1 Copyright © 2016 by ASME

Proceedings of ASME Turbo Expo 2016: Turbine Technical Conference and Exposition GT2016

June 13-17, 2016, Seoul, South Korea

GT2016-57485

COMPARISON OF FLAME TRANSFER FUNCTIONS ACQUIRED BY CHEMILUMINESCENCE AND DENSITY FLUCTUATION

Johannes Peterleithner, Riccardo Basso, Franz Heitmeir, Jakob Woisetschläger Institute for Thermal Turbomachinery and

Machine Dynamics, Graz University of Technology

8010 Graz, Austria

Raimund Schlüßler, Jürgen Czarske,

Andreas Fischer Chair of Measurement and Sensor System

Techniques Technische Universität Dresden

01062 Dresden, Germany

ABSTRACT

The goal of this study was to measure the Flame Transfer Function of a perfectly and a partially premixed turbulent flame by means of Laser Interferometric Vibrometry. For the first time, this technique is used to detect integral heat release fluctuations. The results were compared to classical OH*-chemiluminescence measurements. Effects of equivalence ratio waves and vortex rollup were found within those flames and were then investigated by means of time resolved planar CH*/OH*-chemiluminescence and Frequency modulated Doppler global velocimetry. This work is motivated by the difficulties chemiluminescence encounters when faced with partially premixed flames including equivalence ratio waves and flame stretching. LIV, recording the time derivative of the density fluctuations as line-of-sight data, is not affected by these flame properties.

NOMENCLATURE c [m/s] speed of sound Dexit [mm] burner exit diameter fD [Hz] Doppler shift of laser light FTF [-] flame transfer function G [m³/kg] Gladstone-Dale constant ı [-] Laser incidence direction IOH* ICH* [-] light intensity of OH* and CH* kvib [mm/s/V] vibrometer calibration constant LIV laser interferometric vibrometry LDA laser Doppler interferometry

i-OH*-CL integral OH* Chemiluminescence o [-] observation direction PPM perfectly premixed Q’, [W] fluctuation of heat release qv’ [W/m³] fluctuation of heat release s [-] sensitivity vector TPM technically premixed u’ [m/s] fluctuation of Axial velocity U [V] voltage v [m/s] velocity vector x, y, z [mm] coordinates Φ [-] equivalence ratio α [°] phase angle

[m] length of measurement volume Κ [-] ratio of specific heats λ [nm] laser wave length

[kg/mm³] density

INTRODUCTION

Modern gas turbines for power generation rely on premixed combustion systems to achieve high combustion efficiency and low emissions. As a drawback, high power densities and reduced damping capabilities of the combustor increase the susceptibility to thermoacoustic oscillations. These instabilities arise from the positive coupling between the fluctuations of pressure and heat release [1]. Prediction of gas turbine stability is often achieved by network models, originally used in system dynamics analysis. Within the model the flame remains a ‘black box’. The flame is described as a single input single output block. Usually the data for this block comes from

flameof theexit measucombdetermOH* fluctuhot wapprofor aequivtechn

Ainterfcan d[6, 7]time absenreleasexplois recproceis apknowflow pertur(LDAis disa rescalcuwhereto thepremiis invby mmeasuthe flmoduveloc

EXPE

Fgeomsiren,documfield the sishow(TPMpremi

Itangethrou

e transfer functe flame to pert[2, 3]. In thurement of

bustors have bmining the FTFwithin the flam

uations at the bwire anemometoaches relying adiabatic flamvalence ratio nically premixeAs an alternaferometric vibrdetect heat rele]. The advanta

derivative ofnce of pressurse rate [8]. Thoited before, bucorded. As a ess of traversinplied to a sw

wn flow field ais excited usinrbation level i

A) in the burnercussed for excsult from bo

ulated. This we chemiluminee absence of eixed configura

vestigated. Diffmeans of purements. To ilame, a Dopplulation (FM-DGcity field with l

ERIMENTAL

For the investimetry burner w, mounted intomented in detaand characteriiren has been cn with the sta

M) - to the leftixed (PPM) - t

In the TPM conential air (b) augh a stratifier

tions, which returbations of ahe past, a v

flame transbeen publishedF is measuringme as the measburner exit aretry or laser Don chemilumi

mes with low[4, 5], this m

ed combustion ative Interferomrometry (LIV) ase without th

age of this techf density flucre fluctuationshe link betweut this is the fibenefit, it oveg the whole fie

wirl stabilized and well definng a siren in ths measured wir exit nozzle. S

citation frequenoth, the flamewas done for escence shouldequivalence ratation more releferences in hea

planar time interpret the naler global veloGV) was used line-of-sight-in

SETUP

igations presenwas used and o the axial air-fail in the work istics have beecharacterized iandard configu

and the refereo the right of thnfiguration, thand axial air r in order to

elate the unsteaacoustic velocivariety of mefer functions

d. A method wg the chemilumsure for heat re

e obtained eitheoppler anemominescence are

w strain ratesmethod is notsystems. metric techniqare not affectee restriction ofhnique is that ctuations dires, this is a men those quan

first time, integercomes the teld. In this studburner config

ned operating he main air supith laser DoppSecondly, integncies from 0 He transfer fua perfectly pr

d provide a reltio waves. The

evant to industrat release are fresolved equ

atural heat releocimeter with

to compare thntegrated heat r

nted in this pathe flow was feedline. The bof Giuliani et

en published ren [11]. In Fig.

