CECB423 Numerical Methods for Civil Engineers

Boundary Value Problem (BVP)

a) Boundary condition without derivative

Consider the general linear two point BVP,

𝑦 ′′ (𝑥) + 𝑝(𝑥)𝑦 ′ (𝑥) + 𝑞(𝑥)𝑦(𝑥) = 𝑟(𝑥); 𝑎≤𝑥≤𝑏

With the boundary condition (without derivative)

𝑦(𝑎) = 𝛼 and 𝑦(𝑏) = 𝛽

To derive the finite-difference equation, the interval [a,b] is divided by step size h,

First boundary condition at Second boundary condition at

𝑥=𝛼 𝑥=𝑏
𝑦(𝑎) = 𝛼 𝑦(𝑏) = 𝛽
𝑦(𝑥0 ) = 𝛼 𝑦(𝑥𝑛 ) = 𝛽
𝑦0 = 𝛼 𝑦𝑛 = 𝛽

At grid points 𝑥 = 𝑥𝑖 , the differential equation can be written as

𝑦 ′′ (𝑥𝑖 ) + 𝑝(𝑥𝑖 )𝑦 ′ (𝑥𝑖 ) + 𝑞(𝑥𝑖 )𝑦(𝑥𝑖 ) = 𝑟(𝑥𝑖 )

By using central differentiation

𝑦𝑖+1 − 𝑦𝑖−1
𝑦 ′ (𝑥𝑖 ) ≈
2ℎ
𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1
𝑦 ′′ (𝑥𝑖 ) ≈
ℎ2
Substituting the approximation into above equation,

𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1 𝑦𝑖+1 − 𝑦𝑖−1

[ 2 ] + 𝑝𝑖 [ ] + 𝑞𝑖 𝑦𝑖 = 𝑟𝑖
ℎ 2ℎ
 CECB423 Numerical Methods for Civil Engineers

And multiplying throughout by h2

ℎ
𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1 + 𝑝𝑖 (𝑦𝑖+1 − 𝑦𝑖−1 ) + ℎ2 𝑞𝑖 𝑦𝑖 = ℎ2 𝑟𝑖
2

Rearrange the equation,

ℎ ℎ
(1 − 𝑝𝑖 )𝑦𝑖−1 + (−2 + ℎ2 𝑞𝑖 )𝑦𝑖 + (1 + 𝑝𝑖 )𝑦𝑖+1 = ℎ2 𝑟𝑖
2 2

Expanding the finite difference equation for 𝑖 = 1,2,3, … . , 𝑛

i.e:

at i=1 ℎ ℎ
(1 − 𝑝1 )𝑦0 + (−2 + ℎ2 𝑞1 )𝑦1 + (1 + 𝑝1 )𝑦2 = ℎ2 𝑟1
2 2

at i=2 ℎ ℎ
(1 − 𝑝2 )𝑦1 + (−2 + ℎ2 𝑞2 )𝑦2 + (1 + 𝑝2 )𝑦3 = ℎ2 𝑟2
2 2

at i=3 ℎ ℎ
(1 − 𝑝3 )𝑦2 + (−2 + ℎ2 𝑞3 )𝑦3 + (1 + 𝑝3 )𝑦4 = ℎ2 𝑟3
2 2

Combine all equation in linear system of from 𝐴𝑦 = 𝑏

Solve the value of 𝑦1 , 𝑦2 , 𝑦3 , … … … . , 𝑦𝑛

Example:

Solve the BVP

1 1
𝑦 ′′ + 𝑦 ′ + 2 𝑦(𝑥𝑖 ) = 3; 𝑦(1) = 2, 𝑦(2) = 3
𝑥 𝑥

For x at interval [1,2] with h=0.2. Use four decimal places in all calculation.
 CECB423 Numerical Methods for Civil Engineers

1 1
𝑦 ′′ + 𝑦 ′ + 2 𝑦(𝑥𝑖 ) = 3
𝑥 𝑥
𝑦 ′′ (𝑥) + 𝑝(𝑥)𝑦 ′ (𝑥) + 𝑞(𝑥)𝑦(𝑥) = 𝑟(𝑥)

First boundary condition at Second boundary condition at

𝑦(1) = 2 𝑦(2) = 3
𝑦(𝑥0 ) = 2 𝑦(𝑥5 ) = 3
𝑦0 = 2 𝑦5 = 3

ℎ ℎ
(1 − 𝑝𝑖 )𝑦𝑖−1 + (−2 + ℎ2 𝑞𝑖 )𝑦𝑖 + (1 + 𝑝𝑖 )𝑦𝑖+1 = ℎ2 𝑟𝑖
2 2

at i=1 0.2 1 1 0.2 1

[1 − ( ∗ )] 𝑦0 + [−2 + (0.22 ∗ 2 )] 𝑦1 + [1 + ( ∗ )] 𝑦2 = (0.2)2 ∗ 3
(x1=1.2) 2 𝑥 𝑥 2 𝑥
(y0=2)
0.2 1 1 0.2 1
[1 − ( ∗ )] (2) + [−2 + (0.22 ∗ )] 𝑦1 + [1 + ( ∗ )] 𝑦2 = (0.2)2 ∗ 3
2 1.2 1.22 2 1.2

