C
HAPTER
2
–
C
ONDUCTION
Consider a wall where the surface temperatures are such that T1>T2
qx = flow rate of heat in the x
direction (vector!) [units of Watts]
qx is dependent on THREE factors:
1. A - surface area perpendicular
to the flow of heat (i.e. y x z),
2. k – the capacity of material to transfer heat across its body
(thermal conductivity)
3. ∆T L – temperature gradient across the thickness of the wall, L
T1
T2
qx
L
These three factors are related
together through FOURIER’S LAW:
dx
dT
kA
q
x=
−
Q: why the negative sign?
Or, in terms of heat flux q” (q” = q/A)
dx
dT
k
q
x"
=
−
Although we are only considering 1D heat flow in this case, Fourier’s Law may be expressed in 3 dimensions:
+
+
−
=
∇
−
=
dz
dT
k
dy
dT
j
dx
dT
i
k
T
k
q
"
where ! is the del operator
NOTE: Fourier’s law is typically
written individually for each dimension over which heat flow occurs.
heat transfer always going to go from hot to cold. heat transfer is always a vector quantity
The driving force of conduction is the temperature gradient
dx
dT
[oC/m or K/m]
Average gradient:
L
T
T
run
rise
dx
dT
2−
1=
=
here
dx
<
0
dT
, thus qx>0
Instantaneous
gradient: may be some function of x
If T(x)=Px2+Qx+R
qx = -kA(dT/dx)
= -kA(2Px+Q)
T
1L
T
1T
2dT
dx
How can you determine if the temperature gradient is linear?
• Is there an internal source of heat? • Is the material through which heat
is flowing homogeneous?
o Is k constant?
• Does the surface area
perpendicular to the heat flow change with dx?
o Increasing cross-sectional area
= lower flux, more uniform
temperature as a function of x
r
q
wall T
x
T
q
H
EAT
T
RANSFER
P
ROPERTIES
Thermal Conductivity (k): (W/m•K)
High k values are good conductors Low k values are better insulators
e.g.
Material Temp
(K) (W/mK) k Material Temp (K) (W/mK) k
Brick 300 0.72 Air 300 0.026
Cork 300 0.039 Copper 300 401
Glass 300 1.4 Aluminum 300 237
More values available in Appendix A
k is often given as a constant, but it does vary with T–
get a value valid for
the T range you are
looking at.
A more accurate relationship is:
(
aT
)
k
k
=
o1
+
where ko and a are constants.
! a can be negative or positive
depending on the material.
For blends of materials A and B, use a weighted average to determine k:
(
)
B Aeff
ak
a
k
k
=
+
1
−
(a = fraction of material A in the blend)
For porous media consisting of a mixture of a stationary solid and a fluid, k depends on the geometry of the porous media (size distribution of pores, pore shape, etc.)
Minimum k:heat conducts sequentially through a fluid region of length ɛL and a solid region of length (1- ɛ)L: (fluid forms long, random fingers in solid)
q Solid
ks
Pore (fluid) kf
εL
(1−ε)L
(
)
f s
eff
k k
k
ε ε +
− =
1
1
min ,
Maximum k: heat can conduct through either a continuous fluid region of cross-section ɛw or a continuous solid region of
cross-section (1-ɛ)w: (interconnected solid)
(
)
sf
eff k k
k ,max = ε + 1− ε
Choose best approximation based on the material you are using.
For solids with uniformly distributed, non-contacting spherical pores
(ɛ<0.25), use Maxwell equation:
q Solid - ks
Pore (fluid) kf εL
(1−ε)L
(
)
(
)
sf s
s f
f s
s f
eff k
k k
k k
k k
k k
k
− +
+
− −
+ =
ε ε
2
2 2
min ,
Thermal Diffusivity (α): (m2/s)
p
c k
ρ α =
k = thermal conductivity (W/m•K)
cp = heat capacity of material (J/kg•K)
ρ = density of material (kg/m3)
Heat Capacity = the ability of a material to store heat (per kg)
Density = volume per kg
ρcp is the volumetric heat capacity (per m3 instead of per kg)
Q: What would ρcp and α be for a
good heat conductor? A bad heat conductor?
