Harmattan

2009/2010
SESSION
PREPARED BY:
S.A. ADIO
MECHANICAL ENGINEERING DEPARTMENT
OBAFEMI AWOLOWO UNIVERSITY

MEE 403 MODULE 4

2-D STEADY STATE CONDUCTION
.
3.0 2-Dimensional Steady State Conduction
To this point we have restricted our attention to conduction problems in which the
temperature gradient is significant for only one coordinate direction. We now wish to analyze the
more general case of two-dimensional heat flow. For steady state with no heat generation, the
Laplace equation (equation 2.13) applies
  2T  2T 
k  2  2   0 3.1
 x y 
y
T2 The solution to this equation may be obtained by analytical,

T1 numerical, or graphical techniques.

T2
dy Equation 3.1 will give the temperature in a two-
dimensional body as a function of the two independent
x
dx T1 space coordinate x and y. Then the heat flow in x and y
directions may be calculate from equation 2.1

T
q x  kAx
x
3.2
T
q y  kAy
y
The total heat flow at any point in the material is the resultant of the q x and q y at that

point. The total heat-flow vector is directed so that it is perpendicular to the lines of constant
temperature of the material as in figure 24.0

T
q y  kAy
y
q  qx  q y
T
Isotherm q x  kAx
x

Fig. 24.0
3.1 Numerical Methods
In many practical situation the geometry or boundary conditions are such that analytical
solution are not obtainable, or the solution involves complex series solution. Therefore,
numerical methods such as finite element method, finite difference method, boundary element
method are the best alternatives. As part of its strength, numerical methods can readily be
extended to three-dimensional problems, because of the ease of application of finite-difference
method, we seek to learn its application to two dimensional heat-transfer problems.

3.2 Nodes – Formation (Mesh Generation)

The first step in any numerical analysis is the generation of
(m, n+1)
y
grids (descritization) for the problem at hand as in fig. 24.
(m-1, n) m,n (m+1, n) Consider figure 25.0, the nodal points are designate such
y that the ‘m’ locations indicating x increment and the n
(m,n-1)
locations indicating the y increment
x x
Fig 25.0
Using equation 3.1, it is possible to establish the temperature at any of these nodal points.
Using finite difference method, we seek to approximate differential increments in the
temperature and space coordinates; and the smaller these finite increments are, the more closely
the true temperature distribution will be approximated. For figure 25.0 the temperature gradients
may be written as follows;
T Tm1, n  Tm,n

x 1
m , n x
2

T Tm, n  Tm 1, n
 3.3
x 1
m , n x
2

T Tm, n 1  Tm,n

y m,n
1 y
2

T Tm,n  Tm, n 1

y m,n 
1 y
2

For Equation 3.1, the finite difference form of this equation is given as
 T T

x m 1 ,n x 1
 2T 2 m ,n
 2

x 2 m,n
x

T T
 3.4
y m ,n  1 y 1
 2T 2 m,n 
 2

y 2 m,n
y

Substituting equation (3.3) into equation (3.4), we have

 2T Tm1, n  Tm1n  2Tm,n


x 2 m, n x 2
3.5
 T
2 Tm, n 1  Tm,n 1  2Tm,n

y 2 m, n y 2
If the nodal length is chosen such that
x  y
Equation (3.5) becomes
Tm+1, n + Tm-1,n + Tm,n+1 + Tm,n-1 – 4Tm,n = 0 3.6

Consider the figure below,

y
x Using energy balance method to analyze the heat flow into
x
m, n+1 the central node (m,n). Using Fourier’s law
qcond y T
q  kA ----------------- 
qcond x
m-1, n m,n qcond m+1,n
assume the thickness of the plate in the figure is 1 (unity).
The Area is either
qcond
m, n-1
x, or y for the control volume when considering the heat
flow directions from the neighbouring nodes to node m,n.
Fig. 25
Consider all the heat inflow into the node m, n we have that
 Tm1,n  Tm,n
q m1,nm,n  k y.1
x
Tm1,n  Tm,n
q m1,nm,n  k y.1
x
3.7
Tm,n 1  Tm,n
q m,n 1m,n  k x.1
y
Tm,n 1  Tm,n
q m,n 1m,n  k x.1
y
Adding all these equation and equating x = y.
We realize that equation (3.7) becomes equation (3.6)
For a case of internal energy generation, we have that equation 3.7 becomes.

