Type 1: Using Ritz Approximation to find the solution of functional equation using

algebraic and trigonometric polynomial.

1. Set up the equations for 2 parameters Ritz approximation of the following equations
associated with a string under tension 𝑇 subjected to lateral load 𝑞 for which potential
energy is given by

𝐿 𝐿
𝑇 𝑑𝜙 2
Π = ∫ ( ) 𝑑𝑥 − ∫ 𝑞𝜙𝑑𝑥
2 𝑑𝑥
0 0

with boundary conditions 𝜙(0) = 𝜙(𝐿) = 0.

(i) Algebraic polynomials

(ii) Trigonometric polynomials

Solution:

Assuming two parameter solution: 𝜙 = 𝜙0 + 𝐶1 𝜙1 + 𝐶2 𝜙2

Trial solution: 𝜙̂ = 𝐶1 𝑥(𝐿 − 𝑥) + 𝐶2 𝑥 2 (𝐿 − 𝑥)

Substituting in equation:
𝐿2 𝐿
𝑇 𝑑𝜙̂
Π = ∫ ( ) 𝑑𝑥 − ∫ 𝑞𝜙̂𝑑𝑥
2 𝑑𝑥
0 0

𝐿 𝐿
2
𝑇 𝑑
= ∫ ( (𝐶1 𝑥(𝐿 − 𝑥) + 𝐶2 𝑥 (𝐿 − 𝑥))) 𝑑𝑥 − ∫ 𝑞(𝐶1 𝑥(𝐿 − 𝑥) + 𝐶2 𝑥 2 (𝐿 − 𝑥))𝑑𝑥
2
2 𝑑𝑥
0 0

𝐿
𝑇
= ∫ (𝐶1 𝐿 − 2𝐶1 𝑥 + 2𝐶2 𝐿𝑥 − 3𝐶2 𝑥 2 )2 − 𝑞(𝐶1 𝐿𝑥 − 𝐶1 𝑥 2 + 𝐶2 𝑥 2 𝐿 − 𝐶2 𝑥 3 ) 𝑑𝑥
2
0

Using differentiation under integral

𝐿
∂Π
= ∫ 𝑇(𝐶1 𝐿 − 2𝐶1 𝑥 + 2𝐶2 𝐿𝑥 − 3𝐶2 𝑥 2 )(𝐿 − 2𝑥) − 𝑞(𝐿𝑥 − 𝑥 2 ) 𝑑𝑥
𝜕𝐶1
0
 𝐿

= ∫ 𝑇(𝐶1 𝐿 − 2𝐶1 𝑥 + 2𝐶2 𝐿𝑥 − 3𝐶2 𝑥 2 )(𝐿 − 2𝑥) − 𝑞(𝐿𝑥 − 𝑥 2 )𝑑𝑥

0

𝐿 𝐿

∫ 𝑞(𝐿𝑥 − 𝑥 2 )𝑑𝑥 = ∫ 𝑇{𝐶1 (𝐿2 − 2𝑥𝐿 − 2𝑥𝐿 + 4𝑥 2 ) + 𝐶2 (2𝐿2 𝑥 − 4𝐿𝑥 2 − 3𝐿𝑥 2 + 6𝑥 3 )}𝑑𝑥
0 0

𝐿3 𝐿3 4𝐿3 4𝐿4 3𝐿4

𝑞( − ) = 𝑇 {𝐶1 (𝐿3 − 𝐿3 − 𝐿3 + ) + 𝐶2 (𝐿4 − − 𝐿4 + )}
2 3 3 3 2

𝑞𝐿3 𝐶1 𝐿3 4𝐿4 3𝐿4

= 𝑇{ + 𝐶2 (− + )}
6 3 3 2

𝑞𝐿3 = 𝑇 2𝐶1 𝐿3 + 𝑇𝐶2 𝐿4

𝑞
= 2𝐶1 + 𝐶2 ……………………………………………………………(i)
𝑇

Similarly;
𝐿
∂Π
= ∫ 𝑇(𝐶1 𝐿 − 2𝐶1 𝑥 + 2𝐶2 𝐿𝑥 − 3𝐶2 𝑥 2 )(2𝐿𝑥 − 3𝑥 2 ) − 𝑞(𝑥 2 𝐿 − 𝑥 3 ) 𝑑𝑥
𝜕𝐶2
0

= ∫ 𝑇(𝐶1 𝐿 − 2𝐶1 𝑥 + 2𝐶2 𝐿𝑥 − 3𝐶2 𝑥 2 )(2𝐿𝑥 − 3𝑥 2 ) − 𝑞(𝑥 2 𝐿 − 𝑥 3 ) 𝑑𝑥

0

∫ 𝑞(𝑥 2 𝐿 − 𝑥 3 )𝑑𝑥
0
𝐿

= ∫ 𝑇{𝐶1 (2𝐿2 𝑥 − 4𝑥 2 𝐿 − 3𝑥 2 𝐿 + 6𝑥 3 )
0
+ 𝐶2 (4𝐿2 𝑥 2 − 6𝐿𝑥 3 − 6𝐿𝑥 3 + 9𝑥 4 )}𝑑𝑥

𝐿4 𝐿4 4𝐿4 3𝐿4 4𝐿5 3𝐿5 3𝐿5 9𝐿5

𝑞( − ) = 𝑇 {𝐶1 (𝐿4 − − 𝐿4 + ) + 𝐶2 ( − − + )}
3 4 3 2 3 2 2 5

𝑞𝐿3 𝐶1 𝐿3 𝐶2 (40𝐿5 − 90𝐿5 + 54𝐿5 )

