Question 1

Considert turbulent flow past an isothermal flat plate on lenght L and width b, with
constant (𝜌, 𝜇, c p , k). Assume Pr = 1, that is δ u = 𝛿 T , with uniform stream velocity U
and stream temperature T e at the plate leading edge. At the trailing edge, x = L, the
mean velocity and temperature may be approximated by one-seventh power-law profiles:
T - TW 1/7
u y
≈ ≈
U Te - TW δ

With no further information, estimate the drag force and total heat transfer of one
side of the plate, in terms of the boundary layer thickness δ and other flow parameters.

The flat-plate control-volume analysis is valid for turbulent flow, as long as the mean
profiles are known. Thus, the drag force is given by:
δ 1 7
F = ∫ 𝜌u(U - u)b dy = 𝜌δbU 2 ∫0 η 1/7 1 - η 1/7 dη = 𝜌δbU 2
0 72

In like manner, the total heat transfer is given, for either laminar or turbulent flow, by
the flat-plate thermal integral analysis, neglecting kinetic energy changes:
dQ 𝛿 1
= ∫ 𝜌cp u(T - Te )b dy = 𝜌δcp U(TW - Te )∫ η 1/7 1 - η 1/7 dη
dt 0 0

dQ 7
or: = 𝜌δbcp U(TW - Te )
dt 72

The analysis would be complete if we had an accurate estimate for the growth of the
boundary layer thickness, δ(x). For example, use of Prandtl's flat plate estimate,
δ / L = 0.37 / ReL1/5 , gives the dimensionless drag force formula
2F 2(7 / 72)(0.37) 0.072
CD = ≈ ≈ (turbulent flat-plate flow)
𝜌U 2 bL ReL1/5 ReL1/5

which is popular in the literature. Similarly, the Stanton number could be

approximated as
dQ / dt 0.37(7 / 72) 0.036
CH = = =
bL𝜌Ucp ReL1/5 ReL1/5

Question 2
Known: Water flow over a flat plate; expressions for the dependence of the local
convection coefficient with distance from the plate's leading edge, x; and approximate
temperature of the water.
 ⏨.
Find: Average convection coefficient, h
Schematic:

Laminar Turbulent

u∞

Ts
y
δ t (x )

xc
L
x

Assumptions:
1. Steady-state conditions
2. Transition occurs at a critical Reynolds number of Re x,c = 5 × 10 5

T ≈ 300K) : 𝜌 = 𝜈f-1 = 997 kg / m 3 , 𝜇 = 855 × 10 -6 N. s / m 2

Properties: water ( ⏨
T ≈ 350K) : 𝜌 = vf-1 = 974kg / m 3 , 𝜇 = 365 × 10 -6 N. s / m 2
(⏨

Analysis: The local convection coefficient is highly dependent on whether laminar or

turbulent conditions exist. Therefore, we first determine the extent to which these
conditions exist by finding the location where transition occurs, x c . We know that at 300K
Rex,c 𝜇 5 × 10 5 × 855 × 10 -6 N. s / m 2
xc = = = 0.43 m
𝜌u∞ 997 kg / m 3 × 1 m / s

while at 350K,
Rex,c 𝜇 5 × 10 5 × 365 × 10 -6 N. s / m 2
xc = = = 0.19 m
𝜌u∞ 974 kg / m 3 × 1 m / s

We also know that

1 L 1 xc L
h⏨= ∫
L 0
hdx =
L
∫0 hlam dx + ∫ hturb dx
xc

or
xc L
1 Clam Cturb
⏨
h= x 0.5 + x 0.8
L 0.5 0 0.8 xc

At 300K,
 1 395 W / m 1.5 . K 0.5 0.5 2330 W / m 1.8 . K
⏨
h= × 0.43 m + × 0.6 0.8 - 0.43 0.8 m 0.8
0.6m 0.5 0.8

= 1620 W / m 2 . K
While at 350K,

1 477 W / m 1.5 . K 0.5 0.5 3600 W / m 1.8 . K

⏨
h= × 0.19 m + × 0.6 0.8 - 0.19 0.8 m 0.8
0.6m 0.5 0.8

= 3710 W / m 2 . K

The local and average convection coefficient distributions for the plate are shown in
the figure below.

8000
h, 300K
h, 350K
6000 h⏨, 350K
k .2m / W ⏨

4000
, h ,h

h⏨, 300K
2000

0
0 0.2 0.4 0.6
x (m)

Comments:
1. The average convection coefficient at T ≈ 350K is over twice as large as the
value at T ≈ 300K. This strong temperature dependence is due primarily to the significant
shift of x c that is associated with the smaller viscosity of the water at the higher
temperature. Careful consideration of the temperature dependence of fluid properties is
crucial when performing a convection heat transfer analysis.
2. Spatial variations in the local convection coefficient are significant. The largest
local convection coefficients occur at the leading edge of the flat plate, where the laminar
thermal boundary layer is extremely thin, and just downstream of x c , where the turbulent
boundary layer is thinnest.

Question 3
For water at 20°C and 1 atm, take 𝜌 = 998 kg / m 3 and 𝜇 = 0.001 kg / m. s. Then
 𝜌UL (998)(6.0)(1.0)
ReL = = = 5, 988, 000
𝜇 (0.001)

Thus the first one-sixth of the plate flow is laminar, and the rest turbulent. If we ignore
the laminar portion and assume fully turbulent flow, then the power-law equations give
0.16 L (0.16)(1.0) 𝛿
𝛿 ( L) = = ≈ 0.0172m, 𝛿 * = = 0.0022 m
ReL1/7 (5988000) 1/7 8

0.027 1
C f ( L) = = 0.0029, or: 𝜏W (L) = (0.0029) (998)(6) 2 = 52 Pa
ReL1/7 2

7 1
CD = Cf (L) = 0.00339, or: Drag = (0.00339) (998) 6 2 0.6 m 2 ≈ 36, 5 N
6 2

We don't have a "long-law" estimate for 𝛿(L), but integral analysis predicts
Cf (L) ≈ 0.00278 or 𝜏W (L) = 50 Pa, and CD = 0.00320, or Drag ≈ 34.4 N.
If we account for the laminar leading-edge now, for Re x,tr = 1E6, we equate
momentum thicknesses there to obtain Re xo = 36600, or x(origin) ≈ 0.161 m. Then
Leff = 0.839 m, and the thickness formula predicts
𝜌ULeff 0.16Leff 0.16(0.839)
ReL,eff = = 5025000, or 𝛿(L) = = ≈ 0.0148 m
𝜇 ReL1/7
, eff (5025000) 1/7

Similarly, the trailing edge shear stress is estimated by

0.027 1
C f ( L) = ≈ 0.00298, or: 𝜏W (L) = (0.00298) (998)(6) 2 = 53.5 Pa
ReL1/7
, eff 2

Finally, using Schlichting's idea for drag force with a laminar region
0.031 A 0.031 1440
CD = - ≈ - = 0.003095
ReL1/7 ReL (5988000) 1/7 5988000

1
or: Drag = (0.003096) (998) 6 2 0.6 m 2 ≈ 33.4 N
2

These latter "laminar-leading-edge" results would be the preferred engineering

estimates.

Question 4
The definition of L-averaged Nusselt number tells us what we have to do:
 q''0-L L
Nu 0-L =
k ΔT
Where
x tr L-xtr
1
q''0-L = ∫0 q''lam dx + ∫ q''turb dx
L 0

x tr L - x tr

laminar turbulent

For the laminar section we write

k ΔT
q''lam (x) = 0.332 Pr 1/3 Rex1/2
Pr>1
x
hence
x tr
∫0 q''lam dx = 0.664 k ΔT Pr 1/3 Rex1/2
tr

We do the same for the turbulent section

q''turb (x) = 𝜌 cp U∞ ΔT St x
Pr>1
⏜⏟⏟⏟ ⏝⏟⏟⏟ ⏞
-2/3 1 (Colburn)
Pr Cf,x
2
⏜⏟⏟⏟ ⏝⏟⏟⏟⏞ (Prandtl)
0.0296 Rex-1/5
hence
L-xtr
∫0 q''turb dx = 0.037 𝜌 cp U ∞ ΔT Pr -2/3 (U∞ / 𝜈) -1/5 (L - xtr ) 4/5

In conlusion, we obtain
q''0-L L
Nu 0-L = = Pr 1/3 0.664 Retr1/2 + 0.037(ReL - Rextr ) 4/5
k ΔT
 MAE3401- Aerodynamics II
Problem set 10: Turbulence II
In this problem set you will have the opportunity to put in practice some of the concepts you saw in
both the lecture videos and the workshop. This will be about turbulent flows.

General instructions: Please show all your workings carefully, and make sure that all you’ve written is
intelligible. You will still get marks if you show your workings (and drawings), even if you don’t reach
the final answer.

Question 1: You are asked to estimate the drag and heat transfer from an isothermal flat plate at
temperature 𝑇! . The flow past the isothermal flat plate of length L and width b, with constant
(𝜌, 𝑐" , 𝜇, 𝑘) is turbulent. Assume 𝑃𝑟 = 1, that is 𝛿# = 𝛿$ = 𝛿, with uniform free-stream velocity U
and free-stream temperature 𝑇% at the plate leading edge. At the trailing edge, 𝑥 = 𝐿 , the mean
velocity and mean temperature may be approximated by one-seventh power-law profiles:
𝑢 𝑇 − 𝑇! 𝑦 &/(
≈ ≈4 6
𝑈 𝑇% − 𝑇! 𝛿
Starting with a control volume analysis similar to the one used in the derivation of momentum
thickness:

a) What are the dimensions of momentum and heat flux, and what are these equivalent to,
physically? What product of fluid properties gives a heat flux?
b) Estimate the drag force and total heat transfer rate of one side of the plate, in terms of the
boundary layer thickness and other flow parameters

Question 2: Water flows at a velocity u = 1 m/s over a flat plate of length L= 0.6 m. Consider two cases,
one for which the water temperature is approximately 300K and the other for an approximate water
temperature of 350K. In the laminar and turbulent regions, experimental measurements show that
the local convection coefficients are well described by

ℎ)*+ (𝑥) = 𝐶)*+ 𝑥 ,-./ ℎ0#12 (𝑥) = 𝐶0#12 𝑥 ,-.3

Here, x has units of m, and the coefficients for each temperature are 𝐶)*+,5-- = 395 𝑊/𝑚&./ ⋅ 𝐾 ,
𝐶0#12,5-- = 2330 𝑊/𝑚&.6 ⋅ 𝐾, 𝐶)*+,5/- = 477 𝑊/𝑚&./ ⋅
𝐾 , 𝐶0#12,5/- = 3600 𝑊/𝑚&.6 ⋅ 𝐾. As is evident, the
constant, 𝐶, depends on the nature of the flow as well as
the water temperature because of the thermal
dependence of various properties of the fluid. Consider
the critical Reynolds number 𝑅𝑒01 = 5 × 10/ and the
properties of water: 𝜌5-- = 997𝑘𝑔/𝑚5 , 𝜇5-- = 855 ×
10,7 𝑁 ⋅ 𝑠/𝑚3 , 𝜌5/- = 974𝑘𝑔/𝑚5 , 𝜇5/- = 365 ×
,7 3
10 𝑁 ⋅ 𝑠/𝑚

a) Determine the average convection coefficient, ℎN, over the entire plate for the two water
temperatures.
b) Plot ℎ(𝑥) and ℎN for the two cases and compare the results.
Question 3: An aerodynamic test is being undertaken using a water tunnel at matched Reynolds
number to the air experiment. Consider that the water tunnel experiment will be conducted using
a smooth flat plate 1 m long and 600 mm wide with water at 20∘ 𝐶 and 1 atm, flowing at 6 m/s
past it. Knowing that 𝜌 = 998 𝑘𝑔/𝑚5 and 𝜇 = 0.001 𝑘𝑔/(𝑚 ⋅ 𝑠) Estimate:
a) the trailing edge boundary layer thickness,
b) the trailing edge wall shear stress
c) the drag of one side of the plate, if 𝑅𝑒01 = 107 . Note that the drag may be obtained from
integration of the shear stress.

Question 4: You are tasked with deriving an equation for the length averaged Nusselt number for
the case of heat transfer in boundary layer flow from an isothermal plate at temperature at zero
angle to a constant temperature stream (𝑈9 , 𝑇9 ). The leading laminar section of the boundary
layer has a length comparable with the length of the trailing turbulent section; consequently, the
heat flux averaged over the entire plate length is influenced by both sections. Assume that the
;! :
laminar-turbulent transition is located at a point where 𝑅𝑒: = <
= 3.5 × 10/ .

Question 5: To start your revision: List all the main different flows covered this semester and
write out the assumptions for each one. Which ones have a closed analytical solution?
Question 1
a)
Start from N-S in tensor form
∂ui
=0
∂xi

∂ui ∂ui 1 ∂p ∂ 2 ui
+ uj = - +𝜈
∂t ∂xj 𝜌 ∂xi ∂xj2

Assume u i (x, y, z, t) = U i (x, y, z) + u i (x, y, z, t)

1 T
Properties (where denotes time-average ∫
T 0
★ dt ) T ≫ t *

I) u i = U i + u i

But U i doesn't depend on t ⇒ U i = U i

II) ui = 0 (assumed to have zero-mean)

∂ui ∂
III) = ui (derivative & average are linear operators, or leibniz rule)
∂xj ∂xj

Start with continuity

∂ Ui + ui ∂Ui ∂u i
=0⇒ + =0
∂xi ∂xi ∂xi

∂Ui ∂ui ∂ Ui ∂ ui
Apply the average: + = + =0
∂xi ∂xi ∂xi ∂xi

∂Ui
⇒ =0
∂xi

Now for momentum:

 ∂Ui ∂u i ∂u i ∂Ui ∂Ui ∂u i
+ + Uj + uj + Uj + uj =
∂t ∂t ∂xj ∂xj ∂xj ∂xj

1 ∂P 1 ∂p ∂ 2 Ui ∂ 2 ui
=- - +𝜈 2 +𝜈 2
𝜌 ∂xi 𝜌 ∂xi ∂xj ∂xj

∂Ui
Considering statistically stationary: =0
∂t
Applying the average

∂u i ∂u i ∂Ui ∂Ui ∂u i
+ Uj + Uj + uj + uj
∂t ∂xj ∂xj ∂xj ∂xj

1 ∂p ∂P ∂ 2 ui ∂ 2 Ui
=- + +𝜈 +𝜈
𝜌 ∂xi ∂xi ∂xj2 ∂xj2

As a + b = a + b

∂u i ∂ ∂u i ∂u i ∂
But: = u = 0 and Uj = Uj = Uj u = 0, as
∂t ∂t i ∂xj ∂xj ∂xj i
Uj does not depend on time. Similarly for the other terms.

