0% found this document useful (0 votes)
36 views

SOV Conduction

This document discusses separation of variables techniques for solving partial differential equations in multi-dimensional coordinate systems such as rectangular, cylindrical, and spherical coordinates. It provides several examples of using separation of variables to solve Laplace's equation in different coordinate systems. The solutions involve expanding the dependent variable as a sum of products of functions with each function depending on one spatial variable. This leads to ordinary differential equations that can be solved for the separated functions.

Uploaded by

Ranadip Acharya
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
36 views

SOV Conduction

This document discusses separation of variables techniques for solving partial differential equations in multi-dimensional coordinate systems such as rectangular, cylindrical, and spherical coordinates. It provides several examples of using separation of variables to solve Laplace's equation in different coordinate systems. The solutions involve expanding the dependent variable as a sum of products of functions with each function depending on one spatial variable. This leads to ordinary differential equations that can be solved for the separated functions.

Uploaded by

Ranadip Acharya
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 47

Separation of Variables for Multi-

dimensional Steady Systems


Example 1: Laplaces equation in rectangular
coordinates (p.93):
See solution on page 93-99
) , ( ) (
) ( ) , (
y x u Y b
X a Y X T
+ +
=

0
' ' '
= +
+
k
g
T T
yy xx
T
L
T
L
T
y
=0
T
x
=0
Example 2: Laplaces equation in cylindrical
coordinates
r
0 1

0
1 1
2
= + +

U
r
U
r
U
r rr
U(1,)=f()
1
Use separation of variables:
1 becomes:
Also, it must be periodic in :
So:
This is satisfied only if =n where n=0,1,2,, so:
) ( ) ( = r R U
: or

cos , sin
" ' "
0 "
1
'
1
"
2 2
2
=

= +
= + +
R
R
r
R
R
r
R
r
R
r
R
)] ( cos[ )] ( sin[
cos sin


2 2 + + + =
+
B A
B A
) 2 ( ) ( + =
,... 2 , 1 , 0 cos sin = + = n n B n A
n n n
;
R equation:
This is the equidimesional equation. Characteristic
equation obtained substituting R=r
n
:
For n=0:
For n0:
0 ' "
2 2
= + R n rR R r
2 2 2
0 n n = = + or - 1) - (
r c c R
r
c
R
r c R rR R r
ln '
'
'
ln ln ' ln 0 ' "
0 0
0
0
2
+ = =
= = +
or or
or
n
n
n
n n
r c r c R

+ = '
r c c R ln '
0 0
+ =
solution:
However, u< as r 0, so a
0
= a
n
= b
n
= 0
( )
( )

=
+ +
+ + + =
=
1
1
0 0
0
cos '
sin ' ln '
) ( ) ( ) , (
n
n
n
n
n
n
n
n
n
n
n
n n
n r b r b
n r a r a a r a
r R r U

=
+ + =
1 1
0
cos sin ) , (
n
n
n
n
n
n
n r b n r a a r U
Apply non homog. B.C.:
Use orthogonality:
First multiply by sin(m) and integrate in (0,2):


=

=
+ + = =
1 1
0
cos sin ) ( ) , 1 (
n
n
n
n
n b n a a f U




m
n
n
n
n
a d m f
d n m b d n m a
d m a d m f
=
+ +
=

=
2
0
0
1
2
0
1
2
0
0
2
0
0
2
0
sin ) (
cos sin sin sin
sin sin ) (
or

4 4 4 4 3 4 4 4 4 2 1
4 43 4 42 1
Next multiply by cos(m) and integrate in (0,2)
Next multiply by 1 and integrate in (0,2)
General 2-D cylindrical coordinate problems:

m
b d m f =

2
0
cos ) (

2 ) (
0
2
0
a d f =

) , ( ) , ( ) , ( z T z r T r T
Equidimensional
equation in r:
sin and cos
Bessel function
in r, exponential
in z
Thin walled tubes,
can reduce to
Cartesian coords.
Example 3: Semi-infinite solid cylinder
Let = T - T

Then:
r
r
0
z

) ( 0 ) , (
) ( ) ( ) 0 , (
0 ) , (
) , 0 ( 0 ) , 0 (
0
0
2
2
finite r
r F T r f r
z r
finite z z
r
z
r
r
r
r

or
=
= =
=
= =


+
|

\
|

Let (r,z) = R(r)Z(z)


Need eigenvalue problem in r-direction:
( )
0 ) ( , 0
and
0 ) ( , ) 0 ( or 0
) 0 (
0
"
'
1
2
2
2
0
2
2
= =
= = =
= +
|