= 𝑇{ + }
12 6 30

Multiplying by 60,

5𝑞𝐿3 = 𝑇10 𝐶1 𝐿3 + 8𝑇𝐶2 𝐿4

5𝑞
= 10 𝐶1 + 8𝐶2 ………………………………………………………………….… eq 2
𝑇

Solving (1) and (2)

𝑞
𝐶1 =
2𝑇
𝐶2 = 0

Therefore;
𝑞
𝜙̂ = 𝑥(𝐿 − 𝑥)
2𝑇

ii. Using Trigonometric Polynomials

Solution:

Assuming two parameter solution: 𝜙 = 𝜙0 + 𝐶1 𝜙1 + 𝐶2 𝜙2

𝜋𝑥 3𝜋𝑥
Trial solution: 𝜙̂ = 𝐶1 sin + 𝐶2 sin
𝐿 𝐿

Substituting in equation:

2𝐿 𝐿
𝑇 𝑑𝜙̂
Π = ∫ ( ) 𝑑𝑥 − ∫ 𝑞𝜙̂𝑑𝑥
2 𝑑𝑥
0 0

𝐿 𝐿
𝑇 𝑑 𝜋𝑥 3𝜋𝑥 2 𝜋𝑥 3𝜋𝑥
= ∫ ( (𝐶1 sin + 𝐶2 sin )) 𝑑𝑥 − ∫ 𝑞 (𝐶1 sin + 𝐶2 sin ) 𝑑𝑥
2 𝑑𝑥 𝐿 𝐿 𝐿 𝐿
0 0

𝐿
𝑇 𝜋𝐶1 𝜋𝑥 3𝜋𝐶2 3𝜋𝑥 2 𝜋𝑥 3𝜋𝑥
=∫ ( cos + cos ) − 𝑞(𝐶1 sin + 𝐶2 sin ) 𝑑𝑥
2 𝐿 𝐿 𝐿 𝐿 𝐿 𝐿
0

Using differentiation under integral

𝐿
∂Π 𝜋𝐶1 𝜋𝑥 3𝜋𝐶2 3𝜋𝑥 𝜋 𝜋𝑥 𝜋𝑥
= ∫𝑇( cos + cos ) ( cos ) − 𝑞(sin ) 𝑑𝑥
𝜕𝐶1 𝐿 𝐿 𝐿 𝐿 𝐿 𝐿 𝐿
0
 𝐿
𝜋𝐶1 𝜋𝑥 3𝜋𝐶2 3𝜋𝑥 𝜋 𝜋𝑥 𝜋𝑥
= ∫𝑇( cos + cos ) ( cos ) − 𝑞(sin ) 𝑑𝑥
𝐿 𝐿 𝐿 𝐿 𝐿 𝐿 𝐿
0

𝐿 𝐿
𝜋𝑥 𝜋2 𝜋𝑥 3𝜋 2 𝜋𝑥 3𝜋𝑥
∫ 𝑞(sin ) 𝑑𝑥 = ∫ 𝑇 {𝐶1 ( 2 cos 2 ) + 𝐶2 ( 2 cos cos )} 𝑑𝑥
𝐿 𝐿 𝐿 𝐿 𝐿 𝐿
0 0

𝑞𝐿 𝜋2 1
− (cos 𝜋 − cos 0) = 𝑇 { 𝐶1 ( ) + 𝐶2 (0)}
𝜋 𝐿 2

𝑞𝐿 𝜋2
− (−1 − 1) = 𝑇 { 𝐶1 }
𝜋 2𝐿

2𝑞𝐿2
𝐶1 = ……………………………………………………………(i)
𝜋3 𝑇

Similarly;
𝐿
∂Π 𝜋𝐶1 𝜋𝑥 3𝜋𝐶2 3𝜋𝑥 3 𝜋 3𝜋𝑥 3𝜋𝑥
= ∫𝑇( cos + cos )( cos ) − 𝑞(sin ) 𝑑𝑥
𝜕𝐶2 𝐿 𝐿 𝐿 𝐿 𝐿 𝐿 𝐿
0

𝐿
𝜋𝐶1 𝜋𝑥 3𝜋𝐶2 3𝜋𝑥 3 𝜋 3𝜋𝑥 3𝜋𝑥
= ∫𝑇( cos + cos )( cos ) − 𝑞(sin ) 𝑑𝑥
𝐿 𝐿 𝐿 𝐿 𝐿 𝐿 𝐿
0

𝐿 𝐿
3𝜋𝑥 3𝜋 2 𝜋𝑥 3𝜋𝑥 9𝜋 2 3𝜋𝑥
∫ 𝑞 (sin ) 𝑑𝑥 = ∫ 𝑇 {𝐶1 ( 2 cos cos ) + 𝐶2 ( 2 cos 2 )} 𝑑𝑥
𝐿 𝐿 𝐿 𝐿 𝐿 𝐿
0 0

𝑞𝐿 9𝜋 2 𝐿
− (cos 3𝜋 − cos 0) = 𝑇 {𝐶1 (0) + 𝐶2 ( 2 ∗ )}
3𝜋 𝐿 2

𝑞𝐿 9𝜋 2
= 𝑇 {𝐶2 }
3𝜋 2𝐿

2𝑞𝐿2
𝐶2 = 27𝜋3 𝑇 ……………………………………………………………(i)

