Partial Differential Equations:

Basic Concepts:
An equation involving one or more partial
derivatives of an (unknown) function of two or more
independent variables is called a partial differential
equation. The order of the highest derivative is called
the order of the equation.

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Just as in the case of an ordinary differential equation,
we say that a partial differential equation is linear if it is
of the first degree in the dependent variable (the
unknown function) and its partial derivatives. If each
term of such term of such an equation contains either
the dependent variable or one of its derivatives, the
equation is said to be homogeneous; otherwise it is said
to be nonhomogeneous.

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Example :
Im por tan t linear partial differenti al equation
of the sec ond order :
 2u 2  u
2
(1) 2  c One  dim ensional wave Eq.
t x 2
u  2
u
( 2)  c2 2 One  dim ensional heat Eq.
t x
 2u  2 u
(3) 2  2  0 Two  dim ensional Laplace Eq.
x y
 2u  2u
(4) 2  2  f ( x, y ) Two  dim ensional Poisson Eq.
x y
 2u  2
u  2
u
(5) 2  c 2 ( 2  2 ) Two  dim ensional wave Eq.
t x y
 2u  2 u  2u
(6)  2  2  0 Three  dimEngineering
Tikrit University-Civil ensional Laplace Eq.
x 2
y z Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Here (c) is a constant; (t) is a time; x, y, z are the

Cartesian coordinates and the dimension is the

number of these coordinates in the equation.

Equation (4) [with f(x,y)≠0] is nonhomogeneous,

while the other equations are homogeneous.

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
MODELING:
(1)Vibration of elastic string (wave equation):
Assumptions;
 Vibration takes place in x-y plane.
 No elongation.
 The string can transmit force only in the
direction of its length.
 Constant tension force.
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 y

0
x+x
L x
x

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 Fx  0
T2 cos (   )  T1 cos   0
 is too small   0
T2  T1  0
T2  T1  T
 Fy  0
 y 2
T1 sin   T2 sin (   )   s ( 2 )  0
t
  density per length
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
  y 2
T1 sin   T2 sin (   )   s ( 2 )  0
t
sin    
  is too small
sin (   )     
T1  T2  T
2 y
T   T (   )    s 2
t
2 y
T   s 2
t
   y 2

s T t 2
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan

 curvature
s
 y
2

x 2  2
y
  2
dy 2 2
3
x
[1  ( ) ]
dx
2 y  2 y
 2 
x T t 2

a 2

T
2 y  2
y
a 2
One  dim ensional wave Eq.
x 2
t 2 Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
2 y  2
y
a 2
(1)
x 2
t 2
By separation the var iables :
yX T
X  function of x
T  function of t
2 y 
 X "T 
 x 2 
 2  Subst. int o Eq. (1)
 y
 X T "
t 2 
X " T  a2 X T"
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
X " T  a2 X T"
X" 2 T"
a
X T
 k 2
X " T" 
2
  Cons tan t   0
a X T  k2

X " T"
(1) 2   k 2

a X T
2
d X
X " a k X  0 
2 2
2
 a k X 0
2 2

dx
X  c1 cos kax  c2 sin kax
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 T"
 k 2
T
2
d T
T " k T  0  2  k 2T  0
2

dt
T  c3 cos k t  c4 sin k t
 y ( x, t )  X .T
 (c1 cos k a x  c2 sin k a x)(c3 cos k t  c4 sin k t )

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
(2) Cons tan t  0
X " T"
2
 0
a X T
X"
2
 0  X "  0  X '  c1  X  c1 x  c2
a X
T"
 0  T "  0  T '  c3  T  c3t  c4
T
 y ( x, t )  (c1 x  c2 ) (c3t  c4 )

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
(3) Cons tan t  k 2
X " T"
2
 k 2

a X T
2
d X
X " a k X  0 
2 2
2
 a k X 0
2 2

dx
X  c1e k a x  c2 e  k a x
T"
 k2
T
2
d T
T " k T  0  2  k 2T  0
2

dx
T  c3e k t  c4 e  k t
e  k a x )(Engineering
 y ( x, t )  (c1e k a x Tikritc2University-Civil c3e k t  c4 e  k t )
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
For the string fixed at the both ends
and subjected to the following :
Initial condition :
x
y ( x,0)  yo sin
L
y 
( x,0)  0
t 
y ( x, t )  (c1 cos k a x  c2 sin k a x)(c3 cos k t  c4 sin k t )