uration - technence configurahe setup. e combustor is(c). The axiaensure purely

ady heat releasty at the burneethods for ths in researcwidely used fominescence fromelease. Velociter directly frommetry. Whereaonly applicabl

s and constant applicable t

ques like laseed by this. Thef perfect mixinit measures thctly. With th

measure of heantities has beegral heat releasime consumindy the techniquguration with conditions. Thpply, where thler anemometr

gral Heat releasz to 600 Hz. A

unction can bremixed flameliable result duen a technicallrial applicationfurther analyzeuivalence ratiase spectrum olaser frequenc

he time resolverelease rate.

aper, a variablexcited with

burner has beeal [9], the flow

ecently [10] an. 1 the burner ically premixeation - perfectl

s fed by fuel (aal air is force

axial flow. I

2

se er he ch or m ty m as le nt to

er ey ng he he at en se ng ue a

he he ry se

As be e, ue ly ns ed io of cy ed

le a

en w-nd is

ed ly

a) ed In

contrachambcylindthe buouter c

Ininto thsplit.bores pressuacoustslightlenablerelease

Taboveexit nothe pthermolayersThe moperatnumbemeasuheighttime tsystemthe me

TABL

Axia[g

0.4

FIGTPMCON

ast to this theber (d) and frdrical bores aligurner axis. Metchamber. The n case of perfehe air supply fAs shown inof a diamete

ure drop is prtically stiff anly alter the fued without nece.

The movable cee the exit in ordozzle (e) and c

point of operoacoustic labo

s of low reflectmass flow, wtion, can be er, according ured using a but of z = 8.5 mthe burner ex

m for all measeasurement pla

LE 1: FLOW P

al air g/s]

Tang

422 0

GURE 1: THEM CONFIGURNFIGURATIO

e tangential afrom there, engned tangentiathane is injectesiren modulate

ect mixing (PPMfar upstream bFig. 1, the fu

er of 0.5 mmresent. Therefond the acousticuel flow. Then cessarily genera

enter cone (f) oder to constricconsequently eration. The toratory within tive curtains an

which was thefound in Tabto [2] neglec

urner exit radiumm to z = 17 mxit diameter rsurements origane (g) is defin

PROPERTIES

gential air [g/s]

0.397

E EXPERIMENRATION TO TON TO THE R

Copyright © 2

air passes thenters the plenually and symmeed into the tanges the axial air M), the methan

before tangentiuel is injected. At those injore, they are cs of the plenu

equivalence rating large amp

of the burner wct the flow throensure correct test rig was a 3x3x2.5 m3

nd a sound abse same for bble 1. The sicting the pressus of 8 mm av

mm which is 0respectively. T

ginates at the bned by the x an

Fuel [g/s]

0.0683

NTAL SETUPTHE LEFT ANRIGHT

2016 by ASME

e outer mixingum through 32etrically aroundgential air in thflow only. ne was injecteal and axial ai

d through smalnjectors a larg

assumed to bum would onlyratio waves arplitudes of hea

was set to 1 mmough the burnemomentum foset up in

3 box with twsorbing ceilingboth points oimplified Swirsure term, waveraging over 0.5 times and The coordinatburner exit and z-axis.

Swirl Number [-]

0.54

P WITH THEND THE PPM

E

g 2 d e

d ir ll e e y e

at

m er or a o

g. of rl as a 1 e d

3 Copyright © 2016 by ASME

Measuring Techniques Heat release fluctuations for the FTF were acquired by

means of LIV and OH*-chemiluminescense using a photomultiplier with a filter. Velocity fluctuations for the FTF were acquired by means of laser Doppler anemometry. For planar time resolved velocity fields, FM-DGV was employed, and finally the phase averaged planar CH*/OH*-chemiluminescence measurements were used to visualize equivalence ratio waves. Below the different measurement techniques are explained in detail.

Laser interferometric vibrometry (LIV) detects the line

of sight ( ) integrated density fluctuations of gasses by means of interferometry. This is shown in detail in [12] and applied in [13, 14]. Using a Polytec laser vibrometer (interferometer head OFV-353, velocity decoder OFV-3001, calibration factor 5mm/s/V, 200kHz bandwidth, no filters, Polytec, Waldbronn, Germany), the measured voltage (U) is linked to the derivative of the density fluctuation (ρ) by the Gladstone-Dale constant (G) which is 2.59e-4 m³/kg for the present points of operation and the calibration factor (k) which was set to 5 mm/s:

2 ∗

(1)

The link between density fluctuations and heat release fluctuations has been derived and extensively discussed by [8, 6, 7]. Neglecting pressure fluctuations which - for unconfined flames - are low compared to volumetric heat release rate qv, the following equation applies:

1

(2)

With the ratio of specific heats (κ) and the speed of sound (c). In combustion application LIV is used to locally or globally detect heat release fluctuations. In order to locally resolve the heat release fluctuations, the vibrometer must be traversed in a two dimensional field [14]. Since for FTFs, only the space integrated information of the flame is relevant, the laser beam of the vibrometer was expanded to and collimated at a diameter of 81 mm and centered at a height of 40 mm above the burner exit plane in order to acquire the entire combustion fluctuations at once. For the comparison with vorticity, the vibrometers beam was narrowed down to 5 mm in diameter and traversed in two directions with increments of 5mm. The result was interpolated in order to provide more readable plots. Areas of uncertainty of the LIV technique include the varying Gladstone-Dale constant discussed in [6], dependency on temperature discussed in [12] and a varying intensity of the laser beam over the beam diameter due to a Gaussian distribution of the laser light. Since only the central part of the

beam covered the test section, the accuracy of the LIV detector signal is less than +/- 4 %, this can be further reduced, using a sufficiently strong light source.

