SME 3033 FINITE ELEMENT METHOD

Steady-State Heat Transfer

(Initial notes are designed by Dr. Nazri Kamsah)
SME 3033 FINITE ELEMENT METHOD

One-Dimensional Steady-State Conduction

We will focus on the one-dimensional steady-state conduction problems
only. It is the easiest heat conduction problem.
In one-dimensional problems, temperature gradient exists along one
coordinate axis only.
Objective
The objective of our analysis is to determine; a) the temperature distribution
within the body and, b) the amount of heat transferred (heat flux).

T1 T2 T3
qx
x
SME 3033 FINITE ELEMENT METHOD

The Governing Equation

Consider heat conduction q (W/m2) through a plane wall, in which there is a
uniform internal heat generation, Q (W/m3).

An energy balance across a control volume (shaded area) yields,

 dq 
qA  QAdx   q  dx  A
 dx 
SME 3033 FINITE ELEMENT METHOD
where q = heat flux per unit area (W/m2)
A = area normal to the direction of heat flow (m2)
Q = internal heat generated per unit volume (W/m3)
Cancelling term qA and rearranging, we obtain,

Q
dq
… (i)
dx
For one-dimensional heat conduction, the heat flux q is governed by the
Fourier’s law, which states that,
 dT 
q  k   
 dx 
… (ii)

where k = thermal conductivity of the material (W/m.K)

(dT/dx) = temperature gradient in x-direction (K/m)

Note: The –ve sign is due to the fact that heat flows from a high-temperature to
low- temperature region.
SME 3033 FINITE ELEMENT METHOD

Substituting eq.(ii) into eq.(i) yields,

d  dT 
k Q  0
dx  dx 
The governing equation has to be solved with appropriate boundary conditions
to get the desired temperature distribution, T.

Note:
Q is called a source when it is +ve (heat is generated), and is called a sink when
it is -ve (heat is consumed).
SME 3033 FINITE ELEMENT METHOD
Boundary Conditions
There are three types of thermal boundary conditions:
a) Specified temperature, Ti = To;
b) Specified heat flux, e.g., qi = 0 (insulated edge or surface);
c) Convection at the edge or surface, (h & T∞ are specified).

These are illustrated below.

Note: h is the convective heat transfer coefficient (W/m2K).

SME 3033 FINITE ELEMENT METHOD
Finite Element Modeling
The uniform wall can be modeled using
one-dimensional element.
To obtain reasonably good temperature
distribution, we will discretize the wall into
several 1-D heat transfer elements, as
shown.
Note:
X represents the global coordinate
system.
Can you identify the kind of boundary
conditions present?
There is only one unknown quantity at
any given node, i.e. the nodal
temperature, Ti.
SME 3033 FINITE ELEMENT METHOD

Temperature Function
For a given element in local coordinate (), temperature T varies along the
length of the element.
We need to establish a temperature function so that we can obtain the
temperature T, at any location along the element, by interpolation.

For a one-dimensional steady-state conduction, temperature varies linearly along

the element.
Therefore we choose a linear temperature function given by,

T    N1T1  N2T2 or T    N T

 1    and N2  1 1   
SME 3033 FINITE ELEMENT METHOD

where N1 
1
2 2
We wish to express the (dT/dx) term in the governing equation in terms of
element length, le, and the nodal temperature vector, {T}. Using the chain
rule of differentiation
dT dT d d dT
   
dx d dx dx d
…(i)

d
  x  x1   1  
 x2  x1  dx  x2  x1 
2 2 …(ii)
Recall,

T    1   T1  1 1   T2    T1  T2 …(iii)

d
1 dT 1 1
2 2 2 2
Substitute eq.(ii) and eq.(iii) into eq.(i) we get,
dT

2  1
  
1 
2 
1
 T1  T2 
dx x2  x1  2 2  x2  x1
T1 T
SME 3033 FINITE ELEMENT METHOD

  1 1T
dT 1
dx x2  x1
or, e

  BT T 
dT e

dx

 BT    1 1   1 1
 x2  x1 
1 1
where
le

is called the temperature-gradient matrix. The heat flux, q (W/m2) can then
be expressed as

