Aerodynamics

Turbulent Flow

• Main features of turbulent flow

• Reynolds-averaged Navier-Stokes equations

• Turbulence models

• Turbulent boundary-layers. Near-wall turbulent flow.

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

1. Caothic fluctuations with a wide range of frequencies and

amplitudes. Unsteady flow of aleatory nature.
For enginnering purposes, it is often required to apply
statistical methods (average values, standard deviations,
spatial and/or time correlations)
2

1.8

1.6

1.4

1.2

0.8

0.6

0.4

0.2

0
0 100 200 300 400 500 600 700 800 900 1000
t (ms)

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

2. Three-dimensional flows. Highly distorted eddies of

different shapes and dimensions

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow

3. High mixing properties. Large heat, momentum and

mass transfer rates, which can be orders of magnitude
larger than those due to molecular diffusion.

4. Turbulent flow is highly dissipative. Energy is transferred

from the “mean flow” to the turbulence field (fluctuations)
from the largest eddies. Energy transfer process (small
to high frequencies) due to vortex stretching.

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
1. Aleatory
2. Three-dimensional
3. High diffusion
4. Dissipative
5. Property of the flow
6. Continuous medium
7. High Reynolds number

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Energy cascade with frequency or wavenumber

• Smallest frequencies,
largest wavelengths, correspond
to the eddies with the highest
turbulence kinetic energy.
The largest dimensions of the
eddies is limited by the boundary
conditions. Highly anysotropic
eddies.

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Energy cascade with frequency or wavenumber

• The highest frequencies, lowest

wavelengths, correspond to the
dissipative eddies.
Smallest dimensions of the
eddies is limited by the
molecular shear-stress.
Smallest Reynolds numbers,
higher viscous effects, imply an
increase of the dimensions of
the dissipative eddies. Isotropic
eddies.
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Energy cascade with frequency or wavenumber

• Inertial range in the

intermediate region promoting
the energy transfer by a
mechanism involving
vortex stretching

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Entrainment

• A turbulent boundary-layer grows by molecular diffusion and

entrainment, i.e. entraining “external fluid” in to the
boundary-layer. The entrainment effect is significantly larger
than molecular diffusion.
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Entrainment

• For zero pressure gradient

– Laminar boundary-layer grows approximately 2.5mm per meter

– Turbulent boundary-layer grows about 18mm each meter

Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Entrainment

• Entrainment velocity, VE quantifies the increase of the

volumetric flow rate along the boundary-layer
d   u 
VE 
dQ d 
 
dx dx 0
udy   
dx 0 
U e  U 
e 1  
  dy 
d
 
Ue    * 
 U e  dx
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Coanda effect

Inflection
point

Transition

Separation

Recirculation Re-attachment
bubble

• Re-attachment of a shear-layer to a nearby solid wall

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Coanda effect

Inflection
point

Transition

Separation

Recirculation Re-attachment
bubble

• Laminar free shear-layer includes an inflection point.

Therefore, there is a quick transition to turbulent flow
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Coanda effect

Inflection
point

Transition

Separation

Recirculation Re-attachment
bubble
• Entrainment of the turbulent shear-layer “diffuses” momentum
to the fluid close to the step making the pressure drop.

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Coanda effect

Inflection
point

Transition

Separation

Recirculation Re-attachment
bubble
• Transverse pressure gradient deflects the free shear-layer
to the wall to balance the forces

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Direct Numerical Simulation, DNS

• Navier-Stokes equations solved numerically with a grid

spacing and a time step sufficiently small to resolve the
smallest eddies of the flow. On the other hand, simulation
time must be large enough to capture the effects of the
largest eddies
• Numerical accuracy of the solution is very important (to
avoid misleading results due to numerical diffusion)
• Dependent variables change in space and time, i.e. they
are the instataneous values at a given flow location

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Large-Eddy Simulation, LES

• Navier-Stokes equations filtered in space. Extra

mathematical model required to include the effect of
the filtered scales. Time dependent numerical
solution.

• Numerical accuracy is also important (“on-going

debate” about “how much”). Correct application to
near-wall flows is difficult.

• Dependent variables change with time, but have a

different meaning of DNS, due to filtering in space
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Multi-Scale Simulation

• Navier-Stokes equations split in to small and large

scales. Analytical methods used to obtain an
approximate solution of the small scales. Numerical
solution of the large scales in time with effect of small
scales included from its approximate solution.
• Numerical accuracy is still important. Application near-
walls troublesome (“small scales” become too small...)

