01/04/2022

Chapter 4
FINITE ELEMENT METHOD

Numerical methods for heat transfer

2
Learning Objectives

LEARNING OBJECTIVES
After this chapter you will be able to:
- Understand the Finite Element Method
- Express the Galerkin Method for the FEM
-Solve steady, unsteady one-dimensional conduction
problems numerically using the FEM
- Apply Ansys Workbench to solve one-, two- or three-
dimensional conduction problems using FEM.

Numerical methods for heat transfer

3
4.1 Introduction
4.1 Introduction
In the FEM, the domain is divided into elements (each
element has a certain number of nodes). The PDE is
multiplied by a weight function before being integrated over a
control volume.
4.1.1 Basic steps
1. Grid generation: Discretize the domain into elements. Each
element is formed by a certain number of nodes.
2. Select shape functions(hàm hình dạng): that represents the
variation of the Temperature over an element
3. Form element equations: Develope equations of an element.
Number of equations are equal to number of nodes.
4. Assemble (lắp ghép) the element equations: Assembly of the
element equations to form the global equation.
5. Solve the global equation: temperature of nodes.
Numerical methods for heat transfer

1
 01/04/2022

4
4.1 Introduction
4.1.2 Elements and shape functions
The functions to determine the temperature within an
element by interpolating the nodal temperatures are called
shape functions, or interpolating functions.
+ One-dimensional linear element

The linear temperature field in the element is represented by

(4.1)
where α1, and α2 are constants.

Numerical methods for heat transfer

5
4.1 Introduction
From the above equations, we get

substituting the values α1, and α2 to Eq. (4.1)

where Ni and Nj are called shape functions or Interpolation

functions

Numerical methods for heat transfer

6
4.1 Introduction
is the shape function matrix is the
vector of unknown temperatures( véc tơ nhiệt độ nút).
Example: Calculate the temperature of an 8-cm long bar at a
distance of 5 cm from one end where the temperature is
120oC with the other end at a temperature of 200oC. Assume
the temperature variation between the two end points as
being linear.

Temperature distribution over an element

at x = 5 cm, using local coordinate

Numerical methods for heat transfer

2
 01/04/2022

7
4.1 Introduction
+ One-dimensional quadratic(bậc 2) element

The temperature field in the element is represented by

(4.2)
The temperature at these three points in local coordinate

Numerical methods for heat transfer

8
4.1 Introduction
From the above three equations, we obtain

substituting the values α1, α2 and α3 to Eq. (4.2)

+ One-dimensional cubic(bậc 3) element

The temperature field in the element is represented by

Numerical methods for heat transfer

9
4.1 Introduction
4.1.3 Form element equations
Element equations is in the matrix equation form

where [k] is the thermal stiffness matrix( ma trận dẫn nhiệt),

{T} is the vector of unknown temperatures and {f} is the
thermal load.
-Several methods to obtain the finite element equation:
1. Direct Approach,
2. Variational(biến phân) Approach
3. Weighted Residual(thặng dư có trọng số) Method
The method of weighted residuals provides a very powerful
approximate solution introduced here.
Numerical methods for heat transfer

3
 01/04/2022

10
4.1 Introduction
4.1.4 The method of weighted residuals
The PDE of an element can be represented by
Ex.
with

The method of weighted residual requires

the weighting functions

Different types of weighted residual methods: Point
Collocation Method, Subdomain Collocation Method, Least
Square Method, Galerkin Method.
Numerical methods for heat transfer

10

11
4.1 Introduction
4.1.5 The Galerkin method: is one of the most popular
methods for developing the FE equation. The Galerkin
method requires that
Galerkin method

The Galerkin method requires that

4.1.6 Green’s theorem (Green's first identity): with functions

A and B

Numerical methods for heat transfer

11

12
4.2 FEM for steady heat conduction
4.2 FEM for steady heat conduction
4.2.1 Steady state heat conduction in one dimension
Consider the steady heat conduction in a one-dimensional
domain

The differential equation for elements:

Step 1: The fin is discretized into elements

The shape function

Numerical methods for heat transfer

12

4
 01/04/2022

13
4.2 FEM for steady heat conduction
+ The Galerkin method requires that

Which may expanded as

Apply Green’s theorem

Numerical methods for heat transfer

13

14
4.2 FEM for steady heat conduction
Doing Similar, we have

Combine the intergral equations to obtain

We have

The last term associates with the surface heat flux at the two
ends of the element.
Numerical methods for heat transfer

14

15
4.2 FEM for steady heat conduction
+ For the interior element

Element equations

The element [k]

Numerical methods for heat transfer

15

5
 01/04/2022

16
4.2 FEM for steady heat conduction

Integration formula for one-dimensional element:

If A, λ, P and α are all assumed to be constant throughout the

element, we obtain the following [k] matrix:

Numerical methods for heat transfer

16

17
4.2 FEM for steady heat conduction
Integration over Each Element:

+ For boundary elements

- Convection boundary

• Right element

Numerical methods for heat transfer

17

18
4.2 FEM for steady heat conduction

• Left element

- Specific heat flux boundary

• Right element

• Left element

positive: if heat flux into the boundary

negative: if heat flux out of the boundary
zero: if insulated
Numerical methods for heat transfer

18

6
 01/04/2022

19
4.2 FEM for steady heat conduction
To summarise
+The interior element
-The conductance matrix

- The thermal load matrix

+ Boundary elements
• the right element having convection (element n)
-The conductance matrix

- The thermal load matrix

Numerical methods for heat transfer

19

20
4.2 FEM for steady heat conduction
• the left element having convection (element 1)
-The conductance matrix

- The thermal load matrix

+ Specific heat flux boundary

• Left element

• Right element

Numerical methods for heat transfer

20

21
4.2 FEM for steady heat conduction
Example 4.1 Consider the heat conduction in a large plate of
thickness L = 2 cm with λ = 0.5 W/m.K. The faces A and B are
at temperatures of 200°C and 100°C respectively. Assuming
that the dimensions in the y- and z-directions are so large that
temperature gradients are significant in the x direction only,
calculate the steady state temperature distribution. Compare
the numerical result with the analytical solution.
The differential equation:

Solution: Divide the wall into five equal elements, This gives
∆x=0.004m.

Numerical methods for heat transfer

21

7
 01/04/2022

22
4.2 FEM for steady heat conduction
The shape function

+ The Galerkin method requires that

Apply Green’s theorem

Numerical methods for heat transfer

22

23
4.2 FEM for steady heat conduction

Combine the integral equations to obtain

We have

+ For the interior element without the surface heat flux at the
two ends

Numerical methods for heat transfer

23

24
4.2 FEM for steady heat conduction
Element equations

Integration over the element [k]

Numerical methods for heat transfer

24

8
 01/04/2022

25
4.2 FEM for steady heat conduction
For all elements we have

Assembling elements

Applying the boundary conditions

Numerical methods for heat transfer

25

26
4.2 FEM for steady heat conduction
-Analytical solution (exact solution)

-Comparision the FEM result with the analytical solution

Node 1 2 3 4 5 6
Distance(m) 0 0.004 0.008 0.012 0.016 0.02
FEM 200.00 180.00 160.00 140.00 120.00 100.00
Exact 200.00 180.00 160.00 140.00 120.00 100.00
Error(%) 0 0 0 0 0 0
Numerical methods for heat transfer

26

27
4.2 FEM for steady heat conduction
Example 4.2 A large plate of thickness L = 2 cm with constant
thermal conductivity λ = 0.5 W/m.K and uniform heat
generation q = 1000 kW/m3. The faces A and B are at
temperatures of 200°C and 100°C respectively. Assuming that
the dimensions in the y- and z-directions are so large that
temperature gradients are significant in the x direction only,
calculate the steady state temperature distribution. Compare
the numerical result with the analytical solution.

Solution: Divide wall into five equal elements.

∆x=0.004m.

Numerical methods for heat transfer

27

9
 01/04/2022

28
4.2 FEM for steady heat conduction
The Galerkin method requires that

Apply Green’s theorem

Integration over Each Element

Numerical methods for heat transfer

28

29
4.2 FEM for steady heat conduction
For all elements we have

Assembling elements, we obtain

Applying the boundary conditions

Numerical methods for heat transfer

29

30
4.2 FEM for steady heat conduction
-Analytical solution

-Comparision the FEM result with the analytical solution

Node 1 2 3 4 5 6
Distance(m) 0 0.004 0.008 0.012 0.016 0.02
FEM 200.00 244.00 256.00 236.00 184.00 100.00
Exact 200.00 244.00 256.00 236.00 184.00 100.00
Error(%) 0 0 0 0 0 0

Numerical methods for heat transfer

30

10
 01/04/2022

31
4.2 FEM for steady heat conduction
Example 4.3 Aluminum fins of a rectangular profile are used to
remove heat from a surface whose temperature is
Tbase=100°C. The temperature of the ambient air Tf= 20°C. The
thermal conductivity of aluminum is λ = 160 W/m.K (W/m·0C).
The natural convective heat transfer coefficient associated
with the surrounding air is  =30 W/m2.K (W/m2.°C). The fin
length L= 80 mm, 5 mm wide, and 1 mm thick. Determine the
temperature distribution along the fin, including the heat loss
through the tip of the fin. Compare the numerical result with
the analytical solution.

