Anda di halaman 1dari 30

M.E.

(Product Design and Development)


ED9221- FINITE ELEMENT METHODS IN MECHANICAL DESIGN
University Examination Papers Solved

Part-A

June 2010

1. How will you obtain the total potential energy of a system

The total potential energy of any system is the sum of the strain-energy and the work potential. It
can be written as Total Potential Energy = Strain Energy + Work Potential. The work potential is the
negative of work done by external forces. For example, for a rod subjected to an axial load at one end
and fixed at the other,
2
0 0
1 1
,
2 2
( ) , sec ,
' mod
( )
L L
St xx xx
du
Strain Energy U Adx EA dx where
dx
u x is the axial displacement A the area of cross tion
E the Young s ulus and L the length
The external workdone is Pu L where P the axial force at x L
TotalP
o c
| |
= =
|
\ .


=
} }
2
0
1
( )
2
L
St
du
otentialEnergy U WP EA dx Pu L
dx
t
| |
= + =
|
\ .
}



2. Define discretization error

Discretization error in FEA arises out of meshing and can be reduced by taking more number of
elements. It depends on the mesh size.

3. What is a weighted integral statement? Give examples

When a function to be evaluated by solving a differential equation is to be approximated, it will
not satisfy the governing differential equation. A residue is left. To improve the accuracy of the
approximation, this residue is weighted by another weighting function and the integral of the weighted
residue over the domain is equated to zero, it is called a weighted integral statement.
0
( ) ( ) 0 0
( ) ( ), int
( ) ( ) ( ) 0 ( )
a
L
a
a
d du
For a governing equation given by a x b x u in x L
dx dx
and an approximation u x u x the weighted egral statement is
du d
a x b x u W x dx where W x is the weighting function
dx dx
(
+ = s s
(

~
( (
+ =
( (

}


4. Express the quadratic shape functions of a line element

praveen kumar .A
(Note: If you do not remember, use Lagrangean Interpolation to derive)
( )
1 2 3 2 2 2
,
2 4 2
( ) ; ( ) ( ) ( )
2 2
For a quadratic line element the shape functions are given by
N x x x N x x x and N x x x
| | | |
= = =
| |
\ . \ .



5. What is constant strain triangle? Why is it called so?

A constant strain triangle is a three nodded triangular element with nodes at the vertices. The nodal
approximation will have three shape function which are linear in x & y.(Draw the figure and show)
It is called a CST element because it gives a constant state of strain within the element.

6. How can the bandwidth of a matrix be minimized?

In FEA we come across sparse and banded matrices. For optimal storage space, the bandwidth of
such matrices can be minimized by properly numbering the nodes so that the band becomes narrower.

7. Sketch a typical beam element showing the degrees of freedom

(Sketch a typical beam element with the two nodes and two-degrees of freedom per node, and show
the properties needed like, Youngmodulus and moments of area)

8. Give two examples of time dependent problems in elasticity

1) Longitudinal free vibration of rods
2) Impulsive response of beams

9. State any two nonlinear problems in FEA

1) Contact stresses in contacting solids
2) Large deflection of beams

10. Differentiate between material non-linearity and geometric non-linearity

Material non-linearity comes from plastic behaviour of structures where the stress-strain relation is
non-linear. Geometric nonlinearity comes from large deformations of structures and the stress-
strain relationship is linear.


Nov./Dec. 2010

1. What are non-conforming elements? Name one problem where it is applied.

Elements which have displacement functions which are continuous along the whole interface
between adjacent elements are called non-conforming or incompatible elements.
Some of the commonly used elements in plate bending are non-conforming elements
praveen kumar .A

2. State the importance of patch test

Elements which pass the patch test ensure convergence of solutions to exact solution.
( A patch test is one where a patch of elements are subjected to a constant state of strain and it is
checked whether a constant state of stress is achieved satisfying the constitutive behaviour when these
elements become infinitely small)


3. How is discretization done by quadtree method?

In the quadtree method, the structure to be discretized is enclosed in squares and which ever square is
not fully occupied is further divided into four smaller squares, thus modelling the structure into
elements as closely as possible.

4. Write down the shape function of a one-dimensional line element with two nodes

For a two-noded line element with the local coordinate system of
1 2
0 , ( ) 1 ( )
x x
x the two shape functions are N x and N x s s = =

5. What is mapping of elements?

Mapping of elements is the process of transforming the physical elements into reference or parent
elements using auxiliary or reference coordinates. An example is the iso-parametric element which
transform ( , ) into ( , ) coordinates in 2-D with 1 1 x y and q q s s

6. Compare consistent and lumped mass matrix

Lumped mass matrix is a diagonal matrix where the elements are the total mass of the element
arbitrarily distributed. Consistent mass matrix is one which uses a continuum approach to arrive at a
mass matrix which is fully populated. For example, for a line element with two nodes, these two
matrices are given by:
| |
| |
2 1
,
1 2 6
sec
1 0
0 1 2
Con
Lum
A
Consistent mass matrix m where mass density
A area of cross tion and the length
A
Lumped mass matrix is m

(
=
(


(
=
(



7. What is constant strain element?

Constant strain element ( for a two-dimensional problem) is a triangular element with three nodes
located on the vertices. The approximating functions are bi-linear( liner in x, and linear in y). This
leads to a state of constant strain in the element at any point.
praveen kumar .A

8. Draw an eight nodded shell element


9. Give some practical examples of plasticity computation

Analysis of metal forming processes, Contact problems in elasticity, Crash analysis of vehicles

10. State a few modelling considerations in non-linear FE analysis

Linear analysis should precede non-linear analysis to indicate modelling blunders and locate points of
interest. Avoid very large load steps to ensure convergence. Substructuring could be effectively used.


