Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a...
Transcript of Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a...
![Page 1: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/1.jpg)
1
Murat Tutkun
Institute for Energy Technology (IFE)Process and Fluid Flow TechnologyKjeller, Norway
École Centrale de Lille Laboratoire de Mécanique de Lille (LML) Villeneuve d'Ascq, France
Turbulence and hot-wire measurementsMek 4600
Hot-wire rake of 143 single-wire probes
![Page 2: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/2.jpg)
2
Turbulence! What is it?
Realizations of solutions to the governing equations and boundary conditions.
Most engineering flows are turbulent.
![Page 3: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/3.jpg)
3
Instantaneous Navier-Stokes Equations (NSE):
jj
i
ij
ij
i
xxu
xp
xuu
tu
~~1~~~ 2
Continuity equation:
0~
k
k
xu
Is turbulence still a problem?What about Navier-Stokes Equations?
C.-L. Navier
G.G. Stokes
![Page 4: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/4.jpg)
4
There are some real disadvantages associated with these equations.
Most solutions of interest are ‘random’ (or ‘stochastic’) and ‘chaotic’ in character.
All scales of motion are important to the dynamics; none are negligible.
These Navier-Stokes equations are non-linear due to convective term: jij xuu ~~
![Page 5: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/5.jpg)
5
Do we really bother with this?
Turbulence is almost everywhere:
• Aero/hydrodynamics (Airplanes, ships, submarines, road vehicles, trains)(Pipeline, channels, distribution systems)
• Environmental flows• Industrial processes (chemical and multiphase)• Combustion• Energy technology (gas turbines, wind turbines,…)
The answer is YES!
![Page 6: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/6.jpg)
6
The range of scales in real turbulent flows is enormous… typically 105 to 1021
Turbulent boundary layer Van Dyke, 1982
viscousinertia
Re The higher Re, the greater the separation of scales.
![Page 7: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/7.jpg)
7
It is very difficult to measure every scale and will be many decades before computing them directly.
Cascade of turbulence kinetic energy from scale to scale
Energy
Physical space – Structure functions:
Spectral space – Energy spectra:
![Page 8: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/8.jpg)
8
Flow Physics (DNS & Experiments)
Turbulence Modeling Flow control
Validation
Application (Performance Enhancement and Energy Efficiency)
The primary goal of any turbulence research is to be able to predict or at least model turbulence.
![Page 9: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/9.jpg)
What can we do then?
One obvious way is to divide flow into mean and fluctuations (the so called Reynolds decomposition).
e.g., mean:
fluctuation: ),(),(~),(),(~),(
txUtxutxutxutxU
iii
ii
Ensemble average < > is space and time-dependent.
9
![Page 10: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/10.jpg)
One has to be careful when computing statistical quantitiesin turbulence:
10 Streamwise velocity behind a grid in a wind tunnel
![Page 11: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/11.jpg)
11
![Page 12: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/12.jpg)
12
![Page 13: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/13.jpg)
It is not easy to get the statistics right in turbulencemeasurements; in particular high order moments.
13
![Page 14: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/14.jpg)
Autocorrelation of two random and one periodic process
14
![Page 15: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/15.jpg)
Integral time scale is obtained by integrating the area under the autocorrelation coeffeicient.
15
![Page 16: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/16.jpg)
)(vijji
jii
jj uu
xP
xU
xU
t
is the so-called Reynolds stress. jiuu
Mean momentum equation:
Plugging the Reynolds decomposition into the NSE yields the Reynolds Averaged N-S equations (RANS).
16
• It is a flow property.• Working against mean flow gradient and extracts
energy to turbulence at large scales.
![Page 17: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/17.jpg)
17
Now we have a closure problem due to : No new equations, 9 (6 independent) new unknowns
jiuu
j
itji x
Uuu
• 130 years since the turbulent viscosity,
• k-epsilon models are just another way to guess.
Original “gradient” idea: (Boussinesq (1877))
It has been proven that simple ideas/approaches do not work in this problem
![Page 18: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/18.jpg)
Reynolds stress models are another way…
kij
jkiijkjkij
ki uux
uuupupux
uuDtD
)(1
j
k
j
i
i
k
k
i
j
ijk
j
kji
xu
xu
xup
xup
xUuu
xUuu
1
Using Navier-Stokes equations to `build’ a set of equations for the Reynolds stress tensor.
• The number of unknowns is now 52, but only 13 eqns!
• Presence of pressure presents a huge problem!(non-locality)
18
![Page 19: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/19.jpg)
19
Anybody who does research in turbulence should keep following points in mind:
This presents real problems for turbulence models, since all closures are LOCAL.
The flow at a single point is related to the flow at every other point, and at all previous times (Triadic Interactions).
Even the terms in our averaged equations are NON-LOCAL in both space and time.
![Page 20: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/20.jpg)
20
How much of these can be measured in the lab?--- Unfortunately, not much!