L
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
y ( x, t )  (c1 cos k a x  c2 sin k a x)(c3 cos k t  c4 sin k t )
y (0, t )  0 & y ( L, t )  0
0  (c1 cos 0  c2 sin 0)(c3 cos k t  c4 sin k t )
c1 (c3 cos k t  c4 sin k t )  0
Either ; c1  0
Or; (c3 cos k t  c4 sin k t )  0
 y ( x, t )  c2 sin k a x(c3 cos k t  c4 sin k t )
Or; y ( x, t )  sin k a x( A cos k t  B sin k t )
Such that ; c2 * c3  A & c2 * c4  B
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
y ( x, t )  sin k a x( A cos k t  B sin k t )
y ( L, t )  0
0  sin k a L( A cos k t  B sin k t )
n
sin k a L  0  k a L  n    k 
aL

y ( x, t )   sin k a x( An cos k t  Bn sin k t )
n 1

n x n n
y ( x, t )   sin ( An cos t  Bn sin t)
n 1 L aL aL
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Initial conditions :
x
t  0  y ( x, t )  yo sin
L
x n x

yo sin   sin ( An cos 0  Bn sin 0 )
L n 1 L
x 
n x
yo sin  An sin
L n 1 L
n x
L
2
An   f ( x) sin dx
L0 L
2
L
x n x
An   yo sin sin dx
L0 L L
2
An  yo L  2 yTikrit
o University-Civil Engineering
L Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
y
( x,0)  0
t

n x n n
y ( x, t )   sin ( An cos t  Bn sin t)
n 1 L aL aL
y  n x n n n n
  sin ( An sin t Bn cos t)
t n1 L aL aL aL aL
y
At t  0 ; 0
t

n x n n
0   sin ( An sin 0  Bn cos 0 )
n 1 L aL aL

n x n
0   sin ( Bn )  Bn  0
n 1 L aL

n x n
 y ( x, t )   sin Tikrit2University-Civil
yo cos Engineering t
n 1 L
Department a L Num.
Third Stage Eng.Anal.&
Meth. Dr.Adnan Jayed Zedan
(2) One  dim ensional consolidat ion :
 2u  2
u
a 2

z 2
t 2
u ( z , t )  Z z Tt
 2u
 Z" T
z 2

u
 Z T
t
Z" T  a2 Z T 
 k 2
Z  T  
2
  Cons tan t   0
a Z T  Engineering
2
Tikrit University-Civil
Department Third Stage
k
Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 Z"
  k 2

a2Z
2
d Z
Z " a k Z  0  2  a 2 k 2 Z  0
2 2

dx
Z  c1 cos k a z  c2 sin k a z
T
 k 2
T
dT
T  k T  0 
2
 k 2T  0
dt
 k 2t
T  c3e
 u ( z , t )  Z .T
 k 2t
 (c1 cos kTikritaUniversity-Civil
z  c2 sin k a z ) c3 e
Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
  k 2t
u ( z , t )  c3 e (c1 cos k a z  c2 sin k a z )
Let A  c1 * c2 & B  c1 * c3
 k 2t
u ( z , t )  e ( A cos k a z  B sin k a z ) general Sol.
Boundary Conditions :
u (0, t )  0
u ( L, t )  0
Initial Condition :
u ( z ,0)  f ( z ) given
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 A clay layer :
u ( z ,0)  z ( L  z )
 k 2t
u ( z, t )  e ( A cos k a z  B sin k a z )
(1) u (o, t )  0
 k 2t
0e ( A cos 0  B sin 0)
 k 2t
0e A  A0
(1) u ( L, t )  0
 k 2t
0e B sin k a L
B0
n
 sin k a L  0  k a L  n  k 
aL
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 
u ( z , t )   B sin k a z e  k 2t

n 1

n
  Bn sin z e  k 2t

n 1 L
u ( z ,0)  z ( L  z )

n
z ( L  z )   Bn sin z e0
n 1 L

n
z ( L  z )   Bn sin z
n 1 L
n z
L
2
Bn   z ( L  z ) sin dz
L0 L


Tikrit University-Civil Engineering
H. W. Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Solution of Partial Differential Equations:

Here under we consider the simple

examples, the solution of which depends

up to the meaning of partial

differentiation.
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Example (1) :
z
Solve ; (i )  0
x
 z
2
(ii ) 2  0
y
 z
3
(iii ) 2  0
x y
Noting that z is a function of x & y.
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Solution :
z
(i )  0 (a)
x
Here z is a function of two independen t var iables
x and y. On int egrating Eq. (a), we get that solution as;
z  function independen t of x.
 z   ( y ) which is arbitrary.

