CIV 4234

Advanced Structural Analysis

 Isoparametric plate element

1 1
ξ=1 η=1
B

ξ=-1 η=1 4 3
P
Y 1
(ξp ηp)
ξ

1
1 A 2

ξ=-1 η=-1 ξ=1 η=-1

X
xP =
1
(1 − ξ)(1 − η)x 1 + 1 (1 + ξ)(1 − η)x 2 + 1 (1 + ξ)(1 + η)x 3 + 1 (1 − ξ)(1 + η)x 4
4 4 4 4

yP =
1
(1 − ξ )(1 − η)y 1 + 1 (1 + ξ)(1 − η)y 2 + 1 (1 + ξ )(1 + η)y 3 + 1 (1 − ξ )(1 + η)y 4
4 4 4 4

This suggests the form of the four shape functions for the isoparametric plate
element as:
u = N1 u1 + N2 u2 + N3 u3 + N4 u4
v = N1 v1 + N2 v2 + N3 v3 + N4 v4
where,
N1 = ¼ (1 - ξ) (1 - η)
N2 = ¼ (1 +ξ) (1 - η)
N3 = ¼ (1 +ξ) (1 +η)
N4 = ¼ (1 - ξ) (1 +η)
Substituting the values of ξ and η

node 1: ξ = -1 η = -1 N1 = 1 N2 = 0 N3 = 0 N4 = 0
node 2: ξ=1 η = -1 N1 = 0 N2 = 1 N3 = 0 N4 = 0
node 3: ξ=1 η=1 N1 = 0 N2 = 0 N3 = 1 N4 = 0
node 4: ξ = -1 η=1 N1 = 0 N2 = 0 N3 = 0 N4 = 1
To summarise:

the displacements of the element can be expressed in terms of the shape

functions Ni,
4 
∑ i i 
u   i =1
N u
 =4 
  
v
Ni vi 
∑
 i =1 

The coordinates of the element can also be expressed in terms of the shape
functions Ni,

4 
∑ i i 
x   i =1
N x
 =4 
  
y
N i yi 
∑
 i =1 

If the element’s nodes define two things:

•Displacements {u v} of a point in the element,
•Global coordinates {x y} of a point in the element

where the shape function matrix [N] are the same, then the element is isoparametric
 Mapping

(x,y)

x  ξ 
 = f  
 y η 

We want a simpler way to formulate the equations for irregular shaped elements
 X4 Y4 η X3 Y3
η
(u4 v4) (u3 v3)
1 1
ξ=1 η=1 P (xp yp)
ξ=-1 η=1
4 P3
1
(ξp ηp) ξ Y (v) ξ

1
1 2
ξ=-1 η=-1 ξ=1 η=-1 X1 Y1
(u1 v1) X (u) X2 Y2
template/parent
element (u2 v2)
actual
element

x  ξ 
  = f  
 y η 
 X4 Y4 η X3 Y3
η
(u4 v4) (u3 v3)
1 1
ξ=1 η=1 P (xp yp)
ξ=-1 η=1
4 P3
1
(ξp ηp) ξ Y (v) ξ

1
1 2
ξ=-1 η=-1 ξ=1 η=-1 X1 Y1
(u1 v1) X (u) X2 Y2
template/parent
element (u2 v2)
actual
element

x  ξ 
  = f  
 y η 
 η
ξ

η
ξ
 η η
ξ 3

α
4

Y (v)
η β
ξ 2

X (u)
 η
ξ
η
4 7
3

8 L/3
η Y (v) 6
ξ ξ
1
2 L
5 L/3
X (u)
 y (x,y)
η
(-1,1) (1,1)
ξ
ξ

(-1,-1) (1,-1)

L2= 0 3
L2= 0 3
1 L1 = 0 L1 = 0

L 3= 0
2 1
2
Local co-ordinates L3= 0
(Parent)

Cartesian map x
 η η= 1 y η= 1
ζ η = -1

ζ
η
ξ ξ

η = -1

4
4
L1 = 0
3
1 L1 = 1
1
3
2
Local co-ordinates z
(Parent) 2

Cartesian map x
 Strain-displacement relationship

[ε(x,y)] = [B][U]

