Finite Element Analysis

Dr. Latha Nagendran

Associate Professor
AU-FRG I.C.C
LECTURE 1
ME2353 FINITE ELEMENT
ANALYSIS
INTRODUCTION
Solution to engineering problems – mathematical
modeling- discrete and continuum modeling – need
for numerical methods of solution – relevance and
scope of finite element methods – engineering
applications of FEA
UNIT I FINITE ELEMENT
FORMULATION OF BOUNDARY VALUE
PROBLEMS

Weighted residual methods –general weighted residual

statement – weak formulation of the weighted residual
statement –comparisons – piecewise continuous trial
functions example of a bar finite element –functional
and differential forms –principle of stationary total
potential – Rayleigh Ritz method – piecewise
continuous trial functions – finite element method –
application to bar element
UNIT II ONE DIMENSIONAL FINITE
ELEMENT ANALYSIS

General form of total potential for 1-D applications –

generic form of finite element equations – linear bar
element – quadratic element –nodal approximation –
development of shape functions element matrices and
vectors – example problems – extension to plane truss–
development of element equations – assembly –element
connectivity –global equations – solution methods –
Beam element – nodal approximation – shape functions
– element matrices and vectors – assembly – solution –
example problems–
UNIT III TWO DIMENSIONAL FINITE
ELEMENT ANALYSIS
Introduction – approximation of geometry and field variable
– 3 noded triangular elements – four noded rectangular
elements – higher order elements – generalized coordinates
approach to nodal approximations – difficulties – natural
coordinates and coordinate transformations – triangular and
quadrilateral elements – iso-parametric elements – structural
mechanics applications in 2-dimensions – elasticity
equations –stress strain relations – plane problems of
elasticity – element equations – assembly –need for
quadrature formulae – transformations to natural coordinates
– Gaussian quadrature – example problems in plane stress,
plane strain and axisymmetric applications
UNIT IV DYNAMIC ANALYSIS USING
FINITE ELEMENT METHOD

Introduction – vibrational problems – equations of

motion based on weak form –longitudinal vibration of
bars – transverse vibration of beams – consistent mass
matrices– element equations –solution of Eigen value
problems – vector iteration methods –normal modes –
transient vibrations – modeling of damping – mode
superposition technique – direct integration methods
UNIT V APPLICATIONS IN HEAT
TRANSFER & FLUID MECHANICS

One dimensional heat transfer element – application

to one-dimensional heat transfer problems- scalar
variable problems in 2-Dimensions – Applications to
heat transfer in 2-Dimension – Application to
problems in fluid mechanics in 2-D
TEXT BOOK:
1. P.Seshu, “Text Book of Finite Element Analysis”,
Prentice-Hall of India Pvt. Ltd., NewDelhi, 2007.

REFERENCES:
1. J.N.Reddy, “An Introduction to the Finite Element
Method”,McGraw-Hill International Editions
(Engineering Mechanics Series), 1993.
2. Chandrupatla & Belagundu, “Introduction to Finite
Elements in Engineering”, 3rd Edition, Prentice-Hall of
India, Eastern Economy Editions.
3. David V.Hutton, “Fundamentals of Finite Element
Analysis”, Tata McGraw-Hill Edition 2005.
 Why FEM ?

Predictive Method of Analysis

vs
Experimental Analysis
WHY DO WE SIMULATE?
COST EFFECTIVE
FAST
PROVEN METHOD

WE CANNOT TEST
 What is FEM ?

 Determination of the solution for

a complicated problem by
replacing it by a simpler one.

 Geometrically complex domain

represented as a collection of
smaller manageable domains.
 Solution to these geometrically
simple domains is easier.

 Replacing the original complex

geometry as an assemblage of
smaller simple geometry will
result in only an approximate
solution.
 Lower bound solution
Area = 6 x A1
App.Area < actual
A1
Area area

No of sides

Upper bound solution

Area = 6 x A2
Area A2
App.Area > actual area

No of sides
Where FEM ?
Finalized design
 FEM
FEA (finite element analysis), or
FEM (finite element method), was
primarily developed by engineers
using physical reasoning and can
trace much of its origin to matrix
methods of structural analysis.
The finite element method is a
computer aided mathematical
technique that is used to obtain an
approximate numerical solution to
the fundamental differential and/or
integral equations that predict the
response of physical systems to
external effects/influences.
What is meant by external influence?

