100% found this document useful (2 votes)
75 views28 pages

Finite Difference Method

The document discusses different finite difference methods for approximating derivatives including forward, backward, and central difference formulas. The forward difference formula approximates the derivative at a point using the function value at a point ahead. The backward difference formula uses the point behind. The central difference formula uses points on both sides to get a second-order accurate approximation. Expressions for the derivative operator D are derived in terms of the displacement operator E for each method. Approximations keeping higher-order terms provide increased accuracy but require more points. The central difference formula results in an expression involving sinh and is second-order accurate.

Uploaded by

ramiz
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
100% found this document useful (2 votes)
75 views28 pages

Finite Difference Method

The document discusses different finite difference methods for approximating derivatives including forward, backward, and central difference formulas. The forward difference formula approximates the derivative at a point using the function value at a point ahead. The backward difference formula uses the point behind. The central difference formula uses points on both sides to get a second-order accurate approximation. Expressions for the derivative operator D are derived in terms of the displacement operator E for each method. Approximations keeping higher-order terms provide increased accuracy but require more points. The central difference formula results in an expression involving sinh and is second-order accurate.

Uploaded by

ramiz
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 28

Finite Difference Method

Forward difference formula :- X+


2 2 3
The Taylor series expansion , u ( x  x )  u ( x)  x
u ( x )  x  u ( x ) x u ( x)
 .  .  .......
x 2! x 2 3! x3
May be written in terms of displacement operator, E and derivative operator, D as follows:
x 2 2 x 3 3 Where, D

Eu ( x )  [1  xD  .D  .D  ...]u ( x ) .......(1) x
2! 3!
x  x x  x
x
E=displacement operator i,e it will displace the function by next nodal value i.e. at node
x the function u has value of u(x) and Eu(x) has value of u( x   x) is the value
which
of u at the node ( x). x
From equation(1) : ui  u  xi 

xD
ui 1  u  xi 1   u ( xi  x )
Eu ( x )  e u ( x)
Eui  u ( xi 1 )  u ( xi  x)
xD
therefore E  e therefore Eui  ui 1
1 Eui 1  ui  2 etc from the defintion of E
D ln E
x therefore E 2ui  ui  2
 E nui  ui  n
Finite Difference Method continued
  ui  ui 1  ui  Eui  ui  ( E  1)ui
    E 1  E=1+ 
1
therefore D  ln E
x
1
or , D(u i )= ln E (ui )
x
 u  1
or ,    ln(1    )(ui )
 x i x
2
 u  1     3  4  ....  A 
         ...  ui
 x i x  2 3 4 
 E  1  E  1  E  1
2 3 4
 u  1  
    E  1     ... ui   ( B)
 x i x  2 3 4 
Eui  ui 1 , Eui 1  E 2ui  ui  2 ;.......E nui  ui  n
For 1st order accuracy we will consider only one term on RHS of Equation (B)
 u  1 Eui  ui ui 1  ui
   [ E  1]ui  
x
 i x  x x
Finite Difference Method continued
Backward Difference Formula : u ( x) x 2 u ( x) x3 u ( x)
u ( x  x)  u ( x)  x    ...
x 2! x 2 3! x 2
     
2 3 4
1
  x.D  xD xD 
E u ( x)   1  x.D     ....  u ( x)
 2! 3! 4! 
 
 xD   xD   xD 
2 3 4

 E 1  1  x.D     ....
2! 3! 4!
=e xD
 -xD  ln E 1
1
 D=- ln( E 1 ) ... C 
x
Again,

E 1ui  ui 1 ; E 1ui 1  E 2ui  ui 2 ...........E  nui  ui n

 ui  ui  ui 1  ui  E 1ui  1  E 1  ui

   1  E 1  E 1  1   
Finite Difference Method
c continued
1 1
D ln(1    );  Dui   ln(1    )(ui )
x x
 u  1   (  ) 2 (  )3 (  ) 4 
            ui    (C )
 x i x  2 3 4 

   
 1 2 1 3 
 u  1  1  E 1 E
  
 x i x 
1 
 E 1
  2 3
    ui    ( D)