\
|
= =
Z Z
dr
Z d
r R finite R
dr
dR
rR
dr
dR
r
dr
d
Z
Z
rR
dr
d
Rr

R
n
= A
n
J
o
(
n
r) where
n
are determined from J
o
(
n
r
o
)
Also,
Apply the non-homogeneous condition:
Use orthogonality:

=
=
1
0
) ( ) , (
) (
n
n
z
n
z
n n
r J e a z r
e C z Z
n
n

=
= =
1
0
) ( ) ( ) 0 , (
n
n n
r J a r F r

=
0
0
0
2
0
0
0
) (
) ( ) (
r
n
r
n
n
dr r rJ
dr r J r rF
a

2
) (
0
2
1
2
0
r J r
n

=
Example 4: Laplaces eqn. In spherical coordinates:
Axisymmetric no variation in direction

y
z
T(r,,)
) ( ) , ( , ) , 0 (
0 sin
sin
1
2


f R T finite T
T
r
T
r
r
= =
=
|

\
|

+
|

\
|


Let T(r,) = R(r)()
is the only possible homogeneous direction.
Choose the (-) sign to end up with the following:
Recast 1 as
finite R
R
dr
dR
r
dr
R d
r
d
d
d
d
=
= +
= +
|

\
|

) 0 (
0 2
0 sin
sin
1
2
2
2

and


1
2
0
sin
cos 1
sin
1
2
= +
|
|

\
|

d
d
d
d
Let x = cos ; then and
This is the so called Legendres eqn. If = n:
0 ) 1 ( ) 1 (
2
= + +
|

\
|

43 42 1


dx
d
x
dx
d
dx
d
d
dx
dx
d
d
d []
sin
[] []


= =
3
) ( ), ( : x Q x P
n n

Legendre functions
of first kind (polynomials
Of order n)
Legendre functions
of second kind (in finite
series)
Temperature Variation in a Solid Sphere
Eqn. 2 is equidimensional eqn. Characteristic
eqn. obtained by setting R = r

K , 2 , 1 , 0 ); (cos = = n P A
n n

+
=
+
=
+ +
=
+
=
= + = +
) 1 ( 2
) 1 2 ( 1
2
) 1 ( 4 1 1
2
4 1 1
0 0 2 ) 1 (
2 , 1
2
n
n
n
n n

or or


Hence
Apply non-homog. B.C.:

=
+
=
= = + =
0
) 1 (
) (cos ) , (
0 ) 0 ( ; ) (
n
n
n
n
n n
n
n
n
n n
P r a r T
D finite R r D r C r R


|

\
|
= =
+
=
=

=
1
1 0
0
0
0 sin ) (cos ) (cos 0 ) ( ) (
sin ) (cos ) (
2
1 2
) (cos ) (




d P P dx x P x P
d P f
n
R a
P R a f
n n m n
n
n
n
n
n
n
n
or

For example: if

< <
< <
= =
/2




, 0
2 / 0 ,
) ( ) , (
0
T
f R T
0
1
0
0
1
0
1
0
0
0
1
0
0
2 /
0
0
4
3
2
3
2 2
) (
2
1 2
sin ) (cos
2
1 2
T xdx
T
R a
T
dx
T
a
dx x P
n
T
d P
n
T R a
n
n
n
n
= =
= =
+
=
+
=


[ ]
K +
|

\
|

\
|
+ =
= =
= = =

) (cos
16
7
) (cos
4
3
2
1 ) , (
16
7
) 3 5 (
2
1
2
7
0
4
5
) 1 3 (
2
1
2
5
3
3
1
0
1
0
0
3
0
3
3
1
0
3
0
1
0
2
0
2
2

P
R
r
P
R
r
T
r T
T dx x x T R a
x x
T
dx x
T
R a

.
.
Legendre Polynomials
Consider the following linear second-order ODE with
variable coefficients:
Or equivalently:
0 ) 1 ( ) 1 (
2
= + +
|

\
|
y
dx
dy
x
dx
d

0 ) 1 ( 2 ) 1 (
2
2
2
= + + y
dx
dy
x
dx
y d
x
(5.109a)
(5.109b)
This equation is known as Legendres diferential
equation and its two linearly independent solutions
are called Legendre functions of the first and second
kind, respectively.
When = n, where n is a positive integer or zero, the
Legendre functions of the first kind becomes a
polynomial of degree n.
The general solution of Legendres equation for
n = 0, 1, 2, can be written as
P
n
(x) are polynomials called Legendre Polynomials of
degree n, and Q
n
(x) are the Legendre functions of the
second kind
) ( ) ( ) (
2 1
x Q C x P C x y
n n
+ =
(5.110)
The Legendre polynomials are given by:
...
! 5
) 4 )( 2 )( 3 )( 1 (
! 3
) 2 )( 1 (
) (
...
! 4
) 3 )( 1 )( 2 (
! 2
) 1 (
1 ) (
,... 7 , 5 , 3 ), (
) 1 ( 6 4 2
5 3 1
) 1 ( ) ( , ) (
,... 6 , 4 , 2 ), (
6 4 2
) 1 ( 5 3 1
) 1 ( ) ( , 1 ) (
5 3
4 2
2 / ) 1 (
1
2 /
0