qx 
2
Tm, n+1 + Tm, n-1 + Tm+1, n + Tm-1, n +  4Tm,n  0 3.8
k
If there is no internal generation of heat energy equation (3.8) becomes equation (3.7)

x Consider figure 26, for finite difference equation

m, n+1 with an internal corner of a solid with surface
qcond
y
convection.
qcond m+1,n
m-1, n

qcond qcond
qcond
m, n-1

Fig. 26

The internal node (m,n) represents the three-quarter shaded section and exchanges energy by
convection with an adjoining fluid at T∞. Conduction to the nodal region (m,n) occurs along four
different lanes from neighbouring nodes in the solid. The conduction heat rates qcond may be
expressed as:
 Tm1,n  Tm,n
qm1,n m,n  k y.1
x
Tm,n1  Tm,n
qm,n 1m,n  k x.1
y 3.9

 y  Tm1,n  Tm,n
qm1,n m,n  k  .1
 2  x

 x  Tm,n1  Tm,n
q m,n 1m,n  k  .1
 2  y
Note the Areas for conduction for nodal region

 2 2
 
(m+1, n) & (m, n-1) are  y  and x respectively

 x   y 
qconv  h  1T  Tm,n   h  1T  Tm,n  3.10
 2   2 
Overall heat transfer is the addition of qcond + qconv, therefore we have
hx hx 
Tm1,n  Tm,n1 
1
Tm1,n, Tm,n1  

T   3  Tm,n  0 3.11
2 k  k 
Note x = y
Table 3.1: Nodal Finite-Difference Equation for Different Configurations
m, n+1

y
m, n
m+1, n Tm, m+1 + Tm,n-1 + Tm+1, n + Tm-1, n – 4Tm,n = 0
m-1, n
Case 1. Interior node
m, n -1
x

x
m, n+1 2(Tm-1, n + Tm, n+1) + (Tm+1, n + Tm,n-1)
hx  hx 
m-1, n m, n
m+1, n 2 T  2 3  Tm,n  0
y T∞, h
k  k 
m, n -1 Case 2. Node at an internal corner with convection

m, n+1
y 2(Tm-1, n + Tm, n+1 + Tm,n-1)
m, n
T∞, h
hx  hx 
m-1, n
2 T  2  2 Tm,n  0
m, n -1
k  k 
x
Case 3. Node at a plane surface with convection

m-1, n T∞, h
m, n
hx  hx 
y (Tm, n-1 + Tm-1, n)  2 T  2  1Tm,n  0
k  k 
m, n -1
x Case 4. Node at an external corner with convection

m, n+1
y
m, n
q” 2q " x
m-1, n (2Tm-1, n + Tm, n+1 + Tm,n-1)   4Tm,n  0
k
m, n -1
x Case 5. Node at a plane surface with uniform heat flux

Note : x = y
 To utilize the numerical method, equation (3.6) must be written for each node within the
material and the resultant system of equations solved for the temperatures at the various nodes.
For Example consider the figure below (fig. 27)

T = 500oC Writing the nodal equations fore nodes 1,2,3,&4 we have

using equation 3.6
1 2 100 + 500 + T2 + T3 – 4T1 = 0
100oC T1 + 500 + 100 + T4 – 4T2 = 0
3 4
100oC 100 + T1 + T4 + 100 – 4T3 = 0
T3 + T2 + 100 + 100 – 4T4 = 0
100oC
These equations are solved to determine T1, T2, T3 and T4, which in this case we have T1
= T2 = 250oC and T3 = T4 = 150oC.

3.3 Solution Techniques

* Matrix Inversion Method
Numerical method is simply a means of approximating a continuous temperature
distribution with the finite nodal element. The more nodes taken, the closer the approximation.
Also this means more equations and by inference more cumbersome solutions. In practical
problems the selection of a large number of nodes may be unnecessary because of uncertainties
in boundary conditions.
The nodal equations may be written as
a11T1  a12T2  a13T2  a13T3  . . ..  a1nTn  C1
a21  a22T2  a23T2  a23T3  . . ..  a2 nTn  C2
: 3.12
.
an1T1  an 2T2  an 3T2  an 3T3  . . ..  annTn  Cn

where the quantities a11, a12, . . . , C1, C2 . . . Cn are known coefficients and constants involving
quantities such as x, h, k and T∞. T1, T2 . . .Tn are the unknown nodal temperatures. By using
matrix notation
 [A] [T] = [C] 3.13

where
a11 a12 . . . a1 
a  T1  C1 
 21 a22 . .. a2 n  T  C 
A  : : : ; T   2 ; C   2
  :  : 
: : :     
an1 an 2 ann  Tn  C n 