April/May 2011

1. What is the essence of finite element method?

The essence of FEM is the assumption that a large structure can be considered as an assemblage of
several small structures in which solutions can be approximated by proper interpolation. Discretization
and nodal approximations are the basic concepts of FEM

2. What is the need to perform a patch test? How is it done?

The need for a patch test is to ensure convergence to exact results with the meshes becoming finer.
In this test, a patch of elements is created such that at least one node is internal to the patch. The patch
of elements is subjected to a constant state of strain and as the patch becomes infinitely small it is
checked whether we are able to get a constant state of stress satisfying the constitutive relations.

3. Mention the basic steps in Galerkin method.

In Galerkin method, which is a weighted residual method, the assumed approximations are forced
to satisfy all the boundary conditions ( both essential and natural). The function to be approximated is
written as a series in terms of basis functions which satisfy the boundary conditions and unknown
constants which are to be determined. The basis functions are chosen as weighting functions, and the
weighted residual statement is written for each of the basis functions. This results in an algebraic
equation which is then solved for the unknown constants.

4. Distingush between one dimensional and two dimensional elements

One-dimensional elements have one independent coordinate ( say, x) and is a line element. Two
dimensional elements have two independent spatial coordinates ( say x and y). They are either triangles or
quadrilaterals. (Draw a sketch for each)



praveen kumar .A




5. Specify Jacobian matrix for CST element

| |
3 3
1 1
3 3
1 1
1
;
2
i i i i
i i
i j k i k j
i i i i
i i
x y
x y
J y y and x x
x y A
x y
| |

|

q q
= =
= =
c c ( (
( (
c c
( (
= = = =
c c ( (
( (
c c





6. What is a Mindlin plate element?

Mindlin plate elements takes into account the effect of shear deformation.


7. Specify the mass matrix for a truss element having four degrees of freedom

| |
2 0 1 0
0 2 0 1
; :
1 0 2 0 6
0 1 0 2
A
m Note This is a consistent mass matrix

(
(
(
=
(
(



8. State the required condition for solving dynamic problems as specified by the characteristic
polynomial method

The mode-superposition method holds good only for linear vibrations.

9. Mention two examples for contact non-linearity

(i) Determination of Hertzian contact stresses, (ii) sliding contact with frictional forces

10. State the essential difference between non-linear elastic material and elastic-plastic material

In non-linear elasticity, the material is still in the elastic range, and the non-linearity comes from
large deformations. The strain-displacement relations are non-linear. In the elastic-plastic range, the
material experiences plastic deformation and the stress-strain relations are non-linear.







praveen kumar .A

PART-B

JUNE 2010

11(a) Taking the stepped bar as an example, describe the step-by step procedure of the finite
element solution for obtaining the nodal displacements and stresses. Clearly indicate the
procedure for information and representation of nodes
(b) An axially loaded stepped circular bar is having a diameter of 25 mm for a length of
300mm and a diameter of 20mm for a length of 200mm. The bar is fixed at left end and is
subjected to an axial load of60N at the right end. Determine the element stresses and reaction
force at the left end. Take E = 2 x 10
5
N/mm
2


Since the bar in the problem is fixed at the left end and loaded at the right end, it is taken as a
horizontal rod with the axis in the horizontal direction. Hence, the effect of self-weight is not
considered as contributing to axial stresses

1


Numbering the nodes: The global nodes are numbered from 1 to 3 from left end to the right end.

Numbering the elements: The elements are numbered from 1 to 2 as shown

L
L
1
L
2
Area A
1
Area A
2
Load P
Discretization: Since there is only a concentrated force, for each step of the
rod, let us select one two nodded axial element. There are two elements
for the entire structure
1 2
E,A
2
,L
2

U
1
U
2
U
3
E,A
1
,L
1
Global displacement vector:
{ } { }
| |
1 1
2 2
3 3
1
1
1
;
.
:
1 1
U P
U U The global load vector F P
U P
Next step is to write the element equations
We need to caculate the element stiffness matrices and element load
vectors
The element stiffness matrices are
EA
k
L


= =
` `

) )

| |
2
2
2
1 1
1 1 1 1
, .
EA
and k
L
Since there is no distributed load element distributed load vector is not there
( (
=
( (



Assembly of element equations into global equation

To assemble the global equation, we need to write down the element connectivity matrix.

Element Connectivity

Element No. Global degree of freedom number
Corresponding to
Local degree of freedom number

e 1 2

1 1 2
2 2 3

Using the above, the global stiffness matrix is given by

praveen kumar .A
| |
1 1
1 1
1 1 2 2
1 1 2 2
2 2
2 2
1 1
2 3
0
0
,
0 sin 2,
3.
A A
L L
A A A A
K E
L L L L
A A
L L
For the global concentrated force vector P R reactionat fixed end
P ce no external load is applied at node and P P applied load
at the node
Therefore t
(

(
(
(
= +
(
(
(

(

=
= =
| |{ } { }
1 1
1 1
1 1
1 1 2 2
2
1 1 2 2
3
2 2
2 2
0
0
0
he global equation is K U F which yields
A A
L L
U R
A A A A
E U
L L L L
U P
A A
L L
=
(

(
(

(

+ =
` ` (
(
) )
(

(


Insert the boundary conditions

The essential boundary condition is U
1
=0. On account of this, the first equation which has unknown
reaction at the RHS can not be used at this point. The other two equations will contain only unknowns
U
2
and U
3
. In effect, the first row and first column will not participate in the soluition. We get the
reduced global equation as

1 2 2
1 2 2 2
3 2 2
2 2
1
1 2 2 2 2
1 2 2 2 2 2
1 2
3 2 2 2 1 2
2 2 2
0
,
0
( )
A A A
L L L U
E
U A A P
L L
Solution of reduced global equation
Solving the above equation we get
A A A A A
L L L L L U
L L
P
U A A A A A A
E
L L L

(
+
(

(
=
` `
(
) )

(

(
+

(


(
= =
` `
(
)

)
(

1 2
1 2
0
P
A
E
L L
(
(


(
`
(

)
+
(






Post-Processing

Once the global degrees of freedom have been determined, we evaluate the unknown reactions, and
then go to individual elements and evaluate the strains and stresses.

