Ruhr UniversittBochum FakulttfrBauingenieurwesen

ComputationalEngineering FiniteElementsinStructuralMechanics

ShapeFunctionsgeneration, requirements,etc. i t t
Studentpresentation Student presentation

Student:SiljakEnes S d Silj k E January,2009

p pp DisplacementApproximation
A finite element approximation of displacements is given by ~ u( x) u = N i ( x)ui = N (x)u e

(1)

where N i are element shape or interpolation functions, ui are nodal displacements and the sum ranges over the number of nodes associated with an element. d h h b f d i d ih l N is called shape function matrix. Alternatively, in isoparametric form the expressions are given by : ~ u( ) u ( ) = N i ( )ui = N( ) u e (2)
i

x( ) = N i ( ) xi = N ( ) ~ x
i

The interpolation condition on the element shape function N i states that it must take the value one at the i th node and zero at all others. This insures that the interpolation p is correct at the nodes

q p RequirementsforShapeFunctions
Requirements for shape functions are motivated by convergence: as the mesh is fi d th l ti h ld h th l ti l l ti f th refined the FEM solution should approach the analytical solution of the mathematical model. 1. 1 The requirement for compatibility: The interpolation has to be such that field of displacements is : 1. continual and derivable inside the element 2. 2 continual across the element border
The finite elements that satisfy this property are called conforming, or compatible. (The use of elements that violate this property, nonconforming or incompatible elements is however common)

2. The requirement for completeness: The interpolation has to be able to represent: 1. the i id b d di l 1 th rigid body displacement t 2. constant strain state

q p RequirementsforShapeFunctions
Requirement for Compatibility: The shape functions should provide displacement continuity between elements. Physically this insure that no material gaps appear as the elements deform. As the mesh is refined, such gaps would multiply and may absorb or release spurious energy.

Figure 1. Compatibility violation by using different types of elements. a) Discretization and load; b) Deformed shape (left gap, right overlapping)

Requirement for Completeness: The interpolation has to be able to represent: 1. The rigid body displacement 2. Constant strain state

Rigid body translation

Rigid body rotation

Deformation

Figure 2. a) Deformation of cantilever beam. b) Rigid body displacement and deformation of hatched element

RequirementsforShapeFunctions intermsofhighestderivativeinintegrand i f hi h d i i i i d
If the stiffness integrands involve derivatives of order m, then requirements for shape functions can be formulated as follows:
(m-1) 1. The requirement for compatibility: The shape functions must be C( 1) continuous between elements, and Cm piecewise differentiable inside each element.

2. The requirement for completeness: The element shape functions must represent exactly all polynomial terms of order m in the Cartesian coordinates. A set of shape functions that satisfies this condition is called m complete m-complete.

Highestderivativeinintegrand reminder

Virtual work expression for the general three - dimensional problem is given by : ~ ~ ~ ~ [D u ]T C [D u ]d = u T b d uT td
u u
e e

e e

te

Physicalor Ph i l mechanical problem

Independent primaryvariables

Heatconduction

Truss(N=1) Twodimensionalproblem T di i l bl (N=2) Threedimensional continuum(N=3) Displacements N

EulerBernoullibeam (N=1) Thinplate(N=2) (Kirchhoffplate) Transversedisplacement 1

Thickplates(N=2) (ReissnerMindlin plate)

Temperature 1

Transversedisplacement1 RotationN

Operator Du (forN=2)

/ x / y

0 / x 0 / y / y / x

2 / x 2 2 / y 2 2 2 / xy

/ x 0 0 0 0 / y / y / x 0 1 0 / x / y 0 1

Required continuity

C0

C0

C1

C0

Table 1. Differential operator Du for different types of physical or mechanical problems

Generationofshapefunctions Onedimensionalelements Onedimensional elements Physicalandnaturalcoordinates

In the case of a truss element, the position of a point relative to the longitudinal axis is measured in terms of the physical coordinate X1 or the natural coordinate 1

L L X 1 , 2 2

1 1, 1

The relationship between the coordinates X1 and 1 is described by the equation: ti

X1 =

L 1 2

LocalpolynomialApproximation of One dimensional Onedimensional functions

ei ~ u1 (1 ) u1 (1 ) = u1 N i (1 ) = N (1 )u e i =1 p +1

where the matrix of shape functions N(1 ) and the element displacement vector u e are defined in the following way : N (1 ) = N 1 (1 ) N 2 (1 ) ... N NN (1 )

e e eNN u e = u1 1 u1 2 ... u1

( NN = p + 1 symbolizes the number of element nodes.)