 When a bar is subjected to an axial

pull ‘P’ it elongates  

When a metallic rod is heated its

temperature rises  

When a beam is subjected to an

external harmonic excitation it
vibrates
 In the above examples the force ‘P’, or
heat flux ‘q’ or harmonic excitation force
constitute the “external influence” that
causes the system to change.

 The elongation, temperature rise or

vibration represents the system’s
response to the external influence.
 Why FEM ?
Mathematical modeling to simulate
physical happening
Any Physical
System
Laws of physics
Mathematical
modeling
Solution
 When FEM ?
Complex geometry

Complex loading

Complex material
properties
 Applications
• Structural Engineering
• Aerospace Engineering
• Automobile Engineering
• Thermal applications
• Acoustics
• Flow Problems
• Dynamics
• Metal Forming
• Medical & Dental applications
• Soil mechanics etc.
 NUMERICAL SOLUTION TECHNIQUES

 
• Weighted Residual Methods - Collocation method
• - Sub domain method
• - Least squares method
• - Galerkin method

•    Finite Difference Method

• Rayleigh Ritz Technique
•    Finite Element Method
•    Boundary Element Method
 FEM
Mathematical modeling to simulate
physical happening
Any Physical
System

Laws of physics

Mathematical modeling

Solution
Mathematical modeling

Field Variable/
dependent variable
Cross-sectional Primary variable
Governing
property Equation

d du 
 EA( x)   A( x)  0
 dx dx 
Independent variable
or spatial co-ordinate
Material
Properties
Example of a taper rod subjected a
point load ‘P’ and its own self weight

P
 A(x) = A0 - (A0-A1) x/l

For equilibrium
( +d) A(x) +  A(x) dx -  A(X) = 0 --(1)
i.e) d A(x) +  A(x) dx = 0 ---(2)
from continuum
du
  E  E  (3) mechanics,  = du / dx
dx

Where  - stress,  - strain & E - Young’s Modulus

 (3) in (2 ) & dividing by dx.

d du
Governing EA( x)   A( x)  0
Equation dx dx

2
d u
For a bar of constant cross-section EA 2   A  0
dx
 d du
EA( x )   A( x )  0
dx dx
Boundary conditions

1.

2.
Variables:
Primary
eg. Displacement, u
Temperature, T
Secondary
eg. Force -
Heat flux –

Moment –
BOUNDARY CONDITIONS:
 Essential/ Geometric/ Dirichlet
Boundary conditions
Natural/ Force/ Neumann
Boundary conditions
 BOUNDARY CONDITIONS
CAN BE OF TWO KINDS:
 HOMOGENEOUS BOUNDARY
CONDITION

 NON HOMOGENEOUS BOUNDARY

CONDITION

T(0) = T
Loads:
Volume loads N/m3 ,N/m
eg. Self weight, udl
Point loads N
Problems that could be solved
by the FEM

1. Boundary Value Problems

2. Initial Value Problems
3. Eigen Value Problems
 Boundary Value Problem (BVP)
A boundary value problem is one where the
field variable (e.g., temperature or
displacement) and possibly its derivatives
are required to take on specified values on
the boundary
e.g.,

where K= Thermal conductivity,

A = area of cross-section,
Q = Heat flux
 d  dT(x) 
  KA(x)  + hp [T(x) - T ] = 0
dx  dx 
  
Boundary conditions: @ x = 0, T = T0
@ x = l, -KA (dt/dx) = 0
 Initial Value Problem (IVP)

An Initial value problem is one where

the field variable and possibly its
derivatives are specified initially (i.e., at
time t=0). These are generally time
dependent problems.
Examples include
Unsteady heat conduction
Dynamic problems
Initial conditions: @ time t = 0
i) du/dt = C0
where Velocity = du /dt
ii)
displacement u = a0
T = 300K T = 350K