For the given problem, we have:

2 2
1 1 1 1
2 2
2 2 2 2
1 2
1 2
2
5
3
25 , 490.6 ; 300
4
20 , 314 ; 200
4
490.6 314
1.635 1.57 3.205
300 200
0
1.57 1.57 1.57 1.57
300 200
0.3895
60
1.57 3.205 1.57 490.6 314
2 10
d mm Area A d mm L mm
d mm A d mm L mm
A A
L L
U
U
t
t
= = = =
= = = =
+ = + = + =

(
= =
` `
(

)

)
5
5
5
0
60
3.205
2 10
18.34 10
37.45 10
mm


(
`
(


)

=
`

)

Reaction at fixed end

We make use of the first equation in the global equation now.

1 1
1 2 1
1 1
5 5 1
1 1 2
1
0 2 10 1.635 18.34 10 60
: , .
A A
E U E U R
L L
A
U and R E U N
L
Note This is the same as the applied load but in the opposite direction

| | | |
=
| |
\ . \ .
| |
= = = =
|
\ .


Computation of stresses

The stress at any point in an axially loaded bar element is


praveen kumar .A
praveen kumar .A
{ } | |{ } | || |{ }
( )
1
1 2
2
1 2
1 2
2 1
5
1 1 2 2
5
1 5 2
1 1
( ) 1 , ( )
.
1, 0 18.34 10 ,
18.34 10
2 10 0.122
300
xx ndf
xx
xx
u
dN dN
D D B U E
u dx dx
dN dN x x
Since N x and N x and
dx dx
u u
E for each element
For element u U and u U and we have
N mm
For elem
o o c
o
o


= = = =
`
)
= = = =

=
= = = =

= =
( )
5 5
1 2 2 3
5
2 5
2
2, 18.34 10 , 37.45 10
37.45 18.34 10
2 10 0.0956
200
xx
ent u U u U
N
mm
o

= = = =

= =


Q 14(a) Explain the method for obtaining natural frequencies of longitudinal vibration of a stepped
bar having a cross sectional area of A for a length L, and 2A for a length L. Idealizing the bar with two
elements, take A=100mm
2
and L= 250mm


This is a free vibration problem of a similar stepped bar. For the given set of dimensions we need to
recalculate the global stiffness matrix and also evaluate the global mass matrix. The material of the rod
praveen kumar .A
is not specified. We will take its properties as
| |
| | | |
| |
1 2 1 2
1 2
' mod
, , 2
2 1
,
1 2 6
2 1 2 1
(2 )
1, 2,
1 2 1 2 6 6
e
Young s ulus E and mass density
Here in this problem A A and A A and L L L
AL
The element mass matrices are m
AL A L
For element m and for element m
A
Global mass matrix is M

= = = =
(
=
(

( (
= =
( (

=
| |
2 1 0 2 1 0
1 2 4 2 1 6 2
6 6
0 2 4 0 2 4
( )
1 1 0 1 1 0
1 1 2 2 1 3 2
0 2 2 0 2 2
.
L AL
The global stiffness matrix from the previous problem is
EA EA
K
L L
The boundary conditions are not given Let us assume th

( (
( (
+ =
( (
( (

( (
( (
= + =
( (
( (

1
2 2 2
3 3
0 .
3 2 6 2 0
2 2 2 4 0 6
,
at theleft end is fixed
U The first row and first column do not take part in the solution
The reduced global equation is
U U
EA AL
U U L
EA
Dividing by throughout we get the frequ
L

e
=
( (
=
` ` `
( (

) ) )
2 2
1 2 1
2
3 6 2 2
1
0
2 2 2 4 6
1
, 0.164 1.1087 0.9922 . / sec
1
2.579 . / sec
: mod , .
ency equation as
where L
E
This yields and and radn
L E
and radn
L E
Note Though e shapes are not asked they are given below
Taking the first eq


e

e

= =

= = =
=

( ) ( )
( )
( )
2 3
3 2 2 3
1 1
1 2 3
2 2
2 2 3
, 3 6 2 2 0
3 6
3(1 2 )
. 1, .
2 2 2(1 )
0.164, 1 0.866
1.1087, 1 0.866
ation U U
U U Taking U U
For U andU
For U andU

+ =


= = =
+ +
= = =
= = =



NOV./DEC 2010

11(a). Find the stresses induced in the axially loaded stepped bar shown in Figure.1. The bar has
a cross-sectional area of A
(1)
and A
(2)
over the lengths L
(1)
and L
(2)
respectively. Assume the
following data: A
(1)
=2 cm
2
, A
(2)
=1 cm
2
, L
(1)
=L
(2)
=10 cm E= 2 x 10
7
N/cm
2
, P
3
=1 N


Similar to the previous problem. Please work out. The answers are:

2
7
2 1 2 2
7
3
2.5 10
; : 0.5 / , 1 /
7.5 10
xx
xx
U
cm Element stresses are N cm N cm
U
o o


= = =
` `

) )



April/May 2011

11(b) A horizontal rod ( e=1, A=1, L=1 ) fixed at both ends is subjected to an axially varying
force, f=x
3
N/m. Using a quadratic displacement field, determine the displacement u(x) and
stress ( ) x o in the rod.