∂Ui ∂u i 1 ∂P ∂ 2 Ui
⇒ Uj + uj = - +𝜈 2
∂xj ∂xj 𝜌 ∂xi ∂xj

∂ ui uj ∂u i ∂u j
But = uj + ui (continuity)
∂xj ∂xj ∂xj

∂Ui 1 ∂P ∂ ∂U i
⇒ Uj = - + 𝜈 - ui uj
∂xj 𝜌 ∂xi ∂xj ∂xj

This equation is exactly the same as for steady laminar flows, except for the last term
(Reynold stresses). This term is related to turbulent non-linear interactions and acts like a
turbulent viscosity, increasing momentum diffusion and mixing, in general.
 b) This is done by simply subtracting the rans equations from the equations for
Ui + ui
∂u j
Continuity: =0
∂xj

Momentum:
☆
☆☆
⏜⏟⏟⏟ ⏝⏟⏟⏟ ⏞
∂u i ∂u i ∂Ui ⏜⏟⏟⏟⏟
∂
⏝⏟⏟⏟⏟ ⏞ 1 ∂p ∂ 2 ui
+ Uj + uj + u u - ui uj = - +𝜈 2
∂t ∂xj ∂xj ∂xj i j 𝜌 ∂xi ∂xj

The final equation now contains time-derivatives and extra terms associated with
convection of disturbances ( ☆ ). The equation is also linear, except for term ( ☆☆ ),
which is responsible for the chaotic behaviour of turbulent flow. That term also defines the
flux of energy in the dynamics of turbulent flows.

Question 2
The process is the same as previously. The difference here is that the flow is
axisymmetric. Thus:
∂
=0; W =0; uw = vw =0
∂θ

2 2
with w the azimuthal velocity. Note that w ≠ 0, as w is always positive.

Question 3
In fully developed turbulent pipe flow, the mean temporal derivatives vanish and
velocity statistics only depend on the radial coordinate. Hence, continuity reduces to
∂
(r V ) = 0
∂r
which, with the boundary condition V r=R , integrates to

V = 0.

The radial momentum equation then simplifies to

1∂ p 1∂ w2
0=- - r v2 +
𝜌 ∂r r ∂r r
 1∂ p ∂ v2 w2 - v2
=- - +
𝜌 ∂r ∂r r

Differentiating in the streamwise direction yields

2
1∂ p
0=-
𝜌 ∂x∂r
and thus
∂ p dpw
= = const.
∂x dx

The axial momentum equation simplifies to

1∂ p 1∂ ∂2 U
0=- - (r uv ) + 𝜈
𝜌 ∂x r ∂r ∂r 2

dpw ∂ ∂2 U
= -r - 𝜌 (r uv ) + r𝜈𝜌
dx ∂r ∂r 2

which, using integration by parts, can be integrated to

1 dpw ∂ U
= - r2 - 𝜌r uv + 𝜈𝜌 r - U +C
2 dx ∂r

∂ U
where C is a constant of integration. At r = 0, symmetry implies that = 0, and
∂r
hence C = 𝜈𝜌 U , leading to

1 dpw ∂ U
r = - 𝜌 uv + 𝜈𝜌 ≡ 𝜏( r )
2 dx ∂r

At r = R, uv vanishes and thus

dpw 2 ∂ U 𝜏w
= 𝜈𝜌 ≡ -2
dx R ∂r r=R
R

Question 4:
a) According to the definition of the material derivative we get
D𝜙 ∂𝜙 ∂𝜙
= + Uj (1)
Dt ∂t ∂xj
The Reynolds decomposition of the velocity is
 U = U +u (2)
Substituting (2) into (1), we get
D𝜙 ∂𝜙 ∂𝜙
= + ( Uj + uj )
Dt ∂t ∂xj

∂𝜙 ∂𝜙 ∂(uj 𝜙)
= + Uj + (3)
∂t ∂xj ∂xj

The second step in Eq. (3) follows from ∇ · u = 0. Defining

⏨
D ∂
= + U ·∇ (4)
⏨
Dt ∂t
we get

D𝜙 ⏨𝜙
D
= + ∇ · ( u𝜙 ) (5)
Dt ⏨
Dt

b) From part (a), we have

D𝜙 ∂𝜙 ∂𝜙 ∂(uj 𝜙)
= + Uj + (6)
Dt ∂t ∂xj ∂xj
So the mean is

D𝜙 ∂𝜙 ∂𝜙 ∂(uj 𝜙)
= + Uj +
Dt ∂t ∂xj ∂xj

∂ 𝜙 ∂ 𝜙 ∂ uj 𝜙
= + Uj +
∂t ∂xj ∂xj

⏨𝜙
D
= + ∇ · u𝜙 (7)
⏨t
D

Question 5
a) The Reynolds stress tensor u i u j is symmetric positive semi-definite. Its
diagonal components u12 = u1 u1 , u22 and u32 are non-negative. Thus Eq.(a)

is incorrect because that the value of u 32 is negative.

b) The given components u 1 u 2 and u 2 u 1 are not equal, i.e., u i u j is not
symmetric, so Eq. (b) is incorrect.
c) The Reynolds stresses:
 1/2
Random variables u 1 , u 2 , and u' = u 2

2
u1 u2
± ⩾0
u'1 u'2

u12 u22 u1 . u2
⇒ + ±2 ⩾0
u' 12 u' 22 u'1 . u'2
Take average
u12 u22 u1 . u2
+ ±2 ⩾0
u' 12 u' 22 u'1 . u'2

u1 . u2
⇒1⩾± 1/2
u12 u22

u1 . u2
Define 𝜌 12 =
1/2
u12 u22

In Eq. (c), 𝜌 12 = 1.5 > 1, so it is incorrect.

 MAE3401- Aerodynamics II
Problem set 9: Turbulence I
In this problem set you will have the opportunity to put in practice some of the concepts you saw in
both the lecture videos and the workshop. This will be about turbulent flows.

General instructions: Please show all your workings carefully, and make sure that all you’ve written is
intelligible. You will still get marks if you show your workings (and drawings), even if you don’t reach
the final answer.

Question 1: Starting from the incompressible Navier-Stokes system with constant viscosity, use the
Reynolds decomposition 𝑢(𝑥, 𝑦, 𝑧, 𝑡) = 𝑈(𝑥, 𝑦, 𝑧) + 𝑢,(𝑥, 𝑦, 𝑧, 𝑡), where 𝑈 is the time averaged flow
and 𝑢, is the zero-mean fluctuations around 𝑈 (similarly for the other velocity components and
pressure), and

a) Obtain an equation that describes the behaviour of the time-averaged component of

turbulent flows, justifying each step of your derivation. What are the main differences
between the resulting new set of equations and the original Navier-Stokes equations? What
is the physical significance of the new terms of these equations?
b) From the equations for the total field (𝑢) and the time-averaged field (𝑈), derive a new
equation that describes the evolution of the flow fluctuations around this mean flow. What
are the main differences between the resulting new set of equations and the original Navier-
Stokes equations? What is the physical significance of the new terms of these equations?

Question 2: Derive the Reynolds-Averaged Navier-Stokes equations in cylindrical polar coordinates for
a statistically stationary axisymmetric non-swirling flow. Hint: using these assumptions, you may
consider the azimuthal components 𝑊, 〈𝑢𝑤〉, 〈𝑣𝑤〉 to be zero.

Question 3: Starting from the Reynolds equations in polar-cylindrical coordinates show that, for fully
developed turbulent pipe flow, the shear stress

is given by

where 𝑝! (𝑥) is the mean pressure at the wall. Hence obtain the relation

where the wall shear stress - a positive quantity - is defined as 𝜏! = −𝜏(𝑅).

Question 4: For a random field 𝜙(𝒙, 𝑡), obtain the results

 : /𝐷
Where 𝐷 : 𝑡 indicates the linearised total derivative.

Question 5: Each of the following equations is incorrect. Why?

 Question 1
Known: Airflow over an isothermal plate
Find: Cooling rate per unit width of plate, q' (W / m).
Schematic:

Air
T∞ = 300°C
u∞ = 10 m / s
p ∞ = 5 kN / m 2
Ts = 27°C

L = 0.5m
x

Assumptions:
1. Steady-state conditions
2. Negligible radiation effects

Properties: air (T f = 437 K, p = 1 atm): 𝜈 = 30.84 × 10 -6 m 2 / s,

k = 36.4 × 10 -3 W / m ⋅ K, Pr = 0.687. Properties such as k, Pr and 𝜇 may be
assumed to be independent of pressure to an excellent approximation. However, for a
gas, the kinematic viscosity 𝜈 = 𝜇 / 𝜌 will vary with pressure through its dependence on
density. From the ideal gas law, 𝜌 = p / RT, it follows that the ration of kinematic
viscosities for a gas at the same temperature but at different pressures, p 1 and p 2 is
(𝜈1 / 𝜈2 ) = (p2 / p1 ). Hence the kinematic viscosity of air at 437 K and
p∞ = 6 × 10 3 N / m 2 is

-6 2 1.0133 × 10 5 N / m 2
𝜈 = 30.84 × 10 m /s× 3 2
= 5.21 × 10 -4 m 2 / s
6 × 10 N / m

Analysis: For a plate of unit width, it follows from Newton's law of cooling that the
rate of convection heat transfer to the plate is

q' = ⏨
h L( T ∞ - T s )

To determine the appropriate convection correlation for computing ⏨

h , the Reynolds
number must first be determined
 u∞ L 10 m / s × 0.5 m
ReL = = = 9597
𝜈 5.21 × 10 -4 m 2 / s

Hence the flow is laminar over the entire plate, and the appropriate correlation is
given by:

1/2 1/3
⏨⏨
Nu L = 0.664Re L Pr = 0.664(9597) 1/2 (0.687) 1/3 = 57.4

The average convection coefficient is then

⏨⏨
Nu L k 57.4 × 0.0364 W / m ⋅ K
⏨
h= = = 4.18 W / m 2 ⋅ K
L 0.5 m

and the required cooling rate per unit width of plate is

q' = 4.18 W / m 2 ⋅ K × 0.5 m (300 - 27)°C = 570 W / m

Question 2
Known: Surface dimensions for an array of 10 silicon chips. Maximum allowable chip
temperature. Air flow conditions.
Find: Maximum allowable chip electrical power (a) without and (b) with a turbulence
promoter at the leading edge.
Schematic:

u∞ = 40 m / s
Air
T∞ = 24°C
Ts ⩽ 80°C
p = 1 atm 10mm 10th chip

x L = 0.1m

Assumptions: (1) Steady-state conditions, (2) Film temperature of 52°C, (3)

Negligible radiation, (4) Negligible heat loss through insulation, (5) Uniform heat flux at
chip interface with air, (6) Re x,c = 5 × 10 5 .
Properties: Air (T f = 325K, 1atm): 𝜈 = 18, 4 × 10 -6 m 2 / s, k = 0.0282 W / m ⋅ K,
Pr = 0.703.
Analysis: Re L = u ∞ L / 𝜈 = 40 m / s × 0.1 m / 18.4 × 10 -6 m 2 / s = 2.174 × 10 5 .
Hence, flow is laminar over all chips without the promoter.
(a) For laminar flow, the minimum h x exists on the last chip. Approximating the average
coefficient for Chip 10 as the local coefficient at x = 95 mm, ⏨
h 10 = hx = 0.095 m.

k
h 10 = 0.453 Rex1/2 Pr 1/3
⏨
x

u∞ x 40 m / s × 0.095 m
Rex = = = 1.065 × 10 5
𝜈 18.4 × 10 -6 2
m /s

0.0282 W / m ⋅ K 1/2
⏨
h 10 = 0.453 2.065 × 10 5 (0.703) 1/3 = 54.3 W / m 2 ⋅ K
0.095

W
q10 = ⏨
h 10 A(Ts - T∞ ) = 54.3 2
(0.01m) 2 (80 - 24)°C = 0.30 W.
m ⋅K

Hence, if all chips are to dissipate the same power and T s is not to exceed 80°C.
qmax = 0.30 W.

Comments: It is far better to orient array normal to the air flow. Since ⏨
h 1 > h⏨10 , more
heat could be dissipated per chip, and the same heat could be dissipated from each chip.

