Convection BL YouTube

Theodore Ong

April 29, 2021

Part I
Some Links before we start
https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_equations

Matrices in LaTeX
https://www.overleaf.com/learn/latex/Matrices

Tensors in LaTeX

Navier Stokes Equations

https://www.comsol.com/multiphysics/navier-stokes-equations
https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_equations

Github
https://github.com/theodoreOnzGit/heatTransferTheory_YouTube

Part II
Hydrodynamics
1 Navier Stokes Equations
∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
u + u u + v u + w u − ν( 2 u + 2 u + 2 u) = − + gx
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂x

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
v + u v + v v + w v − ν( 2 v + 2 v + 2 v) = − + gy
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂y

1
∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
w + u w + v w + w w − ν( 2 w + 2 w + 2 w) = − + gz
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂z

2 Boundary Layer Equations (Laminar)

∂P
0=−
∂y

∂ ∂ ∂ ∂2 ∂P
u + u u + v u − ν 2 u = (− )
∂t ∂x ∂y ∂y ∂x
∂u ∂v
+ =0
∂x ∂y

2.0.1 Blasius Similarity Solution

f 0 = 0 at η = 0
f = 0 at η = 0
f 0 = 2 at η → ∞
f 00 f + f 000 = 0

u∞ 0
u= f
2
r
1 u∞
η= y
2 νx
[not derived previously:]
r
∂u u∞ u∞ 00
τx = µ = ρν f
∂y 4 νx
Welty, J., Rorrer, G. L., & Foster, D. G. (2014).
Fundamentals of momentum, heat, and mass transfer. John Wiley & Sons.

2.1 Integral Solution by Theodore Von Karman

Final form of Von Karman equation:
Z δp Z δp
τx ∂u∞ ∂
=( ) (u∞ − u) dy + u(u∞ − u) dy
ρ ∂x 0 ∂x 0

2
3 Solutions to BL
3.1 Von Karman Results (approximate solution)
δp (x) 4.64095
= √
x Rex
0.6464
Cf x ≡ √
Rex
1.2928
Cf L ≡ √
ReL

3.2 Blasius Results (Exact solution)

δp (x) 5
=√
x Rex
0.664
Cf x ≡ √
Rex
1.328
Cf L ≡ √
ReL
7.2% off for BL thickness, and 3% off for skin friction coeff. Pretty good!
This shows that Von Karman method is pretty good, if you can’t use
Blasius results

Welty, J., Rorrer, G. L., & Foster, D. G. (2014).

Fundamentals of momentum, heat, and mass transfer. John Wiley & Sons.

Part III
Energy Equations
h ≡ e + PV
Enthalpy per unit mass (Specific enthalpy) P is pressure V is specific
volume e is specific internal energy
In thermodynamics, U is often used for internal energy, but we avoid
using that because it gets confused with u which is x velocity. So we use e
instead.
Enthalpy Balance

3
 ∂ ∂ ∂ ∂
h+u h+v h+w h = conduction+heat generation+dissipation+radiation
∂t ∂x ∂y ∂z
(1)
neglect heat generation and dissipation terms. Also neglect radiation
heat transfer
Fourier’s law

00
qcond = −k∇T
Conduction term:

00
conduction = −∇qcond
00
Heat flux = qcond ;
Heat flux is heat energy Transferred per unit area

∂ ∂ ∂ ∂ 00
h + u h + v h + w h = −∇qcond
∂t ∂x ∂y ∂z
∂ ∂ ∂ ∂
h + u h + v h + w h = −∇(−k∇T )
∂t ∂x ∂y ∂z
∂ ∂ ∂ ∂
h + u h + v h + w h = ∇(k∇T )
∂t ∂x ∂y ∂z
heat conduction coeffcient is isotropic (it’s a scalar), we also assume k
does not change with x,y or z

∂ ∂ ∂ ∂
h + u h + v h + w h = k∇2 T
∂t ∂x ∂y ∂z

∆h = ρcp ∆T
h − href = ρcp (T − Tref )
h = ρcp (T − Tref ) + href
We note, reference temperature and enthalpy are constant with time, x
y z etc.

∂ ∂ ∂ ∂
ρcp T + u ρcp T + v ρcp T + w ρcp T = k∇2 T
∂t ∂x ∂y ∂z
We assume, volumetric heat capacity does not change with x,y or z and
time

4
 kg J
ρcp = volumetric heat capacity =
m3 kg • k
∂ ∂ ∂ ∂
ρcp T + u ρcp T + v ρcp T + w ρcp T = k∇2 T
∂t ∂x ∂y ∂z
∂ ∂ ∂ ∂ k 2
T +u T +v T +w T = ∇ T
∂t ∂x ∂y ∂z ρcp
∂ ∂ ∂ ∂
T +u T +v T + w T = α∇2 T
∂t ∂x ∂y ∂z
k
α=
ρcp

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2
T +u T +v T + w T = α( 2 T + 2 T + 2 T )
∂t ∂x ∂y ∂z ∂x ∂y ∂z

Part IV
Thermal Laminar BL Equations
First, we assume, 2D, so z derivatives are zero.

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2
T +u T +v T + w T = α( 2 T + 2 T + 2 T )
∂t ∂x ∂y ∂z ∂x ∂y ∂z

∂ ∂ ∂ ∂2 ∂2
T +u T +v T = α( 2 T + 2 T )
∂t ∂x ∂y ∂x ∂y
We need to introduce relative temperature

Trel = T − Ts

∂ ∂
(T − Ts ) = (T )
∂x ∂x
so long as Ts is constant

∂ ∂ ∂ ∂2 ∂2
(T − Ts ) + u (T − Ts ) + v (T − Ts ) = α( 2 (T − Ts ) + 2 (T − Ts ))
∂t ∂x ∂y ∂x ∂y

5
 Let’s see which terms we can neglect

T − Ts
θ=
T∞ − Ts
u
u∗ =
u∞
Let’s nondimensionaise temrperature:

∂ ∂ ∂ ∂2 ∂2
θ(T∞ −Ts )+u θ(T∞ −Ts )+v θ(T∞ −Ts ) = α( 2 θ(T∞ −Ts )+ 2 θ(T∞ −Ts ))
∂t ∂x ∂y ∂x ∂y
Divide throughout,

∂ ∂ ∂ ∂2 ∂2
θ + u θ + v θ = α( 2 θ + 2 θ)
∂t ∂x ∂y ∂x ∂y
x
x∗ =
L
y
y∗ =
δt
Note: in general, thermal boundary layer is different from momentum
boundary layer

∂ 1 ∂ 1 ∂ 1 ∂2 1 ∂2
θ+u θ + v θ = α( θ + θ)
∂t L ∂x∗ δt ∂y ∗ L2 ∂(x∗ )2 δt2 ∂(y ∗ )2
u
u∗ =
u∞

∂ u∞ ∂ 1 ∂ 1 ∂2 1 ∂2
θ + u∗ θ + v θ = α( θ + θ)
∂t L ∂x∗ δt ∂y ∗ L2 ∂(x∗ )2 δt2 ∂(y ∗ )2
We assume no natural convection

∂ u∞ ∂ 1 ∂ 1 ∂2 1 ∂2
θ + u∗ ∗
θ+v ∗
θ = α( 2 ∗
θ+ 2 θ)
∂t L ∂x δt ∂y L ∂(x ) 2 δt ∂(y ∗ )2
If we were to use momentum nondimensionalisation:
y
y∗ =
δp

6
 ∂u ∂v
+ =0
∂x ∂y
u∞ ∂u∗ vc ∂v ∗
+ =0
L ∂x∗ δp ∂y ∗
u∞ δp ∂u∗ ∂v ∗
+ =0
vc L ∂x∗ ∂y ∗
u∞ δp
= O(1)
vc L
u∞ δp
= O(1)vc
L
we can just set:

u∞ δp
vc = O(1)
L
so that
v
v∗ = = O(1)
vc
or else

u∞ δp
vc =
L
So we substitute this back in:

∂ u∞ ∂ u∞ δp ∗ ∂ 1 ∂2 1 ∂2
θ + u∗ θ + v θ = α( θ + θ)
∂t L ∂x∗ Lδt ∂y ∗ L2 ∂(x∗ )2 δt2 ∂(y ∗ )2

t∗
t=
tc
if you talk about hydrodynamic timescales:
L
tc =
u∞
we have to use thermal BL timescales...
For simplicity, we may not want to consider transient timescales in flat
plate BL analysis
We assume steady state

7
 u∞ ∂ u∞ δp ∗ ∂ 1 ∂2 1 ∂2
u∗ θ + v θ = α( θ + θ)
L ∂x∗ Lδt ∂y ∗ L2 ∂(x∗ )2 δt2 ∂(y ∗ )2
∂ δp ∗ ∂ L 1 ∂2 1 ∂2
u∗ θ + v θ = α ( θ + θ)
∂x∗ δt ∂y ∗ u∞ L2 ∂(x∗ )2 δt2 ∂(y ∗ )2
We can make this observation:

L2 >> δt2

∂ δp ∂ L 1 ∂2
u∗ ∗
θ + v∗ ∗ θ = α ( θ)
∂x δt ∂y u∞ δt2 ∂(y ∗ )2
assume steady state, and performing BL analysis: Also, no dissipation
or heat generation

∂ ∂ ∂2
u T +v T = α( 2 T )
∂x ∂y ∂y

Part V
Solution Procedures - Constant
Temp Forced Conv BL
∂ ∂ ∂2
u T + v T = α 2T
∂x ∂y ∂y
∂P
0=−
∂y
∂ ∂ ∂2 ∂P
u u + v u − ν 2 u = (− )
∂x ∂y ∂y ∂x
∂u ∂v
+ =0
∂x ∂y
Solution Procedures:
• Similarity Solution (Blasius Style) [GOLD STANDARD]

• Integral Solution (Von Karman Style)

• Computational Fluid Dynamics (MultiPhysics)

8
4 back to similarity Solution
In case of no pressure gradient,

∂ ∂ ∂2
u T + v T = α 2T
∂x ∂y ∂y

∂ ∂ ∂2
u u+v u=ν 2 u
∂x ∂y ∂y
∂u ∂v
+ =0
∂x ∂y
And nondimensionalising

f 0 = 0 at η = 0

f = 0 at η = 0
f 0 = 2 at η → ∞
f 00 f + f 000 = 0

u∞ 0
u= f
2
r
1 u∞
η= y
2 νx
[not derived previously:]
r
∂u u∞ u∞ 00
τx = µ = ρν f
∂y 4 νx

4.0.1 Reynold’s analogy

Hey these equations look so similar

∂ ∂ ∂2
u T + v T = α 2T
∂x ∂y ∂y

∂ ∂ ∂2
u u+v u=ν 2 u
∂x ∂y ∂y
Let’s copy/paste the solution... (under certain constraints)

9
 ν=α

ν
Pr =
=1
α
If you wanna change up the equations:

∂ ∂ ν ∂2
u T +v T = T
∂x ∂y P r ∂y 2
If Pr=1,

∂ ∂ ∂2
u T + v T = ν 2T
∂x ∂y ∂y
Use the relative temperature trick, and change the variable to θ

∂ ∂ ∂2
u θ + v θ = ν 2θ
∂x ∂y ∂y
If we nondimensionalise the momentum equation

∂ ∗ ∂ ∂2
u u u∞ + v u∗ u∞ = ν 2 u∗ u∞
∂x ∂y ∂y

∂ ∗ ∂ ∂2
u u + v u∗ = ν 2 u∗
∂x ∂y ∂y
We already proved we can reduce the momentum BL equations as fol-
lows:
f 00 f + f 000 = 0
And we note that we can set:

θ = u∗
Note that
u∞ 0
u= f
2

f0
u∗ =
2
Under Reynold’s analogy we can say:

f0
θ=
2

10
 We can pretty much adapt the Blasius solution to thermal boundary
layer... (Reynold’s analogy under Pr=1)
We assume that the nondimensional temperature profile and nondimen-
sional BL profile (and BCs) are exactly equal.
Based on that, we can get the temperature profile and heat flux

f0
θ=
2
f0
T − Ts = (T∞ − Ts )
2
What about heat flux:

∂T
q 00 = −k
∂y
note:
r
∂u u∞ u∞ 00
= f
∂y 4 νx

∂u∗ u∞
r
u∞ u∞ 00
= f
∂y 4 νx

∂u∗
r
1 u∞ 00
= f
∂y 4 νx
under reynold’s analogy
u∗ = θ

r
∂θ 1 u∞ 00
= f
∂y 4 νx
r
∂θ 1 u∞ 00
= f
∂y 4 νx
r
∂θ 1 u∞ x 00
= f
∂y 4 νx2
∂θ 1p 1
= Rex f 00
∂y 4 x
∂(T − Ts ) T∞ − Ts p 1
= Rex f 00
∂y 4 x

11
 ∂T T∞ − Ts p 1
= Rex f 00
∂y 4 x
Heat flux:

∂T
q 00 = −k
∂y
k T∞ − Ts p
q 00 = − Rex f 00
x 4
Heat trf Coefficient and Nu

q 00 = −h(T∞ − Ts )
Substitute

k T∞ − Ts p
h(T∞ − Ts ) = Rex f 00
x 4
k f 00 p
h= Rex
x 4
Define Nusselt Number:

hx
N ux ≡
k

f 00 p
N ux = Rex
4
What is the value of f” (look at textbook, or solve it yourself)
at y=0
f 00 = 1.328

1.328 p p
N ux = Rex = 0.332 Rex
4

5 What if Pr is not 1?
5.1 Similarity Solution by Pohlhausen
Bejan, A. (2013). Convection heat transfer. John wiley & sons.
Welty, J., Rorrer, G. L., & Foster, D. G. (2014).
Fundamentals of momentum, heat, and mass transfer. John Wiley & Sons.

12
 From Blasius Solution
And nondimensionalising

f 0 = 0 at η = 0

f = 0 at η = 0
f 0 = 2 at η → ∞
f 00 f + f 000 = 0

u∞ 0
u= f
2
Some useful derivatives previously derived (Bejan, 2013,Welty et al., 2014)
From hydrodynamic BL:
∂ ∂ 1p y∗ η
∗
η = ∗
Reδ √ = ∗
∂y ∂y 2 x∗ y

∂ ∂ 1p y∗
η = Reδ √
∂x∗ ∂x∗ 2 x∗
∂ ∂ ∂2
u T + v T = α 2T
∂x ∂y ∂y
We nondimensionalised the energy equation during scaling analysis...