For a quadratic displacement field, we can take the approximation as,

praveen kumar .A
1
2 2
1 2 3 2
3
1 2 3
1
( ) , ( ) 1
0 ,
., 0 ,
2
,
a
u x a a x a x or u x x x a
a
For x where is the element length we place the nodes at the two
ends and one node at thecentre x x and x
Substituting the nodal values in the approximations we have
u
u


= + + =
`

)
s s
= = =
1
1 1 1
2 2
2 2 2 2
2 2
3 3 3 3
1 1
2
2 2
3 3
2 2 2
1 0 0 1 0 0
1 . 1
2 4 2 4
1 1
1 0 0 1
3 4 1
( ) 1
2 4 2
a a u
a a u
u a a u
a u
a u and u x x x
a u

( (

( (

= =
( ( ` ` ` `
( (
) ) ) )
( (

(
(

(

(
= =
` `
(

(
) )
(

| |
1
2
3
2 2 2
1
1 2 3 2
3
2 2 2
1 2 3 2 2 2
1 2
0 0
3 4 1
2 4 2
( ) ( ) ( ) ( )
3 2 4 4 2
( ) 1 ; ( ) ; ( )
,
i
u
u
u
u
u x N x N x N x u where the shape functions N are
u
x x x x x x
N x N x N x
N N N
The B matrix is given by B
x x
(
(

(

(

`
(

(
)
(



=
`

)
= + = = +
c c c
=
c c
| | | | | | | | | |
3
2 2 2
0 0
3 4 4 8 1 4
T T
x x x
x
The stiffness matrix is k B E B Adx EA B B dx
| | | | | |
= + +
| | |
c
\ . \ . \ .
= =
} }

praveen kumar .A
| | | | | |
| |
2
2 2 2 2
0 0
2
3 4
4 8 3 4 4 8 1 4
1 4
, int ,
7 8 1
8 16 8
3
1 8 7
T
Stiffness matrix for a quadratic axial element is
x
x x x x
k EA B B dx EA dx
x
This yields on egration
EA
k
The consistent load ve

+



= = + +
`


+

)
(
(
=
(
(

} }
{ }
{ }
1
2
0
3
3
1 2 3
4
( ) ( ) ( )
, , ( 0 (
)
1
8
60
8
i i i
ctor is given by
f
f f where f f x N x dx where N x the shape functions
f
Substituting for N N and N in the above with f x x for the given
loading we get
f


= =
`

)
=


=
`

)
}

For the given problem, E=1., A=1 and L=1 and taking a one-element approximation, we get

| | | | { } { }
{ }
4
1
1 3
3
7 8 1 1
8 16 8 ; 8
3 60
1 8 7 8
0
1 3 .
c
EA L
K k distributed load vector F f
L
R
Concentrated load vector F where R and R are the reaction forces
R
at the nodes and which are fixed No concentrated load is app
(

(
= = = =
`
(

(
)


=
`

)
1 3
2 2
2
0 ( )
16 8
0.025
3 60
lied at node
The boundary conditions are U U fixed ends
Inserting the above in the global equation we get
U or U
= =
= =




1
2 1 3
3
2 2
. sin , 0
4 8
xx
xx xx xx
The strain in the element is given by
U
B U For the gle element approximation U U
U
x
and U and the stress is given by E
c
c o c


= = =
`

)
| |
= =
|
\ .


NOTE

In the case of axially loaded bars, it is possible to have tapered rods with given taper which can
be modelled as equivalent stepped bars

Example: A rod of length 100cm is having a diameter of 2 cm at one end and a diameter of 1 cm at
the other. How would you model this as a two-element approximation?

Taking a two element approximation with equal lengths,

1 2
2 2
2 1
1 3
2
2 3
3
1 3
2
2
2
2
50
2
0,
2
2
1, 3.14 . 3, 1
4 4
0.785
4 4
2, sin var , 1.5
2
1.
4
L
cm
L
The global nodes are located at x x and x L
d
At node area A cm At node d cm
d
area A cm
d d
At the node ce diameter ies uniformly d cm
d
and area A
t t
t t
t
= = =
= = =

= = = =
= = =
+
= =
= =
( ) ( )
2
1 2
1
2 3
2
2 2
1 2
1 2
766
1,
2
2, . mod
2
2 1.766 1.766 0.785
1.883 1.276
2 2
50
av
av
av av
cm
A A
Therefore for element we take an average area of A and
A A
for element A Thus we el the entire tapered bar as
a stepped bar with two areas A cm and A cm
+
=
+
=
+ +
= = = =
= = cm





If the axially loaded bar is positioned vertically, there will be additional load due to self weight.
If this is not negligible compared to the externally applied loads, there will be a distributed load
vector due to this self-weight which has to be assembled into global load vector

1
2
:
1
1 2
dist
Element load vector due to self weight
f
A
where the specific weight of material g
f


= =
` `
) )


Thermal loads due to thermal exapansion being prevented

{ }
0 0
exp ,
( )
( )
xx xx xx xx xx
Th
If is the coefficient of thermal ansion and T the increase in temperature
the thermal strain initial strain in an axial element is given by
T and the corresponding stress is E E E T
This results
o
c c o o c c c o
A
= = A = = A
{ } | | { } { } | |
| |
1
2
1 0
2
1
.
1
;
1 1
Th
xx
in an additional axial loads at the two nodes of a linear element as
f
EA T
f
The element stresses are given by
u
D E B T for theaxial element
u
B for a two noded axial element
o
o c c o o

= A
` `
) )
(
(
= = A
` (

)
=
( )
1 2
, .
xx
and
E
u u E T for eachelement o o = + A


















praveen kumar .A
Example Problem:


Problem



200mm
300mm
P
1
2
3
1
2
Aluminium
Steel
1 2
2 2
1 2
6 0 6 0
1 2
70 200
900 1200
23 10 11.7 10
300 . 60deg
Determine the nodal displacements, stresses and nodal reactions.
E Gpa E GPa
A mm A mm
per C per C
Axial load P kN Temp rise C
o o

= =
= =
= =
= =
praveen kumar .A
{ } { }
{ }
{ }
1 1 1 2 2 2
1 2
9 6 6 3
1
9 6 6 3
2
1 1
1 1
1 86.94
70 10 10 900 23 10 60 10
1 86.94
1 168.48
200 10 10 1200 11.7 10 60 10
1 1
Th Th
Th
Th
Element thermal load vectors are
f E A T and f E A T
That is
f
f
o o



= A = A
` `
) )

= =
` `
) )

= =
`
)
{ }
{ }
3 3
3
1 1
3
1 3
3
3 3
68.48
86.94 86.94
10 86.94 168.48 10 81.54
168.48 168.48
10
10 300
10
Th
C
Global thermal load vector is
F
The global applied load vector is
R R
F P where R and R ar
R R


`
)


= =
` `

) )


= =
` `

)
)
{ } { } { }
3
1
3
3
3
1 3
.
86.94 10
10 81.54 300
168.48 10
0
Th c
e nodal reactions at the fixed nodes
The total global load vector is the sum of the above
R
F F F
R
The boundary conditions are U U Therefore only the middle row
is available

+

= + = +
`

+
)
= =
9 6 9 6
3 3 1 1 2 2
2 2
1 2
2
3 3 1 1
2 1
1
70 10 10 900 200 10 10 1200
218.46 10 ., 218.46 10
200 300
0.196
,
10 86.94 10
for solution and
E A E A
U ie U
U mm
Substituting for all the displacements in the first equation
E A
U R

| | | |
+ = + =
| |
\ . \ .
=
= +

1
3 3 2 2
2 3 3
2
25.2
,
10 168.48 10 325.2
R kN
Substituting in the third equation we get
E A
U R R kN

( =

( = + =



Check: R
1
+P+R
3
=0
Calculation of element stresses

1
1 1 1 1 0 2
2 2 1 1 1
3
1 3 6 2
2
1 1 1
1
3
2 3 6 2 2
2 2
2
0
1 1
:
70 10 0.196
70 10 23 10 60 50.9 /
300
200 10 0.196
200 10 11.7 10 60 2
300
xx xx xx xx
xx
xx
u
U
Element strains are B and E E
u U
U
E E T N mm
E U
E T
c o c c
o o
o o


= = = =
` `
) )

= A = =

= A = =
2
71 / N mm


ONE DIMENSIONAL HEAT TRANSFER PROBLEMS

JUNE 2010
13 (a) (i) Explain the Galerkin approach of deriving finite element equations for the heat transfer
problem.
13 (b) One dimensional element has been used to approximate the convection distribution in a fin. The
solution indicated that the temperature in the node 1 and 2 are 120
0
and 90
0
respectively. Determine
the temperature at a point 4 cm from nodes 1 and 2 which are located at 1.5cm and 6 cm from the
origin.
NOV/DEC 2010
13(b) Find the steady state temperature distribution in the one-dimensional fin shown in Figure 5
considering both convection and radiation losses from its perimeter surface. Assume
8 2 4
2
0.1 5.7 10 /
5 / , 70 /
and W cm K
h W cm K k W cmK
c o

= =
= =







APR/MAY 2011
12(a) Describe Galerkins method of formulation for solving a one-dimensional structural problem
with an illustrative example

13(b) For a one-dimensional composite bar shown in Figure 1, determine the interface temperatures.
For element 1, k=200W/mC, for element 2, k=100W/mC, and for element 3, k=50W/mC. Let a=0.1m
2

The left end has a constant temperature of 100C and the right end has a constant temperature of 300C.
The lengths of the elements 1,2 and 3 are 2m,1m and 0.5m respectively.







GALERKIN METHOD FOR ONE DIMENSIONAL PROBLEMS


0
2
2 2
2
:
( ) ( ) ( ) 0 0
: (0) ( ) ( )
, 5 5 3 5 0 , ( ) 5 , ( ) 3 , ( ) 5
x L
Consider a general equation of the form
d d
x x x in x L
dx dx
d
with boundary conditions of the type and c x d x P
dx
d y dy
For example x xy x Here x x x x x x
dx dx
For
|
o | |
|
| | |
o |
=
(
+ = s s
(

= + =
+ + = = = =
, ,
( ) ( ) 0; ( ) ( ) ( ) ( ) ( ) ,
' mod , ( ) var sec ,
the axial deformation of a rod we have
d du
EA x A x x EA x and x A x where u x axial displacement
dx dx
E Young s ulus A x iable area of cross tion and specific weight
For heat transfer throu
o

(
+ = = =
(


,
( ) ( ) ( ) 0; int, ( ) sec ,
, , , .
For a second order differential equa
gh a fin
d dT
kA x hp x T hp x T T temperature at a po A x tional area
dx dx
k thermal conductivity h convection coefficient p perimeter T ambient temperature

(
+ =
(


tion, specification of the function( )
at a boundary is called the Essential or Geometric Boundary Condition.
Specification of the first derivative of the function at a boundary is called
the Natural or
|
Force Boundary Condition. Thus, the first condition above is
an Essential B.C., and the second is aNatural B.C. If the boundary values are
zero, it is called a homogeneos B.C., and if not, a non-homogeneous b.c.