Question 3
Observations: (i) The determination of the Nusselt number requires the determination
of the velocity and temperature distributions. (ii) Velocity is assumed linear. (iii) Surface
temperature is variable. (iv) Newton's law of cooling gives surface heat flux. This requires
knowing the local heat transfer coefficient.
Problem Definition: Determine the velocity and temperature distribution for boundary
layer flow over a flat plate at variable surface temperature. This reduces to determining
the viscous and thermal boundary layer thickness.
Solution Plan: Start with the definition of local Nusselt number. Equate Newton's law
with Fourier's law to obtain an equation for the heat transfer coefficient h. Apply the
integral form of the momentum and energy equations to determine the velocity and
temperature distribution. Apply Newton's law of cooling to determine surface heat flux.
Plan Execution:
(i) Assumptions: (1) Steady state, (2) constant properties, (3) two-dimensional, (4)
laminar flow Re x < 5 × 10 5 , (5) viscous boundary layer flow (Re x > 100), (6) thermal
boundary layer (Pe > 100), (7) uniform upstream velocity and temperature, (8) flat plate,
(9) negligible changes in kinetic and potential energy, (10) negligible axial condution, (11)
negligible dissipation, (12) no buoyancy β = 0 or g = 0 and (13) Pr ≫ 1.
(ii) Analysis: The local Nusselt number is defined as
 hx
Nu x = (a)
k
The heat transfer coefficient h is given by
∂T(x,0)
-k ∂y
h= (1)
T s ( x) - T ∞

Thus the temperature distribution T(x, y) must be determined. SUrface heat flux is
obtained using Newton's law of cooling
q''s = h(Ts - T∞ ) (b)

The integral form of the energy equation is used to determine temperature distribution

∂T(x, 0) d δ t (x )
-α = ∫ u(T - T∞ )dy (2)
∂y dx 0

Axial velocity distribution u(x, y) for Pr ≫ 1 is assume to vary linearly with normal
distance y from the plate. Thus
y
u = V∞ (c)
δ
where δ is the viscous boundary layer thickness. Before proceeding with the
determination of temperature distribution, δ must be determined. We apply the integral
form of the momentum equation
∂u(x, 0) d δ (x ) d δ (x )
𝜈 = V∞ ∫ udy - ∫ u 2 dy (3)
∂y dx 0 dx 0

Substitute (c) into (3)

V∞ d δ (x ) 1 d δ (x ) 1
𝜈
δ
= 2
V∞ ∫
dx 0 δ
2
ydy - V∞ ∫
dx 0 2
y 2 dy
δ

Evaluate the integrals and simplify

1 V∞ dδ
𝜈 =
δ 6 dx

Seperate variables and integrate

𝜈 δ2
6 dx = + C1 (d)
V∞ 2
 The constant of integration C 1 is obtained from the boundary condition on δ
δ ( 0) = 0
The condition gives C 1 = 0. Substituting into (d) and solving for δ

12𝜈
δ= x (e)
V∞

Turning now to the temperature distribution, we assume a third degree polynomial

T ( x, y) = b o ( x) + b 1 ( x) y + b 2 ( x) t 2 + b 3 ( x) y 3 (f)

The temperature boundary conditions are:

( 1) T ( x, 0) = T s ( x)

( 2) T ( x, δ t ) ≈ T ∞

∂T(x, δt )
( 3) ≈ 0
∂y

∂ 2 T ( x, 0)
( 4) =0
∂y 2

The four boundary conditions are used to determine the coefficients in (f). The
assumed profile becomes

3 y 1 y3
T(x, y) = Ts (x) + [T∞ - Ts (x)] - (g)
2 δt 2 δt3

Substitute (g) into (1)

3k
h( x) = (h)
2 δt
Introducing (h) into (a)
3x
Nu x = (i)
2 δt

Thus the problem reduces to determining the thermal boundary layer thickness δ t .
This is accomplished using the integral form of the energy equation (2). Substituting (c)
and (g) into (2)
 3 α [ T s ( x) - T ∞ ] d δ t (x ) y 3 y 1 y3
= ∫ [ T s ( x) - T ∞ ] 1 - - dy
2 V∞ δt dx 0 δ 2 δt 2 δt3

Evaluate the integral

3 α [ T s ( x) - T ∞ ] 1 d T s ( x) - T ∞
= δt2 (j)
2 V∞ δt 10 dx δ

However, surface temperature is given by

Ts - T∞ = C x (k)

Substitute into (j)

α x d x
15 = δt2 (l)
V∞ δt dx δ

To solve (l) for δ t (x) we frist rewrite (l)

2
α x δ d δt
15 = xδ (m)
V∞ δ δt dx δ

Use (e) to eliminate 1 / δ on the left side of (m) and δ on the right side
2
5α δ d δt
= x (n)
4 𝜈 δt dx δ

To solve (n) for δ t / 𝛿, let

δt
x = z2 (o)
δ
Substitute into (n)
5α dz
x = z2
8𝜈 dx
Separate variables and integrate
10 α 3/2 1 3
x = z + C2 (p)
24 𝜈 3

Where C 2 is a constant determined from the boundary condition δ t (x)

δ t ( 0) = 0 (q)
 Apply (q) to (o)
z( 0 ) = 0 (r)
Apply (r) to (p) gives C 1 = 0. Equation (p) becomes
5 α 3/2
x = z3 (s)
4𝜈
Use (o) to eliminate z in (s)
3
5α δt
=
4𝜈 δ
δt
Use (e) to eliminate δ and solve for
x
1/3
δt 5 3 α 𝜈
= (t)
x 2 𝜈 V∞ x

Note that
𝜈
= Pr (u)
α
and
V∞ x
Rex = (v)
𝜈
Substitute (u) and (v) into (t)

δt 5 3
= [Pr] -1/3 [Rex ] -1/2 (w)
x 2

(w) into (i) gives the local Nusselt number

3 (x)
Nu x = [Pr] 1/3 [Rex ] 1/2
5

To examine surface heat flux we determine the heat transfer coefficient. Substitute
(a) into (w), and solve for h(x)

3k
h= [Pr] 1/3 [Rex ] 1/2
5 x

Use (v) to eliminate the Reynolds number in the above

1/2
3 V∞
h= k[Pr] 1/3 x (y)
5 𝜈
 Substitute (k) and (y) into (b), simplify
1/2
3 1/3
V∞
q''s = Ck[Pr] (z)
5 𝜈

This result shows that surface heat flux is uniform.

Checking: Dimensional check: (1) Equations (i), (t), (w), and (x) are dimensionless.
(2) Equations (e), (g), (h), (y), and (z) are dimensionally correct.
Boundary conditions check: Assumed temperature (g) satisfies the four
boundary conditions.
Limiting check: If surface temperature is the same as free stream
temperature, the heat flux will be zero. According to (k) T s (x) = T ∞ when C = 0. Setting
C = 0 in (z) gives q''s = 0.

Question 4
Known: Liquid metal in parallel flow over a flat plate
Find: An expression for the local Nusselt number
Schematic:

δT
u∞ , T∞ Ts
y
Pr ≪ 1 δ

Assumptions: (1) Steady, incompressible flow, (2) δ ≪ δ t , hence u(y) ≈ u ∞ , (3)

Boundary layer approximations are valid, (4) Constant properties.
Analysis: The boundary layer energy equation is
∂T ∂T ∂2T
u +v =𝛼 2
∂x ∂y ∂y

Assuming u(y) = u ∞ , it follows that v = 0 and the energy equation becomes

∂T ∂2T ∂T α ∂2T
u∞ =𝛼 2 or =
∂x ∂y ∂x u∞ ∂y 2

Boundary Conditions: T(x, 0) = T s , T(x, ∞) = T ∞

Initial Condition: T(0, y) = T ∞

The differential equation is analogous to that for the first Stokes problem. Hence, the
solution is given by
T ( x, y) - T s y
= erf
T∞ - Ts 2(α x / u∞ ) 1/2

k(T s - T ∞ )
q''s =
(𝜋 α x / u∞ ) 1/2
Hence, with
hx q''s x
Nu x = =
k (T s - T ∞ )k
find
1/2
x (xu∞ ) 1/2 1 𝜌u∞ x cp 𝜇
Nu x = = = ⋅
(𝜋 α x / u∞ ) 1/2 𝜋 1/2 (k / 𝜌 cp ) 1/2 𝜋 1/2 𝜇 k

Nu x = 0.564(Rex Pr) 1/2 = 0.564Re 1/2

Where Pe = Re ⋅ Pr is the Peclet number.

Comments: Because k is very large, axial conduction effects may not be negligible.
That is, the 𝛼 ∂ 2 T / ∂x 2 term of the energy equation may be important.
 MAE3401- Aerodynamics II
Problem set 8: Thermal boundary layers and convection
In this problem set you will have the opportunity to put in practice some of the concepts you saw in
both the lecture videos and the workshop. This will be about thermal boundary layers and convection.

General instructions: Please show all your workings carefully, and make sure that all you’ve written is
intelligible. You will still get marks if you show your workings (and drawings), even if you don’t reach
the final answer.

Note: The critical Reynolds number for a flat plate is 5 × 105 . Any use of laminar approximations must
be justified.

Question 1: Air at a temperature of 300°C flows with a velocity of 10m/s over a flat plate 0.5m long.
Estimate the cooling rate per unit width of the plate needed to maintain it at a surface temperature
of 27°C.
𝑊
You may use the following values: 𝜈 = 5.21 × 10−4 𝑚2 /𝑠, 𝑘 = 36.4 × 10−3 𝑚∙𝐾
, 𝑃𝑟 = 0.687

Question 2: It is possible to show that, for laminar flow, the local Nusselt number considering a
uniform surface heat flux imposed at the plate is given by
1/2
𝑁𝑢𝑥 = 0.453 𝑅𝑒𝑥 𝑃𝑟1/3
This expression is valid for 𝑃𝑟 > 0.6.

An array of 10 square silicon chips, each of length L = 10mm on a side, is insulated on one surface and
cooled on the opposite surface by atmospheric air in parallel flow with 𝑇∞ = 24°𝐶 and 𝑢∞ =
40 𝑚/𝑠. When in use, the same electrical power is dissipated in each chip, maintaining a uniform heat
flux over the entire cooled surface.

If the temperature of each chip may not exceed 80°C, what is the maximum allowable power per chip?
Would it be preferable to orient the array normal, instead of parallel, to the airflow?
𝑊
You may use the following values: 𝜈 = 18.4 × 10−6 𝑚2 /𝑠, 𝑘 = 28.2 × 10−3 𝑚∙𝐾, 𝑃𝑟 = 0.703
Question 3: A plate is cooled by a fluid with Prandtl number 𝑃𝑟 ≫ 1, free-stream temperature 𝑇∞
and constant velocity in the streamwise direction. Measurements of the surface temperature 𝑇𝑠
indicate that it varies with distance from the leading edge according to

𝑇𝑠 = 𝑇∞ + 𝐶 √𝑥

where C is constant. Use a third-degree polynomial temperature profile to determine the local
Nusselt number and the heat flux at the plate. Assume laminar boundary layer flow which for this
fluid can be approximated by a linear boundary layer streamwise velocity profile within the
thermal boundary layer.

You will be following a similar process to the final example in the Week 9 lecture videos, with a
𝑢 𝑦 12𝜈𝑥
linear velocity profile 𝑢 = 𝛿 and corresponding boundary layer thickness 𝛿 = √ 𝑢
∞ ∞

Question 4: Consider a liquid metal (𝑃𝑟 ≪ 1), with free stream conditions 𝑢∞ and 𝑇∞ , in parallel flow
over an isothermal flat plate at 𝑇𝑠 . Assuming that 𝑢 = 𝑢∞ throughout the thermal boundary layer,
write the corresponding form of the boundary layer energy equation. Applying appropriate initial (x =
0) and boundary conditions, solve this equation for the boundary layer temperature field, T(x, y). Use
the result to obtain an expression for the local Nusselt number Nu (it will be necessary to look up the
derivative of the error function). Hint: This problem is analogous to the first Stokes problem (plate
suddenly placed in motion). You may use the exact same form of solution.
 Question 1
a) The displacement thickness 𝛿 * of the boundary layer is the amount by which an
ideal inviscid flow would have to be displaced above the solid bboundary to satisfy the
continuity of mass equation for the combined flow through the boundary layer and the free
stream. Thus in Fig. (a) there is a real flow with a free stream velocity u 0 beyond the
boundary layer and a flow within the boundary layer varying in velocity from u 0 to zero.
For the puporses of analysis this can be regarded as equivalent to an inviscid flow only
slipping over the displacement surface as shown in Gig. (b). The displacement thickness
can be determined by applying the continuity of mass equation to the real and inviscid
flows as follows.
In Fig. (a) the mass flow rate m ␒ through the space between the solid surface and
some arbitrary point Y is, for unit width perpendicular to the diagram

y=𝛿
m␒ = 𝜌0 u0 (Y - 𝛿) + ∫ 𝜌0 udy
y=0

u0 u0

Density 𝜌 0
𝜌0

edge y
dary layer
Boun

u < u0 ent surface

𝛿 Displacem
u=0 𝛿*

(a) Real flow field (b) Inviscid flow field

In Fig. (b) the mass flow rate through the space between the 𝛿 * displacement
surface and the same point Y will be, for unit width
m␒ = 𝜌0 u0 Y - 𝛿 *
For these mass flows to be the same
𝛿
𝜌0 u0 𝛿 * = 𝜌0 u0 𝛿 - ∫ 𝜌0 udy
0
But since
𝛿
𝜌0 u0 𝛿 = ∫ 𝜌0 u0 dy
0

𝛿 u
𝛿 = ∫ 1-
*
dy
0 u0
 The displacement thickness 𝛿 * can be considered as a measure of the mass flow
defict due to the boundary layer velocity profile which could be written directly, for unit
𝛿
width, as ∫0 𝜌0 (u0 - u)dy. Equating this to the mass flow which would have occurred
between the solid boundary and the displacement surface for unit width 𝜌 0 u 0 𝛿 *

𝛿
𝜌0 u0 𝛿 * = ∫ 𝜌0 (u0 - u)dy
0

𝛿 u
𝛿 = ∫ 1-
*
dy
0 u0

If 𝛿 * is very small compared with 𝛿 then the velocity profile is very full and the
boundary layer has a minimum effect on the displacement of the external flow. If
𝛿 * approaches 𝛿 / 2 then the external flow is restricted by about half the boundary layer
thickness. In ducted flows this displacement causes fluid in the centre of the duct to
accelerate, while for external flows this displacement causes streamlines to move away
from the body surface.
The momentum thickness 𝜃 is defined as the amount by which an inviscid flow
would have to be displaced from the solid boundary to make the momentum of the
resulting ideal flow equivalent to the momentum of the real flow through the boundary
layer and the free stream. As the velocities in the boundary layer are less than that in the
free stream, the boundary layer produces both a mass flow deficit and a momentum
deficit. At any point in the boundary layer

momentum deficit = local mass flow deficit x local velocity

Integrating across the boundary layer

𝛿
momentum deficit = ∫0 𝜌0 (u0 - u)udy

This will be equal to the momentum of the free stream fluid which would pass
through the space between the momentum surface and the solid boundary so that for unit
width

𝛿
𝜌0 u02 𝜃 = ∫ 𝜌0 (u0 - u)udy
0
 𝛿 u u
𝜃=∫ 1- dy
0 u0 u0

If 𝜃 is small compared with 𝛿, the boundary layer velocity profile is full and the
momentum deficit low.
The rate of growth of the momentum thickness in the x direction d𝜃 / dx is
proportional to the local shear force between the fluid and the surface.

b) Since

𝛿 u u y y2
𝛿* = ∫ 1 - dy and =2 - 2
0 u0 u0 𝛿 𝛿

𝛿 y y2
𝛿 = ∫ 1 - 2 - 2 dy
*
0 𝛿 𝛿

Putting
y/𝛿 = 𝜂 dy = 𝛿d𝜂

The limits of integration become

y = 0, 𝜂=0
y = 𝛿, 𝜂=1
So

1
𝛿 * = 𝛿∫ 1 - 2𝜂 - 𝜂 2 d𝜂
0

1
𝜂3 1
= 𝛿 𝜂 - 𝜂2 + = 𝛿
3 0 3

𝛿* 1
=
𝛿 3

Question 2
a) Considering
 d𝜃
𝜏w = 𝜌0 u02
dx
𝛿
and 𝜃 = ∫0 uu 1-
u
u0
dy
0

Putting

u y
= =𝜂
u0 𝛿

1 𝛿
𝜃 = 𝛿∫ 𝜂 - 𝜂 2 d𝜂 =
0 6

Thus

𝜌0 u02 d𝛿
𝜏w =
6 dx

And since

du u0
𝜏w = 𝜇0 , 𝜏w = 𝜇0
dy 𝛿
y=0

giving

𝜇0 u0 𝜌0 u02 d𝛿
=
𝛿 6 dx

Separating the variables

𝜇0
𝛿. d𝛿 = 6 dx
𝜌0 u0

Integrating
 𝛿2 𝜇0
=6 x+C
2 𝜌0 u0

If 𝛿 = 0 at x = 0, then C = 0

1/2
𝜇0
𝛿 = 3.46 x 1/2
𝜌0 u0

Substituting

1/2
u0 𝜇0 u0 𝜌0 u0
𝜏w = 𝜇0 = x 1/2
𝛿 3.46 𝜇0

which, when plotted along the plate, will appear as in Fig. 2. The shear stress would
be infinite at the leading edge and asumptotic to the x axis.

