Scribd
Upload
120 views

Introduction To Fem

This document discusses higher-order isoparametric finite elements. It introduces serendipity elements, which have additional nodes along element boundaries to better represent geometry. Lagrange elements are also covered, which have additional interior nodes. The shape functions for these elements are constructed from Lagrange polynomials and satisfy interpolation conditions. Higher-order elements increase accuracy but also the problem size. The document provides examples of shape functions for various quad elements.

Uploaded by

Pavan Kishore
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
120 views

Introduction To Fem

This document discusses higher-order isoparametric finite elements. It introduces serendipity elements, which have additional nodes along element boundaries to better represent geometry. Lagrange elements are also covered, which have additional interior nodes. The shape functions for these elements are constructed from Lagrange polynomials and satisfy interpolation conditions. Higher-order elements increase accuracy but also the problem size. The document provides examples of shape functions for various quad elements.

Uploaded by

Pavan Kishore
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 21

Introduction to the Finite Element Method

Lecture 18

P.S. Koutsourelakis
[email protected] 369 Hollister Hall

November 22 2010
Last Updated: November 22, 2010

[email protected]

Lecture 18

Higher-order isoparametric elements

In many problems there might be a need to increase the number of shape functions (without increasing the number of elements) and/or represent more accurately the boundary of the domain. We discuss some higher order isoparametric elements that achieve this goal. There is no free lunch! Every time you increase the number of shape functions you will also increase the number of unknowns and the size of the system you need to solve.

[email protected]

Lecture 18

Serendipity Elements
= (x) (1, 1) 4 (1, 0) 8 1 (1, 1) 5 (0, 1) 7 3 6 2 (1, 1) (1, 0) (x e , y e )8 8 8 y
e e (x4 , y4 ) 4 e e (x7 , y7 ) e e (x3 , y3 )

7 e 5

3 6
e e (x6 , y6 )

1
e e (x1 , y1 )

(0, 1) (1, 1)

2 e e (x5 , y5 )

e e (x2 , y2 )

x x = x()

[email protected]

Lecture 18

Serendipity Elements
Serendipity elements consist of additional nodes along the boundary Observe the numbering! Let the mapping be dened as follows: x(, ) = y(, ) =
8 a=1 8 a=1 e Na (, )xa x() = e Na (, )ya 8

Na ()x e a
a=1

To obtain the functions Na we rst assume a form: x(, ) = b0 + b1 + b2 + b3 + b4 2 + b5 2 + b6 2 + b7 2 y (, ) = c0 + c1 + c2 + c3 + c4 2 + c5 2 + c6 2 + c7 2 where the parameters b and c should be determined by satisfying:
e x(a , a ) = xa e for a = 1, . . . , 8 y (a , a ) = ya

[email protected]

Lecture 18

Serendipity Elements
We obtain the following shape functions: N1 N2 N3 N4 N5 N6 N7 N8 = = = = = = = =
1 4 (1 1 4 (1 1 4 (1 1 4 (1 1 2 (1 1 2 (1 1 2 (1 1 2 (1

)(1 ) + )(1 ) + )(1 + ) )(1 + ) 2 )(1 ) + )(1 2 ) 2 )(1 + ) )(1 2 )

1 2 (N8 1 2 (N5 1 2 (N6 1 2 (N7

+ N5 ) + N6 ) + N7 ) + N8 )

Note that N1 arises from the linear shape function of the 4-node quad (which is equal to 1/2 at nodes 5 and 8) by subtracting 1/2 N5 and 1/2 N8 in order to become zero at nodes 5 and 8. Similarly for N2 , N3 , N4 .

[email protected]

Lecture 18

Serendipity Elements

[email protected]

Lecture 18

Serendipity Elements
To nd the local stiffness matrix k e : ke =
e 1 1

(B e )T D B e d (B e (, ))T D B e (, ) j(, ) d d
1 1

The strain displacement matrix B e :


e dx,1 e dy ,1 . . e dx,16 e dy ,16

= Be
31 316

d e = Be 161

[email protected]

Lecture 18

Serendipity Elements

Lets break B e in two parts:

ux x uy y ux y

= or:

+
uy x

y 1 , 0 j(, ) x,

y, 0 x,

0 x, y,

0 x, y,

ux ux uy uy

= Be 1

ux ux uy uy

[email protected]

Lecture 18

Serendipity Elements
and the second part:
or:
ux ux uy uy

N1, N1, = 0 0

0 0 N1, N1,
ux ux uy uy

... ... ... ...

... ... ... ...

N9, N9, 0 0

0 e 0 d N9, N9,

e e = B2 d

Hence: Be = Be 1
316

Be 2

34 416

[email protected]

Lecture 18

Serendipity Elements

When is the Jacobian determinant not zero?