Using Matrix inversion method, the coefficient matrix [A] is a square (n x n) matrix, the matrices
[T] and [C] are column matrix (column vector)
The solution to equation (3.13) is obtained such that
[T] = [A]-1 [C] 3.14
where [A]-1 is the inverse of [A], and is defined as

b11 b12 . . . b1 
b b . .. b 
 21 22 2n 

[A]-1 = : : : 
 
: : : 
bn1 bn 2 bnn 

Evaluating the right hand side of the equation above (eqn 3.14) we have
T1  b11C1  b12C2  . . .  b1nCn
T2  b12 C1  b22C2  ...  b2 nCn
3.15
:
Tn  bn1Cn  bn 2C2  ...  bnnCn

* Gauss- Seidel Iteration

Equation 3.12 can also be solved using the method of gauss-Seidel Iteration. The following are
the steps.
1. The equations should be recorded to provide diagonal elements whose magnitudes are
larger than those of other elements in the same row. That is, it is desirable to sequence the
equations such that /a11/ > /a12/, /a13/ …. /a1n/; /a22/ > /a21/, /a23/ … /a2n/ etc.
2. Each of the N equations should be written in explicit form for the temperature associated
with its diagonal element. Each temperature in the solution vector would then be of the
form.
Ci i 1 k  n aij
T1k    T j   a T j
k 1

Cii j 1 j  i 1 ii

where i = 1, 2, ………. n. The superscript k refers to the level of the iteration.

3. An initial (k = 0) value is assumed for each temperature Ti . Subsequent computations
may be reduced by selecting values based on rational estimates of the actual solution.
4. Setting k = 1 in no 2 above, values of Ti1 are then calculated by substituting assumed
(second summation, k – 1 = 0). This is the first iteration step (k =1).
5. Using the equation no 2 above the iteration procedure is continued by calculating new
values of Tik  from the T jk  values of the current iteration, where 1  j  i  1 , and the

T k 1 values of the previous iteration, where i  j   j  n

6. The iteration is terminated when a prescribed convergence criterion is satisfied

TIk   T jk 1   3.16

Where ε is the acceptable error in the iteration process

If step 1 above can be accomplished for each equation the resulting system is said to be
diagonal dominant and the rate of convergence is maximized (the number of required iterations
is minimized).

Example 5
A large industrial furnace is supported on a long column of the fireclay brick, which is 1m by 1m
on a side. During steady-state operation, installation is such that three surfaces of the column are
maintained at 500 K while the remaining surface is exposed to an airstream for which T∞ = 300
K and h = 10 W/m2.K. Using a grid of x = y = 0.25 m, determine the two-dimensional
temperature distribution in the column and the heat rate to the airstream per unit length of
column.
Solution 5
Known: Dimensions and surface conditions of a support column
Find: Temperature distribution and heat rate per unit length
Schematic:

x = Ts=500 k
0.25m
1 2 1
Fire clay brick
y = 0.25m
3 4 3

Ts=500 k Ts=500 k
5 6 5

7 8 7

Air T∞=500 k
h = 10 W/m2. k
Assumptions:
1. Steady-state conditions
2. Two-dimensional conduction.
3. Constant properties
4. No internal heat generation.

Properties: fireclay brick (T 478 K): k = 1 W/m. K.

Analysis: The prescribed grid consists of 12 nodal points at which the temperature is unknown.
However, the number of unknowns is reduced to 8 through symmetry, in which case the
temperature of nodal points to the left of the symmetry line must equal the temperature of those
to the right.
Nodes 1, 3, and 5 are interior points for which the finite-difference equations may be inferred
from Equation 4.29. Hence
Node 1: T2 + T3 + 1000 – 4T1 = 0
Node 3: T1 + T4 + T5 + 500 – 4T3 = 0
Node 5: T3 + T6 + T7 500 – 4T5 = 0
Equations for points 2, 4, and 6 may be obtained in a like manner or, since they lie on a
symmetry adiabatic, by using Equation 4.42 with h = 0. Hence
Node 2: 2T1 + T4 + 500 – 4T2 = 0
Node 4 T2 + 2T3 + T6 – 4T4 = 0
Node 6: T4 + 2T5 + T8 – 4T6 = 0
From Equation 4.42 and the fact that h x /k = 2.5, it also follows that

Node7: 2T5 + T8 + 2000 – 9T7 = 0

Node 8: 2T6 + 2T7 + 1500 – 9T8 = 0
Having the required finite-difference equations, matrix inversion solution may be obtained by
ordering them from node 1 to node 8 as follows:

