FEM Dr. T.

FINITE ELEMENT METHOD – BASIC DEFINITIONS

1. DISCRETIZATION PROCESS:
The various considerations to be taken in the discretization process are as given below:
1.1 Type of Elements or Various Elements Shapes or Element Library: The discretization of the
structure or body into finite elements forms the first step in finite element analysis of a complicated
structural system. These elements coincide with the geometry of the structure and represent the geometry
and mechanical properties in the regions. Therefore, the first decision the engineer must make is to select
the shape of the basic elements to be used in the analysis. The selection of finite element in FEM is
depends on the following factors. They are:
1. Dimension of the problem.
2. Shape of the body or continuum.
3. Accuracy of the results required.
4. Cost and time for the analysis.
Fig. 1.1 shows some typical elements used to discretize the continuum.

1 2 1 3 2
(a) 2-noded element (b) 3-noded element

(a) One – dimensional bar element

3
3 7 6
6 5 4
4 2
1 2 8
1 5

(a) (b) (c) (d)

(b) Two – dimensional elements

(c) Three – dimensional elements

Fig. 1.1: Typical elements
1D Elements: When the geometry, material properties and field variable of the problem is described in
terms of only one spatial coordinate (for example x- coordinate), then line elements shown in Fig. 1(a) can
be used. A one – dimensional element may be represented by a straight line whose ends are nodal points
as shown in Fig. 1(a). These nodal points, numbered 1 and 2 are called external nodes because they
represent connecting points to the adjacent elements. Some applications require additional nodal point
such as node-3 as shown in Fig. 1a(b). Because no connection to other elements occurs at this node, it is
called internal node.
2D Elements: When the configuration and other parameters of the problems are described in terms of two
spatial coordinates, then two-dimensional elements are used. Triangular or quadrilateral elements (Fig.1b)
are used for two – dimensional problems such as plane stress or plain strain or plate bending problems.
The corner nodes of these elements are called primary external nodes and additional nodes occur on the
sides of the elements, like 4, 5 and 6 in 6 – nodded triangular element or 5, 6, 7 and 8 in 8 – nodded
curved element are called secondary external nodes. Some times internal node such as 7 as shown in
Fig.1b (b) is also used for these elements.
3D Elements: When the geometry, material properties and other parameters of the problems are
described by three spatial coordinates, then three-dimensional elements shown in Fig. 1(c) are used. For 3

65
FEM Dr. T.N

– dimensional problems, tetrahedron or hexahedron elements are used. A tetrahedron has 4 primary
external nodes and hexahedron has 8 primary external nodes.
1.2 Size of Elements: The size of elements influences the convergence of the solution directly and hence
it has to be chosen with care. If the size of the elements is small, the final solution is expected to be more
accurate. However, the use of elements of smaller size will take more computational time. Sometimes, we
may have to use elements of different sizes in the same body. For example, in the case of stress analysis
of the plate in Figure 1.2(a), the size of all the elements can be approximately the same, as shown in
Figure 1.2(b). However, in the case of stress analysis of a plate with a hole shown in Figure 1.2(c),
elements of different sizes have to be used, as shown in Figure 1.2(d). The size of elements has to be very
small near the hole (where stress concentration is expected) compared to far away places. In general,
whenever steep gradients of the field variable are expected, we have to use a, finer mesh in those regions.
Another characteristic related to the size of elements that affects the finite element solution is the aspect
ratio of the elements. The aspect ratio describes the shape of the element in the assemblage of elements.
For two-dimensional elements, the aspect ratio is taken as the ratio of the largest dimension of the element
to the smallest dimension. Elements with an aspect ratio of nearly unity generally yield best results.

Fig.1.2(b) Finite Element Discretization

Fig.1.2(a) Original Structure

Fig. 1.2(c) Original Structure Fig. 1.2(d) Finite Element Discretization

1.3 Location of Nodes: If the body has no abrupt changes in geometry, material properties, and external
conditions (e.g., load and temperature), the body can be divided into equal subdivisions and hence the
spacing of the nodes can be uniform. On the other hand, if there are any discontinuities in the problem,
nodes have to be introduced at these discontinuities, as shown in Fig 1.3

