Scribd
Upload
2 views

Chapter Two

Chapter Two of the document discusses the formulation of the Finite Element Method (FEM) using various approaches including Direct, Weighted Residual Method (WRM), and Energy Approach. It outlines the steps involved in each method and introduces techniques such as Point Collocation, Sub-domain, Least Square, and Galerkin's Method for minimizing errors in approximate solutions. The chapter concludes with examples demonstrating the application of these methods to solve differential equations.

Uploaded by

Endalamaw Ejigu
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
2 views

Chapter Two

Chapter Two of the document discusses the formulation of the Finite Element Method (FEM) using various approaches including Direct, Weighted Residual Method (WRM), and Energy Approach. It outlines the steps involved in each method and introduces techniques such as Point Collocation, Sub-domain, Least Square, and Galerkin's Method for minimizing errors in approximate solutions. The chapter concludes with examples demonstrating the application of these methods to solve differential equations.

Uploaded by

Endalamaw Ejigu
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 17

Addis Ababa University

Addis Ababa Institute of Technology


School of Mechanical and Industrial Engineering

Introduction to
Finite Element Method

Chapter Two
FEM Formulation

Bililign Amare
FEM Formulation
 The governing equations for problems solved by the FEM are
typically formulated by partial differential equations in their
FEM Formulation
original form.
Direct Approach WRM Approach Energy Approach Variational Approach

 Applies physical concept directly  Method for finding an


 Variational Formulation is
to discretized element. approximate solution to the DE’s.
Differential Formulation.
 It does not need sophisticated  Its objective is to reduce the error.
 Relies on the calculus of
mathematical concepts.  The problem will represented by
variations and involves
 Its applicability is limited to DE within a domain and
extrimizing a function.
certain problems. boundary equation.

WRM == Weight Residual Method


1. Direct Approach
 Analysis is based up on stiffness method(spring , pipe).
 No need of interpolation function because exact method (Hook’s
Law).
 The spring system having four nodes and three elements (spring) is
shown in the next figure as example.

Direct Method Solving Steps


1. Define an element, determine their properties
2. Use direct physical relationship to establish element equations in terms of
intended variables.
3. Combine element equations for various elements as a system of equations.
4. Engineering problems such as spring system, truss, beam, fluid flow in pipe
network, ….
2. WRM Formulation Steps

1. Start with the Governing Differential Equation.

2. Assume an approximate solution equation to the problem.

Note: The assumed solution equation must satisfy the

initial and boundary condition.

3. By Substituting the assumed approximate solution and

BC’s in the DE’s find the domain residual (Rd).

4. Force the residual to zero at various points or intervals.

5. Determine the unknown coefficients.


Weighted Residual Techniques to Minimize Error
 When the domain residual is varying from point to point WR is
not more applicable so there must be another method to
minimize these residual variations and residual error’s .
 The most Popular weighted Residual Methods to get improved
approximate solutions are:
 Point Collocation Method
 Sub-domain Method
 Least Square Method
 Galerkin’s Method (common in using FEM analysis).
2.1. Point Collocation Method

 The residual function is set to zero at chosen discrete points in the domain.

 The DE is satisfied only at the chosen points.

 Have no control over the residuals at all other points in the domain.

 The number of points being equal to the number of coefficients in the approximate

function that need to be determined.


2. 2. Sub-domain Method
 The integral of the residual equation is forced to be zero over subintervals.

 The number of subintervals are bring equal to the number of coefficient's in the

approximate function.
2.3. Least Square Method
 The integral of the square of residual over the domain is required to be
minimum.

 The requirement is as follow’s,


2.4. Galerkin’s Method
 Was invented by Russian Mathematician Boris Grigoryevich Galerkin.
 Includes to minimize the residual in an overall sense through out the domain.
L

w ( x) R ( x)dx 0
0
i d

 Where wi (x) is weighting function and Rd(x) is residual.


 How to select the weighting function??
 Approximating or trial functions are considered to be the weighting functions.

 Consider a system with homogeneous boundary conditions

L( ) g ....in..
2.4. Galerkin’s Method cont..
n

 Assume an approximate function as


   N i i
i 1

Ni i
 Where is assumed function, and are either the unknown parameters or
unknown function of one of the independent variables.
R L ( )  g 0
 Residual is calculated as d

 Galerkin Method

w R d 0....where..w N

i d i i

 This method yields n linear equation to determine n values of the unknown


parameters.
Example
Example 1
Solve the following differential equation using Point collocation,
least square Method and Galerkin Method and then compare your
results from the exact method.
The given DE is and use the boundary condition’s of
and choose the collocation points as
x=0.25 and x=0.5??
Solution to Example 1

Case 1). Using point collocation method.