∂ δp ∗ ∂ L 1 ∂2
u∗ θ + v θ = α ( θ)
∂x∗ δt ∂y ∗ u∞ δt2 ∂(y ∗ )2
We just need to convert it to include similarity variables...
We start by making substitutions

f0
u∗ =
2
r
1 νu∞
v= (ηf 0 − f )
2 x
u∞ δp
vc =
L
r
∗ u∞ δp 1 νu∞
v =− (ηf 0 − f )
L 2 x

13
 r
∗1L ν
v = (ηf 0 − f )
2 δp xu∞
1 L − 12
v∗ = Rex (ηf 0 − f )
2 δp
subs into:

∂ δp ∗ ∂ L 1 ∂2
u∗ θ + v θ = α ( θ)
∂x∗ δt ∂y ∗ u∞ δt2 ∂(y ∗ )2

f0 ∂ L 1 ∂2
 
δp 1L 1
0 ∂
θ + − Rex
2
(ηf − f ) θ = α ( θ)
2 ∂x∗ δt 2 δp ∂y ∗ u∞ δt2 ∂(y ∗ )2

f0 ∂ L 1 ∂2
 
L 1 21 0 ∂
θ + − Rex (ηf − f ) θ = α ( θ)
2 ∂x∗ δt 2 ∂y ∗ u∞ δt2 ∂(y ∗ )2
Now we need to substitute out y ∗ and x∗ for η
y
y∗ =
δt
x
x∗ =
L
we note x is the same definition as the nondimensionalised x∗ used in
∗

deriving hydrodynamic BL, so we use it again...

So we can use:

∂ ∂ 1p y∗
η = Reδ √
∂x∗ ∂x∗ 2 x∗
1p y ∗ −1 η
= Reδ √ ∗
=
2 x∗ 2x −2x∗
Replace left most term...

∂ ∂η ∂ η ∂
∗
= ∗
=− ∗
∂x ∂x ∂η 2x ∂η

f0 ∂ L 1 ∂2
 
L 1 − 12 0 ∂
θ + Re x (ηf − f ) θ = α ( θ)
2 ∂x∗ δt 2 ∂y ∗ u∞ δt2 ∂(y ∗ )2

14
 f0 L 1 ∂2
 
η ∂ L 1 − 12 ∂
(− ∗ ) θ + Rex (ηf 0 − f ) θ = α ( θ)
2 2x ∂η δt 2 ∂y ∗ u∞ δt2 ∂(y ∗ )2

f0 L 1 ∂2
 
η 0 L 1 − 12 0 ∂
(− ∗ )θ + Rex (ηf − f ) θ = α ( θ)
2 2x δt 2 ∂y ∗ u∞ δt2 ∂(y ∗ )2

f0 L ∂2
 
η 0 1 − 12 0 ∂
(− ∗ )θ + L Rex (ηf − f ) θ=α ( θ)
2 2x 2 ∂y u∞ ∂y 2
What about the y terms?
r
1y 1 1 u∞
η= (Rex ) 2 = y
2x 2 νx
∂ ∂η ∂
=
∂y ∂y ∂η
r
∂η 1 u∞ η
= =
∂y 2 νx y
For RHS...

∂2θ ∂ ∂η ∂θ ∂ ∂η 0
2
= ( )= ( θ)
∂y ∂y ∂y ∂η ∂y ∂y

∂θ0 ∂η ∂ 2 η 0
= + θ
∂y ∂y ∂y 2

∂2η ∂ η
2
= ( )=0
∂y ∂y y

∂2θ ∂θ0 ∂η ∂η ∂θ0 ∂η 00 η

2
= = = θ ( )
∂y 2 ∂y ∂y ∂y ∂η ∂y y2
Substitution back in:

f0 L η 2 00
 
η 1 − 12 η
(− ∗ )θ + L − Rex (ηf − f ) θ0 = α
0 0
( θ )
2 2x 2 y u∞ y 2
tidying up:

ηf 0 0 L η 2 00
 
1 − 12 η
(− ∗ )θ + L Rex (ηf − f ) θ0 = α
0
( θ )
4x 2 y u∞ y 2

15
 ηf 0 0 1 η − 12 0 0 L η 2 00
(− )θ + LRex (ηf − f )θ = α ( θ )
4x∗ 2y u∞ y 2

ηf 0 0 1 η − 12 0 0 0 L η 2 00
(− )θ + LRex (ηf θ − f θ ) = α ( θ )
4x∗ 2y u∞ y 2

ηf 0 0 1 η − 12 0 0 1η − 12 0 L η 2 00
(− )θ + LRex ηf θ − LRex f θ = α ( θ )
4x∗ 2y 2y u∞ y 2

1 1η − 12 1η − 12 L η 2 00
θ0 ηf 0 (− + LRex ) − LRex f θ 0
= α ( θ )
4x∗ 2 y 2y u∞ y 2
simplifying...

L η 2 00
r
0 01 1 u∞ −1 1η −1
θ ηf (− ∗ + LRex 2 ) − LRex 2 f θ0 = α ( θ )
4x 4 νx 2y u∞ y 2

1 1p − 12 1η − 12 L η 2 00
θ0 ηf 0 (− + Rex LRex ) − LRex f θ 0
= α ( θ )
4x∗ 4x 2y u∞ y 2

1 L 1η − 12 L η 2 00
θ0 ηf 0 (− + ) − LRex f θ 0
= α ( θ )
4x∗ 4x 2y u∞ y 2

1 1 1η − 12 L η 2 00
θ0 ηf 0 (− + ) − LRex f θ 0
= α ( θ )
4x∗ 4x∗ 2y u∞ y 2

1η −1 L η 2 00
− LRex 2 f θ0 = α ( θ )
2y u∞ y 2

1 1 1 12 −1 L η 2 00
− Rex LRex 2 f θ0 = α ( θ )
22x u∞ y 2

1 1 0 L η 2 00
− f θ = α ( θ )
4 x∗ u∞ y 2
1 1 0 L 1 Rex 00
− ∗
fθ = α ( θ )
4x u∞ 4 x2
1 1 0 1 1 Rex 00
− ∗
fθ = α ( θ )
4x u∞ 4 xx∗

16
 1 1 1 Rex 00
− f θ0 = α ( θ )
4 u∞ 4 x
1 Rex 00
−f θ0 = α ( θ )
u∞ x
1 u∞ x 00
−f θ0 = α ( θ )
u∞ xν
α 00
−f θ0 = θ
ν
1 00
−f θ0 = θ
Pr

θ00 + P r f θ0 = 0

5.2 Similarity Solution BCs

2 Bcs:

y → ∞ ; T → T∞

η→∞; θ→1
Reminder

T − Ts
θ=
T∞ − Ts

y = 0; T = Ts
η = 0 ;θ = 0

5.3 solving the equation

θ00 + P r f θ0 = 0

∂θ0
+ P r f θ0 = 0
∂η
∂θ0
= −P r f θ0
∂η

17
 ∂θ0
= −P r f θ0
∂η

1 ∂θ0
= −P r f
θ0 ∂η
∞ Z ∞
1 ∂θ0
Z
dη = − P r f dη
0 θ0 ∂η 0
Z θ0 |η=∞ Z ∞
1 0
dθ = − P r f (η)dη
θ0 |η=0 θ0 0
Z ∞
θ0 |η=∞
loge ( 0 )=− P r f (η)dη
θ |η=0 0
Z ∞
θ0 |η=∞
= exp(− P r f (η)dη)
θ0 |η=0 0
What if η is not infinity,
Z θ0 |η0 =η Z η
1 0
dθ = − P r f (η 0 )dη 0
θ0 |η0 =0 θ0 0
Z ∞
θ0 |η0 =∞
loge ( 0 )=− P r f (η)dη
θ |η=0 0

θ0 |η0 =η
Z η
= exp(− P r f (η 0 )dη 0 )
θ0 |η=0 0
Z η
θ0 |η = θ0 |η=0 exp(− P r f (η 0 )dη 0 )
0
Integrate again,
Z η Z η Z η
0 0 0 0
θ (η )dη = θ |η0 =0 exp(− P r f (η 0 )dη 0 )dη 0
0 0 0
Z η Z η
θ(η) = θ0 |η=0 exp(− P r f (η 0 )dη 0 )dη 0
0 0
use other conventional dummy variables to avoid confusion
Z γ=η Z β=η
θ(η) = θ0 |η0 =0 exp(− P r f (β)dβ)γ
γ=0 β=0

18
One problem: we don’t know what θ0 |η0 =0 is
we use the boundary condition:

θ = 1 at η = inf inity
the dummy variable representing η is γ
Z γ=∞ Z β=η
0
θ(η = ∞) = θ |η=0 exp(− P r f (β)dβ)γ
γ=0 β=0
Z γ=∞ Z β=η
1= θ0 |η=0 exp(− P r f (β)dβ)γ
γ=0 β=0
Z γ=∞ Z β=η
0 −1
θ (0) = exp(− P r f (β)dβ)γ
γ=0 β=0
Z γ=∞ Z β=η −1
0
θ (0) = exp(− P r f (β)dβ)γ
γ=0 β=0

using correct mathematical symbols:

Z γ=∞ Z β=γ −1
0
θ (0) = exp(− P r f (β)dβ)γ
γ=0 β=0
Z γ=η Z β=γ
0
θ(η) = θ (0) exp(−P r f (β)dβ)γ
γ=0 β=0

We want to find out heat flux...

∂T
q 00 = −k |y=0
∂y
∂(T − Ts )
q 00 = −k |y=0
∂y
∂θ
q 00 = −k(T∞ − Ts ) |y=0
∂y

q 00 ∂θ
h= = k |y=0
−(T∞ − Ts ) ∂y
h ∂θ
= |y=0
k ∂y

19
 h ∂η ∂θ
= |y=0
k ∂y ∂η
h η ∂θ
= |y=0
k y ∂η
r
h 1 u∞ ∂θ
= |y=0
k 2 νx ∂η
r
h 1 u∞ 0
= θ (0)
k 2 νx
r
h 1 u∞ x 0
= θ (0)
k 2 νx2
hx 1p
= Rex θ0 (0)
k 2
1p
N ux = Rex θ0 (0)
2
Z γ=∞ Z β=γ −1
1p
N ux = Rex exp(− P r f (β)dβ)γ
2 γ=0 β=0

In textbook expression:
Z γ=∞ Z β=γ −1
p 1
N ux = Rex exp(− P r f (β)dβ)γ
γ=0 β=0 2
What is the value of:
Z γ=∞ Z β=γ −1
1
exp(− P r f (β)dβ)γ
γ=0 β=0 2
Pohlhausen worked it out for Pr>0.5:
Z γ=∞ Z β=γ −1
1 1
exp(− P r f (β)dβ)γ ≈ 0.332P r 3
γ=0 β=0 2
Liquid metals P r < 0.1
Air P r = 0.72
Water P r = 7

20
 Pr number says that for liquids of same viscosity, a higher prandtl num-
ber means lower thermal diffusivity
Thermal diffusivity intuition, high thermal conductivity means high ther-
mal diffusivity, all else constant.
1 1
N ux = 0.332Rex2 P r 3

Z L
1
N uL = N ux dx
L 0
Z L
1 1 1
N uL = 0.332Rex2 P r 3 dx
L 0
Z L
1 1 1
N uL = P r 3 0.332Rex2 dx
L 0
1 1
N uL = 0.664ReL2 P r 3
Note: For the above solutions, use film temperature to evaluate thermal
properties!

Ts + T∞
Tf ilm =
2
eg. cp , µ etc..

6 Integral Solutions to Thermal BL Constant Temp

Flat Plate
In the momentum BL, we considered a control volume in the BL to get:
Z δp Z δp
τx ∂u∞ ∂
=( ) (u∞ − u) dy + u(u∞ − u) dy
ρ ∂x 0 ∂x 0
Now for thermal BL, we can also consider control volume too (Welty
et al., 2014).
We can consider again four sides of the control volume and perform an
enthalpy balance, neglecting fluid kinetic energy:

Q̇ − Ẇ = net enthalpy f lows out + ∆ enthalpy w.r.t time

W is Work done by system, if Q+W then W is work done on system

21
 Q̇ = q 00 dA

00 ∂T
qwall = −k |y=0
∂y

dA = lz ∆x
For conduction, ignore conduction within fluid. (important assumption)
2nd assumption, no work done W=0,
Let’s consider the enthalpy flows and assume steady state..
Consider net enthalpy flows in:
neglecting kinetic energy
inflow of enthalpy through left
Z δt
= ρuhlz dy|x
0
Z δt
= ρucp (T − Tref )lz dy|x
0
outflow of enthalpy through the right
Z δt
= ρuhlz dy|x+∆x
0
Z δt
= ρucp (T − Tref )lz dy|x+∆x
0
inflow of enthalpy through the top
Specific enthalpy:

h∞ = cp (T∞ − Tref )
mass flowrate into system:
Z δt Z δt
ṁ = ρulz dy|x+∆x − ρulz dy|x
0 0
put this toegther to find enthalpy inflow through top:
Z δt Z δt
ṁh∞ =( ρulz dy|x+∆x − ρulz dy|x )cp (T∞ − Tref )
0 0
Let’s subs everything back:

22
 Q̇ − Ẇ = net enthalpy f lows out + ∆ enthalpy w.r.t time
Z δt Z δt
∂T
−k |y=0 lz ∆x − 0 = −( ρulz dy|x+∆x − ρulz dy|x )cp (T∞ − Tref )
∂y 0 0

Z δt Z δt
+ ρucp (T − Tref )lz dy|x+∆x − ρucp (T − Tref )lz dy|x + 0
0 0

divide throughout by ∆x take limit ∆x → 0

Z δt Z δt
∂T 1
−k |y=0 lz = − ( ρulz dy|x+∆x − ρulz dy|x )cp (T∞ − Tref )
∂y ∆x 0 0

Z δt Z δt
1
+ [ ρucp (T − Tref )lz dy|x+∆x − ρucp (T − Tref )lz dy|x ]
∆x 0 0

Assume T∞ constant with x

Z δt
∂T ∂
−k |y=0 lz = − ( ρulz dy|)cp (T∞ − Tref )
∂y ∂x 0
Z δt
∂
+ [ ρucp (T − Tref )lz dy]
∂x 0
Drop out Tref after bring cp T∞ into integral:
Z δt
∂T ∂
−k |y=0 lz = − ( ρulz dycp T∞ )
∂y ∂x 0
Z δt
∂
+ [ ρucp T lz dy]
∂x 0
Finally get rid of lz
Z δt Z δt
∂T ∂ ∂
−k |y=0 = − ( ρudycp T∞ ) + [ ρucp T dy]
∂y ∂x 0 ∂x 0
Z δt Z δt
∂T ∂ ∂
−k |y=0 =− ( ρudycp T∞ ) + [ ρucp T dy]
∂y ∂x 0 ∂x 0

23
 Tidy up equation:
Z δt
∂T ∂
−k |y=0 = [ ρucp (T − T∞ )dy]
∂y ∂x 0
y is integrated out so the integral on the RHS does not depend on y
Z δt
∂T d
−k |y=0 = [ ρucp (T − T∞ )dy]
∂y dx 0
You can even bring out ρcp :
Z δt
∂T d
−k |y=0 = ρcp [ u(T − T∞ )dy]
∂y dx 0

6.1 The Hardcore Integral Solution

But in other textbooks, von karman’s integral approach literally integrates
the BL equations over y (Bejan, 2013).
we start at:
∂ ∂ ∂2
u T +v T = α( 2 T )
∂x ∂y ∂y
and for we put back u and v into the differential using product rule.

