Finite Element Methods

[DE ZG513]

BITS Pilani
Pilani|Dubai|Goa|Hyderabad Lecture 1 – Course Handout + Introduction lecture
Credentials
PhD, IIT Madras
(Experimental mechanics)

10+ Years Research

Course
Experience

Mechanics of materials,
Digital Image Correlation,
Computer vision, Deep
learning, Photogrammetry,
Material testing, Crystal

Handout
deformation studies,
Analysis of Auxetic
structures and materials

Funding received from

BITS, BIRAC, PIEDS
Elearn link
Published 15+ papers in
the Int Jls, Int. conferences

Granted 1 US patent and Prof. Iniyan Thiruselvam

1 Indian patent (Design)
Government funded
PhD (IIT Madras)
projects on material Assistant Professor, BITS Pilani
testing.
https://web.bits-pilani.ac.in/goa/iniyann/profile 2
Today….
❑ What is FEM?
❑ Applications
❑ Advantages
❑ Brief History
❑ About this course
❑ Basics of approximate solution

3
What is FEM?

4
5
6
 What is FEM?
FEM is a numerical method used
to obtain an approximate solution
to differential equations

7
To obtain an approximate solution, we assume a function called trial
function to be solution of the governing differential equation

We will denote the trial function as 𝑢ො and the true solution as 𝑢

True solution
to GDE, 𝑢
GDE of the
given problem
along with BC

Approximate solution
to GDE, 𝑢ො

8
Recall: Solution of the algebraic equation 𝑥 2 + 2𝑥 + 1 = 0

Solution is 𝑥 = −1

The function here is 𝑦 = 𝑥 2 + 2𝑥 + 1

Minimize this function, we get solution

𝑑𝑦
=0
𝑑𝑥
2𝑥 + 2 = 0 For the algebraic equation, the solution is a value returned
by the function
x = −1
For a differential equation, the solution is a function

9
10
11
Approximate method 1

12
Approximate method 2

13
Approximate method 3

14
Approximate method 4

THESE METHODS PERFORM FUNCTION APPROXIMATION

15
Discretization – an example

Let us compute perimeter of a Circle of unit diameter through discretization

Continuum object | Discrete approximation by inscribed regular polygons | Disconnected element,

16
What is FEM?
Computing Perimeter of a Circle of unit Diameter

17
Possible Sources of Error in FEM

❑ Domain Approximation Error

❑ Element Interpolation/Approximation Errors

❑ Numerical Integration Errors

❑ Computer Errors (Round-Off, Etc., )

18
Where does FEM fit in the Product Cycle ?
Design and Manufacturing

Computer- Computer-
Aided automated drafting
Design and documentation
FEM

product design
drafting
concept engineering

customers order new Computer-

process
and equipment aided process
planning
markets and tooling planning

quality production
production
control scheduling

Computer- Computer-controlled Computer scheduling,

aided quality robots, material requirements planning,
control machines, etc. shop floor control

19
FEM ? or FEA ?

Finite Element Method

OR
Finite Element Analysis

“A numerical method to obtain obtaining approximate solutions to

differential equations”

“A numerical analysis technique for obtaining approximate

solutions to……engineering problems”
20
21
Applications of FEM

22
Aerospace Engineering Automobile Engineering

Structural Engineering Biomedical Engineering

23
What are the advantages?

24
Can readily handle complex geometry:
• The heart and power of the FEM.
Can handle complex analysis types:
• Vibration
• Transients
• Nonlinear
• Heat transfer
• Coupled (Thermal + Structural)
• Fluids
• Wear

25
Can handle complex loading:
• Node-based loading (point loads).
• Element-based loading (pressure, thermal, inertial forces).
• Time or frequency dependent loading.

Can handle complex restraints:

• Indeterminate structures can be analyzed.

26
Can handle bodies comprised of non-homogeneous materials:
• Every element in the model could be assigned a different set
of material properties.

