2/2/2016

Scales of Turbulence
Course: Turbulence (AM2205)

Akshoy Ranjan Paul

Department of Applied Mechanics
MNNIT Allahabad

You are a fluid dynamicist visiting the

famous Louvre Museum in Paris and are
asked by the Curator to comment on the
paintings below. What do you say?

Painting by Van Gogh

Sketch by Leonardo da Vinci

2/2/2016

Non turbulent flow, Van Goghs clouds have

no small scales!
Turbulent illustrates by this
sketch of a free water jet issuing
from a square hole into a pool

The world's first use of visualization as a scientific tool

to study turbulence
thus the water has eddying motions, one
part of which is due to the principal current,
the other to the random and reverse
motion.
L. da Vinci

Leonardo da Vinci

Da Vinci provided the earliest reference to

the importance of vortices in fluid motion:
Finally, da Vinci's words
"... The small eddies are almost numberless,
and large things are rotated only by large
eddies and not by small ones, and small
things are turned by both small eddies and
large .."
Confirmed in Richardson's energy cascade,
which is coherent structures, and large-eddy
simulations, at least.

2/2/2016

Demonstrated by an experiment first reported

by O. Reynolds (1883)
Flow inside a pipe becomes turbulent every time
a single parameter Re would increase

Dye injected on the centerline

No change in time, streamlines
// pipe axis

Flowing water

Re >2300, turbulent
Occurrence of small scales.
Generated by the inertial forces
and dissipated by the viscous forces.

Re=UaxialD/n

From laminar to turbulent flow

2D cylinder (Williamson 1996)

Dynamics of large scale structures

Hydrodynamic stability (cf. lecture F. Gallaire) explains

how structures of a specific frequency and scale are
selected and emerge.

2/2/2016

From Laminar to Turbulent flow

Turbulent flow: Large-scale structures + small-scale turbulence

Turbulence in a Box
Lets consider a periodic box:
u(0)=u(L)

Possible to consider a Fourier expansion

where,

2/2/2016

Properties of Fourier Series

It is a way to represent a wave-like function as
the sum of simple sine waves.
If u(x) is real then
Distinct Fourier modes are orthogonal
Correlations (spatially homogenous):

Fourier Decomposition in 3D
In 3D space,
Physical interpretation:

Energy:

2/2/2016

Energy Spectrum
The energy spectrum can be described as average
energy of eddies plotted against their size. This is energy
) versus wave number (k), which is defined as
density (E
k = 2/.
Wavelength () characterizes eddy size. Small
wavelength corresponds to large k. Smaller eddies have
low energy; hence, the spectrum tapers off rapidly at
high k.

11

Energy Spectrum

Figures show Energy spectra in Fourier space. Energy

is normally plotted against one over the size, as
above. The axes are scaled arbitrarily.
12

2/2/2016

Energy Spectrum
A plot of the vorticity spectrum would go the other way:
small eddies have high vorticity at least until some very
small size is reached, below which there is little
turbulent motion.
The rate of energy dissipation goes like vorticity: it is
largest at small scales. That is because vorticity and
dissipation involve derivatives of velocity. (The rate of
energy dissipation is |u |2 . Differentiation amplifies
at small scales.)

13

Energy Spectrum: A Summary

This is the picture that emerges: On average, the energy is
highest at large scales but unstable, whereas the rate of
dissipation peaks at small scales. We are led to the
concept of an energy cascade: to achieve an equilibrium,
energy must flow from the large scales, where it is
concentrated, to the small scales, where it is being
dissipated. That is what is meant by the energy cascade.
Large eddies might develop instabilities that produce
smaller and smaller scales of motion or straining might
stretch and fold large-scale vortices to create smallerscale structure.

14

2/2/2016

Energy Balance for Forced Turbulence

Stirring can be considered as forcing at small wave
numbers (large scales).
Spoon is much larger than smallest eddies.
u
u u p 2 u f ( x , t )

t
stirring with spoon

f ( x, t )

f ( k , t )d f

2
L

Eqn. for average turbulent kinetic energy: k 12 u u

k

:
u
f u
u

t
=Dissipation rate
Power input

Energy Balance for Forced Turbulence

After statistical stationarity is reached (stirring for a
long time)
k
0
f u

:
u
u

t
=Dissipation rate
Power input
For constant power input:
Decreasing Increasing u : u
At high Reynolds number (Re) flow field has higher
gradients.