0
0
1
Assuming a trial solution ( ) , the weighted residual statement is
( ) ( ) ( ) ( ) 0 ( )
, ( ) ( ) ( )
ap
L
ap
ap
n
ap j j
j
x
d
d
x x x x dx where x is the weighting function
dx dx
In Galerkinmethod wetake x x a x w
| |
|
o | |
|
=
~
( (
+ =
( (

= u + u
}

0
( ) hom ( ) hom
( ) ( ) 1, 2..,
In finite element formulations, the approximations are made over indivi
j
j j
here
x satisfies the non ogeneous part of BC and x satisfy the ogeneous part
The weighting function is taken as x x for j n
u u
= u =
1
dual elements and for each element
( ) ( ) . . ( ) int
, deg
n
ap j j j
j
j
x N x This is called nodal approximation N x are the erpolation functions or
shape functions and are the nodal values of the function called the element rees of free
| |
|
=
=

.
. ( ) ( ) , 1, 2,...,
.
:
Taking the first term( which con
j j
dom
In this case the weighting functions are the shape functions That is x N x j n
To facilitate lower order approximation we go in for a weak form of the equation
Weak Formulation
= =
0 0 0
tains the 2nd order derivative) and integrating by parts
( ) ( ) ( ) ( ) ( )
L
L L
d d d d d
x x dx x x x dx
dx dx dx dx dx
| | |
o o o
( ( (
=
( ( (

} }

0 0 0
With this, the weak form of the equation becomes,
( ) ( ) ( ) ( ) ( ) ( )
This contains only the first derivative of the functions, and also a boundary term
L
L L
ap
ap
d
d d
x x dx x x dx x x
dx dx dx
|
|
o | | o

(
+ = +
`
(

)
} }
.
The LHS is found to be symmetric with repect to & and is called the
Bi-linear Term ( , ). The RHS is called the Linear Term, ( )
Thus the weak form is ( , ) ( ).
,
ap
ap
ap
B
B
If B is symmetric we can write
|
|
| =

( )
2
2
0 0 0
1
, , , ( , ) ( )
2
1
, ( ) ( ) ( ) ( ) ( ) ( )
2
ap ap ap
L
L L
ap
ap ap ap
a QuadraticFunctional defined by B
d
d
Or x x dx x x dx x x
dx dx
and from this quadratic functional we can arrive at the weak f
t t | | |
|
|
t o | | | | o
=
(
| |
(
( = +
` |
(
(
\ .

)
} }
min
. 0
orm by imizing
ie ot =

0
1 1 1
, :
( ){ ( ) } ( ){ ( ) } ( ) ( ) ( ) ( ) ( )
n n n
L
j
i
j j j j i i i j
j j j
Substituting the nodal approximation and the shape function as weighting function
in the weak form of theequation we have
dN
dN d
x N x x N x N x dx x N x dx N x x
dx dx dx
o | | | o |
= = =
(
+ =
(


}
0
0
1, 2,..., .
, deg
.
int lg ,
L
L
j
for i n
In the equation above the unknowns are s which are nodal rees of freedom
There are thus n equations in n unknowns Thus the differential equation has been
transformed o an a ebraic equation which is ea
|
=


}
0 0
1 1
.
int ,
( ) ( ) ( ) ( ) 0
e
e N n
j e e e e e e e i
j i j i i
e j
sy to solve
If we take the summation sign out of the egrals we have
dN
dN
x x N N dx x N x dx P
dx dx
| o |
= =
( (

+ ( = (
`
( (
)

} }

The first summation is over all the elements, numbering N, and the term within the first bracket has
to be zero, since equilibrium is attained in each of the element. In this term, for every I ( i=1,2,..n),
there is one equation, which for all non-zero degree of freedom, has to vanish. This leads us to the
algebraic equation,

1
0 0
1, 2, 3,.., 1, 2,..,
, ( sup ),
, ( ) ( )
n
e e e e
ij j i i
j
j e i
ij i j i i
k f P for every i n for each element e N
Thus for each element dropping the erfix e we get
dN
dN
k N N dx f x N x dx represents the effect of distributed force
dx dx
a
|
o |
=
= + = =
(
= + =
(

} }
| |{ } { }
.
, ,
.
i
nd P represents the boundary force applied at the nodes
Finally we get the element equation
k U F for each element

=


For a structural problem,

sec , ,
,
0 0
Neglecting the effect of the self weight and taking the example of a
uniform rod of tional area A and length L with the left end fixed
and a concentrated load P applied at the other end
d du
The governing equation is EA in x
dx dx

(
= s
(

2
1 1 2 2
1
, ( ) tan . ( ) 0 ( ) 0
, (0) 0
, 0
( ) ( ) ( ) ( ) (
x L
loc
loc j loc j
j
L
Here x EA cons t x and x
du
The boundary conditions are u and EA P
dx
Taking a two noded linear axial element with local coordinate x
u x N x u N x u N x u we drop the suffi
o |
=
=
s
= = = =
= =
s s
= = +

1 2
1 2
0
11
0
)
1 1
( ) 1 ( )
, tan
loc
j
i
ij
j
i av
ij av
x loc in x
dN dN x x
N x and N x and
dx dx
Element stiffness matrix is populated by
dN
dN
k EA dx For a small element we can take an average area as cons t
dx dx
dN
dN EA
k EA dx leading to k k
dx dx
= = = =
=
= = =
}
}
| |
22 12 21
1 1
0 ( ).
1 1
av
av
i
EA
and k k
EA
f no distributed load k
= =
(
= =
(



For one-dimensional heat-transfer through fin problem, the equation is

( ) 0 min int
, .
, ( ) tan ( )
d dT
kA hp T T assu g no ernal heat generation and
dx dx
heat transfer withinthe finis by conduction and from the surface by convection
Here the radiation effect is neglected initially
Here x kA cons t and x o |