Lagrangian interpolation polynomial of degree p can be formed by the product :

1k 1 N (i ) = k i k =1 1 1
i k i

p +1

with 1k characterizing the position of the node k and N i the shape function associated to the node i 1 for i = k N i (1k ) = 0 for i k

Linear, QuadraticandCubicShapefunctions
Here the Lagrange interpolation polynomials used for one-dimensional finite elements must be defined for p = 1, p = 2 and p = 3.
Linear p = 1 Quadratic p = 2 Cubic p = 3

N 1 (1 ) N 2 (1 ) N 3 (1 ) N 4 (1 )

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

1 ( 2 1

1)1

9 (1 1 )(12 16 27 ( 2 16 1

1) 9 + 1) 3

1 12
1 (1 + 1 )1 2

1)(1 1 ) 3

27 (1 12 )(1 16 9 ( 16 1

+ 1)(12 1 ) 9

ShapeFunctionsforBeamElement
Admissible displacements must be C1 continuous. The nodal displacements (3) are used to define uniquely the variation of the transverse displacement

Conditions on the ends :

Figure 3: The two node Beam element with four DOFs.

node 1 : x = L : w = w1 and dw dx = 1 2
(3)

u e = [w1 1 w2 2 ] with four degrees of freedom interpolation will be

node 2 : x = L : w = w2 and dw dx = 2 2 dw / dx = 2 + 2 3 x + 3 4 x 2 calculating f nodal values : l l i for d l l w1 = 1 2 L + 3 ( L ) 4 ( L ) 2 2 2

2 2 3 2 3

w( x) = 1 + 2 x + 3 x 2 + 4 x3 w( x) = [1 x 1 3 2 x ] 3 4

1 = 2 2 3 L + 3 4 ( L ) 2 2

x2

2 = 2 + 2 3 L + 3 4 ( L ) 2 2

w2 = 1 + 2 L + 3 ( L ) + 4 ( L ) 2 2 2
2

ShapeFunctionsforBeamElement
Written in matrix form : L 1 2 w1 0 1 1 = w2 L 1 2 2 0 1 L 2 L 2 2 2 L 2 L 2 2
2

L 2 2 L 1 3 2 2 3 L 3 2 4 2 L 3 2
3

Solving for coeficients : 2a 3 1 1 3a 2 2 = 3 4a 3 0 4 1 a4 a3 a2 a 2a 3 3a 2 0 1 a 4 w1 a 3 1 a 2 w2 a 2 L with : a = 2

ShapeFunctionsforBeamElement
w( x) = [1 x x2 1 2 3 3 x ] 4a 0 1 3 4a a 4 1 4 1 4a 1 4a 2 1 2 3 4a 0 1 4a 3 a 4 1 w1 4 1 1 w2 4 a 1 2 4a 2

w( x) = [N w1

N 1

N w2

w1 N 2 ] 1 w2 2

1 3x x3 1 x3 x N w1 ( x) = + 3 = 3 3 + 2 2 4a 4a 4 a a a x x2 x3 a x3 x 2 x N 1 ( x) = + = + 1 4 4 4a 4a 2 4 a 3 a 2 a 1 3x x3 1 x3 x 3 = 3 3 2 N w2 ( x) = + 2 4a 4a 4 a a a x x2 x3 a x3 x 2 x + = + 1 N 2 ( x) = + 4 4 4a 4a 2 4 a 3 a 2 a

ShapeFunctionsforBeamElement
with = L x a = : a 2 1 N w1 ( ) = 3 3 + 2 4 a N 1 ( ) = 3 2 + 1 4 1 N w 2 ( ) = 3 3 2 4 a N 2 ( ) = 3 + 2 1 4

) )

These functions are called Hermitian cubic shape f b h functions i

Differentiating twice to get curvature :

d 2 w 4 d 2 w 4 dN e u = Bu e = N'' u e = 2 = 2 2 = 2 dx L d L d B=

1 6 3 1 6 3 + 1 L L L

Figure 4. Hermitian cubic shape functions

ShapeFunctions ofPlaneElements
Classification of shape functions according to:
the element form: triangular elements, rectangular elements elements. polynomial degree of the shape functions: linear quadratic cubic type of the shape functions Lagrange shape functions g g p serendipity shape functions

Rectangularelements Lagrangefamily
Lagrange interpolation polynomial in one direction :
n lk (1 )

1 1i = k i i = 0 1 1
n i ik

An easy and systematic method of generating shape functions of any order now can be achieved by simple products of Lagrange polynomials in the two coordinates :
m N a = N IJ = lIn (1 )l J ( 2 )

where :

1 =

2( x xe ) 2( y ye ) 2 = ae be
Figure 5. Generation of a typical shape function for a Lagrangian element

xe , ye are coordinates of the center of the element ae , be are dimensions of the element

Rectangularelements Lagrangefamily

Figure 6: The Quadrilateral Lagrangian elements: a) bilinear, b) biquadratic c) bicubic