 ∂u  ∂  ∂ 
   0  0
ε x   ∂x   ∂x   ∂x 
   ∂v   ∂ u   ∂
{ε} = ε y  =   = 0   = 0  [N ]{δ}
   ∂y   ∂y  v   ∂y 
γ xy   ∂u ∂v  ∂ ∂ ∂ ∂
 +   ∂y ∂x   ∂y ∂x 
 ∂y ∂x   
 Strain-displacement relationship

[ε(x,y)] = [B][U]

 ∂u  ∂  ∂ 
   0  0
ε x   ∂x   ∂x   ∂x 
   ∂v   ∂ u   ∂
{ε} = ε y  =   = 0   = 0  [N ]{δ}
   ∂y   ∂y  v   ∂y 
γ xy   ∂u ∂v  ∂ ∂ ∂ ∂
 +   ∂y ∂x   ∂y ∂x 
 ∂y ∂x   
u 1 
v 
∂   1
 0 u 2 
 ∂x   
 ∂   N1 0 N2 0 N3 0 N4 0  v 2 
{ε} =  0   
 ∂y   0 N1 0 N2 0 N3 0 N 4  u 3 
∂ ∂ v 3 
 ∂y ∂x   
 u 4 
v 
 4
 ∂u ∂ ∂N i
εx = =
∂x ∂x
∑ N u
i i = ∑ ∂x
ui
.
∂v ∂ ∂N i
εy = =
∂y ∂y
∑ i i ∑ ∂y v i
N v =

∂u ∂v ∂N ∂N
γ xy = + = ∑ i u i + ∑ i vi
∂y ∂x ∂y ∂x

To evaluate these expressions we need to be able to

evaluate the partial derivatives ∂N i and ∂N i
∂x ∂y
Using the Chain Rule of Differentiation, write,

∂N i ∂N i ∂x ∂N i ∂y
= +
∂ξ ∂x ∂ξ ∂y ∂ξ
or in matrix form,
∂N i ∂N i ∂x ∂N i ∂y
= +
∂η ∂x ∂η ∂y ∂η

 ∂N i   ∂x ∂y   ∂N i   ∂N i 
 ∂ξ   ∂ξ ∂ξ   ∂x   ∂x 
 =  = [J ] ∂N  , where [J] is the Jacobian matrix.
 i   ∂x
∂N ∂y   ∂N i   i
 ∂η   ∂η ∂η   ∂y   ∂y 

(the Jacobian matrix is the matrix of all first-order partial derivatives of a function)

Invert the matrix equation to get,

 ∂N i   ∂N i   ∂N i   ∂N i 
 ∂x   ∂ξ   ∂ξ   Γ11 Γ12   ∂ξ 
 ∂N  = [J ]  ∂N  = [Γ ] ∂N  = 
−1
  , where [Γ] is the inverse of [J].
 i  i  i  Γ21 Γ22   ∂N i 
 ∂y   ∂η   ∂η   ∂η 
 The terms of [J] (and hence Γ) can be readily evaluated. For our plate shape functions
example,

x = N1 x1 + N2 x2 + N3 x3 + N4 x4
= {¼ (1 - ξ) (1 - η)}x1+{¼ (1 +ξ) (1 - η)}x2 +{¼ (1 +ξ) (1 +η)}x3 +{¼ (1 - ξ) (1 +η)}x4

Therefore,

 ∂x ∂y   1 x y1 
 1
 ∂ξ   4
− (1 − η ) 1
(1 − η) 1
(1 + η) − (1 + η) x 2
1
y 3 
[J] =  ∂∂xξ =
∂y   1
4 4 4


  − (1 − ξ ) − (1 + ξ )
1 1
(1 + ξ) 1 (1 − ξ )   x 3 y3 

 ∂η ∂η   4 4 4 4  x y4 
 4

The Jacobian Determinant |J| provides us with the relationship between the element of area
dx.dy and the corresponding area element dξ.dη. That is, dx.dy = |J| dξ.dη.
 ∂u ∂ ∂N i
εx = =
∂x ∂x
∑ N u
i i = ∑ ∂x
ui
.
∂v ∂ ∂N i
εy = =
∂y ∂y
∑ i i ∑ ∂y v i
N v =

