Advanced Aeronautics

Zhiwei Hu: [email protected]

Ch3. Methods for Boundary
Layers
1. Introduction
2. Blasius & Falkner-Skan Solutions for laminar BL
3. Pohlhausen Solution & Influence of Pressure Gradient
4. Transition to Turbulence
5. Momentum Integral Equation
6. Viscous-Inviscid Interaction
Prandtl’s thin boundary layer theory

𝜕𝜕𝑢𝑢
≈0
𝜕𝜕𝜕𝜕
outside BL
Boundary layer thickness
𝑦𝑦
Non-dimensional velocity profiles by (𝛿𝛿99 , 𝑈𝑈𝑒𝑒 )

δ99 𝑦𝑦/𝛿𝛿99

𝑢𝑢
0.99Ue Laminar
δ99 is the distance from the wall where the
velocity reaches 99% of its free-stream Turbulent
value ----boundary layer thickness.

However, it is difficult to measure exactly 𝑢𝑢/𝑈𝑈𝑒𝑒

so we seek other definitions.
Displacement thickness: δ*
y y
Replace this with this

δ*
A
A
u u
Ue Ue
Match the mass flow defect in area A
∞

∫ e
ρ (U − u ) d y ρ
=U eδ *

0
Displacement thickness: δ*
For the same mass flow rate:
∞
ρ=
U eδ * ∫ ρ (U
0
e − u ) dy

∞
 u 
δ= ∫ 1 −  dy
*

0
Ue 

• δ* is the distance we would have to move the wall for a uniform flow Ue to
get the same mass flow rate as in the boundary layer.

• Adding δ* to an object moving in a viscous fluid gives the equivalent

displacement body, around which potential flow can be assumed for the
same mass flow rate.
Momentum thickness: 𝜃𝜃
We can do the same for momentum:

∞
ρU e2θ
= ∫ ρu (U
0
e − u ) dy

∞
u  u 
=θ ∫ 1 −  dy
0
Ue  Ue 

We will be able to relate momentum thickness to friction drag (later through

more rigorous derivation of the momentum integral equation, MIE).
Shape factor and skin friction
δ*
Shape factor: H=
θ
A single dimensionless number that can be used to indicate the state of the
boundary layer (e.g. exact laminar solution for a flat plate H=2.59)

Skin friction:
 du 
Wall shear stress: τw = µ  
 dy  w

τw
Skin friction coefficient: cf = 1
2 ρU 2
e
Boundary layer on a flat plate

Zero streamwise pressure gradient (i.e. Ue=U∞) and no pressure gradient across the
boundary layer (shown during the derivation of BLE)

Take density ρ=constant (incompressible flow)

Boundary layer on a flat plate
h2

Mass conservation ρU ∞ h1 = ∫ ρ u dy (1)

0
h2

∫ ρ ρ
2 2
Newton’s second law u dy − U ∞ h1 =
− D (2)
0

h2
Substitute U∞(1) into (2):
∫ ρu ( u − U ) dy =
0
∞ −D

h2
Therefore u  u  Since integrand
=D ρU ∫ 2
1 −  dy = 𝜌𝜌𝑈𝑈∞ 𝜃𝜃
2
U ∞ 
∞
U∞  is zero for y>h2
0

The distribution along the surface can be dθ

∫
found by differentiation, since D = τ w dx τ w = ρU 2
∞
dx
Momentum integral equation for a flat dθ τw
plate (‘ZPG MIE’) =
dx ρU ∞2
Revision questions
1. Understand the concept of boundary layer for viscous
flow, and the assumptions of Prandtl’s thin boundary
layer theory.
2. Know well the definitions of basic boundary layer
quantities: BL thickness, displacement and momentum
thicknesses, shape factor and their physical meanings. y
3. Define the wall shear stress and skin friction coefficient，
and how the skin friction coefficient is linked to
momentum thickness for flat plate.
δ
4. Assume the boundary layer velocity profile can be
𝑢𝑢 𝑦𝑦 𝑢𝑢
approximated = for 𝑦𝑦 ≤ 𝛿𝛿, and = 1 for 𝑦𝑦 > 𝛿𝛿 as
𝑈𝑈𝑒𝑒 𝛿𝛿 𝑈𝑈𝑒𝑒
shown in the figure. Find δ*, θ (in terms of δ), H and the
variation of δ*, θ and cf as functions of 𝛿𝛿.
Ue u
Ch3. Methods for Boundary
Layers
1. Introduction
2. Blasius & Falkner-Skan Solutions for laminar BL
3. Pohlhausen Solution & Influence of Pressure Gradient
4. Transition to Turbulence
5. Momentum Integral Equation
6. Viscous-Inviscid Interaction
Contents
• Introduce the variables used to change 2D BL PDEs to ODE for Blasius
and Falker-Skan methods.
• The solution procedure for Blasius and Falkner-Skan methods.
• How to use the solutions to calculate BL quantities.

