Isoparametric formulation

Lecture Notes
Hari Ram Parajuli, Dr. Eng
Element discretization
 Quadrilateral Elements
The term ‘Isoparametric’ is derived from the use of same shape function to
define element shape as well as displacement field.
η
(-1, 1) (1, 1) (x4,y4) 3 (x3,y3)
y 4
4 3

1
(x1,y1) 2
O (0, 0) ξ
(x2,y2)
1 x
2
(-1, 1) (1, -1)
Parent element Quadrilateral element

Displacement field Coordinate transformation

4 4
u = ∑Nu
i i
v = ∑ Ni vi 4
x = ∑Ni xi
4
y = ∑Ni yi
i=1 i =1
i=1 i=1
 Shape Functions
η
(-1, 1) (1, 1)
1
N1 = (1−ξ )(1−η)
4 3

4
O (0, 0) ξ
1
N2 = (1+ξ )(1−η) 1 2
4 (-1, 1) (1, -1)

1
N3 = (1+ξ )(1+η) v4
v3
4 u3
u4 v
u
1
N4 = (1−ξ )(1+η)
Y v1
u1 v2
4
X u2
 Displacement Field and Mapping
Displacements (u, v) in a plane element are interpolated from
nodal displacements (ui, vi) using shape functions Ni as follows

Displacements Coordinate transformation

u1   x1 
v  y 
 1  1
u2  x2 
   
u N1 0 N2 0 N3 0 N4 0  v2  x N1 0 N2 0 N3 0 N4 0  y2 
 =     =  
v  0 N1 0 N2 0 N3 0 N4  u3  y  0 N1 0 N2 0 N3 0 N4  x3 
v3   y3 
   
u4  x4 
v  y 
 4  4
 Stiffness Matrix
1
U= ∫ { } {ε }dV
σ
T

2V
But, {σ } = [ D ]{ε } {ε } = [ B ]{δ }

U = ∫ ([ D][ B]{δ }) [ B]{δ }dV

1 T

2V
1 T  
U = {δ }  ∫ [ B] [ D][ B] dV  {δ }
T

2 V 
1 T
U = {δ } [ k ]{δ } [ k ] = ∫ [ B] [ D][ B] dV
T
where,
2 V

[ k ] = t ∫ [ B] [ D][ B] dxdy
T
For an element with constant thickness t
A
 Strain Displacement Relationship
 ∂u 
  v3
εx   ∂x 
v4
u3
u4 v
   ∂v 
εy  =   u
γ   ∂y 
Y v1
v2
u1
 xy   ∂ u ∂v 
 +  X u2
 ∂y ∂x 
∂N1 ∂N2 ∂N3 ∂N4
{ε x } = u1 + u2 + u3 + u4
∂x ∂x ∂x ∂x
∂N1 ∂N 2 ∂N 3 ∂N 4
{ε y } = ∂y v1 + ∂y v2 + ∂y v3 + ∂y v4
∂N1 ∂N1 ∂N2 ∂N2 ∂N3 ∂N3 ∂N4 ∂N4
{γ xy} = ∂y u1 + ∂x v1 + ∂y u2 + ∂x v2 + ∂y u3 + ∂x v3 + ∂y u4 + ∂x v4
 Jacobian Matrix
Since the coordinates x and y are function of ξ and η
f → u (or )v → f ( x, y ) and f = f  x (ξ ,η ) , y (ξ ,η ) 
Using chain rule
∂f ∂f ∂x ∂f ∂y
= + ,
∂ξ ∂x ∂ξ ∂y ∂ξ

∂f ∂f ∂x ∂f ∂y
= +
∂η ∂x ∂η ∂y ∂η

 ∂f   ∂x ∂y   ∂f   ∂f 
 ∂ξ   ∂ξ 
   ∂ξ  ∂ x   ∂ x 
 =   ∂f  = [J ] ∂f 
 ∂f   ∂x ∂y     
 ∂ η   ∂ η ∂η 
  ∂ y   ∂ y 
 Jacobian Matrix
 ∂x ∂y 
 ∂ξ ∂ξ  J J 12 
[J] is Jacobian matrix [ J ] =   =  11
 ∂x ∂y   J 21 J 2 2 
 ∂η ∂η 
 

 ∂f   ∂f   ∂f 
 ∂ x    − J 2 1   ∂ ξ 
 ∂ξ  1  J 22 
 = [J ]
−1
 ∂f  = − J   
   ∂f  d e t [ ]  12
J J 11   ∂ f 
 ∂ y   ∂ η   ∂ η 

d xd y = d et [ J ]d ξ dη
Please refer to
Chandrapatla and Belegundu,
Introduction to Finite Elements in
Engineering -Appendix
 Jacobian Matrix
In this case f is u or v thus, formation of Jacobian matrix is as follows
f = u  x (ξ ,η ) , y (ξ ,η ) 
Using chain rule
∂u ∂u ∂x ∂u ∂y
= + ,
∂ξ ∂x ∂ξ ∂y ∂ξ