(
+ =
(

= =
| | { }
0 0 0
1 1 2 2 1 2
tan , ( )
k
, ( ) ; ( ) 1 ( )
1 1 2 1 1
, k
1 1 1 2 1 6 2
j
i
ij i j i i
hp cons t and x hpT
dN
dN
kA dx hp N N dx and f hpT N dx
dx dx
x x
For a two noded linear element T x N T N T N x and N x
hpT kA hp
This gives and f
R

= = =
= + =
= + = =
( (
= + =
`
( (

)
} } }
( )
( )( )
| |
4 4
2 2
.
xx
r
r
r
adiation brings in non linearity as anadditional term appears in the
governing equation as
q T T Such problems are solved iteratively by taking
aradiation heat transfer coefficient h T T T T
h
and adding a matrix k
oc
oc

=
= + +
=
{ }
2 1
1 2 6
1
.
1 2
0, , .
r
r
r
p
to stiffness matrix and
h pT
a load vector f to the other load vector Initially start
with h solve for T then iterate

(
(

(
=
(

=



Weak formulation has already been included.

Numerical Problems:

June 2010

13(b) Given: Temperatures at nodes, T
1
=120
o
C and T
2
=90
o
C
Nodal coordinates( in global coordinate system) x
1
=1.5 cm and x
2
=6 cm
Taking a local coordinate system x
*
=x-x
1
, new local coordinates of the nodes are,
x
*
1
=0 and x
*
2
=6-1.5 = 4.5cm . The length of the element
* *
* *
1 2
* *
*
1 1 2 2 1 2
*
*
4.5
, ( ) 1 ( )
int
( ) 1
4 4
int 4 1, 4 (1 ) 120 90 100
6 6
4 2, 4
o
cm
x x
Shape functions are N x and N x
Temperature at any po within the element is
x x
T x N T N T T T
At a po cm from node x cm and T C
At cm from node x
=
= =
| |
= + = +
|
\ .
= = + =
= =
2 2
6 4 2 (1 ) 120 90 110
6 6
o
cm T C = = + =


Nov/Dec.2010
13(b)

2
2
1
2
tan 2 . 3.14 , 6.28
4
5
. 1, 140 , . 40
70 / , 5 /
o o
o o
d
Fin has a cons t diameter d cm Area A cm Perimeter p d cm
Length of the fin L cm
Temp at global node T C Ambient Temp T C
Thermal conductivity k W cm K Film coefficient h W cm K
Emissivity
t
t
c

= = = = =
=
= =
= =
| |
8 2 4
0.1 tan 5.7 10 /
1 1 2 6
,
1 1 6 2 6
1 1 2 1 43.96 43.96 52.34 26.16
70 3.14 5 6.28 5
1 1 1 2 43.96 43.96 26.16 52 5 6
o
and Stefan Boltzmann cons t W cm K
kA hpL
Making a one element approximation K
L
o

= =
( (
= +
( (


( ( (
= + = +
( ( (


96.3 17.8
.34 17.8 96.3
( (
=
( (




{ }
| | { }
1 1 3140
5 6.28 40 5
1 1 3140 2 2
.
0 0 0
int
0
Th
Conv Conv
hpT L
The thermal load vector is F
At the right end there is convective transfer to ambient air This would
roduce additional matrix K and load vector F
hA hA


= = =
` ` `
) ) )
(
= =
(

| | { }
0 0 0 0 0
,
0 0 15.7 628
Conv Conv
T
These are K and F
hA
When these are added to existing matrix and vector we have the equation


`
)
( (
= = =
`
( (
)



( )( )
( )
1 1
1
2
2
1 2 1 2 2
96.3 17.8 3140
140
17.8 112 3768
sec , 55.9
,
195.9
97.9
2 2
o
r
T q
with T
T
From the ond of the equation T C
To account for radiation we now add a radiation heat transfer coefficient
T T
h T T T T with T co

+ (
= =
` `
(

) )
=
+
= + + = = =
1 8 2 2
2
2
5
0.1 5.7 10 (97.95 40 )(97.95 40) .0088
5.0088 .
. .
r
r r
h
With this the new h and we get a new temperature T From that
we get h This process is continued till convergence for h is reached

= + + =
=


NOTE: Exercise problem: Rework the above problem neglecting the effect of radiation heat loss
and taking two equal length elements. Compare the result with the previous one.


APR/MAY 2011
13(b)
2
1 1 1 1
2
2 2 2 1
2
3 3 3
,
2 , 0.1 , 200 / , 0 ( )
1 , 0.1 , 100 / , 0 ( )
0.5 , 0.1 , 50 /
o
o
o
We make a three element approximation
The properties of the elements are
m A A m k W m C h No convective transfer
m A A m k W m C h No convective transfer
m A A m k W m
= = = = =
= = = = =
= = = =
| |
| |
1
1 4
1
1 1
1
1
2 2
2
, 0 ( )
, 100 300
,
1 1 1 1 10 10 1
200 0.1
1 1 1 1 10 10 2 2
1 1 1 1 10 10
100 0.1
1 1 1 1 10 1
o o
C h No convective transfer
Boundary conditions T C and T C
The element matrices are
k A
k
k A
k
=
= =
( ( (
= = =
( ( (


( (
= = =
( (


| |
| |
1
3 3
3
2
10 3
1 1 1 1 10 10 3
50 0.1
1 1 1 1 10 10 4 0.5
. . . .
. .
1 2
1 1 2
2 2 3
3 3 4
1 1 0 0
1 2 1 0
10
0 1 2 1
0 0 1 1
k A
k
The element connectivity is
El No Global d o f corresponding to
local d o f
e
The global matrix is
K
(
(

