|


 
  
 

FEM is based on Direct Stiffness approach or Displacement approach.
A broad procedural outline is listed below

1.     
    
 

Skeletal structures
14 15 16
Skeletal structure gets
11 12 13
discretized naturally.
7 8 9
Member between two
joints are treated as an 4 5 6
element.
1 2 3

Continuums Arbitrary
discretization.
›recautions to be taken while discretization.
£
   


1.›rovide nodes wherever geometry changes
2.›rovide a set of nodes along bimaterial interface,  
£
  
 

so that no single element encompass or cover both
materials. Element should cover one material.  

3. Nodes at such points where concentrated load acts.

4.Nodes at points of specific interest for the analyst.
5.Nodes/|elements are provided such that distributed loads are covered completely
by the element edge. Distributed load shall not be applied partially on any element
edge.
›recautions to be taken while discretization (contd)

6.Nodes to provide prescribed boundary condition.

‰ ‰
1

7.Fine mesh in the regions of steep stress gradient.

8.Use symmetry condition
9 Aspect ratio of element[1 ‰ ‰
1

10. Avoid obtuse/acute angle

11.Node numbering along shorter direction.
 å
 
ü
Fundamental concept of FEM å

     

Consider a bar subjected to some
exicitations like heating at one end. Let
the field quantity flow through the
body as fig2, which has been obtained Fig. 2
ü
by solving governing DE/›DE, In FEM
the domain  is subdivided into å
subdomain and in each subdomain a ü

 
piecewise continuous function is å

assumed.   


    

The fundamental concept of FEM is that å

continuous function of a continuum å

å
å å
(given domain ) having infinite å
ü
degrees of freedom is replaced by a
discrete model, approximated by a set of
piecewise continuous function having a
finite degree of freedom.
Thus the method got the name finite element


coined by Clough(1960). 


ü
x     


 


 


  e

In the displacement approach a displacement function is assumed for

the element. For example ui uj
For a one dimensional element i j
u(x) = 2 > 2
x ui uj
uk
u(x) = 2 > 2
x > 2x2 i
k
j
For two dimensional rectangular elements displacement field at any
interior of element is given by
u(x,y) = 2 > 2
x > 2y > 2xy
v(x,y) = 2> 2x > 2 y > 2
xy 1 Cons n

x y nea
isplacement at any
x2 xy y2  ada
interior point x3 x2 y xy 2 y3 Cub

u(x, y) 
(x, y)  x4 x3 y x 2y2 xy 3 y4 ua

v(x, y) 
›ascal triangle for 2D problems
odal displacement
vector The terms for displacement function is
selected symmetrically from the pascal
u i  triangle to maintain geometric isotropy
d i  

v i 
Element displacement vector
ui  
v 
i  ü

di  u j   

d    ü
j  v j 
de    ü ü


d k  u k  


d l 
 v  ü  

k  ü
ü
u l  ü


v  ü

l  ü


ü


 
ü   
ü
ü 

ü

 ü

ü ü
ü
 ü



›ascal triangle for 3D problems

 25 
21 
2 
22  
( ,
)  €1

 u( ,
) €1

 6
23 27 
24  28 


21 
2 
2
23 
 
u(x, y)  1 x y xy 0 0 0 0  24 
(x, y)  0 0 0 0 1 x y xy  2 (1a)

v(x, y)   5
26 
 
27 


28 
1 x y xy 0 0 0 0
(x, y) 0 0 0 0  2 (1b)
 1 x y xy 
Using the nodal conditions like
x ß x i , y ß yi u(x, y) ß u i , v(x, y) ß v i
x ß x j, y ß y j u(x, y) ß u j , v(x, y) ß v j
(2)
x ß x k , y ß yk u(x, y) ß u k , v(x, y) ß v k
x ß x l , y ß yl u(x, y) ß u l , v(x, y) ß v l

