1-D Thermal Diusion Equation and Solutions

3.185
Fall, 2003
The 1-D thermal diusion equation for constant k (thermal conductivity) is almost identical to the solute
diusion equation:
T
t
=

2
T
x
2
+
q
c
p
(1)
or in cylindrical coordinates:
r
T
t
=

r
_
r
T
r
_
+ r
q
c
p
(2)
and spherical coordinates:
1
r
2
T
t
=

r
_
r
2
T
r
_
+ r
2
q
c
p
(3)
The most important dierence is that it uses the thermal diusivity =
k
cp
in the unsteady solutions, but
the thermal conductivity k to determine the heat ux using Fouriers rst law
q
x
= k
T
x
(4)
For this reason, to get solute diusion solutions from the thermal diusion solutions below, substitute D for
both k and , eectively setting c
p
to one.
1-D Heat Conduction Solutions
1. Steady-state
(a) No generation
i. Cartesian equation:
d
2
T
dx
2
= 0
1
Most texts simplify the cylindrical and spherical equations, they divide by r and r
2
respectively and product rule the
r-derivative apart. This gives
T
t
=
_

2
T
r
2
+
1
r
T
r
_
+
q
cp
for cylindrical and
T
t
=
_

2
T
r
2
+
2
r
T
r
_
+
q
cp
for spherical coordinates. I prefer equations 2 and 3 because they are easier to solve.
1
Solution:
T = Ax + B
Flux magnitude for conduction through a plate in series with heat transfer through a uid
boundary layer (analagous to either 1st-order chemical reaction or mass transfer through a
uid boundary layer):
|q
x
| =
|T
fl
T
1
|
1
h
+
L
k
(T
fl
is the uid temperature, analagous to the concentration in equilibrium with the uid in
diusion; T
1
is the temperature on the side opposite the uid.)
Dimensionless form:

q
= 1
1
1 +
h
where
q
=
qxL
k(T
fl
T1)
and
h
=
hL
k
(a.k.a. the Biot number).
ii. Cylindrical equation:
d
dr
_
r
dT
dr
_
= 0
Solution:
T = Alnr + B
Flux magnitude for heat transfer through a uid boundary layer at R
1
in series with conduc-
tion through a cylindrical shell between R
1
and R
2
:
|r q
r
| =
|T
fl
T
2
|
1
hR1
+
1
k
ln
R2
R1
iii. Spherical equation:
d
dr
_
r
2
dT
dr
_
= 0
Solution:
T =
A
r
+ B
(b) Constant generation
i. Cartesian equation:
k
d
2
T
dx
2
+ q = 0
Solution:
T =
qx
2
2k
+ Ax + B
ii. Cylindrical equation:
k
d
dr
_
r
dT
dr
_
+ r q = 0
Solution:
T =
qr
2
4k
+ Alnr + B
iii. Spherical equation:
k
d
dr
_
r
2
dT
dr
_
+ r
2
q = 0
Solution:
T =
qr
2
6k
+
A
r
+ B
2
(c) (Diusion only) rst-order homogeneous reaction consuming the reactant, so G = kC
i. Cartesian equation:
D
d
2
C
dx
2
kC = 0
Solution:
C = Ae

k
D
x
+ Be

k
D
x
or:
C = Acosh
_
_
k
D
x
_
+ B sinh
_
_
k
D
x
_
ii. Cylindrical and spherical solutions involve Bessel functions, but here are the equations:
D
d
dr
_
r
dC
dr
_
krC = 0
D
d
dr
_
r
2
dC
dr
_
kr
2
C = 0
2. Unsteady solutions without generation based on the Cartesian equation with constant k and c
p
:
T
t
=

2
T
x
2
where =
k
cp
.
(a) Uniform initial condition T = T
i
(or T = T

), constant boundary condition T = T

s
at x = 0,
semi-innite body; or step function initial condition in an innite body.
Solution is the error function or its complement:
T T
s
T
i
T
s
= erf
_
x
2

t
_
T T
i
T
s
T
i
= erfc
_
x
2

t
_
Semi-innite criterion:
L
2

t
2
Note: this also applies to a diusion couple, where two bodies of dierent temperatures (or
concentrations) are joined at x = 0 and diuse into each other; the boundary condition there is
halfway between the two initial conditions. This works only if the (thermal) diusivities are the
same.
(b) Fixed quantity of heat/solute diusing into a (semi-)innite body (same semi-innite criterion as
2a), no ux through x = 0, initial condition T = T
i
everywhere except a layer of thickness if
semi-innite or 2 if fully innite where T = T
0
.
Short-time solution consists of erfs at the interfaces, like a diusion couple.
Long-time solution is the shrinking Gaussian:
T = T
i
+
(T
0
T
i
)

t
exp
_

x
2
4t
_
3
(c) Uniform initial condition T = T
i
, constant boundary condition T = T
s
at x = 0 and x = L (or
zero-ux boundary condition q
x
= kT/x = 0 at x = L/2), nite body; or periodic initial
condition (weve covered sine and square waves) in an innite body.
Solution is the Fourier series:
T = T
s
+ (T
i
T
s
)

n=0
a
n
exp
_

n
2

2
t
L
2
_
sin
_
nx
L
_
For a square wave or uniform IC in a nite body, a
n
=
4
n
for odd n, zero for even n, T
s
is the
average temperature for a periodic situation or the boundary condition for a nite layer, L is the
half period of the wave or the thickness of the nite layer.
The n = 1 term dominates when

2

L
2
t 1.
(d) *Uniform initial condition T = T

, constant ux boundary condition at x = 0 q

x
= k
dT
dx
= q
0
,
semi-innite body (same semi-innite criterion as 2a).
Solution:
T = T

+
q
0
k
_
2
_
t

exp
_

z
2
4t
_
z
_
1 erf
z
2

t
_
_
(e) *Uniform initial condition T = T

, heat transfer coecient boundary condition at x = 0 q

x
=
k
dT
dx
= h(T
fl
T), semi-innite body (same semi-innite criterion as 2a).
Solution:
T T
fl
T

T
fl
= erfc
_
x
2

t
_
exp
_
hx
k
+
h
2
t
k
2
_
erfc
_
x
2

t
+
h

t
k
_
*These solutions are neither covered nor required, but are here for your edication and future reference.
3. Moving body
If a body is moving relative to a frame of reference at speed u
x
and conducting heat only in the
direction of motion, then the equation in that reference frame (for constant properties) is:
T
t
+ u
x
T
x
=

2
T
x
2
+
q
c
p
Note that this is the diusion equation with the substantial derivative instead of the partial derivative,
and nonzero velocity only in the x-direction. Recall the denition of the substantial derivative:
D
Dt
=

t
+u
Applied to temperature with u
y
= u
z
= 0:
DT
Dt
=
T
t
+ u
x
T
x
Therefore:
DT
Dt
=

2
T
x
2
+
q
c
p
When this reaches steady-state, so
T
t
= 0, then the solution in the absence of generation is
T = A + Be
uxx/
4