𝜏w

x
Fig. 2

The total drag on a plate of span b and streamwise length l is obtained by integrating
the elemental wall shear forces along the length of the plate. For a plate wetted on one
side only

x=1
Drag = D = b ∫x=0 𝜏w dx

For a plate wetted on both sides

1
D = 2b∫ 𝜏w dx
0
Substituting for 𝜏 w
 1/2
2 𝜌0 u0 2
Total drag on the plate = 𝜇0 u0 b∫ x 1/2 dx
3.46 𝜇0 0

1/2
4 𝜌0 u0
= 𝜇0 u0 bl 1/2
3.46 𝜇0

The Drag coefficient C D for the plate is defined as

Drag Drag
CD = =
1 1
𝜌 u2
2 0 0
× wetted area 𝜌 u2
2 0 0
× 2bl

CD = 1.155(Rel ) -1/2

Where Re l is the flow Reynolds number based on the length of the plate in the
streamwise direction.

b) For a retangular fin of the given dimensions

1000 × 12 × 0.15
Rel = = 1.2 × 10 6
0.0015
1/2
1
CD = 1.155
6
= 1.054 × 10 -3
1.2 × 10

1
Total drag on fin = C D × 𝜌 0 u 02 × 2bl
2

= 1.054 × 10 -3 × 1000 × 12 2 × 0.15 × 0.8 N

= 18.22 N

Power absorbed = drag force × velocity

= 18.22 × 12 = 218.64 W

Question 3
For air at 20°C, take 𝜌 = 1.2 kg / m 3 and 𝜇 = 1.8E - 5 kg / m. s. The velocities u at
each pitot inlet can be estimated from the Blasius solution:

𝜂1 = y[U / vx1 ] 1/2 = (0.002){ 15 / [1.5E - 5(0.5)]} 1/2 = 2.83, Table: read f' ≈ 0.816
(1)
 Then u 1 = Uf' = 15(0.816) ≈ 12.25 m / s

𝜂2 = y[U / vx1 ] 1/2 = 2.0, f' ≈ 0.630, u2 = 15(0.630) ≈ 9.45 m / s (2)

Assume constant stream pressure, then the manometers are a measure of the local
velocity u at each position of the pilot inlet, so we can find Δp across each manometer:

𝜌 2 1.2
Δp1 = u1 = (12.25) 2 = 90 Pa = Δ𝜌gh1 = (998 - 1.2)(9.81)h1 , h1 ≈ 9.2mm
2 2

𝜌 2 1.2
Δp2 = u2 = (9.45) 2 = 54 Pa = (998 - 1.2)(9.81)h2 , h2 ≈ 5.5mm
2 2

Question 4

y
r
U
𝛿 (x )

u
p = constant

Assume "axisymmetric" flow, that is, v 𝜃 = 0 and ∂ / ∂𝜃 = 0 everywhere. The

boundary layer assumptions are:

∂u ∂u ∂vr ∂vr ∂p
v r ≪ u; ≪ ; ≪ ; hence r-momentum becomes ≈0
∂x ∂r ∂x ∂r ∂r

Thus p ≈ p(x) only, and for a long straight cylinder, p ≈ constant and U ≈ constant.
Then, with ∂p / ∂x = 0, the x-momentum equation becomes

∂u ∂u 𝜇 ∂ ∂u
𝜌u + 𝜌vr ≈ r when 𝛿 = R
∂x ∂r r ∂r ∂r

∂u 1 ∂
plus continuity: + (rvr ) ≈ 0 when 𝛿 = R
∂x r ∂r
 For thick boundary layers (part b) the radial geometry is important.
If, however, the boundary layer is very thin, 𝛿 ≪ R, then r = R + y ≈ R itself, and
we can use (x, y) :

∂u ∂vr
Continuity: + ≈ 0 if 𝛿 ≪ R
∂x ∂y

∂u ∂u ∂2u
x-momentum: 𝜌u + 𝜌vr ≈𝜇 2 if 𝛿 ≪ R
∂x ∂y ∂y

Thus a thin boundary-layer on a cylinder is exactly the same as flat-plate (Blasius)

flow.

Question 5

y
U 𝛿

v0

0 x

Substitute these (u, v) into the x-momentum equation with ∂u / ∂x = 0 :

∂u ∂u ∂2u
𝜌u + 𝜌v ≈ 𝜇 2 , or: 0 + 𝜌(v0 ) -CUo e Cy ≈ 𝜇 -C 2 Uo e Cy ,
∂x ∂y ∂y

or: C = 𝜌v 0 / 𝜇 = constant < 0

If the constant is negative, u does not go to ∞ and the solution represents laminar
boundary-layer flow past a flat plate with wall suction, v o ⩽ 0. It satisfies

at y = 0 : u = 0 (no slip) and v = v o (suction); as y → ∞, u → U o (freestream)

The thickness 𝛿, where u ≈ 0.99U o , is defined by exp(𝜌v o 𝛿 / 𝜇) = 0.01, or

𝛿 = -4.6𝜇 / 𝜌vo .
 MAE3401- Aerodynamics II
Problem set 7: Application of boundary layers
In this problem set you will have the opportunity to put in practice some of the concepts you saw in
both the lecture videos and the workshop. This will be about velocity boundary layers. We will talk
about thermal BLs next week.

General instructions: Please show all your workings carefully, and make sure that all you’ve written
is intelligible. You will still get marks if you show your workings (and drawings), even if you don’t
reach the final answer.

Question 1: Answer the following:

(a) Explain the significance of the boundary layer displacement thickness 𝛿 ∗ and the momentum
thickness 𝜃 and express them as integrals of the boundary layer velocity profiles on a
smooth flat plate.
(b) Calculate the ratio 𝛿 ∗ /𝛿 for a laminar boundary layer with a velocity profile given by

𝑢 2𝑦 𝑦 #
= − #
𝑢" 𝛿 𝛿
Question 2: Answer the following:

(a) Use the momentum integral equation to obtain a general expression for the drag coefficient
𝐶$ of a rectangular plate, span 𝑏 and streamwise length 𝑙, wetted both sides and with
% &
laminar boundary layers having the approximate velocity profile % = '
!
(b) Calculate the power absorbed in skin friction drag by a thin rectangular fin having a span of
0.8 m and a streamwise length of 0.15 m when moving through still water at 12 m/s. For
water, density 𝜌" = 1000 𝑘𝑔/𝑚( and viscosity 𝜇" = 0.0015 𝑁𝑠/𝑚# .

Question 3: Air at 20°C and 1 atm flows past the flat plate in the
figure below. The two pitot tubes are each 2 mm from the wall.
The manometer fluid is water at 20°C. If U = 15 m/s and L = 50
cm, determine the values of the manometer readings ℎ) and ℎ#
in cm. Assume laminar boundary-layer flow. Tabulated
*
solutions for Blasius (taken from the textbook, with 𝜂 = 𝑦:#+,)
are available below.
Question 4: Derive modified forms of the laminar boundary-layer equations for flow along the
outside of a circular cylinder of constant radius R. Consider the two cases (a) 𝛿 ≪ 𝑅 ; and (b) 𝛿 ≈ 𝑅.
What are the boundary conditions?

-.
Question 5: Show that the two-dimensional laminar-flow pattern with -, = 0

𝑢 = 𝑈" (1 − 𝑒 /& ) ; 𝑣 = 𝑣" < 0

is an exact solution to the boundary-layer equations. Find the value of the constant 𝐶 in terms of the
flow parameters. Are the boundary conditions satisfied? What might this flow represent? What is
the expression for the boundary layer thickness of this flow?
 Question 1
a) "When a force is applied tangentially to the surface of a solid, the solid will
experience a finite deformation, and the tangential force per unit area - the shear stress-
will usually be proportional to the amount of deformation. In contrast, when a tangential
shear stress is applied to the surface of a fluid, the fluid will experience a continuosly
increasing deformation, and the shear stress usually will be proportional to the rate of
change of the deformation."

𝜌LV Inertia forces

b) Re = =
𝜇 Viscous forces

c)
DV Du i ∂ui ∂ui
= = + uj
Dt Dt ∂t ∂xj

∂u ∂u ∂u ∂u
+u +v +w
∂t ∂x ∂y ∂z ∂u ∂u
∂
Steady: =0 u +v
∂v ∂v ∂v ∂v ∂t DV ∂x ∂y
+u +v +w ⇒ =
∂t ∂x ∂y ∂z ∂ Dt ∂v ∂v
2D: w = 0, =0 u +v
∂w ∂w ∂w ∂w ∂z ∂x ∂y
+u +v +w
∂t ∂x ∂y ∂z

d)
D e + V2 / 2
1) 𝜌 : Variation in internal energy
Dt

2) 𝜌q␒ : Heat input

∂ ∂T ∂ ∂T ∂ ∂T
3) k + k + k : Heat exchange by conduction
∂x ∂x ∂y ∂y ∂z ∂z

4) -∇. pV : Power of pressure forces

5)
 ∂(u𝜏x x ) ∂(u𝜏y x ) ∂(u𝜏z x ) ∂(v𝜏x y ) ∂(v𝜏y y )
+ + + + +
∂x ∂y ∂z ∂x ∂y
: Power of viscous forces
∂(v𝜏z y ) ∂(w𝜏x z ) ∂(w𝜏y z ) ∂(w𝜏z z )
+ + +
∂z ∂x ∂y ∂z

Question 2

Uw

u ( y)
h
y
x
0
∂P
∂x

Step 1: Write N-S (incompressible, no gravity)

∇⋅u = 0

∂u
𝜌 + (u ⋅ ∇) u = - ∇P + 𝜇∇ 2 u ; u = ( u, v, w )
∂t

Step 2: Assumptions
∂
1) Steady state: =0
∂t

∂u
2) Fully developed: =0
∂x

∂
3) 1-D flow: v = w = 0; =0
∂z

Simplifying:
• Continuity: 0 = 0
∂P ∂2u
• x-momentum: 0 = - +𝜇 2
∂x ∂y
 ∂P
• y-momentum: 0 =
∂y

Step 3: Solve ODE

∂2u 1 ∂P ∂u 1 ∂P 1 ∂P y 2
= ⇒ = y + C1 ⇒ u = + C1 + y + C2
∂y 2 𝜇 ∂x ∂y 𝜇 ∂x 𝜇 ∂x 2

Step 4: Apply BC
BC#1: u(0) = 0 (no slip) ⇒ C 2 = 0
1 ∂P h 2
BC#2: u(h) = U W ⇒ U W = + C1 h
𝜇 ∂x 2

UW 1 ∂P h
⇒ C1 = -
h 𝜇 ∂x 2

1 ∂P y 2 y 1 ∂P hy
⇒ u= + UW -
𝜇 ∂x 2 h 𝜇 ∂x 2

2
1 ∂P h 2
y y y
a) ⇒ u(y) = U W + -
h 𝜇 ∂x 2 h h

b) u(y) ⩾ 0 ∀ y

Define y * = y / h

* 1 ∂P h 2 * 2 1 ∂P h 2
u y = y + UW - y*
𝜇 ∂x 2 𝜇 ∂x 2
⏠⏣⏣⏡
A
⏣⏣⏢ ⏠⏣⏣⏣⏣⏡⏣⏣⏣⏣ ⏢
B

2
= Ay * + By * = y * Ay * + B

Since y * ⩾ 0, u y * ⩾ 0 ∀ y * if
f y * = Ay * + B ⩾ 0
Graphic Solution:
∂P
f y * If >0
∂x
Since y * ⩾ 0 ⇒ B ⩾ 0
1 ∂P h 2
B ⇒ UW - ⩾0
𝜇 ∂x 2
∂P UW .𝜇.2
⇒ ⩽
-B / A 0 y* ∂x h2

∂P
If <0
∂x
f y* B
Since y * ⩾ 0 ⇒ - ⩽1
A
1 ∂P h 2
𝜇 ∂x 2
- UW
B ⇒ ⩽1
1 ∂P h 2
𝜇 ∂x 2