T=500 K T=500 K

t = 0 Sec t=1 Sec

 Eigen Value Problem (EVP)

An eigen value problem is one where

the problem is defined by a
homogeneous differential equation that
is one where the right hand side is zero.
An important class of eigen value
problems is the ‘Vibration of Beams”
or continuous systems
Eigen Value Problems (EVP)
First mode shape

Second mode shape

Third mode shape

 DIMENSIONALITY

Physical problems can be classified into

(i) I dimensional
(ii) II dimensional
(iii) III dimensional problems.
Domain Geometry Boundary

1D Line Points

2D Area Curves

3D Volume Area
 I-D PROBLEMS
When the geometry, material properties and
field variables such as displacement,
temperature, pressure etc can be described in
terms of only one spatial co-ordinate we can
go in for one-dimensional modeling

T
 2D PROBLEMS
When the geometry and other
parameters are described in terms
of two independent co-ordinates we
go in for two-dimensional modeling

T
 3D PROBLEMS
If the geometry, material properties and
other parameters of the body can be
described by three independent spatial co-
ordinates, we can discretize the body
using 3 dimensional modeling
Exact and approximate solutions:
An exact solution satisfies the differential
equation at every point in the domain and
the boundary conditions on the boundary
An approximate solution satisfies the
boundary conditions completely and as
closely as possible the differential
equation
 d du
EA( x )   A( x )  0
dx dx

Boundary conditions
1.

2.
d du
EA( x )   A( x )  0
dx dx

R - RESIDUE
u - Approximate solution
uex – u - Error in solution
NUMERICAL SOLUTION OF BVPs
(i) Choose a trial solution u(x) for u(x)
(ii) Select a criterion for minimising the
error
u(x) can be a trigonometric function such
as A sinx
or a logarithmic function log x
or a hyperbolic function
or polynomial functions
2 3
u(x) = a0 + a1 x + a2 x + a3 x
 2 3
u(x) = a0 + a1 x + a 2 x + a3 x

u(x) =  ai x i

i= 0

u(x) = a0 0 (x) + a1 1 (x) + . . . + an n(x)


u(x) =  ai i

i= 0

1  x i
1. Methods of weighted residuals (WRM)
which are applicable when the governing
equations are differential equations.

2.Ritz variational method which is

applicable when the governing equations
are variational (integral) equations with an
associated quadratic functional.
The WRM criteria seek minimise the error
involved in not satisfying the governing
differential equations.
The most popular methods are

(i) The Collocation method.

(ii) The Sub-Domain method
(iii) The Least squares method.
(iv) The Galerkin method.
COLLOCATION METHOD
For each undetermined coefficient ai ,
choose a point xi in the domain and at each such
point called as collocation point force the residual
to be exactly zero
R(x1 ) = 0
R(x 2 ) = 0
..........
R(x n ) = 0
ie.

The collocation points may be located

anywhere on the boundary or in the domain.
THE SUB-DOMAIN METHOD
For each undetermined parameter
choose an interval x, in the domain.
Then force average of the residual in
each interval to be zero.
1
Δ x1 
Δ x1
R(x) dx = 0

1
 x2 
 x2
R ( x) dx = 0

1
 xn 
 xn
R ( x) dx = 0
LEAST SQUARES TECHNIQUE:
In this method we minimize with respect
to each undetermined coefficient the
integral of the square of the residue over
the entire domain
l

ai 0
R 2 (x) dx = 0

2
R

1
R(x)
 ai
dx = 0
THE GALERKIN METHOD
For each undetermined parameter
we require that a weighted average of R(x)
over the entire domain be zero. The
weighting functions are the trial functions
associated with the generalised coefficients
l


0
R(x) i (x) dx = 0
 GENERAL WRM


Ω
R(x) wi (x) dx = 0 i = 1, 2, . . . , n

(i) The Collocation method - dirac delta

function
(ii) The Sub-Domain method - Unity
(iii)The Least squares method - Residue
(iv) The Galerkin method – coefficient of the
undetermined coefficients in the trial
solution