Professional Documents
Culture Documents
http://www.iup.uni-heidelberg.de/institut/studium/lehre/MKEP4/
8. Turbulence
1
Contents of Today's Lecture
Turbulence
• The phenomenon of turbulence
• Reynolds number
• Reynolds decomposition
• Turbulent kinetic energy and the turbulence spectrum
• Kolmogorov theory
• Autocorrelation
• Taylor's theorem and turbulent diffusion
Literature on Turbulence:
1) Roedel, W., 2000. Physik unserer Umwelt, Die Atmosphäre. Springer Verlag,
Heidelberg, 3rd Edition. (IUP 1780). (Kap. 6)
2) Pedlosky, J., 1987. Geophysical Fluid Dynamics. Springer Verlag, Heidelberg, 2nd
Edition. (IUP 720)
3
2
Laminar and turbulent flow
3
Deterministic Chaos - Edward Lorenz
Lorenz-attractor in 3D phase space
Edward Lorenz
1917 - 2008
Heisenberg was asked what he would like to know from God, given the
opportunity. His reply was:
"When I meet God, I am going to ask him two questions: Why relativity?
And why turbulence? I really believe he will have an answer for the first."
4
Criterion for Turbulence: Reynolds Number
Reynolds number: Ratio of turbulence producing non-linear
term to turbulence destroying friction term
Re
v v U2 L UL
v U L2
Re > Rec ≈ several 1000: Turbulent flow
Atmosphere, ocean, lakes are usually turbulent:
air water
viscosity 10-5 m2/s 10-6 m2/s
Osborne Reynolds velocity U 10 m/s 0.1 m/s
1842-1912 length scale L 1000 m 1000 m
5
Examples for Turbulent Flow
Top: A mixing layer at high Reynolds number. The upper stream is moving at 100 m/s and the lower at 38 m/s, both from
left to right in the image. Two observations: (1) the transition to turbulence is evident on the left side of the image, where
the initially smooth roller structures suddenly develop small-scale detail, and (2) the large-scale organization of the flow is
evident as the flow moves downstream, even though the flow has a lot of small-scale activity. Bottom: The same flow
arrangement as above, but at twice the Reynolds number. There appears to be more small-scale activity, but the large-scale
organization is not greatly affected.
(Images from van Dyke, An Album of Fluid Motion.)
11
r
x
Eulerian measurement
vy
of transverse velocity
T 2r 4r
T
2 v v
x
t
v
12
6
Velocity Fluctuations due to Eddies
Deviation of velocity from mean in a wind channel
big eddy
intermed. eddy
small eddy
from:
Frisch 1995,
Turbulence
13
Turbulent Fluctuations
Laminar flow (stationary): At each point the velocity of flow
is constant in time.
Turbulent flow (stationary): Statistical variations of the flow
velocity due to eddies.
Description by fluctuations in flow velocity, i.e. deviations from the average
velocity:
v '(t)
v(t)
Time
14
7
Reynolds Decomposition
Method to analyse turbulence and to parameterise the non-
linear terms in the Navier-Stokes equation
Basic idea: Separate velocity components into a mean
value and (statistical) fluctuations around this mean
T
1
v(t) v(t) v '(t) with v(t) v(t)dt
T0
v : Mean velocity
v ' : Turbulent velocity fluctuations,
fluctuations "turbulent
turbulent velocities"
velocities
From the definition follows: v ' 0
Description of turbulence by statistical properties of v '(t)
In the following, v denotes one component of the velocity
15
Inverse transform:
F e
2 i t
v '(t) d
16
8
Turbulent Kinetic Energy (TKE)
From the separation of the flow field in a mean flow and
statistical fluctuations follows:
Emean 1
Kinetic energy density of the mean flow: kin
v 2
[J/m3] V 2
turb
Ekin 1
Kinetic energy density of the turbulent flow: v 2
[J/m3] V 2
1 2
Def.: Turbulent kinetic energy (TKE): TKE v
[J/kg = m2/s2] 2
F
2
dE F
2
f and
T v'² d T
T is the averaging time.
18
9
Turbulent Energy Cascade (Richardson)
Turbulent energy
generation at
large scales
Transport of
turbulent energy
down the eddy
scales
Dissipation of
turbulent energy
at small scales Lewis Fry Richardson
due to viscosity (1881 - 1953)
Frisch 1995, Turbulence, Cambridge Univ. Press
Richardson 1922:
Big whirls have little whirls that feed on their velocity,
and little whirls have lesser whirls and so on to viscosity.
19
10
Kolmogorov Scales of Turbulence
Turbulence at high Re and small scales is determined by
1) Kinematic viscosity of the fluid
2) The rate of energy dissipation in the turbulent flow
Idea: Derive by dimensional analysis the universal, smallest scales
off turbulent
t b l t motion
ti from
f andd .