0 -B / A y*

1 ∂P h 2 1 ∂P h 2
⇒ - UW ⩽
𝜇 ∂x 2 𝜇 ∂x 2

⇒ UW ⩾ 0

∂P UW .𝜇.2
Since U W ⩾ 0, ⩽
∂x h2

c) Minimum shear stress at y = 0

∂u
| 𝜏x y | = 𝜇
∂y

∂u 1 ∂P UW 1 ∂P h
= y+ - : Equation of a line
∂y 𝜇 ∂x h 𝜇 ∂x 2
 ∂u UW 1 ∂P h
= -
∂y h 𝜇 ∂x 2
y=0

∂u
is minimum if it's = 0
∂y

UW 1 ∂P h ∂P UW .2.𝜇
- =0 ⇒ =
h 𝜇 ∂x 2 ∂x h2

Part 2:
Simplifying the energy equation using assumptions
2
∂2T ∂u ∂u 1 UW .2.𝜇 h U W . 2. 𝜇 U W
k +𝜇 = 0, = y- +
∂y 2 ∂y ∂y 𝜇 h 2 2𝜇 h 2 h

2
∂2T𝜇 4UW 2
⇒ 2 = - y
∂y k h4

2
∂T 𝜇 UW y 3
⇒ = - 4 4 + C1
∂y k h 3

2
𝜇 UW y 4
⇒T=- 4 4 + C1 y + C2
k h 12

∂T
B.C.1: = 0 ⇒ C1 = 0
∂y
y=0

2
𝜇 UW
B.C.2: [T] y=h = TU ⇒ TU = - + C2
k 3

2 2
𝜇 UW 𝜇 UW
⇒ C2 = TU + = TU +
k 3 k 3

Question 3
 Ro
r
θ

a) Ω = A (constant)

Step 1: Write N-S (incompressible. No gravity)

Step 2: Simplify using assumptions
∂
#1) Steady state: =0
∂t

∂u
#2) Fully developed: = 0 (homogeneous or periodic)
∂θ

∂
#3) 2D flow: = 0, uz = 0, ur = 0
∂z

Simplifying:
• Continuity: 0 = 0
uθ2 ∂P
• r-momentum: -𝜌 = -
r ∂r

1 ∂ ∂uθ uθ
• θ-momentum: 0 = 𝜇 r - 2
r ∂r ∂r r

• z-momentum: 0 = 0

Spet 3: Solve ODE

1 ∂ ∂uθ uθ
r - =0
r ∂r ∂r r2

2
1 ∂ uθ 1 ∂uθ uθ
⇒ r + - =0
r ∂r 2 r ∂r r2
 ∂ 2 uθ ∂ uθ ∂uθ uθ
⇒ + =0⇒ + = C1
∂r 2 ∂r r ∂r r

∂
⇒ (u θ . r) = C 1 . r ⇒ u θ . r = C 1 . r 2 + C 2
∂r

1
⇒ uθ = C1 . r + C2
r

Step 4: Apply B.C

B.C.1: u 𝜃 → for r → ∞
⇒ C1 = 0

B.C.2: u 𝜃 = A. R o at r = R o
C2
A. R o = ⇒ C2 = A. Ro2
Ro

Ro2
u 𝜃 = A.
r

2nd Part
Ro2
⇒ u 𝜃 = A.
r

uθ2 ∂P
using r-momentum: + 𝜌 = +
r ∂r
4
∂P A Ro 11
=𝜌 ⇒ P = 𝜌A 2 Ro4 - + C3
∂r r 3
r2 2

P r → ∞ = PATM ⇒ C3 = PATM

𝜌A 2 Ro4 1
⇒ P = PATM -
2 r2

b) Back to θ-momentum

∂uθ 1 ∂ ∂uθ uθ
=𝜇 r - 2
∂t r ∂r ∂r r
Equation is homogeneous

i) Transient solution
I) Write PDE
II) Apply similarity variable/ separation of variables
III) Obtain ODE and solve it
IV) Apply B.C & I.C
V) Obtain solution as function of (r, t)

ii) Harmonically varying

I) Write PDE
II) Apply separation of variables/ harmonic solution
III) Obtain ODE and solve it
IV) Apply B.C.
V) Obtain solution as function of (r, t)
VI) Take real part
Monash University – Department of Mechanical and Aerospace Engineering

MAE3401- Aerodynamics II
Mid-Semester Test 2023 (Semester 1)

General instructions
• Do not open this test before instructed!
• This test is worth 100 marks, and accounts for 15% of your final
grade.
• This test has 4 pages and 3 questions. Please check all pages.
• This is a closed book test, and your solutions must be written
with pen and paper (no digital version allowed). Calculators of
any kind are not allowed.
• Please show all your working carefully, and make sure that all
you’ve written is intelligible. Marks will be awarded for correct
working, even if the final answer is not correct.
• Remember to follow the procedure to obtain analytical solutions
for the Navier-Stokes equations as seen in the workshops.
• Remember to clearly state your assumptions and the
simplifications that they allow you to make.
• Please include drawings showing your coordinate system,
including its origin.
Monash University – Department of Mechanical and Aerospace Engineering
Monash University – Department of Mechanical and Aerospace Engineering

Question 1 (20 marks): Answer the following questions about viscous fluids.

a) In solids, shear stress is proportional to shear strain. The defining characteristic of a fluid is
that shear stress is proportional to what (your answer should be less than five words)?
b) Reynolds number is a non-dimensional parameter representing the ratio of which two forces
(your answer should be less than five words)?
c) The first term in the incompressible non-dimensional momentum equation, in vector form, is
!𝑽
, where 𝑽 is the velocity vector with three cartesian components, 𝑽 = ( 𝑢(𝑥, 𝑦, 𝑧),
!#
𝑣(𝑥, 𝑦, 𝑧), 𝑤(𝑥, 𝑦, 𝑧) ). Write out then expand and simplify each component of this term,
assuming steady 2D flow.
d) Given the compressible energy equation:

Identify and group together the terms resulting from the same physical mechanisms (hint: there are
five mechanisms), and describe in words their physical interpretations (each description should be less
than five words).

Question 2 (40 marks): Channel-Couette flow is a mix of two canonical wall-bounded flows that is
useful for studies on the effect of pressure gradients in the development of the laminar flow solution.
Consider an incompressible steady flow between two walls. The top wall is moving with positive
"#
velocity 𝑈! while the bottom one is kept at rest. A constant streamwise pressure gradient "$ is applied
to this flow. Neglecting gravitational effects and assuming that the centre of your coordinate system
is at the bottom wall, obtain:

a) The final shape of the velocity profile.

"#
b) The values of "$
for which the flow has no reverse-flow (𝑢(𝑦) < 0) region.
c) The pressure gradient that leads to the minimum shear-stress (and thus to the minimum drag)
at the bottom wall.
d) The temperature distribution in the flow for the minimal shear case if the bottom wall is
adiabatic, and the upper wall has constant temperature 𝑇% . Here, you can consider the
incompressible limit of the energy equation, as provided below.
𝐷𝑇
𝜌𝑐& = 𝑘∇' 𝑇 + 𝜙
𝐷𝑡
Monash University – Department of Mechanical and Aerospace Engineering

𝜕𝑢 ' 𝜕𝑣 ' 𝜕𝑤 ' 𝜕𝑣 𝜕𝑢 ' 𝜕𝑤 𝜕𝑣 ' 𝜕𝑢 𝜕𝑤 '

𝜙 = 𝜇 :2 < > + 2 < > + 2 < > + < + > + < + > +< + > ?
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑧 𝜕𝑧 𝜕𝑥
𝜕𝑢 𝜕𝑣 𝜕𝑤 '
+𝜆< + + >
𝜕𝑥 𝜕𝑦 𝜕𝑧

Question 3 (40 marks): An infinitely long cylinder of radius 𝑅( immersed in an incompressible flow
rotates around its axis with angular velocity given by Ω.

a) Neglecting gravity, obtain the steady analytical solution for the velocity and pressure
distributions of the flow around this cylinder if the angular velocity is constant (Ω = 𝐴). Hint:
you may have to use the product rule a couple of times.
b) Propose a detailed step-by-step approach (including the method to solve the PDE) to obtain
the unsteady analytical solution for the following cases:
i. Transient solution for the start-up regime (cylinder suddenly put in motion)
ii. Unsteady solution for the harmonically-varying angular velocity (Ω = A exp(𝑖𝜔𝑡))
Note: You don’t need to solve the actual PDE!
 Question 1

Assuming 2D, no heat input, steady

∂ ∂ ∂ u2 + v2 ∂ ∂T ∂ ∂T ∂ ∂T
𝜌 u +v +w e+ = k + k + k -
∂x ∂y ∂z 2 ∂x ∂x ∂y ∂y ∂z ∂z

∂ ∂ ∂ ∂u ∂v ∂u ∂ ∂v ∂u
(pu) - (pv) + u 𝜆 + 𝜆 + 2𝜇 + u𝜇 + + (1)
∂x ∂y ∂x ∂x ∂y ∂x ∂y ∂x ∂y

∂ ∂v ∂u ∂ ∂u ∂v ∂v
v𝜇 + + v 𝜆 + 𝜆 + 2𝜇
∂x ∂x ∂y ∂y ∂x ∂y ∂y

In order to perform an order of magnitude analysis, the equation must be normalized.

Introducing:

x y u v 𝜇 𝜆
x' = y' = u' = v' = 𝜇' = 𝜆' =
c c UE UE 𝜇E 𝜇E

T e k p 𝜌
T' = e' = k' = p' = 𝜌' =
TE cv TE kE pE 𝜌E

Substituing in Eq. 1:

∂e' ∂e' -UE2 ∂ ∂

𝜌'u' + 𝜌'v' = 𝜌'u' u' 2 + v' 2 + 𝜌'v' u' 2 + v' 2 +
∂x' ∂y' 2cv TE ∂x' ∂y'

kE ∂ ∂T' ∂ ∂T' pE ∂ ∂
k' + k' - (u'p') + (v'p') +
c. 𝜌 E . U E . c v ∂x' ∂x' ∂y' ∂y' 𝜌E . cv . TE ∂x' ∂y'

𝜇E . UE ∂ ∂u' ∂v' ∂u' ∂ ∂v' ∂u'

u'. 𝜆' + 𝜆' + 2𝜇' + u'. 𝜇' + +
c. 𝜌 E . c v . T E ∂x' ∂x' ∂y' ∂x' ∂y' ∂x' ∂y'

∂ ∂v' ∂u' ∂ ∂u' ∂v' ∂v'

v'. 𝜇' + + v'. 𝜆' + 𝜆' + 2𝜇' }
∂x' ∂x' ∂y' ∂y' ∂x' ∂y' ∂y'
 Checking the coefficients from RHS:

UE2 2
𝛾 ( 𝛾 - 1) U E UE2
= . = 𝛾(𝛾 - 1) 2 = 𝛾(𝛾 - 1). ME2
cv . TE 𝛾 R TE aE

kE k E . 𝛾. 𝜇E 𝜇E kE
= =𝛾 .
c. 𝜌 E . U E . c v c. 𝜌 E . V E . 𝜇 E . c v . 𝛾 𝜌E . VE . c cp . 𝜇E
⏠⏣⏣
⏡⏣⏣
⏢
cp ⏠⏣⏣⏣ ⏡⏣⏣⏣
Re -1
⏢ ⏠⏣⏣ ⏡⏣⏣
Pr -1
⏢

pE ( 𝛾 - 1) p E pE
= . = ( 𝛾 - 1) . = 𝛾-1
𝜌E . cv . TE R 𝜌E . TE pE

𝜇E . UE 𝜇E UE2
1 ( 𝛾 - 1) U E 𝛾
2
= = = . . .
c. 𝜌 E . c v . T E 𝜌E . c. UE cv . TE Re R TE 𝛾

ME2
= 𝛾 ( 𝛾 - 1) .
Re

Consider incompressible and ignore normal 𝜏

∂e' ∂e' ∂u' ∂u' ∂v' ∂v'

𝜌' 𝜇' + v' = - 𝛾(𝛾 - 1). ME2 . 𝜌' u' u' + v' + v' u' + v' +
∂x' ∂y' ∂x' ∂y' ∂x' ∂y'

𝛾 ∂ ∂T' ∂ ∂T' ∂ ∂
k'. + k'. - ( 𝛾 - 1) (u'. p') + (v'. p') + (2)
Re. Pr ∂x' ∂x' ∂y' ∂y' ∂x' ∂y'

ME2 ∂ ∂v' ∂u' ∂ ∂v' ∂u'

𝛾 ( 𝛾 - 1) 𝜇'. u' + + 𝜇'. v' +
Re ∂y' ∂x' ∂y' ∂x' ∂x' ∂y'

1 pE p' p'
Work with LHS: e' = h' - . . = h' - (𝛾 - 1).
cv . TE 𝜌E 𝜌' 𝜌'

∂h' ∂h' (𝛾 - 1) ∂p' ∂p'

LHS = 𝜌' u'. + v'. - u'. + v'.
∂x' ∂y' 𝜌' ∂x' ∂y'
 ∂h' ∂h' ∂p' ∂p'
= 𝜌'. u'. + 𝜌'. v'. - (𝛾 - 1). u'. + v'.
∂x' ∂y' ∂x' ∂y'

Substituing back to (2) (& continuity)

∂h' ∂h'
𝜌'. u'. + 𝜌'. v'. =
∂x' ∂y'

-1 ∂p' 1 ∂ ∂v' ∂u'

𝛾(𝛾 - 1). ME2 . 𝜌' u' + 𝜇' + 𝜇' +
𝜌'𝛾ME2 ∂x' 𝜌'Re ∂y' ∂x' ∂y'
-1 ∂p' 1 ∂ ∂v' ∂u'
v' + 𝜇' + 𝜇' }+
𝜌'𝛾ME2 ∂y' 𝜌'Re ∂x' ∂x' ∂y'

𝛾 ∂ ∂T' ∂ ∂T' ∂ ∂
k'. + k'. - ( 𝛾 - 1) (u'. p') + (v'. p') +
Re. Pr ∂x' ∂x' ∂y' ∂y' ∂x' ∂y'

ME2 ∂ ∂v' ∂u' ∂ ∂v' ∂u' ∂p' ∂p'

𝛾 ( 𝛾 - 1) 𝜇'. u' + + 𝜇'. v' + + (𝛾 - 1) u' + v'
Re ∂y' ∂x' ∂y' ∂x' ∂x' ∂y' ∂x' ∂y'

(May be further simplified)

Order of magnitude analysis:

As in previous cases:

[u', x', 𝜌', ME , k', 𝜇'] = O (1)

1 2 ∂P'
[y', v'] = O (𝛿); Re = O 𝛿 ; ∂y' = 0

Assuming [Pr, h, T] = O (1) (Dimensional analysis), the equation may be simplified to:

∂h' ∂h' 𝛾 ∂ ∂T' ME2 ∂u' ∂u' ∂p'

𝜌'. u'. + 𝜌'. v'. = k'. + 𝛾 ( 𝛾 - 1) . 𝜇'. u' + (𝛾 - 1)u'
∂x' ∂y' Re. Pr ∂y' ∂y' Re ∂y' ∂y' ∂x'

Question 2

This question is report-like. Students should get these solutions from Week 4 slides,
and plot them as fuction of t (instead of x, which we use in B.L. analysis).