• More details can be found in section 4.3.1 and 4.3.3 of White ‘Viscous
flow’
• Scanned files are added to Bb for the detailed in deriving the Blasius
and FS equations for those interested.
Similarity in BL Velocity Profiles
 Blasius (1908) – a PhD student of Prandtl
 He found a solution of the BLE for flat-plate laminar BLs by assuming a
similarity in the shape of the velocity profiles that would not change
when plotted in a normalised form: u/Ue against y/δ.

1
Ue Ue
profiles at
x1 and x2
y/δ

U∞ y u δ
x u
0
x1 x2 0 u/Ue 1
plate

 He confirmed the similarity from a wind-tunnel measurement.

Reduced BLE
 Begin with BLE with zero PG (flat plate):
∂u ∂v
+ =0
∂x ∂y
∂u ∂u dU e ∂ 2u
u +v = Ue +ν 2
∂x ∂y dx ∂y

 Introduce a stream function 𝜓𝜓 ensures the continuity eq is satisfied:

∂u ∂v ∂ 2 Ψ ∂ 2 Ψ
+ = − =0
∂x ∂y ∂x∂y ∂y∂x
 Only one equation is left to be solved:
∂u ∂u ∂ 2u ∂Ψ ∂ 2 Ψ ∂Ψ ∂ 2 Ψ ∂ 3Ψ
u +v =ν 2 ⇒ − =ν
∂x ∂y ∂y ∂y ∂x∂y ∂x ∂y 2
∂y 3

 Two second-order PDEs are replaced by a third-order PDE.

Blasius similarity variables
 Blasius’ similarity variable #1:
ξ δ
= k Re −1 / 2
x ⇐ = O(Re −x1/ 2 ) from the derivation of BLE
x x
 ξ is a crude approximation of δ , however they are not the same.
 Blasius chose k = Sqrt(2) for simplicity in the final equation.
2ν 2ν x dξ 2ν 1 2νx 1 ξ
ξ=x = ⇒ = = =
Ue x Ue dx Ue 2 x Ue 2 x 2 x

 Blasius’ similarity variable #2 (dimensionless):

𝑦𝑦 𝑦𝑦
 𝜂𝜂 = (corresponding to , a function of both 𝑥𝑥 and 𝑦𝑦)
𝜉𝜉 δ
𝜕𝜕𝜕𝜕 1 𝜕𝜕𝜕𝜕 𝑦𝑦 𝜕𝜕𝜕𝜕 𝑦𝑦 𝜉𝜉 𝜂𝜂
⇒ = & =− 2 =− 2 =−
𝜕𝜕𝜕𝜕 𝜉𝜉 𝜕𝜕𝜕𝜕 𝜉𝜉 𝜕𝜕𝜕𝜕 𝜉𝜉 2𝑥𝑥 2𝑥𝑥
Blasius’ Similarity Variables
 Blasius’ similarity variable #3 (dimensionless):
 Normalised stream function:

Ψ
f = (assumed to be a function of η only)
U eξ
df d2 f
f = f (η ) ⇒ f′= , f ′′ = 2 , 
dη dη

 Similarity velocity profiles to be obtained:

∂Ψ u
Ψ = U eξ f ⇒ u= =  = Ue f ′ ⇒ f′=
∂y Ue

 u y
⇒  vs.  is equivalent to [ f ′(η ) vs. η ]
U e δ
Blasius Equation
 Work out the derivatives in the reduced BLE to find the Blasius eq:

𝑓𝑓 ′′′ + 𝑓𝑓𝑓𝑓 ′′ = 0
 Two second-order PDEs are reduced to a third-order ODE.
 Still a nonlinear equation but a pseudo-exact solution is available.

 Boundary conditions:
u
 No slip condition at the wall (y = 0): f′= ⇒ f ′(0) = 0
Ue
 Take the wall as a streamline: Ψ( y = 0) = 0 ⇒ f (0) = 0
 Away from the wall (y → ∞): u(∞) = U e ⇒ f ′(∞) = 1

 Numerical solution:
 Three BCs are required at the wall but only two are available.
 A guess-and-shoot method is necessary.
Guess and shooting method for Blasius BL
f ′′′ + f f ′′ = 0
1. Guess the value of 𝑓𝑓 ′′ 0 .
2. Integrating along η to obtain
values of 𝑓𝑓 ′′ η , 𝑓𝑓 ′ η , 𝑓𝑓 η .
3. If the results does not match
𝑓𝑓 ′ ∞ =1, re-guess 𝑓𝑓 ′′ 0 , repeat
from step 1.