∂ ∂ ∂
(uT ) = u T + T u
∂x ∂x ∂x
∂ ∂ ∂
u T = (uT ) − T u
∂x ∂x ∂x
∂ ∂ ∂
(vT ) = v T + T v
∂y ∂y ∂y
∂ ∂ ∂
v T = (vT ) − T v
∂y ∂y ∂y
subs back in:

∂ ∂ ∂2
u T +v T = α( 2 T )
∂x ∂y ∂y

∂ ∂ ∂ ∂ ∂2
(uT ) − T u+ (vT ) − T v = α( 2 T )
∂x ∂x ∂y ∂y ∂y

24
 ∂ ∂ ∂ ∂ ∂2
(uT ) + (vT ) − T ( v + u) = α( 2 T )
∂x ∂y ∂y ∂x ∂y
Now we consider continuity equation

∂ ∂
v+ u=0
∂y ∂x
And finally get...

∂ ∂ ∂2
(uT ) + (vT ) = α( 2 T )
∂x ∂y ∂y
Now that we have these equations we integrate across y in the BL.
Y Y Y
∂2
Z Z Z
∂ ∂
(uT )dy + (vT )dy = α( T )dy
0 ∂x 0 ∂y 0 ∂y 2
Y Y
∂2
Z Z
∂
(uT )dy + (vT )|Y − (vT )|0 = α ( T )dy
0 ∂x 0 ∂y 2
Z Y
∂ ∂T ∂T
(uT )dy + (vT )|Y − (vT )|0 = α[ |Y − |0 ]
0 ∂x ∂y ∂y
Notice there is a partial derivative with x, we will need to use Leibniz’s
Rule

https://en.wikipedia.org/wiki/Leibniz_integral_rule

Consider the integral

Z y2=b(x)
f (x, y)dy
y1=a(x)

!
Z y2=b(x) Z y2=b(x)
d d d ∂
f (x, y)dy = f (x, y = b(x)) b(x)−f (x, y = a(x)) a(x)+ f (x, y)dy
dx y1=a(x) dx dx y1=a(x) ∂x

We can apply it, we let f(x,y) =uT

!
Z y2=b(x) Z y2=b(x)
d d d ∂
uT dy = uT (x, y = b(x)) b(x)−uT (x, y = a(x)) a(x)+ uT dy
dx y1=a(x) dx dx y1=a(x) ∂x

25
 substitute in y1=0, y2=Y

Z y2=Y  Z y2=Y
d d d ∂
uT dy = uT (x, y = Y ) Y (x)−uT (x, y = 0) 0+ uT dy
dx 0 dx dx 0 ∂x
Z Y  Z Y
d d ∂
uT dy = uT (x, y = Y ) Y (x) + uT dy
dx 0 dx 0 ∂x
Z Y Z Y 
∂ d d
uT dy = uT dy − uT (x, y = Y ) Y (x)
0 ∂x dx 0 dx
substitute back into main equation:
Z Y
∂ ∂T ∂T
(uT )dy + (vT )|Y − (vT )|0 = α[ |Y − |0 ]
0 ∂x ∂y ∂y

Z Y 
d d ∂T ∂T
uT dy −uT (x, y = Y ) Y (x)+(vT )|Y −(vT )|0 = α[ |Y − |0 ]
dx 0 dx ∂y ∂y

Then consider getting v (y velocity) by integrating continuity equation

(Bejan, 2013).

∂u ∂v
+ =0
∂x ∂y
Z Y Z Y
∂u ∂v
dy + dy = 0
0 ∂x 0 ∂y
Z Y
∂u
dy + v|Y − v|0 = 0
0 ∂x
Z Y
∂u
v|Y = v|0 − dy
0 ∂x
We considered Leibniz’s rule before using f=uT,
Z Y Z Y 
∂ d d
uT dy = uT dy − uT (x, y = Y ) Y (x)
0 ∂x dx 0 dx
now consider f(x,y)=u(x,y)
Z Y Z Y 
∂ d d
udy = udy − u(x, y = Y ) Y (x)
0 ∂x dx 0 dx

26
 Z Y 
d d
v|Y = v|0 − ( udy − u(x, y = Y ) Y (x))
dx 0 dx
Z Y 
d d
v|Y = v|0 − udy + u(x, y = Y ) Y (x)
dx 0 dx
Let’s substitute this back into our energy equation:

Z Y 
d d ∂T ∂T
uT dy −uT (x, y = Y ) Y (x)+(vT )|Y −(vT )|0 = α[ |Y − |0 ]
dx 0 dx ∂y ∂y

Z Y 
d d ∂T ∂T
uT dy −uT (x, y = Y ) Y (x)+v|Y T |Y −(vT )|0 = α[ |Y − |0 ]
dx 0 dx ∂y ∂y

Z Y  Z Y 
d d d d
uT dy −uT (x, y = Y ) Y (x)+(v|0 − udy +u(x, y = Y ) Y (x))T |Y −(vT )|0
dx 0 dx dx 0 dx

∂T ∂T
= α[ |Y − |0 ]
∂y ∂y

Z Y  Z Y 
d d d d
uT dy −uT (x, y = Y ) Y (x)+v|0 T |Y − udy T |Y +u(x, y = Y )T |Y Y (x)−(vT )|0
dx 0 dx dx 0 dx

∂T ∂T
= α[ |Y − |0 ]
∂y ∂y

Z Y  Z Y 
d d ∂T ∂T
uT dy + v|0 T |Y − udy T |Y − (vT )|0 = α[ |Y − |0 ]
dx 0 dx 0 ∂y ∂y

Now we need to introduce some BCs to tidy things up:

• no slip

• Y = δt

• conduction at surface >> conduction at BL

27
Firstly no slip v=0 at y=0:
Z Y  Z Y 
d d ∂T ∂T
uT dy − udy T |Y = α[ |Y − |0 ]
dx 0 dx 0 ∂y ∂y
now, Y = δt (x)
Z δt  Z δt 
d d ∂T ∂T
uT dy − udy T |δt = α[ |δt − |0 ]
dx 0 dx 0 ∂y ∂y
note: T∞ = T |δt
Z δt  Z δt 
d d ∂T ∂T
uT dy − udy T∞ = α[ |δt − |0 ]
dx 0 dx 0 ∂y ∂y
in general, T∞ = T∞ (x)

Z δt  Z δt 
d d ∂T ∂T
uT dy − T∞ (x) udy = α[ |δt − |0 ]
dx 0 dx 0 ∂y ∂y
Apply product rule:

 Z δt  Z δt   Z δt
d d dT∞ (x)
T∞ (x) udy = T∞ (x) udy + udy
dx 0 dx 0 dx 0

multiply all by -1, im going to just use T∞ instead of T∞ (x)

 Z δt  Z δt    Z δt
d d dT∞
T∞ udy = T∞ udy + udy
dx 0 dx 0 dx 0

 Z δt  Z δt    Z δt
d d dT∞
− T∞ udy = −T∞ udy − udy
dx 0 dx 0 dx 0

 Z δt    Z δt  Z δt 
d dT∞ d
− T∞ udy + udy = −T∞ udy
dx 0 dx 0 dx 0

Z δt   Z δt   Z δt
d d dT∞
−T∞ udy =− T∞ udy + udy
dx 0 dx 0 dx 0

Substitute back:

28
 Z δt  Z δt 
d d ∂T ∂T
uT dy − T∞ (x) udy = α[ |δt − |0 ]
dx 0 dx 0 ∂y ∂y

Z δt   Z δt   Z δt
d d dT∞ ∂T ∂T
uT dy − T∞ udy + udy = α[ |δ − |0 ]
dx 0 dx 0 dx 0 ∂y t ∂y
bring T∞ into integral

Z δt  Z δt   Z δt
d d dT∞ ∂T ∂T
uT dy − uT∞ dy + udy = α[ | δt − |0 ]
dx 0 dx 0 dx 0 ∂y ∂y

Z δt   Z δt
d dT∞ ∂T ∂T
u(T − T∞ )dy + udy = α[ | δt − |0 ]
dx 0 dx 0 ∂y ∂y
on RHS we assume heat conduction at BL negligible compared to heat
conduction at surface

∂T ∂T ∂T
[ |δt − |0 ] ≈ − |0
∂y ∂y ∂y
So we are left with our integral BL equation:
Z δt   Z δt
d dT∞ ∂T
u(T − T∞ )dy + udy = −α |0
dx 0 dx 0 ∂y
note: α = ρckp
Now assume a power series (cubic) expression for temperature and ve-
locity...
Recall for velocity:
u 3y 1 y
= − ( )3 (2)
u∞ 2 δp 2 δp
with the BCs:

u = 0 at y = 0
u = u∞ at y = δp
∂u
= 0 at y = δp
∂y

29
 ∂2u
= 0 at y = 0
∂y 2
and assuming

u = a + by + cy 2 + dy 3
Near the wall, wall shear stress is constant.
Now we apply the same thing for temperature:

T − Ts = a + by + cy 2 + dy 3
with the BCs:

T − Ts = 0 at y = 0
T − Ts = T∞ − Ts at y = δt

∂(T − Ts )
= 0 at y = δt
∂y
∂ 2 (T − Ts )
= 0 at y = 0
∂y 2
Last BC is saying, near the wall, heat flux is constant..
Now if we do the same thing for temperature profile, you will get:

T − Ts 3y 1 y
= − ( )3
T∞ − Ts 2 δt 2 δt
3y 1 y
T = Ts + (T∞ − Ts )( − ( )3 )
2 δt 2 δt
we will then need to consider ratio of thermal BL and momentum BL.
Let’s try substituting these back into our integral BL equation.
So in case we need: We already know that from hydrodynamic BL inte-
gral and similarity solution

δp 4.64
=√
x Rex
δp 5
=√
x Rex
Let’s substitute back.

30
 Z δt   Z δt
d dT∞ ∂T
u(T − T∞ )dy + udy = −α |0
dx 0 dx 0 ∂y
Assume T∞ is constant w.r.t x
 Z δt 
d ∂T
u(T − T∞ )dy = −α |0
dx 0 ∂y

Subs the Temp and Velocity profiles:

Z δt 
d 3y 1 y ∂T
u∞ ( − ( )3 )(T − T∞ )dy = −α |0
dx 0 2 δp 2 δp ∂y
Z δt 
d 3y 1 y ∂T
u∞ ( − ( )3 )(T − T∞ )dy = −α |0
dx 0 2 δp 2 δp ∂y
subs:
3y 1 y
T = Ts + (T∞ − Ts )( − ( )3 )
2 δt 2 δt

Z δt 
d 3y 1 y 3y 1 y ∂T
u∞ ( − ( )3 )((Ts + (T∞ − Ts )( − ( )3 )) − T∞ )dy = −α |0
dx 0 2 δp 2 δp 2 δt 2 δt ∂y

Z δt 
d 3y 1 y 3y 1 y ∂T
u∞ ( − ( )3 )((Ts − T∞ + (T∞ − Ts )( − ( )3 )))dy = −α |0
dx 0 2 δp 2 δp 2 δt 2 δt ∂y

Z δt 
d 3y 1 y 3y 1 y ∂T
u∞ ( − ( )3 )(Ts − T∞ − (Ts − T∞ )( − ( )3 ))dy = −α |0
dx 0 2 δp 2 δp 2 δt 2 δt ∂y

Z δt 
d 3y 1 y 3y 1 y ∂T
u∞ (Ts − T∞ )( − ( )3 )(1 − ( − ( )3 ))dy = −α |0
dx 0 2 δp 2 δp 2 δt 2 δt ∂y

Z δt 
d 3y 1 y 3y 1 y ∂T
u∞ (Ts − T∞ ) ( − ( )3 )(1 − ( − ( )3 ))dy = −α |0
dx 0 2 δp 2 δp 2 δt 2 δt ∂y

divide both sides by (Ts − T∞ ),

31
 Z δt 
d 3y 1 y 3y 1 y 1 ∂T − Ts
u∞ ( − ( )3 )(1 − ( − ( )3 ))dy = −α( ) |0
dx 0 2 δp 2 δp 2 δt 2 δt (Ts − T∞ ) ∂y

Z δt 
d 3y 1 y 3y 1 y 1 ∂T − Ts
u∞ ( − ( )3 )(1 − ( − ( )3 ))dy = α( ) |0
dx 0 2 δp 2 δp 2 δt 2 δt (T∞ − Ts ) ∂y

Z δt 
d 3y 1 y 3y 1 y ∂θ
u∞ ( − ( )3 )(1 − ( − ( )3 ))dy =α |0
dx 0 2 δp 2 δp 2 δt 2 δt ∂y