2𝑞𝐿2 𝜋𝑥 2𝑞𝐿2 3𝜋𝑥

Therefore: 𝜙 = sin + 27𝜋3 𝑇 sin
𝜋3 𝑇 𝐿 𝐿
 Take Away Questions
1.1 Determine the approximate two term solution for the deflection of a simply supported beam
subjected to a point load 𝑃 at center using Ritz method for which potential energy function is given
by

𝐿
1 2
Π = ∫ 𝐸𝐼(𝜙,𝑥𝑥 ) 𝑑𝑥 − 𝑃𝜙|𝑥=𝐿/2
2
0

1.2 Solve the Poisson equation governing heat conduction in a square region:

−𝑘∇2 𝑇 = 𝑔0

𝑇 = 0 on sides 𝑥 = 1 and 𝑦 = 1

𝜕𝑇
= 0 (insulated) 𝑥 = 0 and 𝑦 = 0
𝜕𝜂

using a one-parameter Ritz approximation of the form

𝑇1 = 𝑐1 (1 − 𝑥 2 )(1 − 𝑦 2 )
 Type 2: Using weak formulation to solve differential equation with subjected boundary
condition.

2. Find two parameter approximate solution of the equation using weak formulation

𝑑2𝜙 𝑑𝜙
+ 3 + 𝜙 = 4𝑥 0≤𝑥≤1
𝑑𝑥 2 𝑑𝑥

subjected to boundary conditions

𝜙(0) = 5, 𝜙′(1) = 2

Solution:

It says the weak formulation:

Using weighted integral statement with weight functions 𝑤1 and 𝑤2 .

1
𝑑2𝜙 𝑑𝜙 𝑑𝜙
∫ 𝑤1 ( 2 + 3 + 𝜙 − 4𝑥) 𝑑𝑥 + 𝑤2 ( |1 − 2)𝑑𝑥 = 0
𝑑𝑥 𝑑𝑥 𝑑𝑥
0

Using integration by parts

∫ 𝑢′ 𝑣𝑑𝑥 = 𝑢𝑣 − ∫ 𝑢𝑣 ′ 𝑑𝑥

1 1
𝑑2𝜙 𝑑𝜙 1 𝑑𝑤1 𝑑𝜙
∫ 𝑤1 ( 2 ) 𝑑𝑥 = 𝑤1 |0 − ∫ 𝑑𝑥
𝑑𝑥 𝑑𝑥 𝑑𝑥 𝑑𝑥
0 0

1 1
𝑑𝜙 𝑑𝑤1
∫ 3𝑤1 ( ) 𝑑𝑥 = 3𝑤1 𝜙|10 − 3 ∫ 𝜙 𝑑𝑥
𝑑𝑥 𝑑𝑥
0 0

Therefore:
𝑑𝜙 1 𝑑𝑤1 𝑑𝜙 1 𝑑𝑤1 1 1 𝑑𝜙
𝑤1 𝑑𝑥 |10 − ∫0 𝑑𝑥 + 3𝑤1 𝜙|10 − 3 ∫0 𝜙 𝑑𝑥 + ∫0 𝑤1 𝜙 𝑑𝑥 − ∫0 4𝑥𝑤1 𝑑𝑥 +𝑤2 𝑑𝑥 |1 −
𝑑𝑥 𝑑𝑥 𝑑𝑥
2𝑤2 |1 = 0

We chose 𝑤2 such that

𝑤2 = −𝑤1 at x=1
and 𝑤1 = 0 at x= 0 is 0
 𝑑𝜙 1 𝑑𝑤1 𝑑𝜙 1 𝑑𝑤1 1 1 𝑑𝜙
𝑤1 𝑑𝑥 |10 − ∫0 𝑑𝑥 + 3𝑤1 𝜙|10 − 3 ∫0 𝜙 𝑑𝑥 + ∫0 𝑤1 𝜙 𝑑𝑥 − ∫0 4𝑥𝑤1 𝑑𝑥 −𝑤1 𝑑𝑥 |1 −
𝑑𝑥 𝑑𝑥 𝑑𝑥
2𝑤2 |1 = 0

The last term becomes + 2𝑤1 |1

First term and second last term equates.
The second limit of first term is zero as 𝑤1 is 0 at x=0.
The second limit of third term also vanishes. So, we get;
1 𝑑𝑤1 𝑑𝜙 1 𝑑𝑤1 1 1
− ∫0 𝑑𝑥 + 3𝑤1 𝜙|1 − 3 ∫0 𝜙 𝑑𝑥 + ∫0 𝑤1 𝜙 𝑑𝑥 − ∫0 4𝑥𝑤1 𝑑𝑥 - 2𝑤1 |1 = 0
𝑑𝑥 𝑑𝑥 𝑑𝑥

After expressing the strong form into weak form;

𝜙 = 𝜙0 + 𝐶1 𝜙1 + 𝐶2 𝜙2

Only satisfy 𝜙0 as it is not incorporated.

others, 𝜙𝑖 = 𝑥𝑖2
Take trial function as 𝜙 = 5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 (for two parameter) and use Galerkin’s method
(𝑤1 = 𝜙𝑖 ) and 𝜙1 = 𝑥 𝑎𝑛𝑑 𝜙2 = 𝑥 2 )

