UNIT 3: Interpolation Models and Higher Order Elements

Interpolation polynomials
 The primary idea of FEM is piecewise approximation that is, the complicated problem is
divided into small regions (finite elements) of interest and the approximate solution is
obtained using simple function.
 Thus, the important solution is in choosing the simple function for the solution in each
element.
 The functions used to represent the behavior of the solutions within an element are called
interpolation functions or approximating functions or interpolation models.

There are two types of displacement function used,

a) Polynomial function
b) Trigonometric function
Generally polynomial functions are preferred due to ease of mathematical calculation. For
selection of the displacement model, three condition has to be satisfied.
1. Type and order of the function: Type indicates either polynomial or trigonometric
functions. But for finite element analysis, preferabily polynomial functions are used. If
polynomial of nth order is used, then the solution will be realistic but the computational
time and cost involved are more. The solution obtained is approximate and are within the
limits. Suppose if the polynomial is truncated to a finite order, calculation becomes easy.
2. Displacement model selected should give particular magnitude of displacement at nodal
points. These are governed by coefficients of polynomial.
3. For the displacement model selected to reach exact solution, certain condition are to be
satisfied called convergence requirements.
Reasons for using polynomial for displacement function
Polynomial interpolation function have been most widely used due to the following reasons.
1. It is easier to formulate and computerize the finite element equations with polynomial
type interpolation function. Specifically it is easier to perform differentiations or
integration with polynomials.
 2. By increasing the order of polynomial, it is possible to increase the accuracy of the
results. Theoretically, a polynomial of infinite order corresponds to the exact solution.
But in Practice we use polynomials of finite order only as an approximation.
 If the interpolation polynomial is of the order one, then it is called a linear element.
 A linear element is called a simplex element if the number of nodes in the element is 2,3,
and 4 in one, two and three dimensions respectively.
 If the interpolation polynomial is of order two or more, the element is called as higher order
element.
 In higher order elements, some secondary nodes are introduced in addition to the primary
nodes in order to match the number of nodal degrees of freedom with the number of
constants in the interpolation polynomial.
 If the order of interpolation polynomial is fixed, the discretization of the region can be
improved by two methods.
1. If the locations of the nodes are altered without changing the total number of
elements, the method is known as r-method.
2. If the number of elements are increased, it is called as h-method.
3. If improvement in accuracy is sought by increasing the order of the interpolation of
polynomial, the method is known as p-method.
 Iso-parametric elements are developed for problems involving curved boundaries. To
derive the iso-parametric element equation, each element shape is represented in local or
natural coordinate system. Then the interpolation or shape function are expressed in terms
of the natural coordinates.

POLYNOMIAL FORM OF INTERPOLATION FUNCTIONS

The polynomial form of interpolation function in one, two and three dimensional finite elements
are represented as
  x   1   2 x  3 x 2 +.......+ m x n
For n=1 (Linear model)
  x   1   2 x .....1D 
  x, y   1   2 x   3 y ..... 2 D 
  x, y, z   1   2 x   3 y+ 4 z .....3D 
For n=2 (Quadratic model)
  x   1   2 x  3 x 2 .....1D 
  x, y   1   2 x   3 y+ 4 x 2 + 5 y 2 + 6 xy ..... 2 D 
  x, y, z   1   2 x  3 y+ 4 z+5 x 2 + 6 y 2 + 7 z 2  +8 xy   9 yz  10 xz .....3D 

For n=3 (Cubic model)

  x   1   2 x   3 x 2   4 x3 .....1D 
  x, y   1   2 x   3 y+ 4 x 2 + 5 y 2 + 6 xy+ 7 x3 + 8 y 3 + 9 x 2 y +10 xy 2 ..... 2 D 
  x, y, z   1   2 x   3 y+ 4 z+ 5 x 2 + 6 y 2 + 7 z 2   8 xy   9 yz  10 xz .....3D 
....+11 x3 +12 y 3 +13 z 3 +14 x 2 y +15 x 2 z +16 y 2 z +17 xy 2 +18 yz 2 +19 yz 2 + 20 xyz

The simplex elements are those for which the approximation polynomial consists of constant and
linear terms. (Linear model)
Eg: Simplex element in 2 dimensions is a triangle with 3 nodes.
The complex elements are those for which the approximating polynomials consists of quadratic,
cubic, and higher order terms. The complex elements may have the same shapes as the simplex
elements but will have additional boundary nodes and sometimes internal nodes.
Eg: A triangular element with three corner nodes and three mid side nodes.