Where θ = TT∞−T s
−Ts
Let’s evaluate the integral:
Z δt
3y 1 y 3y 1 y
( − ( )3 )(1 − ( − ( )3 ))dy
0 2 δp 2 δp 2 δt 2 δt
let’s substitute the integral for dy with d δyt
y
dy = δt d
δt
note: δt and δp are constant w.r.t y
subs the change of variable:
Z δt
3y 1 y 3y 1 y
( − ( )3 )(1 − ( − ( )3 ))dy
0 2 δp 2 δp 2 δt 2 δt
Z δt
3y 1 y 3y 1 y y
= ( − ( )3 )(1 − ( − ( )3 ))d δt
0 2 δp 2 δp 2 δt 2 δt δt
Z 1
3y 1 y 3y 1 y y
= δt ( − ( )3 )(1 − ( − ( )3 ))d
0 2 δp 2 δp 2 δt 2 δt δt
y
let’s replace β = δt , y = βδt
Z 1
3y 1 y 3 1
= δt ( − ( )3 )(1 − ( β − β)3 )dβ
0 2 δp 2 δp 2 2
Z 1
3y 1 y 3 1
= δt ( − ( )3 )(1 − β + β 3 )dβ
0 2 δp 2 δp 2 2

32
 Z 1
3 βδt 1 βδt 3 3 1
= δt ( − ( ) )(1 − β + β 3 )dβ
0 2 δp 2 δp 2 2
Z 1
3 βδt 3 1 1 βδt 3 3 1
= δt (1 − β + β 3 ) − ( ) (1 − β + β 3 )dβ
0 2 δp 2 2 2 δp 2 2

Z 1 Z 1
3 βδt 3 1 1 βδt 3 3 1
= δt (1 − β + β 3 )dβ − δt ( ) (1 − β + β 3 )dβ
0 2 δp 2 2 0 2 δp 2 2

[careless mistake! multiplied in β not β 3 ]

Z 1 Z 1
δt 3 3 1 δt 1 3 3 1
= δt β(1 − β + β 3 )dβ − δt ( )3 β (1 − β + β 3 )dβ
δp 0 2 2 2 δp 0 2 2 2

Z 1 Z 1
δt 3 3 1 δt 1 3 3 4 1 6
= δt (β − β 2 + β 4 )dβ − δt ( )3 (β − β + β )dβ
δp 0 2 2 2 δp 0 2 2 2

Z 1 Z 1
3 δt 3 1 1 δt 3 1
= δt (β − β 2 + β 4 )dβ − δt ( )3 (β 3 − β 4 + β 6 )dβ
2 δp 0 2 2 2 δp 0 2 2

Z 1 Z 1
3 δt 3 1 1 δt 3 1
= δt (β − β 2 + β 4 )dβ − δt ( )3 (β 3 − β 4 + β 6 )dβ
2 δp 0 2 2 2 δp 0 2 2

3 δt 1 1 δt 3
= δt − δt ( )3
2 δp 10 2 δp 140
3 δt 3 δt
= δt − δt ( )3
20 δp 280 δp
Substituting the integral back:
 
d 3 δt 3 δt 3 ∂θ
u∞ δt − δt ( ) = α |0
dx 20 δp 280 δp ∂y

3 δt2 3 δt4
 
d ∂θ
u∞ − ( 3 ) = α |0
dx 20 δp 280 δp ∂y
δt
in approximate method, we consider the ratio δp

33
 if δδpt >> 1 we can ignore first term, otherwise, if δt
δp << 1 we can ignore
the second term

δt
>> 1
δp

3 δt4
 
d ∂θ
u∞ − ( 3 ) = α |0
dx 280 δp ∂y
δt
<< 1
δp

3 δt2
 
d ∂θ
u∞ = α |0
dx 20 δp ∂y
Now we can substitute θ

T − Ts 3y 1 y
θ= = − ( )3
T∞ − Ts 2 δt 2 δt

∂ 31 1 3y 2
θ= − ( 3 )
∂y 2 δt 2 δt
now set y=0

∂ 3
θ|y=0 =
∂y 2δt
subs back in:

δt
>> 1
δp

3 δt4
 
d 3
u∞ − ( ) =α
dx 280 δp3 2δt
δt
<< 1
δp

3 δt2
 
d 3
u∞ =α
dx 20 δp 2δt
note: from momentum integral equation
r
4.64 νx
δp = x √ = 4.64
Rex u∞

34
Try substituting in latter case

δt
<< 1
δp
3 δt2
 
d 3
u∞ =α
dx 20 δp 2δt
 
d 3 δt2 3
u∞ q =α
dx 20 4.64 νx 2δt
u∞

u1.5 δ2
 
3 d 3
√∞ √t =α
ν 20 ∗ 4.64 dx x 2δt
using quotient rule [careless mistake here, forgot to multiply by 2]:
√
d

δ2
 2 xδt dδ t 2 √1
dx − δt (0.5) x
√t =
dx x x
substitute back:
√
u1.5 3 2 xδt dδ t 2 √1
dx − δt (0.5) x 3
∞
√ =α
ν 20 ∗ 4.64 x 2δt
u1.5 3 2 dδt 1 3
√∞ ( √ δt − δt2 (0.5) √ ) = α
ν 20 ∗ 4.64 x dx x x 2δt
u1.5 3 2 dδt 1 3
√∞ ( √ δt2 − δt3 (0.5) √ ) = α
ν 20 ∗ 4.64 x dx x x 2
dδt3 dδt
= 3δt2
dx dx
1 dδt3 dδt
= δt2
3 dx dx
substitute back:

u1.5 3 1 2 dδt3 1 3
√∞ (√ − δt3 (0.5) √ ) = α
ν 20 ∗ 4.64 x 3 dx x x 2
1st order linear ODE in δt3

u1.5 1 2 dδt3 1 20 ∗ 4.64 3

√∞ ( √ − δt3 (0.5) √ ) = α
ν x 3 dx x x 3 2

35
 √
1 2 dδt3 3 1 10 ∗ 4.64 ν
√ − δt (0.5) √ = α
x 3 dx x x 1 u1.5
∞
3 √
dδt 1 15 ∗ 4.64 ν √
− δt3 (0.75) = α 1.5
x
dx x 1 u∞
Let
√
15 ∗ 4.64 ν
A=α
1 u1.5
∞

dδt3 0.75 √
− δt3 =A x
dx x
[Alternate solution: substitute later, may get better solution (less care-
less mistake)]

3 δt2
 
d 3
u∞ =α
dx 20 δp 2δt

3 δt2
 
d 3
u∞ δp 2 = α
dx 20 δp 2δt

3 δt2
 
d 3
u∞ δp 2 = α
dx 20 δp 2δt
δt
let φ = δp
 
d 3 3
u∞ δ p φ2 =α
dx 20 2δt
d 10
δ p φ2 = α


u∞
dx δt
Product rule:

d d 10
φ2 δp + δp φ2 = α
dx dx u∞ δt
d d 10
δ t φ2 δp + δt δp φ2 = α
dx dx u∞
sub δt = φδp

d d 10
δ p φ3 δp + φδp2 φ2 = α
dx dx u∞

36
 d d 10
δp φ3 δp + 2φ2 δp2 φ = α
dx dx u∞
d d 10
φ3 δ p δp + 2δp2 φ2 φ2 = α
dx dx u∞
d 2 d 3 10
φ3 δp δp + δp2 φ =α
dx 3 dx u∞
√
4.64 4.64 νx
δp = x √ = √
Rex u∞
√
4.64 4.64 νx
δp = x √ = √
Rex u∞
4.642 4.642 νx
δp2 = x √ =
Rex u∞
√
d 1 4.64 ν
δp = √
dx 2 u∞ x
√ √
d 1 4.64 νx 4.64 ν
δp δp = √ √
dx 2 u∞ u∞ x
d 1 4.642 ν
δp δp =
dx 2 u∞
subs back:

d 2 d 3 10
φ3 δp δp + δp2 φ =α
dx 3 dx u∞
1 4.642 ν 4.642 νx 2 d 3 10
φ3 + φ =α
2 u∞ u∞ 3 dx u∞
1 4.642 ν 4.642 νx 2 d 3 10
φ3 + φ =α
2 1 1 3 dx 1
1 2 d 3 10 α
φ3 + x φ =
2 3 dx 4.642 ν
3 d 30 1
φ3 + 2x φ3 =
2 dx 4.642 P r
4 d 20 1
φ3 + x φ3 =
3 dx 4.642 P r
First order ODE:

37
https://mathworld.wolfram.com/First-OrderOrdinaryDifferentialEquation.html

y 0 + p(x)y = q(x)
Rx
p(x0 )dx0 q(x)dx
R
e +c
y= Rx
0 0
e p(x )dx
Replace x by dummy variables x’

dδt3 3 0.75
√
− δ t = A x0
dx0 x0
0.75
p(x0 ) = −
x0
√
q(x0 ) = A x0
[second careless mistake: wrong choice of integrating factor]
Z x Z x
−0.75 0
p(x0 )dx0 = dx = −0.75 ln(x)
x0

exp(0.75 ln(x)) = − exp(0.75)x

√
Z Rx Z
p(x0 )dx0
e q(x)dx = − exp(0.75)xA xdx
Z
= −A exp(0.75) x1.5 dx

x2.5
= −A exp(0.75)
2.5
2.5
−A exp(0.75) x2.5 + c
δt3 =
− exp(0.75)x
2.5
−A exp(0.75) x2.5 c
δt3 = +
− exp(0.75)x exp(0.75)x

x1.5 c
δt3 = A +
2.5 − exp(0.75)x
[wrong... ODE solving gone wrong...]

38
https://www.wolframalpha.com/input/?i=solve+y%27+-+0.75+y%2Ft+%3DA+sqrt%28t%29

According to wolfram:
4
y = Ax1.5 + Cx0.75
3
apply BC, x=0 δt = 0, c=0
4
δt3 = Ax1.5
3
√
δt = O( x)

√
15 ∗ 4.64 ν 1.5 4
δt3 =α 1.5
x
1 u∞ 3
δp 4.64
=√
x Rex

α 15 ∗ 4.64 ν 1.5 4 1.5

δt3 = x
ν 1 u1.5
∞ 3
√
x νx
√ =√
Rex u ∞

α 15 ∗ 4.64 x3 4
δt3 =
ν 1 Re1.5
x 3

δp x
=√
4.64 Rex
1 15 ∗ 4.64 δp 3 4
δt3 = ( )
Pr 1 4.64 3

δt3 1 15 ∗ 4.64 4
3
=
δp P r 4.643 3

δt3 1
3
= 0.92895
δp Pr
δt 1
= 0.97573 1/3
δp Pr

39
 [redundant to use similarity soln and wrong number] If you use the sim-
ilarity solution, you will get this:

δp 5
=√
x Rex
1 30 ∗ 5 δp 3 1
δt3 = ( )
Pr 1 5 2.5
δt 1
= 0.78297 1/3
δp Pr
[redundant to use similarity soln and wrong number]
What’s our heat flux?

∂ 3
θ|y=0 =
∂y 2δt
∂ 3
θ|y=0 =
∂y 2δp ∗ 0.97573 P r11/3
∂ 3
θ|y=0 = ∗ P r1/3
∂y 2δp ∗ 0.97573
4.64x
δp = √
Rex
∂ 3
θ|y=0 = 4.64x ∗ P r1/3
∂y 2 Re ∗ 0.97573
√
x

∂ 3
θ|y=0 = ∗ P r1/3 Re1/2
∂y 2 ∗ 4.64x ∗ 0.97573
hx
N ux =
k

00 ∂T
qwall→f luid = −k = −h(T∞ − Ts )
∂y
∂T
−k = −h(T∞ − Ts )
∂y
∂T h
= (T∞ − Ts )
∂y k

40
 T − Ts
θ=
T∞ − Ts
∂θ h
=
∂y k
h 3
|y=0 = ∗ P r1/3 Re1/2
k 2 ∗ 4.64x ∗ 0.97573
hx 3
|y=0 = ∗ P r1/3 Re1/2
k 2 ∗ 4.64 ∗ 0.97573
3
N ux |y=0 = ∗ P r1/3 Re1/2
2 ∗ 4.64 ∗ 0.97573
N ux |y=0 = 0.3313P r1/3 Rex1/2
1/2
N uL |y=0 = 0.663P r1/3 ReL

Part VI
Natural Convection
7 Simplifying N-S equations and Energy Equa-
tions
Mass, momentum and energy balances:
∂ ∂ ∂
u+ v+ w=0
∂x ∂y ∂z

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
u + u u + v u + w u − ν( 2 u + 2 u + 2 u) = − + gx
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂x

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
v + u v + v v + w v − ν( 2 v + 2 v + 2 v) = − + gy
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂y

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
w + u w + v w + w w − ν( 2 w + 2 w + 2 w) = − + gz
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂z

41
 ∂ ∂ ∂ ∂ ∂2 ∂2 ∂2
T +u T +v T + w T = α( 2 T + 2 T + 2 T )
∂t ∂x ∂y ∂z ∂x ∂y ∂z
Note: compressibility applies here! How can we deal with this?