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged equations
• Statistics applied to mass conservation and
momentum balance (continuity and Navier-Stokes).
Type of statistic handling depends on flow properties:
1. Spatial averaging
2. Time averaging
3. Ensemble averaging
• Instataneous velocity components, u~i , (dependent
variables) split in to a mean value, U i , and a
fluctuation, ui
u~i  U i  ui
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged equations
1. Spatial averaging

 u~ x , y , z 
j i i i
U j  lim i 1
n n

Homogeneous turbulence

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged equations
2. Time averaging
2

1.8

u~i dt
t oT

1.6

to 1.4
U i  lim
T  T 1.2

0.8

Statistically steady flow 0.6

0.4

0.2

0
0 100 200 300 400 500 600 700 800 900 1000
t (ms)

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged equations
3. Ensemble averaging

 u~ (t )
n

j i
U j  lim i 1
n  n

Mean values are

time dependent.
Suitable for
periodic flows.

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged equations
3. Ensemble averaging

 u~ (t )
n

j i
U j  lim i 1
n  n

Mean values are

time dependent.
Suitable for
periodic flows.

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged equations
3. Ensemble averaging

 u~ (t )
n

j i
U j  lim i 1
n  n

Mean values are

time dependent.
Suitable for
periodic flows.

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Time averaging applied to the dependent

variables and to “conservation” principles
toT ~
__
~
i  lim
 to
i dt
 i
T  T
~
i stands for any of the dependent variables
(incompressible flow u,v,w,p)

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Decomposition of the instantaneous variables

~
   i  i

~
  Instantaneous variable
 i  Mean value
i  Fluctuation around the mean value

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Consequences of time averaging


 0  Time derivative of mean value is zero
t
__
 
 0 
t
__  Mean value of the time derivative of the
 
 0  fluctuations is zero
xi 

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Linear terms
___
~ ~
   
   0
t t t t
___
~ ~
    
   
xi xi xi xi xi

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Non-linear terms
~ ~ ~ ~~ v~~ w ~ ~
   u
u~ i  v~ i  w
~ i  i
 i
 i
x y z x y z

____ ___
~
u~ ji U i  j u
 
x j x j x j

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Continuity equation

U V W
  0
x y z

- Velocity fluctuations also satisfy

u v w
  0
x y z

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Momentum equations
U U U 1 P   U __   U __   U __
U V W     uu     uv     uw 
x y z  x x  x  y  y  x  z 
V V V 1 P   V __   V __   W __ 
U V W     vu     vv     vw 
x y z  y x  x  y  y  z  z 
W W W 1 P   W __    W __   W __
U V W     wu     wv     ww 
x y z  z x  x  y  y  x  z 
__
•  ui u j Reynolds stresses
• The number of equations is smaller than the number
of unknowns
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

__
• Transport equation of  ui u j
__ __ __ _________
Dui u j ui u j ui u j  __ U i __ U j  p  ui u j 
 Uk   ui u j  ui u k     
Dt t xk  xk xk    x j xi 

____ __ __
 1  pu j pui 
 uiu juk    
xk   xi x j 
__ ____
 2ui u j ui u j
  2
xi2 xk xk

• System remains with less equations than unknowns

Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Reynolds stress models

- 6 additional transport equations

- Most of the terms of the Reynolds stresses
transport equations must be modeled, including
pressure fluctuations
- There are alternative Explicit Algebraic Stress
Models (EARSM) available
- Turbulence anisotropy is taken into account

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity models

- Boussinesq hypothesis: the Reynolds stresses

are proportional to mean velocity derivatives

- The proportionality constant is the eddy-viscosity

- It is difficult to include the anisotropy of turbulence.

Most models are isotropic, i.e. eddy-viscosity is a
scalar quantity

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Momentum equations including an effective viscosity

U U U 1 P   U     U V      U W  
U V W   2  ef    ef       ef   
x y z  x x  x  y   y x   z   z x  
V V V 1 P    U V    V     V W 
U V W    ef      2  ef    ef    
x y z  y x   y x  y  y  z   z y 
W W W 1 P    U W      V W     W 
U V W    ef       ef      2  ef 
x y z  z x   z x   y   z y   z  z 
 ef    t

