Chapter 5

Steady and unsteady diffusion

In this chapter, we solve the diffusion and forced convection equations, in
which it is necessary to evaluate the temperature or concentration fields
when the velocity field is known.

5.1

Spherical coordinates

The solution of the equation:

t c = D2 c

(5.1)

in spherical coordinates can be obtained using the separation of variables. In

spherical coordinates, 5.1 is:
1
c
t c = 2
r2
r r
r

1
c

+ 2
sin ()
r sin ()

1
2c
r 2 sin ()2 2

(5.2)

where T is a general function of (r, , , t). In the separation of variables

technique, the variable T is written as:
c = F (t)R(r)()()

(5.3)

Inserting this into

4215, and dividing the entire equation by F (t)R(r)()(),
we get:
1
d
1
1 1 d
1
d
1
d2
1 1 dF (t)
2 dR
r
+
sin
()
+
=
D F (t) dt
R r 2 dr
dr
r 2 sin () d
d
r 2 sin ()2 d2
(5.4)
!

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

Note that the partial derivatives have now become total derivatives because
F is only a function of t, and R, and are functions of only r, and
only. The left side of the equation is only a function of t, while the right side
is only a function of (r, , ). Therefore, each side has to be equation to a
constant, say 2 . The solution for F can be easily obtained:
F (t) = exp (2 t)

(5.5)

Note that it is necessary for the constant 2 to be negative for the solution
to be bounded in the limit t .
The remainder of the equation 5.4 can now be written as:
1 d2
dR
1 1 d
d
1 d
sin ()
r2
+
sin ()
+ 2 r 2 =
R dr
dr
sin () d
d
d2
(5.6)
Here, the left side is only a function of r and , and the right side is only a
function of . Therefore, both these have to be equal to a constant, say m2 .
This can be easily solved for the function ():
2

"

= A1 sin (m) + A2 cos (m)

(5.7)

Note that m has to be an integer, because the physical system obtained

remains the same when is increased by an angle 2.
The rest of equation 5.6 can now be written as:
dR
1
d
m2
d
1 d
r2
+ 2 r 2 =
sin ()
+
R dr
dr
sin () d
d
sin ()2
!

(5.8)

Here, the right side is only a function of , while the left side is only a function
of r. Therefore, both sides can be set equal to a constant, n(n + 1). The
equation for then becomes
d
d
m2
1
= n(n + 1)
sin ()
+

sin () d
d
sin ()2
!

(5.9)

The solutions of the above equation are called associated Legendre functions
of degree n and order m:
() = B1 Pnm (cos () + B2 Qm
n (cos ()

(5.10)

5.1. SPHERICAL COORDINATES

Note that just as m was constrained to be an integer in 5.7 and

4221 because
the solution is required to remain unchanged when is increased by 2, in
the above equation n is also an integer because of the functions with non integer values of n become infinite at = 0. In addition, there is another
stipulation that n > m, because diverges at = 0 for n < m. Further, the
functions Qm
n (cos ()) also become infinite for = , and therefore the only
solution for which is finite for both = 0 and = is:
= B1 Pnm (cos ())

(5.11)

The above solutions are called Legendre polynomials, because the series
solution for the Legendre equation 5.11 terminates at a certain order. The
first few polynomials are:
P00 = 1
P10 (x) = x
2
P20 (x) = 12 (3x
1)
P11 (x) = 1 x2
P21 (x) = 3x 1 x2 P22 (x) = 3(1 x2 )

(5.12)

Finally, the function R(r) can be determined by solving the left side of
5.8:
"
#
d2 R 2 dR
n(n + 1)
2
+
+
R=0
(5.13)
dr 2
r dr
r2
The solutions of this equation are called spherical Bessel functions, which
are given by:
R(r) = C1 jn (r) + C2 yn (r)
(5.14)
where the functions jn and yn are related to Bessel functions:
jn (x) =
yn (x) =

/2xJn+ 1 (x)

/2xYn+ 1 (x)

(5.15)