Alternatively, for the FTF, the integral OH*-chemiluminescence (i-OH*-CL) intensity emitted by the flame was acquired using an UV filtered photomultiplier (PMM01, Thorlabs Inc., Newton, New Jersey, USA). On the photomultiplier a narrow band OH* interference filter (310 nm CWL, FWHM 10±2 nm Bandwidth, 50mm Mounted Diameter, 18 % Transmission, Edmund Optics, Barrington, NJ, USA) was mounted.

The processing of LIV and the i-OH*-CL signal were acquired with 100 kilosamples per second using a data acquisition with analog input modules NI-91215 (National Instruments, Austin, Texas) and Labview 8.6 software. The spectral analysis was performed using a fast fourier transform (FFT) based on Matlab routines. In order to tackle the scalloping loss of Fourier transforms, a Matlab 2015a implementation of a Flattop filter was used, which is valid if frequencies are known in advance, ensuring a high signal to noise ratio.

To record the velocities for the FTF a classical laser Doppler anemometer (LDA) was used to measure the axial velocity at the burner exit at z=2 mm x=r= 5.5 mm. (FibreFlow, DANTEC Dynamics, Roskilde, Denmark). Since LDA does not provide frequency spectra per se, with the help of a siren trigger the result was phase averaged and then divided into 64 bins. A FFT was performed on the phase averaged result and the base-frequency was used for the FTF.

A particle image velocimetry (PIV) was used as reference. The PIV setup was the same as in [10] with 1200 averaged images and two cameras set up at an angle of 45°.

A Doppler global velocimetry measurement system with laser frequency modulation (FM-DGV) was employed to assess the flame dynamics of the non-excited flame in part two of this article. The FM-DGV technique relies on measuring the Doppler frequency shift fD of laser light, which is scattered by seeding particles moving with the flow:

| |

∗| |

| |

(3)

with as laser wavelength, as observation direction and as laser incidence direction. Hence, the velocity component

∗ along the direction of the sensitivity vector (i.e., along the bisecting line of and ) can be derived from the measured Doppler frequency. For determining the Doppler frequency shift, a frequency stabilized laser source in combination with a molecular cesium absorption cell with frequency dependent transmission was used. The frequency

shift cof the

Tin [16with Photothree three mm aveloccan bthe mfor thresolv

Tratio (Nano310 Göttinfilter MounBarrinwith averaavera

RES

StatiT

FIGTPMFLU

can then be eve scattered lighThe applied FM6]. It features a maximum o

onics AG) for observation dcomponent m

and a maximucity spectra wibe resolved. A tmeasurement uhe velocity speve expected floThe OH* and

waves were oStar, 1280x1nm/430 nm =ngen, Germany(430 2 nm C

nted Diameterngton, NJ, USthe images r

aged. For phasaged according

ULTS

onary FlameThe mean flam

GURE 2: OHM FLAME. UCTUATION

valuated by meht behind the abM-DGV system

a high-power output power achieving a hi

directions weremeasurements wum measuremeith up to a matemporal avera

uncertainty, resectra of 2 kHz,ow structures. CH* acquisitimeasured wit024 Pixel, pho= 65 %, DaVy) and a OH*

CWL, FWHM r, 98 % Tra

SA). The camerecorded phase steps α of 4to:

∗

∗

e Properties me behavior and

H*-CHEMILUMMEAN (LE

(RIGHT)

easuring the intbsorption cell [

m was describecontinuous lasof 1 W at 89igh signal-to-n used, enablinwith a spatial ent rate of 10

aximum frequeaging was applisulting in a cu, which is suff

ion providing tth a light-inteotocathode radVis 7.6 Softw(above) or a C10 2 nm Ban

ansmission, Eera was triggerse resolved a45° one hundre

∗

∗

∗

∗

and Flow Fied flow field wil

MINESCENCEEFT) AND

tensity variatio[15]. d in more detaser illuminatio95 nm (Topticnoise ratio. Alsg simultaneouresolution of

00 kHz. Henceency of 50 kHied for reducin

ut-off frequencficiently high t

the equivalencensified camerdiant sensitivitware, LaVisionCH* interferencndwidth, 50 mmdmund Opticred by the sireand then phased images wer

(4

eld ll be discussed

E OF THERMS OF

4

on

ail on ca so us,

1 e,

Hz ng cy to

ce ra ty n, ce m s,

en se re

4)

belowreactinleft) downs2.5Detakes reactiostrain.