H
EAT
D
IFFUSION
E
QUATION
Consider a control volume (CV) as shown below (no bulk motion):
Heat balance on the control volume:
=
+
−
on accumulati
heat of
rate generation
heat of
rate out
flow
heat of
rate in
flow
heat of
rate
or, E&in − E&out + E&g = E&st
where
(
)
dx
dT
dz
dy
k
q
x=
−
⋅
qx qx+dx
qz
qz+dz
qy
qy+dy
dx
dz dy
First, consider transfer (Ein and Eout) To determine qx+dx, we must use the
Taylor Expansion:
( )
... ) ( ) ( 2 2 2 + ∆ ∂ ∂ + ∆ ∂ ∂ + = ∆ + x x F x x F x F x x FTaking only the first term of the
expansion (since ∆x = dx is small):
( )
q dx xq
qx dx x x
∂ ∂ + =
+
Thus, in the x direction, the difference
− = − out flow heat of in flow heat of rate rate E
E&in &out is given as:
( )
x y zx T k x dx q x q
qx x dx x ∂ ∂ ∂
∂ ∂ ∂ ∂ + = ∂ ∂ − = − +
In parallel, y x y z
T k y q
qy y dy ∂ ∂ ∂
∂ ∂ ∂ ∂ = − +
qz qz dz z k Tz ∂x∂y∂z
Summing up over all dimensions:
(
+ +) (
− + +)
==
− out x y z x+dx y+dy z+dz
in E q q q q q q
E& &
dxdydz z T k z y T k y x T k
x
∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂
The energy source term Eg can be written as:
Eg heattimecontent = q• dxdydz
∂ ∂ = & & ) ( ) (
where q is the rate of heat generation per unit volume (units W/m3)
Heat generation may occur via several different mechanisms, including:
" Electrical resistance, P = I2R [W]
" Nuclear reaction
" Chemical reaction (∆Hrxn )
" Direct thermal heating (furnaces)
Assuming no phase changes (i.e. only
sensible heat effects are present), the energy storage (or accumulation)
term is given as (again in parallel to the other derivations):
(
)
t T dxdydz c time content heatEst p
∂ ∂ = ∂ ∂ = ρ ) ( ) ( &
Substituting each of the derived terms in the energy balance E&in − E&out + E&g = E&st
and dividing out the δxδyδz term:
= + ∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂ q z T k z y T k y x T k
x & t
T cp
∂ ∂
ρ
General Heat Diffusion Equation
Solutions for other coordinates are:
Cylindrical Coordinates (r,θ , z)
Spherical Coordinates (r, θ ,ϕ )
Z r θ r θ ϕ
+
∂
∂
∂
∂
+
∂
∂
∂
∂
θ
θ
ϕ
T
k
r
r
T
kr
r
r
2 22 2
sin
1
1
t T c q T kr p ∂
∂ = + ∂ ∂ ∂ ∂
ρ
ϕ
ϕ
ϕ
ϕ
sin &sin 1 2 = + ∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂ q z T k z T k r r T kr r
r 2 θ θ &
S
IMPLIFYING
A
SSUMPTIONS
1. Semi-Infinite Assumption
Heat flow is always a vector quantity – it has a magnitude and direction
dx
dT
kA
q
x=
−
dy
dT
kA
q
y=
−
dz
dT
kA
q
z=
−
(x-axis) (y-axis) (z-axis)
But, do all the dimensions always matter to a given problem? # NO!
In cases where heat transfer in one or more dimension is negligible relative to heat transfer in another dimension, the semi-infinite assumption may be used to simplify the problem
# neglect edges of large slabs # neglect small contact areas
# neglect smaller cross-section areas
Example: Consider our wall thought experiment but add the floor and the ceiling. Find the wall temperature.
Ceiling = 16°C
air = 20°C outside = -10°C
Floor = 18°C
Over 98% of the wall’s cross-sectional area, the wall is in contact with only the inside and outside air, ∆∆∆∆T = 30°C
At the ceiling contact (1% of the wall’s cross-sectional area), ∆∆∆∆T = 4°C
At the floor contact (1% of the wall’s cross-sectional area), ∆∆∆∆T = 2°C
Does heat flow between the ceiling,
the floor, the wall, and the air? # YES
Is it significant in the context of the problem # NOT REALLY
wall
2. Steady State Assumption
This assumption is useful in situations in which there is a steady heat flow through a plane wall or slab
INSIDE OUTSIDE
air = 20°C air = -10°C
While the inside wall feels cool to our hand, the temperature is constant. Outside the wall will be >-10°C since the wall is heated from the inside.
The process is at steady-state since none of the temperatures vary as a function of time.
In this case, δT/δt = 0 such that
0
= +
∂ ∂ ∂
∂
+
∂ ∂ ∂
∂
+
∂ ∂ ∂
∂ ⋅
p
c q z
T z
y T y
x T
x
ρ
α
&wall
There are two conditions under which you can apply this assumption:
1) An external driving force is acting to either add or remove heat from the surfaces of interest.