For first equation, put 𝜙1 = 𝑥

1 𝑑𝑥 𝑑 1 𝑑𝑥
− ∫0 (5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 )𝑑𝑥 + 3𝑥(5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 )|1 − 3 ∫0 (5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 ) 𝑑𝑥 +
𝑑𝑥 𝑑𝑥 𝑑𝑥
1 1
∫0 𝑥(5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 ) 𝑑𝑥 − ∫0 4𝑥 2 𝑑𝑥 - 2𝑥|1 = 0

For second equation, put 𝜙2 = 𝑥 2

1 𝑑𝑥 2 𝑑 1 𝑑𝑥 2
− ∫0 (5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 )𝑑𝑥 + 3𝑥 2 (5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 )|1 − 3 ∫0 (5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 ) 𝑑𝑥 +
𝑑𝑥 𝑑𝑥 𝑑𝑥
1 1
∫0 𝑥 (5 + 𝐶1 𝑥 + 𝐶2 𝑥 2 ) 𝑑𝑥 − ∫0 4𝑥(𝑥 2 ) 𝑑𝑥 - 2𝑥 2 |1 = 0
2

Solve these to find 𝐶1 𝑎𝑛𝑑 𝐶2

Replace in 𝝓 = 𝟓 + 𝑪𝟏 𝒙 + 𝑪𝟐 𝒙𝟐
Take Away Question

2.1 Find two parameter approximate solution of the equation using weak formulation

𝑑2𝜙 𝑑𝜙
− 3 +𝜙 =𝑥 0≤𝑥≤1
𝑑𝑥 2 𝑑𝑥

subjected to boundary conditions

𝜙(0) = 0, 𝜙′(1) = 1

2.2 Find two parameter approximate solution of the equation using weak formulation

𝑑2𝜙 𝑑𝜙
2
+3 + 𝜙 = 4𝑥 0≤𝑥≤1
𝑑𝑥 𝑑𝑥

subjected to boundary conditions

𝜙(0) = 0, 𝜙(1) = 20

2.3 Find two parameter approximate solution of the equation using weak formulation

𝑑2𝜙 𝑑𝜙
+ 3 + 𝜙 = 4𝑥 0≤𝑥≤1
𝑑𝑥 2 𝑑𝑥

subjected to boundary conditions

𝜙′(0) = 2, 𝜙′(1) = 20

2.4 Construct weak form for the nonlinear differential equation and the associated boundary
conditions given by

𝒅 𝒅𝒖
− (𝒖 ) + 𝒇 = 𝟎 𝟎<𝒙<𝟏
𝒅𝒙 𝒅𝒙
𝒅𝒖
𝒖 𝒅𝒙 |𝒙=𝟎 = 𝟎 , u(1) = √𝟐
2.5 Construct weak form for the nonlinear differential equation and the associated boundary
conditions given by

Type 3: Solving two dimensional problems using Galerkin’s Method

Find the two parameter Galerkin solution of the equation.

𝜕 2𝜙 𝜕 2𝜙
+ = −2 in Ω Ω: −2 ≤ x ≤ 2, −1 ≤ y ≤ 1
𝜕𝑥 2 𝜕𝑦 2

𝜙=0 on Γ.

Formulation of trial function is most important in this problem.

First observe the boundary, the boundary suggests that the domain is rectangular domain. The x
limit is from -2 to 2 and y limit is from -1 to 1. On all the rectangular boundary, the value of
deflection is 0 i.e. 𝜙 = 0 on Γ.

Then observe the parameters. This problem requires two parameters.

𝜙 = 𝐶1 𝜙1 + 𝐶2 𝜙2

𝜋𝑥 𝜋𝑦
𝜙1 = cos cos
4 2

Why this?
because it satisfies boundary condition.
3𝜋𝑥 𝜋𝑦
𝜙2 = cos cos
4 2

𝜋𝑥 3𝜋𝑦
𝜙3 = cos 4
cos 2
 3𝜋𝑥 3𝜋𝑦
𝜙4 = cos cos
4 2

For two parameters;

𝜋𝑥 𝜋𝑦 3𝜋𝑥 𝜋𝑦
𝜙 = 𝐶1 cos cos + 𝐶2 cos cos
4 2 4 2
From weak formulation of 2D heat conduction equation for two parameters.

𝐶 𝐾 𝐾12 𝑞1
{ 1 } = ( 11 ){ }
𝐶2 𝐾21 𝐾22 𝑞2

𝜕 2 𝜙1 𝜕 2 𝜙1
𝐾11 = ∬( 2 + ) 𝑑𝑦 𝑑𝑥
𝜕𝑥 𝜕𝑦 2
𝜕 2 𝜙2 𝜕 2 𝜙1
𝐾21 = ∬( 2 + ) 𝑑𝑦 𝑑𝑥
𝜕𝑥 𝜕𝑦 2

𝜕 2 𝜙1 𝜕 2 𝜙2
𝐾12 = ∬( 2 + ) 𝑑𝑦 𝑑𝑥
𝜕𝑥 𝜕𝑦 2

𝜕 2 𝜙2 𝜕 2 𝜙2
𝐾22 = ∬( 2 + ) 𝑑𝑦 𝑑𝑥
𝜕𝑥 𝜕𝑦 2

𝑞1 = ∬ −2 𝜙1 𝑑𝑦 𝑑𝑥

𝑞2 = ∬ −2 𝜙2 𝑑𝑦 𝑑𝑥

For all, x limit from -2 to 2 and y limit from -1 to 1

𝜋𝑥 𝜋𝑦 3𝜋𝑥 𝜋𝑦
ϕ1 = cos cos and ϕ2 = cos cos
4 2 4 2

𝜕2 𝜙1 𝜋2 𝜋𝑥 𝜋𝑦
= − cos cos
𝜕𝑥 2 16 4 2

𝜕2 𝜙1 𝜋2 𝜋𝑥 𝜋𝑦
= − cos cos
𝜕𝑦 2 4 4 2

𝜕2 𝜙2 9𝜋 2 3𝜋𝑥 𝜋𝑦
= − cos cos
𝜕𝑥 2 16 4 2

𝜕2 𝜙2 𝜋2 3𝜋𝑥 𝜋𝑦
= − cos cos
𝜕𝑦 2 4 4 2

𝐾21 = 𝐾12 = 0 (pretty obvious in these types of problem.)