1.8333 − 2.0278𝑦1 + 1.0833𝑦2 = 0.12

−𝟐. 𝟎𝟐𝟕𝟖𝒚𝟏 + 𝟏. 𝟎𝟖𝟑𝟑𝒚𝟐 = −𝟏. 𝟕𝟏𝟑𝟑

at i=2 0.2 1 1 0.2 1

[1 − ( ∗ )] 𝑦1 + [−2 + (0.22 ∗ 2 )] 𝑦2 + [1 + ( ∗ )] 𝑦3 = (0.2)2 ∗ 3
(x2=1.4) 2 𝑥 𝑥 2 𝑥

0.2 1 1 0.2 1
[1 − ( ∗ )] 𝑦1 + [−2 + (0.22 ∗ )] 𝑦2 + [1 + ( ∗ )] 𝑦3 = (0.2)2 ∗ 3
2 1.4 1.42 2 1.4

𝟎. 𝟗𝟐𝟖𝟔𝒚𝟏 − 𝟐. 𝟎𝟐𝟎𝟒𝒚𝟐 + 𝟏. 𝟎𝟕𝟏𝟒𝒚𝟑 = 𝟎. 𝟏𝟐

at i=3 0.2 1 1 0.2 1

[1 − ( ∗ )] 𝑦2 + [−2 + (0.22 ∗ 2 )] 𝑦3 + [1 + ( ∗ )] 𝑦4 = (0.2)2 ∗ 3
(x3=1.6) 2 𝑥 𝑥 2 𝑥

0.2 1 1 0.2 1
[1 − ( ∗ )] 𝑦2 + [−2 + (0.22 ∗ 2
)] 𝑦3 + [1 + ( ∗ )] 𝑦4 = (0.2)2 ∗ 3
2 1.6 1.6 2 1.6

𝟎. 𝟗𝟑𝟕𝟓𝒚𝟐 − 𝟐. 𝟎𝟏𝟓𝟔𝒚𝟑 + 𝟏. 𝟎𝟔𝟐𝟓𝒚𝟒 = 𝟎. 𝟏𝟐

at i=4 0.2 1 1 0.2 1

[1 − ( ∗ )] 𝑦3 + [−2 + (0.22 ∗ 2 )] 𝑦4 + [1 + ( ∗ )] 𝑦5 = (0.2)2 ∗ 3
(x4=1.8) 2 𝑥 𝑥 2 𝑥
(y5=3)
0.2 1 1 0.2 1
[1 − ( ∗ )] 𝑦3 + [−2 + (0.22 ∗ 2
)] 𝑦4 + [1 + ( ∗ )] (3) = (0.2)2 ∗ 3
2 1.8 1.8 2 1.8

0.9444𝑦3 − 2.0123𝑦4 + 3.1668 = 0.12

𝟎. 𝟗𝟒𝟒𝟒𝒚𝟑 − 𝟐. 𝟎𝟏𝟐𝟑𝒚𝟒 = −𝟑. 𝟎𝟒𝟔𝟖

 CECB423 Numerical Methods for Civil Engineers

Solve the system using trigonal system

−2.0278 1.0833 0 0 𝑦1 −1.7133

0.9286 −2.0204 1.0714 0 𝑦2 0.1200
( ) (𝑦 ) = ( )
0 0.9375 −2.0156 1.0625 3 0.1200
0 0 0.9444 −2.0123 𝑦4 −3.0468

𝑦1
𝑦2
(𝑦 ) = ( )
3
𝑦4

b) Boundary condition with derivative (Initial)

Consider the general linear two point BVP,

𝑦 ′′ (𝑥) + 𝑝(𝑥)𝑦 ′ (𝑥) + 𝑞(𝑥)𝑦(𝑥) = 𝑟(𝑥); 𝑎≤𝑥≤𝑏

With the boundary condition (with derivative 𝑥 = 𝑎)

𝑘1 𝑦(𝑎) + 𝑘2 𝑦′(𝑎) = 𝛼 and 𝑦(𝑏) = 𝛽

To derive the finite-difference equation, the interval [a,b] is divided by step size h,

First boundary condition at Second boundary condition at

𝑘1 𝑦(𝑎) + 𝑘2 𝑦′(𝑎) = 𝛼 𝑥=𝑏
𝑘1 𝑦(𝑥0 ) + 𝑘2 𝑦′(𝑥0 ) = 𝛼 𝑦(𝑏) = 𝛽
𝑘1 𝑦0 + 𝑘2 𝑦′0 = 𝛼 𝑦(𝑥𝑛 ) = 𝛽
𝑦1 − 𝑦−1 𝑦𝑛 = 𝛽
𝑘1 𝑦0 + 𝑘2 ( )=𝛼
2ℎ
2ℎ𝑘1
𝑦−1 =
𝑘2
 CECB423 Numerical Methods for Civil Engineers

At grid points 𝑥 = 𝑥𝑖 , the differential equation can be written as

𝑦 ′′ (𝑥𝑖 ) + 𝑝(𝑥𝑖 )𝑦 ′ (𝑥𝑖 ) + 𝑞(𝑥𝑖 )𝑦(𝑥𝑖 ) = 𝑟(𝑥𝑖 )

By using central differentiation

𝑦𝑖+1 − 𝑦𝑖−1
𝑦 ′ (𝑥𝑖 ) ≈
2ℎ
′′ (𝑥 )
𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1
𝑦 𝑖 ≈
ℎ2
Substituting the approximation into above equation,

𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1 𝑦𝑖+1 − 𝑦𝑖−1

[ ] + 𝑝𝑖 [ ] + 𝑞𝑖 𝑦𝑖 = 𝑟𝑖
ℎ2 2ℎ

And multiplying throughout by h2

ℎ
𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1 + 𝑝𝑖 (𝑦𝑖+1 − 𝑦𝑖−1 ) + ℎ2 𝑞𝑖 𝑦𝑖 = ℎ2 𝑟𝑖
2

Rearrange the equation,

ℎ ℎ
(1 − 𝑝𝑖 )𝑦𝑖−1 + (−2 + ℎ2 𝑞𝑖 )𝑦𝑖 + (1 + 𝑝𝑖 )𝑦𝑖+1 = ℎ2 𝑟𝑖
2 2

Expanding the finite difference equation for 𝑖 = 1,2,3, … . , 𝑛

i.e:

at i=0 ℎ ℎ
(1 − 𝑝0 )𝑦−1 + (−2 + ℎ2 𝑞0 )𝑦0 + (1 + 𝑝0 )𝑦1 = ℎ2 𝑟0
2 2
ℎ 2ℎ𝑘1 ℎ
(1 − 𝑝0 )( ) + (−2 + ℎ2 𝑞0 )𝑦0 + (1 + 𝑝0 )𝑦1 = ℎ2 𝑟0
2 𝑘2 2

at i=1 ℎ ℎ
(1 − 𝑝1 )𝑦0 + (−2 + ℎ2 𝑞1 )𝑦1 + (1 + 𝑝1 )𝑦2 = ℎ2 𝑟1
2 2

at i=2 ℎ ℎ
(1 − 𝑝2 )𝑦1 + (−2 + ℎ2 𝑞2 )𝑦2 + (1 + 𝑝2 )𝑦3 = ℎ2 𝑟2
2 2

Combine all equation in linear system of from 𝐴𝑦 = 𝑏

Solve the value of 𝑦0 , 𝑦1 , 𝑦2 , … … … . , 𝑦𝑛

 CECB423 Numerical Methods for Civil Engineers

Example:

Solve the BVP

1 1
𝑦 ′′ + 𝑦 ′ + 2 𝑦(𝑥𝑖 ) = 3; 𝑦 ′ (1) + 𝑦(1) = 1, 𝑦(2) = 3
𝑥 𝑥

For x at interval [1,2] with h=0.2. Use four decimal places in all calculation.

First boundary condition at Second boundary condition at

𝑦 ′ (1) + 𝑦(1) = 1, 𝑦(2) = 3
𝑦′0 + 𝑦0 = 1 𝑦(𝑥5 ) = 3
𝑦1 − 𝑦−1 𝑦5 = 3
+ 𝑦0 = 1
2ℎ
𝑦1 − 𝑦−1 + 2ℎ𝑦0 = 2ℎ
𝑦−1 = −0.4 + 0.4𝑦0 + 𝑦1

1 1
𝑦 ′′ + 𝑦 ′ + 2 𝑦(𝑥𝑖 ) = 3
𝑥 𝑥
𝑦 ′′ (𝑥) + 𝑝(𝑥)𝑦 ′ (𝑥) + 𝑞(𝑥)𝑦(𝑥) = 𝑟(𝑥)

ℎ ℎ
(1 − 𝑝𝑖 )𝑦𝑖−1 + (−2 + ℎ2 𝑞𝑖 )𝑦𝑖 + (1 + 𝑝𝑖 )𝑦𝑖+1 = ℎ2 𝑟𝑖
2 2

at i=0 ℎ ℎ
(1 − 𝑝0 )𝑦−1 + (−2 + ℎ2 𝑞0 )𝑦0 + (1 + 𝑝0 )𝑦1 = ℎ2 𝑟0
2 2
x0=1.0
ℎ 1 1 ℎ 1
(1 − ∗ ) (−0.4 + 0.4𝑦0 + 𝑦1 ) + (−2 + ℎ2 ∗ 2
) 𝑦0 + (1 + ∗ ) 𝑦1 = ℎ2 ∗ 3
2 𝑥0 (𝑥0 ) 2 𝑥0
0.2 1 1 0.2 1
(1 − ∗ ) (−0.4 + 0.4𝑦0 + 𝑦1 ) + (−2 + 0.22 ∗ ) 𝑦0 + (1 + ∗ ) 𝑦 = 0.22 ∗ 3
2 1.0 1.02 2 1.0 1
0.90(−0.4 + 0.4𝑦0 + 𝑦1 ) − 2.0400𝑦0 + 1.1000𝑦1 = 0.12
−𝟏. 𝟔𝟖𝟎𝟎𝒚𝟎 + 𝟐. 𝟎𝟎𝟎𝟎𝒚𝟏 = 𝟎. 𝟒𝟖𝟎𝟎
 CECB423 Numerical Methods for Civil Engineers