– 4T1 + T2 + T3 + 0 + 0 + 0 + 0 + 0 = – 1000

2T1 + – 4T2 + 0+ T4 + 0 + 0 + 0 + 0 = – 500

T1 + 0 + – 4T3 + T4 + T5 + 0 + 0 + 0 = – 500

0+ T2 + 2 T3 + – 4T4 + 0 + T6 + 0 + 0 =0

0+ 0 + T3 + 0 + – 4T5 + T6 + T7 + 0 = – 500

0+ 0 + 0+ T4 + 2T5 +– 4T6 + 0 + T8 = 0

0+ 0 + 0+ 0 + 2T5 + 0 + – 9T7 + T8 = – 2000

0+ 0 + 0+ 0 + 0 + 2T6 + 2T7 – 9T8 = – 1500

In matrix, notation, following Equation 4.48, these equations are of the form [A][T] = [C], where
  4 1 1 0 0 0 0 0   1000 
 2 4 0 1 0 0 0 0   5000 
   
 1 0 4 1 1 0 0 0   5000 
   
0 1 2 4 0 1 0 0  C    0
A     500 

0 0 1 0 4 1 1 0
   
 0 0 0 1 2 4 0 1   0 
 0 0 0 0 2 0 9 1   2000
   
 0 0 0 0 0 2 2 9   1500 
Using a standard matrix inversion routine, it is a simple matter to find the inverse of [A],
[A]-1, giving
[T] = [A]-1[C]

where
T1  489.30 
T   
 2  485.15 
T3  472.07 
   
[T ]   T4  462.01 
 K
T  436.95 
  
5

T6  418.74 
T  356.99 
 7  
T8  339.05 

The heat rate from the column to the airstream may be computed from the expression
q  x   x  
   2h  Ts  T   xT7  T    T8  T 
L  2   2  
Where the factor of 2 outside the brackets originates from the symmetry condition.
Hence
q
   2  10 W / m .K 10.125m200 K   0.25 m 56.99 K   0.125m39.05 K   883W / m
2

L

Gauss-Seidel
T1k   0.25T2k 1  0.25T3k 1  250
T2k   0.50T1k   0.25T4k 1  125

T3k   0.25T1k   0.25T4k 1  0.25T5k 1  125

T4k   0.25T2k   0.50T3k   0.25T6k 1

T5k   0.25T3k   0.25T6k 1  0.25T7k 1  125

T6k   0.25T4k   0.50T5k   0.25T8k 1

T7k   0.2222T5k   0.1111T8k 1  222.22

T8k   0.2222T6k   0.2222T7k   166.67

Having the finite-difference equations in the required form, the iteration procedure may
be implemented by using a table that has one column for the iteration (step) number and a
column for each of the nodes labeled as Ti. The calculations proceed as follows:
1. For each node, the initial temperature estimate is entered on the row for k = 0 values are
selected rationally to reduce the number of required iterations.
2. Using the N finite-difference equations and values of Ti from the first and second rows,
the new values of Ti are calculated for the first iteration (k = 1). These new values are
entered on the second row.
3. This procedure is repeated to calculate Ti k  from the previous values of Ti k 1 and the

current values of Ti k  , until the temperature difference between iterations meets the

prescribed criterion, ε  0.2 K, at every nodal point.

k T1 T2 T3 T4 T5 T6 T7 T8
0 480 470 440 430 400 390 370 350
1 477.5 471.3 451.9 551.3 428.0 411.8 356.2 337.3
2 480.8 475.7 462.5 453.1 432.6 413.9 355.8 337.7
3 484.6 480.6 467.6 457.4 434.3 415.9 356.2 338.3
4 487.0 482.9 469.7 459.6 435.5 417.2 356.6 338.6
5 488.1 484.0 470.8 460.7 436.1 417.9 356.7 338.8
6 488.7 484.5 471.4 4613 436.5 418.3 356.9 338.9
7 489.0 484.8 471.7 461.6 436.7 418.5 356.9 339.0
 8 489.1 485.0 471.9 461.8 436.8 418.6 356.9 339.0

The results given in row 8 are in excellent agreement with those obtained by matrix inversion,
although better agreement could be obtained by reducing the value of ε. However, given the
approximate nature of the finite-difference equations, the results still represent approximations to
the actual temperatures. The accuracy of the approximation may be improved by using a finer
grid (increasing the number of nodes).