FINITE ELEMENT METHOD

ME751
Mechanical IV/II
Chapter 6
APLICATIONS: ONE
DIMENSIONAL PROBLEMS
6.1 INTRODUCTION
To make the ideas so far expressed more concrete, we shall consider
in this section some one-dimensional examples in which piecewise
defined shape functions are used to solve problems governed by
ordinary second order differential equations.

Consider the problem of finding the function ϕ(x) that satisfies the
differential equation

..... (1)

..... (2a)

..... (2b)
..... (2c)

2
FINITE ELEMENT METHOD
Equation [1] arises in connection with the analytical description of
many physical processes. For example, conduction and convection
heat transfer in plane wall or fin, flow through channels and pipes,
transverse deflection of cables, axial deformation of bars, and many
other physical processes.
We solve this problem by the finite element method for k = 1 and L =
1. A set of 4 nodes is chosen in the region 0 < x < 1, as shown in
Figure, and with each node we associate a piecewise linear global
shape function Nm.

3
FINITE ELEMENT METHOD
An approximation
..... (3)

can be constructed, where ϕm is the value of the approximation at

node m. The boundary conditions at x = 0 and x = 1 can therefore be
satisfied immediately by specification of the appropriate nodal
values.
However, it is more convenient to formulate the weighted residual
equations without initially prescribing the boundary values of ϕ and
then to insert these known conditions when solving the final equation
set. Thus with ϕ1, ϕ2, ϕ3 and ϕ4 all regarded as being unkown at this
stage, the weighted residual form of [1] is then

..... (4)

4
FINITE ELEMENT METHOD
Equation [4] requires continuity of first derivates of the trial
functions if infinite values are to be avoided. Integration by parts
relaxes this requirement on the shape function and leads to a weak
form of the weighted residual statement

..... (5)

Now it is apparent that only C0 continuity of ϕ^ (and hence of Nm)

and wl is demanded. The piecewise linear shape functions satisfy this
requirement, and C0 continuity of the weighting functions is ensured
if the Galerkin method is used. The Galerkin form of this equation set
can be expressed immediately as
..... (6)

where ..... (6a) ..... (6b)

5
FINITE ELEMENT METHOD
We now observe that the contributions to these terms from a typical
element e, associated with nodes numbered i and j , as in Figure, can
be calculated in a general form. On such a typical element e, we can
write

where he = xj – xi. The only nonzero global trial functions on element e

are Ni and Nj, and so Nl = 0 on element e if l does not equal to i or j,
that is, if node l does not belong to element e.
6
FINITE ELEMENT METHOD
Then approximate function given by Equation [3] can also be
expressed as

..... (7)

which can be combined as

..... (8)

7
FINITE ELEMENT METHOD
Solution for Case 1
We solve Equation [1] subjected to the boundary conditions

Then approximate solution given by Equation [8] can be expressed

as
..... (9)
which is equivalent to choosing

for Galerkin formulation.

Since two nodal values and are given, we need two linear
equations to determine the unknown nodal values ϕ2 and ϕ3.

8
FINITE ELEMENT METHOD
Substituting wl = ψ1 = N2 into Equation [5], we get

..... (10)

Since N2 = 0 at x = 0 and x = 1, Equation [10] can be expressed as

..... (11)

Defining

Equation [11] can be expressed as

..... (12)
9
FINITE ELEMENT METHOD
Similarly, substituting wl = ψ2 = N3 into Equation [5], and following
the similar procedure we get
..... (13)

Solution for Case 2

We solve Equation [1] subjected to the boundary conditions

Then approximate solution given by Equation [8] can be expressed

as
..... (14)
which is equivalent to choosing

for Galerkin formulation.