e.g. a Matlab function use this method given at

https://uk.mathworks.com/matlabcentral/fileexchange/58996-blaisus-equation-solution.
Blasius Solution
Blasius BL Properties
• BL thickness
u
= f ′ = 0.99 at η = η99 = 3.5
Ue
y δ δ 5.0
η99 = 99 = 99 = 3.5 ⇒ 99 =
ξ ξ x Re x
• Integral type A

f ′′(0) = 0.4696 & lim f (η ) = η − 1.217

η →∞

∞
∞
𝜂𝜂 ∗ = � (1 − 𝑓𝑓 ′ )𝑑𝑑𝑑𝑑 = (𝜂𝜂 − 𝑓𝑓)�
0
0
= lim [ 𝜂𝜂 − (𝜂𝜂 − 1.217)] − [0 − 𝑓𝑓(0)]
𝜂𝜂→∞
= 1.217
Blasius BL Properties
• Integral type B
∞ ∞ ∞
∞ ∞ ∞ ∞
Θ∗ =� 𝑓𝑓 ′ (1 − 𝑓𝑓 ′ )𝑑𝑑𝑑𝑑 = 𝑓𝑓(1 − 𝑓𝑓 ′ )� +� 𝑓𝑓𝑓𝑓 ″ 𝑑𝑑𝑑𝑑 = 𝑓𝑓(1 − 𝑓𝑓 ′ )� −� 𝑓𝑓 ′″ 𝑑𝑑𝑑𝑑 = 𝑓𝑓(1 − 𝑓𝑓 ′ )� − 𝑓𝑓 ″ �
0 0 0 0
0 0 0
′ ′ ″ ″
= lim [ 𝑓𝑓(1 − 𝑓𝑓 )] − 𝑓𝑓(0)[1 − 𝑓𝑓 (0)] − lim ( 𝑓𝑓 ) + 𝑓𝑓 (0)
𝜂𝜂→∞ 𝜂𝜂→∞
= lim [ 𝑓𝑓(1 − 𝑓𝑓 ′ )] − lim ( 𝑓𝑓 ″ ) + 0.4696
𝜂𝜂→∞ 𝜂𝜂→∞

• In the meantime, we can deduce:

c1 c2 c3
lim f (η ) = η − 1.217 ⇔ f (η ) = η − 1.217 + + + +
η →∞ η η 2
η 3

c1 2c 2 3c 3
⇒ f ′(η ) = 1 − − − +
η 2
η 3
η 4

• This means:
lim f ′(η ) = 1 & lim f ′′(η ) = 0
η →∞ η →∞
Blasius BL Properties
 c c c 
• Also, f (η )[1 − f ′(η )] = η − 1.217 + 1 + 22 + 33 +  ×
 η η η 
 c1 2c 2 3c 3 
 2 + 3 + 4 + 
η η η 
 c 
⇒ lim[ f (1 − f ′)] = lim η × 12  = 0
η →∞ η →∞
 η 
• Therefore,

Θ∗ = lim [ 𝑓𝑓(1 − 𝑓𝑓 ′ )] − lim ( 𝑓𝑓 ″ ) + 0.4696

𝜂𝜂→∞ 𝜂𝜂→∞
= 0.4696
Blasius BL Properties
• Displacement thickness
∞ ∞ ∞
𝑢𝑢 𝜕𝜕𝜕𝜕
𝛿𝛿 ∗ =� 1− 𝑑𝑑𝑑𝑑 = � (1 − 𝑓𝑓 ) 𝑑𝑑𝑑𝑑 = � (1 − 𝑓𝑓 ′ )𝜉𝜉𝜉𝜉𝜉𝜉
′
𝑈𝑈𝑒𝑒 𝜕𝜕𝜕𝜕
0 0 0
∗
𝛿𝛿 1.721
⇒ =
𝑥𝑥 Re𝑥𝑥
• Momentum thickness
∞ ∞ ∞
𝑢𝑢 𝑢𝑢 𝜕𝜕𝜕𝜕
𝜃𝜃 = � 1− 𝑑𝑑𝑑𝑑 = � 𝑓𝑓 (1 − 𝑓𝑓 ) 𝑑𝑑𝑑𝑑 = � 𝑓𝑓 ′ (1 − 𝑓𝑓 ′ )𝜉𝜉𝜉𝜉𝜉𝜉
′ ′
𝑈𝑈𝑒𝑒 𝑈𝑈𝑒𝑒 𝜕𝜕𝜕𝜕
0 0 0
𝜃𝜃 0.664
⇒ =
𝑥𝑥 Re𝑥𝑥

• Shape factor
δ ∗ δ ∗ /x
H= = = 2.592
θ θ /x
Blasius BL Properties
• Skin friction coefficient
τ 2 µ ∂u 2µ ∂η
cf = 1 w 2 = = U e f ′′( 0 ) ⇐ u = Ue f ′
2 ρU e ρU e ∂y y =0 ρU e ∂y
2 2