at i=1 ℎ ℎ
(1 − 𝑝1 )𝑦0 + (−2 + ℎ2 𝑞1 )𝑦1 + (1 + 𝑝1 )𝑦2 = ℎ2 𝑟1
2 2
x1=1.2
ℎ 1 1 ℎ 1
(1 − ∗ ) 𝑦0 + (−2 + ℎ2 ∗ 2
) 𝑦1 + (1 + ∗ ) 𝑦2 = ℎ2 ∗ 3
2 𝑥1 (𝑥1 ) 2 𝑥1
0.2 1 1 0.2 1
(1 − ∗ ) 𝑦0 + (−2 + 0.22 ∗ ) 𝑦1 + (1 + ∗ ) 𝑦 = 0.22 ∗ 3
2 1.2 1.2 2 2 1.2 2
𝟎. 𝟗𝟏𝟔𝟕𝒚𝟎 − 𝟐. 𝟎𝟐𝟕𝟖𝒚𝟏 + 𝟏. 𝟎𝟖𝟑𝟑𝒚𝟐 = 𝟎. 𝟏𝟐

at i=2 ℎ ℎ
(1 − 𝑝2 )𝑦1 + (−2 + ℎ2 𝑞2 )𝑦2 + (1 + 𝑝2 )𝑦3 = ℎ2 𝑟2
2 2
x2=1.4
ℎ 1 1 ℎ 1
(1 − ∗ ) 𝑦1 + (−2 + ℎ2 ∗ 2
) 𝑦2 + (1 + ∗ ) 𝑦3 = ℎ2 ∗ 3
2 𝑥2 (𝑥2 ) 2 𝑥2
0.2 1 1 0.2 1
(1 − ∗ ) 𝑦1 + (−2 + 0.22 ∗ ) 𝑦2 + (1 + ∗ ) 𝑦 = 0.22 ∗ 3
2 1.4 1.4 2 2 1.4 3
𝟎. 𝟗𝟐𝟖𝟔𝒚𝟏 − 𝟐. 𝟎𝟐𝟎𝟒𝒚𝟐 + 𝟏. 𝟎𝟕𝟏𝟒𝒚𝟑 = 𝟎. 𝟏𝟐

at i=3 ℎ ℎ
(1 − 𝑝3 )𝑦2 + (−2 + ℎ2 𝑞3 )𝑦3 + (1 + 𝑝3 )𝑦4 = ℎ2 𝑟3
2 2
x3=1.6
ℎ 1 1 ℎ 1
(1 − ∗ ) 𝑦2 + (−2 + ℎ2 ∗ 2
) 𝑦3 + (1 + ∗ ) 𝑦4 = ℎ2 ∗ 3
2 𝑥3 (𝑥3 ) 2 𝑥3
0.2 1 1 0.2 1
(1 − ∗ ) 𝑦2 + (−2 + 0.22 ∗ ) 𝑦3 + (1 + ∗ ) 𝑦 = 0.22 ∗ 3
2 1.6 1.6 2 2 1.6 4
𝟎. 𝟗𝟑𝟕𝟓𝒚𝟐 − 𝟐. 𝟎𝟏𝟓𝟔𝒚𝟑 + 𝟏. 𝟎𝟔𝟐𝟓𝒚𝟒 = 𝟎. 𝟏𝟐

at i=4 ℎ ℎ
(1 − 𝑝4 )𝑦3 + (−2 + ℎ2 𝑞4 )𝑦4 + (1 + 𝑝4 )𝑦5 = ℎ2 𝑟4
2 2
x4=1.8
ℎ 1 1 ℎ 1
(1 − ∗ ) 𝑦3 + (−2 + ℎ2 ∗ 2
) 𝑦4 + (1 + ∗ ) 𝑦5 = ℎ2 ∗ 3
2 𝑥4 (𝑥4 ) 2 𝑥4
0.2 1 1 0.2 1
(1 − ∗ ) 𝑦3 + (−2 + 0.22 ∗ 2
) 𝑦4 + (1 + ∗ ) (3) = 0.22 ∗ 3
2 1.8 1.8 2 1.8
0.9444𝑦3 − 2.0123𝑦4 + 3.1668 = 0.12
𝟎. 𝟗𝟒𝟒𝟒𝒚𝟑 − 𝟐. 𝟎𝟏𝟐𝟑𝒚𝟒 = −𝟑. 𝟎𝟒𝟔𝟖
 CECB423 Numerical Methods for Civil Engineers

Solve the system using trigonal system

−1.6800 2.0000 0 0 0 𝑦0 0.4800

0.9167 −0.0278 1.0833 0 0 𝑦1 0.1200
0 0.9286 −2.0204 1.0714 0 𝑦2 = 0.1200
0 0 0.9375 −2.0156 1.0625 𝑦3 0.1200
( 0 0 0 0.9444 −2.0123) (𝑦4 ) (−3.0468)

𝑦0 1.9081
𝑦1 1.8428
𝑦2 = 1.9455
𝑦3 2.1837
𝑦
( 4 ) (2.5388)

c) Boundary condition with derivative (End)

Consider the general linear two point BVP,

𝑦 ′′ (𝑥) + 𝑝(𝑥)𝑦 ′ (𝑥) + 𝑞(𝑥)𝑦(𝑥) = 𝑟(𝑥); 𝑎≤𝑥≤𝑏

With the boundary condition (with derivative 𝑥 = 𝑏)

𝑦(𝑎) = 𝛼 and 𝑘1 𝑦(𝑏) + 𝑘2 𝑦′(𝑏) = 𝛽

To derive the finite-difference equation, the interval [a,b] is divided by step size h,