Both numerical and analytical integrations are fine. The quantities to be computed
are 𝛿 9 9 (t),
h h
u ( y, t ) u ( y, t ) u ( y, t )
𝛿 (t) = ∫
*
1- dy and 𝜃(t) = ∫ 1- dy.
0 u ( h, t ) 0 u ( h, t ) u ( h, t )

They should be wise in choosing the "free-stream" position for the pipe case, and
also in defining the integration limits (wall to centerline).

Analysis is kind of direct. For low t, it is expected a B.L.-like behaviour, with 𝛿

growing asumptotically. For large t, they shouldn't vary anymore, leading to fully
developed flow.

Question 3
Let us proceed to derive the momentum integral by integration the boundary layer
momentun equation from y = 0 to y = h. The position y = h is outside the boundary layer
2
where u ≈ u e and where all derivatives ∂u ∂y, ∂ u ∂y 2 ,... are zero. The equation is

h ∂u
h
∂u ∂u h
𝜇 ∂2u
∫0 u
∂x
+v
∂y
dy = ∫ ue
0 ∂x
e
dy + ∫
0 𝜌 ∂y 2
dy

Rearranging and integration the shear stress term produces

h
h ∂u ∂ue ∂u 𝜇 ∂u 𝜏0
∫0 u - ue
∂x ∂x
+v
∂y
dy =
𝜌 ∂y
= -
𝜌
(1)
0

The wall shear stress is denoted as 𝜏 0 . Now, for a solid wall where
v = (y = 0) = 0, we may express v at any position y by integrating the continuity
equation:

y y
∂v ∂u
v=∫ dy' = - ∫ dy'
0 ∂y 0 ∂x

Consider the third term in Eq. 1, and insert the expression above for v. It becomes
 y
h ∂u h ∂u ∂u
∫0 v dy = ∫ -∫
∂y 0 0 ∂x
dy'
∂y
dy

This integral is of the form ∫w dz = wz - ∫z dw, where

y
∂u ∂u
w=∫ dy' ⇒ dw = - dy
0 ∂x ∂x
and

∂u
dz = dy ⇒ z = u
∂y

Hence, integration by parts gives

h h h
∂u ∂u ∂u
∫0 v dy = - ue ∫ dy + ∫ u dy
∂y 0 ∂x 0 ∂x

With this, Eq. 1 becomes

h
∂u ∂ue ∂u ∂u 𝜏o
∫0 u
∂x
- ue
∂x
- ue
∂x
+u
∂x
dy = -
𝜌

Next, we add and subtract u

∂ue
∂x and rearrange the terms into the form

h ∂ue ∂u ∂u h ∂ue 𝜏o
-∫ u - + ( u e - u) dy - ∫ (ue - u) dy = -
0 ∂x ∂x ∂x 0 ∂x 𝜌

or equivalently,

d h
u u due h
u 𝜏o
dx
ue2 ∫0 u 1-
ue
dy + ue ∫
dx 0
1-
ue
dy =
𝜌
e

As h → ∞ the integrals are by definition the displacement and momentum

thicknesses; thus,
 d due 𝜏o
ue2 Θ + ue 𝛿* = (2)
dx dx 𝜌

The profile must meet several boundary conditions. At the wall, the no-slip condition,
the momentum equation, and the derivative of the momentum equation with respect to y
require that

due ∂2u ∂3u

u = 0, -ue =𝜈 , =0 (3)
dx ∂y 2 0
∂y 3 0

A smooth approach of u to u e is enforced at the finite position y = 𝛿. This gives

∂u ∂2u
u = ue , = 0, = 0, ... (4)
∂y ∂y 2
𝛿 𝛿

For a specific boundary layer where u e (x) is given, conditions 3 and 4 together with
the momentum integral (Eq. 2) allow one to find the coefficients a, b, c, and d in the
assumed profile as funcions of x. As profiles with more coefficients are introduced, more
smoothness is required in the approach of u to u e at y = 𝛿; that is, the higher-order
n
derivatives ∂ u ∂y'' are required to be zero.
𝛿
As a relatively simple example, let us solve the Blasius problem where u e = u 0 , a
constant. Applying the first two boundary conditions of Eq. 3 and the first two of Eq. 4
shows that the profile is

3 1
u* = 𝜂 - 𝜂3 (5)
2 2

Inserting Eq. 5 into the definition of the displacement thickness yields

1 3
𝛿 * = 𝛿∫ 1 - u * d𝜂 = 𝛿 (6)
0 8

The same process applied to the definition of the momentum thickness produces

1 117
Θ = 𝛿∫ u * 1 - u * d𝜂 = 𝛿 (7)
0 840
 To find the wall shear stress, we can use

𝜏o ∂u 3 uo
=𝜈 =𝜈 (8)
𝜌 ∂y 2 𝛿
0

Now we are in a position to substitute for all terms in the von Kármán momentum
integral. Equation 2 for u e = u 0 becomes

dΘ 𝜏o
ue2 =
dx 𝜌

117 ∂𝛿 3 𝜈uo
ue2 =
840 ∂x 2 𝛿

Integration gives:

840 𝜈x 𝜈x
𝛿= = 4.64 (9)
39 u0 u0
 MAE3401- Aerodynamics II
Problem set 5: Introduction to boundary layers
In this problem set you will have the opportunity to put in practice some of the concepts you saw in
both the lecture videos and the workshop. This one will be about some initial concepts regarding
boundary layers. The due date for this one is Friday of Week 7

General instructions: Please show all your workings carefully, and make sure that all you’ve written
is intelligible. You will still get marks if you show your workings (and drawings, if necessary), even if
you don’t reach the final answer.

Question 1: Starting from the compressible energy equation:

Where 𝑒 = ℎ − 𝑝/𝜌. Formulate a set of assumptions valid inside a thin boundary layer developing
close to a large flat plate and perform an order of magnitude analysis to obtain the boundary-layer
energy equation.

Question 2: Considering (i) the starting pipe solution and (ii) the start-up flow around a flat plate,
compute (and plot) the equivalent 𝛿99 , 𝛿 ∗ and 𝜃 lengths as function of time. For what times the
behaviour of these flow is similar to a boundary layer? What are the main similarities and
differences between them and a boundary layer forming around a flat plate? Note: you may use the
final solutions from the Week 4 lectures and integrate them either numerically or using
Wolfram/Mathematica (or any software), if you like.

Question 3: Starting from the incompressible boundary layer equations in a flat plate for zero
pressure gradient:

a) Manipulate/integrate these expressions to find a relationship between displacement and

momentum thicknesses and the shear stress at the wall. Note: you may need to use
ℎ ℎ
integration by parts: ∫0 𝑓 𝑔′ 𝑑𝑦 = 𝑓𝑔|ℎ0 − ∫0 𝑔𝑓 ′ 𝑑𝑦
b) Since you’re not able to obtain an analytical solution for the BL equations at this point, you
are allowed to guess a profile that has the same essential features. Consider that you have
guessed a velocity profile (defined inside the boundary layer) that is self-similar or, in other
words, that its overall shape doesn’t change as the flow develops. Your first guess was a
 cubic polynomial 𝑢 = 𝑎 + 𝑏𝜉 + 𝑐𝜉 2 + 𝑑𝜉 3 , with 𝜉 = 𝑦/𝛿(𝑥). Propose a set of 4 boundary
conditions for this profile and obtain the expression for 𝑢. Hint: you may need two BCs at
each border of the domain.
c) Obtain the displacement and momentum thicknesses and the shear stress associated with
this profile and use the equation obtained in (a) to estimate the boundary layer thickness 𝛿
as function of 𝑥 for this profile.
Question 1

y
𝜈 = kinematic viscosity
constant pressure 𝜌 = density y, v

x, u

Uo cos(𝜔t)

= Re Uo e i𝜔t , e i𝜔t = cos(𝜔t) + i sin(𝜔t)

We will assume that the real part of a solution is the physical meaningful solution

Governing equation:
∂u ∂2u
=𝜈 2 (1)
∂t ∂y

B.C.:
y = 0 : u = Uo e i𝜔t (1a)
y→∞ : u = 0 (1b)

a) The velocity solution is assumed of the form

u(y, t) = f(y)e i𝜔t (2)

Since we are interested in the long term response

Substitute (2) into (1):

i𝜔f = 𝜈f"
i𝜔
i.e. f" - f=0 (3)
𝜈

Assume
f(y) = e 𝜎y (4)
Substitute (4) in (3):

i𝜔
𝜎2 - =0
𝜈

i𝜔
𝜎2 =
𝜈

𝜔
𝜎=± i
𝜈

𝜔 (5)
∴ 𝜎=± (1 ± i)
2𝜈

𝜔 𝜔
f(y) = c1 exp (1 + i) y + c2 exp (-1 - i) y
2𝜈 2𝜈

𝜔 𝜔
∴ u(y, t) = e i𝜔t c1 exp (1 + i) y + c2 exp (-1 - i) y
2𝜈 2𝜈

(1b) ⇒ u y → ∞ = 0 ⇒ c 1 = 0 as u must remain finite → 0

𝜔 𝜔
i 𝜔t- y
- y 2𝜈 (6)
∴ u ( y, t ) = c 2 e 2𝜈 e

(1a) ⇒ U o e i𝜔t = c 2 e -0 e i𝜔t

∴ c2 = Uo (7)

𝜔 𝜔
i 𝜔t- y
- y 2𝜈 (8)
∴ u ( y, t ) = U o e 2𝜈 e

The physical meaningful part of (8) is its real part

𝜔
- y 𝜔
∴ u ( y, t ) = U o e 2𝜈 cos 𝜔t - y
2𝜈
b)
2π
Note: 𝜔 = 2πf = T is the period
T

𝜔 2π/𝜔
⏨⏨⏨
u ( y, t ) = ∫ u(y, t)dt = 0
2π 0
c)

𝜔 2π/𝜔 2
⏨⏨⏨
u 2 ( y, t ) = ∫ u (y, t)dt
2π 0

𝜔 2
𝜔 2π/𝜔 2 -2y 𝜔
= ∫ Uo e
2π 0
2𝜈 cos 𝜔t - y
2𝜈
dt

𝜔
-2y
Uo2 e 2𝜈
∴ ⏨⏨⏨
u 2 ( y, t ) =
2

u 2⏨
⏨
Uo2

0.5

0.2 𝜔 1.0
y
2𝜈

d)

du
𝜏w = 𝜇 |y=0
dy

𝜔
∴ 𝜏w = - 𝜇Uo [cos(𝜔t) - sin(𝜔t)]
2𝜈

𝜔 𝜔 2π/𝜔
⏨⏨⏨
𝜏w (+) = - 𝜇Uo ∫ [cos(𝜔t) - sin(𝜔t)]dt
2𝜈 2π 0
 ∴ 𝜏 w ( +) = 0

e)

𝜔 𝜔 2π/𝜔
⏨⏨⏨
𝜏w2 (+) = (𝜇Uo ) 2 ∫ [cos(𝜔t) - sin(𝜔t)] 2 dt
2𝜈 2π 0

(𝜇Uo ) 2 𝜔 𝜌Uo2 𝜇𝜔
∴ ⏨⏨⏨
𝜏w2 (+) = =
2𝜈 2

Question 2
2D laminar Couette flow constant pressure

2D laminar Couette flow

𝜌 = density
constant pressure R2 𝜇 = dynamic visc
u
v 𝜇
𝜈=
h = R2 - R1 ≪ R1 𝜌
Ω
R1
r
h
∴ ≪1
R1

and R 2 = R 1 + h

a) u = u(r, t) only v = 0 as for the steady case

To account for the transient state during "start-up" from quiescent flow state
(u = v = 0) to the steady flow case (u = u(r), v = 0), the local acceleration has to be
inclusided and hence, the net viscous force activity on a cylindrical infinitesimal element
results in the acceleration of the element.