In case we are considering a system at steady state, where the time derivative
of T is zero (so that is zero), the spherical Bessel functions assume relatively
simple forms:
R(r) = C1 r n + C2 r n1
(5.16)
Having obtained the above solutions, we can derive a general solution for
T:
T = R(r)()()F (t)
=

n1
X
X

n=0 m=0

[C1n jn (r) + C2n yn (r)]Pnm (cos ())

[A1m sin (m) + A2m cos (m)] exp (2 t)

(5.17)

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

where the functions C1n , C2n , A1m and A2n are to be determined from the
boundary conditions.
The solutions
4220 and
4224 are eigenfunctions of Sturm - Liouville equations, and therefore the eigenfunctions are complete and orthogonal. This
orthogonality can be used to determine the constants C1n , C2n , A1n and A2n
in
4230. The orthogonality of the sine and cosine functions are expressed as:

0
d sin (m) sin (m0 ) =
0
2
Z 2
0
d cos (m) cos (m0 ) =
0
2
Z

for m 6= m0
for m = m0
for m 6= m0
for m = m0

0
for m 6= n and (m + n) even
2m
d sin (m) cos (m ) =
for m 6= n and (m + n) odd
m2 n 2
0
0
for m = n
(5.18)
Similarly, the Legendre polynomials also satisfy the following orthogonality
constraints in the interval 0 :
Z

Pnm (cos ())Pnm0 (cos ())

0
for n 6= n0
2 (n + m)!
for n = n0
2n + 1 (n m)!

5.1.1

Diffusion from a sphere

5.1.2

Effective conductivity of a composite

(5.19)

A composite material of thickness L consists of a matrix of thermal conductivity Km , embedded with particles of radius R and thermal conductivity
Kp . A temperature gradient T 0 is applied across the composite in the x3
direction, as shown in figure 5.1, and it is necessary to determine the heat
flux q3 as a function of the temperature gradient T 0 in the x3 direction. It is
assumed that the length L is large compared to the thickness of the particles
R, so that the material can be viewed as a continuum with an effective conductivity which is determined by the conduction of heat through the particles
and the matrix,
q3 = Kef f T 0
(5.20)
It is necessary to determine Kef f as a function of the conductivities of the
particles and the matrix.

5.1. SPHERICAL COORDINATES

T+LT
x3
q3
x1

x2

T
Figure 5.1: Effective conductivity of a composite material.
If there were no particles, the temperature gradient in the material would
be uniform, and the conductivity would be equal to the conductivity of the
matrix. The presence of the particles causes a disturbance to the temperature
field around a particle, which in turn results in a variation in the flux in the
vicinity of the particle. If the concentration of particles is small, so that
the temperature field around one particle is not influenced by the presence
of other particles, the solution can be obtained by considering the effect of
an isolated particle in an infinite matrix, and adding this effect for all the
particles in the matrix.
In order to solve this problem, we first consider the effect of a particle on
the temperature field and heat flux in the matrix. The particle is considered
to be located at the origin of the coordinate system, as shown in figure 5.2,
and a uniform temperature gradient T 0 is imposed in the x3 direction. At
steady state, the temperature fields in the particle Tp and in the matrix Tm
satisfy the Laplace equation,
2 Tp = 0
2 Tm = 0

(5.21)

At the interface between the particle and the matrix, the boundary conditions
require that the temperature and the heat flux are equal,
Tp = T m

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

Kp

Tm
Tp
= Km
r
r

(5.22)

Though the temperature field is disturbed by the presence of the particle near
the particle surface, it is expected that the temperature field is unaltered at
a large distance from the particle r , where it is given by
T = T 0 x3 = T 0 r cos ()

(5.23)

The temperature field far from the particle can be expressed in terms of
the Legendre polynomial solutions for the Laplace equation as
T = T 0 rP10 (cos ())

(5.24)

Since the temperature field is driven by the constant temperature gradient

imposed at a large distance from the particle, it is expected that the temperature field near the particle will also be proportional to P10 (cos ()), since
any solution which contains other Legendre polynomials is orthogonal to
(0)
P1 (cos ()). Therefore, the solutions for Tp and Tm have to be of the form
Ap2 0
P (cos ())
r2 1
Am2
= Am1 rP10 (cos () + 2 P10 (cos ())
r