+ +
+
+
=

+ +
+
+
=
=


= =
=


= =

x
n n n n
x
n n
x x V
x
n n n n
x
n n
x U
n x V
n
n
x P x x P
n x U
n
n
x P x P
n
n
n
n
n
n
n
n


(5.111a,b)
(5.111c,d)
(5.112a)
(5.112b)
The first six Legendre Polynomials are readily found
to be:
Rodrigues Formula
) 15 70 63 (
8
1
) ( ) 1 3 (
2
1
) (
) 3 30 35 (
8
1
) ( ) (
) 3 5 (
2
1
) ( 1 ) (
3 5
5
2
2
2 4
5 1
3
3 0
x x x x P x x P
x x x P x x P
x x x P x P
+ = =
+ = =
= =



n
n
n
n
n
x
dx
d
n
x P ) 1 (
! 2
1
) (
2
=
(5.113)
(5.114)
Recurrence formula:
The following property of Legendre Polynomials is
also frequently used:
x n
x P n x nP
x P
n n
n
) 1 2 (
) ( ) 1 ( ) (
) (
1 1
+
+ +
=
+
,... 3 , 2 , 1 ), ( ) 1 2 (
1 1
= + =
+
n x P n
dx
dP
dx
dP
n
n n

(5.115)
(5.116)
If |x| < 1, the Legendre functions of the second kind
are given by
Notice that the Legendre Functions of the second
kind are infinite series, which are convergent when
|x| < 1, but diverges as x 1
,... 7 , 5 , 3 ), (
5 3 1
) 1 ( 6 4 2
) 1 ( ) (
), ( ) (
,... 6 , 4 , 2 ), (
) 1 ( 5 3 1
6 4 2
) 1 ( ) (
), ( ) (
2 / ) 1 (
1 1
2 /
0 0
=


=
=
=


=
=
+
n x U
n
n
x Q
x U x Q
n x V
n
n
x Q
x V x Q
n
n
n
n
n
n




(5.117a,b)
(5.117c,d)
Legendre Functions of the second kind Q
n
(x) also
satisfy recurrence formulas (5.115) and (5.116)
When |x| < 1
Q
n
for any positive integer value of n:
x
x
x
x Q
1
0
tanh
1
1
ln
2
1
) (

=

+
=
(5.118)
)... (
) 1 ( 3
) 5 2 (
) (
1
) 1 2 (
) ( ) ( ) (
3 1 0
x P
n
n
x P
n
n
x Q x P x Q
n n n n

=
(5.119)
If |x| < 1, the substitution x = cos transforms
Legendres differential equation from the form
(5.109b) into the form:
or, equivalently
when = n. Hence the general solution to (5.120a) is
0 ) 1 ( sin
sin
1
= + +
|

\
|
y n n
d
dy
d
d


(5.120a)
0 ) 1 ( cot
2
2
= + + + y n n
d
dy
d
y d


(5.120b)
) (cos ) (cos ) (
2 1

n n
Q C P C y + =
(5.121)
It should be noted here that Q
n
(cos) is not finite
when cos = 1; that is when = k, k = 0, 1, 2, ,
whereas P
n
(cos) is merely a polynomial of degree n
in cos. In particular we have
.
.
.
) cos 3 3 cos 5 (
8
1
) cos 3 cos 5 (
2
1
) (cos
) 1 2 cos 3 (
4
1
) 1 cos 3 (
2
1
) (cos
cos ) (cos
1 ) (cos
3
3
2
2
1
0


+ = =
+ = =
=
=
P
P
P
P
When |x| < 1, the functions
are called the associated Legendre functions of
degree n and order m of the first and second kinds,
respectively. They can be shown to satisfy the
differential equation
) ( ) ( ,
) (
) 1 ( ) (
) ( ) ( ,
) (
) 1 ( ) (
0 2 / 2
0 2 / 2
x Q x Q
dx
x Q d
x x Q
x P x P
dx
x P d
x x P
n n
m
n
m
m m
n
n n
m
n
m
m m
n
= =
= =