INTERPOLATION POLYNOMIAL IN TERMS OF NODAL

DEGREES OF FREEDOM
 The basic idea of the finite element method is to consider a body as composed of several
elements (or subdivisions) that are connected at specified node points.
 The unknown solution or the field variable (e.g.. displacement, pressure, or temperature)
inside any finite element is assumed to be given by a simple function in terms of the
nodal values of that element.
 The nodal values of the solution, also known as nodal degrees of freedom, are treated as
unknowns in formulating the system or overall equations.
 The solution of the system equations (e.g.. force equilibrium equations or thermal
equilibrium equations or continuity equations) gives the values of the unknown nodal
degrees of freedom.
 Once the nodal degrees of freedom are known, the solution within any finite element (and
hence within the complete body) will also be known to us.
 Thus, we need to express the approximating polynomial in terms of the nodal degrees of
freedom of a typical finite element ‘e’.
  For this, let the finite element have M nodes. We can evaluate the values of the field
variable at the nodes by substituting the nodal coordinates into the polynomial equations
For example, the simplex element and can be expressed as
  x   
T

 1 
 
 
where   1 x x 2 ... x n 
T
and    2 
 : 
 n 1 
The evaluation of above equation at the various nodes of element ‘e’ gives
  T  at node 1 
   at node 1 
e
 
 
   at node 2          T  
T
 e  at node 2
       
   
  at node M    T 
    at node M  
Where    is the vector of nodal values of the field variable corresponding to element ‘e’ and the square
e

matrix [ ] can be identified by inverting the above equation

1
      e
T

 
Substituting in the main equation, it gives
1
          e    N    e 
T T

where
1
N  
T
 
 
The above equation expresses the interpolating polynomial inside any finite element in terms of
the nodal unknowns of that element   e 

One dimensional elements

Displacement model for the case of axial or longitudinal displacement is
u  x   a1  a2 x  a3 x 2  a4 x3  .......  am x n
Transverse deflection is
v  x   b1  b2 x  b3 x 2  b4 x 3  .......  bm x n
Where ‘u’ and ‘v’ are the field variables. And a1, a2, a3…….an and b1, b2, b3 ,……. bm are generalized
coordinates.
n = Degree of polynomial
m = Number of generalized coordinates = n+1
Two dimensional CST elements
Generation of the displacement models for the CST
u  x, y   a1  a2 x  a3 y  a4 x 2  a5 xy  a6 y 2  .......   m y x
v  x, y   b1  b2 x  b3 y  b4 x 2  b5 xy  b6 y 2  .......  bm y x

Pascal’s triangle for CST elements

Generation of displacement models for quadrilateral elements

In case of axis symmetric elements

u  x, z   a1  a2 r  a3 z  a4 r 2  a5rz  a6 z 2  .......  am z n
w  x, z   b1  b2 r  b3 z  b4 r 2  b5 rz  b6 z 2  .......  bm z n
In case of plate bending element
w  x, z   a1  a2 x  a3 z  a4 x 2  a5 xz  a6 z 2  .......  am z n

Selection of the order of the interpolation polynomial

1. The number of the generalized coordinates should be equal to the number of nodal degree of
freedom of the elements.
2. The pattern of variation of interpolation polynomial should be independent of the local
coordinate system.
3. The interpolation polynomial should satisfy the convergence requirement.
The field variables representation within an element and hence the polynomial should not change with a
change in the local coordinate system. This property is called geometric isotropy or geometric Invariance
or spatial isotropy.
The polynomial should contain terms that do not violate symmetry which is known as Pascal’s triangle.

SHAPE FUNCTION
The shape function is the function which interpolates the solution between the discrete values
obtained at the mesh nodes. Therefore, appropriate functions have to be used and, and that function
involving low order polynomials are typically chosen as shape functions.

Shape function of 1D bar element

Consider a 1D linear bar element having two nodes with 3-dof at each node. Let u1, v1, w1 and u2,
v2, w2 be the nodal displacements at node 1 and node 2.
Let N1 and N2 be the shape function at node 1 and node 2

The element displacement is given by

 u  x    N1u1  N 2u2 
U  x    v  x     N1v1  N 2v2 
w  x   N w  N w 
   1 1 2 2

 u1 
v 
 u  x    N1 0 0 N2 0 0  1 
    w 
 v  x    0 N1 0 0 N2 0   1 
w  x   0 u
   0 N1 0 0 N 2   2 
 v2 
 
 w2 
U    N  x  q
Where U   Field variable
 N  x    Nodal function matrix
q  Nodal displacement vector
Three-Noded Triangular Element (TRIA 3) – Shape function

The element displacement is given by

u  x, y   N1u1  N 2u2  N 3u3
v  x, y   N1v1  N 2v2  N 3v3
In matrix form, it is written as
 u1 
v 
 1
u  x, y    N1 0 N2 0 N3 0  u2 
   
 v  x, y    0 N1 0 N2 0 N 3   v2 
u3 
 
 v3 

Coordinate system
There are two types of system
1. Cartesian (or) Rectangualar (Plane Surface) – (x, y, z)
2. Polar or Cylindrical (Curved Surface) – (r, θ, z)

In FEM, there are three types of coordinate system

1. Global Coordinate system (x, y, z) or (r, θ, z)
2. Local Coordinate system
3. Natural coordinate system
Four-Noded Quadrilateral Element (QUAD 4),
Shape Functions of 2, 3, and 4 Noded bar element,
Serendipity family,
Hexahedral elements,
Lagrange family,
Shape functions for Higher Order Elements.
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner
Scanned by CamScanner