Indominexit ddecreacaused2xDexi

drivenfrom ambieacquiragreemFM-Dshowsrecord

FlamePr

proceschemivalue aroundfluctuaat thephotom(TPMthe tyfrequefrequeveloci

FIR

w on the examng gasses meashows the m

stream of the exit. The typicaplace in the inon is quenched. n this flame nated by the axdiameter (Dexit)ases slightly dd by the suddeit or 30 mm n by expansion2.5xDexit grad

ent gasses. Thered by FM-DGment between

DGV and stereos the reliabilityd velocity spec

e Dynamics remixed swirl ss. Fig. 2 (riluminescence(left) shows thd the positioations are more spatially inmultiplier (Fig

M and PPM) looypical featuresencies below 2encies. This iities within the

IGURE 3: PIRESULTS OF T

mple of the TPasured by OHmain reaction

burner exit eal V-shape flamnner shear layd in the outer

the fluid flowxial component) after the burdue to swirl inen area change downstream t

n of the hot gadually decelerae mean velocityGV and stereo-P

the two differo-PIV is very y of the novel tra as discusse

stabilized com

right) shows fluctuations.

hat the highest on of mean re extended thntegrated heag. 4), the specok quite differs of a turbul200 Hz and geis expected de flame of arou

IV (LEFT) ATHE MEAN V

Copyright © 2

PM case. The *-chemilumine

n zone just extending to ame indicates thyer of the jet o

shear layer by

w of the react (Fig 3). Fromrner exit nozzlnduced jet opeat the burner e

the fluid startases in the reacates due to my field for reactPIV. As shownrent measuremgood. This coFM-DGV whid below.

mbustion is a hthe RMS vaComparison wamplitudes of

heat releasehan above the fat release recctra for both orent. Both flamlent process.

ently decrease due to the cound 5m/s (Fig.

AND FM-DGVELOCITY FIE

2016 by ASME

distribution oescence (Fig. 2after 2xDexi

a maximum ohat the reactiononly, while thy heat loss an

cting gasses im half the burne

le, the velocityening, which iexit. At arouns to acceleratction zone, and

mixing with thting gasses wan in (Fig. 3) th

ment techniquesonfidence checkich was used to

highly dynamialue of OH*with the meanf fluctuation are. Below thflame. Lookingcorded by thoperating pointme spectra show

They peak atowards highe

omparably low3). The PPM’

GV (RIGHT)ELD.

E

of 2 it

of n e d

is er y is d e d e

as e s, k o

c *-n e e g e ts w at er w s

broadconsiAdditpeak under

Ifluctupeak broadbelow

RecoI

upstreOH*-condiappro

Trecord

dband backgroderably lowertionally the p

at around 1rground signal.In contrast to uation in the at 250 Hz is r

d band spectruw.

ording the FlIn the folowineam excitation-CL flame tranitions are reoaches is perforTo investigatding the FTF i

FIGURE 4:CASE.

FIGUR

ound peaks atr than the o

perfectly prem125 Hz, sitti. this, the TPMlow frequenci

recorded whichum. This effe

ame Transfeng, the linearn is investigatensfer functionsecorded. A rmed. te thermoacois a common to

: NATURAL S

E 5: BURNER

t around 125one of the Tixed flame shng on top o

M case shows ies. Secondly h is about twicect will be fu

er Function r response ofed. By meanss (FTF) under

comparison

ustic stabilityool. FTF reduc

SPECTRUM O

R EXIT VELOC

Hz, which TPM test casehows a distincof the smoot

a lower overaa very distince as high as th

urther discusse

f the flame t of LIV and TPM and PPMbetween bot

y of flameces the comple

OF OH*-CHEM

CITY SPECTR

5

is e. ct th

all ct he ed

to i-

M th

s, ex

dynamplot. Tthe flaincludof thefluctua

w

mean fluctuaThere from acoustthe mvelociwith i-

MILUMINENC

RA OF THE N

mic behavior oThis informatioame is genera

des one input we flame, andation

with the heat r

heat release ation ‘, agaiare several mexperiments. tic method, cal

most widely usity just upstrea-OH*-CL.

CE FOR THE

NON-REACTIN

of combustion on can be useally treated aswhich is the flud one output

FTF′′

release fluctua

. This is then in normalized

methods to acquWhile [17, 1lled multi micrsed approach am of the flam

E PPM (LEFT)

NG FLOW.

Copyright © 2

to an amplitudd for network s a ‘black boxuctuation of vel

which is the

/

/

ation ′ morm

divided by theby its mean

uire the FTF, m8] have develrophone metho

is to measurme and the he

AND TPM (R

2016 by ASME

de and a phasmodels, wher

x‘. This modelocity upstreame heat releas

(5

malized by th

e axial velocitycounterpart

most commonlyloped a purely

od (MMM), stilre the dynamieat release rat

RIGHT) TEST

E

e e

el m e

5)

e

y . y y ll c e

T

Wfluctunot quantmeasuof flausingcompFor pchanganalythe Onormcalibrdepencomepremiequivfluctu[4]. Unot pto mtechnfluctu

FITH

While it is a suations causingso easy to

titatively. I-Ourement tool tames. Mostly

g an appropriaparably easy toperfect premixge with time, ysis. For a conOH* radical is

malizing the iration factor cndency on the es into play whixed’ flames

valence ratio cuations when mUnfortunately mperfectly premimeasure heat rnique not siguations.