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 2 z
(ii ) 2  0
y
z is a function x and y.
Integratin g with respect to y, we get ;
z
 a function independen t of y
y
  ( x) (...the other var iable is x only )
Again int egrating with respect to y, we get ;
z    ( x) dy  a function independen t of y

  ( x)  dy  f ( x)
 z   ( x) y  f ( x) where  ( x) and f ( x) are
arbitrary in the solution
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 3 z
(iii ) 2  0
x y
 z
3
  z2
 { 2}  0
x y y x
2

On int egrating with respect to y, we get ;

2 z
 a function independen t of y
x 2

  ( x) ( say)
 2 z  z
Now; 2  { }   ( x)
x x x
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
  2 z  z
Now; 2  { }   ( x)
x x x
On int egrating with respect to x, we get ;
z
   ( x) dx  a function independen t of x
x
   ( x)dx  q( y )
Again int egrating with respect to x, we get ;
 z  [   ( x)dx] dx   q( y ) dx  F ( y )
Or :
z     ( x)dxdx  x q( y )  F ( y ) which is required solution

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Example (2) : Solve the partial differenti al equations
in the following case;
2 z
(i ) 2  sin x
x
 2u
(ii )  e  t cos x given that u  0 when t  0
x t
u
and  0 when x  0
t
2 z z
(iii )  sin x sin y for which  2 sin y when x  0
xy y

and z  0 when y is an odd multiply of
2
2 z 1
(iv )  given that z  y ln y and
xy x  y
z
 1  ln y when x  0
DepartmentThird
Tikrit University-Civil Engineering
y Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Solution :
2 z
(i ) 2  sin x
x
Integratin g w.r.t. x, we get ;
z
  sin x dx  f ( y )   cos x  f ( y )
x
Again int egrating w.r.t. x, we get ;
z    cos x dx   f ( y ) dx  g ( y )
  sin x  x f ( y )  g ( y )
where f ( y ) and g ( y ) are arbitrary functions .
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
  2u  u  0 when t  0
(ii )  e  t cos x  u  0 when x  0
xt  t
Integratin g partially w.r.t. x, we get ;
u
  e  t cos x dx  a function independen t of ( x)
t
 e  t  cos x dx  f (t ) ( say)
u
  e  t sin x  f (t )
t
u
x0 0
t
0  e  t sin 0  f (t )  f (t )  0
u
  e  t sin x
t Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 u
 e  t sin x
t
Now int egrating partially w.r.t. t , we get ;
u ( x, t )   e  t sin x dt  a function independen t of (t )

 sin x  e  t dt  g ( x) ( say)
u ( x, t )  e  t sin x  g ( x)
t 0u0
0  e sin x  g ( x)  f ( x)  sin x
0

 u ( x, t )  e  t sin x  sin x  sin x (1  e  t )

Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 2 z z
(iii )  sin x sin y for which  2 sin y when x  0
xy y

and z  0 when y is an odd multiply of
2
Integratin g partially w.r.t. x, we get ;
z
 sin y  sin x dx  f ( y )
y
 sin y ( cos x)  f ( y )
z
x0  2 sin y
y
 2 sin y  sin y ( cos 0)  f ( y )
 2 sin y  sin y (1)  f ( y )  f ( y )   sin y
z
   sin y cos x  sin y   sin y (cos x  1)
y Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 z
  (cos x  1) sin y
y
Now int egrating partially w.r.t. y, we get ;
z  (cos x  1)  sin y dy  g ( x)
 (cos x  1) cos y  g ( x)

z  0  y  odd *
2
0  (cos x  1) (0)  g ( x)  g ( x)  0
 z  (cos x  1) cos y

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 2 z 1
(iv )  given that z  y ln y and
xy x  y
z
 1  ln y when x  0
y
Integratin g partially w.r.t. x, we get ;
z 1
 dx  f ( y )
y x y
 ln ( x  y )  f ( y )
z
x0  1  ln y
y
1  ln y  ln (0  y )  f ( y )
 f ( y )  1  ln y  ln y  1
z
  ln ( x  y )  1
y Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 z
 ln ( x  y )  1
y
Now int egrating partially w.r.t. y, we get ;
z   ln ( x  y ) dy  1dy g ( x)

 udv  uv   vdu ln( x  y)  u and dy  dv)