0.664
⇒ cf =
Re x
dθ τw
• Show it beginning with =
dx ρU e2
Generalized Laminar BLs
 Potential flow velocity varies with “x”: U e ( x ) = cx m (c , m : constants)

 It means wedge/corner flows (depending on “m”):

π
α=
m +1

βπ 2m
 Relation between m and β : α + 2 = π ⇒ β=
m +1
 m = 0: flat plate flow (Blasius) ↔ β = 0
 m > 0: accelerating flow (wedge or compression corner) ↔ β > 0
 m < 0: decelerating flow (expansion corner) ↔ β < 0
Special Cases in Laminar BLs
 Flat plate flow

 Stagnation point flow (aerofoil leading edge, etc.)

 Laminar BL separation

β = −0.19884
Falkner-Skan similarity variables
• Falkner-Skan’s similarity variable #1:
2ν x  2ν x  𝑑𝑑𝜉𝜉 𝜉𝜉
ξ=  ξ = for Blasius  = (1 − 𝑚𝑚)
( m + 1)U e  Ue  𝑑𝑑𝑑𝑑 2𝑥𝑥
y
• #2: η = (same as Blasius)
ξ
∂η 1 ∂η η
⇒ = & == ( m − 1)
∂y ξ ∂x 2x
• #3: Same as Blasius but Ue is no longer constant.
∂Ψ u
Ψ = U eξ f ⇒ u= =  = Ue f ′ ⇒ f′=
∂y Ue

 u y
⇒  vs.  is equivalent to [ f ′(η ) vs. η ]
U e δ
Falkner-Skan Equation & its Solution
η* is integral A.
θ* is integral B.

• In the same way, we can derive

the following Falkner-Skan eq for BL
with pressure gradient:

𝑓𝑓 ′′′ + 𝑓𝑓𝑓𝑓 ′′ + 𝛽𝛽(1 − 𝑓𝑓 ′2 ) = 0

• and solve under the same BCs as

were used for Blasius eq.
Falkner-Skan Solution
𝜂𝜂 ∗ *

𝜂𝜂 ∗ is integral A.
θ* is integral B.

https://uk.mathworks.com/matlabcentral/fileexchange/9712-numerical-solution-of-the-
falkner-skan-equation-for-various-wedge-angles
Falkner-Skan Example Question
• Find δ99 near the front stagnation point for flow over circular cylinder:

R = 10 cm
x
U ∞ = 30 m / s
Front stagnation point

• From potential flow theory:

u(θ ) = 2U ∞ sin θ (wall tangential velocity)

Falkner-Skan Example Question
• For a small angle near the stagnation point:
x
θ << 1 ⇒ u(θ ) ≅ 2U ∞θ ⇒ u( x ) = 2U ∞ = 600 x = cx m
R

• This corresponds to the F-S solution for β = m = 1.

• From the table, η99 ≅ 2.4

2ν x
δ 99 = η99ξ = 2.4
( m + 1)U e
2× 1.46 × 10 −5 x
= 2.4
2 × 600 x
= 3.74 × 10 − 4 m = 0.374 mm
Revision questions
1. Know well the definitions of variables introduced in Blasius and Falker-
Skan methods to change the BL equations to ODEs.
2. Find the values of integrals A & B for flat plate boundary layer from
Blasius solutions.
3. Show how values of the integrals A & B are related to boundary layer
thicknesses.
4. Calculate boundary layer displacement and momentum thicknesses
as function of distance along the wedge wall for flow over wedge
when 𝛽𝛽 = 0.3 from Falker-Skan solutions.
Ch3. Methods for Boundary
Layers
1. Introduction
2. Blasius & Falkner-Skan Solutions for laminar BL
3. Pohlhausen Solution & Influence of Pressure Gradient
4. Transition to Turbulence
5. Momentum Integral Equation
6. Viscous-Inviscid Interaction
Pressure Gradient
𝑑𝑑𝑈𝑈𝑒𝑒 𝑑𝑑𝑝𝑝
• Accelerating external flow: > 0, < 0 (FPG)
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑

𝑑𝑑𝑈𝑈𝑒𝑒 𝑑𝑑𝑝𝑝
• Decelerating external flow: > 0, < 0 (APG)
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
• Pressure increase downstream
• Fluid elements retarded leading to boundary-layer separation

dp e
>0
dx

Dividing
streamline

𝜕𝜕𝜕𝜕
=0
𝜕𝜕𝜕𝜕 𝑤𝑤
BL Prediction if Pressure Gradient Known?
• Recall BLEs:
∂u ∂v
+ =0
∂x ∂y
∂u ∂u dU e ∂ 2u
u +v = Ue +ν 2
∂x ∂y dx ∂y
• Try to fit a fourth-order polynomial (5 coefficients) for BL velocity profile

u y
= a0 + a1η + a 2η 2 + a3η 3 + a 4η 4 where η =
Ue δ
d  u 
  = a1 + 2a 2η + 3a3η 2 + 4a 4η 3
dη  U e 
d2  u 
  = 2a 2 + 6a 3η + 12a 4η 2
dη 2  Ue 
Polynomial Coefficients
• Boundary conditions:
u
• No-slip at the wall (η = 0): = 0 = a0
Ue η =0