Can handle bodies comprised of non-isotropic materials:

• Orthotropic
• Anisotropic

27
Special material effects are handled:
• Temperature dependent properties.
• Plasticity
• Creep
• Swelling
• Freezing / Melting

Special geometric effects can be modeled:

• Large displacements
• Large rotations
• Contact (gap) condition

28
Brief History

29
FEM | Brief History
❑ Hrenikoff, 1941 – “framework method”
❑ Courant, 1943: “piecewise polynomial interpolation”
❑ Turner, 1956: Derived stiffness matrices for truss, beam, etc. (Stiffness and
Deflection Analysis of Complex Structures)
❑ Clough, 1960: Coined the term “finite element” (The Finite Element Method in
Plane Stress Analysis)
❑ Zienkiewicz and Cheung, 1965: “Finite Element in the Solution of Field
Problems,”
❑ Zienkiewicz, 1968: “The Isoparametric Finite Element System-A New Concept in
Finite Element Analysis

30
FEM Pioneers

• O. C. Zienkiewicz
• R. L. Taylor
• K. J. Bathe
• J. H. Argyris
• E. Wilson
• J. T. Oden
• T. J. R. Hughes
• R. H. Gallagher
31
There are many software available
Ansys

Fluent But how do they work?

What are the procedures to solve them?
Abacus Any mathematical framework?

Matlab
We learn those techniques in this course
DUNE

OpenFoam
32
About this course

33
Credit Model 1 – 1 – 2

❑ 32 Hours of Class-room Instruction + 32

Hours of Case-studies/Tutorials + 64
Hours of Student Preparation

❑ Lab class recording + assignment + exam

34
Evaluation Scheme

35
Approximation methods

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

What is FEM or FEA?

“A numerical method to obtain obtaining

approximate solutions to differential
equations”

“A numerical analysis technique for

obtaining approximate solutions
to……engineering problems”

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Example problem

1D axial deformation of a bar

Length 𝐿
Cross-sectional area 𝐴
Body load 𝑞(𝑥) Young’s modulus 𝐸
Point load 𝑃 Unknown displacement 𝑢(𝑥)
Equilibrium equation
𝑞 𝑥 . ⅆ𝑥
ⅆ𝐹
ⅆ𝐹 𝐹 = 𝑞 𝑥 ⅆ𝑥 + 𝐹 + ⅆ𝑥
𝐹 𝐹+ ⅆ𝑥 ⅆ𝑥
ⅆ𝑥 ⅆ𝐹
𝑞 𝑥 + =0
ⅆ𝑥 ⅆ𝑥
Differential element ⅆ𝑢 𝑥 𝐹 = σ 𝑥 .𝐴
ⅆ 𝐴𝐸.
ⅆ𝑥
𝑞 𝑥 + =0
ⅆ𝑥 𝐹 = 𝐸. ε 𝑥 . 𝐴
ⅆ2𝑢 ⅆ𝑢 𝑥
We must solve this differential equation 𝐴𝐸 2 + 𝑞 𝑥 = 0 𝐹 = 𝐸. .𝐴
ⅆ𝑥 ⅆ𝑥
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Boundary conditions

1D axial deformation of a bar

Length 𝐿
Cross-sectional area 𝐴
Body load 𝑞(𝑥) Young’s modulus 𝐸
Point load 𝑃 Unknown displacement 𝑢(𝑥)

Governing differential equation (GDE)

ⅆ2 𝑢
𝐴𝐸 2 + 𝑞 𝑥 = 0
ⅆ𝑥
Dirichlet / Necessary /
Boundary condition (BC) 𝑢 𝑥 ቚ = 0 Homogeneous displacement BC
𝑥=0

ⅆ𝑢
𝐴𝐸 ቤ = 𝑃 Neumann / Natural / Force BC
ⅆ𝑥 𝑥=𝐿
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Analogy of solution to an
algebraic equation
Recall: Solution of the algebraic equation 𝑥 2 + 2𝑥 + 1 = 0

Solution is 𝑥 = −1

The function here is 𝑦 = 𝑥 2 + 2𝑥 + 1

Minimize this function, we get solution

ⅆ𝑦
=0
ⅆ𝑥 For the algebraic equation, the solution
2𝑥 + 2 = 0 is a value returned by the function
For a differential equation, the solution
x = −1 is a function
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Analytical solution to GDE