∂u ∂u ∂x ∂u ∂y
= +
∂η ∂x ∂η ∂y ∂η

 ∂u   ∂x ∂y   ∂u   ∂u 
 ∂ξ   ∂ξ 
   ∂ξ  ∂ x   ∂ x 
 =   ∂u  = [J ] ∂u 
 ∂u   ∂x ∂y     
 ∂ η   ∂ η ∂η 
  ∂ y   ∂ y 
 Jacobian Matrix
Similarly
f = v  x (ξ ,η ) , y (ξ ,η ) 
Using chain rule
∂v ∂v ∂x ∂v ∂y
= + ,
∂ξ ∂x ∂ξ ∂y ∂ξ

∂v ∂v ∂x ∂v ∂y
= +
∂η ∂x ∂η ∂y ∂η

 ∂v   ∂x ∂y   ∂v   ∂v 
 ∂ξ   ∂ξ 
   ∂ξ  ∂ x   ∂ x 
 =   ∂v  = [J ] ∂v 
 ∂v   ∂x ∂y     
 ∂ η   ∂ η ∂η 
  ∂ y   ∂ y 
 Jacobian Matrix
1
[J] is Jacobian matrix N1 = (1 − ξ )(1 − η )
4
 ∂x ∂y  1
N 2 = (1 + ξ )(1 − η )
 ∂ξ ∂ξ  J J 12 
4

[J ] =  ∂x ∂y
 =  11
J 2 2 
1
N 3 = (1 + ξ )(1 + η )
   J 21 4
1
 ∂η  N 4 = (1 − ξ )(1 + η )
 ∂η  4

∂x 4 ∂N 1
J 11 = =∑ i
=  − (1 − η ) x1 + (1 − η ) x 2 + (1 + η ) x3 − (1 + η ) x 4 
∂ ξ i =1 ∂ ξ 4
∂y 4 ∂N i 1
J12 = =∑ =  − (1 − η ) y1 + (1 − η ) y2 + (1 + η ) y3 − (1 + η ) y4 
∂ξ i =1 ∂ξ 4
∂x 4 ∂N 1
J 21 = =∑ i
=  − (1 − ξ ) x1 − (1 − ξ ) x2 + (1 + ξ ) x3 − (1 − ξ ) x4 
∂η i =1 ∂η 4
∂y 4 ∂N 1
J 22 = =∑ i
=  − (1 − ξ ) y1 − (1 − ξ ) y 2 + (1 + ξ ) y3 + (1 − ξ ) y 4 
∂η i =1 ∂η 4
 Strain Displacement Relationship
 ∂u  v3
v4
 
εx   ∂x  u4 v
u3

   ∂v  u
ε =
 y    Y v1
γ   ∂y  u1 v2
 xy   ∂ u ∂v  X u2
 + 
 ∂y ∂x 
 ∂u   ∂u 
 ∂ x  1  J 22 − J 1 2   ∂ ξ 

 ∂u = − J  
  d e t [ ]  21
J J 1 1   ∂ u 
 ∂ y   ∂ η 
 ∂v   ∂v 
 ∂ x  1  J 22 − J 1 2   ∂ ξ 

 ∂v = − J   
  d e t [ ]  21
J J 11   ∂ v 
 ∂ y   ∂ η 
 Strain Displacement Relationship

 ∂u 
   ∂ξ 
∂u  
 
∂x  ∂u 
εx     J 22 − J 12 0 0   ∂η 
   ∂v  1   
εy  =  =  0 0 − J 21 J 1 1   
γ   ∂y  J
 − J 2 1 J 11 J 22 − J 1 2   ∂v 
 xy   ∂ u ∂v   ∂ξ 
 +   
 ∂y ∂x   ∂v 
 ∂ η 
 Strain Displacement Relationship
 ∂u 
 ∂ξ 
 
 ∂u 
 ∂η 
 
{ } [ ]
ε = A 
 ∂v 
 ∂ξ 
 
 ∂v 
 ∂ η 

 J 22 − J 12 0 0 
1 
[ ]
A =  0 0 − J 21 J 1 1 
J
 − J 2 1 J 11 J 22 − J 1 2 
 Strain Displacement Relationship
 ∂u 
 ∂ξ 
  {ε } = [ B ]{δ }
 ∂u 
 ∂η


 [ B ] = [ A ][G ]
  = [G ]{δ }
 ∂v 
 ∂ξ 
  ∂u  ∂N1 ∂N2 ∂N3 ∂N4  u1 
 ∂v  ∂ξ   ∂ξ 0 ∂ξ 0 ∂ξ 0 ∂ξ 0  v1 
 ∂ η      
∂u  ∂N1 ∂N2 ∂N3 ∂N4  u2 
∂η  ∂η 0 ∂η 0 ∂η 0 ∂η 0  v 
     2
 =  
∂
  0
v ∂N1 ∂ N2 ∂N3 ∂N4  u3 
 0 0 0
∂ξ  ∂ξ ∂ξ ∂ξ ∂ξ  v3 
    