(a) Discontinuity in Loading

(b) Discontinuity in Geometry

(c) Discontinuity in Material Properties

66
FEM Dr. T.N

2. DISPLACEMENT OR INTERPOLATION FUNCTION:

The basic principle of FEM is to sub – divide the solution domain into number of finite elements and
representing the solution within each element by relatively simple function. This simple function is called
‘displacement function or interpolation function’. The polynomial function is commonly used for this
displacement model in FEM due the following two reasons.
1. The polynomials are easier for the mathematical operation i.e. the differentiation and integration of
polynomials are easy as compared to trigonometric functions.
2. A polynomial of arbitrary order permits us to have a required approximation i.e by selecting the order of
polynomial we can vary the degree of approximation.
The degree of approximation is mainly depends on the order of the polynomial selected for the
displacement model as shown in Fig. 2. In this Fig.2 an exact solution for the displacement u( x ) for a
simple one – dimensional problem is approximated by the following degree of polynomials of the form,
u( x )=α 1 +α 2 x+ α 3 x 2 +−−−−−+α n+1 x n ---(a)
Similarly, for two and three dimensional problems, the polynomial form of interpolation functions can be
expressed as
u( x , y )=α 1 + α 2 x +α 3 y +α 4 x 2 + α 5 xy + α 6 y 2 +−−−−−−+ α m y n --- (b)

u( x , y , z )=α 1 + α 2 x + α 3 y + α 4 z+ α 5 x 2 + α 6 y 2 +α 7 z 2 −−−−−−+ α m z n --- (c)

The greater the number of terms included in the polynomial, the solution is more close to the exact solution
as shown in Fig. 2(a) to 2(c). The coefficient of the polynomial, the α ’s in Eqn. (a) are called generalized
coordinates.

(a) Constant polynomial (b) Linear polynomial

(c) Quadratic polynomial

Fig. 2: Polynomial approximation in one – dimension
In most of the cases, the order of the polynomial in interpolation function is taken as one, two or three.
Thus, the above equations reduce to the following form for various cases.
For Linear Model (i.e. n=1)
For 1D problem:
u( x )=α 1 +α 2 x

For 2D problem:
u( x , y )=α 1 + α 2 x +α 3 y

67
FEM Dr. T.N

For 3D problem:
u( x , y , z )=α 1 + α 2 x + α 3 y + α 4 z
For quadratic Model (n=2)

For 1D problem:
u( x )=α 1 +α 2 x+ α 3 x 2

For 2D problem:
u( x , y )=α 1 + α 2 x +α 3 y +α 4 x 2 + α 5 xy + α 6 y 2

For 3D problem:
u( x , y , z )=α 1 + α 2 x + α 3 y + α 4 z+ α 5 x 2 + α 6 y 2 +α 7 z 2 + α 8 xy + α 9 yz + α 10 zx

3. NODAL DEGREES OF FREEDOM: 3

6 5
7
1 2
4
Fig. 4: 7 – nodded triangular element
The nodal displacement, rotation and/or strains necessary to specify completely the deformation of the
element are referred as degrees of freedom of the element. The nodal degrees of freedom are the nodal
displacement, rotations and/or strains corresponding to the external nodes of an element. On the other
hand, the internal degrees of freedom are corresponds to the internal node of an element.
For example, in a triangular element shown in Fig. 4, the nodes 1, 2 and 3 are the primary external nodes
and nodes 4, 5 and 6 are secondary external nodes. The node 7 is an internal node. The displacements
corresponding to these primary and secondary external nodes 1 to 6 are called nodal degrees of freedom
and the displacement corresponding to the internal node 7 is called internal degrees of freedom.

4 SELECTION OF THE ORDER OF THE INTERPOLATION POLYNOMIAL: While selecting the order of
the polynomial in interpolation function, the following considerations have to be taken into account:
1. The interpolation polynomial should satisfy, as far as possible, the convergence requirements.
2. The pattern of variation of field variable resulting from the polynomial model should be independent of
local coordinate system.
3. The number of generalized coordinates ( ’s) should be equal to the number of nodal degrees of
freedom of the element.

5. SIMPLEX, COMPLEX AND MULTIPLEX ELEMENTS: Finite elements can be classified into three
categories as simplex, complex, and multiplex elements depending on the geometry of the element and
the order of the polynomial used in the interpolation function.
Simplex Elements: The simplex elements are those for which the approximating polynomial consists of
constant and linear terms. Thus, the following polynomials represent the simplex functions for one-, two-,
and three-dimensional elements.
For 1D problem:
u( x )=α 1 +α 2 x --- (a)

For 2D problem:
u( x , y )=α 1 + α 2 x +α 3 y --- (b)

For 2D problem:
u( x , y , z )=α 1 + α 2 x + α 3 y + α 4 z ---(c)
The simplex element in one dimension is a 2-noded bar element and for two dimensions is a triangle with
three nodes (corners).
Complex Elements: The complex elements are those for which the approximating polynomial consists of
quadratic, cubic, and higher order terms, according to the need, in addition to the constant and linear
terms. Thus, the following polynomials denote complex functions.