T1 
q  k   1 1  
1
le T2 
SME 3033 FINITE ELEMENT METHOD

Element Conductivity Matrix

The element conductivity matrix [kT] for the 1-D heat transfer element
can be derived using the method of weighted residual approach.
Recall, the conduction governing equation with internal heat generation,

d  dT 
k Q  0
dx  dx 

and q x L  hTL  T 
Imposing the following two boundary conditions,

Tx0  To

and solving the equation yields the functional, pT given by

pT   0  
1  dT 
k    
2
1
2  dx 
L L 2
dx QTdx h TL T
0 2
Substitute for dx and (dT/dx) in terms of  and {T}e,
SME 3033 FINITE ELEMENT METHOD

x x
d  dx  dx  2 1 d  e d
2 l
x2  x1 2 2

T   N T   BT T
(e) dT (e)
and
dx

within the element, the functional pT becomes

Assuming that heat source Q = Qe and thermal conductivity k = ke are constant

pT      1 T T  T
   
( e )  ke le 
 
1 1

 2 
T (e)
T B B d
2

e  2 1    T  h TL  T 

 Qele 
e


1 1

(e) 2
N d
2
Note: The first term of the above equation is equivalent to the internal strain
energy for structural problem. We identify the element conductivity matrix,

 kT     B   B   d
kele 1 T

2 1 T T
SME 3033 FINITE ELEMENT METHOD

Solving the integral and simplifying yields the element conductivity

matrix, given by

 1 1
 kT    1 1 
ke
 
(W/m2K)
le

Note: If the finite element model comprises of more than one element, then the
global conductivity matrix can be assembled in usual manner to give

 K11 K1L 
K K2 L 
K12

 KT    21
K22

(W/m2K)
 
 KL1 KL 2 ... KLL 
SME 3033 FINITE ELEMENT METHOD
Exercise1
A composite wall is made of material A and B as shown. Inner surface of
the wall is insulated while its outer surface is cooled by water stream with
T∞ = 30C and heat transfer coefficient, h = 1000 W/m2K. A uniform heat
generation, Q = 1.5 x 106 W/m3 occurs in material A. Model the wall using
two 1-D heat transfer elements.
Question: Assemble the global conductivity matrix, [KT].
SME 3033 FINITE ELEMENT METHOD
Element Heat Rate Vector
If there is an internal heat generation, Qe (W/m3) within the element,
then it can be shown that the element heat rate vector due to the
internal heat generation is given by

rQ   Qe  le 1 W

e


2
2 1 m
Note:
1. If there is no internal heat generation in the element, then the heat rate vector

 0   le 1 0
rQ 
for that element will be,

 
  
e W
2 1 0 m2
2. If there are more than one element in the finite element model, the global heat
rate vector, {RQ} is assembled in the usual manner.
SME 3033 FINITE ELEMENT METHOD
Global System of Linear Equations
The generic global system of linear equation for a one-dimensional
steady-state heat conduction can be written in a matrix form as

 K11 K1L   T1   RQ1 

K  
K2 L  T2   RQ 2 
K12
 21 K22
  
    
   
 KL1 KL 2 ... KLL  TL   RQL 

Note:
1. At this point, the global system of linear equations have no solution.
2. Certain thermal boundary condition need to be imposed to solve the equations
for the unknown nodal temperatures.
SME 3033 FINITE ELEMENT METHOD
Exercise 2
Reconsider the composite wall in Exercise 6-1. a) Assemble the global
heat rate vector, {RQ}; b) Write the global system of linear equations for
the problem.
SME 3033 FINITE ELEMENT METHOD
Temperature Boundary Condition
Suppose uniform temperature T =  C is specified
at the left side of a plane wall.

T1   oC
To impose this boundary condition, modify the
global SLEs as follows:
1. Delete the 1st row and 1st column of [KT] matrix;
2. Modify the {RQ} vector as illustrated.
x

 K11 K1L      RQ1   K11 

L
K  
K2 L  T2   RQ 2   K21 
K12
 21 K22
   
      
   
 KL1 KL 2 ... KLL  TL   RQL   KL1 

Note: Make sure that you use a consistent unit.

