MANE 4240 & CIVL 4240

Introduction to Finite Elements

Prof. Suvranu De

Four-noded
rectangular element
Reading assignment:

Logan 10.2 + Lecture notes

Summary:

• Computation of shape functions for 4-noded quad

• Special case: rectangular element
• Properties of shape functions
• Computation of strain-displacement matrix
• Example problem
•Hint at how to generate shape functions of higher order
(Lagrange) elements
Finite element formulation for 2D:

Step 1: Divide the body into finite elements connected to each

other through special points (“nodes”)
py
3 px v3
4 3
u3
v4 u 1 
v 2 Element ‘e’ v2 v 
1  1
4 u 2 
u ST u4
v1 2 u2  
v 2 
y d 
x y u3 
Su 
u1 v 3 
1
v  
x x u 4 
u v 
 4
Summary: For each element
Displacement approximation in terms of shape functions
uNd

Strain approximation in terms of strain-displacement matrix

εBd
Stress approximation
  DB d
Element stiffness matrix
k   e B D B dV
T
V

Element nodal load vector

f   e N X dV   e N T S dS
T T

V     ST    
f f
b S
 Constant Strain Triangle (CST) : Simplest 2D finite element

v1
v3
1 u1
(x1,y1) (x3,y3)
v2 v u3
y u 3
(x,y)
u (x, y)  N1(x, y) u1  N 2(x, y) u 2  N 3(x, y) u 3
u2
2 (x2,y2) v (x, y)  N1(x, y) v1  N 2(x, y) v 2  N 3(x, y) v3
x
• 3 nodes per element
• 2 dofs per node (each node can move in x- and y- directions)
• Hence 6 dofs per element
 Formula for vthe shape functions are
a1  b1 x  c1 y
1
v3 N1 
1 u1 2A
(x1,y1) (x3,y3) a2  b2 x  c2 y
N2 
v2 v u3 2A
y u 3 a  b x  c3 y
(x,y)
N3  3 3
u2 2A
where 2 (x2,y2)

1 x1 y1 
1
A  area of triangle  det 1 x 2 y2 
2
1 x 3 y3 
a1  x2 y3  x3 y2 b1  y2  y3 c1  x3  x2
a2  x3 y1  x1 y3 b2  y3  y1 c2  x1  x3
a3  x1 y2  x2 y1 b3  y1  y2 c3  x2  x1
Approximation of displacements
 u1 
uNd v 
 1
u (x, y)  N1 0 N2 0 N3 0  u 2 
u    
 v (x, y)  0 N1 0 N2 0 N 3  v 2 
u 3 
Approximation of the strains  
 v 3 
 u 
 x 
 x   
   v 
   y    y   Bd
   
 xy   u v 
 y  
 x 
 N1(x, y) N 2(x, y) N 3(x, y) 
 0 0 0 
  x x x  b1 0 b2 0 b3 0 
N1(x, y) N 2(x, y) N 3(x, y)  1 
B 0 0 0   0 c1 0 c2 0 c3 
 y y y  2 A
 N (x, y) N (x, y) N 2(x, y) N 2(x, y) N 3(x, y) N 3(x, y)  c1 b1 c2 b2 c3 b3 
 1 1

 y x y x y x 
Element stiffness matrix
t
k   e B D B dV
T
V

Since B is constant
A

k  B D B e dV  B D B At
T T t=thickness of the element
V A=surface area of the element

Element nodal load vector

f   e N X dV   e N T S dS
T T

V     ST    
f f
b S
Class exercise
For the CST shown below, compute the vector of nodal loads due to surface traction

  e N T S dS
T
f S ST

1
f S  t
T
e
N T S dS
l13 along 2 3
y fS3y
fS2y fS2x fS3x
2 3 x
(0,0) py=-1 (1,0)
 Class exercise
f S  t
T
e
N T S dS
l13 along 2  3
1
0
TS   
y fS3y  1
fS2y fS2x fS3x The only nonzero nodal loads are
1
2 3 x f S2 y  t  N 2 along 23 p y dx
x1 0
py=-1 f S3 y  t  N 3 along 23 p y dx
x 0

 a  b x  c2 y  a2  b2 x  x3 y1  x1 y3    y3  y1  x
N 2 along 23   2 2   
 2A  y 0 2A 2A
y1  y1 x y1 (1  x) y1 (1  x)
  
1 x1 y1  1 x1 y1  y1 ( x3  x2 )
det 1 x 2 y2  det 1 x 2 0 
1 x 3 y3  1 x 3 0 
 1 x (can you derive this simpler?)
 1
 f S2 y  t  N 2 along 23 p y dx
x 0
1
 t  (1  x)(1) dx
x 0

t

2

Now compute

1
f S3 y  t  N 3 along 23 p y dx
x 0
 4-noded rectangular element with edges parallel to the
coordinate axes:

(x4,y4) 4 3 (x3,y3)
4
u (x, y)   N i (x, y)ui
v i 1
u 2b 4
(x,y) v (x, y)   N i (x, y)vi
y i 1

1 2a 2 (x2,y2)
(x1,y1)

• 4 nodes per element

• 2 dofs per node (each node can move in x- and y- directions)
•8 dofs per element
 Generation of N1: x  x2
At node 1 l1 ( x) 
y x1  x 2

has the property

3 l1(y) l1 ( x1 )  1
4
l1 ( x 2 )  0

2b Similarly y  y4
l1 ( y ) 
y1  y 4
N1 1 2 1 has the property
2a x l1 ( y1 )  1
l1(x) l1 ( y 4 )  0
1
Hence choose the shape function at node 1 as

 x  x2  y  y 4  1
N 1  l1 ( x)l1 ( y )       x  x 2  y  y 4 
 x1  x 2  y1  y 4  4ab
Using similar arguments, choose
1
N1   x  x 2  y  y 4 
4ab
1
N2    x  x1  y  y3 
4ab
1
N3   x  x 4  y  y 2 
4ab
1
N4    x  x3  y  y1 
4ab
 Properties of the shape functions:
1. The shape functions N1, N2 , N3 and N4 are bilinear functions
of x and y
2. Kronecker delta property

 1 at node ' i '

N i ( x, y )  
0 at other nodes
3. Completeness
4

N
i 1
i 1
4

N x
i 1
i i x
4

N
i 1
i yi  y
3. Along lines parallel to the x- or y-axes, the shape functions
are linear. But along any other line they are nonlinear.

4. An element shape function related to a specific nodal point is

zero along element boundaries not containing the nodal point.

5. The displacement field is continuous across elements

6. The strains and stresses are not constant within an element

nor are they continuous across element boundaries.
 The strain-displacement relationship
x 
 
   y 
 xy 
 
u 1 
v 
 N 1(x, y) N 2 (x, y) N 3 (x, y) N 4 (x, y)  1 
 0 0 0 0  u 2 
  x  x  x  x  
N (x, y) N (x, y) N (x, y) N (x, y)  v 2 
 0 1
0 2
0 3
0 4
 
 y y y y  u 3 
 N (x, y) N (x, y) N (x, y) N 2 (x, y) N 3 (x, y) N 3 (x, y) N 4 (x, y) N 4 (x, y)  v 
 1 1 2
 3
 y x y x y x y x  u 
                                         4
B v 
 4
 y  y4 0 y3  y 0 y  y2 0 y1  y 0 
1  
B 0 x  x 0 x  x 0 x  x 0 x  x
4ab  
2 1 4 3

 x  x 2 y  y 4 x1  x y3  y x  x 4 y  y 2 x3  x y1  y 

Notice that the strains (and hence the stresses) are NOT constant within an element
 Computation of the terms in the stiffness matrix of 2D elements (recap)
 
v3  
v4
4
u4
3 The B-matrix (strain-displacement) corresponding to this element is
  u3
 
  u1 v1 u2 v2 u3 v3 u4 v4
v  
     N1 (x,y) N 2 (x,y) N 3 (x,y) N 4 (x,y) 
y    0 0 0 0 
u v2  x x x x 
v1 (x,y)    N1 (x,y) N 2 (x,y) N 3 (x,y) N 4 (x,y) 
   0 0 0 0 
 y y y y 
 
1 u1 2 u2  N1 (x,y) N1 (x,y) N 2 (x,y) N 2 (x,y) N 3 (x,y) N 3 (x,y) N 4 (x,y) N 4 (x,y) 
   
   y x y x y x y x 
x

We will denote the columns of the B-matrix as

 
 
   
 N
  1 (x,y)   
    0
 x  
    N1 (x,y) 
B u1     0 B v1  
; ; and so on...
 y 
 N (x,y)   
 1   N (x,y) 
 y  
1

x
 The stiffness matrix corresponding to this element is

k   e B D B dV
T
which has the following form
V

u2 v2 u3 v3 u4 v4
u1 v1

 k11 k12 k13 k14 k15 k16 k17 k18  u1

k k 22 k 23 k 24 k 25 k 26 k 27 k 28  v1
 21 
 k31 k 32 k 33 k 34 k 35 k 36 k 37 k 38  u2
 
k k 42 k 43 k 44 k 45 k 46 k 47 k 48  v2
k   41
 k51 k52 k53 k54 k 55 k 56 k 57 k 58  u3
 
 k61 k 62 k 63 k 64 k 65 k 66 k 67 k 68  v3

k k 72 k 73 k 74 k 75 k 76 k 77 k 78  u4
 71 

 k81 k82 k83 k84 k85 k86 k87 k 88  v4

The individual entries of the stiffness matrix may be computed as follows

 Bu1 D Bu1 dV; k12   e Bu1 D Bv1 dV; k13   e Bu1 D Bu2 dV,...
T T T
  k11  e
V V V

k21   e Bv1 D Bu1 dV; k21   e Bv1 D Bv1 dV;.....

