3.

Partial Differential Equations

Classification: The most general linear partial differential

equation of the second order with two independent variables is,

 2  2  2  
A 2 B C 2  D E  F  G  0
x xy y x y

where A, B, C, D, E, F, G are functions of x and y including constants.

(i) If B 2  4 AC the equation is said to be elliptic.

(ii) If B 2  4 AC the equation is hyperbolic.

(iii) If B 2  4 AC the equation is said to be parabolic.

 2  2
Examples: (i)   0 which is known as Laplace equation is elliptic.
x 2 y 2

 2  2
(ii) c 2  which is known as one dimensional wave
x 2 t 2
equation is hyperbolic.(c is a constant)

 2 
(iii) k 2  which is known as the diffusion equation
x 2 t
is parabolic.( k is a constant)

The order of a partial differential equation is the order of the highest partial derivative it contains. A
partial differential equation is said to be linear if all partial derivatives and the dependent variable occur
as first order terms.

In general , a partial differential equation of order n has a solution which contains at most n arbitrary
functions. Therefore the general solution can be written as the linear combination of n arbitrary
functions. This general solution can be particularized to a unique solution if appropriate extra conditions
are provided. These are classified as boundary conditions. The kind of boundary conditions we need to
specify depend on the nature of the problem.

1
Classification of problems
(i) Equilibrium problems
(ii) Propagation problems

Equilibrium problems relate to steady state conditions. The problems in this category include steady
state temperature distributions, steady flow of electric current, equilibrium stress situations steady
ideal fluid flows. These problems are boundary value problems. We need to specify conditions which
exist along the entire boundary.

Examples

(i) Laplace equation in two dimensional Cartesian coordinates.

 2  2
i.e.  0
x 2 y 2

(ii) Poisson equation

 2  2
i.e.   f ( x, y )
x 2 y 2

Two Dimensional Heat Flow

Suppose we want to find the temperature distribution in a rectangular metal plate under certain conditions.
The plate is covered on its top and bottom faces by layers of thermal insulating material so that heat is
constrained to flow mainly in the X and Y directions. Along the edges of the plate various conditions are
applied. These are known as boundary conditions.

  00 C
layers
plate

  00 C
X Y
 0 C0

  100 0 C
2
When formulating a simple mathematical model the following assumptions are made.

(i) The metal is uniform in the sense that its thermal conductivity is the
same at all points of the plate.
(ii)The plate is sufficiently thin so that we neglect any heat flow in the
directions perpendicular to its face.

(iii) The temperature distribution is in the steady state. i.e. temperature

at any point in the plate does not depend on the time.

Let temperature function depends on x and y. Therefore,

   ( x, y)
 2  2
It can be shown that  ( x, y) satisfies Laplace equation   0.
x 2 y 2
Laplace equation possesses infinitely many solutions. For a unique solution the following boundary
conditions are used.

Boundary conditions

 0 when x  0 for 0  y  b …………..(i)

 0 when y  0 for 0  x  a ………….(ii)

 0 when x  a for 0  y  b …………(iii)

  100 when y  b for 0  x  a ……….(iv)

The analytical solution i.e. A formular for  is obtained by the method of separation of variables.

Assume that we can express  ( x, y) as a product of a function of x and a function of y .

  X ( x).Y ( y )
  2
 X ( x)Y ( y )   X ( x)Y ( y )
x x 2
  2
 X ( x)Y ( y )   Y ( y ) X ( x)
y y 2

 2  2
  0  YX   XY   0
x 2 y 2

3
 X Y   XY 
X  Y 
  k 2 , k  0
X Y
X 
( If k=0 we get  0  X ( x)  Ax  B
X
 ( x, y)  ( Ax  B)Y ( y)
When x=0,   0  0  B(Y ( y))  B  0.
When x=a,   0  0  ( Aa )Y ( y)  A  0
X=0 . This is not possible for non trivial solutions.)

X  Y 
If   k2
X Y
d2X
X   k 2 X  2
 k 2 X  0  X  Ae kx  Be kx
dx
2
d Y
Y   k 2Y  2  k y  0  Y  CCosky  DSinky
dy

 ( x, y)  ( Ae kx  Be  kx )( Cosky  DSinky ) (1)

X  Y 
If   k 2
X Y

X   k 2 X  X  ACoskx  BSinkx

Y   k 2Y  Y  C e ky  De  ky

 ( x, y)  ( ACoskx  BSinkx) )( C e ky  De  ky ) (2)

(1)  when x  0,  0

0=(A+B)Y(y) ,for 0<y<b  B=-A ( for non trivial  )

Then  ( x, y)  ( Ae kx  Ae  kx )Y ( y)
 ( x, y)  A(Sinhkx)Y ( y)

  0 when x=a.
0= A(Sinhkx)Y ( y)  Sinhka  0 (Since A  0, Y ( y)  0 for non trivial  . )
ka=0 . This is a contradiction since k  0, a  0.
(1) is not possible.
The possible solution is (2).