Tp = Ap1 rP10 (cos () +

Tm

(5.25)

The constant Ap2 is identically zero because Tp is finite at r = 0, while

Am1 = T 0 in order to satisfy the boundary condition in the limit r .
The constants Ap1 and Am2 are determined from the boundary conditions at
the surface of the sphers,
3T 0
2 + KR
(1 KR )R3
=
2 + KR

Ap1 =
Am2

(5.26)

The temperature fileds in the particle and the matrix are

3T 0 r
P 0 (cos ()
2 + KR 1
(1 KR )R3 T 0 0
P (cos ())
= T 0 rP10 (cos () +
(2 + KR )r 2 1

Tp =
Tm

(5.27)

5.1. SPHERICAL COORDINATES

x3

r
x2

x1
Figure 5.2: Coordinate system for analysing temperature field around the
sphere.

The effective conductivity is determined using an effective equation of the

form
hj3e i = Kef f T 0
(5.28)
where hj3e i is the heat flux along the direction of the temperature gradient,
which is the direction of the imposed temperature gradient. The average
heat flux can be calculated by taking an average over the entire volume of
the matrix,
1 Z
dV j3e
V
!
Z
Z
1
dV Kp (T ).e3
dV Km (T ).e3 +
=
V
particles
matrix
!
Z
Z
1
=
dV K(T ).e3 +
dV (Kp Km )(T ).e3
V
composite
particles
!
N Vparticle
1
3
0
0
=
(Kp Km )
T
Km T +
(5.29)
V
V
2 + KR

hj3e i =

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

This gives the effective conductivity

3(KR 1)

1+
2 + KR

K = Km

5.2
5.2.1

(5.30)

Steady diffusion in an infinite domain

Temperature distribution due to a source of energy

Consider a hot sphere of radius R which is continuously generating Q units

of energy per unit time into the surrounding fluid, whose temperature at a
large distance from the sphere is T , as shown in figure 5.3. The temperature
field in the fluid surrounding the sphere can be determined from the energy
balance equation at steady state,
2 T = 0

(5.31)

Since the configuration is spherically symmetric, the conservation equation

5.31 is most convenient to solve in a spherical coordinate system in which the
origin is located at the center of the sphere. The energy balance equation is
then given by
!
1
T
r
=0
(5.32)
r 2 r
r
The solution for this equation, which satisfies the condition T = T at a
large distance from the sphere, is
T =

A
+ T
r

(5.33)

The constant A has to be determined from the condition that the total heat
flux at the surface of the sphere is Q,


dT 
(4R2 ) = Q
K 
dr r=R

(5.34)

This equation can be solved to obtain A = (Q/4K), so that the temperature

field is
Q
T =
+ T
(5.35)
4Kr

5.2. STEADY DIFFUSION IN AN INFINITE DOMAIN

x3

x1

r>Infinity
T>T
inf

x2

r=R
e
R2 )
j=(Q/4

Figure 5.3: Temperature field due to heat generation at the surface of a

sphere.

10

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

Note that the solution for the temperature, 5.35 does not depend on the
radius of the sphere, but only on the total amount of heat generated per unit
time from the sphere. If the object that generates the heat were some other
irregular object, then the temperature field would depend on the dimensions
of the object. However, if we are sufficiently far from the object, so that the
distance from the object is large compared to the characteristic length of the
object, the object appears as a point source of heat, and the observer cannot
discern the detailed shape of the object. In this case, it is expected that the
temperature distribution will not depend on the dimensions of the object,
but only on the total heat generated per unit time. Mathematically, a point
source of energy located at the position x0 is represented by a delta function,
S(x) = Q(x x0 )

5.2.2

(5.36)

Dirac delta function

The one-dimensional Dirac delta function, (x), is defined as

(x) = 0 for x 6= 0
Z

and

(5.37)

dx(x) = 1

(5.38)

dx(x)g(x) = g(0)

(5.39)

