Chapter -3

INTERPOLATION(SHAPE) FUNCTION

Yilma T(PhD)

1
 Shape functions
• The vertices of finite element (e.g.triangular)are numbered
to indicate that these points are nodes. A node is a specific
point in the finite element at which the value of the field
variable is to be explicitly calculated.

• Exterior nodes are located on the boundaries of the finite

element and may be used to connect an element to
adjacent finite elements.

• Nodes that do not lie on element boundaries are interior

nodes and cannot be connected to any other element.
2
Example

3
 Shape functions cont..
• Finite element method involves the discretization of
both the domain and the governing equations. In this
process, the variables are represented in a piece-wise
manner over the domain.
• By dividing the solution region into a number of small
regions, called elements, and approximating the
solution over these regions by a suitable known
function, a relation between the differential equations
and the elements is established.
• The functions employed to represent the nature of the
solution within each element are called shape
functions, or interpolating functions, or basis functions.
4
• They are called interpolating functions as they are used
to determine the value of the field variable within an
element by interpolating the nodal values.
• They are also known as basis functions as they form the
basis of the discretization method.
• Polynomial type functions have been most widely used
as they can be integrated, or differentiated, easily and
the accuracy of the results can be improved by increasing
the order of the polynomial as shown in Figure 3.3(c) and
(d).

5
• Figure 3.3 One-
dimensional
finite elements.
(a) A linear
element, (b) a
quadratic
element, (c)
linear and (d)
quadratic
variation of
temperature
over an element

6
 Decisive point in FEM
If the values of the field variable are computed only at
nodes, how are values obtained at other points within a
finite element? The answer contains the crux of the
finite element method:
The values of the field variable computed at the nodes
are used to approximate the values at non nodal points
(that is, in the element interior) by interpolation of the
nodal values.
• For the three-node triangle example, the nodes are all
exterior and, at any other point within the element, the
field variable is described by the approximate relation

7
 Shape functions cont..
(1.1)

• Where are the values of the field variable

at the nodes, and are the interpolation
functions, also known as shape functions or blending
functions. In the finite element approach, the nodal
values of the field variable are treated as unknown
constants that are to be determined.

• The interpolation functions are most often polynomial

forms of the independent variables, derived to satisfy
certain required conditions at the nodes.
8
 Shape functions cont..
 The major point to be made here is that the interpolation
functions are predetermined, known functions of the
independent variables; and these functions describe the
variation of the field variable within the finite element.
• This would be the case if the field variable represents a scalar
field, such as temperature in a heat transfer problem. If the
domain of Figure 1.1 represents a thin, solid body subjected to
plane stress, the field variable becomes the displacement vector
and the values of two components must be computed at each
node.
• In the latter case, the three-node triangular element has 6
degrees of freedom.

9
 Shape functions cont..
• In general, the number of degrees of freedom associated with a
finite element is equal to the product of the number of nodes
and the number of values of the field variable (and possibly its
derivatives) that must be computed at each node.

10
 FEM objective
• The objective of the finite element method is to
discretize the domain into a number of finite
elements for which the governing equations are
algebraic equations.
• Solution of the resulting system of algebraic
equations then gives an approximate solution to the
problem
• As with any approximate technique, the question,
How accurate is the solution? must be addressed.

11
 FEM Solution accuracy
• In finite element analysis, solution accuracy is
judged in terms of convergence as the element
“mesh” is refined.
• There are two major methods of mesh refinement.
i) h-refinement, mesh refinement refers to the process
of increasing the number of elements used to model a
given domain, consequently, reducing individual
element size.

12
 FEM Solution accuracy
ii) p-refinement : element size is unchanged but the
order of the polynomials used as interpolation
functions is increased.
• The objective of mesh refinement in either method
is to obtain sequential solutions that exhibit
asymptotic convergence to values representing the
exact solution.

13
 Unstructured meshes

• Mathematical proofs of convergence of finite

element solutions to correct solutions are based on
a specific, regular mesh refinement procedure.
• Although the proofs are based on regular meshes
of elements, irregular or unstructured meshes
(such as in Figure 1.2) can give very good results.