(2)   ( x, y)  ( ACoskx  BSinkx) )( C e ky  De  ky )

When x  0,  0 for 0  y  b

4
0  A(Y ( y))  A  0
 ( x, y)  ( BSinkx)(Ceky  De ky )

When y=0,   0 for 0  x  a

0  B(Sinkx)(C   D)  C   D  0 for non trivial solutions.
D  C 

 ( x, y)  (C BSinkx)(e ky  e  ky )
 ( x, y)  ( ESinkx)(Sinhky)
When x  a, ( x, y)  0 for 0  y  b
0  E SinkaSinhky  Sinka  0 for a non trivial solution.
Sin ka=0  ka  n where n is an integer.
n nx n
Then k=   E Sin Sin ( ) y , , n  1,2,.......
a a a

When n is negative , let n  n

nx nx nx nx
  E Sin ( ) Sinh ( )  E Sin ( ) Sinh ( )
a a a a
Therefore, negative values for n can be neglected.

nx ny
General Solution is    An Sin ( ) Sinh ( )
n 1 a a
  100 when y=b for 0  x  a

nx nb
 100=  An Sin Sinh
n 1 a a
This represents fourier half –range Sine series for the function f ( x)  100.
nb 2 nx
a
200
An Sinh   100Sin dx  (1  Cosn )
a a0 a n

nb  
0 , evenn
Hence An Sinh  400
a , oddn

 n
 0 n  even
 400 1
An   . n  odd
n nb
 Sinh ( )
 a

400 1 nx ny
General Solution is    Sin ( ) Sinh ( )
n 1 n nb a a
Sinh ( )
a
Analytical method of the General Solution is made to yield qualitative and quantitative
information. It will be seen later, even for a geometrically simple region straightforward
boundary conditions and the resulting formula for  is complicated.

5
Then if the geometry of the region is more awkward the analytical method may become vertually
impossible. In such a case we can resort to a numerical method.

Exercise: A rectangular plate with insulated surfaces is

8cm wide and so long compared to its width that it may be
considered infinite in length. If the temperature along one
short edge y=0 is given by, U(x,0)= 100Sin( 8x ) , 0  x  8, while
the two long edges x=0 and x=8, as well as the other short
edge are kept at 0 C , find steady state temperature U(x,y).
0

Laplace equation in polar coordinates

It can be shown that Laplace equation in polar co-ordinates is

2  T T  2T
2
r r   0.
r 2 r  2

Example Solve the above equation by the method of separation of

variables.

Let T (r , )  R(r ) ( )

T  2T
 R   2  R 
 r

T  2T
 R    R 
  2

 2T T  2T
r  2 0  r 2 R  rR  R   0
2
r
r 2
r 

6
 r 2 R   rR    
(r R  rR)  R   0 
2
  h (say)
R 

If h>o , let h  k , k  0
2

r 2 R  rR  k 2 R  0
dr
Let
r  e z  z  ln r   ez  r
dz
dR dR dz dR 1 dR dR
 .  . r 
dr dz dr dz r dr dz
d  dR  d  dR  d  dR  dz d 2 R 1
 r       2 .
dr  dr  dr  dz  dz  dz  dr dz r
d 2 R dR d 2 R 1
r 2   2 .
dr dr dz r

d 2R dR d 2 R
r2
2
r  2
dr dr dz

d
Let D 
dz then
r 2 R  rR  k 2 R  0
 (D 2  k 2 )R  0
Let R  e z  auxiliary equation is
2  k 2  0    k
7
  R  Ae kz  Be  kz  Ar k  Br  k
   k 2    CCosk  DSink
T (r , )  ( Ar k  Br  k )( CCosk  DSink )
…………(1)
If h=0    0    C   D
 D2R  0
R(r )  Ar  B
T ( r, )  ( Ar  B)( C   D ) …………….(2)

let h  k , k  0.
2
If h<o

   k 2    C e k  De k


 
r R  rR  k R  0 
2 2
(D 2  k 2 )R  0

R  e z  2  k 2  0    ki
R  ( Ae kiz  Be  kiz )( C e k  De k )
R  ( ACoskz  BSinkz)( C e k  De k )
R  ( ACos(k ln r )  BSin(k ln r )(
C e k  De k ) ………..(3)

8
 There are three possible solutions given by (1),(2) and
(3).

Example

The diameter of a semi circular plate of radius a, is kept

at 0 C0

and the temperature at the semi circular boundary is T C. 0

Find the steady state temperature in the plate.