𝑑2 𝑢 𝑑𝑢
GDE: 𝐴𝐸 2 +𝑞 𝑥 =0 BC: 𝑢 𝑥 ȁ𝑥=0 = 0 𝐴𝐸 ቚ =𝑃
𝑑𝑥 𝑑𝑥 𝑥=𝐿

ⅆ2 𝑢
Integrating GDE, we get, න 𝐴𝐸 ⅆ𝑥 2 + 𝑞 𝑥 ⅆ𝑥 = 0

ⅆ𝑢
𝐴𝐸 + න 𝑞 𝑥 ⅆ𝑥 = 𝐶1
ⅆ𝑥

Integrating again, we get, 𝐴𝐸. 𝑢 𝑥 + ඵ 𝑞 𝑥 ⅆ𝑥 = 𝐶1 𝑥 + 𝐶2

1
𝑢 𝑥 = 𝐶 𝑥 + 𝐶2 − ඵ 𝑞 𝑥 ⅆ𝑥
𝐴𝐸 1
Apply BC to solve for 𝐶1 and 𝐶2
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Is analytical solution always
possible?

𝑑2 𝑢 𝑑𝑢
GDE: 𝐴𝐸 2 +𝑞 𝑥 =0 BC: 𝑢 𝑥 ȁ𝑥=0 = 0 𝐴𝐸 ቚ =𝑃
𝑑𝑥 𝑑𝑥 𝑥=𝐿

Analytical solution of this GDE is the function 𝑢 𝑥

1
𝑢 𝑥 = 𝐶1 𝑥 + 𝐶2 − ඵ 𝑞 𝑥 ⅆ𝑥
𝐴𝐸

Analytical solution is NOT always possible for real world

problems in engineering
Therefore, we need approximation methods to find the
solution for the given GDE
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Approximation methods

Here, we assume a function called trial function to be solution

of the governing differential equation

We will denote the trial function as 𝑢ො and the true solution as 𝑢

True solution
to GDE, 𝑢
GDE of the
given problem
along with BC

Approximate solution
to GDE, 𝑢ො

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Trial function must obey a set
of rules
A trial function may be chosen to be a polynomial function, for
example, 𝑢ො = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2 … or trigonometric function, e.g.,
𝑢ො = 𝑐0 + 𝑐1 sin 𝜋𝑥…

In general, 𝑢ො = 𝑐0 ϕ0 + 𝑐1 ϕ1 + 𝑐2 ϕ2 + 𝑐3 ϕ3 + 𝑐4 ϕ4 +…+𝑐𝑛 ϕ𝑛

1. ϕ1 …ϕ𝑛 are known/assumed functions of 𝑥

2. ϕ𝑗 𝑥 must be complete polynomials/functions
3. polynomial order of ϕ𝑛 𝑥 ≥ order of GDE (for 𝑢ො to be
sufficiently differentiable)
4. ϕ0 𝑥 … ϕ𝑛 𝑥 must be independent polynomials/functions
5. 𝑢ො must satisfy both natural and essential BCs
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Approximation using trial functions

Let us consider a GDE defined in the domain Ω while satisfying

the BC on the boundary Γ

Assume 𝑢ො = 𝑐0 ϕ0 + 𝑐1 ϕ1 + 𝑐2 ϕ2 + 𝑐3 ϕ3 + 𝑐4 ϕ4 +…+𝑐𝑛 ϕ𝑛

By substituting 𝑢ො in GDE and BC, find the error/residual

Error in satisfying GDE is called domain residual

Error in satisfying BC is called boundary residual

Minimize both of these residuals, if possible, make them zero

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Example 1

𝑑2 𝑢 𝑑𝑢
GDE: 𝐴𝐸 2 +𝑞 𝑥 =0 BC: 𝑢 𝑥 ȁ𝑥=0 = 0 𝐴𝐸 ቚ =𝑃=0
𝑑𝑥 𝑑𝑥 𝑥=𝐿

For the given problem, body load 𝑞 𝑥 is a constant load per unit length 𝑞0

𝑑2 𝑢 𝑑𝑢
GDE: 𝐴𝐸 2 + 𝑞0 = 0 BC: 𝑢 𝑥 ȁ𝑥=0 = 0 𝐴𝐸 ቚ =𝑃=0
𝑑𝑥 𝑑𝑥 𝑥=𝐿

Assume the trial function 𝑢ො = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2