Figure 7: The Quadrilateral Lagrangian elements: a) quadratic-linear, b) linear-cubic c) quadratic-cubic, d) quartic-quadratic

Rectangularelements Lagrangefamily

Figure 8: Complete two-dimensional Lagrange ansatz polynomials in the Pascal triangle

TheFourNodeBilinearQuadrilateral

Figure 9: The four-node bilinear quadrilateral: (a) element geometry; (b) perspective view of shape function

1 N 1 ( ) = (1 1 )(1 2 ) 4 1 N 2 ( ) = (1 + 1 )(1 2 ) 4

1 N 3 ( ) = (1 + 1 )(1 + 2 ) 4 1 N 4 ( ) = (1 1 )(1 + 2 ) 4

TheFourNodeBilinearQuadrilateral ec o co pa b y Checkofcompatibility

Figure 10: Assemblage of four bilinear quadrilateral elements

Change of N5 along the edge is linear and it is uniquely defined by two nodes Derivative inside element exists, and on the boundary has finite discontinuity
Figure 11: Partial derivatives with respect to x and y of the shape functions N5

Checkofcompleteness
A set of shape functions is complete for a continuum element if they can represent exactly any linear displacement motions such as : u x = 0 + 1 x + 2 y , u y = 0 + 1 x + 2 y (1)

The nodal point displacements corresponding to this displacement field are : u xi = 0 + 1 xi + 2 yi , u yi = 0 + 1 xi + 2 yi (2)

The displacements (1) have to be obtained within the element when the element nodal point displacements are given by (2). In the isoparametric formulation we have the displacement interpolation : ux =
i =1 n

u xi N ie

u y = u yi N ie
i =1

(3)

Computatio n for the displacement u x : ux = (

i =1 e

0 + 1 xi + 2 yi N ie

= 0
i =1

N ie

+ 1
i =1

xi N ie

+ 2 yi N ie
i =1

(4)

Since in the isoparametric formulation the coordinates are interpolated in the same way as the displacements, we can use : x=
i =1 n

xi N ie

y = yi N ie
i =1

to obtain : u x = 0 N ie + 1 x + 2 y
i =1 n

(5)

The displacements d fi d i (5) are the same as those given i (1) provided Th di l defined in h h i in (1), id d that for any point in the element :

Ni = 1
i =0

(6)

The relation (6) is the condition on the interpolation functions for the completeness requirements to be satisfied

TheNineNodeBiquadratic Quadrilateral
i i N i = 1 (1 + 1i1 )1i1 (1 + 2 2 ) 2 2 4

2 i N i = 1 (1 2 )(1 + 1i1 )1i1 , 2 = 0 2 i i N i = 1 (1 12 )(1 + 2 2 ) 2 2 , 1i = 0 2

Figure 12: The nine-node biquadratic quadrilateral

2 N 9 = (1 12 )(1 2 )

1 N ( ) = (1 1 )(1 2 ) 1 2 4
1

1 N ( ) = (1 12 ) ( 2 1) 2 2
5

2 N 9 ( ) = (1 12 )(1 2 )

Figure 13: Perspective view of shape functions for nodes 1, 5 and 9 of the nine-node biquadratic quadrilateral

The16NodeBicubic Quadrilateral
Ni = Ni = Ni =
Figure 14: The 16-node bicubic quadrilateral
i 81 (1 + 1i1 )(1 + 2 2 )( 1 9 256

2 12 )( 1 2 ) 9

2 243 (1 12 )( 2 256

i 1 )( 1 + 31i1 )(1 + 2 2 ) 9 3

2 243 (1 2 )(12 256

i 1 )( 1 + 3 2 2 )(1 + 1i1 ) 9 3

i 2 N i = 729 (1 12 )(1 2 )( 1 + 31i1 )( 1 + 3 2 2 ) 256 3 3

Figure 15: Perspective view of shape functions for nodes 1, 5 and 13 of the 16-node bicubic quadrilateral

Serendipityelements
Serendipity elements are constructed with nodes only on the element b d boundary

Figure 16. Serendipity quadrilateral elements: a) bilinear , b) biquadratique, c) bicubic

Figure 17 Two dimensional serendipity polynomials 17. of quadrilateral elements in Pascal triangle

SerendipityBiquadratic Shapefunctions
For mid-side nodes a lagrangian interpolation of quadratic x linear type suffices to determine Ni at nodes 5 to 8. For corner nodes start with bilinear lagragian family (step 1), and successive subtraction (step 2, step 3) ensures zero value at nodes 5, 8

N =
5

1 (1 12 )(1 2 ) 2

N =
8

2 1 (1 1 )(1 2 ) 2

N 1 ( ) = 1 (1 1 )(1 2 )(1 1 2 ) 4
i i N i ( ) = 1 (1 + 1i 1 )(1 + 2 2 )(1 + 1i 1 + 2 2 ) 4