It is clear from equation 5.39 that (x) has dimensions of inverse of length.
The delta function can be considered the limit of the discontinuous function
(figure 5.4)
f (x) = (1/h) for(h/2 < x < h/2)
(5.40)
in the limit h 0. In this limit, the width of the function tends to zero,
while the height becomes infinite, in such a way that the area under the
curve remains a constant. It is clear that the function f (x) satisfies all three
conditions, 5.37 to 5.39, in the limit h 0.
In a similar fashion, the three dimensional Dirac delta function, (x1 , x2 , x3 )
is defined by
(x1 , x2 , x3 ) = 0 for x1 6= 0; x2 6= 0; x3 6= 0
(5.41)
Z

dx1

dx2

dx3 (x1 , x2 , x3 ) = 1

(5.42)

11

5.2. STEADY DIFFUSION IN AN INFINITE DOMAIN

f(x)
(1/h)

x
h __
__
h
2 2
Figure 5.4: Dirac delta function in one dimension.
and
Z

dx1

dx2

dx3 (x1 , x2 , x3 )g(x1 , x2 , x3 ) = g(0, 0, 0)

(5.43)

The three dimensional Dirac delta function can be considerd as the limit of
the function
f (x1 , x2 , x3 ) = (1/h3 ) for(h/2 < x1 < h/2)and(h/2 < x2 < h/2)and(h/2 < x3 < h/2)
= 0 otherwise
(5.44)
when h 0. It is easy to see that this function satisfies all of the above
properties.

5.2.3

Temperature distribution due to a point source

The Greens function for an infinite domain is defined as the temperature (or
concentration) distribution due to a point source of unit strength.
K2 G = (x x0 )

(5.45)

The solution for the Greens function on an infinite domain can be obtained
by first shifting the origin of the coordinate system to the position x0 , so that

12

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

the radius vector r = x x0 , and then solving equation 5.45 in a spherical

coordinate system in this coordinate system. In this new coordinate system,
the configuration is spherically symmetric, and so the conservation equation
5.45 is
1 G
r
= (r)
(5.46)
K
r r r
For r 6= 0, the right side of equation 5.46 is equal to zero, so the solution for
G (upto an additive constant) is
G=

1
4Kr

(5.47)

where A is a constant to be determined from the condition at the origin.

The condition at the origin is most conveniently determined by integrating
equation 5.46 over a small radius  around the origin,
K


0

A
(4r )dr 2
r


= 1S(x) =

dx0 (x x0 )S(x0 )

(5.48)

The solution for the temperature field is then given by

T (x) =

1
4K

dx0

S(x0 )
|x x0 )

(5.49)

Example
A wire of length 2L immersed in a fluid generates heat at the rate of Q
per unit length of the wire per unit time, as shown in figure 5.5. Determine
the temperature field due to this wire.
A cylindrical coordinate system is used, where the z axis is along the
length of the wire, and the origin is at the center of the wire. The wire is
considered to be a line of infinitesimal thickness in the x and y directions, so
that the energy source due to the wire per unit length is given by
S(x) = Q(x)(y) for L < z < L

(5.50)

The temperature field, in terms of this source, is given by

T (x) =

dx

dx0

Z
Z

dy

dy 0

Z
Z

L
L
L
L

dz 0 G(x x0 , y y 0 , z z 0 )S(x0 , y 0 , z 0 )
dz 0

1
Q(x0 )(y 0 )
0
2
4K
((x x ) + (y y 0 )2 + (z z 0 )2 )1/2

5.2. STEADY DIFFUSION IN AN INFINITE DOMAIN

13

x3

x2
2L
x1
Figure 5.5: Heat generation due to a wire.
Q
=
4K

= log

L
L

dz 0

(x2

L+z+

L + z +

y2

1
+ (z z 0 )2 )1/2

r 2 + (L + z)2

r2

+ (z

L)2

(5.51)

The formal solution 5.51 can be more conveniently examined as a function

of r at the center of the wire, z = 0,

!
L + r 2 + L2

T (x) = log
(5.52)
L + r 2 + L2
In the limit r  L, this solution reduces to
T (x) =

2QL
4Kr

(5.53)

This is the solution for the temperature field due to a point source of energy,
as expected when the distance from the source is large compared to the
characteristic length of the source. In the opposite limit r  L, the solution
for the temperature field reduces to
4L2
Q
T (x) =
log
4K
r2

14

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

Q
2L
=
log
2K
r


(5.54)

We will see, a little later, that this is the temperature field due to an infinite
line source of energy in three dimensions, or a point source in two dimensions.