For 1D problem:
u( x )=α 1 +α 2 x+ α 3 x 2 ---(d)
2 2
For 2D problem:
u( x , y )=α 1 + α 2 x +α 3 y +α 4 x + α 5 xy + α 6 y ---(e)
2 2 2
For 3D problem:
u( x , y , z )=α + α x + α y + α z+ α x + α y +α z + α xy + α yz + α zx
1 2 3 4 5 6 7 8 9 10
The complex elements may have the same shape as the simplex elements but they have additional
external and internal nodes. For example, eq. (e) represents the quadratic interpolation function for a two-
dimensional element. Since this equation has 6 ’s, the corresponding element should have six nodes and
six-noded triangular element forms a complex element. 3
68
6 5
1 3 2

1 2
FEM Dr. T.N

(a) For 1 – D problem (b) For 2 – D problem

Fig (a) Complex Elements
Multiplex Elements: Multiplex elements are those whose boundaries are parallel to the coordinate axes to
achieve interelement continuity as shown in Fig. (b) and whose polynomials contain higher order terms.
y

x
Fig. (b) Multiplex Element

6. Higher Order Elements:

In certain applications, additional degrees of freedom behind the minimum number may be included for any
element. This can be achieved by adding secondary external nodes. Then, the elements with these
additional degrees of freedom are called as higher order elements. Fig. 5 shows the higher order elements
used in 1 – D and 2 – D problems.
3

6 5
1 3 2
1 2
4
(a) For 1 – D problem (b) For 2 – D problem

Fig. 5: Higher order elements

7. CONVERGENCE REQUIREMENTS: In FEM sequence of approximate solutions are obtained as the
element size is reduced successively. These solutions are converge to the exact solutions if the
interpolation polynomial satisfies the following convergence requirements:

1. The displacement model must be continues within the element and the displacement must be
compatible between adjacent elements: The condition can be satisfied by choosing polynomials which are
inherently continues functions for the displacement model. The second condition indicates that when the
elements deforms there should not be any discontinuity between the elements (elements should not
overlap or separate).

2. The displacement model must be capable of representing rigid body displacement of the element: This
condition states that when the nodes are given such displacements corresponding to a rigid body motion,
the element should not experience any strains and hence leads zero nodal forces. The constant term
present in the polynomial functions satisfy this condition.

3. The displacement model must be capable of representing constant strain state within the element: When
the elements are made smaller and smaller size, the strain in each elements approach constant value. In
such case, the assumed displacement model should include the terms to represent the constant strain
α
state within the element. For one, two and three - dimensional elasticity problems, the linear terms (i.e. 2
) present in the polynomial functions satisfy this requirement.
In order to satisfy the above mentioned three requirements, the displacement model must contain at least
constant and linear terms in the polynomials.

69
FEM Dr. T.N

8. GEOMETRIC INVARIANCE OR GEOMETRIC ISOTROPY:

In addition to the convergence and compatibility requirements, one of the important consideration in
choosing proper terms in the polynomial displacement model is that the element have no preferred
direction. That is the displacement shapes should not change with change in local coordinate system. This
property of the displacement model is known as geometric invariance or geometric isotropy.
Geometric isotropy is achieved if the polynomial selected in the displacement model includes all the terms
i.e. if the polynomial is complete one. However, the geometric isotropy can also be achieved if the
polynomial is balanced in case all the terms can’t be included. This balanced representation can be
achieved using ‘Pascal Triangle’ shown in Fig. 7 for the case of two – dimensional polynomials.

Fig. 7: Pascal triangle for 2 – D problems

For the polynomial functions used in displacement model to be balanced one, it should not
includes any terms from one side of the axis of the symmetry of the triangle without including its
counter part from the other side. For example, if we wish to construct a cubic model with 8 terms,
the following are the geometrically isotropic.
3 3
1. All the constant, linear and quadratic terms plus x and y .

i.e.
u( x , y )=α 1 + α 2 x +α 3 y +α 4 x 2 + α 5 xy + α 6 y 2 +α 7 x 3 + α 8 y 3
2. All the constant, linear and quadratic terms plus x2 y and xy 2 .
2 2 2 2
i.e.
u( x , y )=α 1 +α 2 x +α 3 y +α 4 x +α 5 xy +α 6 y +α 7 x y +α 8 xy
The displacement model formed by any of the above two balanced polynomials will have the
geometric isotropy.

9. Coordinate System:
The following coordinate systems are used in FEM.
1. Global coordinate system
2. Local coordinate system
3. Natural coordinate system
Global coordinate system: The coordinate system used to define entire body is called global
coordinate system. Fig. 8(a) shows the cartesian global coordinate system used for some of the
typical elements. y
4
2 4

1 3 5
1 2 3 7
x1 x2 x3 2 6
70
1 x
3 5
FEM Dr. T.N

Fig. 8(a): Global coordinate system

Local coordinate system: A local coordinate system is one that is defined for a particular
element and not for the entire body. This local coordinate system simplifies the derivation of
element formulations. For example, for typical truss elements shown in Fig. 8(a), the local
coordinates ( x̄ , ȳ ) may be as shown in Fig. 8(b). However, for this truss problem the final
equations derived in terms of local coordinate are to be transformed to the common global
coordinate system.