Sample way of doing:

𝜕 2 𝜙1 𝜕 2 𝜙1
𝐾11 = ∬( + ) 𝑑𝑦 𝑑𝑥
𝜕𝑥 2 𝜕𝑦 2
𝜋2 𝜋𝑥 𝜋𝑦 𝜋 2 𝜋𝑥 𝜋𝑦
𝐾11 = ∬(− cos cos − cos cos ) 𝑑𝑦 𝑑𝑥
16 4 2 4 4 2
2
5𝜋 𝜋𝑥 𝜋𝑦
𝐾11 = ∬(− cos cos ) 𝑑𝑦 𝑑𝑥
8 4 2
5𝜋 2 2 𝜋𝑥 1 𝜋𝑦 5𝜋 2
𝐾11 = − ∫−2 cos 𝑑𝑥 ∫−1 cos 𝑑𝑦 = −
8 4 2 8

Similarly find All K and q value. Solve from matrix above and calculate C1 and C2.Replace those
to
𝜋𝑥 𝜋𝑦 3𝜋𝑥 𝜋𝑦
𝜙 = 𝐶1 cos cos + 𝐶2 cos cos
4 2 4 2
Take Away Question

3.1 Find the one parameter Galerkin solution of the equation for the deflection of a membrane
subjected to uniform transverse load q and tension T

𝜕 2𝜙 𝜕 2𝜙 𝑞
2
+ 2 =− in Ω Ω: −2 ≤ x ≤ 2, −1 ≤ y ≤ 1
𝜕𝑥 𝜕𝑦 𝑇

𝜙=0 on Γ.

3.2 Find the two parameter Galerkin solution of the equation

𝜕 2𝜙 𝜕 2𝜙
+ = −2 in Ω Ω: −3 ≤ x ≤ 3, −2 ≤ y ≤ 2
𝜕𝑥 2 𝜕𝑦 2

subjected to boundary conditions 𝜙 = 0 on Γ

Type 4: Finding the variation form of the differential equation

A string of length L, is subjected to a lateral distributed load of intensity 𝑞 per unit length. The
governing differential equation is given by

𝑑2 𝑤
𝑇 +𝑞 =0
𝑑𝑥 2
Derive the variational form of the problem.

Solution:

Governing differential equation is:

𝑑2 𝑤
𝑇 2 +𝑞 =0
𝑑𝑥
Taking weighting function as 𝛿𝑤 and using weighted residual statement.

𝐿 𝑑2 𝑤
∫0 (𝑇 𝑑𝑥 2 + 𝑞) 𝛿𝑤𝑑𝑥 =0

𝐿 𝑑2 𝑤 𝐿
∫0 (𝑇 𝑑𝑥 2 ) 𝛿𝑤𝑑𝑥 + ∫0 (𝑞)𝛿𝑤𝑑𝑥 =0

Using integration by parts

∫ 𝑢′ 𝑣𝑑𝑥 = 𝑢𝑣 − ∫ 𝑢𝑣 ′ 𝑑𝑥

𝑑𝑤 𝐿 𝑑𝑤 𝑑𝑤 𝐿
𝑇 𝑑𝑥 𝛿𝑤|𝐿0 - ∫0 𝑇 ( ) 𝛿( )𝑑𝑥 + 𝛿 ∫0 𝑞𝑤𝑑𝑥 =0
𝑑𝑥 𝑑𝑥

𝑑𝑤 1 𝐿 𝑑𝑤 2 𝐿
𝑇 𝑑𝑥 𝛿𝑤|𝐿0 - 𝛿 2 ∫0 𝑇 ( 𝑑𝑥 ) 𝑑𝑥 + 𝛿 ∫0 𝑞𝑤𝑑𝑥 =0

𝑑𝑤 𝐿 𝑇 𝑑𝑤 2
𝑇 𝑑𝑥 𝛿𝑤|𝐿0 - 𝛿 ∫0 (2 ( 𝑑𝑥 ) + 𝑞𝑤)𝑑𝑥 =0

We get the differential equation

𝑇 𝑑𝑤 2
( ) + 𝑞𝑤 = 0
2 𝑑𝑥

Boundary condition

Natural: All conditions not containing 𝛿

𝑑𝑤 𝑑𝑤
| =0 and |𝐿 =0 and
𝑑𝑥 0 𝑑𝑥

Essential: Inside of 𝛿

𝑤|0 = 0 𝑎𝑛𝑑 𝑤|𝐿 = 0

Take Away Question

4.1 A 1-Dimensional beam of length L, is subjected to a lateral distributed load of intensity 𝑞 per
unit length. The governing differential equation is given by

𝑑4𝑤
𝐸𝐼 4 + 𝑞 = 0
𝑑𝑥

Derive the variational form of the problem.