T0C
.P
r

00C

LET u (r ,  ) be the steady state temperature at any point

P satisfies the equation,

 2T T  2T
r 2
r  2 0
r 2
r 
The boundary conditions are,

(i) U (r, )  0, 0  r  a

(ii) U (r ,  )  0, 0r a
(iii) U (a, )  T , 0    
9
For conditions (i) and (ii) U  0 when r=0 and U  0 when
r=a.

It can be shown that possible solution is ,

U (r , )  ( Ar k  Br  k )(CCosk  DSink )

U (r,0)  0  0  R(r )C  C  0 (Since R(r )  0)

U (r , )  ( Ar k  Br  k )( DSink )
U (r,  )  0  0  R(r ) DSink  Sink  0
 k  n , n  1,2,.......
k  n , n  1,2,.......
U (r , )  ( Ar n  Br  n )( DSinn )
U  0 when r  0, thenB  0 .


Then U ( r ,  )   A
n 0
n r n
Sinn

When r=a , U =T,


T   An a n Sinn ,0     .
n 0
This is the fourier half range Sine series of T at all points.

 4T
 n  even
An a n   TSinnd  1  (1 )   n
2 T 2T n

 0 n  0 n  0dd

10
 
4T 1 n
U (r ,  )  
 n1,3,5... a n
n
r Sinn

Exercise: An infinitely long plane uniform plate is bounded

by two parallel edges and an end at right angles to them.
The breadth is  . This end is maintained at a temperature
U at all points and other ends are at a zero temperature.
0

Determine the temperature at any point of the plate.

Heat Flow in one dimension

Suppose that we have a long thin bar of length l which is

aligned along the x-axis . We wish to determine the
temperature distribution  ( x, t ) in the bar.

11
Assume the bar is insulated along its sides , and that the
heat flows in the X direction only.
The following laws of heat flow are used.
(1) The amount of heat in a body is proportional to its mass and to its
temperature.

(2) Tha heat flows from a point at a higher temperature at a lower

temperature.

(3) The rate of flow of heat through a plane surface is proportional to the
area of the surface and to the rate of change of temperature with
respect to the distance in a direction perpendicular to the plane.

It can be shown that the temperature distribution  ( x, t ) satisfies

the following.

 2 1 
 where k is a positive constant.
x 2
k t ,

 2 u 1 
Example: Obtain the solution of the equation  ,
x 2 k t
where k is a positive constant satisfying the boundary
conditions,
U (0, t )  0,U (l , t )  0, t  0 , U(x,0)=f(x), 0  x  l . f(x) is a given
function and l is a constant.

Let U ( x, t )  X ( x).T (t )
 2 u 1 u 1
  X 
 ( x )T 
 (t )  X ( x).T (t )
x 2
k t k

12
 X  T 
   ( since x and t are independent )
X kT
f n of x f n of t

Since the temperature of a point on the bar is decreasing

when t increases,  is
negative.

Let  = -
2
, 0
1 d2X
. 2   2  X ( x)  ACosx  BSin x , A,B are constants.
X dx
T  2 kt
 T  kdt
    ln T   kt  T (t )  Ce
2
2
dt , where C is
a constant.

U ( x, t )  ( ACosx  BSinx) Ce  kt 2

 ( ACosx  BSinx)e  kt
2

(1)

U (0, t )  0, t  0 , (1)  0  Ae  kt ,

2
t  0  A  0
U ( x, t )  BSinx.e 
2 kt

U (l , t )  0, t  0  B  0 or Sinl  0
B is not possible for a non trivial solution.
 Sinl  0  l  r
13
 ( r is an integer)
r
 , r  1,2,...........
l
r 2 2 kt
rx 
U ( x, t )  B Sin ( )e l2
, r  1,2,.........
l
r 2 2 kt

rx 
U ( x, t )   Br Sin e l2
where Br is a constant depending on
r 1 l
r.

U ( x,0)  f ( x),0  x  l  f ( x) 

rx
 Br Sin
r 0 l
,o  x  l

This represents Fourier half range Sine series of f(x).

 rx 
l
2
 Br   f ( x) Sin dx, r  1,2,3.....
l 0  l 

r  kt 2 2
 2 l  rx   rx  l 2
U ( x, t )     f ( x) Sin dx Sin e
r 1  l 0  l   l

Example:

A rod of length l has its A and B kept at 0 0 C and 100 0 C respectively, until
steady state conditions prevail. If the temperature at B is reduced suddenly
to 0 0 C and kept so , while that of A is maintained . Find the temperature
U ( x, t ) at a distance x from A and at a distance x.

14
 Equation of Vibrating String( One Dimensional Wave Equation)

P(x,y)
y

Consider a perfectly flexible homogeneous string tightly stretched between two

points O and A. We assume the tension in the string to be so large that gravity
may be neglected in comparison with it.