Fig. 8(b): Local coordinate system

Natural Coordinate Systems: The natural coordinate system is a local system used to specify
the position of a point within the element by a set of dimensionless numbers. The magnitudes of
these number never exceed unity. Natural coordinate systems are so defined that, their values
are zero or one at the nodal points of an element. This coordinate system facilitates the
integration while computing element stiffness matrix.
Fig. 8 shows some of the natural coordinate systems used for one, two and three – dimensional
elements.
le
l2 l1 
1 2 1 2
x1 P ( L1 , L2 ) x2   1  1
(1,0) (0,1)  ( L1 , L2 )

(a) For 1 – D bar elements

(0,0,1)

(-1,+1) (+1,+1)
3 4 3
L2 0 L1 0

P ( L1 , L2 , L3 )

1 L3 0 2
1 2
(1,0,0) (0,1,0) (-1,-1) (+1,-1)
(b) For 2 – D elements
4 (0,0,0,1)

P( L1 , L2 , L3 , L4 )

71
3 (0,0,1,0)
1
(1,0,0,0)
2 (0,1,0,0)
FEM Dr. T.N

(c) For 3 – D element

Fig. 8: Natural coordinate systems
Advantage of natural coordinate system:
1. Natural coordinate systems generalize and simplify the element formulations
2. They simplify the integration and differentiation, which are required to obtain the element
stiffness matrix and load vectors.

Area coordinate: If any point P(x,y) divides the total area of the triangular element A into A 1, A2
and A3 as shown in Fig. 9, then areas A1, A2 and A3 are called area coordinates. These are also
L ,L
natural coordinates and these can be used to express the natural coordinates 1 2
& L3 as
3
A1 A2 A3
L1 = ; L2 = ; and L3=
A A A A2 A1
P ( L1 , L2 , L3 )
A3
1 2
Fig. 9
10. Shape Function or Interpolation Function:
In FEM, it is necessary in many cases to deal with functions whose analytical form is totally
unknown. In such case, the given function is replaced by another function which can be more
easily handled. This operation of replacing a given function by simple one is known as
Interpolation or shape function.
Shape function is a function used to interpolate nodal variables to field variables at any point
inside the element. Using these shape functions, the field variable can be expressed in terms of
nodal variables as
u=N 1 u1 + N 2 u 2 +−−−N n un
where n is the number of nodes in a element.
Properties of Shape functions:
The shape functions, which are used to interpolate nodal variable to field variable, are having the
following properties
1. The shape functions are having unit value at one nodal point and zero value at all other nodal
points For example, in a 1 – D bar element having nodal points (i , j) , the above mentioned
property of shape functions can be written in mathematical form as

At node i ,
N i =1 N j=0 i j

At node j , N i =0 N j=1
Ni Nj

1 1
i j

2. The sum of the shape functions is equal to unity

72
FEM Dr. T.N

N + N =1
.i.e i j
3. The derivatives of a shape function is constant
dN i dN j
= =
i.e. dx dx Constant

11. Derivation of Expressions for Shape Functions:

(a) For a 2-noded bar element:
(i) In terms of natural coordinate:


u1 u2
1 2
  1  1

Fig. shows a two nodded bar element. The values of natural coordinates corresponding to node
u u
1 and 2 are as shown in Fig. Let the displacement at node 1 and 2 be 1 and 2 respectively.
Then the field variable (i.e. displacement) u at any point inside the element can be expressed in
natural coordinate as
u=α 1 +α 2 ξ ---(a)

u=[ 1
α
ξ] 1
α2{ } ---(b)
Substitute the values of ξ corresponding to node 1 and 2 into Eqn. (a), we get nodal
displacements
i.e. at node 1, ξ=−1 and
u=u1
∴ u1 =α 1 −α 2 ---(1)
at node 2, ξ =+1 and
u=u2
∴ u2 =α 1 +α 2 ---(2)
Add Eqn (1) and (2)
u1 + u2=2 α 1
u u
∴ α 1= 1 + 2
2 2 ---(i)
Substitute into Eqn (1) we get
u1 u2
α 2 =− +
2 2 ---(ii)
Eqns. (i) and (ii) can be written in matrix form as

73
FEM Dr. T.N

[ ]{ }
1 1

{ }
α1
α2
=
−
2
1
2
2
1
2
u1
u2

Substitute into Eqn. (b)

[ ]{ }
1 1
2 2 u1
1 1 u2
−
u=[ 1 ξ] 2 2

u= [ 1
2
(1−ξ )
1
2
u
( 1+ξ ) 1
u2 ]{ }
u=[ N 1 N2 ]
{}
u1
u2
=N 1 u1 + N 2 u2
---(c)
1 1
N 1 = ( 1−ξ ) N 2 = ( 1+ξ )
where 2 and 2 are shape functions for 2 – nodded bar element and is
u u
interpolate the nodal variables 1 and 2 to field variable u as seen from Eqn. (c).