• Match edge velocity at η = 1: u

= 1 = a1 + a 2 + a 3 + a 4
Ue η =1

• Asymptote smoothly to freestream:

d  u 
  = 0 = a1 + 2a 2 + 3a3 + 4a 4
dη  U e  η =1
d2  u 
  = 0 = 2a 2 + 6a3 + 12a 4
dη 2  U e  η =1
• One more condition to be considered.
BLE at Wall
• Due to the no-slip condition, at the wall u = v = 0, the momentum Eq
becomes
dU e ∂ 2u
0 = Ue +ν 2
dx ∂y wall
• Working with scaled variables:
𝑑𝑑𝑈𝑈𝑒𝑒 𝑈𝑈𝑒𝑒 𝜕𝜕 2 (𝑢𝑢/𝑈𝑈𝑒𝑒 ) 𝑑𝑑𝑈𝑈𝑒𝑒 𝑈𝑈𝑒𝑒 𝑑𝑑2 𝑢𝑢
𝑈𝑈𝑒𝑒 + 𝜈𝜈 2 2
� = 𝑈𝑈𝑒𝑒 + 𝜈𝜈 2 2 � =0
𝑑𝑑𝑑𝑑 𝛿𝛿 𝜕𝜕(𝑦𝑦/𝛿𝛿) 𝑑𝑑𝑑𝑑 𝛿𝛿 𝑑𝑑𝜂𝜂 𝑈𝑈𝑒𝑒
wall 𝜂𝜂=0

• A new dimensionless pressure gradient parameter is defined as:

𝛿𝛿 2 𝑑𝑑𝑈𝑈𝑒𝑒
𝜆𝜆 =
𝜈𝜈 𝑑𝑑𝑑𝑑 𝑤𝑤
• With the given velocity profile: 𝜆𝜆 = −2𝑎𝑎2
Pohlhausen Solution (1921)
• Working out the coefficients w.r.t. the PG parameter:
a1 + a 2 + a3 + a 4 = 1
a1 + 2a 2 + 3a3 + 4a 4 = 0
2a 2 + 6a3 + 12a 4 = 0
2a 2 = − λ

• Pohlhausen laminar BL solution:

u λ
= 2η − 2η 3 + η 4 + η (1 − η )3
Ue 6
Pohlhausen Solution Profile
u η (1 − η )3
= F + λG ⇐ F(η ) = 2η − 2η 3 + η 4 & G(η ) =
Ue 6

Flat-plate BL profile with no PG The additional effect of PG

Effect of Pressure Gradient (λ)

u/Ue

• Laminar separation (du/dy = 0 at wall) when λ = -12

• Favourable pressure gradient (FPG): λ > 0
• Zero pressure gradient (ZPG): λ = 0
• Adverse pressure gradient (APG): λ < 0
Effect of Pressure Gradient (λ)

δ 2 dU e δ dU e δ dp e
λ= = Re δ = − Re δ
ν dx U e dx ρU e2 dx

• More negative and hence more likely to separate for

• Stronger APG (higher pressure downstream → dpe/dx > 0)
• Higher Reynolds number
• Thicker boundary layer
• Lower freestream velocity
Delay of BL Separation

From Van Dyke’s “An Album of Fluid Motion”

Laminar vs. Turbulent BL Separation
• Laminar BL separates much sooner.
• Additional mixing due to turbulent eddies keep such flows attached
for longer.
• Applications:
• Golf ball dimples
• Trip wind-tunnel test models to be more representative of flight conditions
(effectively higher Reynolds numbers)
Turbulent BLs for Pressure Drag Reduction
• Retarding flow separation by turbulent BLs
• Smaller separation area with TBLs ⇒ lower pressure drag
• Skin friction drag may increase but the reduction in pressure drag is
much bigger in bluff body cases.
Why Dimples on a Golf Ball?
• Patents on size, shape, density & layout of the dimples
Sharp Expansion Corner

From Van Dyke’s “An Album of Fluid Motion”

