29 aug 2006molecular tagging velocimetry molecules, the ultimate flow tracers (or are they?) nico...

29 aug 2006 Molecular Tagging Velocimetry

Molecules, the ultimate flow tracers(or are they?)

Nico DamApplied Molecular Physics

Radboud University of NijmegenNijmegen, NL

The real work:Jeroen BominaarThijs ElenbaasMira PashtrapanskaCoralie SchoemaeckerMargriet VerkuijlenWillem van de Water


Writing in air

• Introduction

• Molecules vs. particulates

• Practical stuff

• Examples & image processing

• Conclusions


563: Anonymous, unknown technique• 1984: Hiller et al., biacetyl• 1997: Noullez et al., RELIEF

– first application to turbulence2000: 3 new NO-based techniques• present:

– ca. 10 MTV-schemes (gas phase)– few in-depth studies

Writing in air: a concise history


Molecular Tagging Velocimetry

Three steps:

1. Create well-defined tracer distribution

2. Wait

3. Visualize advected tracer distribution

molecular

A velocimetry technique based on displacement ofmolecular clouds


Phosphorescence MTVbiacetyl Room temperature

vapour pressure:5 kPa (50 mbar)

Legend: internal (vibronic) energy stateoptical transitioninternal transition (relaxation)

Ene

rgy


Phosphorescence MTVbiacetyl

Absorption spectrum Luminescence spectrum

355 532416



Example: microjet visualisation (1 mm orifice)(flow control, orbit maintenance)

N2 in

Experimental setup



laserbeam

nozzle

1 mm


Ps 100 kPa; P0 few kPa (estimate)Center-line velocity: 330.6 0.7 m/s

(false colour scale)

s

s

s



Observations:• Tracers are created locally, seeding is global• Very good flow visualization• Single write, multiple read• Precise velocity determination• Tracer distribution broadens in time• Relatively straightforward laser & detection system

But:• Single shot images are rather poor in contrast• Requires O2-free flow• Biacetyl will ruin your social life


Molecules versus Particulates

Particularities of MTV:

• Create

• Well-defined

• Molecular tracers

local seeding

taylored to situation

the real flow (?)

non-intrusive (?)


Molecules versus ParticulatesThe role of diffusion - 1

caseourin

20][

1151

1)(

2

2s

uu

Du

(x0; t0)

(x; t)



Written structure blurredby molecular diffusion


11 Sc2/Sc44)( Dd


(x0; t0)

(x; t)

Schmidt number: Sc = /D 1 (air)

Komogorov scale remains unresolved


Molecular Tagging Velocimetrypractical stuff

Two steps:

1. Create well-defined tracer distribution

2. Visualize advected tracer distribution

Crucial issues How to create molecular tracers? How to visualize them?


Tracers for MTV

Requirements:

• distinguishable

• visible (“visualizable”)

• persistent

• producible in sufficiently large amounts

• convenient (non-toxic!)


Marking Molecules

Modify the default composition . . .

• . . . by creating new molecules

• . . . by changing the internal energy of existing molecules

Both can be done locally and instantaneously by (laser-)optical means

chemistry

physics

photochemistry

photophysics

e.g. N2 + O2 → 2NO

e.g. biacetyl → biacetyl*


MTV implementations(gas phase)


APaRTAir Photolysis and Recombination Tracking

• N2 + O2 + lots of h (193 nm) → NO

• Chemical pathway unknown (but it works)

• Highly localized NO creation

• Visualisation of NO by LIF

• Applicable to “air” and combustion

APaRTHardware – 1

C am era

X

Y

flow fie ld

226 nm pulseddye laser‘READ ’

t = t

Exim er N O lif

193 nm pulsed A rFexcim er laser

‘W RITE’

t = 0

Dye laser

APaRTHardware – 2

Excimer laser• pulsed gas discharge laser (rare gas halides)• for NO creation: 193 nm (ArF), ca. 50 mJ/pulse of 20 ns• not or hardly (< 1 nm) tunable

Nd:YAG-pumped dye laser (or OPO)• pulsed liquid (or solid) state laser• for NO visualisation: 226 nm, ca. 5 mJ/pulse of 5 ns• widely tunable

Rep. rate limited by visualisation laser & camera (10 Hz)


