q = −k∇T
∂T ∂2T
ρc = k 2 or Tt = κTxx
∂t ∂x
κ = k/(ρc) (m2 /sec) the diffusivity, ρ the density (km/m3 ), c the
specific heat.
T0 T1
Tt = κTxx
L x
L T0
Figure : Simple Rod Heating
Particular Cases:
1. L = 20m, T0 = 10C ◦ , T1 = 150C ◦
2. L = 2cm, T0 = 20C ◦ , T1 = 25C ◦
Scaling
We write
x = Lx 0 , t = τ t 0 , T = T0 + (T1 − T0 )T 0 (x 0 , t 0 )
with τ as yet unknown. This gives
h κτ i
Tt00 = 2 Tx0 0 x 0
L
L2
There is one dimensionless group [·] and if we choose τ = κ we
get the dimensionless problem
t'
0 1
Tt00 = Tx0 0 x 0
1 0 1 x'
n
D
T = Ts Tt = κ∇2 T + Q T = T0
∂D
L
Tt = κ∇2 T + Q, in D
where κ = k/(ρc), Q = q/(ρc)
Assume Newtonian cooling from the boundary so
We write
ht κi
0 02 q0 t 0
θt 0 − ∇ θ= q 0 (x 0 , t 0 ),
L2 ρc∆T
There are three dimensionless groups ([·]) with just two scales
(t0 , ∆T ) to choose. We can equate two of the [·] terms to unity
leaving one dimensionless group defining the problem. Which two?
At this stage it is important to think about the physics.
The temperature rise rate is due to the heat input and conduction
carries the heat out to the surface so the choice
L2 q0 t0 q0 L2
t0 = , ∆T = ≡ ,
κ ρc κρc
seems sensible here. A good choice is important.
Does the scaling make physical sense? If L is large would you
expect ∆T to be very large?...
The Defining Dimensionless Group of the problem
θt 0 − ∇02 θ = q 0 (x 0 , t 0 ),
Note:
I In order to experimentally determine the time to reach steady
state tst and the final temperature rise one must duplicate (a)
the geometry and (b) the dimensionless cooling parameter
µ = γLk . Explicitly we can write the
L2
tst = fn(µ, geometry )
κ
where the function fn is related to θ.
I Note that the required time increases in proportion to L2 .
I To determine the effect of µ on tst one could solve the
equations analytically, numerically or experimentally.
Further Comments
∂T (x, t) ∂ 2 T (x, t) q0
=κ 2
+ ,
∂t ∂x ρcA