To satisfy first BC, we have, 𝑐0 = 0 𝑢ො must obey the
rules defined earlier
To satisfy second BC, we have, 𝑐1 = −2𝑐2 𝐿
Substituting back in the trial function, we get, 𝑢ො = 𝑐2 (𝑥 2 − 2𝐿𝑥)
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Find domain residual

𝑑2 𝑢 𝑑𝑢
GDE: 𝐴𝐸 2 + 𝑞0 = 0 BC: 𝑢 𝑥 ȁ𝑥=0 = 0 𝐴𝐸 ቚ =𝑃=0
𝑑𝑥 𝑑𝑥 𝑥=𝐿

Trial function, 𝑢ො = 𝑐2 (𝑥 2 − 2𝐿𝑥)

0
𝑑2 𝑢
ෝ 𝑑2 𝑢
Domain residual, 𝑅𝑑 = 𝐴𝐸 2 + 𝑞0 − 𝐴𝐸 2 + 𝑞0
𝑑𝑥 𝑑𝑥

𝑅𝑑 = 𝐴𝐸 2𝑐2 + 𝑞0

What about boundary residual?

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Minimize the residual
Trial function, 𝑢ො = 𝑐2 (𝑥 2 − 2𝐿𝑥)
𝑅𝑑 = 𝐴𝐸 2𝑐2 + 𝑞0

𝑑2 𝑢 𝑑𝑢
GDE: 𝐴𝐸 2 + 𝑞0 = 0 BC: 𝑢 𝑥 ȁ𝑥=0 = 0 𝐴𝐸 ቚ =𝑃=0
𝑑𝑥 𝑑𝑥 𝑥=𝐿

Since there is one parameter to compute and one equation,

one can readily estimate the unknown coefficient 𝑐2 by
setting the residue to zero
𝑞0
𝑅𝑑 = 𝐴𝐸 2𝑐2 + 𝑞0 = 0 𝑐2 = −
2𝐴𝐸
𝑞0
Substitute 𝑐2 in trial function, 𝑢(𝑥)
ො = (2𝐿𝑥 − 𝑥2)
2𝐴𝐸
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Check the solution
𝑞0
Final solution, 𝑢(𝑥)
ො = (2𝐿𝑥 − 𝑥 2)
2𝐴𝐸

𝑑2 𝑢 𝑑𝑢
GDE: 𝐴𝐸 2 + 𝑞0 = 0 BC: 𝑢 𝑥 ȁ𝑥=0 = 0 𝐴𝐸 ቚ =𝑃=0
𝑑𝑥 𝑑𝑥 𝑥=𝐿

ⅆ 2 𝑢ො −𝑞0
=
ⅆ𝑥 2 𝐴𝐸

𝑑2𝑢
ෝ −𝑞0
Therefore, 𝐴𝐸 2 + 𝑞0 = 𝐴𝐸. + 𝑞0 = 0
𝑑𝑥 𝐴𝐸

For this simple example, since we could make the residual

identically zero everywhere in the domain, approximate
solution tallies with the exact solution
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Example 2
Fully-developed steady laminar flow
𝑑2𝑣
GDE: μ 2 + 𝜌𝑔 cos 𝜃 = 0
𝑑𝑥

𝑑𝑣(𝑥)
BC: ቚ = 0 (zero shear stress) 𝑣(𝑥)ȁ𝑥=𝐿 = 0 (no slip)
𝑑𝑥 𝑥=0

Trial function, 𝑣ො = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2 𝑣ො = 𝑐2 (𝑥 2 − 𝐿2 )
Applying BC, we get, 𝑐1 = 0 and 𝑐0 = −𝑐2 𝐿2
Domain residue, 𝑅𝑑 = 2μ𝑐2 + 𝜌𝑔 cos 𝜃
−𝜌𝑔 cos 𝜃
Setting domain residue to zero, 𝑐2 =
2μ
−𝜌𝑔 cos 𝜃 2
𝑣ො = (𝑥 − 𝐿2 )
2μ
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Example 3
Cantilever beam under udl
𝑑4𝑤
GDE: 𝐸𝐼 4 − 𝑞0 = 0
𝑑𝑥

