TURBULENT MIXING 337

Chair of Aerodynamics and Fluid Mechanics

TUM Department of Mechanical Engineering
Technical University of Munich

TurbulentFlows Lecture3
Isotropic Turbulence

Annu. Rev. Fluid Mech. 2005.37:329-356. Downloaded from www.annualreviews.org

by WIB6105 - Technische U. Muenchen on 03/23/12. For personal use only.
TURBULENT FLOWS
1. Introduction
Figure 2 Liquid-phase turbulent-jet symmetry-plane slices of passive scalar field. Grey
1.2 Isotropic turbulence scale (contrast-enhanced) codes jet-fluid mole fraction. Left: Re ≃ 2.5 × 103 , right: Re ≃
1.0 × 104 (Dimotakis et al. 1983).

origins must then be sought in the dynamics over the internal range of scales, i.e.,
for scales λ, such that λK ≪ λ ≪ δ.
The transition coincides with when energy spectra start to osculate an approxi-
mately −5/3 logarithmic slope with increasing Re, that may, in turn, be identified
with the emergence of an inertial inviscid range of eddies. This suggests that
the post-transition regime requires a sufficient scale separation to support quasi-
inviscid dynamics. Expanding on the H. Liepmann idea on the similarity between
the Taylor scale and the viscous-layer scale dependence on Re, we define the
Liepmann-Taylor scale, λLT , as a laminar scale,
λLT = 5.0 δ Re−1/2 , (13)

where the numerical prefactor corresponds to the thickness of an internal laminar

PD Dr.-Ing. habil. Stefan Adami shear layer developing across the δ extent of the flow, as well as for a Blasius
boundary layer (Dimotakis 2000). Because viscous effects span from the outer
scale, δ, to λLT and influence inner scales below λν ≈ 50 λK , inviscid dynamics
require, at a minimum, room for λ-size eddies such that,
λLT 1/4 4
δ > λLT > λ > λν > λK ⇒ ! 0.1 Re ! 1 ⇒ Re ! 10 , (14)
λν
May 2nd, 2024 | 1. Introduction to turbulent flows in accord with observation. 1
1.2 Isotropic turbulence 10 Nov 2004 13:11 AR AR235-FL37-13.tex AR235-FL37-13.sgm LaTeX2e(2002/01/18) P1: IBD

REVIEW
Reynolds decomposition TURBULENT MIXING 337

)( *, , = )( *, , + )′( *, ,

Annu. Rev. Fluid Mech. 2005.37:329-356. Downloaded from www.annualreviews.org

Turbulent mixing ↔ Molecular diffusion

by WIB6105 - Technische U. Muenchen on 03/23/12. For personal use only.

!! #&#$% /% & ' ##$%
= = = '(
!" #/& ' % Figure 2 Liquid-phase turbulent-jet symmetry-plane slices of passive scalar field. Grey
scale (contrast-enhanced) codes jet-fluid mole fraction. Left: Re ≃ 2.5 × 103 , right: Re ≃
1.0 × 104 (Dimotakis et al. 1983).

Homogeneity origins must then be sought in the dynamics over the internal range of scales, i.e.,
for scales λ, such that λK ≪ λ ≪ δ.
The transition coincides with when energy spectra start to osculate an approxi-
• Statistical quantities invariant under translation mately −5/3 logarithmic slope with increasing Re, that may, in turn, be identified
with the emergence of an inertial inviscid range of eddies. This suggests that
the post-transition regime requires a sufficient scale separation to support quasi-
inviscid dynamics. Expanding on the H. Liepmann idea on the similarity between
the Taylor scale and the viscous-layer scale dependence on Re, we define the

Isotropy
Liepmann-Taylor scale, λLT , as a laminar scale,
λLT = 5.0 δ Re−1/2 , (13)

• Statistical quantities invariant under reflection and rotation where the numerical prefactor corresponds to the thickness of an internal laminar
shear layer developing across the δ extent of the flow, as well as for a Blasius
boundary layer (Dimotakis 2000). Because viscous effects span from the outer
scale, δ, to λLT and influence inner scales below λν ≈ 50 λK , inviscid dynamics
require, at a minimum, room for λ-size eddies such that,