 
1ST order accuracy :- Only 1st term of equation (A) or equation (B) is considered.
 u  1 1 ui  E 1ui ui  ui 1
   [1  E ]ui  
 x i x x x
2nd order accuracy:- upto 2nd term of equation (A) or equation (B) is considered.

 u  1 
 1  E 
1 2 

  
 x i x 
1 E  
1

2
 ui

 
 u  3ui  4ui 1  ui  2 x 2  3 u
   
 x i 2x 3 x 3
Finite Difference cMethod continued
Central difference Formula 1 1
E 2 ui u 1 ; E 2 ui u 1
i i
2 2

 1 1 
 ui  u 1  u 1   E  E 2  ui
2
i i  
2 2  
1 1
  E2  E2
1 1
E2  E2
Using the definition of mean,  
2
from forward difference we get, E  ex.D
1
E 2  ex.D /2
1
From backward difference , E2  ex.D / 2
1
E 2  ex.D /2
e x  e x
  E 1/2
E 1/2
e xD /2
e xD/2
 2sinh xD / 2  using sinh x 
2
c
Finite Difference Method continued

xD / 2  sinh 1
2
1
D
x
 2sinh 1
 /2 
 u  1   3 3 5 5 7 
         ........  ui
 x i x  24 640 7168 
 5 7 
2 2    2 2 2   
 3 1 .3 .  1 .3 .5   
 u  2  1 2
 / 2  2
   2
  ......    ( E )
    / 2  
 x i x  3! 5! 7! 
 
 

 
 3 
1/ 2 1/2
E  E
 u 
 
 x i

x 

1  1/2
E  E 1/2
  24
      u ....................( F )
 i
 
2nd order accuracy with 1st term only:
 u  1  1/2 1 1
 u
1/2  1/2 1/2 3
   E u  E u  . E  E
 x i x   x 24
i i i

e 
3
xD /2
ui 1/2  ui 1/2  e xD /2
  ui
x 24.x
ui 1/2  ui 1/2 x 2 D 3ui
 
x 24 24
ui 1/2  ui 1/2 x 2  3u
 
x 24 x 3
E1/2  E 1/2
To avoid the half integer mesh point, the definition of mean may be employed.where  =
2
2 2
  1  E1/ 2  E 1/2   1
 1/2
 E 1/2 
  
2  2  E 4
  

  1  2 4
2
  1 2 /4
Now from equation F we have
 u    3 5 7 
     3  5 ......
    
 x  24 640 7168
 i x 1 2 /4  

1
 2 2  3 5 7 
 1        3  5 ......
x  4   24 640 7168 
   
   2 3 4 15 6   3 3 5 5 7 
 1     ...      ......   ui
x  8 816 4864  24 640 7168




   3 3. 5 5. 7    3  5 3. 7   3. 5 3. 7   15. 7 


    ...     ...   ...     ...   ui
x  24 640 7168   8 824 8640   816 81624   4864  
       
 
   1 1 5  3 1 3  7 5 3 3 15  
   ³                  ui
x   24 8   640 8  24 8  16   7168 8  640 8  16  24 48  64  
   3 12  22 5 50  42  70  350 7 
       ui
x  3! 5! 71680 
 u     3 12  22 5 12.22.32 7 
          .... ui ......(G )
 x i x  3! 5! 7! 
2nd order accuracy(with one term from R.H.S. of Equation (G):-
u   3
  ui  . ui 
x i x x 3!
 12 1
 
1 1
 

1 1  E  E 2  E 2  E 2 
  2     u
  E  E 2  ui  i
x   2  3!  x
ui 1  ui 1 x 2  3 u
 
2x 6 x 3
Higher order derivative by Finite Difference Method
Backward difference:-
u 1
  ln 1     ui
x i x
nu 1
 n   n ln 1    ui
 n

x i x
2 3 4 n
1      
 n      ....  ui
x  2 3 4 
nu 1   n n n 1 n 3n  5   n  2 n  n  2  
 n  n        ......(2)
x x  2 24 48 