N 1 = 1 (1 1 )(1 2 ) 4

N1 1 N 5 2

N1 = N1 1 N 5 1 N 8 2 2
2 N 1 = 1 (1 1 )(1 2 ) 1 1 (1 12 )(1 2 ) 1 1 (1 1 )(1 2 ) = 4 2 2 2 2 1 (1 1 )(1 2 ) 1 (1 1 ) 4

N 5 ( ) = 1 (1 12 )(1 2 ) 2
i N i ( ) = 1 (1 12 )(1 + 2 2 ) , for 1i = 0 2 i 2 N i ( ) = 1 (1 + 1i 1 )(1 2 ) , for 2 = 0 2

(1 2 )]

N 1 = 1 (1 1 )(1 2 )(1 1 2 ) 4
Figure 18. Construction of serendipity element shape functions Figure 19. Biquadratic serendipity element shape functions

SerendipityShapefunctions
In general serendipity shape functions can be obtained with the following expression:

N i (1 , 2 ) = 1 (1 2 ) N i ( 1 ,-1) + 1 (1 + 1 ) N i (1, 2 ) , 1) 2 2 + 1 (1 + 2 ) N i ( 1 ,1) + 1 (1 1 ) N i (-1, 2 ) 2 2 1 (1 1 )(1 2 ) N i (-1,-1) 1 (1 + 1 )(1 2 ) N i (1,-1) 4 4 1 (1 + 1 )(1 + 2 ) N i (1,1) 1 (1 1 )(1 + 2 ) N i (-1,1) 4 4

where functions N i (1 ,1), N i (1, 2 ), N i (1 ,1), N i (1, 2 ) are lagrangian interpolations along the coresponding boundary and values N i (1,1), N i (1,1), N i (1, 1), N i (1, 1) have values 0 or 1 and represent values of interpolation on corrners

SerendipityShapefunctions
As an example shape function of the node N3 of the element on Fig. 20 will be generated .

Figure 20. Construction of N3

N 3 (1 ,1) N 3 ( 1, 2 ) N (1, 2 )
3

N 3 (1, 2 ) = 1 1 (1 + 1 )(1 + 2 ) 4
Figure 21. View of shape function N3

N 3 (1 ,1)

N 3 (1, 2 ) = 1 (1 + 1 ) N 3(1, 2 ) + 1 (1 + 2 ) N 3( 1 ,1) 1 (1 + 1 )(1 + 2 ) N 3(1,1) = 4 2 2

1 (1 + 1 ) 1 (1 + 2 ) + 1 (1 + 2 ) 1 1 (1 + 1 ) 1 (1 + 1 )(1 + 2 ) 1 = 2 2 2 2 4 1 (1 + 1 )(1 + 2 ) 4 1

Quadrilateralelementwith a ab e u be o odes ( o 6 odes) variablenumberofnodes(4to16nodes)

Shape functions of nodes 1 to 4 N 1 = 0.25(1 1 )(1 2 ) N 2 = 0.25(1 + 1 )(1 2 ) N 3 = 0.25(1 + 1 )(1 + 2 ) N 4 = 0.25(1 1 )(1 + 2 )
N 5 = 0 if node 5 is absent N 6 = 0 if node 6 is absent N 7 = 0 if node 7 is absent N 8 = 0 if node 8 is absent

Shape functions of nodes 5 to 8 p N 5 = 0.5(1 12 )(1 2 ) if node 5 is present;

2 N 6 = 0.5(1 + 1 )(1 2 ) if node 6 is present;

N 7 = 0.5(1 12 )(1 + 2 ) if node 7 is present;

2 N 8 = 0.5(1 1 )(1 2 ) if node 8 is present;

Corrections of shape functions of corner nodes : p N 1 N 1 0.5 ( N 8 + N 5 ) N 2 N 2 0.5 ( N 5 + N 6 ) N 3 N 3 0.5 ( N 6 + N 7 ) N 4 N 4 0.5 ( N 7 + N 8 )

Shape function of internal node 9

2 N 9 = (1 12 )(1 2 ) if element is biquadratic lagrangian; if not N 9 = 0

Correction of shape functions of corner nodes and in the middle of the sides : N i = N i + 0.25 N 9 (i = 1, 2, 3, 4) N i = N i 0.5 N 9 (i = 5, 6, 7, 8)

Quadrilateralelementwith a ab e u be o odes ( o 6 odes) variablenumberofnodes(4to16nodes)

Shape functions of nodes 9 to 12 on the border N 9 = 0.28125(1 12 )(1 2 )(1 + 31 ) if node 9 is present; N 9 = 0 if node 9 is absent
2 N 10 = 0.28125(1 + 1 )(1 2 )(1 + 3 2 ) if node 10 is present; N 10 = 0 if node 10 is absent