Type 5: Euler Equation derivation and its use

5. Derive Euler equation for the function below.

𝑏
I = ∫𝑎 𝐹(𝑢, 𝑢′ , 𝑢′′ ) 𝑑𝑥

F = 𝐹(𝑢, 𝑢′ , 𝑢′′ )

From multivariable calculus (total differential);

𝜕𝐹 𝜕𝐹 𝜕𝐹
dF = 𝜕𝑢 𝑑𝑢 + 𝑑𝑢′ + 𝑑𝑢′′
𝜕𝑢′ 𝜕𝑢′′

This is valid for both exact and inexact differential

Using differentiation under integral

𝑏
𝛿I = ∫𝑎 𝛿𝐹(𝑢, 𝑢′ , 𝑢′′ ) 𝑑𝑥

𝑏 𝜕𝐹 𝜕𝐹 𝜕𝐹
= ∫𝑎 𝛿𝑢 + 𝛿𝑢′ + 𝛿𝑢′′ 𝑑𝑥
𝜕𝑢 𝜕𝑢′ 𝜕𝑢′′

𝑏 𝜕𝐹 𝑏 𝜕𝐹
∫𝑎 𝛿𝑢 𝑑𝑥 = ∫𝑎 𝛿𝑢 𝑑𝑥 … (1)
𝜕𝑢 𝜕𝑢

𝑏 𝜕𝐹 𝜕𝐹 𝑏 𝑑 𝜕𝐹 𝑏 𝑑 𝜕𝐹
∫𝑎 𝛿𝑢′ 𝑑𝑥 = =𝜕𝑢′ 𝛿𝑢|𝑏𝑎 − ∫𝑎 ( ) 𝛿𝑢𝑑𝑥 = − ∫𝑎 ( ) 𝛿𝑢𝑑𝑥 (2)
𝜕𝑢′ 𝑑𝑥 𝜕𝑢′ 𝑑𝑥 𝜕𝑢′

u is constant at the boundary so, 𝛿𝑢=0.

 𝑏 𝜕𝐹 𝜕𝐹 𝑏 𝑑 𝜕𝐹 𝑏 𝑑 𝜕𝐹 𝑑 𝜕𝐹
∫𝑎 𝛿𝑢"𝑑𝑥 = 𝜕𝑢" 𝛿𝑢′|𝑏𝑎 − ∫𝑎 ( ) 𝛿𝑢′𝑑𝑥 = − ∫𝑎 ( ) 𝛿𝑢′ 𝑑𝑥 = − ( ) 𝛿𝑢|𝑏𝑎
𝜕𝑢" 𝑑𝑥 𝜕𝑢′′ 𝑑𝑥 𝜕𝑢" 𝑑𝑥 𝜕𝑢"
𝑏 𝑑 𝜕𝐹
+∫𝑎 𝑑2 𝑥 (𝜕𝑢") 𝛿𝑢 𝑑𝑥 (3)

Adding equation 1 2 and 3

𝑏 𝜕𝐹 𝜕𝐹 𝜕𝐹 𝑏 𝜕𝐹 𝑏 𝑑 𝜕𝐹 𝑏 𝑑 𝜕𝐹
∫𝑎 𝛿𝑢 + 𝛿𝑢′ + 𝛿𝑢′′ 𝑑𝑥 = ∫𝑎 𝛿𝑢 𝑑𝑥 + − ∫𝑎 ( ) 𝛿𝑢 𝑑𝑥 + +∫𝑎 ( ) 𝛿𝑢 𝑑𝑥
𝜕𝑢 𝜕𝑢′ 𝜕𝑢′′ 𝜕𝑢 𝑑𝑥 𝜕𝑢′ 𝑑2 𝑥 𝜕𝑢"

𝑏 𝜕𝐹 𝑑 𝜕𝐹 𝑑 𝜕𝐹
𝛿𝐼 = ∫𝑎 ( 𝜕𝑢 − 𝑑𝑥 (𝜕𝑢′ ) + ( )) 𝛿𝑢 𝑑𝑥
𝑑2 𝑥 𝜕𝑢"

For optimization; 𝛿𝐼 = 0
𝑏
𝜕𝐹 𝑑 𝜕𝐹 𝑑 𝜕𝐹
∫( − ( ′) + 2 ( )) 𝛿𝑢 𝑑𝑥 = 0
𝜕𝑢 𝑑𝑥 𝜕𝑢 𝑑 𝑥 𝜕𝑢"
𝑎

𝜕𝐹 𝑑 𝜕𝐹 𝑑 𝜕𝐹
− ( ′) + 2 ( )
𝜕𝑢 𝑑𝑥 𝜕𝑢 𝑑 𝑥 𝜕𝑢"
Take Away Questions

5.1 Derive Euler equation for the function below.

𝑏
I = ∫𝑎 𝐹(𝑢, 𝑢′ , 𝑢′′ , 𝑢′′′) 𝑑𝑥

5.2 Derive Euler equation for the function below.

𝑏
I = ∫𝑎 𝐹(𝑢, 𝑢′ ) 𝑑𝑥

5.3 Find the governing differential equation for a uniform beam string subjected to lateral load q
for which potential energy is given by

𝐿 𝐿
1 2
Π = ∫ 𝐸𝐼(𝑤,𝑥𝑥 ) 𝑑𝑥 − ∫ 𝑞𝑤𝑑𝑥
2
0 0

5.4 Find the governing differential equation of a string under tension T subjected to lateral load q
for which potential energy is given by
𝐿 𝐿
𝑇 𝑑𝜙 2
Π = ∫ ( ) 𝑑𝑥 − ∫ 𝑞𝜙𝑑𝑥
2 𝑑𝑥
0 0
Type 6: Using Weighted Residual Method