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Momentum equations including an effective viscosity

t is the eddy-viscosity

- Eddy-viscosity is obtained from velocity and length

scales of turbulence

- Several models proposed in the last 50 years

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- Algebraic models

- Turbulence length scale

l  y  Mixing length

- Turbulence velocity scale

 
l    is the vorticity vector

t  l2 
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- Algebraic models
- Damping function is applied to the turbulence
length scale in the near-wall region. Length scale
must be changed for the outer region of the
boundary-layer, wakes and jets
- Simplest model available, but with severe limitations.
Numerical implementation in complex flows may be
troublesome
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- 1-equation models (“old style”)

- Turbulence length scale is identical to that used
in the algebraic models
- Turbulence velocity scale is the square root of
the turbulence kinetic energy, which has its
own transport equation
________
1 2 2
k  u  v  w2
2
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Turbulence kinetic energy, k

- Transport equation (k balance)

Dk k k __ U i    1 __ 1 ___ 
 U j   u i u j   pu j  ui ui u j 
Dt t x j  x j  x j  2 

__ 2
 k
2  u 
  2   i 
x j  x 
 j 

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Turbulence kinetic energy, k

k
Uj  Convection
x j

__
 U i 
 ui u j  Production
 x 
 j 

 1
__ 1
___ 
  pu j  ui ui u j   Turbulent diffusion
x j  2 

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Turbulence kinetic energy, k

 2k
 2 Viscous diffusion
x j
__
2
 ui 
 
 x
 
 Dissipation rate, e
 j 

• Most of the terms include unknown quantities

and so they must be modeled

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- 1-equations models

Spalart & Allmaras  t  f v1~

~ ~ ~ 1  ~ 2

U V  cb1~S     ~ ~  cb 2 ~  ~   cw1 f w  
x y s d 
cb1 , cb 2 , cw1  Constants f v1 , f w  Functions

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- 1-equation model of Spalart & Allmaras

- Valid down to the wall

- Eddy-viscosity is proportional to the dependent

variable of the model

- It requires the distance to the wall,d, and in its

original version the location of transition
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- 2-equation models: turbulence velocity scale is k
k2
- k-e model  t  C
e
k k   t  
U V 2

  t S       k  e
x y  
  k  
e e e      e 2
U V  C1  t S 2       t e   C 2
x y k  e   k
C  , C1 , C2 ,  k ,  e  Constants
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- k-e model

- Widely used, specially in flows without walls and

in heat transfer problems

- Poor results for flows with adverse pressure

gradients

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- k-e model
- Can not be applied in the near-wall region
- Two-layer models combine the k-e model in the
outer region with a 1-equation model in the
near-wall region
- There are (too many) Low-Reynolds number versions
of the model for its extension to the near-wall region
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

k
- k- model t 

k k    
U V   t S      
2 t
k    *k
x y   k  
      F
U V  S      
2 t
  
   k       2

x y      
 * ,  ,  ,  k ,    Constants F  Function
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- k- model

- May be applied down to the wall

-  goes to infinity at the wall (smooth walls)

- Several formulations available. One of the most

popular is the SST (Shear-Stress Transport) version
that includes a limiter for the eddy-viscosity
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Eddy-viscosity turbulence models

- k- model

- Widely used for the calculation of adverse pressure

gradient flows

- Numerical implementation is not trivial (w wall

boundary condition) and some versions (as for
example SST) require the distance to the wall

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Boundary-layer approximations
U V
 0
x y
U U 1 dP 1    U __  
U V       uv  
x y  dx   y  y 

- Number of equations is smaller than the number

of unknowns __
- Only one Reynolds stress retained:-uv
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Boundary-layer approximations

- Assessment of negligible Reynolds stresses must

be based on experimental data
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Reynolds-averaged (RANS) equations

• Boundary-layer approximations

- von Kármán remains identical to laminar flow

h
1 h  T  T  w
 0
dy     
y   0 
U __
 T   lam   turb with  lam  ,  turb   uv
y

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall layer:

- Region of local equilibrium.