 1 1 
Where,   1  E  E  1 
Higher order derivative by Finite Difference Method
Forward Difference
u 1
 ln 1     ui ...... First order forward difference 
x i x
nu 1 
 n ln 1     ui
n

n

 taking nth power on both the sides 
x i x
n
 n u 1    2  3  4 
n
 n 
     ....  ui
x x  2 3 4 
1  n  
2
nu   n 1   2
 3
 4
 n n  1  n  2   2
 
n
 n   n.    ...      ....  
x i x   2 3 4  2!  2  

1  n   n 1  
2
   n 1 
3
  n 1 
4
n  n  1  n  2    4  6  8  
 n   n.     n. .  n. .  .....       ...   ... ui
x   2  3 4 2!  4 9 16  
1   n n  n 1 n 3n  5   n  2  n  n  2  n  3  n  3 
 n         .... ui       1
x  2 24 48 
Where,    E  1  E=1+ 
Higher order derivative by Finite Difference Method
Central difference for n =even
u 1  1  
  2sin h  ui
x x  2
n 3 5 7 n
 u 1   3 5 
 n  n   24  640  7168  ...... ui
x x  
n  n 2 n  22  5n  4 n  5 n  1  
 n
x 1  24   64  90    45  7  5   .....(3)
    
 1 1 
Where,  ui  u 1  u 1   E 2  E 2  ui
i i  
2 2  
1 1
  E2  E2
Higher order derivative by Finite Difference Method
Central difference for n=odd

n
n
u  2 1  
n  
2 x
 sin  ui
x   2
1
4
n  
 n  3 2 5n2  52n  135 4
  n 1       ui   4
x  24 5760 
1 1
E2 E2
1 1
Where, mean,   and  E E
2 2
2
Finite Difference Method

For mixed derivatives :


Using Equation either (1) or (2) or(3) or(4) depending on the Backward,
or Forward or Central Diff (n=even) or Central Diff (n=odd) respectively.
e.g. for central difference and n=odd.
  2u  1    2
 4
    y
2
 4

    x x 1    O(x )   y y 1    O(y )  ui , j
x

 x y i , j x y 6  6  
   
Finite Difference Method: SUMMERY
 nu 1  n n   n 1 n 3n  5    n  2  n  n  2  n  3    n  3 
1.          .... ui  1 Forward
x n x n
 2 24 48 
Where,    E  1  E=1+ 
nu 1   n n   n 1 n 3n  5    n  2  n  n  2  
2.         ......(2)Backward
x n x n  2 24 48 
Where,    1  E 1  E 1  1   
 nu n  n 2 n  22  5n  4 n  5 n  1  
3.  1 
 24      5    .....(3) Central Diff, fro n= even
x n
x n  64  90  4 7 5  
1 1
  E  E 2 2

 nu n  n  3 2 5n 2  52n  135 4 
4. n
 1        ui    4  Central Diff, fro n= odd
x x n  24 5760 
 12 1
 1 1
Where, mean,    E  E  / 2 and
2
  E E 2
2

 
5. For mixed derivatives :
Using Equation either (1) or (2) or(3) or(4) depending on the Backward,
or Forward or Central Diff (n=even) or Central Diff (n=odd) respectively.
e.g. for central difference and n=odd.
  2u  1   x2  4
   y2  4

    x x  1    O( x )   y  y  1    O(y )  ui , j
 x y i , j  x y  6    6  
Finite Difference Method

u ui1, j  ui, j u ui , j 1  ui , j
Forward Difference, = and =
x x y y
u ui , j  ui1, j u ui , j  ui , j 1
Backward Difference, = and =
x x y y
u ui 1, j  ui 1, j u ui1, j  ui 1, j
Central Difference, = and =
x 2 x y 2y
 2u ui2, j  2ui1, j  ui, j
Forward Difference, =
x 2 x 2
 2u ui, j2  2ui, j1  ui, j
and =
y 2 y 2

Backward Difference,
 2u ui, j  2ui1, j  ui2, j
=
x2 x 2
 2u u  2u  u
i, j i , j 1 i , j 2
and =
y 2 y 2

2u u
 i 1, j  2ui , j  ui1, j
Central Difference, =
x2 x2
2u u
 i , j 1  2ui , j  ui , j 1
and =
y 2 y 2
Problem -1: 2-d heat diffusion
T=0K