∂u ∂v ∂N ∂N
γ xy = + = ∑ i u i + ∑ i vi
∂y ∂x ∂y ∂x

To evaluate these expressions we need to be able to

evaluate the partial derivatives ∂N i and ∂N i
∂x ∂y

 ∂N i   ∂N i   ∂N i ∂N i 
Γ + Γ
 ∂x   Γ11 Γ12   ∂ξ   ∂ξ 11 ∂η 12 
 ∂N  ==   = 
 i Γ21 Γ22   ∂N i   ∂N i Γ + ∂N i Γ 
 ∂y   ∂η   ∂ξ 21 ∂η 22 
 ∂N i  ∂N ∂N i 
εx = ∑ u i = ∑  i Γ11 + Γ12 u i
∂x  ∂ξ ∂η 
∂N i  ∂N ∂N i 
εy = ∑ v i = ∑  i Γ21 + Γ22  v i
∂y  ∂ξ ∂η 
∂N i ∂N i  ∂N ∂N i   ∂N ∂N i 
γ xy = ∑ ui + ∑ v i = ∑  i Γ21 + Γ22 u i + ∑  i Γ11 + Γ12  v i
∂y ∂x  ∂ξ ∂η   ∂ξ ∂η 

 ∂N 1 ∂N 2 ∂N 3 ∂N 4 
 ∂ξ 0 0 0 0 
∂ξ ∂ξ ∂ξ
 
0   ∂N 1 ∂N 2 ∂N 3 ∂N 4
 Γ11 Γ12 0 0 0 0 0 
 ∂η ∂η ∂η ∂η 
[B(ξ, η)] =  0 0 Γ21 Γ22  
∂N 1 ∂N 2 ∂N 3 ∂N 4 
Γ21 Γ22 Γ11 Γ12   0 0 0 0 
 ∂ξ ∂ξ ∂ξ ∂ξ 
 ∂N 1 ∂N 2 ∂N 3 ∂N 4 
 0 ∂η
0
∂η
0
∂η
0
∂η 

N1 = ¼ (1 - ξ) (1 - η)

∂N 1 ∂N 1
= − (1 − η) = − (1 − ξ )
1 1
∴
∂ξ 4 ∂η 4

N2 = ¼ (1 +ξ) (1 - η)

∂N 2 1 ∂N 2
= (1 − η) = − (1 + ξ )
1
∴
∂ξ 4 ∂η 4

N3 = ¼ (1 +ξ) (1 +η)
∂N 3 1 ∂N 3 1
∴ = (1 + η) = (1 + ξ )
∂ξ 4 ∂η 4

N4 = ¼ (1 - ξ) (1 +η)

∂N 4 ∂N 4 1
= − (1 + η) = (1 − ξ )
1
∴
∂ξ 4 ∂η 4
 Integration of the stiffness matrix

vol
[K ] = ∫ [B]T [D][B]dV

vol
[K ] = ∫ [B(ξ, η)]T [D][B(ξ, η)]dV

dV = t.dx.dy = t.|J|.dξ.dη

+1 +1
[K ] = ∫ ∫ [B(ξ, η)]T [D][B(ξ, η)].t. | J | .dξ.dη
−1 −1
 Numerical Integration Example
external node
Y internal node numbering system
element
numbering system numbering system

1 3 5 7
4 3 4 3 4 3
y y y
x x x
1 2 3

1 X
2 1 2 1 2

2 4 6 8
local element
global coordinate
coordinate system
system
 Numerical Integration Example
external node
Y internal node numbering system
element
numbering system numbering system

1 3 5 7
4 3 4 3 4 3
y y y
x x x
1 2 3

1 X
2 1 2 1 2

2 4 6 8
local element
global coordinate
coordinate system
system
Y
0.5773

-0.5773 3 5 7
1
4 3 4 3 4 3
η
0.5773
ξ 1 unit deep
2 3
-0.5773
X
1 2 1 2 1 2

2 4 6 8
Gauss point

3 units long
Evaluate for the first Gauss Point:
N1 = ¼ (1 - ξ) (1 - η) = ¼ (1 – (-0.5774)) (1 – (-0.5774)) = 0.6221
N2 = ¼ (1 +ξ) (1 - η) = ¼ (1 +(-0.5774)) (1 – (-0.5774)) = 0.1667
N3 = ¼ (1 +ξ) (1 +η) = ¼ (1 +(-0.5774)) (1 +(-0.5774)) = 0.0447
N4 = ¼ (1 - ξ) (1 +η) = ¼ (1 – (-0.5774)) (1 +(-0.5774)) = 0.1667