∂F𝜃 ∂u
= 2𝜋r𝜌
∂r ∂t

∂u
since F 𝜃 = 2𝜋r 𝜏r𝜃 = 2𝜋r 𝜇
∂r
 ∂u 𝜈 ∂ ∂u
∴ = r (1)
∂t r ∂r ∂r

I.C.: t⩽0 :
R1 ⩽ r ⩽ R2 : u = 0 (2)

B.C.: t>0 :
r = R1 : u = 0 (3)
r = R2 : u = R2 Ω (4)

∂u
b) Steady state flow ⇒ → 0 as t → ∞
∂t

d du
∴ r =0
dr dr

With solution:

ln(r / R1 )
as t → ∞ : u r, t → ∞ = u ∞ (r) = R 2 Ω
ln(R2 / R1 )

h r h
Since the gap ≪ 1 and 1 ⩽ ⩽ 1+
R1 R1 R1

r
Expanding ln(r / R 1 ) around ln(1 + ϵ) ϵ= -1
R1
ϵ2
i.e. ln(1 + ϵ) = ϵ - + 𝜃 ϵ3 ϵ = 0, h R 1
2

To first order for small gap h R ≪ 1

1

R1 + h h h h2 h3
and ln = ln 1 + = - +𝜃
R1 R1 R1 2R12 R13

r
R1 - 1 (r - R 1 )
u ∞ (r) = R 2 Ω = R2 Ω
(R 2 - R 1 ) (R 2 - R 1 )
R1
 For this approximation which applies when h = R 2 - R 1 ≪ R 1

Uo y
u ∞ (r) = y = r - R1 y ∈ [0, R2 - R1 ]
h

U0 = R2 Ω

h = R2 - R1

The flow is essentially fully developed planar Couette flow.

c) Under the assumption of h ≪ R 1 and the above steady flow answer, the transient
solution follows from the transient start-up Couette solution (see lecture notes) governed
by
∂u ∂2u
=𝜈 2
∂t ∂y

I.C.: t⩽0 :
0⩽y⩽h : u=0

B.C.: t>0 :
y=0 : u=0
y = h : u = Uo

Solution as per the lecture:

∞
u(𝜂, 𝜏) 2(-1) n+1 -(n𝜋) 2 𝜏
= 𝜂+ ∑ e sin(n𝜋𝜂)
Uo n=1
n𝜋

y r - R1
𝜂= = with h = R 2 - R 1
h h
t𝜈
𝜏=
h2

Uo - R2 Ω

Note: This solution only applies where h = R 2 - R 1 ≪ R 1

d) The characteristic time scale is associated with the exponential decay of the
transient

-(n𝜋) 2
exp t
h2
𝜈
If we take the characteristic time as the time where the velocity is within 1% of the
steady state:

-(n𝜋) 2
exp t = 0.01
2
h
𝜈

(n𝜋) 2
∴ t = 4.6
2
h
𝜈

0.47 h 2
∴ characteristic t = n = 1(longest time)
𝜈

An alternative characteristic is the time when the transient velocity has decayed to
1 :
e
(n𝜋) 2
⇒ t=1
h2
𝜈

0.1 h 2
∴ characteristic t = n = 1(longest time)
𝜈

Question 3 (Solution outline)

The first difficulty is that the basic differential equation contains the inhomogeneity
(-dp / dx) which must be removed before separation of variables can work. To do this,
define velocity V such that
2
* 2, dp r o
V = u - umax 1 - r where r = r / r o and u max = -
dx 4𝜇

We thus work with the difference between u and the final Poiseuille-flow profile at
t = ∞. This gives us the following homogeneous PDE for V :
∂V ∂2V 1 ∂V
= + , where r * = r / ro and t * = 𝜈t / ro2 (1)
* *2 * *
∂t ∂r r ∂r

subject to: V 1, t * = 0 (no-slip) (2)

 2
V r * , 0 = umax 1 - r * (3)

The solutions to (1) subject to condition (2) are all of the form
V = J 0 λn r * exp -λn2 t * , where λn are the roots of the Bessel function J 0 (see values
in Table 1). The general solution of the linear PDE (1) is thus a Fourier-Bessel series of
such terms:
∞
V = ∑ Cn J 0 λn r * exp -λn2 t * (4)
n=1

To evaluate C n , we use condition (3) and the Bessel function orthogonality condition:
1 1 2
∫0 x J0 (λi x) J0 (λj x) dx = J (λ i )
2 1
if i=j

and = 0 if i≠j

We thus multiply Eq.(3) by r * J 0 λ j r * and integrate the whole equation from 0 to

1. The orthogonality condition enable us to pick off the constant one by one, wth the result

8umax
Cn =
λn3 J1 (λn) 0.02 t * = .00

With V thus known, we return to the original .004

u / umax
variable u and the desired solution:

u
∞ 8 J0 λn r * .003

= 1-r *2 -∑ exp -λn2 t * 0.01

umax 3 (λ n )
n=1 λ n J 1 .002

The profiles for very early t * are shown at right .001

ans are seen to consist of a linearly accelerating

0
core with thin growing boundary layers. 1.0 r / r0

Solution for the working velocity profile is given as

 ∞ 8 J0 λn r *
u r
= 1-r - ∑ 3
*2 exp -λn2 t * r* =
umax n=1 λ n J 1 (λ n ) r0
⏠⏣⏣
⏡I ⏣⏣
⏢
⏠⏣⏣⏣⏣⏣⏣
⏡⏣⏣⏣⏣⏣⏣
II
⏢

λn = roots of Bessel function J0 (Table 1)

u 2
Notes: (I) = 1 - r* = steady state solution as t → ∞
umax

(II) the contribuition from the unsteady part is longest felt by the n = 1
contribution to the ∑.
Because all other contribuitions decay to 0 much quicker.

Table 1: n = 1 : λ n = 2.4048 J 1 (λn ) = 0.5191

r0
Volume flow rate = V␒ = u max ∫o 2πru(r) dr

to first order
2 *

r0 r0 8 J0 λn r * e -2.4048 t

= umax ∫ 2πr 1 - r * dr - umax ∫ 2πr dr

o o (2.4048) 3 0.5191

r = r o r * ⇒ dr = r o dr *

1 1
∴ V␒ = 2πr 2 umax ∫0 r * 1 - r * dr * - ∫ (1.11)r * J0 2.4048 r * e -5.783 t dr *
2
0

1 π r02 umax
V␒ s = V␒ t → ∞ = 2πr umax ∫ r 1 - r
* 2 * *2 *
dr =
0 2

V␒ t * 1
= 1 - 4.44∫ r * J0 2.4048r * dr * e -5.783 t
*
∴
V␒ s 0

= 1 - 0.9568 e -5.783 t
 V␒ t *
= 0.99 → t * = 0.79
V␒ s

0.79 r02
∴ the flow rate reaches 99% of the steady state flow rate at t = .
𝜈
 MAE3401- Aerodynamics II
Problem set 4: Analytical solutions for unsteady flows
In this problem set you will have the opportunity to put in practice some of the concepts you saw in
both the lecture videos and the workshop. This will be about analytical solutions of the Navier-
Stokes equations for unsteady flow.

General instructions: Please show all your workings carefully, and make sure that all you’ve written
is intelligible. You will still get marks if you show your workings (and drawings), even if you don’t
reach the final answer. Important: drawings will be particularly important in this problem set, so
make sure that all of them are clear.

Question 1: An infinitely long and wide flat plate is oscillating with a velocity amplitude 𝑈0 and
angular frequency 𝜔 in its plane in the 𝑥 direction. An order-of-magnitude analysis indicates that the
resulting long-time behaviour of the flow is dominated by a balance between the local acceleration
and viscous stresses.

a) Derive a closed form expression for the air velocity distribution normal to the plate surface.
b) Determine and plot the mean velocity distributions.
c) Determine and plot the mean squared velocity distributions.
d) Determine the mean wall shear stress.
e) Determine the mean squared wall shear stress.

Question 2: In order to calculate the necessary torque during the start-up of a gas aero turbine, you
are asked to calculate the viscous torque exerted by the journal bearings used to support the gas
aero turbine shaft. A reasonable first order model for the journal bearings is laminar 2D Taylor-
Couette flow with both cylinders initially stationary. The start-up process is modelled with the inner
cylinder fixed to the aero turbine shaft at 𝑡 = 0 rotating with a constant angular velocity of Ω.

a) Simplify the cylindrical NS equations to obtain the simplified PDE for this phenomenon and
state the relevant boundary and initial conditions.
b) Compute the steady state solution and simplify it by using the assumption that the fluid gap
is much smaller than the radius of the inner cylinder. You may need to use the
𝜖2
approximation ln(1 + 𝜖) = 𝜖 − 2
+ 𝑂(𝜖 3 ); if 𝜖 ≪ 1 the 𝑂(𝜖 3 ) terms may be neglected.
c) Compute the transient solution for the same case.
d) Is there a characteristic time scale associate with the time it takes the flow to reach steady-
state (or, say, 1% deviation from steady-state)? You may truncate your series to order 1 in
this calculation.

Question 3: We are interested in investigating the way that the velocity profile in pipe flow and
hence, volume flow rate develops during the pipe flow start-up process during early times. Starting
𝜈𝑡
from the solution we derived in the last workshop, plot the resulting velocity profile up to =
𝑅02
0.005: how does the centreline velocity varies in time? What is the cause of variation of the velocity
closer to the wall? How long does it take for the volume flow rate to reach 99% of the steady state
flow rate? You may truncate your summation to order one for the last calculation.
 Question 1

Givens:
• Slug pipe flow in a pipe of diameter D ⇒ u(r) = U = constant
• Constant heat flux along the pipe
– Define constant heat flux ≝ q␒ w (1)
• Calculate the temperature distribution and convective heat transfer coefficient

Diagram:

Assumptions:
u(r) = U = constant (1a)

• Fully developed thermal field with constant property fluid

• Viscous dissipation is negligible
• Newton's Law of Cooling

q␒ w = h(Tw - Tm ) (2)
with h = local convective heat transfer coefficient
Tw = wall temperature
Tm = convenient reference temperature of the fluid, mean temperature over a
cross-section of pipe

Equations:
Energy:

dT
𝜌cp = h∇ 2 T
dt
 ∂T h ∂ 2 T 1 ∂ ∂T
U = + r (3)
∂x 𝜌cp ∂x 2 r ∂r ∂r

h
Define 𝛼 =
𝜌cp

∂T ∂ 2 T 1 ∂ ∂T
∴U =𝛼 + r (4)
∂x ∂x 2 r ∂r ∂r

Thermally fully developed flow:

∂ T w ( x) - T ( r , x)
=0 (5)
∂x Tw (x) - Tm (x)

Note:
since
q␒ w = constant ⇒ Tw (x) (6)

Analysis:

∂ ∂ T w ( x) - T ( r , x)
( 5) ⇒ =0
∂r ∂x Tw (x) - Tm (x)

∂ ∂ T w ( x) - T ( r , x)
i. e . =0
∂x ∂r Tw (x) - Tm (x)

∂ - ∂T ∂r
i. e . =0
∂x Tw (x) - Tm (x)
at r = D 2 the pipe inside surface

- ∂T ∂r
∂ r=D/2 (7)
∴ =0
∂x Tw (x) - Tm (x)
 - ∂T ∂r
r=D/2 (8)
⇒ ≠ f ( x)
T w ( x) - T m ( x)

D
If y is measured from the pipe surface, then y = - r and Fourier's Law applied at
2
the pipe internal surface

∂T ∂T
q␒ w = - k =k (9)
∂y y=0
∂r r=D/2

But

∂T
( 2) ⇒ k = h( T w - T m ) (10)
∂r r=D/2

h T w ( x) - T m ( x)
(8) & (10) ⇒ = ≠ f ( x)
k ∂T (11)
∂r
r=D/2

In thermally fully developed flow of fluid with constant properties, the local convective
heat transfer coefficient is independent of x.
Since q␒ w = constant and h = constant from (11)

dTw dTm
( 2) ⇒ = (12)
dx dx

∂ Tw - T 1 ∂T dTw (T - Tw ) dTw dTm

( 5) ⇒ = - - - =0
∂x Tw - Tm (Tw - Tm ) ∂x dx (Tw - Tm ) 2 dx dx

∂T dTw Tw - T dTw Tw - T dTm

i. e . = - + (13)
∂x dx Tw - Tm dx Tw - Tm dx

Using (12):

∂T dTw dTm
= = = f(x) only (at most) (14)
∂x dx dx

Consider an element of the axial pipe flow of length dx

 D/2
Q␒ c (x) = ∫ (2𝜋rdr). u. 𝜌cp T(x, r) = enthalphy into C.V.
0

D/2
using (1a) i.e. = 2𝜋U𝜌c p ∫0 rT(x, r)dr (15)

dQ␒ c
␒ (x + dx) = Q␒ c (x) +
Now to first order: Q dx (16)
c
dx

The heat transfer from the wall of the pipe to the fluid over dx

⇒ Q␒ w = 𝜋Dq␒ w dx (17)

Conservation of energy for C.V.

Q␒ c (x) + Q␒ w - Q␒ c (x + dx) = 0

Using (16):

dQ␒ c
dx = Q␒ w
dx

Using (17):

dQ␒ c
using (17): dx = 𝜋Dq␒ w dx
dx
Using (15):

D/2
d
2𝜋𝜌cp U ∫
dx 0
rT(x, r)dr = 𝜋Dq␒ w

d D/2 Dq␒ w
∴ ∫
dx 0
rT(x, r)dr =
2𝜌cp U
(18)

D/2
Enthalpy flow at any x = 2𝜋𝜌c p U ∫0 rT(x, r)dr

= mass flow rate x cp x Tm

= m␒ cp Tm (19)

∴ m␒ = 𝜌 x volume flow rate = 𝜌 x area x mean velocity

But mean velocity = U in slug flow

𝜋D 2 (20)
∴ m␒ = 𝜌 U
4

(19) & (20):

D/2 𝜋D 2
2𝜋𝜌cp U∫ rT(x, r)dr = 𝜌 Ucp Tm (x)
0 4
D/2
8
∴ Tm (x) = 2 ∫ rT(x, r)dr (21)
D 0

(18) & (21):

D 2 dTm Dq␒ w
=
8 dx 2𝜌cp U

dTm 4q␒ w
i.e. = = constant (22)
dx 𝜌cp DU
Integrating with respect to x with T m1 = T m (x 1 )

4q␒ w
Tm (x) - Tm1 = (x - x 1 ) (23)
𝜌cp DU

From (14) and (22):

∂T 4q␒ w
= = constant (24)
∂x 𝜌cp DU

∂2T
⇒ =0 (25)
∂x 2

We still need to find T(x, r) by solving (4):

dTm 1 ∂ ∂T
U =𝛼 r
dx r ∂r ∂r

∂ ∂T U dTm
r = r (26)
∂r ∂r 𝛼 dx

∫dr
∂T U dTm r 2
r = + c 1 ( x)
∂r 𝛼 dx 2

∂T U dTm c 1 ( x)
= r+
∂r 2𝛼 dx r

U dTm 2 (27)
T= r + c1 (x)ln(r) + c2 (x)
4𝛼 dx

c1 (x) = 0 for T to remain finite

U dTm 2
T= r + c 2 ( x)
4𝛼 dx
 D
at r = : T = Tw
2

U dTm D 2
⇒ Tw = + c 2 ( x)
4𝛼 dx 4

UD 2 dTm
∴ c2 = Tw -
16𝛼 dx

U dTm U dTm
∴T= r2 - D 2 + Tw
4𝛼 dx 16𝛼 dx
2
U dTm 2 D
i.e. T = T w + r -
4𝛼 dx 2