May 2nd, 2024 | 1. Introduction to turbulent flows δ > λLT > λ > λν > λK ⇒
λLT
λν
! 0.1 Re
1/4 4
! 1 ⇒ Re ! 10 , (14) 7
 Short answer is No

It includes vortices

158 9 moving outer part

B 1
Moving
inner part
 1m or 10m behind the inlet if theflow is homogeniers you
see result
can measure those pants and get the

net effecting statistial

result
Direction
1.2 Isotropic turbulence
TABLE OF CONTENTS
1. Introduction to turbulent flows
1. Statistical description
2. Isotropic turbulence
3. Kolmogorov hypothesis
4. Spectal energy transfer
5. Transition
6. Transport equation of turbulent flows

2. Canonical flows
1. Round jet
2. Turbulent channel flow
3. Turbulent boundary layer

3. Statistical turbulence modeling

1. Turbulent-viscosity hypothesis
2. Algebraic turbulence models
3. One- and Two-equation models
4. Wall modeling

May 2nd, 2024 | 1. Introduction to turbulent flows 8

1.2 Isotropic turbulence
GENERATION OF HOMOGENEOUS ISOTROPIC TURBULENCE
Advection trough a homogeneous lattice generates homogeneous, isotropic turbulence (HIT):

Homogeneous isotropic turbulence structures are observed downstream the lattice for…

• … a reasonably large distance

• … a sufficiently small domain moving with the average flow velocity ⟨"⟩ (Taylor Hypothesis)

I
May 2nd, 2024 | 1. Introduction to turbulent flows 9
 men
1.2 Isotropic turbulence
TAYLOR-HYPOTHESISavg.ae

given

point
its

savoritise

May 2nd, 2024 | 1. Introduction to turbulent flows 11

 fluctuations fated
are moderefble.ee
1.2 Isotropic turbulence matters
TWO-POINT CORRELATION any

z
f0 ! " + $, &

$
$

38

"+
matters
" only
y ! ", & distance

x dueto
isotropy
Generalization of the two-point correlcation for HIT
• Arbitrary vector $ denotes principle direction
• Spatial correlation of two velocity-vector compontents at two points with distance |$|

Two-point correlation (Complete) Statistical description of homogeneous, isotropic turbulence

May 2nd, 2024 | 1. Introduction to turbulent flows 12

1.2 Isotropic turbulence
TWO-POINT CORRELATION
Longitudinal two-point correlation

%# "!" !, # $!" , + $, '

,# $! ! + ', # %!
' % $, ' = )!! ($, ') =
!
+ "! ′"! ′
! " + $, &
$#
z #
%# $
!
!, !,
$

$ # Lateral / transverse two-point correlation

"+

$# %
!
"#" ,, ' "#" , + $, '
" / $, ' = )## $, ' =
"# ′"# ′
y
! ", &
x Continuity (divergence-free velocity field)

$ 1%
/ $ =% $ +
2 1$

May 2nd, 2024 | 1. Introduction to turbulent flows 14

1.2 Isotropic turbulence
TAYLOR MICROSCALE
Spatial profiles of f(r) and g(r):
r 15M
(for lattice-generated, isotropic turbulence (mesh size M)
measure

Integral lengthscales quantifiestypical

h * *
/% = 0 1(3, ,)53 , /+ = 0 6(3, ,)53
) )

y Taylor microscales 2$ and 2%

7+ = − &
2
; 6 3, , /;3 & |#,)
.-
&

Fistence
May 2nd, 2024 | 1. Introduction to turbulent flows 18
1.2 Isotropic turbulence
TAYLOR MICROSCALE
Direct connection between the Taylor microscale and the dissipation rate in isotropic turbulence:

• Dissipation rate: ! ()!" ()#"

3&' = +
! = 2% )*+ )*+ # (*# (*!
Symmetric, deviatoric rate-of-strain tensor

(
• Taylor microscale: &'
! = 15%
') (

May 2nd, 2024 | 1. Introduction to turbulent flows 19

1.2 Isotropic turbulence
FOURIER TRANSFORMATION
Statistical averaging (ensemble) can be replaced with spatial averaging (due to homogeneity)

Fourier transformation of velocity fluctuations uj(x,t)

*
• Spectral averaging 0(1·3
)</ (=, ,) = 0 )/ *, , ( 5*
• Statistical means as function of
0*
spatial scales (wave modes) !