Revision questions
1. Explain why 𝐶𝐶𝑓𝑓 = 0 at the flow separation point.
𝑑𝑑𝑈𝑈𝑒𝑒 𝜈𝜈 𝑑𝑑 2 𝑢𝑢
2. Show that = − 2 2 �
𝑑𝑑𝑑𝑑 𝛿𝛿 𝑑𝑑𝜂𝜂 𝑈𝑈𝑒𝑒 𝜂𝜂=0
𝛿𝛿 2 𝑑𝑑𝑈𝑈𝑒𝑒
3. Explain why 𝜆𝜆 = is a non-dimensional parameter for pressure
𝜈𝜈 𝑑𝑑𝑑𝑑
gradient.
4. Use Pohlhausen solution, compare BL velocity profiles and explain
influence of pressure gradient.
5. Use Pohlhausen solution, find the value of non-dimensional pressure
gradient when boundary layer separate.
6. Discuss influence of pressure gradient on boundary layer separation,
and methods to delay boundary layer separation.
Ch3. Methods for Boundary
Layers
1. Introduction
2. Blasius & Falkner-Skan Solutions for laminar BL
3. Pohlhausen Solution & Influence of Pressure Gradient
4. Transition to Turbulence
5. Momentum Integral Equation
6. Viscous-Inviscid Interaction
Contents
• Introduce flow transition to turbulence, and discuss different transition
scenario
• Roadmap through transition: summarize different route of flow transition
• Introduce linear stability theory
• A practical transition prediction method: en method.
Reynolds Pipe Flow Experiment (1883)
• Dye visualisations of transition

ρUD
• Critical value of
µ
• Later called the Reynolds number Re.

• Critical Re
• Up to 13,000 (and higher) with careful attention to inlet design
• 2,300 for straight inlet
Factors Affecting Transition
• Environment
• Surface roughness, ice, bugs, etc.
• Free-stream turbulence/gust
• Surface vibration
• Noise (e.g. engine)
• Aerodynamic design factors:
• Sweep angle (spanwise cross-flow)
• Pressure gradient (favourable/adverse)
• Surface curvature
• Compressibility (high Mach flows)
• Heat transfer (thermal boundary layers)
Transition in a Low Disturbance Environment
• More representative of cruise flight
• Experimental visualization (snapshot) of Tollmien-Schlichting-wave-
induced breakdown on an axisymmetric body by smoke
Flat Plate Transition
Λ Vortices
• Direct numerical simulation of Meyer & Rist (1998)
 Roadmap
through
transition

Recommended reading :
FM White, Viscous Fluid Flow,
3rd ed., McGraw - Hill, Chap. 5.
Elements of Stability Theory
• Decompose into base flow and small perturbation:
u = u + u′, v = v + v ′ & p = p + p ′
• Substitute in the Navier-Stokes equations and linearise (remove
products from small quantities).
• Assume a wavy solution (normal mode approximation):
u′ = uˆ ( y )e i (αx−ωt )
• Wavenumber: α = 2π / λ (λ = wavelength)
• Angular frequency: ω = 2π /T (T = period)
• The shape function contains all y-independence.
• Final form: Orr-Sommerfeld Eqn. (refer to textbooks)
Orr-Sommerfeld Equation
 d 2 vˆ  d 2
u i ν  d 2
ˆ
v d 4
vˆ 
( u − c ph ) 2 − α vˆ  − vˆ 2 = − α vˆ − 2α
2 4 2
+ 4
 dy  dy α  dy 2
dy 
ω
c ph = (phase speed)
α

• Boundary layers are usually convectively unstable (disturbances grow as

they convect downstream).

• In this case, ω is a real quantity and we solve for the (in general complex)
α (eigenvalue) and 𝑣𝑣� (eigenvector).
Waviness and Growth
• For ω real and α complex:

v ′( x , y , t ) = vˆ ( y )e i[(α r + iα i ) x−ωt ]
ˆ
= v( y ) e e
i (α r x −ωt ) −α i x

Shape
Wave
Growth
(growth rate = −α i )
Instability Diagram
• Inflectional, inviscid unstable flow:
Boundary Layer Instability (Viscous)
• Blasius boundary layer as an example:

Dimensionless
Frequency
Amplitude Variation at a Fixed Frequency
• Going through the unstable region means that the amplitude grows
(vice versa in the stable region).
“n” Factor
• Local growth from linear stability theory:

A = Aref e −α i x
dA
= −α i Aref e −α i x = −α i A
dx

• Now allow αi to vary with x and integrate over x from neutral point
(x=x0, A=A0) to x:
A x
dA dA
A
= −α i dx ⇒ ∫A A = − x∫ α i dx
0 0

x
 A
n ≡ ln  = − ∫ α i dx (the definition of n-factor)
 A0  x0
en Method
• It was found that transition occurs when n=9 is reached.
• Smith & Gamberoni (1956) and van Ingen (1956)

ncrit =

• ncrit can be lower in a noisy condition (e.g. upstream turbulence).