1
1
3 4 L2 3 L2 1 L6 T 3 4
length scale LK: LK 3 3 2 L4
T T T L
1
1
2 L2 L 2
1
L T 2 2 3
1
1.5 10 5 3
4
3 4
LK 2.5 10 4 m 0.25 mm
1
1 1
2 1.5 10 2
5
3
tK 4 10 s
1
1
1
m
v K LK tK 4 1.5 10 5 1 4 0.06
s
22
11
Kolmogorov's Energy Spectrum
In the inertial subrange, the energy density spectrum should depend
only on the length scale (LI) and the rate of energy dissipation ().
The wavenumber k is related to the length scale by k = 2/LI.
Dimensional analysis yields the shape of the power spectrum E(k):
L2 1 d TKE L3
k E k E k T 2
T3 L dk
L3 L2a 1 2 5
E k c a k b a ,b
T 2 T 3a Lb 3 3
It follows: E k c 2 3 k 5 3
23
Turbulence Spectrum
Energy
input
E(k)
Inertial
subrange
Dissi-
pation
24
12
Autocorrelation of v'(t)
v t v t 1
v t
v x v t 2
t, x
Sh t times
Short ti : v t v t
25
Autocorrelation Functions
Lagrange's Autocorrelation Function:
Observer moves with fluid parcel and determines velocity fluctuations. The
autocorrelation is determined by spatially averaging over many parcels:
v '(t) v '(t )
R L ( )
v ' 2 (t)
v '(t) v '(t )
R E ( )
v ' 2 (t)
26
13
Eulerian Autocorrelation
RE f()e
2 i
d
R e
2 i
f( ) E d
28
14
The Theorem of Taylor
How far do fluid parcels move due to turbulence? Over which distance
does turbulence mix constituents (e.g., dissolved conc.) of the fluid?
1-D: We use a (moving) coordinate system where v x v x 0
i.e. dx/dt = v = vv',, and x(t
x(t=0)
0) = 0
Since fluctuations v' are statistical, the mean displacement is 0: x t 0
But not the variance: 2x x 2 t 0
Turbulence produces displacement x x 2 analogous to diffusion!
The theorem of Taylor connects the variance of the displacement of a
fluid parcel with the Lagrangian Autocorrelation Function:
d 2x d 2 d 2 d
We have: x (t) x (t) 2x(t) x(t) 2 x(t) v ' x (t)
dt dt dt dt
t
with x t v ' x t ' dt '
0
29
0
v ' x (t) v ' x (t ) t' t
2 v ' 2x (t)
t v ' 2x (t)
d
dt ' d
t t
v ' x (t) v ' x (t )
2 v ' 2x (t)
0 v ' 2x (t)
d 2 v ' 2x (t) R
0
L,x ( ) d
t t'
2x (t) 2 v ' 2x (t) R
0 0
L,x ( ) d dt '
30
15
The Theorem of Taylor - Discussion
Taylor theorem:
t t'
2x ((t)) 2 v ' 2x R L,x
L x ( ) d dt '
0 0
t'
R L,x d L const.
const L: Lagrangian time scale
0
t 1 2
t L R L 1 1 ddt ' t ' dt ' t x (t) v ' 2 t
0 2
t
t L R L 0 R L,x ddt ' L dt ' L t x (t) 2 v ' 2 L t
0
31
Turbulent „Diffusion“
Molecular diffusion (see lecture 4):
1 1
2 (t) 2Dt t 2Dt, D v G v 2
3 3
Turbulent diffusion: mixing length
l' = v' λL.
1 d2 t L K x ( t ) v ' 2x t
K
2 dt t L K x ( t ) v ' 2x L v 'x l'x
16
Molecular versus Turbulent Diffusion
Molecular Diffusion Turbulent Diffusion
t = x/<vx>
Molecular diffusion:
σ2 = 2/3 v t with v: thermal velocity; : mean free path
Turbulent diffusion:
v ' t for t L
2 2
2
2 v ' L t for t L
2
In analogy to molecular diffusion, define mixing length l' = v' λL. For large
t, this yields: 2
2 v ' 2 L t 2 v ' 2 L t 2 v ' v ' L t 2 v 'l' t
Typical values for Lagrangian scale time in case of exhaust from a chimney:
λL ≈ 15-30 min.
33
Summary
• Turbulence is a result of the instability of fluid flow brought about by the
non-linearity of the Navier-Stokes equation
• Turbulence leads to chaotic, unpredictable behaviour (determin. chaos)
• Turbulence occurs for large Reynolds numbers
• Turbulent energy is passed from larger to smaller eddies, until molecular
friction dominates and energy is dissipated (turbulent energy cascade)
• Reynolds decomposition: Mean and fluctuations of velocity
• Important concepts for a statistical description of turbulent fluctuations:
– Power spectrum (energy density of turbulence)
– Autocorrelation of velocity fluctuations
• Turbulence produces statistical displacements and hence transport:
turbulent diffusion
• For sufficiently large diffusion times, turbulent diffusion is analogous to
molecular diffusion (but with different diffusion coefficients)
34
17