(5.122a)
(5.122b)
0
1
) 1 ( 2 ) 1 (
2
2
2
2
2
=
(

+ + y
x
m
n n
dx
dy
x
dx
y d
x (5.123)
Non-Homogeneous Problems
May result from non-homog. boundaries and/or non-
homog. Governing equation
Example 5: Non-homog. Boundaries
The solutions can be added only when expressed in
a common coordinate system. Sign for upper and
lower B.C. depend on whether heat flux is in or out

2
= 0
=

2

1
= 0

2

2
= 0

2

3
= 0 +
+
F(y)

o
F(y)
0
0
0

0
0
=


h
y
k
" q
y
k =

1
1
=


h
y
k
0
1
=


y
0
3
=


y
"
2
q
y
k =

2
2
=


h
y
k
3
3
=


h
y
k
Example 6: Non-homog. Term in the G.E. (use of
partial solutions)
L L
x
y
l
l
0
u
0
0
0
0
u
x
y
=0
=0

y
=0

x
=0
0 ) , ( ) , ( ) 0 , ( ) , 0 (
0
" '
= = = =
= + +
y L l x x y
k
u
y x
yy xx
1
Assume soln. of following form:
With form given by 2 :
If u is function of x and y, try variation of parameters
) ( ) , ( ) , ( or
) ( ) , ( ) , (
4 3
2 1
y y x y x
x y x y x
+ =
+ =
2
3
(


2
2 1
1
1 1
2
1
2
2
1
2
2
2
2
2
) ,
0 ) , ( ) 0 , ( ) , 0 ( ; 0
0 ) ( ; 0 ) 0 ( ; 0
" '
=
= =


= =

= +

l x
y L x
y
y
x y x
L
dx
d
k
u
dx
d
Two-Dimensional Transients
Example 7: Separation of variables (p134)
Let U = W(r,)()
4 becomes:
) ( ) 0 , , ( ; 0 ) , , 1 (
) , , (
1
2



r f r U U
r U U
U U
r r
U
U
r
rr
= =
=
= + +

where
4
already normalized
2
2
' 1 1 1

=
(

+ + W
r
W
r
W
W
r rr
exp. decay
In time
Also:
Let W = R(r)()

2

= e
0
1 1
2
2
= + + + W W
r
W
r
W
r rr

2 2 2 2
2
2
" ' "
0 "
1
'
1
"

= + +
= + + +
r
R
R
r
R
R
r
R R
r
R
r
R
or

need periodic
behavior in
So;
Now , so satisfies:
) , ( ), , ( :
0 ) ( ' "
) 2 ( ) ( , cos , sin :
2 2 2 2
r Y r J R
R n r rR R r
n
n n




for where
= + +
+ = =
must be absent for
finite temp. at r=0
0 ) ( 0 ) , , 1 ( = =
mn n
J U
0 ) ( =
mn n
J
where
mn
is the m
th
zero crossing of J
n
m = 1, 2, 3, ; n = 0, 1, 2,
eigenvalues in r-dir.
eigenvalues in -dir.

Unknown constants are again evaluated using


orthogonality:
first:

+ =
1 0
2
) ( ) cos sin ( ) , , (
m n
mn n mn mn
mn
e r J n B n A r U

=
+ =
=
1 0
) ( ) cos sin (
) , ( ) 0 , , (
m n
mn n mn mn
r J n B n A
r f r U


4 43 4 42 1 4 4 4 3 4 4 4 2 1


d n r J A d n r f
m
mn n mn
n r F


=
=
2
0
2
1
' ' '
) ' , (
2
0
' sin ) ( ' sin ) , (
next:
Drop primes for convenience:



=
=

=
1
0
' '
2
' ' '
1
0
1
1
0
' ' ' ' ' ' ' ' '
) (
) ( ) ( ) ( ) ' ; (
dr r rJ A
dr r J r rJ A dr r J n r rF
n m n n m
m
n m n mn n mn n m n



{ }
1
0
2
1 1
2
1
0
) ( ) ( ) (
2
) ; ( ) (
(

=
+

r J r J r J
r
A
dr n r F r rJ
mn n mn n mn n mn
mn n


or
Determine B
mn
similarly. See p. 138. Note n=0 has
to be treated separately (see p. 90)
{ }
(

=
+
) ( ) ( ) (
2
1
2
1 1 mn n mn n mn n mn
J J J A L.H.S.
0
) (
2
) (
1 mn n
mn
mn n
J
n
J

+ =
+

=
(

=
+
+
1
0
2
0
2
1
2
1
1
0
sin ) , ( ) (
) (
2
) (
2
1
) ; ( ) (
dr d n r f r rJ
J
A
r J A dr n r F r rJ
mn n
mn n
mn
mn n mn mn n