Production of k ≡ Dissipation of k (e

- At the wall, y=0, the momentum balance in the x

direction is reduced to
 T dP

y dx

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall layer:
- Assuming the convection is negligible close
to the wall (U0)

dP
T   w  y
dx

- For pressure gradients close to zero the wall

layer exhibits a constant total shear stress, T≈w
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall layer:
- The 3 fundamental variables (LMT) to define
dimensionless parameters are:

- Density of the fluid, 

- Viscosity of the fluid, 

- Shear-stress at the wall, w

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall layer:
- Friction velocity

w Cf
u   Ue
 2

- Reference length

Lref 
u

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Linear sub-layer:
- For very small values of y (-uv0)
U
 T   w   lam 
y

- Integration and the no slip condition (y=0 U=0)

lead to
 w y  w  y
U   
   

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Linear sub-layer:
- In dimensionless variables

U    y  u
u   
u     u 
U u y
  U   y
u 
U u y
U  y 
u 

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Linear sub-layer valid for y  5
- u=1m/s, air=1.5×10-5 y<7.5×10-5m

• Consequences:
- Experimentally, it is very hard to determine the
wall shear-stress from  U 
 y 
  y 0
- Numerically, the direct application of the no slip
condition requires near-wall grid line spacings
satisfying y2  1
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Linear sub-layer valid for y   5

- Spalart & Allmaras model

~   y  , ~   ~ 
- Turbulence kinetic energy, k
 
*
  0.5  0.25 6  
k  Ck y , k   k u2
- Turbulence “frequency”, 
 
   6  y   ,      u2
2

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Buffer-layer, 5  y   30  50

- In this region, the main contribution to the total

shear-stress changes from the laminar stress
to the Reynolds (turbulent) stress

- For y  5 the Reynolds stress (turbulent)
is negligible

- For y  30  50 the Reynolds stresses
(turbulent) the predominant contribution to the
total stress
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Buffer-layer, 5  y   30  50

- The region of the mean velocity profile with y+

smaller than 30-50 is the viscous sub-layer

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall (log) law, y   30  50

- Reynolds stress (turbulent) is predominant

- Dimensional analysis applied to the region with

approximately constant shear-stress
U u y
 f  
u   
- The velocity gradient is given by
U u2  u y  u u y  u y  u  u y 
 f '  f '   g 
y     y     y   
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall (log) law, y   30  50

- Experimental data leads to

u y 1
g     const 
   
U u
so 
y y
- Integration gives
U 1  u y 
 ln
u    
 1 
  C  U  ln y  C

 
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall (log) law, y   30  50

  0.41
C  5 .2

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall (log) law, y   30  50

- It is possible to determine the wall shear-stress

experimentally measuring the mean velocity profile
at a region sufficiently away from the wall

- The boundary conditions of a numerical calculation

may be applied at the wall (log) law region. This avoids
the viscous sub-layer and it simplifies significantly the
numerical calculation of near-wall flows. However, it
relies on the validity of the wall (log) law

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Turbulence quantities profiles

• Wall (log) law, y   30  50

- Spalart & Allmaras model

~   y  , ~   ~ 
- Turbulence kinetic energy, k
k 1 C , k   k u2

- Turbulence “frequency”, 
 
   1  C y  ,      u2

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall (log) law for rough walls

- Dimensional analysis applied to the region with

approximately constant shear-stress in the
fully-rough regime
U  y
 f  
u  er 
that leads to
U 1  y
 ln   B
u   e r 
  0.41 B  8.5
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall (log) law for rough walls

- Reynolds number based on the roughness height

ue
Ree   r
r

- For Ree  5 the flow behaves as if the wall is
smooth, “hidraulically smooth” regime (roughness
r

height smaller that the height of the linear sub-layer)

- Ree r  70 corresponds to the fully-rough regime. The

local flow becomes independent of the viscosity
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wall (log) law for rough walls

- For 5  Ree  70 the constant of the wall (log) law

depends on the roughness size and viscosity of the
r

fluid

U  y ue  U u y u e 
 f  ,  r  or  f  ,  r 
u  er   u    

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

1. Linear profile in the linear (laminar) sub-layer, y   5

U   y

2. Semi-logarithmic profile in wall (log) law,

y   30  50, y  0.1  0.2


U 
1

 
ln y   C k  0.41 C  5.2

3. Continuous shift from 1 to 2 in the buffer-layer

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

3. Continuous shift from 1 to 2 in the buffer-layer

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

4. Outer part of the profile where the velocity tends

to U e defined from the difference to the log-law

 Ue 2
U 
e 
u Cf

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wake component
   y
1
 
U   U    ln y   C   w 
   '
• Coles (empirical) wake profile
 y  y
w   1  cos  
'  '
• y=’ is the location where the maximun difference
to the log law occurs

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Mean velocity profile, U

• Wake component magnitude/intensity

 
U max 2

• Mean velocity profile outside the viscous sub-layer

  y 



1
 
U  ln y  C  1  cos   


   ' 

