Lecture 01-Weighted Residual Methods
Lecture 01-Weighted Residual Methods
Lecture 01-Weighted Residual Methods
OF STRUCTURAL SYSTEMS
Muhammad Fahim
hmf.engr2@gmail.com
Class Contents
• Introduction
• Least Squares Method
• Collocation Method
• Galerkin Method
• Modified Galerkin Method
UET Prospectus Description
• Relationship between the finite element method technology, and the
Rayleigh‐Ritz method.
• Derivation of element stiffness matrices.
• Construction of general stiffness matrices in global coordinates.
• Problems in plane stress, plates, and shells under static and dynamic
loads
Resources
• Chandrupatla, T. R., and Belegundu, A. D., Introduction to Finite
Elements in Engineering, 3rd Ed, Prentice Hall, 2002
• Reddy, J.N., An Introduction to the Finite Element Method, 3rd Edition,
McGraw‐Hill Higher Education, 2006.
• Cook, R. D., Malkus, D. S., Plesha, M.E., and Witt, R.J., Concepts and
Applications of Finite Element Analysis, 4th Edition, Wiley, 2007
• Bathe, K.J., Finite Element Procedures, Prentice Hall, 1996.
• Class notes of Dr. Shahzad Rahman
• Class notes of Dr. B. N. Rao of IIT Madras
Introduction
• The finite element method is a numerical approach for solving
differential equations in an approximate manner.
Classical Approximate Methods
1. Weighted Residual Methods
The weighted residual methods work on a boundary value
problem (BVP) directly and try to minimize some measure of
error (residual) in the solution
2. Variational methods
The BVP is first expressed in an equivalent Variational form which
is used to construct the solution
Weighted Residual Methods :Example 1
• Consider the boundary value problem:
sin 0 1
• The boundary conditions (BCs) are: 0 0 1 0
• The exact solution of the problem is as follows:
1
sin
Weighted Residual Methods: Example 1
Trial Solution
• Assume that an approximate solution of the boundary value problem
has the following general form
..
Sub Topic: Essential Vs. Natural BCs
• For a general boundary value problem in which the highest derivative
present is of the order 2p the classification of boundary conditions is
as follows:
• those with the order from 0 to p‐1 are called essential
• those with the order from p to 2p‐1 are called natural
Weighted Residual Method: Example 1
Admissible Trial solution
• From the boundary conditions:
0 0→ 0
1 0→
• Thus the trial solution satisfying the given boundary conditions
becomes:
1 2 2
Weighted Residual Methods: Example 1
Residual
• The residual (error) for the given problem is defined as:
sin 0 1
sin
Weighted Residual Methods: Example 1
• Weighted residual:
• where W(x) is a suitable weighting function.
• One simple example of W is the residual itself i.e.
• Squaring E will avoid error cancellation.
• Different methods of weighted residuals differ in the way they define
the weighting functions.
Common Weighted Residual Methods
• Some of the commonly employed weighted residual methods are as
follows:
• Least Squares
• Collocation
• Galerkin
Least Squares Weighted Residuals: Example 1
1 2 2
• Substituting the trial solution into the differential equation, the
residual can be written as follows:
sin 2 sin
Least Squares Weighted Residuals: Example 1
• Total residual is given as:
• After integrating, R is a function of only unknown parameters.
• Therefore the necessary conditions for minimizing total residual are
as follow:
0 1,2 …
2 0 1,2 …
Least Squares Weighted Residuals: Example 1
2 0 1,2 …
• Thus the least squares weighted residual criteria can be written as:
0 1,2 …
• This statement can be interpreted as a special case of general
weighted residuals statement with weigh function as follows:
1,2, …
Least Squares Weighted Residuals: Example 1
• For the example BVP
2 sin
1
Least Squares Weighted Residuals: Example 1
• Therefore the approximate solution of the BVP becomes:
1
• The exact solution is given as:
1
sin
Least Squares Weighted Residuals: Example 1
Collocation Method: Example 1
0
• where are the Collocation points.
• Thus in the Collocation method, the weighting function is the dirac
delta function.
1,2, …
Sub Topic: Dirac delta function
• Dirac delta function has the following
properties:
0,
1
,
• The integral of dirac delta function over the
domain is 1.
Collocation Method: Example 1
2 sin
• Choosing 1/2 as a Collocation point gives:
1/2 1 2 0
1/2
• Thus the approximate solution is given as:
1
Collocation Method: Example 1
Galerkin Method: Example 1
• The weight functions in the Galerkin method are defined as follows
1,2, …
• The unknown parameters in the trial solution are computed using the
following criteria:
0 1,2, …
Galerkin Method: Example 1
2 sin
12
0
• The approximate solution becomes:
12
Galerkin Method: Example 1
Modified Galerkin Method
• The basic Galerkin criteria is given as:
0 1,2, …
• The computation of requires differentiating trial solution as
many times as the order of the highest derivative in the BVP.
• Thus a second order BVP would require differentiating E twice.
• Therefore if the basic Galerkin criteria is used directly, we must use at
least a quadratic polynomial as a trial solution for a second order BVP.
Modified Galerkin Method
• However, the order of derivatives required in the trial solution can be
reduced by using integration by parts.
• This makes it possible to use even linear trial solutions, if desired.
• This is important in finite elements since a large class of elements are
based on linear assumed solutions.
Modified Galerkin Method: Example 2
• Consider the boundary value problem:
0 0 1
• The boundary conditions are:
0 0
1 1
Modified Galerkin Method: Example 2
• The basic weighted residual statement becomes:
• Integrate the first term in the integral by parts:
0
Sub Topic: Integration by parts
• Integration by parts for a definite integral of two functions is given as:
Weighted Residual Methods: Weight Functions
• Least Squares Weighted Residual Method
1,2, …
• Collocation Weighted Residual Method
1,2, …
• Galerkin Weighted Residual Method
1,2, …
Weighted Residual Methods: Example 1
1
sin 0 1 sin
• Least Squares Weighted Residual Method
1
sin
• Collocation Weighted Residual Method
1
• Galerkin Weighted Residual Method
12