14
Fig.1.2 A mesh of finite elements over a rectangular region
having a
central hole.
15
 Unstructured meshes cont..
 In fact, use of unstructured meshes is more often
the case, since
(1) The geometries being modeled are most often irregular and
(2) The auto meshing features of most finite element software
packages produce irregular meshes.
 An example illustrating regular h-refinement as well
as solution convergence is shown in Figure 1.3a,
which depicts a rectangular elastic plate of uniform
thickness fixed on one edge and subjected to a
concentrated load on one corner.
16
Figure 1.3 Example showing convergence
as element mesh is refined. 17
 Shape functions cont..
• The need for convergence during regular mesh
refinement is rather clear. If convergence is not
obtained, the engineer using the finite element
method has absolutely no indication whether the
results are indicative of a meaningful approximation
to the correct solution. For a general field problem
in which the field variable of interest is expressed
on an element basis in the discretized form

(1.1)

18
• Where M is the number of element degrees of
freedom.
The interpolation functions must satisfy two primary
conditions to ensure convergence during mesh
refinement:
i) The compatibility and
ii) Completeness requirements

19
 Compatibility
• Along element boundaries, the field variable and its
partial derivatives up to one order less than the
highest-order derivative appearing in the integral
formulation of the element equations must be
continuous.
• Given the discretized representation of Equation
1.1, it follows that the interpolation functions must
meet this condition, since these functions
determine the spatial variation of the field variable.
20
 Compatibility
• In addition to satisfying the criteria for convergence,
the compatibility condition can be given a physical
meaning as well. In structural problems, the
requirement of displacement continuity along element
boundaries ensures that no gaps or voids develop in
the structure as a result of modeling procedure.
• Similarly, the requirement of slope continuity for the
beam element ensures that no “kinks” are developed in
the deformed structure. In heat transfer problems, the
compatibility requirement prevents the physically
unacceptable possibility of jump discontinuities in
temperature distribution. 21
 Completeness
• In the limit as element size shrinks to zero in mesh
refinement, the field variable and its partial
derivatives up to, and including, the highest-order
derivative appearing in the integral formulation
must be capable of assuming constant values.
Again, because of the discretization, the
completeness requirement is directly applicable to
the interpolation functions.

22
 POLYNOMIAL FORMS:
ONE-DIMENSIONAL ELEMENTS
• Formulation of finite element characteristics
requires differentiation and integration of the
interpolation functions in various forms. Owing to
the simplicity with which polynomial functions can
be differentiated and integrated, polynomials are
the most commonly used interpolation functions.

23
24
1D LINEAR ELEMENT
1. Kronecker delta property: The shape function at any node
has a value of 1 at that node and a value of zero at ALL other
nodes.
1 1
x2 - x x - x1
N1 (x)  N 2 (x) 
x 2  x1 x 2  x1
x1 x2 x
El #1
Check
x2 - x
N1 (x) 
x 2  x1
x 2 - x1
 N1 (x  x1 )  1
x 2  x1
x2 - x2
and N1 (x  x 2 )  0
x 2  x1 25
The approximation is continuous across element boundaries
x3 - x x - x2
d (2) (x)  d 2x  d 3x
x3  x 2 x3  x 2
x2 - x x - x1
d(x) 1  d1x  d 2x
x 2  x1 x 2  x1

x1 x2 x3 x
El #1 El #2

26
Completeness

N1 (x)  N 2 (x)  1 for all x

N1 (x)x1  N 2 (x)x 2  x for all x

x2 - x
Use the expressions N1 (x)  ;
x 2  x1
x - x1
N 2 (x) 
x 2  x1
And check
x2 - x x - x1
N1 (x)  N 2 (x)   1
x 2  x1 x 2  x1
x2 - x x - x1
and N1 (x) x1  N 2 (x) x 2  x1  x2  x
x 2  x1 x 2  x1 27
1D LINEAR SHAPE FUNCTIONS ON MASTER ELEMENT

1 1

N1
N2
1 2


N1 (x) 
1-  
Node at which N1 is 0

N2 (x) 
 -1 
2 28
1D QUADRATIC MASTER ELEMENT

29
30
A slightly fancier assumption:
displacement varying quadratically inside each bar
N1 ( ) N 2 ( ) N 3 ( )

1 1 
3
El #1

d  N1 ( )d1  N 2 ( )d 2  N3 ( )d 3

This is a quadratic finite element in

1D and it has three nodes and three
associated shape functions per element.
31
 2D elements - bilinear element

v3
4 3
u3
v4
η v2 η
ζ
u4 u2
ζ

y v1
u1 1 2
x

32
2D linear rectangular element

33
 Shape Functions of rectangular linear element

Bilinear element has shape functions generated by multiplying

linear expressions in ζ and η direction. Due to multiplication,
the product is not a linear function:
N1 = Nζ1 . Nη1 = (1 - ζ).(1 - η ) / 4
N2 = Nζ2 . Nη1 = (1 + ζ).(1 - η ) / 4
N3 = Nζ2 . Nη2 = (1 + ζ).(1 + η ) / 4
N4 = Nζ1 . Nη2 = (1 - ζ).(1 + η ) / 4
,

34