(ii) In terms of Cartesian coordinates

u1 u2
1 P(x) 2
x 0 x l e
le

Fig. shows a two nodded bar element. The values of cartesian coordinates corresponding to
node 1 and 2 are as shown in Fig. Let the displacement at node 1 and 2 be 1 and 2
u u
respectively. Then the field variable (i.e. displacement) u at any point inside the element can be
expressed in terms of cartesian coordinate as
u=α 1 +α 2 x ---(a)

u=[ 1 x]
{ }
α1
α2
---(b)

Let us consider a coordinate system as shown in Fig. (i.e. at node1, x=0 and at node2
x=l e ).
Then substitution of the nodal coordinates yields nodal displacements as
i.e. at node 1, x=0 and
u=u1

∴ u1 =α 1
α =u 1
or 1 ---(i)

at node 2,
x=l e and u=u2
∴ u2 =α 1 +α 2 l e=u1 +α 2 l e ∵ α 1=u1
u1 u2
∴ α 2=− +
le l e ---(ii)

74
FEM Dr. T.N

Eqns (i) and (ii) can be written in matrix form as

[ ]{ }
1 0
u1
{ }
α1
α2
=
−1 1
le le
u2
---(c)

Substitute in to Eqn.(b), we get

[ ]{ }
1 0
u1
u=[ 1 x ]
−1 1
le le u 2

[
u= 1−
x
le
x
le ]{ } [
u1
u2
= N1 N2 ]
{}
u1
u2
=N 1 u1 + N 2 u 2

x x
N 1 =1− N 2=
where,
l e and l e are shape functions for 2–noded bar element and is interpolate
u u
the nodal variables 1 and 2 to field variable u as seen from the above equation. The variation
of shape functions along the length of element is as shown in Fig. below.

 

 

1 2

Note: Students are advised to practice the derivation of shape functions either in terms of
cartesian or natural coordinate system and to remember the expressions for shape functions in
terms of both coordinates.

(b) For a 3–noded bar element (Quadratic bar element):

u1 ξ u2
u3
1 3 2
ξ=−1 ξ=0 ξ =+1
Fig. shows a 1 – D three nodded bar element. The values of natural coordinates corresponding
to node 1, 2 and 3 are as shown in Fig. Let the displacement at node 1, 2 and 3 be
u1 , u2 and
u3 respectively. Then the field variable (i.e. displacement) u at any point inside the element can
be expressed in natural coordinate as
u=α 1 +α 2 ξ+ α 3 ξ2 ---(a)

{}
α1
u=[ 1 ξ ξ 2
] α2
α3
---(b)
Substitute the values of ξ corresponding to node 1, 2 and 3 into Eqn. (a), we get nodal
displacement
75
FEM Dr. T.N

i.e. at node 1, ξ=−1 and

u=u1
∴ u1 =α 1 −α 2 + α 3 ---(1)
at node 2, ξ =+1 and
u=u2
∴ u2 =α 1 +α 2 +α 3 ---(2)
at node 3, ξ=0 and
u=u3
∴ u3 =α 1 ---(3)
∴ α 1=u 3 ---(i)
Subtract Eqn (1) from (2)
u2 −u1 =2 α 2
u u
∴ α 2=− 1 + 2
2 2 ---(ii)
α & α 2 into Eqn (1) we get
Substitute 1

(
u1 =u3 − − +
u 1 u2
2 2
+α 3)
u u
∴ α 3= 1 + 2 −u 3
2 2 ---(iii)
Eqns. (i) to (iii) can be written in matrix form as

{ }[ ]{ }
0 0 1
α1 1 1 u1
− 0
α2 = 2 2 u2
α3 1 1
−1 u 3
2 2
Substituting into Eqn. (b), we get

[ ]{ }
0 0 1
1 1 u1
2 − 0
[
u= 1 ξ ξ ] 2 2 u2
1 1
−1 u3
2 2

]{ }
u1
[(
1 1
u= − ξ+ ξ 2
2 2 ) ( 1
2
1
ξ+ ξ2
2 ) ( 1−ξ )
2
u2
u3

{}
u1
u=[ N 1 N2 N3 ] u2
u3
where,

( 1
2
1
2
1
)
N 1 = − ξ+ ξ 2 = ξ( ξ−1 )
2

(
1
2
1
2
1
)
N 2 = ξ+ ξ 2 = ξ( ξ +1 )
2

76
FEM Dr. T.N

N 3 =( 1−ξ 2 )
are the shape functions for the three nodded bar element and are varies over the length of
element as shown below.