∂ ∂ ∂
ρu + ρv + ρw = 0
∂x ∂y ∂z
(note: steady state only)
 
∂ ∂ ∂ ∂ ∂ ∂
u ρ+v ρ+w ρ+ρ u+ v+ w =0
∂x ∂y ∂z ∂x ∂y ∂z
 
∂ ∂ ∂ ∂ ∂ ∂
u ρ+v ρ+w ρ+ρ u+ v+ w =0
∂x ∂y ∂z ∂x ∂y ∂z
We do term by term analysis:
∂ ∂u
u ρ+ρ
∂x ∂x

0.02ρ umax
= O(umax ) + O(ρ )
δt δt
∂ ∂u ∂u
u ρ+ρ ≈ρ
∂x ∂x ∂x
The change in density is considered small in comparison to original fluid
density... (Boussinesq approximation)
Therefore,
∂ ∂ ∂
u+ v+ w=0
∂x ∂y ∂z
With this equation, we can still use essentially the same form of the
(mostly) incompressible momentum equations:

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
u + u u + v u + w u − ν( 2 u + 2 u + 2 u) = − + gx
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂x

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
v + u v + v v + w v − ν( 2 v + 2 v + 2 v) = − + gy
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂y

42
∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
w + u w + v w + w w − ν( 2 w + 2 w + 2 w) = − + gz
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂z
For simplicity, we assume a 2D type flow, where convection goes upwards
in the positive y axis. This is to say, w=0 everywhere, gz = 0

1 ∂P
0=−
ρ0 ∂z
we can also apply the same for the x and y momentum eqns

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
u + u u + v u + w u − ν( 2 u + 2 u + 2 u) = − + gx
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂x

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2 1 ∂P
v + u v + v v + w v − ν( 2 v + 2 v + 2 v) = − + gy
∂t ∂x ∂y ∂z ∂x ∂y ∂z ρ0 ∂y
No variation of u and v in z direction Thus resulting in:

∂ ∂ ∂ ∂2 ∂2 1 ∂P
u + u u + v u − ν( 2 u + 2 u) = − + gx
∂t ∂x ∂y ∂x ∂y ρ0 ∂x

∂ ∂ ∂ ∂2 ∂2 1 ∂P
v + u v + v v − ν( 2 v + 2 v) = − + gy
∂t ∂x ∂y ∂x ∂y ρ0 ∂y
With gravity being only in y direction,

∂ ∂ ∂ ∂2 ∂2 1 ∂P
u + u u + v u − ν( 2 u + 2 u) = −
∂t ∂x ∂y ∂x ∂y ρ0 ∂x

∂ ∂ ∂ ∂2 ∂2 1 ∂P
v + u v + v v − ν( 2 v + 2 v) = − + gy
∂t ∂x ∂y ∂x ∂y ρ0 ∂y
With steady state assumption:

∂ ∂ ∂2 ∂2 1 ∂P
u u + v u − ν( 2 u + 2 u) = −
∂x ∂y ∂x ∂y ρ0 ∂x

∂ ∂ ∂2 ∂2 1 ∂P
u v + v v − ν( 2 v + 2 v) = − + gy
∂x ∂y ∂x ∂y ρ0 ∂y
Thus we have our full set of hydrodynamic equations to begin:

43
 ∂ ∂
u+ v=0
∂x ∂y

∂ ∂ ∂2 ∂2 1 ∂P
u u + v u − ν( 2 u + 2 u) = −
∂x ∂y ∂x ∂y ρ0 ∂x

∂ ∂ ∂2 ∂2 1 ∂P
u v + v v − ν( 2 v + 2 v) = − + gy
∂x ∂y ∂x ∂y ρ0 ∂y

∂ ∂ ∂ ∂ ∂2 ∂2 ∂2
T +u T +v T + w T = α( 2 T + 2 T + 2 T )
∂t ∂x ∂y ∂z ∂x ∂y ∂z
We make the assumption that the change in density is so small it does not
significantly impact thermal diffusivity. Again if we expand things out via
product rule, the terms with change in density are assumed small compared
to other terms.

∂ ∂ ∂2 ∂2
u T +v T = α( 2 T + 2 T )
∂x ∂y ∂x ∂y

8 Boussinesq approximation and dimensionless anal-

ysis
Now where does the buoyancy force come from?
It comes from surrounding fluid of heavier density ”pushing” the lighter
fluid upwards.
We expect buoyancy force to come from the pressure term. How do we
then put this into our equations?
Consider this dimensionless analysis that as long as we are within the
boundary layer, (Bejan, 2013):

O(x) = δT
O(y) = H (wall height)
H >> δT
(why thermal boundary layer and not momentum boundary layer? be-
cause the underlying driving force is temperature, that’s the more important
BL in natural convection)
Also static pressure should change only with y, not x.

44
 What’s the expression then?

P∞ (y) = (L − y)ρ∞ (−gy )

L is the length from the ”surface” so to speak.

dP∞
= −ρ∞ (−gy ) = −ρ∞ g
dy
substitute this back and we get:

∂ ∂ ∂2 ∂2 1 ∂P
u v + v v − ν( 2 v + 2 v) = − + gy
∂x ∂y ∂x ∂y ρ0 ∂y

∂ ∂ ∂2 ∂2 1
u v + v v − ν( 2 v + 2 v) = − (−ρ∞ (−gy )) + gy
∂x ∂y ∂x ∂y ρ0
of course gy is negative, we can simply state

gy = −g

∂ ∂ ∂2 ∂2 1
u v + v v − ν( 2 v + 2 v) = − (−ρ∞ g) − g
∂x ∂y ∂x ∂y ρ0
Note that the ρ0 here refers to fluid density at the BL, not in the
freestream.
The change in ρ0 was small compared to other terms. But in this case,
on the RHS, the only driving force for the fluid is this density change, so it
cannot be neglected as before.

∂ ∂ ∂2 ∂2
ρu v + ρv v − ρν( 2 v + 2 v) = (ρ∞ g) − ρg
∂x ∂y ∂x ∂y

∂ ∂ ∂2 ∂2
ρu v + ρv v − ρν( 2 v + 2 v) = (ρ∞ − ρ)g
∂x ∂y ∂x ∂y
Here’s were the boussinesq approximation comes in, only applies for
small density changes due to temperature, not pressure. Constant static
pressure is assumed (Bejan, 2013)

ρ = ρ∞ (1 − β(T − T∞ ))
This is a linear relationship for a constant β value (thermal expansion
coefficient).

45
  
1 ∂ρ
β=−
ρ ∂T P
When we substitute all this in, we are ready to start deriving the equa-
tions.

∂ ∂ ∂2 ∂2
ρu v + ρv v − ρν( 2 v + 2 v) = ρ∞ β(T − T∞ )g
∂x ∂y ∂x ∂y
Once again, for the sake of simplification, density changes are not im-
portant except that the buoyancy forces are accounted for, therefore,

ρ ≈ ρ∞

∂ ∂ ∂2 ∂2
u v + v v − ν( 2 v + 2 v) = β(T − T∞ )g
∂x ∂y ∂x ∂y
This substitution is known as the boussinesq approximation.
First we start with continuity

∂ ∂
u+ v=0
∂x ∂y

x = x∗ δT

y = y∗H
What’s the characteristic velocity scale? We don’t have a u∞ to really
look to here...
Okay, so maybe, it’s easier to start with energy equations because this is
our driving force! In forced convection, the driving force is sort of implicit
in the momentum and mass conservation equations (freestream velocity is
driving force).

∂ ∂ ∂2 ∂2
u T +v T = α( 2 T + 2 T )
∂x ∂y ∂x ∂y
So now, we can nondimensionalise T, x and y. u and v will remain.
Now combine this with the continuity equation we have two equations
with two unknowns (order of magnitude so to speak).

1 ∂ 1 ∂ 1 ∂2 1 ∂2
u ∗
T +v ∗
T = α( 2 ∗
T+ 2 T)
δT ∂x H ∂y δT ∂(x )2 H ∂(y ∗ )2

46
 Now we can also nondimensionalise temperature.

T − T∞
θ=
Ts − T∞

(Ts − T∞ )dθ = d(T − T∞ ) = dT

1 ∂ 1 ∂ 1 ∂2 1 ∂2
u T + v T = α( T + T)
δT ∂x∗ H ∂y ∗ δT2 ∂(x∗ )2 H 2 ∂(y ∗ )2

1 ∂ 1 ∂ 1 ∂2 1 ∂2
u (T −T∞ )+v (T −T∞ ) = α( (T −T∞ )+ (T −T∞ ))
δT ∂x∗ H ∂y ∗ δT2 ∂(x∗ )2 H 2 ∂(y ∗ )2

(Ts − T∞ ) ∂ (Ts − T∞ ) ∂ (Ts − T∞ ) ∂ 2 (Ts − T∞ ) ∂ 2

u θ+v θ = α( θ+ θ)
δT ∂x∗ H ∂y ∗ δT2 ∂(x∗ )2 H2 ∂(y ∗ )2

Now since H >> δT , we can assume that

(Ts − T∞ ) ∂ 2 (Ts − T∞ ) ∂ 2 (Ts − T∞ ) ∂ 2

( θ + θ) ≈ ( θ)
δT2 ∂(x∗ )2 H2 ∂(y ∗ )2 δT2 ∂(x∗ )2

(Ts − T∞ ) ∂ (Ts − T∞ ) ∂ (Ts − T∞ ) ∂ 2

u θ + v θ = α( θ)
δT ∂x∗ H ∂y ∗ δT2 ∂(x∗ )2

And since the nondimensionalised terms are all O(1)

(Ts − T∞ ) (Ts − T∞ ) (Ts − T∞ )

O(u ) + O(v ) = O(α )
δT H δT2
So now we can work with the mass balance equations:

∂ ∂
u+ v=0
∂x ∂y

x = x∗ δT

y = y∗H

47
 1 ∂ 1 ∂
∗
u+ v=0
δT ∂x H ∂y ∗
O(u) ∂ ∗ O(v) ∂ ∗
u + v =0
δT ∂x∗ H ∂y ∗

u = O(u)u∗

u v
O( ) = O( )
δT H
Where u and v are typical velocity values in the BL. This is thermal BL
btw
From this, we find that we cannot just throw out any of the terms here...
However, we do know that the x and y velocity terms are on the same order
of magnitude. We can then perform elimination to find:
u v α
O( ) + O( ) = O( 2 )
δT H δT
u α
2O( ) = O( 2 )
δT δT
α
2O(u) = O( )
δT
if i eliminate u,

αH
2O(v) = O( )
δT2
Of course we can easily get rid of the factor of 2, since it is still in the
same order of magnitude

2 = O(1)

α
O(u) = O( )
δT
αH
O(v) = O( )
δT2
Now let’s move on to the momentum equations

48
 ∂ ∂ ∂2 ∂2
u v + v v − ν( 2 v + 2 v) = β(T − T∞ )g
∂x ∂y ∂x ∂y
Now here, we only nondimensionalise the position and velocity coordi-
nates, for density and viscosity, while they are changing withing the fluid,
we choose not to nondimensionalise as the changes are not as significant
as the terms contributing to inertial forces, frictional forces and buoyancy
forces.
The easiest is to start with the temperature.

∂ ∂ ∂2 ∂2
u v + v v − ν( 2 v + 2 v) = β(Ts − T∞ )gθ
∂x ∂y ∂x ∂y
Then the advection terms,

1 ∗ ∂ ∗ 1 ∗ ∂ ∗ ∂2 ∂2
O(u)O(v) u v +O(v)O(v) v v −ν( v+ v) = β(Ts −T∞ )gθ
δT ∂x∗ H ∂y ∗ ∂x2 ∂y 2

Then the frictional (viscous) forces

1 ∗ ∂ ∗ 1 ∗ ∂ ∗ O(v) ∂ 2 ∗ O(v) ∂
2
O(u)O(v) u v +O(v)O(v) v v −ν( v + v ∗ ) = β(Ts −T∞ )gθ
δT ∂x∗ H ∂y ∗ δT ∂(x∗ )2
2 H 2 ∂(y ∗ )2

We make an approximation:

O(v) ∂ 2 ∗ O(v) ∂ 2 ∗ O(v) ∂ 2

( v + v ) ≈ ( v∗)
δT2 ∂(x∗ )2 H 2 ∂(y ∗ )2 δT2 ∂(x∗ )2
So we are left with:

1 ∗ ∂ ∗ 1 ∗ ∂ ∗ O(v) ∂ 2
O(u)O(v) u v +O(v)O(v) v v −ν( v ∗ ) = β(Ts −T∞ )gθ
δT ∂x∗ H ∂y ∗ δT2 ∂(x∗ )2

α 1 ∂ 1 ∂ O(v) ∂ 2
O(v) u∗ ∗ v ∗ +O(v)O(v) v ∗ ∗ v ∗ −ν( 2 v ∗ ) = β(Ts −T∞ )gθ
δT δT ∂x H ∂y δT ∂(x∗ )2

αH
α αH 1 ∗ ∂ ∗ αH αH 1 ∗ ∂ ∗ 2
δT ∂2
u v + v v − ν( v ∗ ) = β(Ts − T∞ )gθ
δT δT2 δT ∂x∗ δT2 δT2 H ∂y ∗ δT2 ∂(x∗ )2

49
 α2 H ∗ ∂ ∗ α2 H ∗ ∂ ∗ αH ∂ 2
u v + v v − ν( v ∗ ) = β(Ts − T∞ )gθ
δT4 ∂x∗ δT4 ∂y ∗ δT4 ∂(x∗ )2

Viscous forces are always significant in laminar BL flow.

Or in its dimensional form,

∂ ∂ ∂2
u v + v v − ν( 2 v) = β(T − T∞ )g
∂x ∂y ∂x
Now if we want to nondimensionalise everything, we found the charac-
teristic velocity scaling to be:
u v
O( ) = O( )
δT H
α
O(u) = O( )
δT
αH
O(v) = O( )
δT2
Hence, we can use these to nondimensionalise the velocity scales

1 ∗ ∂ ∗ 1 ∗ ∂ ∗ O(v) ∂ 2
O(u)O(v) u v +O(v)O(v) v v −ν( v ∗ ) = β(Ts −T∞ )gθ
δT ∂x∗ H ∂y ∗ δT2 ∂(x∗ )2

α αH 1 ∗ ∂ ∗ αH αH 1 ∗ ∂ ∗ αH 1 ∂ 2
u v + v v −ν( v ∗ ) = β(Ts −T∞ )gθ
δT δT δT ∂x∗
2 δT δT H ∂y ∗
2 2 δT2 δT2 ∂(x∗ )2

Now collecting terms together,

α2 H ∗ ∂ ∗ α2 H ∗ ∂ ∗ αH ∂ 2
4 u ∗
v + 4 v ∗
v − ν( 4 ∗ 2
v ∗ ) = β(Ts − T∞ )gθ
δT ∂x δT ∂y δT ∂(x )

α2 H
we divide both sides by, 4
δT

∂ ∗ ∗ ∂ ν ∂2 δT4
u∗ v + v v ∗
− ( v ∗
) = β(Ts − T∞ )gθ
∂x∗ ∂y ∗ α ∂(x∗ )2 α2 H
We introduce the prandtl number,

50
 ∂ ∗ ∗ ∂ ∂2 δT4
u∗ v + v v ∗
− (P r v ∗
) = β(Ts − T∞ )gθ
∂x∗ ∂y ∗ ∂(x∗ )2 α2 H
Also we notice the ratio δHT should be important, it describes the rela-
tionship between thermal BL thickness and wall height.