IGURE 6: FLUHEIR AMPLIT

simple task to g the flame per

acquire theOH*-CL is ato assess the d

the OH*-radiate narrow bao use and low xing where the

it is the mosstant equivalens proportional intensity by can be avoided

equivalence rhen partly pre

are investigan cause an o

measuring the Omost industriaixed, and thererelease fluctua

gnificantly aff

UCTUATIONTUDE, PHASE

record the uprturbations, un

e heat releasa commonly ynamics and hical emission and filter. Gecost measurem

e equivalence st preferred tonce ratio the lito the heat rethe mean v

d. The major ratio and the semixed, so calgated. Then overestimation OH*-chemiluml combustion aefore an alternations is needfected by equ

S OF HEAT E AND UNWR

pstream velocitnfortunately it se fluctuation

used opticaheat distributiois recorded b

enerally it is ment technique

ratio does noool for stabilitight intensity oelease [19]. Balue even thdrawback is itstrain rate. Thled ‘technicallfluctuations oof heat releas

minescence onlapplications ar

native techniquded. LIV is uivalence rati

RELEASE, VRAPPED PHA

6

ty is ns al

on by

a e. ot ty of

By he ts is ly of se ly re ue a

io

Land tuthe tefluctuais barinfluenconstaheat rthe FTflame The mprior flame

Tin (Figamplitstrongthan modulexit. Tminorat 550alsmoan inc

VELOCITY ANASE RESPECT

LIV has provenurbulent flameschnique is capations better th

rely affected bnce is presentant. Following release not onlTF the result w

(3,4 kW) assmean temperatu

temperature marea of 1000 m

The fluctuationg. 5) recorded tude spectrum

g overall modu5 percent oflation is strongThree significar peak at 125 H0Hz. The stronost 40 percent creasing lag wit

ND THE FTFTIVELY.

n to measure hs [6, 7, 20]. Forpable of recorhan chemilumiby equivalencet due to a chthe above equ

ly qualitativelywas normalizesuming a combure of the flammeasurements mm2 was used.ns of velocity mwithout the re

m is plotted froulation level of the mean fg enough to vaant peaks can bHz with about 2ngest fluctuatiofluctuation. Thth increasing f

F FOR THE P

Copyright © 2

heat release in r technically pr

rding the overainescence teche ratio waves. hange in the Guations, the LIVy but also quaed by the meanbustion efficie

me was set to 1[21] and a

measured by Lacting flame (fom 25 Hz up of the bulk veflow proves ary the velocitbe found in the20 percent flucon is however he phase (Fig. frequency.

PPM TEST CA

2016 by ASME

laminar flameremixed flameall heat releas

hniques, since i Only a minoGladstone-DalV measures thantitatively. Fon power of th

ency of 100 %1200 °C due tcross-sectiona

LDA are shownflow only). Th

to 600 Hz. Aelocity of morthat the siren

ty at the burnee spectrum, ontuation and onat 212 Hz with5 right) show

ASE WITH

E

s s e it

or e e

or e

%. o al

n e

A e n

er e e h

ws

Iand thPPM similasuppoupstredifferitself.

CflamePPM,well. themsOtherfluctuat 21technover t

Was forrespo

FITH

In Fig. 6 and Fhe FTF calculand TPM. Fo

ar trend as seorts the expeceam of the flrent injection . Considering hes acting as a , the spectra o

The velocityselves in the hr frequenciesuations can be 12 Hz. Over niques match vthe whole frequWhen discussinr PPM can be

onds to the velo

IGURE 7: FLUHEIR AMPLIT

Fig. 7 the relatlated from bothor both reactin

een in the nonctation, that thflame is not ssetups for TP

eat release rlow-pass filte

of the photomuy-fluctuations eat release cau

s are widelyfound due to ththe whole s

very well. The uency band. ng the TPM tee found. Againocity fluctuatio

UCTUATIONTUDE, PHASE

tive heat releah, is shown fo

ng cases, the vn-reacting flowhe flow field significantly c

PM and PPM

rate, the typicer is observedultiplier and L

discussed eausing a strong py suppressedhe intense velospectrum, bothphase general

st case (Fig. 7)n at 125 Hz thons and again t

NS OF HEAT E AND UNWR

ase, the velocitor the test casevelocity showsw (Fig. 5). Th

in the plenumchanged by thor the reactio

cal behavior od. Starting witLIV match verarlier, manifepeak at 125 Hz, only minoocity fluctuatioh measuremenly also matche

) a similar trenhe flame clearlthe 550 Hz pea

RELEASE, VRAPPED PHA

7

ty es a is m he on

of th ry st z. or on nt es

nd ly ak

is damflame behavfluctuadeflecthis, i-than 3assumcompasmalllarge pflow h

Adiffer,very wOH*-Coveresmeasuis podocum

VELOCITY ANASE RESPECT

mpened due tbut now the f

vior. Althoughations of heat

ction compared-OH*-CL sens30 % higher t

mption is thatared to the LIheat release flpressure drops

hardly affect thAlthough the am, the phase bewell. This meaCL is roughly stimation isurements whenossible in pamented for exam

AND THE FTFTIVELY.

to the low-pasfluctuation at 2h the LIV

at release, it dd to neighborinses high amplitthan the fluctui-OH*-CL ov

IV. This stronfluctuations cans over the injehe fuel mass flomplitude value

etween both mans that the adin phase with

s expected n equivalence rartially premixmple by [7].

F FOR THE T

Copyright © 2

ss-filter charac212 Hz also af

clearly detedoes not showng frequenciestude at 212 Hzuation at 125

verestimates thng overshoot in arise in confector. Then fluow. es of i-OH*-C

measurement sydditional effectthe heat fluctu

for chemratio waves arexed flames a

TPM TEST C

2016 by ASME

cteristic of thffects the flamects increasew a significans. In contrast to, which is morHz. Thus, th

he heat releasin company ofigurations withctuations of ai

CL and LIV dystems matchet recorded by iuations. Such anmiluminescence present, whichand has been

CASE WITH

E

e e d

nt o e e e

of h ir

o s i-n e h n

FIGUDIFF

URE 8: EQFERENT FREQ

QUIVALENCEQUENCIES FO

E RATIO OR BOTH TES

WAVES ATST CASES.