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Self-preserved flows

• von Kármán’s integral equation

d
dx
 
U e    w  
2 * dP

dx
• History/memory parameter
 * dP

 w dx
• Self-preserved flow, =constant

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Self-preserved flows

• Clauser’s equilibrium parameter

2
U e  U 
h

2 H 1 0  u  dy
G    
Cf H h U  U 
0  eu  dy
• G is constant for zero pressure gradient and
sufficiently large Reynolds number
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Simplified forms of the mean velocity profile

• Power-law profile
1
U  y n
 
Ue   

• Integral parameters

* 1  n 2
  H  1
 n  1  n  1n  2 n

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Simplified forms of the mean velocity profile
1

• Power-law profile U  y n
 
Ue   
1. It does not safisty the linear sub-layer

2. It does not satisfy the wall (log) law

U
3.  0 at y  
y
4. U   at y  0
y

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Simplified forms of the mean velocity profile
1

• Power-law profile U  y n
 
Ue   
- From experimental data:
dP
0n7
dx
dP
 0  n  7  10
dx
dP
 0  n  37
dx

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Zero pressure gradient (flat plate) boundary-layer

• Integration of von Kármán’s integral equation

d 
 w2
dx U e
• Power-law profile 1
U  y 7
 
Ue   
• w from a friction law for fully-developed flow in pipes
4 w 1 / 4 U med D
  0 . 3164 R R 
1 2 U med 
2 e e

U med  0.8U max   R U e  U max

Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Zero pressure gradient (flat plate) boundary-layer

• Shear-stress at the wall

1
w    4
 0 .0225 
U e2
 U e 
• von Kármán’s integral equation with  as the
dependent variable
1
7 d    4
 0.0225 
72 dx  U e 

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Zero pressure gradient (flat plate) boundary-layer

• Assuming turbulent flow from x=0 (=0)

 1 * 1
 0.37 R
ex
5
 0.046 R
ex
5
x x
 1
 0.036 R ex
5
H  1.29
x
1 1
C f  0.0576 R ex
5
C D  0.072 R ex
5

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Effect of the pressure gradient

• Qualitatively, the effect is similar to that discussed

previously for laminar flow

• H decreases for favourable pressure gradient and it

increases for adverse pressure gradient
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Effect of the pressure gradient
• Entrainment is strongly dependent on the gradients of
the mean velocity profile in the outer region of the
boundary-layer (production of turbulence kinetic energy
is proportional to the gradients of the mean velocity)

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Effect of the pressure gradient

•Effect on the wall (log) law

• Validity of wall (log) law at flow separation is extremely

doubtful
Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Effect of the pressure gradient

• Turbulent flow is significantly more resistent to flow

separation than laminar flow.

1. Velocity close to the wall is larger than in laminar

flow

2. Diffusion is significantly higher than in laminer

flow (separation depends on the ratio between
the pressure force and diffusion)

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Head’s method

• von Kármán’s integral equation

d H  2 dU e C f
 
dx U e dx 2
• Entrainment velocity, VE

VE 
d 
dx 0
Udy 
d
dx
 
U e    *

• Shape parameter, H1
  *
H1 

Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Head’s method

• Additional equation proposed

d
U eH1   U e F H1  H1  G ( H )
dx
• Experimental fit to determine F(H1) and G(H)

F H1   0.0306H1  3
0.6169

0.8234H  1.11.287  3.3  H  1.6

H1  G ( H )  
1.5501( H  0.6778) 3.064  3.3  H  1.6

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Head’s method

• Ludwieg-Tillman correlation to determine Cf

C f  0.246 10 0.678 H  Re0.268

• Flow separation predicted for H  2.4  2.8

• Numerical solution with a Runge-Kutta scheme

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Boundary-layer control

• Forced transition: roughness or trip wire

The objective is to delay or avoid flow separation

U e d arame
Rearame   826 Gibbings’s criterion

Masters of Mechanical Engineering
 Aerodynamics

Turbulent Flow
Boundary-layer control

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Boundary-layer control
• Suction at the wall. Delays (or avoids) flow separation
and it also delays transition from laminar to turbulent flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Boundary-layer control
• Blowing. Delays or avoids flow separation, but it favours
transition from laminar to turbulent flow

Masters of Mechanical Engineering

 Aerodynamics

Turbulent Flow
Boundary-layer control

Masters of Mechanical Engineering