en Method
• For different frequencies, compute the local growth rates and
integrate to give n-factors.
• Construct the envelope of the maximum n-factors at any x (over all F).
• For a given ncrit, read off the transition location.
• Limitations:
• Ignores receptivity and nonlinear stages of transition.
• Doesn’t fully include non-parallel effects.
• Recent research: parabolised stability equations (PSE) includes weak
non-parallel and nonlinear effects.
Effect of Pressure Gradient
• Stabilising with FPG (pressure decreasing downstream)
• Destabilising with APG

Dimensionless (Decelerating flow)

Frequency

(Flat plate flow)

(Accelerating flow)
Effect of Pressure Gradient
• Adverse pressure gradient: destabilising
• Lower critical Re
• Larger unstable region
• Unstable as Re→∞ (inflectional profiles)
• Favourable pressure gradient: stabilising
• Higher critical Re
• Smaller unstable region
• Stable as Re→∞
Revision questions
1. Discuss the natural transition process, use flow transition on flat plate
with low disturbance as example.
2. Familiar with the roadmap through transition and understand
different routes for flow transition.
3. Explain the idea of linearization, and linear stability theory for flow
transition analysis.
4. Understand the ‘thumb plot’
5. Be able to use en method to prediction flow transition using solutions
of from linear stability analysis.
6. Influence of pressure gradient on flow transition.
Ch3. Methods for Boundary
Layers
1. Introduction
2. Blasius & Falkner-Skan Solutions for laminar BL
3. Pohlhausen Solution & Influence of Pressure Gradient
4. Transition to Turbulence
5. Momentum Integral Equation
6. Viscous-Inviscid Interaction
Contents
• Explain the logics for developing MIE for BL.
• Derive the MIE for 2D incompressible BL.
Momentum Integral Eq.: von Kármán (1921)
• For many practical applications, details of the flowfield, e.g. velocity
profile within the BL, are of secondary importance, the overall
behaviour of the flow are the main concern.

• The BLE are PDEs, flow quantities depend on both coordinates (x, y).

• If integrated across BL, it is possible to change the BLE to ODE where

quantities depend on x only.

• We have already defined several variables by integrating BL velocity,

and those variables do have physical meanings.

• Aiming to develop ODE (x only) for physical quantities of BL.

Momentum Integral Eq.: von Kármán (1921)
• Beginning with the BLEs:
∂u ∂v ∂u ∂u dU e ∂ 2u
+ =0 & u +v = Ue +ν 2
∂x ∂y ∂x ∂y dx ∂y

• Multiply continuity by u − Ue:

∂u ∂v
(u − U e ) + (u − U e ) =0
∂x ∂y

• Add to momentum:
∂u ∂u ∂u ∂v dUe ∂ 2u
u + (u − Ue ) + v + (u − Ue ) = Ue +ν 2
∂x ∂x ∂y ∂y dx ∂y
∂(u − Ue ) ∂Ue ∂u ∂(u − Ue ) ∂Ue ∂v dUe ∂ 2u
u +u + (u − Ue ) + v +v + (u − Ue ) = Ue +ν 2
∂x ∂x ∂x ∂y ∂y ∂y dx ∂y
Momentum Integral Equation (MIE)
• Continue to work out:
∂u ∂[ v( u − U e )] dU e ∂ 2u
( 2u − U e ) + = Ue +ν 2
∂x ∂y dx ∂y
⇓
∂u ∂( u − U e ) dU e ∂[ v( u − U e )] dU e ∂ 2u
(u − U e ) + u +u + = Ue +ν 2
∂x ∂x dx ∂y dx ∂y
⇓
∂  2 u  u   u  dU e ∂[ v( u − U e )] ∂ 2u
Ue  − 1 +  − 1U e + =ν 2
∂x  U e U
 e   e U  dx ∂ y ∂y

• Integrate in y (coordinate normal to the wall):

∞
∂  2 u  u 
∞
 u  dU e
∞
∂[ v( u − U e )] ∞
∂ 2u
∫0 ∂x U e U e 
 Ue
− 1dy + ∫ 
 0  Ue
− 1U e
 dx
dy + ∫
0 ∂y
dy = ν ∫ 2 dy
0 ∂y
Momentum Integral Equation (MIE)
• First term: ∞
∂  2 u  u  d ∞ 2 u  u 
∫0 ∂x  e U e  U e 
U  − 1 dy = ∫ U e 
dx 0  U e  U e
− 1dy

d  2∞ u  u   d(U e2θ )
dx  ∫0 U e
= Ue  − 1dy  = −
U
 e   dx
• Second term:
 dU e dU e ∞  u dU e ∗
∞
 u 
∫0  U e  e dx
− 1 U dy = U e ∫ 
dx 0  U e
− 1

dy = −U e
dx
δ

• Third term: ∞
∂[ v( u − U e )]
∫0 dy = [ v ( u − U e 0 = Ve (U e − U e ) − 0 × ( 0 − U e ) = 0 − 0 = 0
)]∞