N 11 = 0.28125(1 12 )(1 + 2 )(1 31 ) if node 11 is present; N 11 = 0 if node 11 is absent

2 N 12 = 0.28125(1 1 )(1 2 )(1 3 2 ) if node 12 is present; N 12 = 0 if node 12 is absent

Corrections of shape functions of corner nodes N 1 N 1 + 0.125 N 8 0.25 N 5 ( N 12 N 9 ) / 3 N 2 N 2 + 0.125 N 5 0.25 N 6 ( N 9 N 10 ) / 3 N 3 N 3 + 0.125 N 6 0.25 N 7 ( N 10 N 11 ) / 3 N 4 N 4 + 0.125 N 7 0.25 N 8 ( N 11 N 12 ) / 3
Corrections of shape functions of nodes on the border N 5 1.125 N 5 N 9 N 6 1.125 N 6 N 10 N 7 1.125 N 7 N 11 N 8 1.125 N 8 N 12

Quadrilateralelement a ab e u be o odes ( o 6 odes) withvariablenumberofnodes(4to16nodes)

Shape functions of internal nodes 13 to 16
2 N 13 = 81(1 12 )(1 31 )(1 2 )(1 3 2 ) / 256 2 N 14 = 81(1 12 )(1 + 31 )(1 2 )(1 3 2 ) / 256 2 N 15 = 81(1 12 )(1 + 31 )(1 2 )(1 + 3 2 ) / 256 2 N 16 = 81(1 12 )(1 31 )(1 2 )(1 + 3 2 ) / 256

if node 13 is present; if node 14 is present; if node 15 is present; if node 16 is present;

N 13 = 0 if node 13 is absent N 14 = 0 if node 14 is absent N 15 = 0 if node 15 is absent N 16 = 0 if node 16 is absent

Corrections of shape functions of corner nodes N 1 N 1 + 4( N 13 + 0.5 N 14 + 0.25 N 15 + 0.5 N 16 ) / 9 N 14 + 0.5 N 15 + 0.25 N 16 ) / 9 N 15 + 0.5 N 16 ) / 9 N 16 ) / 9 N 2 N 2 + 4( 0.5 N 13 +

N 3 N 3 + 4(0.25 N 13 + 0.5 N 14 +

N 4 N 4 + 4( 0.5 N 13 + 0.25 N 14 + 0.5 N 15 +

Corrections of shape functions of the nodes on the border N 5 N 5 ( 2 N 13 + N 16 ) / 3 N 6 N 6 (2 N 14 + N 13 ) / 3 N 7 N 7 (2 N 15 + N 14 ) / 3 N 8 N 8 (2 N 16 + N 15 ) / 3 N 9 N 9 (2 N 14 + N 15 ) / 3 N 10 N 10 ( 2 N 15 + N 16 ) / 3 N 11 N 11 (2 N 16 + N 13 ) / 3 N 12 N 12 ( 2 N 13 + N 14 ) / 3

Triangularelements g ThreeNodeTriangleElement

Triangularelements g ThreeNodeTriangleElement
N 1 = 1 1 2 N 2 = 1 N 3 = 2

N 1 = 1 1 2

N 2 = 1
Figure 24. Isometric view of shape functions,

N 3 = 2

SixNodeQuadraticTriangle
Shape f ti Sh functions can b expressed as products be d d t of linear functions : N i = ci L1 L2 ... Ln where L j = 0 , j = 1,..., n are the homogeneous equation of lines on which N i vanishes and ci is a normalization coefficient N 1 = (1 1 2 )(1 21 2 2 ) N 2 = 1 (21 1) N 3 = 2 (2 2 1) N 4 = ci1 (1 1 2 ) N 4 ( 1 ,0) = ci 1 (1 1 ) = 1 ci = 4 2 2 2 N 4 = 41 (1 1 2 ) N = 41 2
5

N 1 = (1 1 2 )(1 21 2 2 )

N 4 = 41 (1 1 2 )

N 6 = 4 2 (1 1 2 )

Figure 25. Shape functions N1 and N 4 for the quadratic triangle

TenNodeQubic Triangle
Calculating as products of edges on which shape function vanishes :
N 1 = c1 (1 31 3 2 )(2 31 3 2 )(1 1 2 ) N 1 (0,0) = c1 2 = 1 c1 = 1 2 N 1 = 1 (1 31 3 2 )(2 31 3 2 )(1 1 2 ) 2
2 N 2 = c21 (1 1 )(1 2 ) N 2 (1,0) = c2 9 = 1 c2 = 9 3 3 2