SME 3033 FINITE ELEMENT METHOD
Convection Boundary Condition
Suppose that convection occurs on the right side of a
plane wall, i.e. at x = L.
The effect of convection can be incorporated by
modifying the global SLEs as follows: T ; h

1. Add h to the last element of the [KT] matrix;

2. Add (hT∞) to the last element of {RQ} vector.
We get,
 K11   T1   
 
x
K K2 L  T2  
K12 K1L RQ1

 21 

    
K22 RQ 2 L
 
...  KLL  h   TL   RQL  hT 
   
 KL1 KL 2
 

Note: Make sure that you use a consistent unit.

SME 3033 FINITE ELEMENT METHOD

The Heat Flux

Once the temperature distribution within the wall is known, the heat flux
through the wall can easily be determined using the Fourier’s law.

T1 
q  k   1 1  
We have,
1
T2 
W/m2
le
Note:
1. At steady-state condition, the heat flux through all elements has the same
magnitude.
2. T1 and T2 are the nodal temperatures for an element.
3. le is the element length.
SME 3033 FINITE ELEMENT METHOD
Exercise 3
Reconsider the composite wall problem in Exercise 6-2. a) Impose the
convection boundary conditions; b) Solve the reduced SLEs, determine
the nodal temperatures; c) Estimate the heat flux, q through the
composite wall.
SME 3033 FINITE ELEMENT METHOD
Exercise 3: Nastran Solution

413 K

407 K

388 K

378 K
SME 3033 FINITE ELEMENT METHOD
Heat Flux Boundary Condition
Suppose heat flux q = qo W/m2 is specified at the left
side of a plane wall, i.e. at x = 0.
The effect of specified heat flux is incorporated into the

q  q0
analysis by modifying the global SLEs, as shown.

 K11 K1L   T1   RQ1   qo  

K    
K2 L  T2   RQ 2   0 
K12
 21 K22
   
      
   
x

 KL1 KL 2 ... KLL  TL   RQL   0  L

Note:
q0 is input as +ve value if heat flows out of the body and as –ve value if heat is
flowing into the body. Do not alter the negative sign in the global SLEs above.
SME 3033 FINITE ELEMENT METHOD
Exercise 4
Reconsider the composite wall problem in Exercise 6-3. Suppose there is
no internal heat generation in material A. Instead, a heat flux of q = 1500
W/m2 occurs at the left side of the wall.
Write the global system of linear equations for the plane wall and impose
the specified heat flux boundary condition.

q  1500 W/m2
kA  75 W/m  K
SME 3033 FINITE ELEMENT METHOD
Example 1
A composite wall consists of three
layers of materials, as shown. The
ambient temperature is To = 20 oC.
Convection heat transfer takes
place on the left surface of the wall
where T∞ = 800 oC and h = 25
W/m2oC.
Model the composite wall using
three heat transfer elements and
determine the temperature
distribution in the wall.
SME 3033 FINITE ELEMENT METHOD
Solution
1. Write the element conductivity matrices

20  1 1 50  1 1
 kT 
1
  kT 
 3

0.3  1 1  0.15  1 1 
W W
2o
;
m C m 2 oC

30  1 1
 kT 
 2

0.15  1 1  m 2 oC
W

2. Assemble the global conductivity matrix

 1 1 0 0 
 1 4 3 0 
 KT   66.7  0 3 8 5
W

 
m 2 oC
 0 0 5 5 
SME 3033 FINITE ELEMENT METHOD

3. Write the global system of linear equations

 KT T  RQ 
 1 1 0 0  T1   R1 
 1 4  3 0  T   R 
66.7    2    2 
 0  3 8  5 T3   R3 
    
 0 0  5 5  T4   R4 

4. Write the element heat rate vector

Since there is NO internal heat generation, Q in the wall, the heat rate vector

r   r   r 
for all elements are
0 
 
1 2 3

0 
Q Q Q
SME 3033 FINITE ELEMENT METHOD

5. Write the global system of linear equations

 1 1 0 0  T1  0 
 1 4 3 0  T  0 
66.7    2    
 0 3 8 5 T3  0 
    
 0 0 5 5  T4  0 

6. Impose convection and specified temperature boundary conditions (T4 = 20C)

results in modified system of linear equations

1.375 1 0 0  T1   (25  800) 

 1   T   
   
66.7   2   
4 3 0 0
 0 3 8 5 T3  0  (5  66.7)  20 
   
 0 0 5 5  T4   0 
SME 3033 FINITE ELEMENT METHOD

7. Solving the modified system of linear equations yields

T1  304.6 
T  119.0 
 2  o
   C
 3  
T4   20.0 
T 57.1
SME 3033 FINITE ELEMENT METHOD
Example 2
Heat is generated in a large plate (k = 0.8 W/moC) at a rate of 4000 W/m3.
The plate is 25 cm thick. The outside surfaces of the plate are exposed to
ambient air at 30oC with a convection heat transfer coefficient of 20 W/m2oC.
Model the wall using four heat transfer elements and determine: (a) the
temperature distribution in the wall, (b) heat flux, and (c) heat loss from the
right side of the wall surface.