1
N 1 = ξ (ξ−1)
2

N 3 =(1−ξ 2 )

1
N 2 = ξ (ξ+ 1)
2

(c) For a 3–noded triangular element (CST element):

(i) In terms of natural coordinates:

3 (0,0)
 0
  2 (0,1)
1
(1,0)
Fig. shows a 3-noded triangular element. The values of natural coordinates corresponding to
node 1, 2 and 3 are as shown in Fig. Let the displacement at node 1, 2 and 3 in x-direction be
u1
u u
, 2 and 3 respectively. Then the field variable (i.e. displacement) u at any point inside the
element can be expressed in natural coordinate as
u=α 1 +α 2 ξ+ α 3 η ---(a)

{}
α1
u=[ 1 ξ η ] α2
α3
---(b)
Substitute the values of ξ and η corresponding to node 1, 2 and 3 into Eqn. (a), we get the
nodal displacements
i.e. at node 1, ξ=1 ;η=0 and
u=u1
∴ u1 =α 1 +α 2 ---(1)
at node 2, ξ=0 ;η=1 and
u=u2
∴ u2 =α 1 +α 3 ---(2)
at node 3, ξ=0 ;η=0 and
u=u3

77
FEM Dr. T.N

∴ u3 =α 1
∴ α 1=u 3 ---(i)
Substitute into Eqn. (1), we get
u1 =u3 + α 2
∴ α 2=u 1−u 3 ---(ii)
Similarly, substitute into Eqn. (2), we get
α 3 =u2 −u3 ---(iii)
Eqns. (i) to (iii) can be written in matrix form as

{ }[ ]{ }
α1 0 0 1 u1
α2 = 1 0 −1 u2
α3 0 1 −1 u
3
Substitute into Eqn. (b), we get

[ ]{ } {}
0 0 1 u1 u1
u=[ 1 ξ η] 1 0 −1 u2 [ξ η ( 1−ξ−η ) ] u2
0 1 −1 u u3
3
=

{}
u1
u=[ N 1 N2 N3 ] u2
u3

where, 1
N =ξ ; N =η2 and 3
N =1−ξ−η are the shape functions for a three nodded triangular
(CST) element.
(ii) In terms of cartesian coordinate:

y 3 ( x3 , y 3 )

P ( x, y )

2 ( x2 , y 2 )
1
( x1 , y1 )
x

Fig. shows a 3-noded triangular element. The values of natural coordinates corresponding to
node 1, 2 and 3 are as shown in Fig. Let the displacement at node 1, 2 and 3 in x-direction be
u1
u u
, 2 and 3 respectively. Then the field variable (i.e. displacement) u at any point inside the
element can be expressed in terms of cartesian coordinate as
u=α 1 +α 2 x + α 3 y ---(a)

{}
α1
u=[ 1 x y ] α2
α3
---(b)
Substitute the values of x and y correspond to node 1, 2 and 3 into Eqn. (a), we get the nodal
displacements

78
FEM Dr. T.N

i.e. at node 1,
x=x 1 ; y= y 1 and u=u1
∴ u1 =α 1 +α 2 x1 +α 3 y 1 ---(1)
at node 2,
x=x 2 ; y= y 2 and u=u2
∴ u2 =α 1 +α 2 x 2 +α 3 y 2 ---(2)
at node 3,
x=x 3 ; y= y 3 and u=u3
∴ u3 =α 1 +α 2 x 3 +α 3 y 3 ---(3)
Eqns. (1) – (3) can be written in matrix form as

{ }[ ]{ }
u1 1 x1 y1 α1
u2 = 1 x2 y2 α2
u3 1 x3 y3 α3

or { q }=[ A ] { α }
∴ { α }=[ A ]−1 { q } ---(c)
The inversion of matrix A is given by

[ ]
a1 a2 a3
−1 1
A = b b2 b3
2 Ae 1
c1 c2 c3
---(d)
1 x1 y1
1
A e= |1 x 2 y 2|
2
where,
1 x3 y3
a 1=x 2 y 3 −x 3 y 2 b1 = y 2− y 3 c1 =x3 −x 2
a 2=x 3 y 1 −x 1 y 3 b2 = y 3 − y 1 c 2 =x1 −x 3
a 2=x 1 y 2 −x 2 y 1 b 3 = y 1− y 2 c 3=x 2 −x 1
where, 1,2,3 are in cyclic order.

Note:
1 x1 y1
x y2 a 1=|1 x2 y 2|
a 1=M 11 =(−1)1+1| 2 |=+ 1( x 2 y 3 −x 3 y 2 )
x3 y3 1 x3 y3
1 x1 y1
1 y2 b 1=|1 x2 y 2|
b 1=M 12=(−1 )1+2| |=−1( y 3 − y 2 )= y 2− y 3
1 y3 1 x3 y3
1 x1 y1
1 x2 c 1 =|1 x2 y 2|
c 1 =M 13=(−1)1+3| |=x 3−x 2
1 x3 1 x3 y3

Substitute Eqn. (d) into (c), we get

{} [ ]{ }
α1 a1 a2 a3 u1
1
α2 = b b2 b3 u2
2 Ae 1
α3 c1 c2 c3 u3
---(e)
substitute Eqn. (e) into Eqn. (b), we get