Find two parameter approximate solution of the equation

𝒅𝟐 𝝓 𝒅𝝓
𝟐
−𝟑 +𝝓=𝒙 𝟎≤𝒙≤𝟏
𝒅𝒙 𝒅𝒙

subjected to boundary conditions 𝝓(𝟎) = 𝟎, 𝝓′ (𝟏) = 𝟏 . Use

(a) the Galerkin method
(b) the Petrov-Galerkin method
(c) the least squares method.
(d) the point collocation method

Solution

For a weighted residual method, 𝝓𝟎 and 𝝓𝒊 should satisfy the following conditions:
𝝓𝟎 (𝟎) = 𝟎
𝝓′𝟎 (𝟏) = 𝟏

satisfy actual boundary conditions.

Let us assume that 𝝓𝟎 = 𝒂 + 𝒃𝒙 . Substituting the specified boundary conditions and assuming and ,
We get,
𝝓𝟎 = 𝒙

Selection of 𝝓𝟏 :

Let us assume that 𝝓𝟏 = 𝒃𝟎 + 𝒃𝟏 𝒙 + 𝒃𝟐 𝒙𝟐 . Substituting the specified boundary conditions and

assuming 𝑏2 = 1,
We get,
𝑏0 = 0 𝑎𝑛𝑑 𝑏1 = −2

𝝓𝟏 = −𝟐𝒙 + 𝒙𝟐

Selection of 𝝓𝟐 :

Let us assume that 𝝓𝟐 = 𝒄 + 𝒄𝟏 𝒙 + 𝒄𝟐 𝒙𝟐 + 𝒄𝟑 𝒙𝟑 . Substituting the specified boundary conditions

and assuming 𝑐2 = 0 𝑎𝑛𝑑 𝑐3 = 1,
We get,
𝑐1 = −3 𝑎𝑛𝑑 𝑐2 = 0

𝜙2 = −3𝑥 + 𝑥 3
(Cheat idea: satisfy the ϕ0 with the boundary condition and chose higher
ϕi with increasing degree and satisfying equation)

Approximate solution:

𝜙̂ = 𝜙0 + 𝐶1 𝜙1 + 𝐶2 𝜙2

𝜙̂ = 𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)

Finding Residual:
Take differential equation in one side and use 𝜙̂

𝑑2 𝜙
̂ ̂
𝑑𝜙
𝑅 = − 3 ̂−𝑥
+ 𝜙
𝑑𝑥2 𝑑𝑥
𝑑2 𝑑
𝑅= 2
(𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)) − 3 (𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)) + 𝑥
𝑑𝑥 𝑑𝑥
+ 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥) − 𝑥

𝑅 = 2𝐶1 + 6𝐶2 𝑥 − 3[1 + 𝐶1 (−2 + 2𝑥) + 𝐶2 (−3 + 3𝑥 2 )]2 + [𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)]

−𝑥

Galerkin Method

𝜓1 = 𝜙1 = 𝑥 2 − 2𝑥

𝜓2 = 𝜙2 = 𝑥 3 − 3𝑥

Weighted residual statement.

1

∫ 𝜓1 𝑅 𝑑𝑥 = 0
0

1
Or, ∫0 𝜙1 𝑅 𝑑𝑥 = 0

and
1

∫ 𝜓2 𝑅 𝑑𝑥 = 0
0

1
∫0 𝜙2 𝑅 𝑑𝑥 =0

Two equations two unknowns.

Solve to find C1 and C2.

And replace in 𝜙 = 𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)

Petrov Galerkin Method

𝜓1 = 𝑥

𝜓2 = 𝑥 2

Weighted residual statement.

1

∫ 𝜓1 𝑅 𝑑𝑥 = 0
0

1
Or, ∫0 𝑥 𝑅 𝑑𝑥 = 0

and
1

∫ 𝜓2 𝑅 𝑑𝑥 = 0
0

1
∫0 𝑥 2 𝑅 𝑑𝑥 =0

Two equations two unknowns.

Solve to find C1 and C2.

And replace in 𝜙 = 𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)

The Least-Squares method

 𝜕𝑅
𝜓1 =
𝜕𝐶1

𝜕𝑅
𝜓2 =
𝜕𝐶2

Weighted residual statement.

1

∫ 𝜓1 𝑅 𝑑𝑥 = 0
0

1 𝜕𝑅
Or, ∫0 𝑅 𝑑𝑥 = 0
𝜕𝐶1

and
1

∫ 𝜓2 𝑅 𝑑𝑥 = 0
0

1 𝜕𝑅
∫0 𝑅 𝑑𝑥 =0
𝜕𝐶2

Two equations two unknowns.

Solve to find C1 and C2.

And replace in 𝜙 = 𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)

The point collocation Method

𝜓1 = 𝛿(𝑥 − 𝑎)

𝜓2 = 𝛿(𝑥 − 𝑏)

Weighted residual statement.

1

∫ 𝜓1 𝑅 𝑑𝑥 = 0
0

1
Or, ∫0 𝛿(𝑥 − 𝑎)𝑅 𝑑𝑥 = 0

Or, R(a) = 0 i.e. replace x = a in residual

and
1

∫ 𝜓2 𝑅 𝑑𝑥 = 0
0

1
∫0 𝛿(𝑥 − 𝑏) 𝑅 𝑑𝑥 =0

R(b) = 0 i.e. replace x = b in residual

a and b trisect the boundary.