8

Tthe lasmeasumeasudifferefluctuabecaufluctuaamplifhigh, spite measuamplittransferesult

EquivA

correcthe heHz fchemioveresequivabe valan effe

Creleaseequivaratio measunormaof lowcase amixingcase, cexactlof reaca minequivauncertthe siguncertflame reliabl

Cfor thagreemat 212is exacompastrongheat re

T

The flame transt line of plot

urement techniurements was uence is only caations. The se the LIV ation. Apparfication is not but at around of low velo

urement technitude is high an

fer function ofof high amplif

valence RatioAlthough heat ctly by both meeat release recofor the TPMiluminescence stimates the healence ratio walidated in ordefect. Chemiluminesc

e reliably, but alence ratio [4fluctuations ϕ

urements, folloalized ϕ’-distriw signal-to-noat 212 Hz, the vg far upstreamchanges in equly zero, becausctants due to thnor order efalence ratio wtainty of such gnal is low (e.tainty increase

- where the le.

Considering TPhe 125 Hz cament of both m2 Hz the TPM actly where theared to the LIg evidence thaelease fluctuati

nsfer functions in Fig. 7 indiques. Since thused for the Faused by differstronger amplpredicts a h

rently, for at 125 Hz whe100 Hz wher

ocity fluctuatiiques matches

nd a good corrf the TPM casfications where

o Waves release fluctuaeasurement tecordings by i-OHM case. The

intensity recoeat release of tave being preseer to correctly i

cence might it is a very goo

4]. Fig. 8 showϕ’ acquired byowing equationibution across oise-ratio werevariations of ϕ

m of the sourcuivalence ratiose ambient air he unconfined ffect comparewithin the bulmeasurements.g. at the edgees. Therefore

signal-to-nois

PM, the variatioase. This is emeasurement tcase exhibits v

e i-OH*-CL siIV recordings.at the OH*-Intions of the flam

Copyright © 2

ns of the PPM dicate similarhe same set ofTF (i-OH*-CLrent recordingslification at thhigher overallboth systemere both velocire heat fluctuaions. The phs well at regielation is possse also shows e velocity is low

ations are genchniques, a stroH*-CL and LIe hypothesis orded by the pthe flame at 21ent. This assumidentify the inf

not always od tool to meaws one cycle y means of On 4. For eight the flame is p

e masked. Forϕ are close to zce of excitatioo cannot appea

can mix with flame setup. T

ed to possibllk flow. Withs it can be seenes of the flame

values in these-ratio is hig

ons in ϕ are alsexpected, bectechniques is gviolent oscillatignal overshoo This measure

tensity overestme.

2016 by ASME

case shown intrends for bothf LDA velocityL and LIV), ths of heat releashe LIV-FTF il heat releas

ms the majoity and heat aration is high inhase for bothions where thible. The flamthe interesting

w.

erally recordeong deviation inIV exists at 212

is that thphotomultiplie

12 Hz due to anmption needs tofluence of such

represent heasure changes inof equivalencOH*/CH*-ratiphase steps th

plotted. Regionr the PPM teszero. Due to thn for the PPM

ar. They are nothe swirled je

This however ile changes in

h regard to thn that wherevee) measuremene center of thgh - are mor

so close to zeroause there th

good. Howevertions in ϕ. Thi

ots significantlyement provideimates the tru

E

n h y e e is e

or e n h e e g

d n 2 e

er n o h

at n e o e

ns st e

M ot et is n e

er nt e e

o e r, is y

es e

FHT

FIGURE 9: VHEAT RELEATHE PULSATI

VORTICITY ASE FLUCTUA

ION AT 250HZ

(LEFT, FMATIONS (RIGHZ.

M-DGV) ANDHT, LIV) FOR

9

CompVeloc

Insignalwas idraises the TP250 His releIn thisto tracfluctuaflame releasearoundspectr

Fdimenresolvpresenreferen

Cthe methat thof the rises tbe fouin couby thetwo cothe coand itTherefthe unvorticefrom t

oudetachvortex= 10 mfrom touter this thwhichexactlIt seemjust rigmean vortice

Tregionpulsatof memm (6reachehigher

DR

parison of city Fluctuatn the previousl in the FTF wdentified as athe question a

PM case wherHz. To identify evant, not onlys chapter the hece the origin oations (FM-DGfront. The ve

e signal fromd 250 Hz withrum. or the domin

nsional cross-sved evolution nted in Fig. 9. nce.