First boundary condition at Second boundary condition at

𝑥=𝑎 𝑘1 𝑦(𝑏) + 𝑘2 𝑦′(𝑏) = 𝛽
𝑦(𝑎) = 𝛼 𝑘1 𝑦(𝑥𝑛 ) + 𝑘2 𝑦′(𝑥𝑛 ) = 𝛽
𝑦(𝑥0 ) = 𝛼 𝑘1 𝑦𝑛 + 𝑘2 𝑦′𝑛 = 𝛽
𝑦0 = 𝛼 𝑦𝑛+1 − 𝑦𝑛−1
𝑘1 𝑦𝑛 + 𝑘2 ( )=𝛽
2ℎ
2ℎ𝑘1 2ℎ
𝑦𝑛+1 = 𝑦𝑛−1 − 𝑦𝑛 + 𝛽
𝑘2 𝑘2
 CECB423 Numerical Methods for Civil Engineers

At grid points 𝑥 = 𝑥𝑖 , the differential equation can be written as

𝑦 ′′ (𝑥𝑖 ) + 𝑝(𝑥𝑖 )𝑦 ′ (𝑥𝑖 ) + 𝑞(𝑥𝑖 )𝑦(𝑥𝑖 ) = 𝑟(𝑥𝑖 )

By using central differentiation

𝑦𝑖+1 − 𝑦𝑖−1
𝑦 ′ (𝑥𝑖 ) ≈
2ℎ
′′ (𝑥 )
𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1
𝑦 𝑖 ≈
ℎ2
Substituting the approximation into above equation,

𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1 𝑦𝑖+1 − 𝑦𝑖−1

[ ] + 𝑝𝑖 [ ] + 𝑞𝑖 𝑦𝑖 = 𝑟𝑖
ℎ2 2ℎ

And multiplying throughout by h2

ℎ
𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1 + 𝑝𝑖 (𝑦𝑖+1 − 𝑦𝑖−1 ) + ℎ2 𝑞𝑖 𝑦𝑖 = ℎ2 𝑟𝑖
2

Rearrange the equation,

ℎ ℎ
(1 − 𝑝𝑖 )𝑦𝑖−1 + (−2 + ℎ2 𝑞𝑖 )𝑦𝑖 + (1 + 𝑝𝑖 )𝑦𝑖+1 = ℎ2 𝑟𝑖
2 2

Expanding the finite difference equation for 𝑖 = 1,2,3, … . , 𝑛

i.e:

at i=1 ℎ ℎ
(1 − 𝑝1 )𝑦0 + (−2 + ℎ2 𝑞1 )𝑦1 + (1 + 𝑝1 )𝑦2 = ℎ2 𝑟1
2 2

at i=2 ℎ ℎ
(1 − 𝑝2 )𝑦1 + (−2 + ℎ2 𝑞2 )𝑦2 + (1 + 𝑝2 )𝑦3 = ℎ2 𝑟2
2 2

at i=n ℎ ℎ
(1 − 𝑝𝑛 )𝑦𝑛−1 + (−2 + ℎ2 𝑞𝑛 )𝑦𝑛 + (1 + 𝑝𝑛 )𝑦𝑛+1 = ℎ2 𝑟𝑛
2 2
ℎ ℎ 2ℎ𝑘1 2ℎ
(1 − 𝑝𝑛 )𝑦𝑛−1 + (−2 + ℎ2 𝑞𝑛 )𝑦𝑛 + (1 + 𝑝𝑛 )(𝑦𝑛−1 − 𝑦 + 𝛽) = ℎ2 𝑟𝑛
2 2 𝑘2 𝑛 𝑘2

Combine all equation in linear system of from 𝐴𝑦 = 𝑏

Solve the value of 𝑦1 , 𝑦2 , 𝑦3 , … … … . , 𝑦𝑛

 CECB423 Numerical Methods for Civil Engineers

Example:

Solve the BVP

𝑦 ′′ + 2𝑥𝑦 ′ − 3𝑦 = −6𝑒 −𝑥 (1 + 𝑥 2 ); 𝑦(0) = 0, 𝑦 ′ (1) = 0,

For x at interval [0,1] with h=0.2. Use four decimal places in all calculation.

First boundary condition at Second boundary condition at

𝑦(0) = 0 𝑦 ′ (1) = 0
𝑦(𝑥0 ) = 0 𝑦 ′ (𝑥5 ) = 0
𝑦0 = 0 𝑦6 − 𝑦4
=0
2ℎ
𝑦6 − 𝑦4 = 0
𝑦6 = 𝑦4

𝑦 ′′ + 2𝑥𝑦 ′ − 3𝑦 = −6𝑒 −𝑥 (1 + 𝑥 2 )

𝑦 ′′ (𝑥) + 𝑝(𝑥)𝑦 ′ (𝑥) + 𝑞(𝑥)𝑦(𝑥) = 𝑟(𝑥)