 ∂v   0 ∂N1 0 ∂N2 0 ∂N3 0 ∂N4  u4 
∂η  ∂η ∂η ∂η ∂η  v4 
 Strain Displacement Relationship
 ∂u 
 ∂ξ 
  {ε } = [ B ]{δ }
 ∂u 
 ∂η 



 = [G ]{δ } [ B ] = [ A ][G ]
 ∂v 
 ∂ξ 
 
 ∂v 
 ∂ η 
−(1−η) 0 (1−η) 0 (1+η) 0 −(1+η) 0 
 
1 ( )
− 1−ξ 0 −( )
1+ ξ 0 ( )
1+ ξ 0 ( )
1− ξ 0
[G] =  0 − 1−η 0 
4 ( ) (1−η) 0 (1+η) 0 −(1+η)
 
 0 −(1−ξ ) 0 −(1+ξ) 0 (1+ξ ) 0 (1−ξ) 
 Stiffness Matrix
1
U= ∫ { } {ε }dV
σ
T

2V
But, {σ } = [ D ]{ε } {ε } = [ B ]{δ }

U = ∫ ([ D][ B]{δ }) [ B]{δ }dV

1 T

2V
1 T  
U = {δ }  ∫ [ B] [ D][ B] dV  {δ }
T

2 V 
1 T
U = {δ } [ k ]{δ } [ k ] = ∫ [ B] [ D][ B] dV
T
where,
2 V

[ k ] = t ∫ [ B] [ D][ B] dxdy
T
For an element with constant thickness t
A
1 1
[ k ] = t ∫ ∫ [ B] [ D][ B] J dξ dη
T

−1 −1
 Numerical Integration
One dimensional Integral
1 1
I = ∫ f {ξ } dξ = ∫ φ dξ
−1 −1
Gaussian quadrature approach to evaluate the integral is given by
1
I = ∫ φ dξ = W1φ1 + W2φ2 + W3φ3 + .......... + Wnφn
−1

where, W1, W2, W3 are weights

For two dimensional Integral,

Numerical Integration
 Numerical Integration
For two dimensional Integral,
 Example
See Matlab program and calculate stiffness matrix
 Constant Strain Triangle (CST)
Natural coordinate system

Shape function for N1

 Constant Strain Triangle (CST)
In isoparametric representation

4
x = ∑Ni xi
i=1

4
y = ∑Ni yi
i=1
 Constant Strain Triangle (CST)
Jacobian Matrix
Displacements u and v can be expressed in terms of (x, y) or (ξ, η). Using
chain rule of derivatives

[J] is Jacobian matrix

Constant Strain Triangle (CST)
 Constant Strain Triangle (CST)
Natural coordinate system
 ∂u 
 
εx   ∂x 
   ∂v  [ε ] = [ B]{δ }
εy  =  
γ   ∂y 
 xy   ∂ u ∂v 
 + 
 ∂y ∂x 

Where,
xij = xi - xj
yij = yi - yj
Axisymmetric Element
Axisymmetric Element
Axisymmetric Element
Axisymmetric Element
Axisymmetric Element
 Strain Displacement Relationship
Displacement field

u (r , z ) = ∑ N i (r , z )ui
m

i =1

w (r , z ) = ∑ N i (r , z ) wi
m

i =1

Strain components for 3 node triangular element

∂u 3 ∂N
εr = = ∑ ui
∂r i =1 ∂ r

u 3 N
εθ = = ∑ i ui
r i =1 r

∂w 3 ∂N
εz = = ∑ i
wi
∂z i =1 ∂ z

∂u ∂w 3 ∂N 3 ∂N
γ xy = + = ∑ i
ui + ∑ i
wi
∂z ∂r i =1 ∂ z i =1 ∂ z
 Strain Displacement Relationship

 ∂N 1 ∂N 2 ∂N 3 
 ∂r 0 0 0   u1 
∂r ∂r u 
 ε r   N N2 N3

  2
ε   1 0 0 0
  u 3 
 θ  r r r
 =
    
ε
 z   0 ∂N 1 ∂N 2 ∂N 3  w
 1
0 0
 γ r z   ∂z ∂z ∂z   w2 
 ∂N ∂N 1 ∂N 2 ∂N 2 ∂N 3 ∂N 3   
 1
  w 3 
 ∂z ∂r ∂z ∂r ∂z ∂r 
 Nodal Load Vector
 f r( p )   pr 
{f }
( p)
=  ( p )  = 2π ∫ [ N ]
T
  rdS
 fz  S  pz 

Stiffness Matrix

[k ] = ∫∫∫
V
([ B ] T
[ D ][ B ])d V

[ k ] = 2 π ∫∫
A
( [B ]
T
[ D ][ B ])r d r d z