7 4 8 1 L/4 5

3 6

2 L/4 acceptable position for node 5

[email protected]

Lecture 18

Serendipity Elements

Example: Find local force vector q4 4 8 5 7 3 q3

[email protected]

Lecture 18

Lagrange Elements
= (x) 3 4 8 1 9 5 6 2 y x x = x() 8 1 9 5 7 3 6 2
e e (x2 , y2 )

[email protected]

Lecture 18

Lagrange Elements

Lagrange elements consist of additional nodes in the interior of the element Observe the numbering!
Lagrange polynomials. Suppose one is given n pairs of values (i , i = (i )). The the function can be approximated as follows: L1 1 + L2 2 + . . . + L n n where: L1 = L2 = ... Ln =
(2 )(3 )...(n ) (2 1 )(3 1 )...(n 1 ) (1 )(2 )...(n ) (1 2 )(3 3 )...(n 3 ) (1 )(2 )...(n1 ) (1 n )(2 n )...(n1 n )

Note: Li (j ) = i,j

[email protected]

Lecture 18

Lagrange Elements

For example, for n = 2, 1 = 1, 2 = +1: L1 = L2 = For n = 3 and 3 = 0: L1 =


(3) L2 (3) L3 (3) (2 )(3 ) (2 1 )(3 1 ) (1 )(3 ) (1 2 )(3 2 ) (1 )(2 ) (1 3 )(2 3 ) 1 = 2 (1 ) 1 = 2 (1 + ) 1 = 2 (1 2 ) (2) (2) (2 ) (2 1 ) (1 ) (1 2 )

= 1 (1 ) 2 = 1 (1 + ) 2

= =

[email protected]

Lecture 18

Lagrange Elements
4 8 1 7 9 5 3 6 2

To generate the shape functions for plane isoparametric elements, it sufces to multiply the Lagrange polynomials w.r.t. to and of appropriate order. For example, for the 9 node element: N1 (, ) = L1 () L1 () = 1 (1 )(1 ) 4 (3) (3) 1 N6 (, ) = L2 () L3 () = 4 (1 + )(1 2 ) or : N9 (, ) = L2 () L2 () = (1 2 )(1 2 )
(3) (3) (3) (3)

bubble function

[email protected]

Lecture 18

Lagrange Elements

[email protected]

Lecture 18

Lagrange Elements
Naturally, we could derive the same shape functions if we follow the procedure discussed for serendipity elements and use interpolating functions of the form: b0 + b1 + b2 + b3 + b4 2 + b5 2 + b6 2 + b7 2 + b8 2 2 Conversely, we can use the Lagrange polynomials to re-derive the shape functions of the isoparametric elements we have seen before.

For example, the 4-node quadrilateral: N1 (, ) = 1 (1 )(1 ) 4 = L1 ()L1 ()


(2) (2)

Or the shape functions of lateral nodes of the 8-node serendipity element, e.g.: N5 (, ) = L3 ()L1 () =
(3) (2)

1 (1 2 )(1 ) 2

One has to be more careful for the shape functions of the corner nodes!
[email protected] Lecture 18

Isoparametric Quadrilateral Elements with Variable Number of nodes

[email protected]

Lecture 18

Isoparametric Quadrilateral Elements with Variable Number of nodes


5-node quadrilateral: = (x) 3 4 1 5 2 y x x = x()
[email protected] Lecture 18

Isoparametric Quadrilateral Elements with Variable Number of nodes


The shape function of lateral node can be found from Lagrange polynomials: N5 = L3 () L1 () =
(3) (2)

1 (1 2 )(1 ) 2

To nd the shape functions of the corner nodes we use the ones from the 4-node quadrilateral and subtract 1/2N5 IF NEEDED so that they become 0 at all other nodes. Hence: N1 = 1 (1 )(1 ) 1 N5 4 2 1 N2 = 4 (1 + )(1 ) 1 N5 2 and:

1 N3 = 4 (1 + )(1 + ) 1 N4 = 4 (1 )(1 + )

[email protected]

Lecture 18

Summary of Shape functions for Isoparametric Quadrilateral Elements


1 (1 4 1 (1 4 1 (1 4 1 (1 4 1 (1 2 1 (1 2 1 (1 2 1 (1 2

N1 N1 N1 N1 N5 N5 N5 N5

= = = = = = = =

)(1 ) )(1 ) )(1 ) )(1 ) 2 )(1 ) + )(1 2 ) 2 )(1 + ) )(1 2 )

i=5 1 N5 2 1 N5 2

i =6 1 N6 2 1 N6 2

i=7

i=8 1 2 N8 1 N8 2

1 N7 2 1 N7 2

i=9 1 N 4 9 1 N 4 9 1 N 4 9 1 N 4 9 1 N9 2 1 N9 2 1 N9 2 1 N9 2

N9 = (1 2 )(1 2 )

The shape functions in columns i = 5 9 are activated only if the respective node appears in the element and they are added to the appropriate shape function in the rst column.

[email protected]

Lecture 18

You might also like