5.2.4

Greens function for finite domains

The Greens function 5.47 is the temperature field due to a point source of
unit strength in a fluid of infinite extent. Most practical problems involve
finite domains, and it is necessary to obtain a Greens function which satisfies
the boundary conditions at the boundaries of the domain. In the case of
planar domains, this Greens function is obtained by using image charges.
For example, consider a point source located at xs = (L, 0, 0) on a semiinfinite domain bounded by a surface at x3 = 0, as shown in figure 5.6,
in which the surface has constant temperature equal to T . In this case,
the Greens function solution of the type 5.47 does not satisfy the condition
(T Tinf ty ) = 0 at the surface. However, the boundary condition can be
satisfied if we replace the finite domain by an infinite domain, in which
there is a source of strength +1 at (L, 0, 0), and a source of strength 1
at xI (L, 0, 0), as shown in figure 5.6 (a). It is easily seen that due to
symmetry, (T T ) is equal to zero everywhere on the plane x3 = 0, and
the Greens function which satisfies the zero temperature condition is called
the Dirichlet Greens function GD ,
GD (x =

1
1

4K|x xs | 4K|x xI |

(5.55)

The diffusion equation is satisfied in the semi-infinite domain x3 > 0 (since

the Greens function 5.55 satisfies the diffusion equation), and the boundary conditions are identical to the required boundary conditions at x3 = 0,
therefore, the solution GD is the required solution for the Greens function.
Of course, GD predicts a spurious temperature field in the half plane x3 < 0,
but this is outside the physial domain.
A similar Greens function can be obtained if the boundary condition at
the surface is a zero normal flux condition, j3e = 0, at x3 = 0, instead of the
zero temperature condition. In this case, the zero flux condition is identically
satisfied by imposing a source of equal strength +1 at xI = (L, 0, 0), as
shown in figure 5.6 (b). The solution for the Greens function with zero flux

15

5.2. STEADY DIFFUSION IN AN INFINITE DOMAIN

+Q

+Q

(0,0,L)

x3

(0,0,L)
x2

T=0



 
 
 
 
 
 
 
 
 
 
 
 
 

x1
(0,0,L)
+Q

(a)

+Q

+Q

(0,0,L)

x3

(0,0,L)
x2

e
j=0
3
















x1
(0,0,L)
Q
(b)

Figure 5.6: Greens functions for source near a wall with (a) zero temperature
conditions at the wall and (b) zero flux conditions at the wall.

16

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

condition is called the Neumann Greens function GN ,

GN (x =

1
1
+
4K|x xs | 4K|x xI |

(5.56)

A similar procedure could be used for more complicated geometries. The

Greens function for a source in a corner with zero flux conditions, as shown in
figure 5.7(a), could be obtained by using four sources of equal strength placed
symmetrically in an infinite domain. Similarly, the Greens function for a
source in a corner with zero temperature conditions, as shown in figure 5.7(b),
could be obtained by placing two sources and two sinks symmetrically in an
infinite domain. The Greens function for a source in a finite channel, as
shown in figure 5.8, would require an infinite number of sources.