2/2/2016

Turbulent Eddy Distribution

Self Similar Energy Cascade

2/2/2016

Time scales in Energy Cascade

Energy Cascade in Turbulence

Kinetic energy enters the turbulence at the largest scales at
which energy is extracted from the mean flow. This energy is then
transferred to smaller and smaller scales.
At the smallest scales, the turbulent kinetic energy is dissipated
into heat due to viscous stresses.

20

10

2/2/2016

Energy Cascade in Turbulence

The idea of energy transfer from large to subsequently smaller scales
was introduced by Lewis F. Richardson in 1922.
Energy is transferred from the large to the small scales: that is a central
tenet of turbulence theory. It is also an underpinning of all types of eddy
simulation.
Large eddies grind down into smaller and smaller ones, until energy is
dissipation by viscous action at the smallest scales.

Big whorls have little whorls,

little whorls have smaller whorls,
that feed on their velocity,
and so on to viscosity...

21

Scales of Turbulence
Integral scales: Scale for largest eddies.
Dissipation/Taylor scales: Scale at which
turbulence is isotropic.
Kolmogorov scales: Scale for smallest eddies
(after Kolmogorov, 1941).
Characteristic length scale ( Lc )
Characteristic time scale ( tc )
Characteristic mean velocity (Uc )
For the characteristic time, tc , the following ratio
holds:
22

11

2/2/2016

Scales of Turbulence
Log E

Integral
scale

Kolmogorov scale
Taylor
scale

Wave number (k)

23

Length Scales of Turbulence

Length scales:

For largest eddies, Integral length scale is (k3/2/e).

At which turbulence is isotropic, Taylor microscale is
(15nu2/e)1/2.
For smallest eddies, Kolmogorov length scale is
(n3/e)1/4. These eddies have a velocity scale (n.e)1/4
and a time scale (n/e)1/2.
is the energy dissipation rate (m 2 /s3 )
k is the turbulent kinetic energy (m 2 /s 2 )

is the kinematic viscosity coefficient (m 2 /s)

24

12

2/2/2016

Length Scales of Turbulence

The Taylor length scale proves always to be larger
than the Kolmogorov micro length, and the
difference between the two becomes larger with
increasing Reynolds number. From the above
relationships, one can compute
14

T k Re )

For a Reynolds number of approximately Re=104,

T is
approximately 10 times larger than
k (Following fig.).
25

Microstructure of Turbulence

26

13

2/2/2016

Jet at two Reynolds Numbers

Source: Tennekes &
Lumley. Page 22.

Integral Scale vs. Characteristic Scale

If one considers the turbulent velocity
fluctuations that occur superimposed on the
mean flow field, it is easy to see that the integral
time-scale of the turbulence always has to be of
the order of magnitude of the characteristic time
scale of the mean flow field, i.e. the largest
vortices that the turbulent part of a flow field
possesses have time scales which correspond to
those of the mean flow field. Generally, the
following relationship holds:
28

14

2/2/2016

Integral Scale
For integral length scale l0,
Here, is the energy dissipation rate. The
proportionality constant is of the order one.
The Reynolds number associated with these large
eddies is referred to as the turbulence Reynolds
number ReL, which is defined as:

Taylor Micro-scale
From these scales, characterizing the smallest vortices of
a turbulent flow, the Taylor micro-scale has to be
distinguished, which is defined as follows:

Taylor micro-scale defines an eddy size which is located

between the smallest viscous eddies and the large eddies
having quantities of the dimension of the geometric
extension of the mean flow. Taking this into account, it
can be shown that the following hold:

30

15

2/2/2016

Taylor Micro-scale vs. Characteristic Scale

Considering that the following holds:

Then inserting this relationship into the

equation for
and taking into account
, a further important relationship follows
from the above derivations:

31

Turbulent Length Scales

32

16

2/2/2016

Production vs. Dissipation Term

The ratio of production term to the dissipation term of the
general turbulence flow energy equation is

where,
represents the square of the
corresponding turbulence intensity (Tu).
As turbulence always occurs for large Reynolds numbers,
e.g. Re = 104, above equation shows that even for Tu =
20%, a comparatively large degree of turbulence, the
viscous dissipation is negligible compared with the
turbulence production.
33