∂ ∗ ∗ ∂ ∂2 δT4
u∗ v + v v ∗
− (P r v ∗
) = β(Ts − T∞ )H 3 gθ
∂x∗ ∂y ∗ ∂(x∗ )2 α2 H 4

∂ ∗ ∂ ∂2 δ 4 β(Ts − T∞ )H 3 g
u∗ ∗
v + v ∗ ∗ v ∗ − (P r ∗ 2
v ∗ ) = T4 θ
∂x ∂y ∂(x ) H α2
Divide throughout by Pr,

∂2 δT4 β(Ts − T∞ )H 3 g 1
 
1 ∗ ∂ ∗ ∗ ∂ ∗ ∗
u v + v v − v = θ
Pr ∂x∗ ∂y ∗ ∂(x∗ )2 H4 α2 Pr

∂2 δT4 β(Ts − T∞ )H 3 g α
 
1 ∗ ∂ ∗ ∗ ∂ ∗ ∗
u v + v v − v = θ
Pr ∂x∗ ∂y ∗ ∂(x∗ )2 H4 α2 ν

∂2 δT4 β(Ts − T∞ )H 3 g
 
1 ∂ ∂
u∗ ∗ v ∗ + v ∗ ∗ v ∗ − v ∗
= θ
Pr ∂x ∂y ∂(x∗ )2 H4 αν
Here we see the occurence of an important number:

β(Ts − T∞ )H 3 g
RaH =
αν
This is called the Rayleigh number. Which is THE important term for
natural convection, it is dependent on both thermal diffusivity and momen-
tum diffusivity.

∂2 δ4
 
1 ∂ ∂
u∗ ∗ v ∗ + v ∗ ∗ v ∗ − ∗ 2
v ∗ = T4 RaH θ
Pr ∂x ∂y ∂(x ) H
It mismashes all the effects in one term. Both thermal and momentum
diffusivity, so it’s not quite like the Reynold’s number.
We can see that Pr is important in whether we consider advection of
momentum an important effect. In fact P1r is important in this equation.

51
We can remove these terms, but what are they small in comparision to? We
need another term with P1r coefficient to compare it to.
Normally if Prandtl number is really big, and the Rayleigh number is
big, this comparison is not needed. But if the prandtl number is small, we
got a problem. So we’ll need to do comparisons another way.

∂2 δT4 β(Ts − T∞ )H 3 g
 
1 ∗ ∂ ∗ ∗ ∂ ∗ ∗
u v + v v − v = θ
Pr ∂x∗ ∂y ∗ ∂(x∗ )2 H4 αν
note: Pr is very important, It describes the relationship between mo-
mentum BL thickness and thermal BL thickness
δp
= f (P r)
δt
If Pr low, (liquid metals) let’s try this trick

∂2 δT4 β(Ts − T∞ )H 3 g P r
 
1 ∗ ∂ ∗ ∗ ∂ ∗ ∗
u v + v v − v = θ
Pr ∂x∗ ∂y ∗ ∂(x∗ )2 H4 αν Pr

∂2 δT4 β(Ts − T∞ )H 3 g 1
 
1 ∗ ∂ ∗ ∗ ∂ ∗ ∗
u v + v v − v = θ
Pr ∂x∗ ∂y ∗ ∂(x∗ )2 H4 α2 Pr
Only within thermal BL, advection for liquid metals or low Pr fluids, we
can neglect momentum transport away from thermal BL through viscous
action

δT4 β(Ts − T∞ )H 3 g 1
 
1 ∗ ∂ ∗ ∗ ∂ ∗
u v + v v = θ
Pr ∂x∗ ∂y ∗ H4 α2 Pr
δT4 β(Ts − T∞ )H 3 g
 
∗ ∂ ∗ ∗ ∂ ∗
u v + v v = θ
∂x∗ ∂y ∗ H4 α2
Now we find this new dimensionless number(Bejan, 2013):

β(Ts − T∞ )H 3 g
Bo = = RaH P r
α2
This is called the boussinesq number. Important for fluids with low Pr,
so that the viscous term can then be neglected.

(Ts − T∞ ) ∂ (Ts − T∞ ) ∂ (Ts − T∞ ) ∂ 2

u ∗
θ+v ∗
θ = α( θ)
δT ∂x H ∂y δT2 ∂(x∗ )2

52
9 solution procedure - integral
Now to solve the equation. First we always need to know what we are solving
for, otherwise we will go in circles.
We have some questions of interest.

• What is the Nu (heat transfer coeff)?

• Velocity profile

• BL thickness of thermal and momentum BL

Remember, we may not want to solve the equation in totality, but rather
go for special cases, eg. high or low Pr to make life easier for ourselves.
For example, in high Pr fluids, we want to just neglect the momentum
advection terms, making life easier for ourselves.

9.1 High Pr fluids

In this case, very simply, we eliminate the momentum advection terms.

∂2 δT4
 
1 ∂ ∂
u∗ ∗ v ∗ + v ∗ ∗ v ∗ − v ∗
= RaH θ
Pr ∂x ∂y ∂(x∗ )2 H4

∂2 ∗ δT4
− v = RaH θ
∂(x∗ )2 H4
From this we can see that:

δT4
RaH = O(1)
H4
(remember all dimensionless terms are of the order of magnitude of 1)
−1
δT = O(HRaH 4 )
Another way to write this is:
−1
δT = O(1)HRaH 4

β(Ts − T∞ )H 3 g − 1
δT = O(1)H( ) 4
αν
β(Ts − T∞ )g − 1
δT = O(1)H 1/4 ( ) 4
αν

53
Based on this, we can guess the velocity profile. Given that

aH
v = O( )
δT2
aH
v = O(1)
δT2

...

v ∼ H 0.5
What about the Nusselt number or heat transfer coeffcient?
We need to then look at our energy equations:

(Ts − T∞ ) ∂ (Ts − T∞ ) ∂ (Ts − T∞ ) ∂ 2

u θ + v θ = α( θ)
δT ∂x∗ H ∂y ∗ δT2 ∂(x∗ )2

The heat transfer in the x direction is governed by,

∂T (Ts − T∞ )
= O(u )
∂x δT
Recall:
α
u = O( )
δT
∂T α (Ts − T∞ )
= O( )
∂x δT δT
∂T (Ts − T∞ )
= O(α )
∂x δT2
aH
v = O( )
δT2
∂T (Ts − T∞ )
= O(1)v
∂y H
∂T aH (Ts − T∞ )
= O(1) 2
∂y δT H

54
 ∂T a(Ts − T∞ )
= O(1)
∂y δT2

∂ ∗ ∂ ∂2
u∗ θ + v θ = θ
∂x∗ ∂y ∗ ∂(x∗ )2
With this we can see how the temperature gradient scales as we go up
the wall.
Why is all this important? Because we assume a velocity or temperature
profile shape when we do integral solution analysis.
So how can we get a temperature profile across x? It’s more or less
qualitative. Just like for the original von karman solution, a polynomial
solution was assumed.
We can try looking at the energy equations and come up with some
qualitative shapes:
Consider the energy equation, hold y constant.

∂ ∗ ∂ ∂2
u∗ θ + v θ = θ
∂x∗ ∂y ∗ ∂(x∗ )2
As we go further away from the wall, intuitively u and v increase. Or
more correctly advection increases. Up to some certain maximum point,
and then falls off.
For temperature, there is conduction effect but with some advection
which effectively acts like a heat sink.
In Bejan’s textbook, the recommended temp profile is(Bejan, 2013):
x
T − T∞ = (Ts − T∞ ) exp(− )
δT
x
θ = exp(− )
δT
If we were to substitute this back,

∂2 δT4
 
1 ∗ ∂ ∗ ∗ ∂ ∗ ∗
u v + v v − v = RaH θ
Pr ∂x∗ ∂y ∗ ∂(x∗ )2 H4

∂2 δ4
− ∗ 2
v ∗ = T4 RaH θ
∂(x ) H

∂2 ∗ δT4 x
− ∗ 2
v = 4
RaH exp(− )
∂(x ) H δT

55
 ∂2 x
− ∗ 2
v ∗ = O(1) exp(− )
∂(x ) δT
x∗
v ∗ = f (exp(− ))
δT
In Bejan’s textbook, the recommended velocity profile is (Bejan, 2013):
x x
v(x, y) = vY (y) exp(− )(1 − exp(− ))
δp δT
Now this is just an estimate and assumption, based on the rough shapes
of the BL, there are no exact reasoning why there must be exponential decay
shapes.
BUT.... one can think of it this way, consider the control volume in the
BL,
For exponential equations, we will have the form
dT
= −kT
dx
There’s conduction in and out (right term), and advection in and out
(left terms)
along x axis, velocity is likely to go in negative x direction. Fluid flows
up, has to come in from somewhere. Also will draw in cold fluid and release
hot fluid at temperature T.
Along y axis, velocity goes in positive y direction, again, draw in cold
fluid, release hot fluid at temperature T.
Basically, we do have a heat loss term in the form kT.
Note that there is a heat flux term
Now for integral solutions, we need to integrate momentum and energy
equations across BL in x direction:

∂ ∂ ∂2
u v + v v − ν( 2 v) = β(Ts − T∞ )gθ
∂x ∂y ∂x

x∞ x∞ x∞ x∞
∂2
Z Z Z Z
∂ ∂
dx u v+ dx v v− dx ν( v) = dx β(Ts − T∞ )gθ
0 ∂x 0 ∂y 0 ∂x2 0

The choice of x∞ is such that the conditions are close enough to freestream.
Ie u = u∞ = 0, T = T∞ .
We can do the same for energy equation

56
 ∂ ∂ ∂2
u
T + v T = α( 2 T )
∂x ∂y ∂x
Z x∞ Z x∞ Z x∞
∂ ∂ ∂2
dx u T + dx v T = dx α( 2 T )
0 ∂x 0 ∂y 0 ∂x
We can start by sorting out the non advection terms first, these are easier.
We assume the thermal diffusivity, expansion and momentum diffusivity take
on average values at film temperature Tf = Ts +T
2
∞

For the energy equation, we can combine back the terms:

∂ ∂ ∂ ∂ ∂2
(uT ) + (vT ) − T ( v + u) = α( 2 T )
∂x ∂y ∂y ∂x ∂x
Fortunately,

∂u ∂v
+ =0
∂x ∂y

∂ ∂ ∂2
(uT ) + (vT ) = α( 2 T )
∂x ∂y ∂x
Z x∞ Z x∞ Z x∞
∂ ∂ ∂2
dx (uT ) + dx (vT ) = dx α( 2 T )
0 ∂x 0 ∂y 0 ∂x
Z x∞ Z x∞
∂ ∂2
(uT )|∞ − (uT )|0 + dx (vT ) = dx α( 2 T )
0 ∂y 0 ∂x
Z x∞ Z x∞
∂ ∂2
dx (vT ) = dx α( 2 T )
0 ∂y 0 ∂x
Z x∞
∂ ∂T ∂T
dx (vT ) = α( |∞ − |0 )
0 ∂y ∂x ∂x
Z x∞
∂ ∂T
dx (vT ) = α(− |0 )
0 ∂y ∂x
And we’ll have to use Leibniz’s rule again...

!
Z y2=b(x) Z y2=b(x)
d d d ∂
f (x, y)dy = f (x, y = b(x)) b(x)−f (x, y = a(x)) a(x)+ f (x, y)dy
dx y1=a(x) dx dx y1=a(x) ∂x

Note that our x and y is swapped around

57
 !
Z x2=b(y) Z x2=b(y)
d d d ∂
f (y, x)dx = f (y, x = b(y)) b(y)−f (y, x = a(y)) a(y)+ f (y, x)dx
dy x1=a(y) dy dy x1=a(y) ∂y

x1 = 0 ; a(y) = 0
x2 = δp (y) ; b(y) = δp

f (y, x) = vT
Substitute into leibniz’s rule:

!
Z x2=δp (y) Z x2=δp (y)
d d d ∂
vT dx = vT (y, x = δp (y)) δp (y)−vT (y, x = 0) 0+ vT dx
dy x1=0 dy dy x1=0 ∂y

!
Z x2=δp (y) Z x2=δp (y)
d d ∂
vT dx = vT (y, x = δp (y)) δp (y) + vT dx
dy x1=0 dy x1=0 ∂y

v = 0 at δp
!
Z x2=δp (y) Z x2=δp (y)
d ∂
vT dx = vT dx
dy x1=0 x1=0 ∂y
Change some variables to make it easier to nondimensionalise

!
Z x2=δp (y) Z x2=δp (y) Z x2=δp (y)
d ∂ ∂
v(T − T∞ )dx = v(T −T∞ )dx = vT dx
dy x1=0 x1=0 ∂y x1=0 ∂y

subs back in energy equation:

Z x∞
∂ ∂T
dx (vT ) = α(− |0 )
0 ∂y ∂x
Z x2=δp (y) !
d ∂T
v(T − T∞ )dx = α(− |0 )
dy x1=0 ∂x

58
 !
Z δp (y)
d ∂T
v(T∞ − T )dx =α |x=0
dy 0 ∂x
Now we are done with energy equation, let’s move on to momentum...

x∞ x∞ x∞ x∞
∂2
Z Z Z Z
∂ ∂
dx u v+ dx v v− dx ν( v) = dx β(Ts − T∞ )gθ
0 ∂x 0 ∂y 0 ∂x2 0

x∞ Z x∞ Z x∞ Z x∞
∂2
Z
∂ ∂u ∂
dx ( (uv)−v )+ dx v v− dx ν( 2 v) = dx β(Ts −T∞ )gθ
0 ∂x ∂x 0 ∂y 0 ∂x 0

For advection terms, we want to eliminate u by considering 2D continuity

equation.