𝑑𝑤 𝑑2 𝑤 𝑑3 𝑤
BC: 𝑤ȁ𝑥=0 = 0, ቚ = 0, ቚ = 0, ቚ =0
𝑑𝑥 𝑥=0 𝑑𝑥 2 𝑥=𝐿 𝑑𝑥 3 𝑥=𝐿

ෝ = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2 + 𝑐3 𝑥 3 + 𝑐4 𝑥 4
Trial function, 𝑤
Apply BC ෝ ȁ𝑥=0 = 0 implies 𝑐0 = 0
𝑤
ෝ
𝑑𝑤
and ቚ =0
𝑑𝑥 𝑥=0
→ 𝑐1 + 2𝑐2 𝑥 + 3𝑐3 𝑥 2 + 4𝑐4 𝑥 3 ȁ𝑥=0 = 0
→ 𝑐1 + 2𝑐2 0 + 3𝑐3 0 + 4𝑐4 (0) = 0
→ 𝑐1 = 0
Trial function becomes 𝑤 ෝ = 𝑐2 𝑥 2 + 𝑐3 𝑥 3 + 𝑐4 𝑥 4
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Example 3
ෝ = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2 + 𝑐3 𝑥 3 + 𝑐4 𝑥 4
Trial function, 𝑤

𝑑2 𝑤
ෝ 𝑑3 𝑤
ෝ
ቚ =0 ቚ =0
𝑑𝑥 2 𝑥=𝐿 𝑑𝑥 3 𝑥=𝐿
→ 2𝑐2 + 6𝑐3 𝑥 + 12𝑐4 𝑥 2 ȁ𝑥=𝐿 = 0 → 6𝑐3 + 24𝑐4 𝑥ȁ𝑥=𝐿 = 0
→ 2𝑐2 + 6𝑐3 𝐿 + 12𝑐4 𝐿2 = 0 → 6𝑐3 + 24𝑐4 𝐿 = 0
→ 𝑐2 = −3𝑐3 𝐿 − 6𝑐4 𝐿2 → 𝑐3 = −4𝑐4 𝐿
ෝ = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2 + 𝑐3 𝑥 3 + 𝑐4 𝑥 4
Trial function, 𝑤
ෝ = 0 + 0 𝑥 + −3𝑐3 𝐿 − 6𝑐4 𝐿2 𝑥 2 −4𝑐4 𝐿 𝑥 3 +𝑐4 𝑥 4
𝑤
ෝ = −3𝑐3 𝐿𝑥 2 − 6𝑐4 𝐿2 𝑥 2 −4𝑐4 𝐿 𝑥 3 +𝑐4 𝑥 4
𝑤
ෝ = −3(−4𝑐4 𝐿)𝐿𝑥 2 + 𝑐4 (𝑥 4 − 6𝑥 2 𝐿2 − 4𝐿𝑥 3 )
𝑤
ෝ = 𝑐4 (𝑥 4 + 12𝑥 2 𝐿2 − 6𝑥 2 𝐿2 − 4𝐿𝑥 3 )
𝑤
ෝ = 𝑐4 (𝑥 4 + 6𝑥 2 𝐿2 − 4𝐿𝑥 3 )
𝑤
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Example 3
Cantilever beam under udl
𝑑4𝑤
GDE: 𝐸𝐼 4 − 𝑞0 = 0
𝑑𝑥

𝑑𝑤 𝑑2 𝑤 𝑑3 𝑤
BC: 𝑤ȁ𝑥=0 = 0, ቚ = 0, ቚ = 0, ቚ =0
𝑑𝑥 𝑥=0 𝑑𝑥 2 𝑥=𝐿 𝑑𝑥 3 𝑥=𝐿

Trial function, 𝑤(𝑥)

ෝ = 𝑐4 (𝑥 4 + 6𝑥 2 𝐿2 − 4𝐿𝑥 3 )
𝑑4 𝑤
ෝ
𝑞0
Domain residue, 𝑅𝑑 = 𝐸𝐼 4 − 𝑞0 = 𝐸𝐼 24𝑐4 − 𝑞0 = 0 → 𝑐4 =
𝑑𝑥 24𝐸𝐼