1
z  ln ( x  y ) y   ( y )dy  y g ( x)
x y
x
 y ln ( x  y )   (1  ) dy  y  g ( x)
x y
dy
 y ln ( x  y )  1dy  x   y  g ( x)
Tikrit University-Civil x y
Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 dy
z  y ln ( x  y )  1dy  x   y  g ( x)
x y
 y ln ( x  y )  y  x ln( x  y )  y  g ( x)
 y ln ( x  y )  x ln( x  y )  g ( x)
 z ( x, y )  ( x  y ) ln ( x  y )  g ( x)
x  0  z  y ln y
y ln y  (0  y ) ln (0  y )  g ( x)  g ( x)  0
 z  ( x  y ) ln ( x  y )
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Numerical Solution of Partial Differenti al Equations :
The general sec ond  order linear partial
differenti al equation is of the form :
 2u  2u  2u u u
A 2 B C 2  D E  Fu  G
x xy y x y
Which can be written as :
Au xx  Bu xy  Cu yy  Dux  Eu y  Fu  G ...........(1)
Where A, B, C , ......, G are all functions of x & y

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
au xx  2bu xy  cu yy  F ( x, y, u , u x , u y )
u is the unknown function .
Equations of form (1) can be classified with respect
to the sign of the discri min ant :
s  B 2  4 AC
in the following ways :
(1) If s  0 at a po int in the ( x, y ) plane, the equation
is said to be elliptic type.
(2) If s  0 at that po int is said to be hyperbolic type.
(3) Parabolic type when s  0 .

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Elliptic type 4ac  b 2  0 Laplace equation
Parabolic type 4ac  b 2  0 Heat equation
Heperbolic type 4ac  b 2  0 Wave equation
In the following , we will restrict our solves to three
simple particular cases of Eq. (1); namely :
u xx  u yy  0 ( the Laplace equation )
1
u xx  2 utt  0 ( the wave equation )
c
u xx  ut  0 ( the heat conduction equation )
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Finite  Difference Approximations to derivatives :
Let the ( x, y ) plane be divided int o a network of
rec tan gles of sides x  h and y  k by drawing
the sets of lines :
x  ih ; i  0, 1, 2, 3, ......
y  jk ; j  0, 1, 2, 3, ......
The po int s of int er sec tion of these families of lines are
called mesh po int s, lattice po int s or grid po int s.

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Similarly we have the approximations :
ui , j 1  ui , j
uy   0 (k ) Forward differenc
k
ui , j  ui 1, j
uy   0 ( h) Backward differenc
h
ui 1, j  ui 1, j
ux   0 (h 2 ) Central differenc
2h
and ;
ui 1, j  2ui , j  ui 1, j
u xx  2
 0 (h ) 2

h
Where ui , j  u (ih, jk )  u ( x, y )
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Similarly we have the approximations :
ui , j 1  ui , j
uy   0 (k ) Forward differenc
k
ui , j  ui , j 1
uy   0 (k ) Backward differenc
k
ui , j 1  ui , j 1
uy   0 (k 2 ) Central differenc
2k
and ;
ui , j 1  2ui , j  ui , j 1
u yy   0 (k 2 )
k2
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
We can now obtain the finite-difference
analogues of partial differential equations by
replacing the derivatives in any equation by
their corresponding difference approximation
given above.

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Thus, the Laplace equation in two dim ension, namely;
u xx  u yy  0
has its finite  difference ana log ue;
1 1
2
[ui 1, j  2ui , j  ui 1, j ]  2 [ui , j 1  2ui , j  ui , j 1 ]  0
h k
If h  k , this gives;
1
ui , j  [ui 1, j  ui 1, j  ui , j 1  ui , j 1 ] (a)
4
Which shows that the value of u at any po int is the mean
of its values at the four neighbouring po int s. This is called
the s tan dard five  po int formula.
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
The s tan dard five  po int formula is written :
ui 1, j  ui 1, j  ui , j 1  ui , j 1  4ui , j  0

Dirichlet
problem

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 1 
 
1  4 1
 
 1 
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Also, instead of formula (a), we may use the following formula :
1
ui , j  [ui 1, j 1  ui 1, j 1  ui 1, j 1  ui 1, j 1 ]
4
Which uses the function values at the diagonal po int s.
And is therefore called the diagonal five  po int formula.