Data:

k  0.8
W
moC
h  20
W
moC
T  30 o C
SME 3033 FINITE ELEMENT METHOD
Example 2: Nastran Solution

55 C 55 C

84.3 C 84.3 C

94 C
SME 3033 FINITE ELEMENT METHOD
Solution
The finite element model for the plane wall is shown below.

h, T
T1 1 T2 2 T3 3 T4 4 T5
x
h, T
1 2 3 4 5

1. Element conductivity matrices.

Since the element length and thermal conductivity are the same for all elements,
we have
 12.8 12.8
 kT 
1  12.8 12.8
   kT 
 3
  m 2 oC
W

W
 12.8 12.8  m 2 oC  12.8 12.8 

 12.8 12.8
 kT 
 2  12.8 12.8
   kT 
 4
 
W
 12.8 12.8  m C
W
  
2o
12.8 12.8 m 2 oC
SME 3033 FINITE ELEMENT METHOD

2. Assemble the global conductivity matrix,

1 2 3 4 5
 12.8 12.8 0 
 12.8 25.6 12.8 
0 0
 
 KT    0 12.8 25.6 12.8 0 
0 0

 
   
 0 12.8 12.8 
0 0 12.8 25.6 12.8
0 0

Note: Connectivity with the global node numbers is shown.

SME 3033 FINITE ELEMENT METHOD

3. Heat rate vector for each element

Since the magnitude of internal heat generation and length of all
elements are the same, we have

r  4000  0.0625 1 125

1
   
1 125
Q
2 4. Assemble the global heat rate

rQ  4000  0.0625 1 125

vector, we get
 2
    125 
2 1 125 250 

RQ   250
 
r  4000  0.0625 1 125
 3
   
W
1 125 250  m2
Q

 
2

rQ  4000  0.0625 1 125 125 

 4
   
2 1 125
5. Write the system of linear equation,  KT T  RQ 
SME 3033 FINITE ELEMENT METHOD

 12.8  12.8 0  T1  125 

 12.8 25.6  12.8  T  250
0 0
 0 0   2   
 0  12.8 25.6  12.8 0  T3   250
    
      
 0 
0 0 12 .8 25 .6 12 .8 T 250
 12.8 12.8  T5  125 
4

0 0

6. Impose convection boundary conditions on both sides of the wall,

12.8  20  12.8  T1  125  2030

  12.8  T   
0 0 0
 25 .6  12 .8 0 0   2   250 
  12.8 25.6  12.8  T3    
    
0 0 250
      
  12.8 12.8  20 T5  125  2030
0 0 12 .8 25 .6 12 .8 T4 250
0 0 0
SME 3033 FINITE ELEMENT METHOD
7. Solving the modified system of linear equations by using Gaussian
elimination method, we obtain the temperatures at the global nodes
as follows,

T1  55.0 
T  84.3
 2   
T3   94.0 
T  84.3
o
C

 4  
T5  55.0 

h, T
T1 T2 T3 T4 T5
x
h, T
1 2 3 4 5

Note: Notice the symmetry of the temperature distribution.

SME 3033 FINITE ELEMENT METHOD

8. Compute the heat flux and heat loss.

a) Heat flux through the wall
Consider the 4th element. Using the Fourier’s law, we have

T1 
q  k   1 1  
1
le T2  The heat flux through the

 1 1 55.0
84.3
wall is not constant due to
q  0.8 
1
 
the heat generation Q that
0.0625 occurs in the wall.
q  375
W
m2

b) Heat loss from the right side of the wall, per unit surface area.
Using the Newton’s law of cooling, we have

q  h Twall  T   20   55  30   500
W
m2