Numerical methods for heat transfer

31

32
4.2 FEM for steady heat conduction
The differential equation for elements:

For elements (1), (2), and (3), the conductance and thermal
load matrices are given by

Numerical methods for heat transfer

32

33
4.2 FEM for steady heat conduction
+ For the the element 4

Assembling elements, and apply boundary condition for node

1, we obtain

Numerical methods for heat transfer

33

11
 01/04/2022

34
4.2 FEM for steady heat conduction
-Analytical solution
where

-Comparision the FEM result with the analytical solution

4 elements

Node 1 2 3 4 5
Distance(m) 0 0.02 0.04 0.06 0.08
FEM 100.00 74.37 58.82 50.48 47.79
Exact 100.00 74.57 59.12 50.81 48.13
Error(%) 0.00 -0.28 -0.50 -0.65 -0.70

Numerical methods for heat transfer

34

35
4.2 FEM for steady heat conduction
4.2.2 Steady State Heat Conduction in 2 or 3 Dimensions
+Two-dimensional linear triangular elements

The temperature field in the element is represented by

Equations for the temperature at these three points,

Numerical methods for heat transfer

35

3
6
4.2 FEM for steady heat conduction
+Triangular elements (phần tử tam giác)

Quadratic(bậc 2) triangular elements Ten-node cubic(bậc 3) triangular

+Three-dimensional elements

+Quadratic triangular three-dimensional elements

Numerical methods for heat transfer

36

12
 01/04/2022

3
7
4.2 FEM for steady heat conduction

Numerical methods for heat transfer

37

38
4.2 FEM for steady heat conduction
Example 4.4 A major objective in advancing gas turbine engine
technologies is to increase the temperature limit associated
with operation of the gas turbine blades. This limit determines
the permissible turbine gas inlet temperature, which, in turn,
strongly influences overall system performance. In addition to
fabricating turbine blades from special, high-temperature,
high-strength superalloys, it is common to use internal cooling
by machining flow channels within the blades and routing air
through the channels. We wish to assess the effect of such a
scheme by approximating the blade as a rectangular solid in
which rectangular channels are machined. The blade, which
has a thermal conductivity of = 25W/m.K, is 6mm thick, and
each channel has a 2mm× 6mm rectangular cross section, with
a = 4mm spacing between adjoining channels.
Numerical methods for heat transfer

38

39
4.2 FEM for steady heat conduction

Under operating conditions for which α0 = 1000W/m2.K,

T∞ =1700K, αi=200 W/m2.K, and Ti =400K, determine the
temperature field in the turbine blade and the rate of heat
transfer per unit length to the channel. At what location is the
temperature a maximum?

Numerical methods for heat transfer

39

13
 01/04/2022

40
4.2 FEM for steady heat conduction
Result using Finite Difference Method [*]

Result using Steady thermal analysis, ANSYS Workbench

[*] Introduction to Heat Transfer, 6th Edition, Frank P. Incropera (Author), David P.
DeWitt (Author)
Numerical methods for heat transfer

40

41
4.2 FEM for steady heat conduction
Result using ANYS APDL

Steady thermal analysis

ANSYS Workbench
ANYS APDL

(K)

Numerical methods for heat transfer

41

42
4.3 FEM for unsteady heat conduction
4.3 FEM for unsteady heat conduction
4.3.1 One-dimensional unsteady heat conduction
Consider the unsteady heat conduction in a 1D case

The Galerkin method requires that

Apply Green’s theorem

Numerical methods for heat transfer

42

14
 01/04/2022

43
4.3 FEM for unsteady heat conduction

We have

a more convenient form on each element

(4.3)

The element [k]

Numerical methods for heat transfer

43

44
4.3 FEM for unsteady heat conduction

The element

Apply

We have

Numerical methods for heat transfer

44

45
4.3 FEM for unsteady heat conduction
4.3.1.1 Explicit scheme

Replace to Eq.(4.3) and integration over the element

The above equation can be rearranged as follows

For stability condition

for the 1D case, reducing the mesh size by one-half increases

the computer time eight-fold a major disadvantage.
Numerical methods for heat transfer

45

15
 01/04/2022

46
4.3 FEM for unsteady heat conduction
Examples
A thin plate is initially at a uniform temperature of 200°C. At a
certain time t = 0 the temperature of the right side of the
plate is suddenly reduced to 0°C. The other surface is
insulated. Use the explicit and implicit methods in conjunction
with a suitable time step size to calculate the transient
temperature distribution of the slab and compare it with the
analytical solution at time (i) t = 40 s, (ii) t = 80 s and (iii) t =
120 s. The data are: plate thickness L = 2 cm, thermal
conductivity λ = 10 W/m.K and ρC=10×106 J/m3.K.