Kolmogorov Hypothesis
Kolmogorovs hypothesis of local isotropy states that at
sufficiently high Reynolds numbers, the small-scale turbulent
motions (l << Lc) are statistically isotropic.
Kolmogorovs first similarity hypothesis states that in every
turbulent flow at sufficiently high Reynolds number, the
statistics of the small scale motions (l < LEI ) have a universal
form that is uniquely determined by and .
Kolmogorovs second similarity hypothesis states that in
every turbulent flow at sufficiently high Reynolds number,
the statistics of the motions of scale l in the range Lc >> l >>
l have a universal form that is uniquely determined by
independent of .
Note: LEI is the length scale that forms the demarcation between
the large scale anisotropic eddies. Lc is the characteristic or
Integral length scale. l is the Kolmorogov length scale.

17

2/2/2016

Kolmogorov Scale
Following a suggestion of Kolmogorov, so-called microscales can be introduced to characterize the turbulent
flow field:

The length, velocity and time scales introduced by

Kolmogorov are determined in such a way that they
characterize that part of the spectrum of the turbulent
velocity fluctuations in which the energy production of
the turbulent vortices is equal to the dissipation. Thus,
assuming isotropic turbulence, one can introduce:
35

Kolmogorov Scale

From these results, we can deduce:

From the equality of the terms for production

and dissipation, it follows that:

36

18

2/2/2016

Kolmogorov Scale
For the Kolmogorov time scale, the following
expression results:
The Reynolds number resulting on the basis of the
above-introduced micro-length scale and microvelocity scale is:
Note that the fact that the Kolmogorov Reynolds
number Re of the small eddies is 1, is consistent
with the notion that the cascade proceeds to smaller
and smaller scales until the Reynolds number is small
enough for dissipation to be effective.
37

Kolmogorov Scale
The characteristic turbulent eddy quantities,
determined by Kolmogorovs scales of turbulence,
are those that represent the viscous effects which
damp the turbulent velocity fluctuations.
These smallest eddies are assumed to convert the
kinetic energy of turbulence into heat. Because of
these characteristic properties, the following
definitions are available in the literature for the
smallest scales of turbulence:
Kolmogorov scales = micro-scales = viscous eddy
scales:
38

19

2/2/2016

Turbulent Scale: A Summary

The different length scales of turbulence have
proven to be very useful in formulations of
turbulence models, which are summarized in the
subsequent sections.
The above derivations indicate the differences in
the structure of turbulent flows at small and high
Reynolds numbers.
For flows with the same integral dimensions, the
flow at a large Reynolds number proves to be
micro-structured, i.e. the smallest eddies have
small dimensions, whereas for the small Reynolds
number the flow appears macro-structured.
39

In order to characterize the complex nature of turbulent

flows, the following Reynolds numbers are often
employed:

Moreover, for the relationships of the characteristic length

scales, the following expression holds:

These relationships are often employed when considering

turbulent flows, in order to carry out order of magnitude
considerations regarding the characteristic properties of
turbulence.
40

20

2/2/2016

Correlations for various Turbulence Scales

Here large scale refers integral scales and
small scale is Kolmogorov scale.

41

Ratio between large scale and small scale

Ratio of Integral to Kolmogorov Length scales:

l l0 ~ Re L 3 4
Ratio of Integral to Kolmogorov velocity scales:

u u0 ~ Re L 1 4
Ratio of Integral to Kolmogorov Length scales:

0 ~ Re L 1 2
Note: As expected, at high Reynolds numbers, the
velocity and time scales of the smallest eddies are
small compared to those of the largest eddies.