∂y

• Last term: ∞
 ∂U e ∂u 
∞
∂ u 2
∂u  ∂u
 ∂u    τw
ν ∫ 2 dy = ν =ν
 ∂y
−

=ν −
 ∂y ∂y 
=−
0 ∂y ∂y 0  y =∞ ∂y y =0   y =0 
ρ
Momentum Integral Equation (MIE)
• Taking them all back to the earlier equation:
d(U e2θ ) dU e ∗ τ w
+ Ue δ =
dx dx ρ
⇓
dθ dU e τ w
U e2 + ( 2θ + δ ∗ )U e =
dx dx ρ
⇓
dθ θ dU e c f dθ c f
+ (2 + H ) = ⇒ = for a flat plate
dx U e dx 2 dx 2

• One equation with three unknowns (assuming that Ue is known)

• Applicable to both laminar and turbulent BLs
• Only for integral quantities (BL velocity profiles are unavailable)
Summary
• Understand well the process of developing the MIE for BLs.
• SESA3043: Be able to follow the process and understand well in
derivation of the MIE
• SESA6087: Be able to derive the MIE for 2D incompressible BLs.
Ch3. Methods for Boundary
Layers
1. Introduction
2. Blasius & Falkner-Skan Solutions for laminar BL
3. Pohlhausen Solution & Influence of Pressure Gradient
4. Transition to Turbulence
5. Momentum Integral Equation
6. Viscous-Inviscid Interaction
Contents
• Developing two VII methods.

• Discuss their advantages and disadvantages.

Viscous-Inviscid Interaction (VII)
• An efficient approach to predict aerodynamic forces on a streamlined
body such as aerofoil, interactively combining inviscid (potential)
solutions with the viscous (boundary layer) solutions around the body.

• Two different types available

• Surface displacement method: re-panelling with new δ∗ after every
iteration.

• Surface transpiration method: no need to re-panel, imposing imaginary

wall blowing/suction to compensate δ∗.
Surface Displacement Method (SDM)
• Displacement thickness (δ∗) is the distance the wall has to be displaced
to set the same mass flow rate with a slip-wall boundary (no BL).

Locate panels here and

solve the potential flow
above with slip-wall BC.

∞ ∞
 u 
A = ∫ (U e − u)dy = U eδ ∗ ⇒ δ ∗ = ∫ 1 − dy
0 0  Ue 
Surface Displacement Method (cont.)
• When there is a BL around an aerofoil, the outer potential flow sees the
aerofoil as if it is thicker than the original shape by δ∗.

• 1. Use panel method to solve Ue & dUe/dx with the displaced surface
and slip-wall BC → CL is obtained.

• 2. Use viscous method (MIE + Thwaites for example) to find new δ∗ with
the given Ue & dUe/dx → CD is obtained.

• 3. Re-panel and iterate to converge.

Surface Transpiration Method (STM)
• A convenient method not to require the re-panel step
• The panels stay on the original body surface.
• Introduce blowing/suction velocity through the wall to generate the same
effect as the surface displacement method.
U e (x ) U e ( x + ∆x )

Panel location if the

surface displacement
method were used.

δ (x )
∗
vs δ ∗ ( x + ∆x )
Current panel with
wall-normal velocity
artificially created

• The “blue” line must coincide with a streamline. Why??

Surface Transpiration Method (cont.)
• Slip-wall boundary condition on the panel (SDM) means that there is
no flow penetration through it, and therefore the SDM panel is a
streamline.

• The blowing/suction velocity on the STM panel is intended to adjust

one of the streamlines in the flow exactly aligned to the SDM panel,
hence the STM generates the same solution as SDM.

• So, what would be the blowing/suction velocity?

Surface Transpiration Method (cont.)
• Consider the tunnel between the two panels:
U e (x ) U e ( x + ∆x )

δ (x )
∗
vs δ ∗ ( x + ∆x )

∆x
• Mass conservation through the tunnel:
U e ( x )δ ∗ ( x ) + v s ∆x = U e ( x + ∆x )δ ∗ ( x + ∆x )

d(U eδ ∗ )
vs =
dx
Surface Transpiration Method (cont.)
• STM requires updating vs after every iteration instead of re-panelling.

• The wall normal velocity can be easily specified as a Dirichlet

boundary condition in the panel code.

• XFoil uses this approach to produce viscous solutions around aerofoils.

Revision questions
• Understand the two VII approaches, and their pros and cons.
𝑑𝑑 𝑈𝑈𝑒𝑒 𝛿𝛿 ∗
• Show that in STM, blowing/suction with velocity 𝑣𝑣𝑠𝑠 = should be
𝑑𝑑𝑑𝑑
applied on the wall to mimic the surface displacement effect of
viscous flow.