Numerical methods for heat transfer

46

47
4.3 FEM for unsteady heat conduction
The differential equation for interior nodes:

the initial conditions are T=200 at t=0

and the boundary conditions are

The analytical solution is given in Özivik (1985) as

where and

Numerical methods for heat transfer

47

48
4.3 FEM for unsteady heat conduction

+ Explicit scheme
(4.4)
Consider

Numerical methods for heat transfer

48

16
 01/04/2022

49
4.3 FEM for unsteady heat conduction

Assembling the global stiffness matrix, [K]

The global matrix [C] is

Numerical methods for heat transfer

49

50
4.3 FEM for unsteady heat conduction
Substituting into Eq. (4.4), we get at ∆t =0.2s
[C]

([C]-[K]∆t)

Numerical methods for heat transfer

50

51
4.3 FEM for unsteady heat conduction
Apply boundary condition

+1
13333 6667 0 0 0 0 ⎡ 1 +1 ⎤
⎡ 6667 26667 6667 0 0 0 ⎤⎢ 2 ⎥
⎢ ⎥ +1
⎢ 0 6667 26667 6667 0 0 ⎥⎢ 3 ⎥
⎢ 0 0 6.667 26667 6667 0 ⎥ ⎢ 4 +1 ⎥
⎢ 0 0 0 6.667 266676 6667⎥ ⎢ 5 +1 ⎥
⎣ 0 ⎢ ⎥
0 0 0 0 1 ⎦ ⎣ +1 ⎦
6
8333 11667 0 0 0 0 ⎡ 1⎤
⎡11667 16667 11667 0 0 0 ⎤⎢ 2 ⎥
⎢ ⎥
0 11667 166676 11667 0 0 ⎥⎢ 3 ⎥
=⎢
⎢ 0 0 11667 216667 11667 0 ⎥⎢ 4 ⎥
⎢ 0 0 0 11667 16667 11667⎥ ⎢ 5 ⎥
⎣ 0 0 0 0 0 0 ⎦⎣ 0 ⎦

Numerical methods for heat transfer

51

17
 01/04/2022

52
4.3 FEM for unsteady heat conduction
Results for Explicit method (∆X=0.004, ∆T=2s)
Time =40s Time =80s Time =120s
Node
Numerical Analytical %error Numerical Analytical %error Numerical Analytical %error
1 194.29 189.86 2.28 156.94 154.46 1.58 122.63 121.36 1.03
2 187.72 183.81 2.08 149.49 147.27 1.49 116.64 115.46 1.02
3 166.24 163.71 1.52 127.68 126.08 1.25 99.26 98.30 0.97
4 126.68 125.71 0.77 93.23 92.33 0.96 72.16 71.50 0.91
5 69.12 69.04 0.10 49.21 48.85 0.73 37.95 37.62 0.86
6 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

(∆X=0.004, ∆T=2s)
Numerical methods for heat transfer

52

53
4.3 FEM for unsteady heat conduction
4.3.1.2 The fully implicit scheme

Replace to Eq. (4.3) and integration over the element

-Advantage: unconditionally stable with ∆t. smaller ∆t, the

better the accuracy.

Numerical methods for heat transfer

53

54
4.3 FEM for unsteady heat conduction
4.3.1.3 Crank-Nicolson scheme

From Eq. (4.3) we have

Rearranging

 The time step limitation is less restrictive than the Explicit

method
Numerical methods for heat transfer

54

18
 01/04/2022

55
Exercise
Exercise 1: A wall of industrial oven consists of three different
materials. The first layer is composed of 5 cm of insulating
cement with a clay binder that hat thermal conductivity of
0.08 W/m. oC. The second layer is made from 15cm of 6-ply
asbestos board with a thermal conductivity of 0.074 W/m.oC.
The exterior consists of 10cm brick with a thermal conductivity
of 0.72W/m2.oC. The inside wall temperature is 200oC, and the
outside air is 30oC with a convection coefficient of 40W/m2. oC.
Determine the temperature distribution along the wall.

Numerical methods for heat transfer

55

56
Excercise
Exercise 2: Analysis of cold plate used to thermally control IBM
multi-chip, thermal conduction module.

- Heat dissipated in the chips is transferred by conduction through

spring-loaded aluminum pistons to an aluminum cold plate.
- Nominal operating conditions, assume a uniformly distributed heat
flux of q=105W/m2 at the base of the cold plate.
Find: (a) Cold plate temperature distribution for the prescribed
conditions. (b) Options for operating at larger power levels while
remaining within a maximum cold plate temperature of 40°C
Numerical methods for heat transfer

56

19