5.2.5

Greens function for a sphere

The Dirichlet Greens function GD for a sphere is the solution for the temperature field due to a source of unit strength which satisfies the zero temperature
condition at the surface of the sphere. This Greens function can be derived
using spherical coordinates (r, , ). Consider a source of strength 1, which
is located, without loss of generality, at xS = (0, 0, r) in a sphere of radius
1, as shown in figure 5.9. The image, by symmetry has to be located along
the line joining the source point and the origin, at xI = (0, 0, r 0 ), but the
strength QI can, in general, be different from that of the source. The temperature at a point on the surface, x = (r, , ) in spherical coordinates,
or x = (sin () cos (), sin () sin (), cos ()), due to the source and sink, is
given by
1
Q
+
4K|x xI | 4K|x xI |
1
1
=
2
2
2
4K (sin () cos () + sin () sin ()2 + (cos () r)2 )1/2
!
Q
+
(sin ()2 cos ()2 + sin ()2 sin ()2 + (cos () r 0 )2 )1/2
!
1
Q
1
+
(5.57)
=
4K (1 + r 2 2r cos ())1/2 (1 + r 02 + 2r 0 cos ())1/2

T =

The values of Q and r 0 are determined from the boundary condition on the
surface of the sphere, which requires that T = 0 for all values of and .

5.2. STEADY DIFFUSION IN AN INFINITE DOMAIN

+Q

17

+Q

(0,L 2,L3)

+Q

(0,L 2,L 3)

(0,L2 ,L3)















T=0
(0,L ,L
2
3)
+Q

(0,L2 ,L3 )
+Q

(a)
+Q
(0,L 2,L3)

Q
(0,L 2,L 3)

+Q
(0,L2 ,L3)

(0,L ,L
2
3)
+Q

(0,L2 ,L3 )


 
 
 
 
 
 
 
 
 
 
je3=0

(b)
Figure 5.7: Greens functions for source near a corner with (a) zero temperature conditions at the walls and (b) zero flux conditions at the walls.

18

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

Q (x1 ,x2 ,x3 +3L)

+Q (x1 ,x2 ,x +3L)

+Q (x1 ,x2 ,x3 +2L)

+Q (x1 ,x2 ,x3 +2L)

Q (x1 ,x2 ,x3 +L)

+Q (x1 ,x2 ,x +L)

 

 

 

 

 
 
 
 
 
 
 
 
 

T=0

+Q (x1 ,x2 ,x3 )

Q (x1 ,x2 ,x L)
3




 

 

 

 

 

 

 

 

 

 

 

 

 
j=0
3

+Q (x1 ,x2 ,x3 )

















+Q (x1 ,x2 ,x L)
3

+Q (x1 ,x2 ,x 2L)

3

+Q (x1 ,x2 ,x 2L)

3

Q (x1 ,x2 ,x3 3L)

+Q (x1 ,x2 ,x3 3L)

(a)

(b)

Figure 5.8: Greens functions for source in a channel of finite width with (a)
zero temperature conditions at the walls and (b) zero flux conditions at the
walls.

19

5.2. STEADY DIFFUSION IN AN INFINITE DOMAIN

x3
r
r
1
x2

x1
Figure 5.9: Greens function for a sphere.
This condition is satisfied only if
Q2 (1 + r 2 ) = 1 + r 02
2Q2 r = 2r 0

(5.58)

From these two conditions, the solution for r 0 and Q are

r 0 = r 1
Q = r 1

(5.59)

Thus, the solution for the Greens function is

T =
where xI = (0, 0, 1/r).

1
1
+
4K|x xI | 4Kr|x xI |

(5.60)

20

5.2.6

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

Temperature distribution due to a dipole

Consider a source and sink of energy, of equal strength, Q of energy per unit
time, located at the positions (0, 0, L) and (0, 0, L), as shown in figure 5.10.
The temperature field due to the combination of source and sink is given by
Q
Q
+
(5.61)
T (x) =
2
2
2
1/2
2
2
4K(x + y + (z L) )
4K(x + y + (z + L)2 )1/2
If the distance of the observation point from the origin is sufficiently large,
so that the (x, y, z)  L, the temperature field is given by
Q
2Lz
T (x) =
2
4K (x + y 2 + z 2 )1/2
2QL cos ()
=
4K r 2
2QL 2
r P1 (cos ())
(5.62)
=
4K
Thus, the temperature distribution due to the combination of a source and
a sink, in the limit where the distance between the two reduces to zero, is
identical to the second spherical harmonic solution that was obtained for the
Laplace equation. Similarly, it can be shown that the combination of two
sources and two sinks, arranged in such a way that the net dipole moment
is zero, corresponds to the third spherical harmonic solution for the Laplace
equation.