ℎ ℎ
(1 − 𝑝𝑖 )𝑦𝑖−1 + (−2 + ℎ2 𝑞𝑖 )𝑦𝑖 + (1 + 𝑝𝑖 )𝑦𝑖+1 = ℎ2 𝑟𝑖
2 2

at i=1 ℎ ℎ
(1 − 𝑝1 )𝑦0 + (−2 + ℎ2 𝑞1 )𝑦1 + (1 + 𝑝1 )𝑦2 = ℎ2 𝑟1
2 2
x1=0.2 ℎ ℎ
(1 − ∗ 2𝑥1 ) 𝑦0 + (−2 + ℎ2 ∗ (−3))𝑦1 + (1 + ∗ 2𝑥1 ) 𝑦2 = ℎ2 ∗ −6𝑒 −𝑥1 (1 + (𝑥1 )2 )
2 2
0.2 0.2
(1 − ∗ 2(0.2)) (0) + (−2 + 0.22 ∗ (−3))𝑦1 + (1 + ∗ 2(0.2)) 𝑦2 = 0.22 ∗ −6𝑒 −0.2 (1 + (0.2)2 )
2 2
−𝟐. 𝟏𝟐𝒚𝟏 + 𝟏. 𝟎𝟒𝒚𝟐 = −𝟎. 𝟐𝟎𝟒𝟒
 CECB423 Numerical Methods for Civil Engineers

at i=2 ℎ ℎ
(1 − 𝑝2 )𝑦1 + (−2 + ℎ2 𝑞2 )𝑦2 + (1 + 𝑝2 )𝑦3 = ℎ2 𝑟2
2 2
x2=0.4 ℎ ℎ
(1 − ∗ 2𝑥2 ) 𝑦1 + (−2 + ℎ2 ∗ (−3))𝑦2 + (1 + ∗ 2𝑥2 ) 𝑦3 = ℎ2 ∗ −6𝑒 −𝑥2 (1 + (𝑥2 )2 )
2 2
0.2 0.2
(1 − ∗ 2(0.4)) 𝑦1 + (−2 + 0.22 ∗ (−3))𝑦2 + (1 + ∗ 2(0.4)) 𝑦3 = 0.22 ∗ −6𝑒 −0.4 (1 + (0.4)2 )
2 2
𝟎. 𝟗𝟐𝒚𝟏 − 𝟐. 𝟏𝟐𝒚𝟐 + 𝟏. 𝟎𝟒𝒚𝟑 = −𝟎. 𝟏𝟖𝟔𝟔

at i=3 ℎ ℎ
(1 − 𝑝3 )𝑦2 + (−2 + ℎ2 𝑞3 )𝑦3 + (1 + 𝑝3 )𝑦4 = ℎ2 𝑟3
2 2
x3=0.6 ℎ ℎ
(1 − ∗ 2𝑥3 ) 𝑦2 + (−2 + ℎ2 ∗ (−3))𝑦3 + (1 + ∗ 2𝑥3 ) 𝑦4 = ℎ2 ∗ −6𝑒 −𝑥3 (1 + (𝑥3 )2 )
2 2
0.2 0.2
(1 − ∗ 2(0.6)) 𝑦2 + (−2 + 0.22 ∗ (−3))𝑦3 + (1 + ∗ 2(0.6)) 𝑦4 = 0.22 ∗ −6𝑒 −0.6 (1 + (0.6)2 )
2 2
𝟎. 𝟖𝟖𝒚𝟐 − 𝟐. 𝟏𝟐𝒚𝟑 + 𝟏. 𝟏𝟐𝒚𝟒 = −𝟎. 𝟏𝟕𝟗𝟏

at i=4 ℎ ℎ
(1 − 𝑝4 )𝑦3 + (−2 + ℎ2 𝑞4 )𝑦4 + (1 + 𝑝4 )𝑦5 = ℎ2 𝑟4
2 2
x4=0.8 ℎ ℎ
(1 − ∗ 2𝑥4 ) 𝑦3 + (−2 + ℎ2 ∗ (−3))𝑦4 + (1 + ∗ 2𝑥3 ) 𝑦5 = ℎ2 ∗ −6𝑒 −𝑥3 (1 + (𝑥4 )2 )
2 2
0.2 0.2
(1 − ∗ 2(0.8)) 𝑦3 + (−2 + 0.22 ∗ (−3))𝑦4 + (1 + ∗ 2(0.8)) 𝑦5 = 0.22 ∗ −6𝑒 −0.8 (1 + (0.8)2 )
2 2
𝟎. 𝟖𝟒𝒚𝟐 − 𝟐. 𝟏𝟐𝒚𝟑 + 𝟏. 𝟏𝟔𝒚𝟒 = −𝟎. 𝟏𝟕𝟔𝟗

at i=5 ℎ ℎ
(1 − 𝑝5 )𝑦4 + (−2 + ℎ2 𝑞5 )𝑦5 + (1 + 𝑝5 )𝑦6 = ℎ2 𝑟5
2 2
x5=1.0 ℎ ℎ
(1 − ∗ 2𝑥5 ) 𝑦4 + (−2 + ℎ2 ∗ (−3))𝑦5 + (1 + ∗ 2𝑥5 ) 𝑦6 = ℎ2 ∗ −6𝑒 −𝑥5 (1 + (𝑥5 )2 )
2 2
0.2 0.2
(1 − ∗ 2(1.0)) 𝑦4 + (−2 + 0.22 ∗ (−3))𝑦5 + (1 + ∗ 2(1.0)) 𝑦6 = 0.22 ∗ −6𝑒 −1 (1 + (1.0)2 )
2 2
𝟎. 𝟖𝟎𝒚𝟒 − 𝟐. 𝟏𝟐𝒚𝟓 + 𝟏. 𝟐𝒚𝟔 = −𝟎. 𝟏𝟕𝟔𝟔
Based on second derivative
0.80𝑦4 − 2.12𝑦5 + 1.2𝑦4 = −0.1766
𝟐. 𝟎𝟎𝒚𝟒 − 𝟐. 𝟏𝟐𝒚𝟓 = −𝟎. 𝟏𝟕𝟔𝟔
 CECB423 Numerical Methods for Civil Engineers