The differential equation governing the motion when the string is set vibrating in
the vertical plane can be shown as,
2 y 2  y
2
 c where c is a constant.
t 2 x 2
This is known as the one dimensional wave equation.

Example:

A string is stretched and fastened to two points at a distance l apart . Motion is started by
x
displacing the string in the form y  aSin from which it is released at a time t=o. Show that
l
the displacement of any point at distance x from one end at time t is given by,

kt kt
y( x, t )  aSin Cos where k is a constant.
l l

Let y( x, t )  X ( x)T (t )
2 y 2  y
2
X  1 T 
y( x, t ) satisfies the equation  k  XT 
  k 2
X 
T   2 h
t 2 x 2 X k T

When h  0 i.e. h  2

X   2 X  X ( x)  C1e x  C2 e x
T   2 k 2T  T (t )  C3 e kt  C4 e kt
15
Y ( x, t )  X ( x).T (t )  (C1 e x  C2 e x )(C3 e kt  C4 e kt ) ----------------(1)

When h  0

X   0  X ( x)  C5 x  C6
T   0  T (t )  C7 t  C8
Y ( x, t )  X ( x).T (t )  ( C5 x  C6 ) (C7 t  C8 ) ----------------(2)

When h  0 i.e. h  2

X   2 X  X ( x)  C9 Cosx  C10 Sinx

T   2 k 2T  T (t )  C11Coskt  C12 Sinkt
Y ( x, t )  X ( x).T (t )  ( C9 Cosx  C10 Sinx )(C11Coskt  C12 Sinkt) ---------------(3)

Since this is a problem on vibrating string , the solution should be a periodic function of t.
Therefore, the solution is given by (3).

Boundary conditions are:

(i) y(0, t )  y(l , t )  0, t

x
(ii) y ( x,0)  aSin
l
 y 
(iii)    0
 t  t 0

(3)  y(0, t )  C9 (T (t ))  0  C9  0 for non-trivial solutions.

 y( x, t )  C10 Sinx( C11Coskt  C12 Sinkt)
y(l , t )  0  C10 Sinl (T (t ))  0  Sinl  0 for non-trivial solutions.
l  n , n is an integer
n
 , n  1,2,....
l
nx  nkt nkt 
 y( x, t )  Sin  b1Cos  b2 Sin  , b1 ,b2 are non zero real numbers.
l  l l 
y nx   nk  nkt  nk  nkt 
 Sin   b1   Sin  b2  Cos 
t l   l  l  l  l 

 y 
   0  b2  0
 t  t 0

nx nkt
 y( x, t )  b1 Sin Cos , n  1,2,.....
l l

16
 
nx nkt
General solution is  y ( x, t )   An Sin Cos where An is a constant depending on n.
n 1 l l
x x  nx
y ( x,0)  aSin  aSin   An Sin
l l n 1 l
A1  a, Ai  0 for all i  2 , where i is an integer.
x kt
Solution is  y ( x, t )  aSin Cos .
l l
Example:
A string is stretched and fastened to two points l apart. Motion is started by displacing the string
into the form y  c(lx  x 2 ) from which it is released at time t=0. Find the displacement at any
point on the string at a distance x from one end at time t.

Exercises

1. Solve the following equation.

 2u u
 24 x 2 (t  2) given that at x  0, u  e 2t and  4t .
x 2
x

2. A rectangular plate OPQR is bounded by the lines x=0,y=0, x=4, y=2.

Determine the potential distribution u(x,y) over the rectangle using the Laplace
 2u  2u
equation   0 , subject to the following boundary conditions
x 2 y 2
u(0, y)  0 , 0 y2
u(0, y)  0 , 0 y2
u( x,2)  0 , 0 x4
u( x,0)  x(4  x) , 0  x  4 .

3. Solve Laplace’s equation in plane polar coordinates

 2 v(r , ) 1 v(r , ) 1  2 v(r , )
  2  0 in the circular region x 2  y 2  1 of the plane
r 2
r r r  2

where,
(1) v(r , ) is finite for 0  x  1 and for all 
(2) v(1, )  Sin 2  4Cos
(3) v(r,  2 )  v(r, ) for 0  x  1 .

4. An insulated uniform metal bar , 10 units long , has the temperature of its ends
maintained at 0 0 C and at t=0 the temperature distribution f (x) along the bar
 2 u 1 u
is defined by f ( x)  x(10  x). Solve the heat conduction equation 
x 2 c 2 t
with c 2  4 to determine the temperature u of any point in the bar at time t.

17
5. A perfectly elastic string is stretched between two points 10cm apart. Its centre
point is displaced 2cm from its position of rest at right angles to the original direction of the
 2u 1  2u
string and then released with zero velocity. Applying the equation  with
x 2 c 2 t 2
c 2  1, determine the subsequent motion u ( x, t ).

18