T T
V V
  Notice that these formulae are quite general (apply to all kinds
of finite elements, CST, quadrilateral, etc) since we have not
used any specific shape functions for their derivation.
 
 
 Example 1000 lb
y 300 psi

4 3
Thickness (t) = 0.5 in
2 in E= 30×106 psi
=0.25
1 2
x
3 in

(a) Compute the unknown nodal displacements.

(b) Compute the stresses in the two elements.

This is exactly the same problem that we solved in last class, except
now we have to use a single 4-noded element
Realize that this is a plane stress problem and therefore we need to use

 
 1  0  3.2 0.8 0 
E 0.8 3.2 0  107 psi
D  1 0  
1  2  1    
0 0   0 0 1.2
 2 

Write down the shape functions

x y
N1 
1
 x  x2  y  y4   ( x  3)( y  2) 0 0
4ab 6
N2  
1
 x  x1  y  y3    x( y  2) 3 0
4ab 6
3 2
1
N3   x  x4  y  y2   xy 0 2
4ab 6
N4  
1
 x  x3  y  y1    ( x  3) y
4ab 6
We have 4 nodes with 2 dofs per node=8dofs. However, 5 of these are fixed.
The nonzero displacements are
u2 u3 v3

Hence we need to solve

u2 u3 v3
 k11 k12 k13  u 2   0 
k    
 21 k 22 k 23  u3    0 
 k31 k32 k33   v  f 
   3y 
3

Need to compute only the relevant terms in the stiffness matrix

k11   e Bu2 D Bu2 dV; k12   e Bu2 D Bu3 dV; k13   e Bu2 D Bv3 dV
T T T
V V V

k21   e Bu3 D Bu2 dV; k22   e Bu3 D Bu3 dV; k13   e Bu3 D Bv3 dV
T T T
V V V

k31   e Bv3 D Bu2 dV; k22   e Bv3 D Bu3 dV; k13   e Bv3 D Bv3 dV
T T T
V V V
Compute only the relevant columns of the B matrix
 N 2   (2  y ) 
 x   6 
   
B u2  0  0 
 N 2    x 
 y   6 
   

 N 3   y 
 x   6 
   
B u3   0   0
 N 3   x 
 y   6 
   
   
 0  0
 N   x 
   
B v3   3    
 y   6 
 N 3   y 

 x   
6 
 k11   e Bu2 D Bu2 dV
T
V
3 2
 2 y 5 2
 0.5    (0.1067  10 8
 0.533  10 7
)( )  3.33  10 x  dxdy
x 0 y 0 
6 
 0.656  107

Similarly compute the other terms

How do we compute f3y

f 3 y  1000  f S3 y  xy  x
N 3 edge  N 3 y 2
  
3
4 3
 6  y 2 3
f S3 y  t  N 3 along (300) dx
x 0 edge 3 4

x 3
 (0.5)(300)  dx 4 3
x 0 3

3
 150 
2
 225 lb

 f 3 y  1000  f S3 y  1225 lb
How about a 9-noded rectangle?
Corner nodes
 x(a  x)   y (b  y )   x(a  x)   y (b  y ) 
y 5 N1  
 2 a 2 
 2 b 2 

N 2  


2a 2   2b 2 
2 a a 1  x(a  x)   y (b  y )   x(a  x)   y (b  y ) 
N 3       N     
b 2b 2 
2 2 4 2
 2 a  2b   2 a 
9 8
6 x
b Midside nodes
7  a 2  x 2   y (b  y )   x(a  x)   b  y 
2 2
3 4 N5     N 6    
 a 2
 2b 2
  2a 2   b 2 
 a 2  x 2   y (b  y )   x( a  x)   b  y 
2 2
N7      N8    
 2a   b
2 2 2 2
 a  2 b  

Center node
 a 2  x2   b2  y 2 
N9   2  2 
 a   b 

Question: Can you generate the shape functions of a 16-noded rectangle?

Note: These elements, whose shape functions are generated by multiplying the
shape functions of 1D elements, are said to belong to the “Lagrange” family