Comparing the ean flow and the vorticity is e conical swirl-to a height of aund on the outeunterphase to the velocity fluctounter rotatingonical jet (rightts shear layerfore the inner ndisturbed jetes on both sidethe corner of thuter vortex mhed 180° earliex is about to demm, z = 5 mmthe center bodyvortex has reahe inner vorte

h interacts withly at this point ms that at thisght to enable thvelocity rema

es can be founThe heat releasn as the mean htions are closeran heat release68° - 113°), thed at x = 5 mmr levels. Due

Heat Releions s chapter, an

was shown at 2an equivalenceabout the nature the most ob

the formationy as integral daeat release dist

of this instabilitGV) which areelocity signal

m the LIV alsh several high

nant frequencyection of the fof heat and vThe photomult

vorticity flucthe mean heat most dominan

-jet (illustratedapproximately 3er shear layer. he inner shear tuation coming

g vortices, botht side of the flrs outwards, ashear layer is

t thins towardes merge. Howhe burner exit, merges with er. Looking atetach from the

m) and a negatiy at (x = 0), buached the closex street is amh the flame, thof operation, t

s point of operhis merge of coains high abo

nd. se fluctuationsheat release of r to the burnere. While there he strongest am

m and z = 15 mme to lower m

Copyright © 2

ease Fluctu

overshoot of 12 Hz for linee ratio wave. ural spectrum obvious effect cn of this peak, tata but also spatribution (LIV)ty and compar

e expected to infrom FM-DGV

so identify theher harmonics

y of around 2flame was anavorticity at thitiplier signal s

ctuation in Figdistribution in

nt along the ind in the second3 x Dexit. A secIt is less devellayer vortex. B

g from the plenh veering off thame). The swiaway from thlonger than th

ds the tip of wever, quickly

the the co-rotatin

t the time step outer corner oive vortex is a

ut already at timest inner co-ro

mplified. Sincehis could be tthe amplificatiration, velocityorrelated swirl

ove the flame,

s show pulsatiothe flame (Fig

r axis and just is no pulsation

mplitudes of hm, slowly decr

measurement r

2016 by ASME

uations and

the i-OH*-CLar excitation. IThis, howeve

of i-OH*-CL acan be found athe heat releasatially resolved) is investigatere it to vorticitynteract with thV and the heae oscillation ain the velocity

250 Hz a twoalyzed. A phasis frequency ierved as phase

g. 9 (left) withn Fig. 2 revealnner shear layed phase plot). Icond vortex canloped and startBoth are causenum generating

he center axis oirl bends the jehe burner axise outer one. Athe cone, th

after detaching

ng one whichp 23° a positivof the burner (xabout to detachmestep 113° thotating one. Bye it is the onthe reason whyion is so strongy and swirl arls. Although th, no correlated

ons in the samg. 2). The strongbelow the peakn below z = 1heat release arreasing towardresolution les

E

d

L It er at at e

d. d y e

at at y

o e is e-

h ls er It n ts d g

of et s.

As e g

h e x h e y e y

g. e e d

e g k 0 e

ds s

10 Copyright © 2016 by ASME

obvious but still distinct, the fluctuations in heat release start at x = 5 mm, z = 10 mm and float down- and outwards following the inner shear layer of the right-side jet. For interpretation of the results, it is important to keep in mind that the measurement resolution is only 5 mm in comparison to 1 mm for velocity measurements.

Comparison of results from two measurement techniques shows that both fluctuations occur within the upper inner shear layer of the jet, which is also the lower part of the main heat release, and the region where the flame is anchored. Looking at the phase evolution, it shows that negative vorticity leads to positive heat release, which means that a vortex veering away from the jet curls up the surface of the flame. This leads to an increased surface area and consequently to an increased burn rate. At x = 5 mm and z = 15 mm the heat release is highest and this is also the region where the vortex fully hits the flame just after it merged and grew in strength. Here both measurement techniques acomplish each other, in order to gain insight into the flame dynamics.

CONCLUSION

The focus of this study was to discuss an alternative way of recording heat release rate and the FTF. It was the first time integral LIV was used for this application. The technique is promising of a better prediction of heat release in partially premixed (technically premixed) flames, where it is less receptive to equivalence ratio waves than the classical i-OH*-CL method.

Heat release spectra and FTF’s for both perfectly premixed and technically premixed flames were analyzed. The trend of both systems correlated very well for the perfectly premixed flame as the literature suggests. While for the technically premixed case the agreement of the overall trends were good as well, a strong overshoot at one peak in the spectra of the i-OH*-CL signal was found. This was related to the known dependency of OH*-emission on equivalence ratio and correctly identified as such by visualizing the ratio between OH* and CH* fluctuations. A significant overshoot of OH*-CL without a considerable fluctuation of heat release can occur when the system features a high pressure drop over the injector. This acoustically stiff fuel line is then less sensitive to fluctuations of air flow. In the special case of this swirl-stabilized flame, the overshoot of the photomultiplier signal does not affect the FTF significantly because the trend of the FTF is mainly dominated by velocity fluctuations. Therefore the LIV method is an interesting alternative to i-OH*-CL in order to measure and quantify heat release fluctuations in perfectly as well as partially premixed flames. Secondly, naturally excited frequencies of the technically premixed flame were investigated, since the LIV technique clearly identified this effect as a heat release fluctuation. Time resolved velocity was an obvious quantity to investigate, in order to identify the root of these fluctuations. By means of FM-DGV the heat release fluctuation was tracked back to flame front roll up and

consequently vorticity. Pulsations of heat release can be observed in the shear layer where the flame is anchored and where vortices hit the flame, heat release is in phase with the vortices. Detailed analysis of the plots identifies the vortices as the source of the heat release pulsations. This leads to the final conclusion that LIV is a promising technique for full field heat release measurements. It has been shown, that it is possible to measure the heat release rate of an unconfined flame. Similar to chemiluminescence, it is not free of restrictions. These include sensitivity of the Gladstone-Dale constant to mixture fluctuations, and sensitivity to temperature. Additionallly, care must be taken when adjusting the laser beam in order to maintain a homogenous illumination of the measurement volume.