21

2/2/2016

A Practical Problem: Mixing

43

Kolmogorov Hypothesis
Kolmogorovs hypothesis of local isotropy states that at
sufficiently high Reynolds numbers, the small-scale turbulent
motions (l << l0) are statistically isotropic.
Kolmogorovs first similarity hypothesis states that in every
turbulent flow at sufficiently high Reynolds number, the
statistics of the small scale motions (l < LEI ) have a universal
form that is uniquely determined by and .
Kolmogorovs second similarity hypothesis states that in
every turbulent flow at sufficiently high Reynolds number,
the statistics of the motions of scale l in the range l0 >> l >>
l have a universal form that is uniquely determined by
independent of .
Note: LEI is the length scale that forms the demarcation between
the large scale anisotropic eddies. L0 is the Integral length scale.
l is the Kolmorogov length scale.

22

2/2/2016

Introduction to Energy Spectrum

We introduce a new length scale LDI , (with LDI 60 l for
many turbulent high Reynolds number flows) so that this
range can be written as LEI > l > LDI .
This length scale splits the universal equilibrium range
into two sub-ranges:
The inertial sub-range (LEI > l > LDI) where motions are
determined by inertial effects and viscous effects are
negligible.
The dissipation range (l < LDI) where motions experience
viscous effects.

Eddy Sizes
The bulk of the energy is contained in the larger
eddies in the size range LEI = l0 /6 < l < 6l0 , which
is therefore called the energy containing range.
The suffixes EI and DI indicate that LEI is the
demarcation line between energy (E) and inertial
(I) ranges, as LDI is that between the dissipation
(D) and inertial (I) ranges.

23

2/2/2016

Energy Transfer Rate

The rate at which energy is transferred from the larger
scales to the smaller scales is T(l).
Under the equilibrium conditions in the inertial subrange this is equal to the dissipation rate , and is
proportional to u(l)2/ .

Wave Numbers
The wave number is defined as = 2/l
Different ranges can be shown as a function of wave no.
The wave no. can also be made non-dimensional by
multiplying it with the Kolmogorov length scale l (or ) to
result in the commonly used dimensionless group ( ).

24

2/2/2016

Turbulent Kinetic Energy Spectrum

49

Turbulent Kinetic Energy Spectrum

The energy spectrum of fully developed homogeneous turbulence is
thought to be composed of three distinct wave-number regions (see
previous fig.).
A. In this region the large energy-containing eddies are found. These
eddies interact with the mean flow and extract energy from the mean
flow. The energy is transferred to slightly smaller scales and eventually
into region B.
B. This region is the inertial sub-range. In this region turbulent kinetic
energy is neither produced nor dissipated. However, there is a net flux
of energy through this region from A to C. The existence of this region
requires that the Reynolds number is high.
C. This is the dissipative region where turbulent kinetic energy is
dissipated into heat. Eddies in this region are isotropic and the scales
are given by the Kolmogorov scales.
50

25

2/2/2016

Turbulent Kinetic Energy Spectrum

However, the small eddies, which are of very high
frequency, in the region of d , dissipate energy. Here,
turbulent kinetic energy is dissipated into heat by
molecular viscosity. The viscous stresses prevent the
generation of eddies with higher frequency.
In wave-number space, the energy of eddies from to
d can be expressed as E ) d
The total turbulent kinetic energy, k, which is the sum
of the kinetic energies of the three fluctuating velocity
components, i.e.
is obtained by
integrating over the whole wave-number space

k E ) d
0

51

Turbulent Kinetic Energy Spectrum

One important tool for analysing the different regions of
turbulence is the energy spectrum.
It is common practice to use wave numbers ( ) instead of
length scales. The dimension of a wave number is one
over length, thus we can think of the wave number as
inversely proportional to the eddy radius, i.e.

1 r
This means that large wave numbers correspond to small
eddies and small wave numbers to large eddies. Eddies
with wave numbers in the region of e , the previous figure
contains the largest part of the energy and contribute little
to the energy dissipation.
52

26

2/2/2016

Turbulent Kinetic Energy Spectrum

The energy contained in eddies with wave numbers
between A and B is then
The full spectrum is given by:
E () in Inertial sub-range:
According to the second similarity hypothesis E () will
solely depend on k and .

The last equation describes the famous Kolmogorov 5/3

spectrum. C is the universal Kolmogorov constant, which
experimentally was determined to be C = 1.5.