Exercise…

FFT FFT-1

Temporal signal (physical space) Fourier coefficients (phase space) Temporal signal (physical space)

May 2nd, 2024 | 1. Introduction to turbulent flows 20

1.2 Isotropic turbulence
FOURIER TRANSFORMATION
Statistical averaging (ensemble) can be replaced with spatial averaging (due to homogeneity)

Fourier transformation of velocity fluctuations uj(x,t)

*
• Spectral averaging 0(1·3
)</ (=, ,) = 0 )/ *, , ( 5*
• Statistical means as function of
0*
spatial scales (wave modes) !

Exercise…

FFT FFT-1

Cut off high frequencies

Temporal signal (physical space) Fourier coefficients (phase space) Temporal signal (physical space)

May 2nd, 2024 | 1. Introduction to turbulent flows 21

1.2 Isotropic turbulence
FOURIER TRANSFORMATION
Statistical averaging (ensemble) can be replaced with spatial averaging (due to homogeneity)

Fourier transformation of velocity fluctuations uj(x,t)

*
• Spectral averaging 0(1·3
)</ (=, ,) = 0 )/ *, , ( 5*
• Statistical means as function of
0*
spatial scales (wave modes) !

Exercise…

FFT FFT-1

„Low-pass filter

Temporal signal (physical space) Fourier coefficients (phase space) Temporal signal (physical space)

May 2nd, 2024 | 1. Introduction to turbulent flows 22

1.2 Isotropic turbulence
FOURIER TRANSFORMATION
Statistical averaging (ensemble) can be replaced with spatial averaging (due to homogeneity)

Fourier transformation of velocity fluctuations uj(x,t)

*
• Spectral averaging 0(1·3
)</ (=, ,) = 0 )/ *, , ( 5*
• Statistical means as function of
0*
spatial scales (wave modes) !

1
Spectral kinetic energy ?> =, , = )</ (=, ,) A )</ ∗ (=, ,)
2

May 2nd, 2024 | 1. Introduction to turbulent flows 23

1.2 Isotropic turbulence k
3
FOURIER TRANSFORMATION
Isotropy simplifies the wavenumber k= (k1,k2,k3) in spherical coordinates to k k
z 2

5=. 5=& 5=5 = = & BCDE

5E 5G 5= y
= H 0, ∞ , G H 0, K , E H [0,2K]
x

k
3D Energy spectra Ê(k,t) for a shell of width k=||k||: 1

1 & &6 6 1
?> =, , = = 0 0 )</ )</ ∗ BCDE 5E 5G = '>// = ($% = $$" $%"
2 ) ) 2
Remember Parseval‘s theorem:
• Gesamte Energie im Fourier-Dualraum entspricht gesamter Energie im Realraum
0
1 3 # 1
4 ' = "'" "'" ' = "+,- ' = .8 :9 ;, ' 1;
2 2 27 /

May 2nd, 2024 | 1. Introduction to turbulent flows 24

α
α
α

None
Related
with

opposite

NO
 %#
1.2 Isotropic turbulence !
+
' ,#
$! ! + ', # %!
LEARNING TARGETS $# ! " + $, &
z !,
#
%#
$
!
!,
$ $# #

+$
Homogeneous isotropic turbulence

"
%
!
"
• Invariant under reflection and rotation (isotropy) y ! ", &
• Invariant under translation (homogeneity) x
• Taylor hypothesis
• Two-point correlation fully describes the flow field
• Length scale approximations
• Connection between dissipation and Taylor microscale

May 2nd, 2024 | 1. Introduction to turbulent flows 30