N 2 = 9 1 (1 1 )(1 2 ) 2 3 3
2 N 3 = 9 2 ( 2 1 )( 2 3 ) 2 3

N 4 = 9 1 (2 31 3 2 )(1 1 2 ) 2 N 5 = 9 1 2 (31 1) 2 N 6 = 9 2 (3 2 1)(1 1 2 ) 2 N 7 = 9 1 (31 1)(1 1 2 ) 2 N 8 = 9 1 2 (3 2 1) 2 N 9 = 9 2 (2 31 3 2 )(1 1 2 ) 2 N 10 = 271 2 (1 1 2 ) Figure 26. Shape functions N1, N4 and N10 for the Qubic triangle

ShapefunctionsofThreedimensionalelements

Figure 27. Examples of Solid finite elements: a) Hexahedral b) Prismatic c) Tetrahedral element

RectangularPrisms Lagrangefamily
Shape functions for this element will be generated by a direct product of three Lagrange polynomials:
m p N a = N IJK = lIn (1 )l J ( 2 )lK (3 )

for n, m, p, subdivisions along each side and : 2 2 2 1 = ( x xe ) , 2 = ( y ye ) , 3 = ( z ze ) a b c

TheeightNodeHexahedron

The shape functions are:

N 1 = 1 (1 1 )(1 2 )(1 3 ) 8 N 2 = 1 (1 + 1 )(1 2 )(1 3 ) 8 N 3 = 1 (1 + 1 )(1 + 2 )(1 3 ) 8 N 4 = 1 (1 1 )(1 + 2 )(1 3 ) 8 N 5 = 1 (1 1 )(1 2 )(1 + 3 ) 8 N 6 = 1 (1 + 1 )(1 2 )(1 + 3 ) 8 N 7 = 1 (1 + 1 )(1 + 2 )(1 + 3 ) 8 N 8 = 1 (1 1 )(1 + 2 )(1 + 3 ) 8

These eight formulas can be summarized in a single expression:

i i N i = 1 (1 + 1i1 )(1 + 2 2 )(1 + 33 ) 8

i i where 1i , 2 , 3 denote the coordinates of the ith node.

The20Node(Serendipity)Hexahedron
The 20-node hexahedron is equivalent to the 8-node serendipity q p y quadrilateral.

The shape functions are:

For the corner nodes i = 1, 2, ..., 8 :
i i i i N i = 1 (1 + 1i1 )(1 + 2 2 )(1 + 33 )(1i1 + 2 2 + 33 2) 8

For the midside nodes i = 9, 11, 13, 15 :

i i N i = 1 (1 12 )(1 + 2 2 )(1 + 33 ) 4

For the midside nodes i = 10, 12, 14, 16 :

i 2 N i = 1 (1 2 )(1 + 1i1 )(1 + 33 ) 4

For the midside nodes i = 17, 18, 19, 20 :

i 2 N i = 1 (1 3 )(1 + 1i1 )(1 + 2 2 ) 4

ShapefunctionsofHexahedron withvariablenumberofnodes(827nodes)
Shape functions of nodes 1 to 8 N 1 = 0.125(1 1 )(1 2 )(1 3 ) N 2 = 0.125(1 + 1 )(1 2 )(1 3 ) N 3 = 0.125(1 + 1 )(1 + 2 )(1 3 ) N 4 = 0.125(1 1 )(1 + 2 )(1 3 ) N 5 = 0.125(1 1 )(1 2 )(1 + 3 ) N 6 = 0.125(1 + 1 )(1 2 )(1 + 3 ) N 7 = 0.125(1 + 1 )(1 + 2 )(1 + 3 ) N 8 = 0.125(1 1 )(1 + 2 )(1 + 3 )

Shape functions of nodes 9 to 20 along the border N 9 = 0.25(1 12 )(1 2 )(1 3 ) if node 9 is present; N 9 = 0 if not present;
2 N 10 = 0.25(1 + 1 )(1 2 )(1 3 ) if node 10 is present; N 10 = 0 if not present; p ; p ;

N 11 = 0.25(1 12 )(1 + 2 )(1 3 ) if node 11 is present; N 11 = 0 if not present;

2 N 12 = 0.25(1 1 )(1 2 )(1 3 ) if node 12 is present; N 12 = 0 if not present;

N 13 = 0.25(1 12 )(1 2 )(1 + 3 ) if node 13 is present; N 13 = 0 if not present;

2 N 14 = 0.25(1 + 1 )(1 2 )(1 + 3 ) if node 14 is present; N 14 = 0 if not present;

N 15 = 0.25(1 12 )(1 + 2 )(1 + 3 ) if node 15 is present; N 15 = 0 if not present;