∂N 1 ∂N 1
= − (1 − η) = −0.3944 = − (1 − ξ ) = −0.3944
1 1
∴
∂ξ 4 ∂η 4

∂N 2 1 ∂N 2
= (1 − η) = +0.3944 = − (1 + ξ ) = −0.1057
1
∴
∂ξ 4 ∂η 4

∂N 3 1 ∂N 3 1
∴ = (1 + η) = +0.1057 = (1 + ξ ) = +0.1057
∂ξ 4 ∂η 4

∂N 4 ∂N 4 1
= − (1 + η) = −0.1057 = (1 − ξ ) = +0.3944
1
∴
∂ξ 4 ∂η 4
  ∂N 1 ∂N 2 ∂N 3 ∂N 4 
 ∂ξ 0 0 0 0 
∂ξ ∂ξ ∂ξ
 
0   ∂N 1 ∂N 2 ∂N 3 ∂N 4
 Γ11 Γ12 0 0 0 0 0 
 ∂η ∂η ∂η ∂η 
[B(ξ, η)] =  0 0 Γ21 Γ22  
∂N 1 ∂N 2 ∂N 3 ∂N 4 
Γ21 Γ22 Γ11 Γ12   0 0 0 0 
 ∂ξ ∂ξ ∂ξ ∂ξ 
 ∂N 1 ∂N 2 ∂N 3 ∂N 4 
 0 ∂η
0
∂η
0
∂η
0
∂η 


Substitute for the first Gauss Point

 −0.3944 0 +0.3944 0 +0.1057 0 −0.1057 0 
 2 0 0 0  
 − 0.3944 − 0.1057 + 0.1057 + 0.3944
[B(ξ, η)] = 0 0 0 2 
 0 0 0 0 
0 − 0.3944 0 + 0.3944 0 + 0.1057 0 − 0.1057 
0 2 2 0  
 0 − 0.3944 0 − 0.1057 0 + 0.1057 0 + 0.3944
2 × (− 0.3944) 0 2 × 0.3944 0 2 × 0.1057 0 2 × (− 0.1057 ) 0 
 
= 0 2 × (− 0.3944) 0 2 × (− 0.1057 ) 0 2 × 0.1057 0 2 × 0.3944 
2 × (− 0.3944) 2 × (− 0.3944) 2 × (− 0.1057 ) 2 × 0.3944 2 × 0.1057 2 × 0.1057 2 × 0.3944 2 × (− 0.1057 )
− 0.7888 0 0.7888 0 0.2114 0 − 0.2114 0 
 
= 0 − 0.7888 0 − 0.2114 0 0.2114 0 0.7888 
− 0.7888 − 0.7888 − 0.2114 0.7888 0.2114 0.2114 0.7888 − 0.2114

Repeat for Gauss Points 2,3,4.

To determine the stiffness matrix for element 1 we must evaluate:

2 2
k = h ∑∑ [B] [D][B] J H i H j
T

i =1 j=1

 B(ξ1η1 )T [D]B(ξ1η1 ) × 0.25 × 1 × 1 + B(ξ 2 η1 )T [D]B(ξ 2 η1 ) × 0.25 × 1 × 1 

 
k = h + 
 
( )T
[ ] ( ) ( )T
[ ] ( )
 B ξ 2 η 2 D B ξ 2 η 2 × 0.25 × 1 × 1 + B ξ1η 2 D B ξ1η 2 × 0.25 × 1 × 1

Jacobian
determinant


1 ν 0 
 Gauss quadrature

[D] = E 2 ν 1 0  weighting factors
1− ν  1− ν
0 0 
 2 

and repeat for each element....

 Interpretation and display

Exact average
40 shear stress

30 Nodal values extrapolated

from Gauss points
20

10

Gauss point
-10
values

1. You should only CALCULATE values at the Gauss points

(the FE program does this work anyway)

2, You should be cautious when extrapolating calculated values to the

nodes
Case Vertical σx at B σx at C
deflection at A

(a) 0.091 0.048 0.096

(b) 0.682 0.727 0.727 B A
(c) 0.494 0.604 0.301
P
(a)

A
C
P
(b)

A
C
P

(c)
 Case Vertical σx at B
deflection at
A
P
(d) 2x2 GP 0.968 1.000
(d) 3x3 GP 0.930 1.129
(e) 2x2 GP 0.366 0.051 B
(e) 3x3 GP 0.161 0.048 (d)
(f) 0.796
(g) 0.801 P

(e)

(f)

(g)