79
FEM Dr. T.N

[ ]{ }
a1 a2 a3 u 1
1
u=[ 1 x y] b b2 b3 u 2
2 Ae 1
c1 c2 c3 u3

]{ }
u1
u=
[
1
( a +b x+ c1 y )
2 Ae 1 1
1
( a +b x+ c2 y )
2 Ae 2 2
1
( a +b x+ c 3 y ) u 2
2 Ae 3 3
u3

{}
u1
u=[ N 1 N2 N3 ] u 2 =N 1 u1 + N 2 u 2 + N 3 u 3
u3
1
N i= (a +b x +c i y )
where shape functions
2 Ae i i (i=1 , 2, 3 )

(d) For 4-noded Tetrahedron Element:

y 4 ( x4 , y 4 , z4 )

3 ( x 3 , y 3 , z 3)
P( x , y , z)

1 2 ( x2 , y2 , z2 )
( x1 , y1 , z1 )
x

z
Fig. shows a 4-noded tetrahedron element. The values of natural coordinates corresponding to
node 1, 2,3 and 4 are as shown in Fig. Let the displacement at node 1, 2,3 and 4 in x-direction
u u u u
be 1 , 2 , 3 and 4 respectively. Then the field variable (i.e. displacement) u at any point
inside the element can be expressed in terms of cartesian coordinate as
u=α 1 +α 2 x + α 3 y + α 4 z ---(a)

{}
α1
α
u=[ 1 x y z] 2
α3
α4
---(b)
Substitute the values of x and y correspond to node 1, 2 and 3 into Eqn. (a), we get the nodal
displacements
u1 =α 1 +α 2 x 1 +α 3 y 1 + α 4 z 1 ---(1)
u2 =α 1 +α 2 x 2 +α 3 y 2 + α 4 z 2 ---(2)
u3 =α 1 + α 2 x 3 + α 3 y 3 +α 4 z 3 ---(3)
u 4 =α 1 +α 2 x 4 + α 3 y 4 + α 4 z 4 ---(4)
Eqns. (1) – (4) can be written in matrix form as

80
FEM Dr. T.N

{ }[ ]{ }
u1 1 x1 y1 z1 α 1
u2 1 x2 y2 z2 α 2
=
u3 1 x3 y3 z3 α 3
u4 1 x4 y4 z4 α4

or { q }=[ A ] { α }
∴ { α }=[ A ]−1 { q } ---(c)
The inversion of matrix A is given by

[ ]
a1 a2 a3 a4
1 b1 b2 b3 b4
A−1 =
6Ve c1 c2 c3 c4
d1 d2 d3 d4
---(d)
1 x1 y 1 z1
1 1 x2 y 2 z2
V e= | |
6 1 x3 y3 z 3
where,
1 x4 y 4 z 4 ; Volume of element
In general,
x2 y2 z2 1 y2 z2
a 1=M 11 =+ 1| x 3 y 3 z 3 | b 1=M 12=−1|1 y3 z3|
x 4 y4 z4 ; 1 y4 z4 ;
1 x2 z2 1 x2 y2
c 1 =M 13 =+ 1|1 x 3 z 3 | d 1 =M 14=−1|1 x3 y3 |
1 x4 z4 ; 1 x4 y4
where, 1,2,3,4 are in cyclic order.

Substitute Eqn. (d) into (c), we get

{} [ ]{ }
α1 a1 a2 a3 a4 u1
α2 1 b1 b2 b3 b4 u2
=
α3 6V e c1 c2 c3 c4 u3
α4 d1 d2 d3 d4 u4
---(e)
Substitute Eqn. (e) into Eqn. (b), we get

[ ]{ }
a1 a2 a3 a4 u1
1 b1 b2 b3 b4 u2
u=[ 1 x y z]
6Ve c1 c2 c3 c4 u3
d1 d2 d3 d4 u4

{}
u1
u2
u=[ N 1 N2 N3 N4 ] =N 1 u1 + N 2 u2 + N 3 u3 + N 4 u 4
u3
u4

81
FEM Dr. T.N

1
N i= ( a +b x +c i y +d i z )
where shape functions
6V e i i (i=1,2,3,4 )

(e) For a 4-noded quadrilateral element:


(-1,+1) (+1,+1)
4 3

1 2
(+1,-1)
(-1,-1)

Fig. shows a 2 – D, 4 – noded quadrilateral element. The values of natural coordinates

corresponding to node 1, 2, 3 and 4 are as shown in Fig. Let the displacement at node 1, 2, 3
u u u u
and 4 in x-direction be 1 , 2 , 3 and 4 respectively. Then the field variable (i.e.
displacement) u at any point inside the element can be expressed in natural coordinate as
u=α 1 +α 2 ξ+ α 3 η+ α 4 ξη ---(a)