2 N 16 = 0.25(1 1 )(1 2 )(1 + 3 ) if node 16 i present; N 16 = 0 if not present; d is t t t 2 N 17 = 0.25(1 1 )(1 2 )(1 3 ) if node 17 is present; N 17 = 0 if not present; 2 N 18 = 0.25(1 + 1 )(1 2 )(1 3 ) if node 18 is present; N 18 = 0 if not present; 2 N 19 = 0.25(1 + 1 )(1 + 2 )(1 3 ) if node 19 is present; N 19 = 0 if not present;

N 20 = 0.25(1 1 )(1 + 2 )(1 32 ) if node 20 is present; N 20 = 0 if not present;

ShapefunctionsofHexahedron withvariablenumberofnodes(827nodes)
Correction of shape functions of corner nodes : N 1 N 1 0.5( N 12 + N 9 + N 17 ) N 2 N 2 0.5( N 9 + N 10 + N 18 ) N 3 N 3 0.5( N 10 + N 11 + N 19 ) N 4 N 4 0.5( N 11 + N 12 + N 20 ) N 5 N 5 0.5( N 16 + N 13 + N 17 ) N 6 N 6 0.5( N 13 + N 14 + N 18 ) N 7 N 7 0.5( N 14 + N 15 + N 19 ) N 8 N 8 0.5( N 15 + N 16 + N 20 )

Shape functions of mid - face nodes 21 to 26 :

2 N 21 = 0.5(1 12 )(1 2 )(1 3 ) if node 21 is present; 2 N 22 = 0.5(1 12 )(1 2 )(1 + 3 ) if node 22 is present;

N 21 = 0 if node 21 is absent N 22 = 0 if node 22 is absent N 23 = 0 if node 23 is absent N 24 = 0 if node 24 is absent N 25 = 0 if node 2 i absent d 25 is b N 26 = 0 if node 26 is absent

N 23 = 0.5(1 12 )(1 2 )(1 32 ) if node 23 is present;

2 N 24 = 0.5(1 + 1 )(1 2 )(1 32 ) if node 24 is present;

N 25 = 0.5(1 12 )(1 + 2 )(1 32 ) if node 2 i present; d 25 is

2 2 N 26 = 0.5(1 1 )(1 2 )(1 3 ) if node 26 is present;

Correction of shape functions of corner nodes : N 1 N 1 + 0.25( N 21 + N 26 + N 23 ) N 2 N 2 + 0.25( N 21 + N 23 + N 24 ) N 3 N 3 + 0.25( N 21 + N 24 + N 25 ) N 4 N 4 + 0.25( N 21 + N 25 + N 26 ) N 5 N 5 + 0.25( N 22 + N 26 + N 23 ) N 6 N 6 + 0.25( N 22 + N 23 + N 24 ) N 7 N 7 + 0.25( N 22 + N 24 + N 25 ) N 8 N 8 + 0.25( N 22 + N 25 + N 26 )

ShapefunctionsofHexahedron withvariablenumberofnodes(827nodes)
Correction of shape functions on the border : N 9 N 9 0.5( N 21 + N 23 ) N 10 N 10 0.5( N 21 + N 24 ) N 11 N 11 0.5( N 21 + N 25 ) N 12 N 12 0.5( N 21 + N 26 ) N 13 N 13 0.5( N 22 + N 23 ) N 14 N 14 0.5( N 22 + N 24 ) N 15 N 15 0.5( N 22 + N 25 ) N 16 N 16 0.5( N 22 + N 26 ) N 17 N 17 0.5( N 26 + N 23 ) N 18 N 18 0.5( N 23 + N 24 ) N 19 N 19 0.5( N 24 + N 25 ) N 20 N 20 0.5( N 25 + N 26 )

Shape function of the internal node 27 :

2 2 N 27 = (1 12 )(1 2 )(1 3 ) if node 27 is present; N 27 = 0 if node 27 is absent

Correction of shape functions of nodes 1 to 26 : N i N i 0.125N 27 N i N i + 0.25 N 27 N i N i 0.5 N 27 (i = 1, 2,..., 8) (i = 9, 10,..., 20) (i = 21, 22,..., 26)

ShapefunctionsofPrismaticelement withvariablenumberofnodes(618nodes)
Shape functions of nodes 1 to 6 N 1 = 0.5(1 1 2 )(1 3 ) N 2 = 0.5 N 3 = 0.5 N 4 = 0.5(1 1 2 )(1 + 3 ) N 5 = 0.5 N 6 = 0.5

1 2

(1 3 ) (1 3 )

1 2

(1 + 3 ) (1 + 3 )

Shape functions of nodes 7 to 15 along the edge N 7 = 21 (1 1 2 )(1 3 ) if node 7 is present; N 7 = 0 if not present; N 8 = 2 1

(1 3 ) if node 8 is present; N 8 = 0 if not present;