T=0K 20 cm X 20 cm T=0K

T=400K General Governing Equation is :


 k T   0
For 2  D, Governing Equation is :
d  dT  d  dT 
k   k   0...( A)
dx  dx  dy  dy 
Problem -2: 2-d heat diffusion
T=0 K

T=300K 20 cm X 20 cm T=0K

T=400K
General Governing Equation is :
  k T   0
For 2  D, Governing Equation is :
d  dT  d  dT 
k  k  0...( A)
dx  dx  dy  dy 
Problem -3: 2-d heat diffusion
T=500K

T=300K 20 cm X 20 cm T=200K

T=400K
General Governing Equation is :
  k T   0
For 2  D, Governing Equation is :
d  dT  d  dT 
k  k  0...( A)
dx  dx  dy  dy 
Problem -6.5: 2-d heat diffusion
T
k  h(T  T f ); (Convective heat transfer )
y

T=T0 20 cm X 20 cm
T
k  q, ( Heat flux  q )
x

T k=205 W/m-K
 0, ( Insulated ; i.e. flux, q  0) q=50 w/m2
x
General Governing Equation is : T0=100oC
  k T   0 Fluid Temp,Tf=50oC
For 2  D, Governing Equation is : (Bi), Biot No.=hL/k=2
d  dT  d  dT 
k  k  0...( A)
dx  dx  dy  dy 
Problem -4: 2-d heat diffusion
Non-dimensionalising the governing equation and all the BCs with the following:
x y T  Tf
x*  ; y*  ,  We get the following
L L T0  T f
Governing Equation is :
 d 2   d 2 
 
2   2 
 0...(1)
 dx *   dy * 
Left Boundary , x*  0;   1
 qL
Right Boundary , x*  1; 
x * k (T0  T f )

Bottom Boundary , y*  0; 0
y *

Top Boundary , y*  1;   Bi *
y *
Problem -6.4: 1-d Conduction-Convection
Tf
scr_conduction_convection.m
T=Tw L T
at x  L,  0 (Adiabatic condition)
x
For steady state 1-d conduction-convection without source is:
 2T hP
2
 (T  T f )  0 (1)
x kA
Boundary conditions are:
T
At x=0, T=Tw and at x= L ,  0 (2)
x
 2
Nondimensional form are: 2
 Bi b   0 (1a)
x *

At x*=0,   1, and at x*= 1 , *  0 (2a)
x
hL PL AP
Where, Bi  ( Biot no.); b   ( Area ratio), P  Perimeter ,
k A Ac
(1) Take Bi  0.3, 0.4, 0.5; b  3; solve temperature field .
(2) Take Bi  0.3; b  3, 4,5 ; solve temperature field .
Problem -6: 1-d UNSTEADY
Unsteady_1D-Heatflow.m
T=Tw T=Ti T=Tw
L
For 1-d transient heat transfer without source is:
 2T  cP T
2
 (1)
x k t
Boundary conditions are:
At x=0, T=Tw and at x= L , T=Tw (2)
Initial Condition i.e. at t=0, T=Ti i.e. at all the nodes except boundaries.
Non-dimensionalising the governing equation and all the BCs and ICs with the following:
x y T  Tw t
x*  ; y*  ,   and  = 2 , where  = Thermal difusivity;We get the following
L L Ti  Tw L

 2 
Nondimensional form are: 2
 (1a)
x * 
At x*=0,   0 and at x*= 1 ,   0 (2a)
And Initial Condition i.e. at  =0,   1
Continueed

By discretising the governing equation using FTCS scheme we get,


 in 1   in  in11  2*in 1  in11

 x 2
n 1 n 1 n 1 n 
   i 1  (1  2 ) i    i 1   where   2
i
x
n 1 n
1 0 0 0 .............0  1  0 
   1  2   ..........0     
   2   2 
 0   1  2   ....0   3  3 
     
......................................  .  . 
 0 0.....   1  2     .   nx 1 
     
 0 0......................0 1   nx  0 
Thomas Algorithm forTriDiagonalMatrixalgorithm(TDMA)

BTi n1,
1
j  DTi , j
n 1
 ATi n1,
1
j  Ci
n

n 1 n 
 T1, j   0 
  
1 0 0 0 . . . . 0 0   
   T2, j   C2, j  
0 D A 0 . . . . 0 0     
0  T3, j  C
B D A 0 . . . 0 0  
 3, j  
    
0 0 B D A 0 . . . .  T4, j  C
 4, j  
0  
0 0 B D A 0 . . .   . 