 Let approximate solution that satisfies the BC be the form of
  x(1  x)(a1  a2 x...)or
 a1 ( x  x 2 )  a2 ( x 2  x 3 ...)
d
a1  2 x(a2  a1 )  3a2 x 2
dx
d 2
2(a2  a1 )  6a2 x
dx 2

Substitution of d
2(a  a )and
2

 a1 ( x  x 2 )  a2 ( x 2  x3...) dx 2
 6a x
2 1 2

into yields
   ( x) a ( x  x  2)  a (6 x  x  x  2)  x
2 2 3

An error, , where
1 2

choosing x=0.25 and x=0.5 as collocation points, we


Solution to Example 1 cont. …
 ( x 0.25)  140a1  29a2  16 0
 ( x 0.5)  18a1  9a2  4 0
Solving the above equation simultaneously gives that a1=-
0.1459, a2=-0.15263
Hence the solutionis x( x  1)(0.1459  0.15263x)

Case 2). Using Least 2Square Method


  ( x) a1 ( x  x  2)  a2 (6 x  x 2  x3  2)  x
 2
[ a1 ( x 2  x  2)  a2 (6 x  x 2  x 3  2)  x]2( x 2  x  2)
a1
 2
And a [a1 ( x  x  2)  a2 (6 x  x  x  2)  x]2( x  6 x  x  2)
2 2 3 2 3

2
Solution to Example 1 cont. …

[
 1a ( x 2
 x  2 )  a 2
( 6 x  x 2
 x 3
 2 )  x ]2 ( x 2
 x  2) dx 0
0
L

[
 1a ( x 2
 x  2 )  a 2
( 6 x  x 2
 x 3
 2 )  x ]2 ( x 2
 6 x  x 3
 2) dx 0
0

Simplifying the above equation will give


4.67a1  2.35a2  1.084
 2.35a1  4.276a2 1.05
a1  0.1500, a2  0.16309
Hence the Solution is
 x( x  1)(0.15  0.16309 x
Solution to Example 1 cont. …
Case 3). Using Galarkin’s Method
 a1 ( x  x 2 )  a2 ( x 2  x 3 ...)

  ( x) a1 ( x  x  2)  a2 (6 x  x  x  2)  x
2 2 3

 Make the error orthogonal to the trial function


 Select weighting functions as, 2
N1 wi ( x  x )
N 2 wi ( x 2  x 3 )
 Then from the Galerkin equation we have the followings,
1 1

w R
i d dx 0,   ( x  x 2
)dx 0
0 0
And
1

 ( x  x )dx 0

2 3

0
Solution to Example 1 cont. …
Proper simplification of the above two equations will give the following equations,
 0.36667a1  0.18333a2 0.083333
 0.18333a1  0.14286a2 0.05
 Solving the above equations yields the value of the coefficients as,

a1  0.14588,
a2  0.16278
 x( x  1)(0.14588  0.16278 x).. Answer.

 Compare the results obtained by the point collocation, sub-domain and Galerkin
Method and select which method yields better result against the exact solution?
[Galerkin Method.. Answer]
Class Work
Solve the differential equation  ''   x 2with boundary conditions of
 ( x 0) 0, ( x 1) 0 using point collocation, Least Square and Galerkin
Method, finally compare your results for better approximate result against the exact
solution? and choose the collocation points as x=0.25 and x=0.5??

Assignment 1.
Assignment
A uniform rod is subjected to the axial load. The deformation of the bar is governed

by the differential equation of with a boundary condition of

u(x=0)=0 and du/dx (x=L)=0 Find an approximate solution of the above

differential equation by using Point collocation Method, Sub-domain Method and

Least Square Method? and choose the collocation points as x=0.25L and x=0.5L??
3. Variational approach

FEM steps in variational approach


1. Discretize the domain into elements
2. Insert the trial function into the functional for each element
3. Differentiate the functional with respect to each parameter and set the
resulting equation equal to zero.
4. Solve all the system of equations to determine unknown variables.

The End.

You might also like