• Particle Image Velocimetry:– 2, or 3 component of velocity at very high spatial resolution,– Most of the systems provide low temporal resolution,– Measurement field is often small,– Near-wall and low turbulence measurements are very difficult.
• Laser Doppler Anemometry- Very good at high turbulence measurements,- Handles very near-wall region,- Single point measurements,- Can provide reasonable sampling frequencies.
![Page 21: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/21.jpg)
We will focus on the most common measurement methodology used in turbulence reasearch (even today).
21
• Hot – wire anemometry:
– 1, 2, or 3 component of velocity at very high temporal resolution,– Based on heat balance along the sensor element,– Single point measurements,– Disturbance to the flow, – Poor response in high turbulence and recirculation,– Cheap compared to the others,– Easy to manufacture in-house.
![Page 22: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/22.jpg)
22
Operating principle (as visualized by Dantec)
![Page 23: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/23.jpg)
23
Last lecture: TurbulenceStatisticsAutocorrelationIntegral time scaleEffective number of samplesRecord lengthVariability of estimator
Hot wire anemometry
Today: Hot wire anemometry
CalibrationSome examplesSpectral measurementsPractical design of experiment.
![Page 24: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/24.jpg)
24
Finite probe size limits the resolvable smallest scale.
Non-resolvable signal
Resolvable signal
Exp. Fluid Mech. 2007
Cut-off frequency in practice :
![Page 25: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/25.jpg)
25
Velocity U
Current I
Sensor (thin wire)
Sensor dimensions:length ~1 mmdiameter ~5 micrometer
Wire supports (St.St. needles)
![Page 26: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/26.jpg)
26
All dates back to 1914: L.V. King (Phil. Trans. Roy. Soc., A214, 373-432).
On the convection og heat from small cylinders in a stream of fluid: Determination of the convection constants of small platinum wires with applicationto hot-wire anemometry.
where the dimensionless heat transfer rate (Nusselt number):
The Reynolds number: Overheat ratio:
I2Rw2 = E2 = (Tw -Ta)(A + B · Un) “King’s law”
![Page 27: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/27.jpg)
27
Hot-wire temperature profiles:
Freymuth, 1979
![Page 28: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/28.jpg)
28
Schematics of the anemometer circuit: Wheatstone bridge
![Page 29: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/29.jpg)
29
How it looks in reality:
![Page 30: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/30.jpg)
30
Calibration before and/or after the experiment is needed in order to convert voltages to velocities.
E2 = A + Bun King’s Law
U = C0 + C1E + C2E2 + C3E3 + C4E4 Polynomial calibration
![Page 31: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/31.jpg)
31
The relation is found by curve fitting to the calibration data using least square method.
![Page 32: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/32.jpg)
32
Two component measurements need cross-wires;and angular calibration
![Page 33: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/33.jpg)
33
U1
U2
U3 θ1
θ2
Calibration of cross-wires may be more troublesome and difficult than expected.
![Page 34: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/34.jpg)
We put all your angular calibrations onto one single curve for each of the sensors!
34
![Page 35: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/35.jpg)
More difficult and time consuming to perform this way!
35
![Page 36: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/36.jpg)
36
U∞ =15 m/sD = 20 mmReD = 20000x/D = 50
Axisymmetric far wake is very difficult to measure because of small velocity deficit.
![Page 37: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/37.jpg)
37
Bridge oscillations and quantization error (even for 16 bit A/D converter) are big problems in this case.
Johansson et al, JFM, 2006
![Page 38: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/38.jpg)
38
21.6 m long wind tunnel of Laboratoire de Mécanique de Lille (LML) is unique to conduct boundary layer research.
![Page 39: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/39.jpg)
39
z-position : 0, ± 4.0 mm, ± 12.0 mm, ± 28.0 mm, ± 60.0 mm, ±100.0 mm, ± 145.0 mm
y-position : 0.3 mm, 0.9 mm, 2.1 mm, 4.5 mm, 9.3 mm, 18.9 mm, 38. 1 mm, 76.5 mm, 153.3 mm, 230.1 mm, 306.9 mm
Probe positioning is crucial
A hot-wire rake of 143 single wire probes to get both spatial and temporal information about the flow.
![Page 40: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/40.jpg)
40
Some practical info: Fast (spectral) or slow measurementExample: pipe flow
Large scale ~ RCharacteristic velocity ~ UcenterlineTime scale of large scales ~ R/Ucl
Small scale: Kolmogorov microscale
Cut-off frequency in practice :
![Page 41: Turbulence and hot-wire measurements Mek 4600 · 13. Autocorrelation of ... (RANS). 16 • It is a flow property. • Working against mean flow gradient and extracts energy to turbulence](https://reader035.fdocuments.in/reader035/viewer/2022071603/613dd3832809574f586e3576/html5/thumbnails/41.jpg)
41