.  

 . 
   
. . 0 0 B . A 0 . .  .  
     
. . 0 0 0 . . A 0 0  .  
 .  
0 . 0 0 0 0 . . A 0   C 
T
 N 1, j  N  1, j 
    
0 . 0 0 0 . 0 B D A T  C  
N , j N , j
0
 0 0 0 0 . . 0 0 1 ( N 1) X ( N 1)     
TN 1, j   1  
  ( N 1,1) N 1,1

Thomas Algorithm forTriDiagonalMatrixalgorithm(TDMA)

d x ax
1 1 1 2
c 1
.................  A.1
b x d x a x
2 1 2 2 2 3
c
2
..............A.2
b x d x a x
3 2 3 3 3 4
c
3
.................A.3
b x d x a x
4 3 4 4
c
4 5 4

........................................................................
bN-1 x  d x  a x  c
N -2 N -1 N -1 N -1 N N -1
..............A.4 
b x d x
N N 1 N N
c N
...............A.5
 A.2   d   A.1  b
1 2
 db x d d x d a x c d
1 2 1 1 2 2 1 2 3 2 1

- b d x  ab x  cb
2 1 1 1 2 2 1 2

................................................................ ...
d d  a b  x  d a x  c d  c b
1 2 1 2 2 1 2 3 2 1 1 2

d 2
 b a / d 1 x  a x   c  b c / d
2 1 2 2 3 2 2 1 1 
 d ' x  a x  c'
2 2 2 3 2
........................  A.6 
d '  d - b a / d , c'  c -b c / d
2 2 2 1 1 2 2 2 1 1

d '2 x2  a2 x3  c '2 .............. A.6


By  A.3 d ' - b A.6 we get
2 3

d '2 b3 x2  d '2 d3 x3  d '2 a3 x4  d '2 c3



  d '2 b3 x2  b3a2 x3  b3c ' 
...............................................................................
 
d3  b3a2 / d '2 x3  a3 x4  c3  b3c '2 / d '2  
d '3 x3  a3 x4  c '3 .......................... A.7 
The pattern which is developing is-
(A.6) can be obtained from (A.2) by dropping
first term, replacing d 2 by d - b a / d   d '  keeping third term unchanged and replacing
2 2 1 1 2

c by c - b c / d  c ' .
2 2 2 1 1 2

 
Similarly,(A.7) is obtained from(A.3) bdropping first term, replacing d3 by d3 - b3 a2 / d2  d '3
Keeping third term unchanged and by replacing c by c - b c ' / d '   c ' 
3 3 3 2 2 3

So if we generalize the pattern


d 'i  di - ba
i i-1 / d 'i-1 where i  2,3..............N

& c 'i  ci - bc
i 'i-1 / d 'i-1 where i  2,3..............N
for i=3 to N
c(i  1,1) b(i, i  1)
c '(i,1)  c(i,1) 
d '(i  1, i  1)
a(i  1, i ) b(i, i  1)
d '(i, i )  d (i, i ) 
d '(i  1, i  1)

n 1
1 0 0 0 . . . . 0 0   T1   0 
     
0 d 22 a23 0 . . . . . 0   T2   C 21 
' '
0 0 d33 a34 0 . . . . 0   T3   C 31 
 '     ' 
0 0 0 d 44 a45 0 . . . .  T
 4   C 41 
0 . 0 0 . . 0 . . .   T5   . 
     
. . . 0 0 . . 0 . .   .   . 
. . . . 0 0 . . 0 .   .   . 
     
0 . . . . 0 0 . aN 1, N .  T
 N 1   . 
0 . . . . '
. 0 0 d NN aN , N 1  T  C 
  N   N ,1 
 0 0 0 . . . . . 0 1  N 1XN 1 TN 1  N 1 X 1
 1 

a(i, i  1).T (i  1,1)


By Back substitution: T (i,1)  c(i,1)  for i=N to 2.
d (i, i)

You might also like