Solve the system using trigonal system

−2.12 1.04 0 0 0 𝑦1 −0.2044

0.92 −2.12 1.04 0 0 𝑦2 −0.1866
0 0.88 −2.12 1.12 0 𝑦3 = −0.1791
0 0 0.84 −2.12 1.16 𝑦4 −0.1769
( 0 0 0 2.00 −2.12) (𝑦5 ) (−0.1766)

𝑦1 0.4897
𝑦2 0.8017
𝑦3 = 0.9838
𝑦4 1.0724
𝑦
( 5 ) (1.0950)

d) Boundary condition with derivative (Both)

 CECB423 Numerical Methods for Civil Engineers

Partial Differential Equations (PDE).

Partial differential question (PDE) are differential equations involving more than one
independent variable. We consider a general second-order PDE in two independent
variable x and y which is written as

𝛿 2𝑢 𝛿 2𝑢 𝛿 2𝑢 𝛿𝑢 𝛿𝑢
𝐴 2
+ 𝐵 + 𝐶 2
+𝐷 +𝐸 + 𝐹𝑢 = 𝐺
𝛿𝑥 𝛿𝑥𝛿𝑦 𝛿𝑦 𝛿𝑥 𝛿𝑦
Or
𝐴𝑢𝑥𝑥 + 𝐵𝑢𝑥𝑦 + 𝐶𝑢𝑦𝑦 + 𝐷𝑢𝑥 + 𝐸𝑢𝑦 + 𝐹𝑢 = 𝐺

Where A, B, C, D, E, F and G are constants or functions of x and y. These PDEs are

classified into three group.

Elliptic <0
2
PDE is called Parabolic if 𝐵 − 4𝐴𝐶 =0
Hyperbolic >0

Determine either the equation is elliptic, parabolic or hyperbolic.

𝛿2 𝑢 𝛿2 𝑢
1. Laplace equation, 𝛿𝑥 2 + 𝛿𝑦 2 = 0

𝛿2 𝑢 𝛿2 𝑢
2. Poisson equation 𝛿𝑥 2 + 𝛿𝑦 2 = 𝑓(𝑥, 𝑦)

𝛿2𝑢 𝛿𝑢
3. Heat equation 𝑐 𝛿𝑥 2 = 𝛿𝑡

𝛿2 𝑢 𝛿2 𝑢
4. Wave equation 𝑐 𝛿𝑥 2 = 𝛿𝑡 2
 CECB423 Numerical Methods for Civil Engineers

Elliptic Equations
Finite difference method for the numerical solution of the PDEs of elliptic type.
Consider the Poisson equation,
𝛿 2𝑢 𝛿 2𝑢
+ = 𝑓(𝑥, 𝑦)
𝛿𝑥 2 𝛿𝑦 2
With boundary condition
𝑢(𝑥, 0) = 𝑓1 (𝑥), 𝑢(𝑥, 𝑏) = 𝑓2 (𝑥), 0<𝑥<𝑎
𝑢(0, 𝑦) = 𝑔1 (𝑥), 𝑢(𝑎, 𝑦) = 𝑔2 (𝑥), 0<𝑦<𝑏

Create Domain

In order to approximate the solution to PDE elliptic using finite difference method the
𝑎 𝑏
rectangular domain is partitioned into a grids ∆𝑥 = ℎ = 𝑛 and ∆𝑦 = 𝑘 = 𝑚 as shown in
figure below

𝛿 2𝑢 𝛿 2𝑢
+ = 𝑓(𝑥, 𝑦)
𝛿𝑥 2 𝛿𝑦 2
At point (𝑥𝑖 , 𝑦𝑗 )

𝛿 2 𝑢(𝑥𝑖 , 𝑦𝑗 ) 𝛿 2 𝑢(𝑥𝑖 , 𝑦𝑗 )
+ = 𝑓((𝑥𝑖 , 𝑦𝑗 )
𝛿𝑥 2 𝛿𝑦 2
 CECB423 Numerical Methods for Civil Engineers

using finite difference approximation

𝑢𝑖+1,𝑗 − 2𝑢𝑖,𝑗 + 𝑢𝑖−1,𝑗 𝑢𝑖,𝑗+1 − 2𝑢𝑖,𝑗 + 𝑢𝑖,𝑗−1

+ = 𝑓𝑖,𝑗
ℎ2 𝑘2

multiply both side ℎ2

ℎ2
𝑢𝑖+1,𝑗 − 2𝑢𝑖,𝑗 + 𝑢𝑖−1,𝑗 + 2 (𝑢𝑖,𝑗+1 − 2𝑢𝑖,𝑗 + 𝑢𝑖,𝑗−1 ) = ℎ2 𝑓𝑖,𝑗
𝑘

Create equation and solve using Ax=b system

Example 1:
𝛿 2𝑢 𝛿 2𝑢
+ = 𝑥𝑒 𝑦 0 <𝑥 < 2; 0<𝑦<1
𝛿𝑥 2 𝛿𝑦 2
With boundary conditions

𝑢(0, 𝑦) = 0, 𝑢(2, 𝑦) = 2𝑒 𝑦 0<𝑦<1

𝑢(𝑥, 0) = 𝑥, 𝑢(𝑥, 1) = 𝑥𝑒 𝑦 0<𝑥<2

Using finite difference method with ∆𝑥 = ℎ = 0.5 and ∆𝑦 = 𝑘 = 0.5

Solve the linear system 𝐴𝑥 = 𝑏 using Gauss Jordan method. Use three decimal places in
all calculation.
 CECB423 Numerical Methods for Civil Engineers