10
FINITE ELEMENT METHOD
Since only on nodal values is given, we need three linear
equations to determine the unknown nodal values ϕ2, ϕ3 and ϕ4.
Weak formulation of the given problem to include the derivative
boundary conditions can be expressed as

..... (15)
Substituting

..... (16)

11
FINITE ELEMENT METHOD
Substituting wl = ψ1 = N2 into Equation [16], we get

..... (17)

Since N2 = 0 at x = 0 and x = 1, Equation [17] can be expressed as

..... (18)

Equation [18] can be expressed as

..... (19)

Similarly substituting wl = ψ2 = N3 into Equation [16], and

following the similar procedure we get
..... (20)
12
FINITE ELEMENT METHOD
Substituting wl = ψ3 = N4 into Equation [16], we get

..... (21)

Since N4 = 0 at x = 0 and N4 = 1 at x = 1, Equation [21] can be

expressed as

..... (22)

Equation [22] can be expressed as

..... (23)

13
FINITE ELEMENT METHOD
Solution for Case 3
We solve Equation [1] subjected to the boundary conditions

Then approximate solution given by Equation [8] can be expressed

as
..... (24)
which is equivalent to choosing

for Galerkin formulation.

Since none of the nodal values are given we need four linear
equations to determine the unknown nodal values ϕ1, ϕ2, ϕ3 and ϕ4.

14
FINITE ELEMENT METHOD
Weak formulation of the given problem to include the derivative
boundary conditions can be expressed as

..... (25)

Substituting

..... (26)
15
FINITE ELEMENT METHOD
Substituting wl = ψ1 = N1 into Equation [26], we get

..... (27)

Since N1 = 1 at x = 0 and N1 = 0 at x = 1, Equation [27] can be

expressed as

..... (28)

Equation [28] can be expressed as

..... (29)

16
FINITE ELEMENT METHOD
Similarly, substituting wl = ψ2 = N2 and wl = ψ3 = N3 into Equation
[26], and following the similar procedure we get
..... (30)
..... (31)
Similarly, substituting wl = ψ3 = N3 into Equation [26], and
following the similar procedure we get

..... (32)

Since N4 = 0 at x = 0 and N4 = 1 at x = 1, Equation [27] can be

expressed as

..... (33)
17
FINITE ELEMENT METHOD
Equation [33] can be expressed as
..... (34)

Evaluation of Elements of Stiffness Matrix

Using

elements of stiffness matrix are determined as

18
FINITE ELEMENT METHOD
 19
FINITE ELEMENT METHOD
Similarly,

20
FINITE ELEMENT METHOD
 21
FINITE ELEMENT METHOD
Similarly,

22
FINITE ELEMENT METHOD
Similarly,

Similarly,

23
FINITE ELEMENT METHOD
Similarly,

24
FINITE ELEMENT METHOD
 25
FINITE ELEMENT METHOD
 26
FINITE ELEMENT METHOD
Similarly,

27
FINITE ELEMENT METHOD
 28
FINITE ELEMENT METHOD
Following similar procedure, other remaining coefficients can be
found to be

29
FINITE ELEMENT METHOD
Then finite element equation for Case 1 is given by

Then finite element equation for Case 2 is given by

30
FINITE ELEMENT METHOD
Then finite element equation for Case 3 is given by

Element-wise Calculation of Stiffness Coefficients

Global stiffness matrix for the problem can also be obtained by the
assembly of stiffness matrix for a single element obtained from the
relation

31
FINITE ELEMENT METHOD
Then finite element equation for Case 3 is given by

Element-wise Calculation of Stiffness Coefficients

Global stiffness matrix for the problem can also be obtained by the
assembly of stiffness matrix for a single element obtained from the
relation

32
FINITE ELEMENT METHOD
For Element 1, i = 1 and j = 2

For Element 2, i = 2 and j = 3

For Element 3, i = 3 and j = 4

By assembly of stiffness matrix of each element, we get the global

stiffness matrix as above.

33
FINITE ELEMENT METHOD
EXAMPLE 6.1
Solve the differential equation using the Galerkin finite element
method. Assume that the domain 0 < x < 1 is divided into equal
length elements.

a) Use two linear elements.

b) Use three linear elements.
c) Use four linear elements.
d) Compare the obtained value ϕ(0.5) of in each case with exact
value.

34
FINITE ELEMENT METHOD
 35
FINITE ELEMENT METHOD
 36
FINITE ELEMENT METHOD