Full Energy Spectrum E()

The full spectrum is given by:

The production range is governed by fL (which

goes to unity for large ( l0):

27

2/2/2016

Full Energy Spectrum E()

The dissipation range is governed by f (which
goes to unity for small ( ):
The model constants were determined
experimentally and based on the constraint that
E() integrate to . Their values are:
For given values of , , and ; the full spectrum
can now be calculated based on these
equations.

Normalized Energy Spectrum

It is, however common to normalize the spectrum in one of
two ways:
Based on Kolmogorov scale:
Measure of length scale becomes ( ).
E() is made dimensionless as E() /( u2)
Based on Integral scale:
Measure of length scale becomes (l0 ).
E() is made dimensionless as E() /(l0 )
Instead of having three adjustable parameters (, , ), the
normalized spectrum then has only one adjustable parameter:
Taylor-scale Reynolds number R.
Note: R can be related to the turbulence Reynolds number as
follows:

28

2/2/2016

The normalized energy spectrum for R = 500

The energy spectrum as a function of R

29

2/2/2016

The energy spectrum as a function of R

Measurements of Turbulent Energy Spectra

The figure shows exp.
measured 1D spectra (one
velocity component was
measured only, as indicated by
the 1 and 11 subscripts).
The number at the end of the
reference denotes the value of
R for which the measurements
were done.
Source: Pope, page 235.
Energy Containing Range:
Most of the energy (~80%) is
contained in eddies of length
scale lEI = l0/6 < l < 6l0.

30

2/2/2016

Measurements of Turbulent Energy Spectra

Determination of the spectrum requires
simultaneous measurements of all velocity
components at multiple points, which is usually
not possible.
It is common to measure one velocity component
at one point over a certain period of time and
convert the time signal to a spatial signal using x
= Ut with U being the time averaged velocity.
This is commonly referred to as Taylors
hypothesis of frozen turbulence. It is only valid
for u/U << 1, which is not always the case.
Spectrum measurements remain a challenging
field of research.

Turbulent Kinetic Energy Spectrum

A spectral analysis of the turbulence scales often reveals a region where
the distribution obeys the following relationship:

This equation is called the Kolmogorov spectrum law, or simply the 5/3
law. The equation states that, if the flow is fully turbulent, the energy
spectra should exhibit a 5/3 decay. The region where this law applies
is known as the inertial sub-range and is the region where the energy
cascade proceeds in local equilibrium.
This law is often used in experiments and simulations (DNS, LES) to
verify that the flow is fully turbulent.
Note that the largest eddies, e , contain most of the turbulence energy
and are therefore responsible for most of the turbulent transport.
Nevertheless, the small eddies are responsible for mixing on the small
scales.
62

31

2/2/2016

Energy Spectrum captured by DNS & LES

The direct numerical simulation (DNS) of turbulent flows at
technically relevant Reynolds numbers will remain unfeasible
due to its huge computational cost in the foreseeable future.
Large Eddy Simulation (LES) offers here an attractive
compromise. As shown in figure, the basic idea of LES is that it
resolves numerically only the large structures of the turbulent
motion, which are associated with small wave numbers range,
while it models unresolved small so-called sub-grid scale (SGS)
structures.

Dissipation Rate Spectrum

We now know which eddies contain most of the energy. The
question remains, which eddies exactly dissipate the energy?
This question can be answered by constructing a dissipation
rate spectrum D(). The integral of D() over the full
wavelength range is by definition the energy dissipation rate :

Furthermore, with being defined as the multiple of the

kinematic viscosity and squared velocity gradients [of order
(du/dx)2 ~ k/l2 ~ k 2 ~ 2 E() ] we can then deduce:

32

2/2/2016

Dissipation Rate Spectrum

Therefore, dissipation rate would be

Here (0, ) is the cumulative dissipation; the energy dissipated

by eddies with a wavelength between 0 and .
The unit of D() is m3/s3 and it can thus be normalized with a
velocity scale cubed, typically the Kolmogorov velocity scale.
Just as the normalized E() only depended on Re, so does the
normalized D() depend only on Re.