For eg.

If boundary is from 0 to 1.
a = 1/3 and b = 2/3

If boundary is from x to y

a = (x+y)/3 and b = 2(x+y)/3

Two equations two unknowns. Solve to find C1 and C2.

And replace in 𝜙 = 𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)

Take Away Questions

6.1 Find two parameter approximate solution of the equation using Galerkin method

𝑑2𝜙 𝑑𝜙
2
+3 + 𝜙 = 4𝑥 0≤𝑥≤1
𝑑𝑥 𝑑𝑥

subjected to boundary conditions

𝜙(0) = 0, 𝜙(1) = 20

Trial: 𝜙 = 20𝑥 + 𝐶1 𝑥(𝑥 − 1) + 𝐶2 𝑥 2 (𝑥 − 1)

6.2 Find two parameter approximate solution of the equation using Galerkin Method.

𝑑2𝜙 𝑑𝜙
2
+3 + 𝜙 = 4𝑥 0≤𝑥≤1
𝑑𝑥 𝑑𝑥
subjected to boundary conditions

𝜙(0) = 5, 𝜙′(1) = 2

Trial: 𝜙 = 5 + 2𝑥 + 𝐶1 (𝑥 2 − 2𝑥) + 𝐶2 (𝑥 3 − 3𝑥)

6.3 Find two parameter approximate solution of the equation using Galerkin Method.

𝑑2𝜙 𝑑𝜙
2
+3 + 𝜙 = 4𝑥 0≤𝑥≤1
𝑑𝑥 𝑑𝑥

subjected to boundary conditions

𝜙′(0) = 2, 𝜙′(1) = 20

Trial: 𝜙 = 2𝑥 + 9𝑥 2 + 𝐶1 (2𝑥 3 − 3𝑥 2 ) + 𝐶2 (2𝑥 4 − 4𝑥 2 )

6.4 A certain physical problem has the functional

Essential boundary conditions are 𝜙 = 0 at x = 0 and 𝜙 = 100 at x = L. Express the general form of 𝜙
and find a 2-term solution for 𝜙.
100𝑥
Trial: 𝜙 = + 𝐶1 𝑥(𝐿 − 𝑥) + 𝐶2 𝑥 2 (𝐿 − 𝑥)
𝐿

6.6 Find two parameter approximate solution of the equation

𝜙(1) =0 and 𝜙(4) = 12

Trial: 𝜙 = −4 + 4𝑥 + 𝐶1 (𝑥 − 1)(𝑥 − 4) + 𝐶2 (𝑥 − 4)(𝑥 − 1)2

Type 7: Formulating Shape function

7. The nodal values of the element, as shown in Figure below, in a heat conduction
problem are: T4 = 1200C, T9 = 2100C, T10 = 1500C.

Determine where the 1600C isotherm cuts the boundaries of the element. Also
determine the [K] matrix for the element.

Solution:
Area(A) = 1/2 * 6*6 = 18

1 𝑥1 𝑦1
In general, 2*A = det [1 𝑥2 𝑦2 ]
1 𝑥3 𝑦3

1 𝑥1 𝑦1
A = ½ det [1 𝑥2 𝑦2 ]
1 𝑥3 𝑦3
𝑛𝑜𝑑𝑒 4 = (𝑥1 , 𝑦1 ) = (2,4)
𝑛𝑜𝑑𝑒 9 = (𝑥2 , 𝑦2 ) = (8,10)
𝑛𝑜𝑑𝑒 10 = (𝑥3 , 𝑦3 ) = (2,10)

𝑎1 = 𝑥2 𝑦3 − 𝑦2 𝑥3 , 𝑏1 = 𝑦2 − 𝑦3 , 𝑐1 = 𝑥3 − 𝑥2

𝑎2 = 𝑥3 𝑦1 − 𝑥1 𝑦3 , 𝑏2 = 𝑦3 − 𝑦1 , 𝑐2 = 𝑥1 − 𝑥3

𝑎3 = 𝑥1 𝑦2 − 𝑥2 𝑦1, 𝑏3 = 𝑦1 − 𝑦2 , 𝑐3 = 𝑥2 − 𝑥1
1
𝑁1 = 2𝐴 (𝑎1 + 𝑏1 𝑥 + 𝑐1 𝑦),
 1
𝑁2 = (𝑎 + 𝑏2 𝑥 + 𝑐2 𝑦)
2𝐴 2
1
𝑁3 = (𝑎 + 𝑏3 𝑥 + 𝑐3 𝑦)
2𝐴 3
𝑇 = 𝑁1 𝑇1 + 𝑁2 𝑇2 + 𝑁3 𝑇3 … (i)

three sides of triangle = three equations.

Solve each with equation (i) to find where isotherm cuts the boundary.

Take Away Question

7.1 Evaluate the element shape functions and calculate the value of the pressure at point
A in Figure P.6, if the nodal values are 𝜙𝑖 = 60 𝑁/𝑐𝑚2, 𝜙𝑗 = 50 𝑁/𝑐𝑚2 and 𝜙𝑘 =
70 𝑁/𝑐𝑚2 . Point A is located at (2, 1.5). Also determine the location of the 65 𝑁/𝑐𝑚2
contour line for the element.

Figure P.6

7.2 Derive expression for shape function for two nodded linear elements.

7.3 Derive expression for shape function for three nodded linear elements.