5.2.7

Boundary integral technique

In order to solve the steady temperature equation,

we can use a Greens function

2 T = S(x)

(5.63)

2 G = (x)

(5.64)

which is defined according to the required boundary conditions on the domain, as follows.
Z

dV 0 .(T (x0 )G(x x0 ) G(x x0 )0 T (x0 ))

=
=

dV 0 (T (x0 )02 G(x x0 ) G(x x0 )02 T (x0 ))

(5.65)

21

5.3. PROBLEMS

x3

+Q
x2

2L
Q
x1

Figure 5.10: Temperature field due to a source and a sink.

The left side of the above equation can be written as an integral over the
boundary of the domain using the divergence theorem,
Z

dV 0 .(T (x0 )G(x x0 ) G(x x0 )0 T (x0 ))

=

dS 0 n.(T (x0 )G(x x0 ) G(x x0 )0 T (x0 ))

(5.66)

where n is the boundary of the domain. Therefore, the equation 5.66 reduces
to
T (x) +
=

5.2.8

5.3

dV 0 G(x x0 )S(x0 )

dS 0 n.(T (x0 )G(x x0 ) G(x x0 )0 T (x0 ))

(5.67)

Greens function for the unsteady diffusion equation

Problems

1. Derive the harmonic expansion for a two dimensional cylindrical coordinate system with coordinates (r, ).

22

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

(a) First, solve the heat equation K2 T = Q(x) in cylindrical coordinates, to obtain the temperature distribution due to a point
source.
(b) What is the temperature field when two sources are located as
shown in figure 1(a) and 1(b), and L  r? Compare with the
second terms in the cylindrical harmonic expansion.
(c) What is the temperature field when four sources are located as
shown in figures 1(c), and (d)? Compare with the third terms in
the cylindrical harmonic expansion.
(d) Determine the second and third terms in the harmonic expansion
by successively taking gradients of the temperature field due to
the point source.

+Q
Q

+Q

2L

x
Q

2L
(a)

(b)
y

y
+Q
x

+Q
(c)

Q
+Q

+Q x
Q

(d)

2. A point source of heat of strength Q (in units of heat energy per unit
time) is placed at a distance L from a wall.

23

5.3. PROBLEMS

(b)

(a)
Figure 5.11:
(a) If the wall is perfectly conducting, so that the flux lines at the wall
are perpendicular to the wall as shown in Figure 1(a), determine
the temperature as a function of position.
(b) If the wall is perfectly insulating, so that the flux lines at the wall
are parallel to the wall as shown in Figure 1(b), determine the
temperature as a function of position.
(c) If the wall is not perfectly conducting, but only a fraction f of the
heat on the wall penetrates it, while a fraction (1 f ) does not
penetrate the wall, determine the temperature field as a function
of position.
3. A heater coil in the form of a ring of radius a in the xy plane generates
heat Q per unit length of the coil per unit time, as shown in figure 2.
(a) If the heater is placed in an unbounded medium of thermal conductivity K, write an equation for the temperature as a function
of position in the medium.
(b) Plot the temperature as a function of position along the symmetry
axis of the heater (z axis in the figure). Simplify the expressions for

24

CHAPTER 5. STEADY AND UNSTEADY DIFFUSION

x
Figure 5.12:
the temperature for z  a and z  a. What does the expression
for z  a correspond to?
4. Determine the effective thermal conductivity for a dilute array of infinitely long circular cylinders along the plane perpendicular to the axis
of the cylinders, when the area fraction of the cylinders is . Use the
following steps.
(a) Consider an infinitely long cylinder with conductivity Kp in a
matrix of conductivity Km , and determine the temperature field
around the cylinder when a uniform gradient T 0 is imposed in the
x direction perpendicular to the axis of the cylinder.
(b) Write the heat flux as the sum of the flux over the matrix and
the sum over the cylinders. When the array is dilute, write the
integral as the sum over one cylinder, and determine the thermal
conductivity.
(c) What is the effective conductivity along the axis of the cylinders?