This results in as many conditions as the number of

unknown constants.
Using nodal boundary condition listed in eq. (2) in eq. 1a, following matrix
eqn. Can be obtained
1  1 1
1

1 1 0 0 0 0  21 
u  0 0 0 0 1
1
 2 
1   1 1 1
2
2  1 2
2 2
2 0 0 0 0  23 
    
u2  0 0 0 0 1
2 2 2 

24 
 ß
2

3  1 3
3 3
3 0 0 0 0   25 
u3  0 0 0 0 1
3

 
3 3  26
  
3
 
4  1 4
4 4
4 0 0 0 0  27 
    

4  0
u 0 0 0 1 4
4 4 4


28 

a ß € 2 
2 ß € a      (3)
1

For quadrilateral element [A] is of size 8 X 8

 1
substituting 2 ß €A d in eq. 1b
1 x y xy 0 0 0 0  1
 (x, y)ß   €A d
 0 0 0 0 1 x y xy  2X8 8X8
U(x, y)ß €N(x, y) d
2X8

N(x, y) ß is called displacement function

or interpolation function
or Shape function
 u i 
vi
i , 0, , 0, k , 0, l, 0 
(x, y)    u
0, i , 0, , 0, k , 0, l 
.

. 
Properties o shape unction
i  1.0 at node 'i', and zero at all other nodes
i  0 at all the sides on hich node o interest does not all.
n
 i  1.0, n number o nodes per element
i 1
 
  

 
  
 
For one dimensional element   
ù     ( )
  
For two dimensional element
    
   

ü   ü  ü

     
 
      

      


ü       

    

 ü    ü 
ƒ 

ğ  


ü ƒ

   

ƒ 
ƒ 


  
 r    

 r     

    



 
 

  r  r

  

r  
r  
¦  
 
 
 
    
 
  

!irtual work principle of a deformable body in equilibrium is subjected to arbitrary
virtual displacement satisfying compatibility condition (admissible displacement), then
the virtual work done by external(loads will be equal to virtual strain energy of internal
stresses. ó Y  ó
Y
Y

 YY  Y 
 ƒ ƒ 


Y


Y ƒ 
ƒ ƒ   ƒ
 
Y Y
 Y 
 ƒ 
ƒ ƒ   ƒ 


Y

 Y 
 ƒ
 ƒ 

ƒ
 ƒ  


Y
Y

ƒ ƒ  


Y

ƒ 
ƒ   ƒ 
 ƒ
 Y 
 ƒ

 
ƒ 


Y


 ƒ

 ƒ  

 ƒ
 ƒ 


Y
Y
 O

 Oü 

 Oü 

 ü 

ü ü






 ü
ü ü 


 
 


 
 



Y
 
YY
š


  
ƒ 
 
Y
 ƒ   
ƒ
    
Y Y  

Y
 
Y
YY
š


  
ƒ
 
Y
 ƒ   
ƒ
    
  

Y
 
Y 
Y

 Y 
ƒ ƒ
Y   Y ƒ  ƒ
   
    
For equilibrium internal virtual work = external virtual work

 d (  €B €D €B dv d-  €B €D 0 dv

T T T

ve ve

  €B 0 dv) ß
T

ve

bx  px  
 d   €N
T
 e
T
 dv   €N
T

 dv  › 
e

v
b y  s
p y  
€ e
K d e

ß f e 
where

 €B €D €B dv ß Element stiffness matrix

T
€K e ß
ve

f e ß Total nodal force vector

 here

€ € €
T
€ e  dv
ve

 e   € € 0 dv   €  0 dv)

T T

ve ve

b x  p x 
€
T
 dv €
T
 dv 
Pe
ve
b y  s
p y 
First term in { e} is equivalent nodal orce vector due to initial
strain. Second term is equivalent nodal orce vector due to
initial stress. Third term is equivalent nodal orce vector due to
body orce. Fourth term is equivalent nodal orce vector due to
sur ace traction. Last term is applied concentrated load vector