# In the case of the wall, the
furnace adds heat as it flows out of the room to keep the room
temperature constant while the air flow continuously replaces the air adjacent to the wall to keep the outdoor temperature constant
! TRUE STEADY STATE
2) When T1 and T2 do change over time, if the rate of change is
sufficiently slow or the time period we are considering is sufficiently short, we can assume T1 and T2 to
be constant over this period
! “PSEUDO” STEADY STATE
K
EY
P
HRASES
T
O
C
HECK
F
OR
FOR STEADY STATE ASSUMPTION
TRUE STEADY STATE:
“with a uniform temperature gradient” “with a constant heat flux”
“with constant surface temperatures”
PSEUDO-STEADY STATE:
“find the instantaneous temperature” “find the instantaneous heat flow”
“find the temperature after x minutes”
FOR SEMI-INFINITE ASSUMPTION “thin”
“slab” “wall”
“narrow” “plane”
“negligible thickness” “neglect edge effects”
P
ROBLEM
A
SSESSMENT
= + ∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂ q z T k z y T k y x T kx & t
T cp
∂ ∂
ρ
the result will be T(x,y,z,t) (T profile)
ASK YOURSELF:
1. Steady State? ∂∂Tt = 0
2. No Heat Generation?
q
&
=
0
3. 2-D Heat Flow? ∂∂Tz = 0
4. 1-D Heat Flow? , ∂ = 0
∂ ∂ ∂ z T y T
5. k constant? 2
2 x T k x T k x ∂ ∂ = ∂ ∂ ∂ ∂ Then, solve the simplified equation for
T and solve
dx
dT
kA
q
x=
−
for heat flow.MUST USE Heat Diffusion Equation
" To determine the temperature
profile of a system (T(x,y,z,t)) –
integrate over significant variable(s)
" To determine heat flows if k or A
vary along a dimensional axis
Use Heat Diffusion Equation OR Energy Balance Ein – Eout + Eg = Est
" To determine heat flows in cases
with volumetric heat generation
" To account for non steady-state
heat flow within the object (Est ≠ 0)
You only need Fourier’s Law if:
" Heat flow is steady state AND
" Cross-sectional area is constant AND
" k is constant over full volume AND
" Temperature profile is known (i.e.
T(x)=ax2 + bx + cz) or assumed to be a certain shape (line, parabola...)
! T(x,y,z,t) solved for you!
B
OUNDARY
C
ONDITIONS
From calculus, remember that every time you integrate a function, one integration constant is generated.
If 2 0 2
=
∂ ∂
x T
# C1
x T
= ∂
∂
# T = C1x + C2
To define the temperature profile (i.e. to find the integration constants C1
and C2), we need to know T(x) at two points at the boundary of our object (i.e. at x=0 and x=L in a wall or x=0 and x=ro in a cylinder)
! 2 equations in 2 unknowns
e.g. If T=T1 at x=0 and T=T2 at x=L
x=0 # T(x=0)=T1= C2
x=L # T(x=L)=T2=C1L+C2 #
Therefore, the temperature profile is:
L
T
T
C
1=
2−
11 1
2 2
1
)
(
x
T
L
T
T
C
x
C
x
T
=
+
=
−
+
The heat equation has three 2nd
derivatives as a function of position (d2T/dx2, d2T/dy2, d2T/dz2) and one 1st
derivative as a function of time(dT/dt)
= +
∂ ∂
+
∂ ∂
+
∂ ∂
q z
T y
T x
T
k &
2 2
2 2
2 2
t T cp
∂ ∂
ρ
Thus, to solve for T(t), we need 1
initial condition (generally, T or q at
t=0, the start of transient heat flow) to find 1 integration constant
" Initial Conditions # time
oDefined in problem statement
To solve for T(x,y,z), we need 2
boundary conditions per dimension of heat flow (T or q defined at x=0 and
x=L) to find 2 integration constants
" Boundary Conditions #### position
oDefined in problem but typically
one of three common conditions
Typical Boundary Conditions
1) Constant surface temperature (Dirichlet condition)
T(0,t) = Ts
2) Constant surface heat flux (Neumann condition)
(a) Finite heat flux at surface - heater
(b) Adiabiatic/insulated surface
3) Convection surface condition
Ts
T(x,t) x
T(x,t) x
qs”
x "
0
s x
q dx
dT
k =
−
=
0
0
=
=
x
dx dT
x
T(0,t)
T∞, h
( (0, )) 0
t T T
h dx
dT k
x
−
=
− ∞
=
T(x,t)
T(x,t)
EXAMPLE
A plane wall has constant
properties, no internal heat generation, and is initially at a uniform
temperature of 20°C. The back of the wall is attached to a thin electric
heater, backed by insulation, which is initially off. Suddenly, the outer surface is heated by a fluid at 50°C with a convection coefficient of 1000W/m2K. 1) Write the simplified heat diffusion equation to find the temperature distribution in the wall a few seconds after the fluid flow begins.
2) Specify the boundary and initial conditions.
3) Sketch the temperature profile throughout the system (a) before heat or fluid flow; (b) with fluid flow but no heater; (c) with heater and fluid flow (assume the q” of heater is higher than q” of fluid). Draw curves just after the heater/flow is started, in the middle of the heating, and at steady state.
4) Sketch the heat flux at x=L as a function of time under the same conditions.