You are on page 1of 3

Consider the numerical solution for Advection equation,

T
T
U
0
t
x

Where U > 0
a) We use Taylor series and truncate the higher order derivatives.
Forward Difference in time, is given as,
n
T jn 1 T jn
T

t
t j

Central Difference in space is given as,


T j 1 T j 1
T


2 x
x j
n

Substituting in the Advection equation, we get,

T jn 1 T jn

T
T
U
0
t
x
T jn1 T jn1
U
0

2x

U t n
T j 1 T jn1
2x
c
T jn 1 T jn T jn1 T jn1
2
T jn 1 T jn

where

U t
x

b) Now, by Fourier series, error is given as,


i kx

nj G n e j

Substituting in the Finite Difference equation, we get,

G n 1eikx G n eikx

c n ik x x
ik x x
Ge
G ne
2

G n 1
c
1 eikx e ikx
n
G
2

i sin kx
We know that,
get

eikx e ikx
2

and

G n 1
G
Gn

. Substituting in above equation, we

G 1 ic sin kx

c) For stability, amplification factor G should always be less than or equal to 1.


G n 1
1
Gn
1 ic sin kx 1

ic sin kx 0
c0
U , t , x 0

But we know that

. Hence,

c0

d) Applying Backward difference in space


n
T jn T jn1
T


x
x j

Substituting in the Advection equation, we get,

T jn 1 T jn

T
T
U
0
t
x
T jn T jn1
0
x
U t n
T j T jn1

x
n 1
n
T j T j c T jn T jn1
T jn 1 T jn

where

U t
x

e) Now, by Fourier series, error is given as,


i kx

nj G n e j

Substituting in the Finite Difference equation, we get,

G n 1eikx G n eikx c G n eikx G n eik x x


G n 1
1 c 1 e ikx
n
G

G 1 c 1 eikx

G 1 c 1 e ikx
2

1 2c 1 eikx c 2 1 eikx

G 1 2c 1 c 1 cos kx
2

For Stability
2

G 1
1 2c 1 c 1 cos kx 1

c 1 c 1 cos kx 0

This implies that

c0
And

1 c 0
c 1
c 0,1
Hence this method is stable for

You might also like