0  u1  0 
  
 1 2 1 0  u2  0 
10 / h   
 0 1 2 1 u3  0 
 
 0 0 1 1  u4  10 

Putting h=1/3 & u1 =0, we will get,

u2 =0.333, u3 =0.667 & u4 =1.000

Since u(1), is always coming out to be 1.000,and hence the solution is converging.

4
ANS 4.9
Variational form is given

 du  2 
I 
1

   (5  u )  dx
0  dx  

Derivative in variational form is first order

So minimum C0 continuity in displacement

Take two noded element

u  N 1u 1  N 2u 2

Shape functions of the linear element

N1  1 x
N2  x
N1
N    
 N 2

U can be written in matrix form as

u  une   N 
Element displacement matrix = u ne

u1 
T

une    
u 2 

Substitute this in variational form

5
 I    une   N , x  N , x  une   5  une   N   dx
1

According to Ritz method derive I by element

0

displacement matrix u  ne

d une  0 
  N , x  N , x  une   5   N   dx
dI
1

By solveing above equation

 1 1 u1  1/ 2 0 
 1 1  u 2  5 1/ 2   0 
      

u1 is already known so eliminate first row and column

u2=2.5.

--------------------------------------------------------------

%Matlab program for solveing the problem

clear all
clc
n=input('enter upto how many elements :');
a=zeros(2,2);
a(1,1)=1;
a(2,1)=-1;
a(1,2)=-1;
a(2,2)=1;
m=1
for i=1:n
b=a;
c=zeros(i+1,i+1);

6
 c(1,1)=a(1,1);
c(2,1)=a(2,1);
c(1,2)=a(1,2);
c(2,2)=a(2,2);

for k=2:m

c(k,k)=c(k,k)+b(1,1);
c(k,k+1)=c(k,k+1)+b(1,2);
c(k+1,k)=c(k+1,k)+b(2,1);
c(k+1,k+1)=c(k+1,k+1)+b(2,2);
end

ANS 4.10
N1=0.1, N2=N3=0.2.
Using N1  N 2  N 3  N 4  1 , we get N4=0.5.

ANS 4.11
Given problem in variational form is,

  du  2 
I       5u dx
1

 dx  
0  
Lagrangian shape functions are

7
By substituting we get,

N1  1  3 x  2 x 2
N2  4x  4 x2
N3   x  2 x 2

Let u e be the approximation inside n-noded element

u e  u ne   N 

Putting this value in I we get,

I   u ne   N , x   N , x  u ne  dx  5 u ne   N  dx
1 1

0 0

For extremizing the condition,

I
0
 u ne 

8
   N   N  u  dx  5  N  dx  0
1 1
ne
,x ,x
0 0

This is the element equation.

For 3-noded element ,
 7 8 1   u1  1 
1    5 
8 16 8 u2    4 
3 6
 1 8 7  u3  1 

Applying boundary condition u  0   0 and solving 2 by 2 matrix we get,

u2  1.875
u3  2.5

ANS 4.12
d2 y
he given differential equation is  5 ( x  0.5)  0
dx 2
By Ritz method,
Variational form of above differential equation is

d 2 y
h
 I n e
 (  5  ( x  0 .5 )) y ) d x  0
0
d x 2

It is for any general element of range (0 to h)

 ) d x  5  (  ( x  0 .5 )  y ) d x
dy dy dy
h h h
 I ne =  y  ( 
dx 0 0
dx dx 0

1
Using the variational property, u u   u 2
2

 I =  y    ( ) 2 dx  5 ( ( x  0.5) y)dx
dy 1 dy
h h h
ne

dx 0 2 0 dx 0

 I   y    ( ) 2 dx   5 y ( x  0.5)dx
dy 1 dy
h h h
ne

dx 0 2 0 dx 0

 I ne  0

y   ( )2 dx   5 y ( x  0.5)dx
dy 1 dy
h h h
I 
ne

dx 0 2 0 dx 0

9
  dy 
  dx  h
1
  
 y ne   N , x  N , x  y ne ne  h
   5  N   y   ( x  0.5)dx
h
I  dx   y   ne
ne

20  dy  0
 dx 0 

dI
ne

0
d  y ne 
 dy 
  dx  h
  N , x  N , x   y     5  N  ( x  0.5) dx
 h
h
ne
dx  
0  dy  0
 dx 0 

Using Dirac delta property…….. 0

1
 ( x  a ) f (t )  f (a)
…we get,

 dy 
  dx 
  N , x  N , x   y    5 N  0.5  …………….(1)
 h
h
ne
dx  
0  dy 
 dx 0 
Where,
0 
 N  0.5   
1 
 x x 
 N   (1  ) ( ) 
 h h 
 x 
(1  h ) 
N    
 ( x) 
 h 
1
 N , x    1 1
h
1 1
 N , x   
h1 
So now for first element the range of element is (0 to 0.5)
 dy 
  dx  0
1 1 1  y1   
0 h  1  h  1 1  y2  dx   dy   5
 1
0.5

 
 dx 2 

10
For first element on integration of equation between limits (0 to 0.5) we get,
 dy 
 
 1 1  y1   dx 1  0 
2     
 1 1   y2   dy  5 
 dx 2 
For second element range is (0.5 to 1)
 dy 
  dx 
1 1 1  y1 
0.5 h  1  h  1 1  y2  dx   dy 
 2
1

 
 dx 3 
For second element on integration of equation between limits (0.5 to 1) we get,

 dy 
 
 1 1  y2   dx 2 
2    
 1 1   y3   dy 
 dx 3 
On assembling the two elements and applying given boundary conditions
y=0, at x  0 …………..(1)
dy
 0, at x  1 ………….(2)
dx
We get,
 dy 
 
 1 1 0   y1   dx 1 
   
2  1 2 1  y2    5 
 0 1 1   y3   0 
 
 
 2 1 5
2  
 1 1   0 
On Solving we get,
y2  y3  2.5
y1  0

11