How non-intrusive is MTV?APARTWrite-laser-induced temperature rise

Beam diameter ca. 60 mPulse energy ca. 50 mJ in 20 nsBroad-bandIntensity ca. 17 MJ/m2

Instantaneous power ca. 25 MW

ambient


1 cm

40 c

m

0.65 cm

Nozzle diameter d = 1 cm

Measurements at x/d = 40

U = 40 m/s, u’/U = 25%

460R

10Re 5

Jet turbulence


Jet turbulence: snap shots

1ms

15ms

25ms

3ms

20ms

30ms(1.5


Analysis: velocity extraction

Cross section

Sub-pixel resolution through gaussian fit

5,0 5,5 6,0 6,5 7,00

200

400

600

Inte

nsity

[a.

u.]

position [mm]2

0

xx

o Ieyy


Analysis: velocity extraction

Perpendicular Cross section

Sub-pixel resolution through gaussian fit

5,0 5,5 6,0 6,5 7,00

200

400

600

Inte

nsity

[a.

u.]

position [mm]2

0

xx

o Ieyy


Velocity

urms

U


Structure functions

pyyvGp

p )(

(p=2)=0.75

p=8

3/1/1

3

)( yG pp

pp


Scaled exponents

t = 10 st = 30 s


t

Material line stretching

DNS: S. Kida and S Goto, Phys. Fluids 14, 352 (2002)

/exp)0()( *t

LtL


Issues in data processing

Finding the correct line centers can be hard in extremely curved lines with low intensity regions

Vertical and perpendicular fit


Issues in data processing

Vertical and perpendicular fits

??

Highly contorted or interupted lines


Examples of badly fitted line

• Result of the fit program with vertical and perpendicular fit

• Fit done by hand


‘Snakes’

1,0,)(),()( ssysxsv

1

0

dsEEE extinsnake

• Any line v (the ‘snake’) is assigned an energy value Esnake (a cost function)

• This ‘energy’ depends on– the shape of the curve (smoothness constraint)– its position within the image (quality of fit)

• The best fit is the curve v with the lowest energy


Snake fit: example



t

ti

tt eLLilL*

0),(

lt(1)

lt(n)

* 0.17



/exp)0()( *t

LtL

DNS: S. Kida and S Goto, Phys. Fluids 14, 352 (2002)


0 20 40 60 80 1000.0

0.5

1.0

1.5

still air turbulent flow

(t)

2 [10

-8m

2 ]

t [s]

t

e)t( 0 exponential spreading

diffusionDtt 4)( 20

2

tDt )24()( 20

20

2

At short times, t~ combined

Diffusion & stretching combined

= 0.25(DNS: 0.17)


At long times exponential spreading becomes clear

Again = 0.25

t

et 0)(

Exponential spreading at larger t


Clouds

Why can you make them at all?

• NO creation process is (strongly) non-linear (in intensity)


• Richardson: Mean-square separation should grow as t3 for long times:

320

2 tg)t(

t• Batchelor: Mean-square separation should grow as t2 for :

23

2

023

720

2 )()( tCt

Spreading of molecular clouds


READ

WRITE

WRITE

Pulsed excimer

laser

lens

lens

beamsplitter

Pulsed dye laser

spherical mirror for the “writing” of the second dot

WRITEt0+t

t0

Two clouds


t = 10 s t = 20 s

t = 30 s t = 40 s

Two clouds: examples


0 1 2 3 4 5 6 7 8 9 100

10

20

30

40

50

0'

0'

0'

0''

0'''

Batchelor

(<2 (t

)>-

2 (0))

/ (

1/3

)2 [-

]

(t / )2[-]

Batchelor rules


Molecules: the perfect flow tracers?

• yes & no, by definition• diffusion

– can(?) be tuned by Sc

• chemical & thermal intrusiveness:– can be tuned by tracer

• local tracer (cloud) creation• image processing is an issue• implementations exist for gas, liquid, flames

29 aug 2006molecular tagging velocimetry molecules, the ultimate flow tracers (or are they?) nico...

Documents

situationthe real flow

kpa p0

ultimate flow tracersor

new molecules

o2free flow biacetyl

based techniquespresent

nlthe real work

single shot images