∂u ∂v
+ =0
∂x ∂y
∂v ∂u
=−
∂y ∂x
Substitute back in To get:

x∞ Z x∞ Z x∞ Z x∞
∂2
Z
∂ ∂v ∂
dx ( (uv)+v )+ dx v v− dx ν( 2 v) = dx β(Ts −T∞ )gθ
0 ∂x ∂y 0 ∂y 0 ∂x 0

x∞ Z x∞ Z x∞ Z x∞
∂2
Z
∂ ∂
dx ( (uv))+ dx 2v v− dx ν( 2 v) = dx β(Ts −T∞ )gθ
0 ∂x 0 ∂y 0 ∂x 0

x∞ Z x∞ Z x∞ Z x∞
∂2
Z
∂ ∂ 2
dx ( (uv))+ dx v − dx ν( 2 v) = dx β(Ts −T∞ )gθ
0 ∂x 0 ∂y 0 ∂x 0

The first integral disappears because of no slip and freestream conditions

Z x∞
∂
dx ( (uv)) = −uv|wall + uv|f reestream = 0 − 0 = 0
0 ∂x
So we are left with:

59
 x∞ x∞ x∞
∂2
Z Z Z
∂ 2
dx v − dx ν( 2 v) = dx β(Ts − T∞ )gθ
0 ∂y 0 ∂x 0

We’ll need to have a go with leibniz’s rule

!
Z y2=b(x) Z y2=b(x)
d d d ∂
f (x, y)dy = f (x, y = b(x)) b(x)−f (x, y = a(x)) a(x)+ f (x, y)dy
dx y1=a(x) dx dx y1=a(x) ∂x

Note that our x and y is swapped around

!
Z x2=b(y) Z x2=b(y)
d d d ∂
f (y, x)dx = f (y, x = b(y)) b(y)−f (y, x = a(y)) a(y)+ f (y, x)dx
dy x1=a(y) dy dy x1=a(y) ∂y

x1 = 0 ; a(y) = 0
x2 = δp (y) ; b(y) = δp

f (y, x) = v 2
Substitute into leibniz’s rule:

!
Z x2=δp (y) Z x2=δp (y)
d 2 2 d 2 d ∂ 2
v dx = v (y, x = δp (y)) δp (y)−v (y, x = 0) 0+ v dx
dy x1=0 dy dy x1=0 ∂y
!
Z x2=δp (y) Z x2=δp (y)
d 2 ∂ 2
v dx = v dx
dy x1=0 x1=0 ∂y

x∞ x∞ x∞
∂2
Z Z Z
∂ 2
dx v − dx ν( 2 v) = dx β(Ts − T∞ )gθ
∂y 0 0 ∂x 0

δp δp δp
∂2
Z Z Z
∂ 2
dx v − dx ν( 2 v) = dx β(Ts − T∞ )gθ
∂y 0 0 ∂x 0
The second term is easy:
δp
∂2
Z
∂v ∂v ∂v
dx ν( 2
v) = ν( |δp − |wall ) = ν(− |wall )
0 ∂x ∂x ∂x ∂x

60
 Subs back in:
Z δp Z δp
∂ 2∂v
dx v − ν(− |wall ) = dx β(Ts − T∞ )gθ
∂y 0 ∂x 0
For the last term:

θ(Ts − T∞ ) = (T − T∞ )

Z δp Z δp
dx β(Ts − T∞ )gθ = gβ dx(T − T∞ )
0 0
Substitute back in:
Z δp Z δp
∂ 2∂v
dx v − ν(− |wall ) = gβ dx(T − T∞ )
∂y 0 ∂x 0

9.1.1 BL equation integral soln

The final form of the BL equation is (Bejan, 2013):
Z X Z X
d ∂v
v 2 dx = −ν( )|x=0 + gβ (T − T∞ )dx
dy 0 ∂x 0
Z X
d ∂T
v(T∞ − T )dx = α( )|x=0
dy 0 ∂x
X is an x value big enough where all values T, u and v are in their
freestream values.

9.1.2 Substitution of velocity profiles and temperature profiles

into integral soln
Now we can start to try and solve this, substitute
x
T − T∞ = (Ts − T∞ ) exp(− )
δT
x x
v(x, y) = vY (y) exp(− )(1 − exp(− ))
δp δT
For momentum equation first,
Z X Z X
d 2 ∂v
v dx = −ν( )|x=0 + gβ (T − T∞ )dx
dy 0 ∂x 0

61
 x 2 x
v 2 (x, y) = vY2 (y)[exp(− )] [(1 − exp(− ))]2
δp δT
2x x 2x
v 2 (x, y) = vY2 (y)[exp(− )][1 − 2 exp(− ) + exp(− )]
δp δT δT

2x 2x x 2x 2x
v 2 (x, y) = vY2 (y)[exp(− ) − 2 exp(− ) exp(− ) + exp(− ) exp(− )]
δp δp δT δp δT

2x 2 1 2 2
v 2 (x, y) = vY2 (y)[exp(− ) − 2 exp(−x( + )) + exp(−x( + ))]
δp δp δT δp δT

Z X Z X Z X Z X
2 2x 2 1 2 2
v dx = vY2 (y)[ dx exp(− )− dx 2 exp(−x( + ))+ dx exp(−x( + )]
0 0 δp 0 δp δT 0 δp δT

Before we continue, combine into single fraction...

2 1 2δT + δp
( + )=
δp δT δT δp
2 2 2δT + 2δp
( + )=
δp δT δT δp
now we are ready:

X
−δp −2δp δT −δP δT
Z
2x 2 1 2 2
v 2 dx = vY2 (y)[ exp(− )− exp(−x( + ))+ exp(−x( + )]|x=X
0 2 δp 2δT + δp δp δT 2δT + 2δp δp δT x=0

X
−δp
Z
2x 2δp δT 2 1 δP δT 2 2
v 2 dx = vY2 (y)[ exp(− )+ exp(−x( + ))− exp(−x( + )]|x=X
0 2 δp 2δT + δp δp δT 2δT + 2δp δp δT x=0

First evaluate at x = X where X is such that velocity will be at freestream

value.
When x is large, the terms in the bracket drop to near 0.

−δp 2X 2δp δT 2 1 δP δT 2 2
[ exp(− )+ exp(−X( + ))− exp(−X( + )] ≈ 0
2 δp 2δT + δp δp δT 2δT + 2δp δp δT

62
 But we cannot just get rid of it yet... you have to compare it to something
else.
let’s also evaluate it at x=0,

−δp 2δp δT δP δT
[ + − ]
2 2δT + δp 2δT + 2δp
The integral evaluates to:

−δp 2X 2δp δT 2 1 δP δT 2 2
exp(− )+ exp(−X( + ))− exp(−X( + )]
2 δp 2δT + δp δp δT 2δT + 2δp δp δT

−δp 2δp δT δP δT
−[ + − ]
2 2δT + δp 2δT + 2δp
−δp 2δp δT δP δT
≈ −[ + − ]
2 2δT + δp 2δT + 2δp
δp 2δp δT δP δT
=[ − + ]
2 2δT + δp 2δT + 2δp

δp (2δT + δp )(2δT + 2δp ) 2δp δT (2)(2δT + 2δp ) 2(2δT + δp )δP δT

=[ − + ]
2(2δT + δp )(2δT + 2δp ) 2(2δT + δp )(2δT + 2δp ) 2(2δT + δp )(2δT + 2δp )

δp (2δT + δp )(2δT + 2δp ) − 4δp δT (2δT + 2δp ) + 2(2δT + δp )δP δT

=
2(2δT + δp )(2δT + 2δp )

δp (2δT + δp )(δT + δp ) − 4δp δT (δT + δp ) + (2δT + δp )δP δT

=
2(2δT + δp )(δT + δp )

To make life easier, we introduce this ratio (Bejan, 2013)

δp
q=
δT
divide top and bottom by δT2

δp (2 + q)(δT + δp ) − 4δp (δT + δp ) + (2δT + δp )δP

=
2(2 + q)(δT + δp )
δp (2 + q)(1 + q) − 4δp (1 + q) + (2 + q)δP
=
2(2 + q)(1 + q)

63
 (2 + q)(1 + q) − 4(1 + q) + (2 + q)
= δp
2(2 + q)(1 + q)
(2 + q)(1 + q) − 4(1 + q) + (2 + q)
= δp
2(2 + q)(1 + q)
q2
= δp
2(2 + q)(1 + q)
substitute into integral:

X
−δp
Z
2x 2δp δT 2 1 δP δT 2 2
v 2 dx = vY2 (y)[ exp(− )+ exp(−x( + ))− exp(−x( + )]|x=X
0 2 δp 2δT + δp δp δT 2δT + 2δp δp δT x=0

X
q2
Z
v 2 dx = vY2 (y)δp
0 2(2 + q)(1 + q)
subs back into equation:
Z X
vY2 (y)δp q 2
 
d ∂v
= −ν( )|x=0 + gβ (T − T∞ )dx
dy 2(2 + q)(1 + q) ∂x 0
Let’s do the viscosity term

∂v ∂ x x
= (vY (y) exp(− )(1 − exp(− )))
∂x ∂x δp δT
∂ x x
= vY (y) (exp(− )(1 − exp(− )))
∂x δp δT
∂ x x x
= vY (y) (exp(− ) − exp(− ) exp(− )))
∂x δp δp δT
∂ x 1 1
= vY (y) (exp(− ) − exp(−x( + )))
∂x δp δp δT
−1 x −1(δT + δp ) 1 1
= vY (y)( exp(− ) − exp(−x( + )))
δp δp δT δp δp δT
1 x (δT + δp ) 1 1
= vY (y)(− exp(− ) + exp(−x( + )))
δp δp δT δp δp δT
Now evaluate at x=0,

64
 ∂v 1 (δT + δp )
( )|x=0 = vY (y)(− + )
∂x δp δT δp
∂v 1 (1 + q)
=( )|x=0 = vY (y)(− + )
∂x δp δp
∂v vY (y)q
( )|x=0 = ( )
∂x δp
substitute back:
Z X
vY2 (y)δp q 2
 
d vY (y)q
= −ν + gβ (T − T∞ )dx
dy 2(2 + q)(1 + q) δp 0

x
T − T∞ = (Ts − T∞ ) exp(− )
δT
Z X Z X
x
(T − T∞ ) = (Ts − T∞ ) dx exp(− )
0 0 δT
x x=X
= (Ts − T∞ )(−δT ) exp(− )|
δT x=0
X
= (Ts − T∞ )(−δT )[exp(− ) − 1]
δT

= (Ts − T∞ )(δT )[(1)]

= (Ts − T∞ )(δT )
substitute back:

vY2 (y)δp q 2
 
d vY (y)q
= −ν + gβ(Ts − T∞ )(δT )
dy 2(2 + q)(1 + q) δp

9.1.3 Momentum equation transformed... after substitution

vY2 (y)δp q 2
 
d vY (y)q δp
= −ν + gβ(Ts − T∞ )
dy 2(2 + q)(1 + q) δp q

65
9.1.4 now onto the energy equation
Z X
d ∂T
v(T∞ − T )dx = α( )|x=0
dy 0 ∂x
Z X
d ∂T
− v(T − T∞ )dx = α( )|x=0
dy 0 ∂x
x
T − T∞ = (Ts − T∞ ) exp(− )
δT
x x
v(x, y) = vY (y) exp(− )(1 − exp(− ))
δp δT

x x x
(T − T∞ )v(x, y) = (Ts − T∞ ) exp(− )vY (y) exp(− )(1 − exp(− ))
δT δp δT

x x x
= (Ts − T∞ )vY (y) exp(− ) exp(− )(1 − exp(− ))
δT δp δT
x x 2x
= (Ts − T∞ )vY (y) exp(− )(exp(− ) − exp(− ))
δp δT δT

1 1 2 1
= (Ts − T∞ )vY (y)(exp(−x( + )) − exp(−x( + )))
δT δp δT δp

δT + δp δT + 2δp
= (Ts − T∞ )vY (y)(exp(−x ) − exp(−x )
δT δp δT δp
Now we can integrate

Z X Z X
δT + δp δT + 2δp
v(T∞ −T )dx = dx (Ts −T∞ )vY (y)(exp(−x )−exp(−x )
0 0 δT δp δT δp

Z X Z X
δT + δp δT + 2δp
v(T∞ −T )dx = (Ts −T∞ )vY (y) dx (exp(−x )−exp(−x )
0 0 δT δp δT δp

Z X
δT δp δT + δp δT δp δT + 2δp x=X
v(T∞ −T )dx = (Ts −T∞ )vY (y)(− exp(−x )−(− ) exp(−x )|x=0
0 δT + δp δT δp δT + 2δp δT δp

66
Z X
δT δp δT + δp δT δp δT + 2δp x=X
v(T∞ −T )dx = (Ts −T∞ )vY (y)(− exp(−x )+( ) exp(−x )|x=0
0 δT + δp δT δp δT + 2δp δT δp

Evaluate at x=X, we find that it is negligible with respect to the first

term.

Z X
δT δp δT + δp δT δp δT + 2δp x=X
v(T∞ −T )dx = (Ts −T∞ )vY (y)(− exp(−x )+( ) exp(−x )|x=0
0 δT + δp δT δp δT + 2δp δT δp

Z X
δT δp δT + δp δT δp δT + 2δp
v(T∞ −T )dx = (Ts −T∞ )vY (y)(−1)(− exp(−x )+( ) exp(−x )|x=0
0 δT + δp δT δp δT + 2δp δT δp

Z X
δT δp δT δp
v(T∞ − T )dx = (Ts − T∞ )vY (y)(−1)(− + )
0 δT + δp δT + 2δp

We are going to reintroduce q, divide top and bottom by δT

Z X
δp δp
v(T∞ − T )dx = (Ts − T∞ )vY (y)(−1)(− + )
0 1 + q 1 + 2q

Z X
1 1
v(T∞ − T )dx = (Ts − T∞ )vY (y)(−1)δp (− + )
0 1 + q 1 + 2q

X
−(1 + 2q) + (1 + q)
Z
v(T∞ − T )dx = (Ts − T∞ )vY (y)(−1)δp ( )
0 (1 + q)(1 + 2q)

X
−q
Z
v(T∞ − T )dx = (Ts − T∞ )vY (y)δp (−1)( )
0 (1 + q)(1 + 2q)

substitute back
Z X
d ∂T
− v(T − T∞ )dx = α( )|x=0
dy 0 ∂x

67
 d −q ∂T
− (−1)(Ts − T∞ )vY (y)δp ( ) = α( )|x=0
dy (1 + q)(1 + 2q) ∂x
d −qvY (y)δp ∂T
(Ts − T∞ ) ( ) = α( )|x=0
dy (1 + q)(1 + 2q) ∂x
divide both sides by (Ts − T∞ )

d qvY (y)δp ∂T
(Ts − T∞ ) ( ) = −α( )|x=0
dy (1 + q)(1 + 2q) ∂x
d qvY (y)δp ∂θ
( ) = −α( )|x=0
dy (1 + q)(1 + 2q) ∂x
x
T − T∞ = (Ts − T∞ ) exp(− )
δT
x
θ = exp(− )
δT
∂θ 1 x 1 q
( )|x=0 = [− exp(− )]x=0 = − =−
∂x δT δT δT δp
subs back to obtain:

d qvY (y)δp q
( )=α
dy (1 + q)(1 + 2q) δp
One more assumption to make, q is independent of y since q = f(Pr)
only.

d vY (y)δp α
( )=
dy (1 + q)(1 + 2q) δp

9.1.5 Energy and momentum equation for integral solution

d vY (y)δp α
( )=
dy (1 + q)(1 + 2q) δp

vY2 (y)δp q 2
 
d vY (y)q δp
= −ν + gβ(Ts − T∞ )
dy 2(2 + q)(1 + q) δp q
Solution proceudre, integrate equations. we will need suitable BCs for
vY , δp
we don’t really know how both of these will vary except for:

δp ∼ y 1/4 ; v ∼ y 1/2

68
 [correction to last video]

v ∼ y 1/2 6= vy (y) ∼ y 1/2

x x
v(x, y) = vY (y) exp(− )(1 − exp(− ))
δp δT
x x
v ∼ y 1/2 → vY (y) exp(− )(1 − exp(− )) ∼ y 1/2
δp δT

y 1/2 exp( δxp )

vY (y) ∼
1 − exp(− δxT )

y 1/2
vY (y) ∼
exp(− δxp ) − exp(−x( δ1p + 1
δT ))

y 1/2
vY (y) ∼ δT +δp
exp(− δxp ) − exp(−x( δp δT ))

y 1/2
vY (y) ∼
exp(− δxp ) − exp(−x( q+1
δp ))

Now this means we have only δp = f (y) here, so we can just substitute,

y 1/2
vY (y) ∼ q+1
exp(− Ayx1/4 ) − exp(−x( Ay 1/4 ))

so we will use the form,

By 1/2
vY (y) = q+1
exp(− Ayx1/4 ) − exp(−x( Ay 1/4 ))

yikes! more work to do...