N 9 = 2 2 (1 1 2 )(1 3 ) if node 9 is present; N 9 = 0 if not present; N 10 = 21 (1 1 2 )(1 + 3 ) if node 10 is present; N 10 = 0 if not present; N 11 = 2 1 N 13 = N 14 = 1 N 15 = 2

(1 + 3 ) if node 11 is present; N 11 = 0 if not present;

N 12 = 2 2 (1 1 2 )(1 + 3 ) if node 12 is present; N 12 = 0 if not present; p ; p ;

2 (1 1 2 )(1 3 ) if node 13 is present; N 13 = 0 if not present; 2 (1 3 ) if node 14 is present; N 14 = 0 if not present; 2 (1 3 ) if node 15 is present; N 15 = 0 if not present;

Correction of shape functions of corner nodes : N 1 N 1 0.5( N 9 + N 7 + N 13 ) N 2 N 2 0.5( N 7 + N 8 + N 14 ) N 3 N 3 0.5( N 8 + N 9 + N 15 ) N 4 N 4 0.5( N 12 + N 10 + N 13 ) N 5 N 5 0.5( N 10 + N 11 + N 14 ) N 6 N 6 0.5( N 11 + N 12 + N 15 )

ShapefunctionsofPrismaticElement withvariablenumberofnodes(618nodes)
Shape functions of nodes 16 to 18 on the face
2 N 16 = 41 (1 1 2 )(1 3 ) if node 16 is present; N 16 = 0 if not present;

N 17 = 4 1

2 (1 3 ) if node 1 i present; N 17 = 0 if not present; d 17 is

N 18 = 4 2 (1 1 2 )(1 32 ) if node 18 is present; N 18 = 0 if not present; Correction of shape functions of corner nodes : N 1 N 1 + 0.25( N 18 + N 16 ) N 2 N 2 + 0.25( N 16 + N 17 ) N 3 N 3 + 0.25( N 17 + N 18 ) N 4 N 4 + 0.25( N 18 + N 16 ) N 5 N 5 + 0.25( N 16 + N 17 ) N 6 N 6 + 0.25( N 17 + N 18 )

Correction of shape functions of corner on the edge : N 7 N 7 0.5 N 16 N 10 N 10 0.5 N 16 N 8 N 8 0.5 N 17 N 11 N 11 0.5 N 17 N 15 N 15 0.5( N 17 + N 18 ) N 9 N 9 0.5 N 18 N 12 N 12 0.5 N 18

N 13 N 13 0.5( N 18 + N 16 )

N 14 N 14 0.5( N 16 + N 17 )

ShapefunctionsofTetrahedronElement withvariablenumberofnodes(410nodes)
Shape functions of corner nodes 1 to 4 : N 1 = 1 1 2 3 N 2 = 1 N 3 = 2 N 4 = 3

Shape functions of nodes 5 to 10 on the border : N 5 = 41 (1 1 2 3 ) N 6 = 41 2 if node 5 is present; N 5 = 0 if node 5 is absent if node 6 is present; N 6 = 0 if node 6 is absent

N 7 = 4 2 (1 1 2 3 ) if node 7 is present; N 7 = 0 if node 7 is absent N 8 = 43 (1 1 2 3 ) if node 8 is present; N 8 = 0 if node 8 is absent N 9 = 413 N 10 = 4 23 if node 9 is present; N 9 = 0 if node 9 is absent if node 10 is present; N 10 = 0 if node 10 is absent

Corrections of shape functions of corner nodes : N 1 N 1 0.5( N 5 + N 7 + N 8 ) N 2 N 2 0.5( N 5 + N 6 + N 9 ) N 3 N 3 0.5( N 6 + N 7 + N 10 ) N 4 N 4 0.5( N 8 + N 9 + N 10 )

Literature
1. BatheK.J.:FiniteElementProcedures,PrenticeHallofIndia,NewDelhi2007 2. Felippa C.:IntroductiontoFiniteElementMethods, 2 C : Introduction to Finite Element Methods
http://www.colorado.edu/engineering/CAS/courses.d/IFEM.d/

3. GmrT. :Mthode deslments finis,Pressespolytechniques et universitaires romandes,2000 4. Knothe K.,Wessels H.:FiniteElemente.Eine EinfrungfrIngenieure. SpringerVerlagBerlin2007 5. KuhlD.,MeschkeG.:LectureNotes,FiniteElementMethodsinLinear StructuralMechanics,6.Edition,InstituteforStructuralMechanics,Ruhr UniversittBochum,2008 6. Zienkiewicz O.C.,TaylorR.L.,ZhuJ.Z.:TheFiniteElementMethod:ItsBasis andFundamentals,6.Edition,ElsevierButterworthHeinemann,2005 d d l 6 di i l i hH i 200