ACKNOWLEDGEMENT

This research was partially funded by the Austrian Science Fund FWF within grant FWF-24096-N24 “Interferometric Detection of Thermoacoustic Oscillations in Flames”. The authors would like to thank Prof. Thomas Sattelmayer, TU Munich for the fruitful discussions and all the support during this work. References [1] Rayleigh, "The explanation of certain acoustical

phenomena," Nature, vol. 18, no. 455, pp. 319-321, 1878.

[2] S. Candel, D. Durox, T. Schuller, J. .. Bourgouin and J. P. Moeck, Dynamics of swirling flames, vol. 46, 2014, pp. 147-173.

[3] T. Sattelmayer and W. Polifke, "Assessment of methods for the computation of the linear stability of combustors," Combustion Science and Technology, vol. 175, no. 3, pp. 453-476, 2003.

[4] M. R. W. Lauer, Determination of the Heat Release Distribution in Turbulent Flames by Chemi-luminescence Imaging, Technische Universität München, 2011, pp. 147-173.

[5] B. Schuermans, F. Guethe, D. Pennell, D. Guyot and C. O. Paschereit, "Thermoacoustic modeling of a gas turbine using transfer functions measured under full engine pressure," Journal of Engineering for Gas Turbines and Power, vol. 132, no. 11, 2010.

[6] T. Leitgeb, T. Schuller, D. Durox, F. Giuliani, S. Köberl and J. Woisetschläger, "Interferometric determination of heat release rate in a pulsated flame," Combustion and Flame, vol. 160, no. 3, pp. 589-600, 2013.

[7] J. Peterleithner, N. V. Stadlmair, J. Woisetschläger and T. Sattelmayer, "Analysis of measured flame transfer functions with locally resolved density fluctuation and OH-Chemiluminescence Data," Journal of Engineering for Gas Turbines and Power, vol. 138, no. 3, 2016.

[8] A. P. Dowling and A. S. Morgans, "Feedback Control of Combustion Oscillations," Annu. Rev. Fluid Mech., vol.

11 Copyright © 2016 by ASME

37, pp. 151-182, 2005.

[9] F. Giuliani, J. W. Woisetschläger and T. Leitgeb, "Design and validation of a burner with variable geometry for extended combustion range," in Proceedings of the ASME Turbo Expo, 2012.

[10] J. Peterleithner, A. Marn and J. Woisetschläger, "Interferometric In estigation of the thermo-acoustics in a swirl stabilized Methane flame," in Proceedings of the ASME Turbo Expo, 2015.

[11] F. Giuliani, A. Lang, K. Johannes Gradl, P. Siebenhofer and J. Fritzer, "Air flow modulation for refined control of the combustion dynamics using a novel actuator," Journal of Engineering for Gas Turbines and Power, vol. 134, no. 2, 2012.

[12] J. Peterleithner and J. Woisetschläger, "Laser vibrometry for combustion diagnostics in thermoacoustic research," Technisches Messen, vol. 82, no. 11, pp. 549-555, 2015.

[13] N. Mayrhofer and J. Woisetschlaeger, "Frequency analysis of turbulent compressible flows by laser vibrometry," Experiments in Fluids, vol. 21, pp. 153-161, 2001.

[14] B. Hampel and J. Woisetschläger, "Frequency- and space-resolved measurement of local density fluctuations in air by laser vibrometry," Measurement Science and Technology, vol. 17, pp. 2835-2842, 2006.

[15] A. Fischer, J. König, J. Czarske, J. Peterleithner, J. Woisetschläger and T. Leitgeb, "Analysis of flow and density oscillations in a swirl-stabilized flame employing highly resolving optical measurement techniques," Experiments in Fluids, vol. 54, no. 12, 2013.

[16] R. Schlueßler, M. Bermuske, J. Czarske and A. Fischer, "Simultaneous three-component velocity measurements in a swirl-stabilized flame," Experiments in Fluids, vol. 56, no. 10, 2015.

[17] C. O. Paschereit, B. Schuermans, W. Polifke and O. Mattson, "Measurement of transfer matrices and source terms of premixed flames," Journal of Engineering for Gas Turbines and Power, vol. 124, no. 2, pp. 239-247, 2002.

[18] M. L. Munjal and A. G. Doige, "Theory of a two source-location method for direct experimental evaluation of the four-pole parameters of an aeroacoustic element," Journal of Sound and Vibration, vol. 141, no. 2, pp. 323-333, 1990.

[19] C. S. Panoutsos, Y. Hardalupas and A. M. K. P. Taylor, "Numerical evaluation of equivalence ratio measurement using OH* and CH* chemiluminescence in premixed and non-premixed methane-air flames," Combustion and Flame, vol. 156, no. 2, pp. 273-291, 2009.

[20] J. Li, D. Durox, F. Richecoeur and T. Schuller, "Analysis of chemiluminescence, density and heat release rate fluctuations in acoustically perturbed laminar premixed flames," Combustion and Flame, vol. 162, no. 10, pp.

3934-3945, 2015.

[21] T. Leitgeb, "On the Design and Validation of a Variable Geometry Burner Concept," Graz, 2012.