( ( (
= = =
( ( (


(

1 1
2
1
3
4 4
.
1 1 0 0
0 1 2 1 0
10
0 0 1 2 1
0 0 1 1
There is no convection matrix
There is no distributed load vectors The global equation is
T q
T
where q is the heat flux entering
T
T q
the bar at its l
(
(
(
(

(
(


(
=
` `
(

(

) )
3
1 4
1 4
2 3
: 100 300
167 233
o o
o o
eft end and q is the heat flux leaving the bar at its right end
Given T C and T C
Solutionof the reduced equation after substituting for T and T yields
T C and T C
= =
= =

Two-Dimensional Problems

Elements and Shape Functions

1. Bi-linear Triangular Element( 3-noded Triangle)



X
O
Y
1
2
3


P(x,y)

2 2
( , ) x y
3 3
( , ) x y
1 1
( , ) x y
e
A
For a 3-noded triangular element, if we assume the approximation as


Substituting the values at the nodes,
We have,
1 2 3
( , ) x y a a x a y | = + + , | |
1
2 1 2 3
3
( , ) 1
a
x y x y a a a x a y
a
|


= = + +
`

)

Substituting the values at the nodes,
We have

1 1 1 1
2 2 2 2
3 3 3 3
1
1
1
x y a
x y a
x y a
|
|
|
(

(
=
` `
(

(
) )


Rearranging this equation, we get,
| |
1
1 1 1
2 2 2
3 3 3
1
1 2 3 2
3
1
( , ) 1 1
1
( , ) ( , ) ( , )
x y
x y x y x y
x y
N x y N x y N x y
|
| |
|
|
|
|

(

(
=
`
(

(
)


( =
`


)














Example:

For a triangular element with the nodes at (0,0),(2,0),(1,2) we have,

( )
1 2 3 1 2 3
1 2 3 3 2 2 3 1 1 3 3 1 2 2 1
1 2 3
1 2 3 2 3 1 3 1 2
1 3 2 2 1 3 3 2 1
1
0 2 1 ; 0 0 2
1 0 0
2 1 2 0 1 2 2 0 1 4 2
1 1 2
4; 0 0
: 2
2 , 2 0
1 , 1 2
(
e e
e
x x and x y y and y
A or A
x y x y x y x y and x y x y
Check A
y y y y and y y
x x x x and x x
N
o o o
o o o
| | |

= = = = = =
= = = =
= = = = = =
+ + =
= = = = = =
= = = = = =
( )
( )
( )
1 1 1
2 2 2 2
3 3 3 3
1 2 3
1 1
, ) (4 2 )
2 4
1 1 1
( , ) (0 2 ) (2 )
2 4 4
1 1
( , ) (0 0 2 )
2 4 2
: 1
e
e
e
x y x y x y
A
N x y x y x y x y
A
y
N x y x y x y
A
Check N N N
o |
o |
o |
= + + =
= + + = + =
= + + = + =
+ + =










( , ) .
( )
( , )
2
; ;
i
i i i
i e
i j k k j i j k i k j
The functions N x y are called the shape functions
x y
N x y
A
x y x y y y x y
o |
o |
+ +
=
= = =
1 1
2 2
3 3
1
2 1
1
e
x y
A x y
x y
=
Exercise Problem: Find the shape functions for a triangular element with three nodes, one at
each vertex, for an element having the nodes at (1,1),(3,2) and(2,3).


June 2010
12(b). For the triangular element having coordinates at (0,0),(2,5), and (4,0.5), obtain the shape
functions and the intensity of pressure at a point (2,1.5). The nodal values of pressures are
60N/m
2
,51N/m
2
, and69N/m
2
at nodes 1,2 and 3, respectively

(Note:Pressure at a node is not an appropriate example. However we will do the problem)

1 2 3 1 2 3
1 2 3
1
1 2 3
1
: 0, 4 2; 0, 0.5 5(
, )
4 5 2 0.5 19; 2 0 0 5 0; 0 0.5 4 0 0
2 19
0.5 5 4.5, 5 0 5, 0 0.5 0.5
2 4
Given x x and x y y and y numbering them
in theclockwise direction after plotting the triangle
A
and
o o o
| | |

= = = = = =
= = = = = =
=
= = = = = =
= =
( ) ( )
( ) ( )
( ) ( )
2 3
1 1 1 1
2 2 2 2
1 3 3 3
1 2 3
1
2, 0 2 2 4 0 4
:
1 1
( , ) 19 4.5 2
2 19
1 1
( , ) 5 2
2 19
1 1 1
( , ) 0 0.5 4 ( 0.5 4 )
2 19 19
: 1
: 60
and
Shape functions are
N x y x y x y
A
N x y x y x y
A
N x y x y x y x y
A
Check N N N
Nodal values of pressure are p N

o |
o |
o |
= = = =
= + + =
= + + =
= + + = + = +
+ + =
=
( ) ( )
2 2 2
2 3
1 1 2 2 3 3
1 2 3
1 1 2
/ , 69 / 51 /
: 2(4, 0.5) 3(2, 5) int
int int ( , ), ( , )
1
( , ) 19 4.5 2 5 2 ( 0.5 4 )
19
1
19 4.5 5 0.5
19
m p N m and p N m
Note The pressures at node and are taken o account
At any erior po x y p x y N p N p N p
p x y x y p x y p x y p
p x p p
= =
= + +
= + + + (

= + + ( ) ( )
( ) ( ) | |
| |
3 1 2 3
2
2 2 4
1 1
1140 49.5 54 1140 49.5 54
19 19
1
2, 1.5 1140 49.5 2 54 1.5 60.95 /
19
p y p p p
x y x y
At x y p N m
+ + (

= + + = + (

= = = + =

Anda mungkin juga menyukai