Dissipation Rate Spectrum

33

2/2/2016

Dissipation Rate Spectrum

Most of the dissipation (~ 90%) occurs in eddies of length
scales LDI / = 60 > l/ > 8 .
This means that most of the dissipation occurs at scales
that are larger than the Kolmogorov scale . The
Kolmogorov scale should be interpreted as a measure of
the smallest eddies that are present in a turbulent flow at
high Reynolds numbers.
How long does it take for a large scale eddy to break up and
be dissipated? The spectra can be further analyzed to show
that eddies spend about 90% of their total lifetime =k/ in
the production range, and that once eddies enter the
inertial sub-range it takes only about /10 before the
energy is being dissipated. This time /10 is also referred to
as the cascade timescale.

Intermittency
Neither k nor are constant in time or space.
Within a turbulent flow field, k and may vary
widely in space, sometimes by orders of
magnitude.
Also, at a given point in space the instantaneous
values of may vary in time. This is called
intermittency. The peak values of relative to the
mean tend to increase with Reynolds number.
Peak values may be of the order of 15 times the
average in laboratory scale flows and 50 times
the average in atmospheric flows.

34

2/2/2016

Validity of Kolmogorovs Theory

Kolmogorovs theory is an asymptotic theory: it has been
shown to work well in the limit of very high Reynolds no.
The exact shape of the normalized spectra may deviate from
Kolmogorovs model spectra for intermediate Reynolds
numbers. E.g. for many laboratory scale flows which have
Reynolds numbers on the order of 10,000 with Re~ 250, the
exponent of E() ~ p in the inertial subrange is often
measured to be p ~ 1.5 instead of 5/3 (~1.67).
Kolmogorovs theory assumes that the energy cascade is one
way: from large eddies to small eddies. Experimental studies
have shown that energy is also transferred from smaller scales
to larger scales (a process called backscatter), albeit at a much
lower rate and the dominant energy transfer is indeed from
large to small.

Validity of Kolmogorovs Theory

The theory assumes that turbulence at high Reynolds
numbers is completely random. In practice, large scale
coherent structures may form.
Research into the fundamental aspects of turbulence
continues, both experimentally and by means of large
computer simulations using DNS (direct numerical
simulation); and the theory continues to be refined.

35

2/2/2016

Vorticity and Vortex Stretching

Existence of eddies implies rotation or vorticity.
Vorticity concentrated along contorted vortex lines or bundles.
As end points of a vortex line move randomly further apart the vortex
line increases in length but decreases in diameter. Vorticity increases
because angular momentum is nearly conserved. Kinetic energy
increases at rate equivalent to the work done by large-scale motion that
stretches the bundle.
Viscous dissipation in the smallest eddies converts kinetic energy into
thermal energy.
Vortex-stretching cascade process maintains the turbulence and
dissipation is approximately equal to the rate of production of turbulent
kinetic energy.
Typically energy gets transferred from the large eddies to the smaller
eddies. However, sometimes smaller eddies can interact with each
other and transfer energy to the (i.e. form) larger eddies, a process
known as backscatter.
71

Vortex Stretching
The interaction between vorticity ( )and velocity
gradients ui ) is an essential ingredient to create and
,i
maintain turbulence.
Consider an fluid element rotates on x-y plane about
z-axis in addition to a stretching along z-axis.
Therefore, an extension in z-direction can decrease
the length scales and increase the velocity
components in the x-y plane.
Neglecting viscous dissipation, Conservation of
angular momentum says that r 2 Constant
According to Kelvins Circulation theorem:
2
Circulation, r Constant
72

36

2/2/2016

Vortex Stretching
Vorticity will cause stretching along z-direction. Thus, there will
be decrease in length in x and y directions. This will increase
vorticity in these directions, hence stretching will occur in x and ydirections.

In this fashion, stretching in

one direction will promote stretching
In other directions.
Soon stretching will be rapidly
promote d in all directions making the
turbulence isotropic in nature.
(See family chart in next slide).

Vortex stretching on a circular cylinder

73

Vortex Stretching
Thus, the cascade of energy of turbulent motion
estimates to smaller scales.
z

(i.e., larger and larger velocity gradients)

Vortex stretching tends

to make the smaller eddies
which is isotropic, hence,
their contribution to
Reynolds uv ) shear stress
is zero.
Family Tree showing how vortex
x

stretching produces small-scale

isotropy
74

37

2/2/2016

Vortex Stretching

t2

t1

t4
(Baldyga and Bourne, 1984)

t5

t3

t6
75

38