𝑞0
𝑤(𝑥)
ෝ = (𝑥 4 + 6𝐿2 𝑥 2 − 4𝐿𝑥 3 )
24𝐸𝐼
It can be seen that this solution tallies with the exact solution
because the domain residue is zero over the entire domain
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Summary
In Approximation method,
• We assume a trial function as solution, and it has some
unknown coefficients
• Trial function must follow certain rules
• Select it such that the boundary residual is zero already
• Substitute it in the governing differential equation
• Finally, minimize the domain residual or make it to zero (if
possible) to obtain unknown parameters in the trial function
• If the domain residual reduces to a one parameter solution,
find the unknown by setting the domain residual to zero
• In this case, domain residual is zero at all points in the domain
What if there is more than one parameter in the domain residual
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Exemplar problem
𝑑 𝑑𝑢
GDE: − 𝑥 +𝑢 =0 for 0<𝑥<1
𝑑𝑥 𝑑𝑥
𝑑𝑢
BC: 𝑢 0 =1 𝑥 ቚ =0
𝑑𝑥 𝑥=1

Assume trial function

𝑢ො = ϕ0 (𝑥) + 𝑐1 ϕ1 (𝑥) + 𝑐2 ϕ2 (𝑥) + 𝑐3 ϕ3 (𝑥) + 𝑐4 ϕ4 (𝑥)+…+𝑐𝑛 ϕ𝑛 𝑥

For this problem, 𝑢ො = ϕ0 𝑥 + 𝑐1 ϕ1 𝑥 + 𝑐2 ϕ2 𝑥

ASSUME ϕ0 𝑥 , ϕ1 𝑥 and ϕ2 𝑥 such that 𝑢ො satisfies BC
ϕ0 𝑥 = 1 ϕ1 𝑥 = 𝑥 2 − 2𝑥 and ϕ2 𝑥 = 𝑥 3 − 3𝑥
ෝ
𝑑𝑢
𝑢ො 0 = 1 and 𝑥 = 𝑥 𝑐1 2𝑥 − 2 + 𝑐2 (3𝑥 2 −3) = 0 when 𝑥 = 1
𝑑𝑥
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Exemplar problem

Trial function 𝑢ො = 1 + 𝑐1 𝑥 2 − 2𝑥 + 𝑐2 𝑥 3 − 3𝑥
Substitute u in the GDE with 𝑢ො
𝑑 ෝ
𝑑𝑢
− 𝑥 + 𝑢ො =
𝑑𝑥 𝑑𝑥
ⅆ
− 𝑐1 2𝑥 2 − 2𝑥 + 𝑐2 (3𝑥 3 −3𝑥) + [1 + 𝑐1 𝑥 2 − 2𝑥 + 𝑐2 𝑥 3 − 3𝑥 ]
ⅆ𝑥

= − 𝑐1 4𝑥 − 2 + 𝑐2 (9𝑥 2 −3) + 1 + 𝑐1 𝑥 2 − 2𝑥 + 𝑐2 𝑥 3 − 3𝑥
= 𝑐2 𝑥 3 + (𝑐1 − 9𝑐2 )𝑥 2 − 6𝑐1 + 3𝑐2 𝑥 + (1 + 2𝑐1 + 3𝑐2 )

Can we make this domain residual zero at all points in the domain?

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Exemplar problem

ⅆ ⅆ𝑢ො
− 𝑥 + 𝑢ො
ⅆ𝑥 ⅆ𝑥
= 𝑐2 𝑥 3 + (𝑐1 − 9𝑐2 )𝑥 2 − 6𝑐1 + 3𝑐2 𝑥 + (1 + 2𝑐1 + 3𝑐2 )

To make this zero at all points in the zero, all the four terms
in the above domain residual must be zero
𝑐2 = 0
𝑐1 − 9𝑐2 = 0 A viable alternative is
6𝑐1 + 3𝑐2 = 0 weighted residual method
1 + 2𝑐1 + 3𝑐2 = 0
The above relations are inconsistent; hence, there is no
solution to the equations
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 BITS Pilani
Pilani | Dubai | Goa | Hyderabad

Thank You