1

Finite Element Analysis
Bijan Sobhi
2
Summary
 Introduction
 Background
 Electromagnetics: governing equations
 Other disciplines
3
History
 Goes back to Archimedes’ famous
studies on circle 250 B.C./
approximating circle to polygon.
 Structural analysis….
 …since of the appearance of first
computer around 1950.
4
Analytical vs Numerical
 Analytical (closed)
 Simplifying assumptions
 Numerical/discrete
 No simplifying assumptions
 Allow us to solve problems with complex
geometry
 Error/approximations can be minimised by
increasing disceretisation
5
Other Numerical Methods
/Discretisation Method
 Finite Difference Method (FDM) (only node
values)
 Finite Element Method (all the triangle
surface points are known) (FEA is the most
accepted method (due to its simplicity and
generality)
 Finite Volume Method
 Boundary Element Method
 Meshless Method


6
Design Models
 Lumped parameter models (coarse)
 Distributed parameter (or discrete)
models
 Field models

7
Lumped Parameter Models
 Insensitive to topological changes
 Based on simplifying and linear
assumptions
 Computationally inexpensive
8
Distributed Parameter (or
Discrete) Models
 Sensitive to topological changes
 Moderate computational expense
 Easy to interpret
9
FE, FD, … Models
 Sensitive to topological changes
 Computationally very expensive
 Difficult to interpret
 Accounts for non-linearities
10
Types of FEM Analysis
 Static/Quasi steady-state
 Dynamic/Transient
 Linear
 Nonlinear
 1D
 2D
 3D
11
FEM shapes
 Triangle (they can be fitted into any
shape
 Quadrantal
 Curvilinear
12
Preliminary Math
Requirements
 Vectors
 Dot products
 Cross products
 Coordinate systems (Cartesian,
Spherical, cylindrical)
 Etc…
13
Math Sample Equations
z
z
f
y
y
f
x
x
f
ˆ ˆ ˆ
c
c
+
c
c
+
c
c
= V
z
x
A
y
A
y
z
A
x
A
x
y
A
z
A
A
y
x x z z
y
ˆ ˆ ˆ
(
¸
(

¸

c
c
÷
c
c
+
(
¸
(

¸

c
c
÷
c
c
+
(
¸
(

¸

c
c
÷
c
c
= × V

z
A
y
A

x
A
A
z
y
x
c
c
+
c
c
+
c
c
= V

.
z
z
f
y
y
f
x
x
f
f
ˆ ˆ ˆ
c
c
+
c
c
+
c
c
= V
z A y A x A A
z
2
y
2
x
2 2
ˆ ˆ ˆ
V + V + V = V

2
2
2
2
2
2
2
z
f
y
f
x
f
f f .
c
c
+
c
c
+
c
c
= V = V V
14
Maxwell’s Equations

 In order to solve electromagnetic field
problems, it is necessary to solve
Maxwell’s Eq in differential or Integral
ρ D . = V

0 E . = V

J H
 
= × V
0 B . = V

(1)
(2)
(3)
(4)
15
Poisson’s Equation: Magnetics
A ) A . ( A
2
  
V ÷ V V = × V × V
J A
2
= V µ ÷
÷1
0 A = × V V

.
A B
 
× V =
z J J
z
ˆ
=

0
z
=
c
c
z A
2
ˆ
z
V In 2D FEA
From (1)
From (2)
16
Boundary Conditions
 Boundary is where the surface do not
feel what is happening inside.
 Neumann(dA/dn=0) at boundary
interfaces
 Dirichlet (A=0) at boundary interfaces

17
FEM Solution Process
 Starting from known nodes (eg. Boundary
conditions and material properties
 Setting the algebraic equations for the
triangles
 Solving the unknowns values of nodes by
using unknown values
 Using triangle properties to interpolate the
node values to the whole surface
 Equipotential lines do not intersect each other
18
Steps in FEA
 Pre Processing
 Forming the model geometry/drawing the model
geometry/
 Define physical parameters
 Define boundary conditions
 Meshing
 Solving
 Post processing
 All the above steps can be
 Programmed
 Manual
19
Other disciplines
J A
2
= V µ ÷
÷1
0 or c b a = V ÷
2
Magnetics
Others (eg, fluid flow, Heat flow, etc)
20
Equations for Incompressible
Fluid Flow with Heat Transfer
Continuity equation
x-momentum
y-momentum
Energy equation
where o is the angle between the gravity force vector
and x-axis

Sign up to vote on this title
UsefulNot useful