Example 8: Reduction of multi-dimensional to one-
dimensional transients.
- Problem must be homog. and linear
- I.C. should be constant or expressible in
product form: e.g. R
i
(r)Z
i
(z) or X
i
(x)Y
i
(y),
etc.
L
z
r
t
o
t
o
t
o
r
o
0
0
2
2
2
2
) 0 , , (
0 ) , , (
0 ) , 0 , (
0 ) , , (
1 1
t t z r T
L r T
r T
z r T
T
z
T
r
T
r r
T
i
=
=
=
=


1
Let
Substitute in 1 :
This will be satisfied if:
) , ( ) , ( r Z r R T =
2
0
1 1 1
) (
1 1
=
|

\
|
+
|

\
|
+
+ = + +


Z Z R Z R R
r
R
RZ Z R RZ Z R
r
Z R
zz r rr
zz r rr
or

Z Z
R R
r
R
zz
r rr
1
1 1
=
= +
and

B.C.:
These are satisfied when:
I.C.:
One choice that will satisfy this:
) , ( ) , ( ) , , (
) , 0 ( ) , ( ) , 0 , (
) , ( ) , ( ) , , (
0 0



L Z r R L r T
Z r R r T
z Z r R z r T
=
=
=
0 ) , ( ) , 0 ( 0 ) , (
0
= = L Z Z r R , ,
) 0 , ( ) 0 , ( ) 0 , , (
0
z Z r R t t z r T
i
= =
1 ) 0 , ( ) 0 , (
0 0
= = z Z t t r R
i
,
This reduces the originial problem into 2 one-dim.
Problems.
Z is the solution to a 1-D transient in a slab: (p.84)
R is the solution to a 1-D transient in an infinitely long
cylinder: (p.120)
where
m
satisfy J
0
(
m
r
0
) = 0.
Solution for T is obtained from 2

=
+
+
+
=
0
/ ) 1 2 (
2 2 2
1 2
] / ) 1 2 sin[( 4
) , (
n
L n
e
n
L z n
z Z

( )

=
1
/ ) (
0 1 0
0
0
2
0
2
0
) ( ) (
) (
2 ) , (
m
r r
m m
m
i
m
e
r J r
r J
t t r R

Many other problems can be handled in a similar


manner, e.g.
See other examples, also p.265 I&P
Can also include other homog. Conditions such as
convection. Cannot handle flux.
Multi-D I.C. must be of product form.
=


Some Multi-D Steady Problems
Solvable By Separation of Variables
Rectangular:
2D (s.s):
1
2
3
q
f
1
(x)
f
4
(x)
f
3
(x)
f
2
(x) T
0
T
0
T
0
f
2
(x)
h, T
0
3D (s.s.):
A: 1 at T
1
, 2 at T
2
, all others at T
0
.
B: 1 at T
1
, 2 at T
2
, all others convective conditions
C: 1 at T
1
, convection at all others
D: Region extends infinitely in the x-dir
n
from X=0.
Surface at x=0 at T
1
. Convection at other four
surfaces
x
y
z
2
6
3
5
4
1
Cylindrical:
2D:
(a) Short cylinder (r,z)
(b) Circumferential temperature variation
variation (r,) in a half infinite solid
long half cylinder: z > 0
z

r
f(r) = T-T
r
T
S
or h
(for r=R or z=L)
f()
T
0
r

z
T

(c) Disk shaped input B.C. (flux or temperature)


Can apply to bottom surface of a semi-infinite
region or between two semi-infinite regions:
Spherical (r,); axisymmetric
(a) hollow sphere (b) solid hemisphere
T
0
f()

T
0
f()
Multi-Dimensional Transients Using
Separation of Variables
Rectangular Regions:
(i) Assuming surface temperature
is T
0
at all faces for t > 0
T
i
(x,y,z)
2H
2W
2L
z
x
y
( ) ( ) ( ) [ ]
( ) dxdydz
H
z m
L
y m
W
x m
T z y x T A
H p L n W m
H
z m
L
y m
W
x m
A e
T t z y x T
W L H
i mnp
mnp
m n p
mnp
t
mnp


sin sin sin ) , , (


/ / /
sin sin sin
) , , , (
0 0 0
0
2 2 2
2 2
1 1 1
0
2

=
+ + =
=
=


(ii)
Cylindrical Regions:
(i) (ii)
L
T
0
x
y
T
i
(x,y)
h, T
0
h, T
0
T
i
(r,) h(T
s
- T

)
T
0
T
0
T
0
T
i

You might also like