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Laplace’s equation:
We wish to solve Laplace's equation:
uxx+uyy=0
in a boundary region R with boundary C. As in Dirichlet's
problem, let the value of u be specified every where on C.
For simplicity, let R be a square region so that it can be
divided into a network of small squares of side h. Let the
values of u(x, y) on the boundary C be given by Ci and let
the interior mesh points and the boundary points be as in
the figure below: Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 C13 C12 C11 C10 C9

u7 u8 u9
C14 C8

C15 u4 u5 u6
C7

u1 u2 u3
C16 C6

C1 C2 Tikrit University-Civil Engineering

C3 C4 C5
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
We first use the diagonal five  po int formula and
compute; u5 , u7 , u9 , u1 and u3 in this order , Thus we obtain ;
1
u5  [C1  C5  C9  C13 ];
4
1
u7  [C15  u5  C11  C13 ];
4
1
u9  [u5  C7  C9  C11];
4
1
u1  [C1  C3  u5  C15 ] and
4
1
u3  [C3  C5  C7  u5 ]
4
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
We then compute, in this order , the remaining
quantities , such u8 , u4 , u6 and u2 by the s tan dard
five  po int formula. Thus we have;
1
u8  [u5  u9  C11  C7 ];
4
1
u4  [u1  u5  u7  C15 ];
4
1
u6  [u5  u3  C7  u9 ] and
4
1
u2  [C3  u3  u5  u1 ]
4 Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
When once all the ui (i  1,2,3,......9) are computed ,
their accuracy can be improved by any of the
iterative methods described below :
(1) Jacobi' s method :
Let ui(,nj) denote the n th iterative value of ui , j .
An iterative procedure to solve the Eq. (a ) is :
( n 1)1 (n)
ui, j [ui 1, j  ui(n1), j  ui(,nj)1  ui(,nj)1 ]
4
for the int erior mesh po int s. This is called the
po int Jacobi method .

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
(2) Gauss  Seidal method :
The method uses the latest iterative values
available and scans the mesh po int s systematic ally
from left to right along successive rows.
The iterative frmula is :
( n 1) 1 ( n 1)
ui, j  [ui 1, j  ui(n1), j  ui(,nj11)  ui(,nj)1 ]
4

It can be shown that the Gauss  Seidal scheme

converges twice as fast as the Jacobi scheme.
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
Example:
Solve Laplace equation for the figure given below:
50 100 100 100 50

u7 u8 u9
0 0

u4 u5 u6
0 0

u1 u2 u3
0 0

Tikrit University-Civil Engineering

0 0
Department Third Stage Eng.Anal.& Num.
0 0
Meth. Dr.Adnan Jayed Zedan
0
Solution :
We first compute the quantities ; u5 , u7 , u9 , u1 and u3
by u sin g the diagonal five  po int formula :
1
u5  [0  0  50  50]  25
(1)

4
1
u7  [0  25  100  50]  43.75
(1)

4
1
u9  [25  0  50  100]  43.75
(1)

4
1
u1  [0  0  25  0]  6.25
(1)

4
1
u3  [0  0  0  25Tikrit] University-Civil
(1)
6.25 Engineering
4 Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
We now compute ; u8 , u4 , u6 and u2 successively
by u sin g the s tan dard five  po int formula :
1
u(1)
8  [25  43.75  100  43.75]  53.125
4
1
u4(1)  [0  6.25  25  43.75]  18.75
4
1
u (1)
6  [25  6.25  0  43.75] 18.75
4
1
u2(1)  [6.25  0  6.25  25]  9.375
4
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
We have thus obtained the first approximations of all the
nine mesh po int s and we can now use one of the aterative
formula, by u sin g Gauss  Seidal formula :
( n 1) 1 ( n 1)
ui, j  [ui 1, j  ui(n1), j  ui(,nj11)  ui(,nj)1 ]
4

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
n u1 u2 u3 u4 u5 u6 u7 u8 u9

1 6.25 9.38 6.25 18.75 25.00 18.75 43.75 53.13 43.75

2 7.03 9.57 7.08 18.94 25.10 18.98 43.02 52.97 42.99

3 7.13 9.83 7.20 18.81 25.15 18.84 42.94 52.77 42.90

4 7.16 9.88 7.18 18.81 25.08 18.79 42.89 52.72 42.88

5 7.17 9.86 7.16 18.78 25.04 18.77 42.88 52.70 42.87

Tikrit University-Civil Engineering

Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan
 1 ( 2)
u  [u0,1  u2,1  u1, 0  u1, 2 ]
( 2)
1, 1
(1) ( 2) (1)

4
1
u1  [0  9.38  0  18.75]  7.03
4
1 ( 2)
u2,1  [u1,1  u3,1  u2, 0  u2, 2 ]
( 2) (1) ( 2) (1)

4
1
u2  [7.03  6.25  0  25]  9.57
4
Tikrit University-Civil Engineering
Department Third Stage Eng.Anal.& Num.
Meth. Dr.Adnan Jayed Zedan