δp = Ay 1/4
Z
4 4y
dyδp = Ay 5/4 = δp +C
5 5
using this logic,
[correction to last video]

By 1/2
Z Z
vY (y) vY (y) 4y
dy = dy x q+1 6= C2 +
δp 1/4
Ay [exp(− Ay1/4 ) − exp(−x( Ay1/4 ))] δp 5

69
 Now i’m NOT going to integrate this...
and also,
Z Z
α α α 4 α 4 α 4y
dy = dy 1/4 = y 3/4 = y =
δp Ay A 3 Ay 1/4 3 δp 3
okay, so we will integrate from y ∗ = 0 to y ∗ = y

d vY (y)δp α
( )=
dy (1 + q)(1 + 2q) δp

vY2 (y)δp q 2
 
d vY (y)q δp
= −ν + gβ(Ts − T∞ )
dy 2(2 + q)(1 + q) δp q
Integrating... with dummy variable y ∗

d vY (y)δp α
( )=
dy (1 + q)(1 + 2q) δp

vY2 (y)δp q 2
 
d vY (y)q δp
= −ν + gβ(Ts − T∞ )
dy 2(2 + q)(1 + q) δp q
d vY (y ∗ )δp α
∗
( )=
dy (1 + q)(1 + 2q) δp

vY2 (y ∗ )δp q 2 vY (y ∗ )q
 
d δp
∗
= −ν + gβ(Ts − T∞ )
dy 2(2 + q)(1 + q) δp q
vY (y ∗ )δp
Z y Z y
d α
dy ∗ ∗
( ) = dy ∗
0 dy (1 + q)(1 + 2q) 0 δp
[correction to last video]
I’m not going to integrate this equation, too hard. I’ll do derivatives of
the left hand side instead...

y
vY2 (y ∗ )δp q 2 ∗
Z   Z y Z y
∗ d ∗ vY (y )q δp
dy ∗
=− dy ν + dy ∗ gβ(Ts −T∞ )
0 dy 2(2 + q)(1 + q) 0 δp 0 q

vY2 (y)δp q 2
 
d vY (y)q δp
= −ν + gβ(Ts − T∞ )
dy 2(2 + q)(1 + q) δp q
let’s evaluate the derivatives using

70
 By 1/2
vY (y) = q+1
exp(− Ayx1/4 ) − exp(−x( Ay 1/4 ))

δp = Ay 1/4
We will need to evaluate,

vY2 (y)δp q 2 q2
 
d d 2
= (v δp )
dy 2(2 + q)(1 + q) 2(2 + q)(1 + q) dy Y

d 2 d By 1/2
(vY δp ) = ( )2 Ay 1/4
dy dy exp(− x1/4 ) − exp(−x( q+1 ))
Ay Ay 1/4

d 2 d B2y 1/4
(vY δp ) = q+1 Ay
dy dy exp(− 2x1/4 ) + exp(−2x q+1 ) − 2 exp(− x
) exp(−x )
Ay Ay 1/4 Ay 1/4 Ay 1/4

d 2 d B2y
(vY δp ) = Ay 1/4
dy dy exp(− 2x1/4 ) + exp(−2x q+1 ) − 2 exp(− x
(1 + q + 1))
Ay Ay 1/4 Ay 1/4

d 2 d B2y
(vY δp ) = Ay 1/4
dy dy exp(− 2x1/4 ) + exp(−2x q+1 ) − 2 exp(− x
(q + 2))
Ay Ay 1/4 Ay 1/4

Now this is a BIG thing to differentiate, i may not want to do it this

way since it’s very tedious. Let’s try something easier...
we’ll differentiate vy directly, and use product rule

By 1/2
vY (y) = q+1
exp(− Ayx1/4 ) − exp(−x( Ay 1/4 ))

d d By 1/2
vY (y) =
dy dy exp(− x1/4 ) − exp(−x( q+1 ))
Ay Ay 1/4
now we note,

δp = Ay 1/4

d d By 1/2
vY (y) =
dy dy exp(− δx ) − exp(−x( q+1
δp ))
p

71
 i make an assumption here,
x
x+ = 6= f (y)
δp
using what we derived,

vY (y)δp 4y α
( )=
(1 + q)(1 + 2q) 3 δp

vY2 (y)δp q 2
 
4y vY (y)q 4y δp
=− ν + gβ(Ts − T∞ )
2(2 + q)(1 + q) 5 δp 5 q

9.1.6 one more equation...

Problem, we have 3 unknowns, q, δp and vy (y), we only have 2 equations.
Where is equation 3?
[CORRECTION, X is not BL thickness, but just x value big enough for
all properties to reach freestream values]
From textbook (Bejan, 2013),
consider near the wall, in the region just next to the wall. Since velocity
is so slow, we can consider that buoyancy forces balance viscous forces.
Advection is small compared to viscous forces.

∂2v
0=ν + gβ(T − T∞ )
∂x2
This becomes our third equation to play with. So let’s substitute:
x
T − T∞ = (Ts − T∞ ) exp(− )
δT
x x
v(x, y) = vY (y) exp(− )(1 − exp(− ))
δp δT
we are going to differentiate twice the velocity,
x x
v(x, y) = vY (y) exp(− )(1 − exp(− ))
δp δT
x 1 1
v(x, y) = vY (y)(exp(− ) − exp(−x[ + ]))
δp δT δp
x δp + δT
v(x, y) = vY (y)(exp(− ) − exp(−x ))
δp δT δp

72
Reintroducing q,
x q+1
v(x, y) = vY (y)(exp(− ) − exp(−x ))
δp δp

Differentiating twice wrt x,

∂v −1 x q+1 q+1
= vY (y)( exp(− ) − (− ) exp(−x ))
∂x δp δp δp δp

∂2v 1 x q+1 2 q+1

2
= vY (y)( 2 exp(− ) − ([− )] exp(−x ))
∂x δp δp δp δp

∂2v 1 x (q + 1)2 q+1

= v Y (y)( exp(− ) − ( ) exp(−x ))
∂x2 δp2 δp δp2 δp
Now remember, we are evaluating near the wall x ≈ 0

∂2v 1 (q + 1)2
= v Y (y)( − ( ))
∂x2 δp2 δp2
Let’s do likewise for the temperature, x ≈ 0
x
T − T∞ = (Ts − T∞ ) exp(− )
δT

T − T∞ |x≈0 = (Ts − T∞ )
Substitute back

∂2v
0=ν + gβ(T − T∞ )
∂x2
1 (q + 1)2
0 = νvY (y)( 2
−( )) + gβ(Ts − T∞ )
δp δp2

0 = νvY (y)(1 − (q + 1)2 ) + δp2 gβ(Ts − T∞ )

0 = νvY (y)(1 − (q 2 + 2q + 1)) + δp2 gβ(Ts − T∞ )

0 = νvY (y)(−(q 2 + 2q)) + δp2 gβ(Ts − T∞ )

νvY (y)q(q + 2) = δp2 gβ(Ts − T∞ )

73
9.1.7 Solving
νvY (y)q(q + 2) = δp2 gβ(Ts − T∞ )
vY (y)δp 4y α
( )=
(1 + q)(1 + 2q) 3 δp

vY2 (y)δp q 2
 
4y vY (y)q 4y δp
=− ν + gβ(Ts − T∞ )
2(2 + q)(1 + q) 5 δp 5 q
So now with 3 equations, 3 unknowns, we should be in theory, able to
solve this completely.
First strategy: eliminate vY , so let’s square the energy equation

vY (y)δp 4y α
( )=
(1 + q)(1 + 2q) 3 δp

vY2 (y)δp2 42 y 2 α 2
( ) =
(1 + q)2 (1 + 2q)2 32 δp2

vY2 (y)δp2 1 16y 2 α2

( ) =
(1 + q)(1 + 2q) (1 + q)(1 + 2q) 9 δp2

vY2 (y)δp2 1 8y 2 α2
( ) =
2(1 + q)(1 + 2q) (1 + q)(1 + 2q) 9 δp2

vY2 (y)δp2 q 2 1 8y 2 α2 2
( ) = q
2(1 + q)(1 + 2q) (1 + q)(1 + 2q) 9 δp2

vY2 (y)δp2 q 2 8y 2 α2 2
( )= q (1 + q)(1 + 2q)
2(1 + q)(1 + 2q) 9 δp2
This will be a natural fit to replace the LHS of the momentum equation.
So we will just multiply the momentum equation thru by δp
" #
vY2 (y)δp2 q 2 4y vY (y)q 4y δp2
=− ν + gβ(Ts − T∞ )
2(2 + q)(1 + q) 5 1 5 q
And replace the term accordingly

8y 2 α2 2 4y vY (y)q 4y δp2
q (1 + q)(1 + 2q) = − ν + gβ(Ts − T∞ )
9 δp2 5 1 5 q

74
 Now then, we then make vY the subject of equation 3, and eliminate
that from the leftmost term on RHS

δp2 gβ(Ts − T∞ )
vY (y) =
νq(q + 2)
And substituting,

8y 2 α2 2 4y δp2 gβ(Ts − T∞ ) 4y δp2

q (1 + q)(1 + 2q) = − νq + gβ(Ts − T∞ )
9 δp2 5 νq(q + 2) 5 q

8y 2 α2 2 4y δp2 gβ(Ts − T∞ ) 4y δp2

q (1 + q)(1 + 2q) = − + gβ(Ts − T∞ )
9 δp2 5 (q + 2) 5 q

8y 2 α2 2
 
4y 2 1 1
q (1 + q)(1 + 2q) = δ gβ(Ts − T∞ ) − +
9 δp2 5 p (q + 2) q

8y 2 α2 2
 
4y 2 q q+2
q (1 + q)(1 + 2q) = δ gβ(Ts − T∞ ) − +
9 δp2 5 p (q + 2)q q(q + 2)

8y 2 α2 2
 
4y 2 2
q (1 + q)(1 + 2q) = δ gβ(Ts − T∞ )
9 δp2 5 p q(q + 2)

2y α2 2
 
1 2 2
q (1 + q)(1 + 2q) = δp gβ(Ts − T∞ )
9 δp2 5 q(q + 2)

10y α2 2
 
2 2
q (1 + q)(1 + 2q) = δp gβ(Ts − T∞ )
9 δp2 q(q + 2)

5y α2 3
q (1 + q)(1 + 2q)(2 + q) = δp2 gβ(Ts − T∞ )
9 δp2
5y 2 3
α q (1 + q)(1 + 2q)(2 + q) = δp4 gβ(Ts − T∞ )
9
Now this is decently neat. We will need to further eliminate terms. δp
So we will combine the third equation with the energy equation

δp2 gβ(Ts − T∞ )
vY (y) =
νq(q + 2)

75
 vY (y)δp 4y α
( )=
(1 + q)(1 + 2q) 3 δp
Substituting,

δp2 gβ(Ts − T∞ ) δp 4y α
=
νq(q + 2) (1 + q)(1 + 2q) 3 δp

δp4 gβ(Ts − T∞ ) 1 4y
= α
νq(q + 2) (1 + q)(1 + 2q) 3
4y
δp4 gβ(Ts − T∞ ) =ανq(q + 2)(1 + q)(1 + 2q)
3
Now we can substitute this in,
5y 2 3
α q (1 + q)(1 + 2q)(2 + q) = δp4 gβ(Ts − T∞ )
9
to obtain,

5y 2 3 4y
α q (1 + q)(1 + 2q)(2 + q) = ανq(q + 2)(1 + q)(1 + 2q)
9 3

5y 2 3 4y
α q = ανq
9 3
Introducing Pr
5y 3 4y
q = P rq
9 3
5y 2 4y
q = Pr
9 3
5 2
q = Pr
3∗4
5 2
q = Pr
12
5 2
q = Pr
6
So we get a result that is consistent as before, q only depends on Pr
(momentum and temperature BL thickness ratio depends only on Pr)

76
Part VII
Resources Online
For Momentum BLs:

http://web.mit.edu/fluids-modules/www/highspeed_flows/ver2/bl_Chap2.pdf
https://community.dur.ac.uk/suzanne.fielding/teaching/BLT/sec3.pdf

for Von Karman

https://nptel.ac.in/content/storage2/courses/112104118/lecture-29/29-3_momentum.htm

For Thermal BLs:

http://mech.sut.ac.ir/People/Courses/18/Chapter3-%20Part2.pdf
http://raops.org.in/epapers/june15_9.pdf

https://nptel.ac.in/content/storage2/courses/112101001/downloads/lec25.pdf

Part VIII
Github Repo
https://github.com/theodoreOnzGit/heatTransferTheory_YouTube

Look under convection heat transfer...

Part IX
Bibliography
References
Bejan, A. (2013). Convection heat transfer. John wiley & sons.
Welty, J., Rorrer, G. L., & Foster, D. G. (2014). Fundamentals of momentum,
heat, and mass transfer. John Wiley & Sons.

77