{}
α1
α
u=[ 1 ξ η ξη ] 2
α3
α4
---(b)
Substitute the values of ξ and η corresponding to node 1, 2, 3 and 4 into Eqn. (a), we get the
nodal displacements
i.e. at node 1, ξ=−1 ;η=−1 and
u=u1
∴ u1 =α 1 −α 2 −α 3 +α 4 ---(1)
at node 2, ξ =+1 ;η=−1 and
u=u2
∴ u2 =α 1 +α 2−α 3 −α 4 ---(2)
at node 3, ξ =+1 ;η =+1 and
u=u3
∴ u3 =α 1 +α 2 + α 3 + α 4 ---(3)
at node 4, ξ=−1 ;η =+ 1 and
u=u4
∴ u3 =α 1−α 2 + α 3 −α 4 ---(4)
Add Eqns. (1) and (2)
u1 + u2=2 α 1 −2 α 3 ---(5)
∴ 2 α 3 =2 α 1−u 1−u 2 ---(6)
Add Eqns. (3) and (4)
u3 +u4 =2 α 1 +2 α 3
Substituting for
2 α 3 , we get
u3 +u4 =2 α 1 +2 α 1−u 1−u 2=4 α 1−u 1−u 2
1
∴ α 1= ( u 1 +u2 + u3 +u 4 )
4 ---(i)
82
FEM Dr. T.N

α 1 into Eqn. (6), we get

Substitute
1 2 u + 2u 2 +2 u3 +2 u 4 −4 u1 −4 u2
2 α 3=2∗ ( u1 + u2 +u3 +u4 ) −u1 −u2 = 1
4 4
1
α 3 = (−u1 −u2 +u 3 +u 4 )
4 ---(iii)
Subtract Eqn. (1) from Eqn. (2)
u2 −u1 =2 α 2 −2 α 4
∴ 2 α 4 =2 α 2 + u1−u 2 ---(7)
Subtract Eqn. (3) from Eqn. (4)
u 4 −u3 =−2 α 2 −2 α 4
2 α 4 from Eqn. (7), we get
Substitute for
u 4 −u3 =−2 α 2 −( 2 α 2 +u 1−u 2 )=−4 α 2 −u1 +u 2
1
∴ α 2= ( −u1 +u2 +u3 −u4 )
4 ---(ii)
Substitute into Eqn. (7), we get
1
∴ 2 α 4 =2∗
4
(−u 1 +u2 + u3 −u4 ) +u 1−u 2
1
∴ α 4= ( u −u + u −u )
4 1 2 3 4 ---(iv)
Eqns. (i) to (iv) can be written in matrix form as

{} [ ]{ }
α1 1 1 1 1 u1
α 2 1 −1 1 1 −1 u2
=
α 3 4 −1 −1 1 1 u3
α4 1 −1 1 −1 u
4

Substitute into Eqn. (b), we get

[ ]{ }
1 1 1 1 u1
1 u
u=[ 1 ξ η ξη ] −1 1 1 −1 2
4 −1 −1 1 1 u 3
1 −1 1 −1 u
4

{}
u1
u=
[ 1
4
( 1−ξ−η+ ξη )
1
4
( 1+ ξ−η−ξη )
1
4
( 1+ξ+ η+ξη )
1
4 ]
( 1−ξ+η−ξη )
u2
u3
u4

{}
u1
u2
u=[ N 1 N2 N3 N4]
u3
u4

1 1
N 1 = ( 1−ξ−η+ ξη )= ( 1−ξ ) ( 1−η )
where, 4 4
83
FEM Dr. T.N

1
N 2 = ( 1+ξ )( 1−η )
4
1
N 3 = ( 1+ξ ) (1+ η )
4
1
N 4 = (1−ξ )( 1+ η )
4

Note:
In general, the above shape functions for a 4-noded quadrilateral element can be written as
1
N i=
4
( 1+ξξ i )( 1+ ηηi ) i=1,2,3,4
;
ξ η
where, i and i are the values of natural coordinates corresponding to node (i=1,2,3,4 ).
For example, at node1(i.e.
ξ =−1 ,η 1=−1 then shape function
), 1
1 1
N 1 = ( 1+ξ (−1 ) )( 1+η(−1) )= ( 1−ξ )( 1−η )
4 4
ξ =+ 1, η 2=−1 then shape function
Similarly, at node 2 (i.e. i=2 ), 2
1
N 2 = ( 1+ξ )( 1−η )
4 so on.
For a 2-noded bar element, the general expression for shape functions can also be written as
1
N i=
2
( 1+ξξ i ) i=1 , 2
;
1 1
ξ =−1 , therefore
N 1 = ( 1−ξ ) ξ =+ 1
N 2 = ( 1+ξ )
at node 1, 1 2 and at node 2, 2 , therefore 2 as
we obtained in section (a).

84