2
UD 2 dTm 2r
∴ T = Tw + -1 (28)
16𝛼 dx D

Using (21):

8 D/2
Tm = ∫
2 0
rTdr
D

2
8 D/2 UD 2 dTm 2r
i.e. = ∫0 r Tw +
16𝛼 dx
-1 dr
D2 D

8 D/2 dTm UD 2 4r 3 UD 2
= ∫ Tw r + - r dr
D 2 0 dx 16𝛼 D 2 16𝛼

D/2
8 Tw dTm U r 4 UD 2
= r2 + - r2
D 2 2 dx 4𝛼 4 32𝛼 0

8 Tw D 2 dTm U D 4 UD 2 D 2
= + -
D2 8 dx 16𝛼 16 32𝛼 4

UD 4 UD 2 dTm
= Tw + - (29)
32𝛼 16𝛼 dx
 UD 2 dTm
∴ Tm = Tw -
16𝛼 dx

UD 2 dTm UD 2 4q␒ w q␒ w D
∴ Tw - Tm = = = (30)
16𝛼 dx 4 16𝛼 𝜌c p DU 4k

From q␒ w = h(T w - T m )

q␒ w
⇒h=
Tw - Tm

q␒ w
=
q␒ w D
4k

4k
∴h= (31)
D

hD for constant heat flux

Nu = =4
k slug flow in a pipe

Question 2
a)
Considering steady behaviour, the 2D N-S equations can be written as:

∂u ∂v
+ =0 (1)
∂x ∂y

∂u ∂u ∂p 1 ∂ 2 u ∂ 2 u
u +v = - + + (2)
∂x ∂y ∂x Re ∂x 2 ∂y 2

∂v ∂v ∂p 1 ∂ 2 v ∂ 2 v
u + v = - + + (3)
∂x ∂y ∂y Re ∂x 2 ∂y 2

∂u ∂v
From (1) ⇒ = -
∂x ∂y
 ∂p
Assume =0
∂x

∂2u
Assume slowly varying: =0
∂x 2

1 ∂2u ∂u ∂v
( 2) ⇒ - v + u=0 (4)
Re ∂y 2 ∂y ∂y

Comparing (4) with Prandtl's equations:

1
1) ϵ = : Thus, small ϵ represents high Re (inertia forces ≫ viscous forces)
Re

∂v
2) v is considered to be negative andpositive. Both will have similar orders of
∂y
1
magnitude. We will consider them to be of order 1 compared to and approximate them
Re
by 1.

b) Illustrate "boundary-layer" behavior with Prandtl's model differential equation:

d2u du
ϵ + + u = 0,
dy 2 dy

with u(0) = 0 and u(∞) bounded

assuming that ϵ ≪ 1. For finite ϵ, we assume a solution u = K exp(my), leading to

1
ϵm 2 + m + 1 = 0, or: m1,2 = -1 ± (1 - 4ϵ)
2ϵ

Then the general solution has the form u = A exp(m 1 y) + B exp(m 2 y). To satisfy
the initial condition u(0) = 0, we must have B = -A. Then, our desired "boundary layer"
type solution is given by

u = C e m1 y - e m2 y

For ϵ > 0, both m 1,2 are negative and thus the solution is bounded at large y. For
very small ϵ, we may approximate m 1 ≈ - 1 and m 2 ≈ - 1 ϵ .

c) If we neglect the second derivative term entirely (ϵ = 0, corresponding to

"inviscid" or "slip" conditions), the basic differential equation reduces to

du
+ y = 0, with solution u = Ce -y
dy

This is equivalent to the first term of the boundary-layer solution. It is impossible in

this "inviscid" case to satisfy the "no-slip" initial condition, u(0) = 0.
These two solutions, inviscid and boundary-layer, are compared in the figure below.
The boundary-layer solutions satisfy the no-slip condition, and, for small ϵ, merge into the
inviscid solution at larger y. The "boundary-layer" is very thin.

Question 3
a) Assumptions:
• 2D flow
• Constant viscosity
• Steady
∂
• Homogeneous (or fully developed, = 0)
∂x
• Incompressible (liquid)
Thus, the equations are given by

∂u ∂v
+ =0 (1)
∂x ∂y
 ∂u ∂u ∂u ∂p 1 ∂ 2 u ∂ 2 u
+u +v = - + + + fx (2)
∂t ∂x ∂y ∂x Re ∂x 2 ∂y 2

∂v ∂v ∂v ∂p 1 ∂ 2 v ∂ 2 v
+u +v = - + + + fy (3)
∂t ∂x ∂y ∂y Re ∂x 2 ∂y 2

b)
∂v
From (1) = 0 ⇒ v = g( x)
∂y

But the flow is fully developed ⇒ v = 0, thus:

∂p 1 ∂2u
(2) ⇒ - fx = (4)
∂x Re ∂y 2

∂p
(3) ⇒ - fy = 0
∂y

(2) is the differential equation that governs the flow velocity.

c)

∂u
• = 0 (rate of change of u in y should be zero)
∂y liquid surface
• Pressure at the surface should be P atm
• No slip (u(y = 0) = 0) at the wall

∂p
d) Since p = P atm at all x of the surface, = 0. Now fx is given by gravitational
∂x
force per unit volume:

𝜌g sin 𝜃
fx =
𝜌U 2 / h

𝜌U 2
where the term comes from the normalization, thus:
h

1 ∂2u gh
(4) ⇒ = - sin 𝜃
Re ∂y 2 U2
 Re. g. h 1
⇒ u=- sin 𝜃 y 2 + Ay + B
U2 2

In dimensional form:

2
u* -𝜌 U h h 1 y* y*
= . 2 g sin 𝜃 2 + A + B
U 𝜇 U 2h h

-𝜌. g. sin 𝜃 2 A. U
⇒ u* = y* + y * + B. U (5)
2𝜇 h

B.C.1: u * y * = 0 = 0 ⇒ B = 0
B.C.2:
* -𝜌. g. sin 𝜃 A. U A. U 𝜌. g. sin 𝜃
∂u
y* = h = 0 ⇒ 0 = .h+ ⇒ = .h
∂y * 𝜇 h h 𝜇

𝜌. g. sin 𝜃
(5) ⇒ u * = y(2h - y) (For 0 ≤ y ≤ h)
2𝜇

e)

h
h
𝜌. g. sin 𝜃 y2 y3
Q=∫ u * dy * = 2h -
0 2𝜇 2 3 o

𝜌. g. sin 𝜃 3h 3 - h 3 𝜌. g. sin 𝜃. h 3
= =
2𝜇 3 3𝜇

You may also start from the dimensional equations and reach the same final value.

Question 4
In this question, you may also start from the cilindrical NS equations.

Consider a cylindrical element of radius r in the fluid (figure (a)) and let the velocity
be v at this radius. The forces acting are

Force causing motion due to pressure difference

= (p1 - p2 ) x 𝜋r 2 = p𝜋r 2
Viscous drag on cylindrical surface

= area x viscous shear stress

dv
= 2𝜋rL𝜇
dr

Since flow is steady there is no resultant force and

dv
p𝜋r 2 + 2𝜋rL𝜇 =0
dr

prdr
dv = -
2𝜇L

Integrating:
pr 2
v=- +A
4𝜇L

pd 2
At the wall v = 0 and r = d / 2, thus A =
16𝜇L

p d2
v= - r2
4𝜇L 4

Consider an annular element (figure (b))

Flow through element = 𝛿Q = area x velocity

2𝜋p d 2
𝛿Q = 2𝜋r𝛿r x v = - r 2 r𝛿r
4𝜇L 4
Integrating,

2𝜋p d/2 d 2
Total discharge Q = ∫
4𝜇L 0 4
- r 2 r𝛿r

𝜋pd 4
Q=
128𝜇L

pd 2
Q
Mean velocity = V = =
𝜋 / 4d 2 32𝜇L

Thus,
32𝜇LV
p=
d2
Critical speed v c corresponding to Re = 2500 is given by
𝜌vc d
= 2500
𝜇

2500 x 1.44
vc = = 160 m/s
0.9 x 10 3 x 0.025

vc
Mean velocity in pipe = V = = 0.32 m/s
500

Thus,
32𝜇LV
p=
d2

32𝜇LV
𝜌gh = =1
d2

32𝜇LV
h=
𝜌gd 2

Substituting the values from the question, h = 8.02m

Question 5

Equation (1) gives the point linear velocity v θ in a vertical annulus for a Newtonian
liquis in steady laminar flow with the inner and outer walls rotatins at ω 1 and ω 2 radians
per second respectively.
ω2 - γ 2 ω1 r γ 2 di2 (ω2 - ω1 )
vθ = - (1)
1 - γ2 4r 1 - γ 2

Where 𝛾 is the diameter ratio.

For the case of no inner cylinder γ = 0. Put 𝜔 = 𝜔 2 and rewrite equation (1) as

v𝜃 = 𝜔r (2)

For steady laminar rotational flow of a Newtonian liquid about a vertical axis with no
vertical or radial velocity, the modified Navier-Stokes equations in vertical cylindrical
coordinates, reduce to:

v𝜃2 ∂P
𝜌 = (3)
r ∂r

and

∂P
0=- + 𝜌g (4)
∂z

Combine equations (3) and (2) to give

∂P
= 𝜌𝜔 2 r (5)
∂r

Write the differential pressure function as

∂P ∂P
dP = dr + dz (6)
∂r ∂z

Substitute quations (4) and (5) into equation (6) to give

dP = 𝜌𝜔 2 r dr + 𝜌g dz (7)

which can be integrated to give

 𝜌𝜔 2 r 2 (8)
P= + 𝜌gz + C
2

where C is a constant.

Figure 1: Rotating Newtonian liquid producing a parabolic vortex.

Let z = 0 at r = d i / 2 as shown in the picutre above. At this point on the liquid

surface, the pressure is atmospheric, i.e. P = P A . Therefore in equation (8) the constant
C is given by the equation

2
𝜌𝜔 2 di (9)
C = PA -
2 2

Substitute equation (9) into equation (8) to give

2
𝜌𝜔 2 2 di
P - PA = r - + 𝜌gz (10)
2 2

Since the pressure P is atmospheric at every point on the free surface, equation (10)
for the free surface can be written as

2
𝜔2 di
z = - r2 (11)
2g 2

which is the equation of a parabola. At r = 0

𝜔 2 di2
zo = (12)
8g

Equation (11) gives the depth of the free surface at any radial position r for a
Newtonian liquid of constant density and dynamic viscosity in steady laminar rotational
flow.
 MAE3401- Aerodynamics II
Problem set 3: Analytical solutions for steady flow and heat transfer
In this problem set you will have the opportunity to put in practice some of the concepts you saw in
both the lecture videos and the workshop. This will be about analytical solutions of the Navier-
Stokes equations for steady flow and heat transfer. Questions about elements of the analytical
solutions and heat transfer can be found in this week’s Quiz.

General instructions: Please show all your workings carefully, and make sure that all you’ve written
is intelligible. You will still get marks if you show your workings (and drawings), even if you don’t
reach the final answer. Important: drawings will be particularly important in this problem set, so
make sure that all of them are clear.

Question 1: “Slug” flow in a long tube is described as flow in which the fluid velocity is constant
across the entire area of the tube. For a tube with an internal diameter 𝐷 derive the fluid
temperature distribution and obtain an expression for the heat transfer coefficient assuming a
constant heat flux condition is maintained along the inside of the tube.

Question 2: To illustrate “boundary-layer” behaviour, i.e., the effect of the no-slip condition for large
Reynolds numbers, Prandtl in a 1932 lecture proposed the following model (linear) differential
equation:

d! 𝑢 𝑑𝑢
𝜖 + +𝑢 =0
𝑑𝑦 ! 𝑑𝑦
where 𝜖 ≪ 1 mimics the effects of small fluid viscosity. The boundary conditions are 𝑢(0) = 0, and
𝑢 remains bounded as y becomes large.

a. From the normalised 2D incompressible Navier-Stokes equations and continuity, obtain an

expression similar to Prandtl’s, and comment on the assumptions made by this researcher
regarding the wall-normal velocity. You may consider the flow to vary very slowly in 𝑥 and
that the pressure gradient in that direction is negligible. What non-dimensional parameter
does 𝜖 represent?
b. Solve this equation considering the appropriate boundary conditions. Plot (or sketch) the
resulting profile in the interval 0 < 𝑦 < 2 for varying 𝜖 (0.01, 0.05, 0.1 and 1)
c. What happens when 𝜖 = 0? Plot (or sketch) the solution for this case. Is it possible for the
system to comply with the required boundary condition? Comment about the kind of flow
(or flow approximation) this case may represent.
Question 3: A fluid of constant thickness ℎ flows steadily down a flat surface that is inclined at an
angle 𝜃, as shown below.

a. Make the necessary assumptions to reduce the Cartesian

Navier–Stokes equations.
b. Write the differential equation of motion for this problem.
c. Specify a judicious assortment of boundary conditions.
d. Solve for the steady-state velocity profile u(y) for 0 ≤ y ≤ h.
e. Calculate the volumetric flow rate per unit width.

Question 4: Show from first principles that the loss of pressure due to laminar flow in a horizontal
"!#$%
circular pipe is given by 𝑝 = where 𝜇 is the coefficient of viscosity of the fluid, 𝑉 the mean
&!
velocity of the flow, 𝐿 and 𝑑 length and diameter of the pipe. Oil with viscosity 1.44 kg/m-s or N-s/m
and specific gravity (density of fluid divided by density of water) 0.9 flows in a pipe 25 mm diameter
and 3 m long at 1/500 of the critical speed, for which the Reynolds number (based on the mean
velocity) is 2500. Calculate the head (or the height of liquid column in an upstream reservoir) in
metres of oil required to maintain the flow.

Question 5: Find the equation representing the shape of the surface

in a liquid inside a rotating cylinder. What factors impact the depth
of the surface? Hint: you may start with the solution for two co-
rotating cylinders and revisit the momentum equations. As in the 2-
cylinder case, you may consider that the velocity in the axial
direction to be negligible (steady 2D flow).