Approximate 𝑢1,1 , 𝑢2,1 𝑎𝑛𝑑 𝑢3,1

At point (𝑥𝑖 , 𝑦𝑗 )

𝛿 2 𝑢(𝑥𝑖 , 𝑦𝑗 ) 𝛿 2 𝑢(𝑥𝑖 , 𝑦𝑗 )
+ = 𝑥𝑖 𝑒 𝑦𝑖
𝛿𝑥 2 𝛿𝑦 2

using finite difference approximation

𝑢𝑖+1,𝑗 − 2𝑢𝑖,𝑗 + 𝑢𝑖−1,𝑗 𝑢𝑖,𝑗+1 − 2𝑢𝑖,𝑗 + 𝑢𝑖,𝑗−1

+ = 𝑥𝑖 𝑒 𝑦𝑖
ℎ2 𝑘2

multiply both side ℎ2

ℎ2
𝑢𝑖+1,𝑗 − 2𝑢𝑖,𝑗 + 𝑢𝑖−1,𝑗 + (𝑢 − 2𝑢𝑖,𝑗 + 𝑢𝑖,𝑗−1 ) = ℎ2 𝑥𝑖 𝑒 𝑦𝑖
𝑘 2 𝑖,𝑗+1

Given ℎ = 𝑘 = 0.5

0.52
𝑢𝑖+1,𝑗 − 2𝑢𝑖,𝑗 + 𝑢𝑖−1,𝑗 + (𝑢 − 2𝑢𝑖,𝑗 + 𝑢𝑖,𝑗−1 ) = 0.52 𝑥𝑖 𝑒 𝑦𝑖
0.52 𝑖,𝑗+1

𝑢𝑖+1,𝑗 − 2𝑢𝑖,𝑗 + 𝑢𝑖−1,𝑗 + 𝑢𝑖,𝑗+1 − 2𝑢𝑖,𝑗 + 𝑢𝑖,𝑗−1 = 0.52 𝑥𝑖 𝑒 𝑦𝑖

𝑢𝑖+1,𝑗 + 𝑢𝑖−1,𝑗 − 4𝑢𝑖,𝑗 + 𝑢𝑖,𝑗+1 + 𝑢𝑖,𝑗−1 = 0.52 𝑥𝑖 𝑒 𝑦𝑖

(𝑥1 , 𝑦1 ) 𝑢2,1 + 𝑢0,1 − 4𝑢1,1 + 𝑢1,2 + 𝑢1,0 = 0.52 𝑥1 𝑒 𝑦1

at i=1, j=1 𝑢2,1 + 0 − 4𝑢1,1 + 1.359 + 0.5 = 0.52 (0.5)𝑒 0.5
𝑢2,1 − 4𝑢1,1 = −1.653

(𝑥2 , 𝑦1 ) 𝑢3,1 + 𝑢1,1 − 4𝑢2,1 + 𝑢2,2 + 𝑢2,0 = 0.52 𝑥2 𝑒 𝑦1

at i=2, j=1 𝑢3,1 + 𝑢1,1 − 4𝑢2,1 + 2.718 + 1 = 0.52 (1)𝑒 0.5
𝑢3,1 + 𝑢1,1 − 4𝑢2,1 = −3.306

(𝑥3 , 𝑦1 ) 𝑢4,1 + 𝑢2,1 − 4𝑢3,1 + 𝑢3,2 + 𝑢3,0 = 0.52 𝑥3 𝑒 𝑦1

at i=3, j=1 3.297 + 𝑢2,1 − 4𝑢3,1 + 4.077 + 1.5 = 0.52 𝑥3 𝑒 𝑦1
𝑢2,1 − 4𝑢3,1 = −8.256
 CECB423 Numerical Methods for Civil Engineers

Equation (1) till (3) can be express in matrix form

−4 1 0 𝑢1,1 −1.653
𝑢
( 1 −4 1 ) ( 2,1 ) = (−3.306)
0 1 −4 𝑢3,1 −8.256

Consider the Laplace equation,

𝛿 2𝑢 𝛿 2𝑢
+ =0
𝛿𝑥 2 𝛿𝑦 2
The method to solve Laplace equation similar to solve Poisson’s equation with
𝑓(𝑥, 𝑦) = 0

Example 2:
𝛿 2𝑢 𝛿 2𝑢
+ =0 0< 𝑥 < 3; 0<𝑦<1
𝛿𝑥 2 𝛿𝑦 2
With boundary conditions

𝑢(0, 𝑦) = 0, 𝑢(3, 𝑦) = 0 0 < 𝑦 < 1

𝑢(𝑥, 0) = 𝑥(3 − 𝑥), 𝑢(𝑥, 1) = 0 0 < 𝑥 < 3

Using finite difference method with ∆𝑥 = ℎ = 1 and ∆𝑦 = 𝑘 = 1/3

Solve the linear system 𝐴𝑥 = 𝑏 using Gauss Jordan method. Use three decimal places in
all calculation.