Vortex Generators in Lean-Premix Combustion - Thermothgr/gasturbiner/Material_for_generating_slide… · Vortex Generators in Lean-Premix Combustion A novel fuel-air mixing technique

lopedBB’s

ent ofces,n. Intexlysisinglearied.ratorsureuni-aturesess ofs de-liquid

A. Eroglu

K. Dobbeling

ABB Corporate Research Ltd.,Baden-Dattwill, Switzerland

F. Joos

P. Brunner

ABB Power Generation Ltd.,Baden, Switzerland

Vortex Generators in Lean-PremixCombustionA novel fuel-air mixing technique on the basis of vortex generators has been deveand successfully implemented in the worlds first lean-premix reheat combustor of AGT24/GT26 series industrial gas turbines. This technique uses a special arrangemdelta-wing type vortex generators to achieve rapid mixing through longitudinal vortiwhich produce low pressure drop and no recirculation zones along the mixing sectiothis paper, after a short introduction to the topic, the motivation for utilizing vorgenerators and the main considerations in their design are explained. A detailed anaof the flow field, pressure drop and the strength of the vortices generated by a svortex generator are presented as one of the three main geometrical parameters is vThe results obtained through water model tests indicate that an optimum vortex genegeometry exists, which produces the maximum circulation at a relatively low presdrop price. Moreover, the axial velocity distribution along the mixing section staysform enough to assure flash-back free operation despite the elevated inlet temperencountered in a reheat combustor. After selecting this optimized geometry, the procthe arrangement of multiple vortex generators in an annular combustor segment iscribed. The optimum arrangement presented here is suitable both for gaseous andfuel injection, since it requires only one injection location per combustor segment.@DOI: 10.1115/1.1335481#

em,

f

m

d

t

m

a

rt

i

g.ABB

forre-

tol-be asclu-

een

hethe

24/pre-

itht ofents

n oned,enta

ual-ese

avehichamuressuretion

withtoaininonof

a

s

IntroductionOn the way to reach the ultra low emission targets of mod

gas turbines, lean premixed combustion appears to be thepromising technique available today. This technique requiresthe one hand, passing the maximum amount of air throughcombustor, and on the other, a complete mixing of the air andinjected into it. Only after complete and uniform mixing of awith fuel, locally lean conditions can be achieved within the cobustion zone, which in turn guarantees low NOX formation. Thesuccess of this technique depends primarily on the quality of ming that can be achieved prior to combustion.

However, achieving sufficiently good mixing within limitespace and residence time available in the mixing section of aturbine combustor is not a simple task. Difficulties arise dueconflicting requirements from different aspects of the combusdesign, such as mixing, flashback safety, pressure drop, robusand reliability of the design.

In order to obtain proper mixing of the fuel and the air streaboth large-scale distribution and fine-scale mixing are necessHowever, given the fact that the mass flowrate of the fuelcounts only a few percent of the mass flowrate of the air, andmomentum of the fuel injection is limited with the available suply pressure, it is not possible to distribute the fuel uniformwithin the air stream when a limited number of injection poinare employed. Unfortunately, a potential multi-point injection slution runs into its limits too, when the size of injection orificefall below an allowed limit. Additionally, suitability of such asolution for liquid fuel injection, which requires special atomizeis questionable. Another problem which led to abandonmenthis path in the past has been the combustion instabilitiesserved, especially under high pressure conditions.

A simple way of overcoming several problems associated wachieving proper mixing quality is utilizing the momentum of thair stream via vortex generators. By this way, large scale vortcan be created by the vortex generators which are employed

Contributed by the International Gas Turbine Institute~IGTI! of THE AMERICANSOCIETY OF MECHANICAL ENGINEERSfor publication in the ASME JOURNAL OFENGINEERING FOR GAS TURBINES AND POWER. Paper presented at the Interntional Gas Turbine and Aeroengine Congress and Exhibition, Stockholm, SweJune 2–5, 1998; ASME Paper 98-GT-487. Manuscript received by IGTI March1998; final revision received by the ASME Headquarters October 20, 1999. Asate Technical Editor: R. Kielb.

Journal of Engineering for Gas Turbines and PowerCopyright © 2

rnostontheuelir

-

ix-

gastotorness

sary.c-

thep-lytso-s

s,of

ob-

ithecesfirst

for bulk distribution of the fuel and subsequent fine scale mixinA comprehensive research program has been undertaken atCorporate Research Center to develop fast mixing techniqueslow emission combustion, based on vortex generators. Thequirements that no recirculation or low velocity zones can beerated and the pressure drop due to vortex generators has tolow as possible led to vortex generators which generate exsively streamwise vortices.

A design on the basis of these vortex generators have bdeveloped and implemented in the second combustor~SEV! ofGT24/GT26 as shown in Fig. 1. A more detailed view of tvortex generators, as pictured from the downstream end ofcombustor, is presented in Fig. 2.

Previously, a detailed account of the development of the GTGT26 machines, EV burners and SEV combustors have beensented@1–5#. In this paper the fundamental findings acquired wvarious vortex generator geometries during the developmenthe SEV combustor are described. First, flow field measuremwith laser Doppler anemometry~LDA ! and mixing quality inves-tigations with laser induced fluorescence~LIF! techniques fromwater model tests are presented. Then, conclusions are drawthe optimum vortex generator geometry based on mixing spepressure drop and flashback danger. Additionally, the arrangemof multiple vortex generators with integrated fuel injection incombustor segment and resulting vortex pattern and mixing qity are presented. Finally, main results and conclusions from thinvestigations are summarized.

Experimental TechniquesThe mixing and aerodynamics investigations reported here h

been carried out in a water channel with transparent models whave full optical access from all sides and from the downstreend for LIF and LDA measurements. Additionally, static pressmeasurements have been conducted in these models with pretabs installed upstream and downstream of the mixing secmodel.

The water model has been operated in closed-cycle mode,the exception of the LIF tests, during which it was switchedopen-loop mode in order to prevent contamination of the mstream with injected dye. The mean axial velocity of the maflow in water model was 1 m/s, resulting in a Reynolds numberthe basis of the hydraulic diameter of about 54,000. Upstream

-den,25,oci-

JANUARY 2001, Vol. 123 Õ 41001 by ASME

42 Õ Vol. 123,

Fig. 1 Combustion system of GT24 ÕGT26 gas turbines

e

a

wo

w

.

torer-omgle

ionsn,ntubeegip,tosec-

the test section, a settling chamber, equipped with honeycoscreen type of flow straighteners, followed by a 9:1 area ramatched-cubic contraction ensured nonuniformities in the maxial velocity of less than 2 percent and a free-stream turbuleintensity of about 1.3 percent.

The LIF tests have been conducted with the blue line~488 nm!of a 5 W Argon-Ion laser beam, transmitted to the test rig vfiber-optic cable and expanded to a sheet of about 0.6 mm thness via a cylindrical lens. The injected fluid is a weak solutiondisodium fluorescein, as the main stream is free of dye. Thejectant concentrations visualized with this technique have brecorded with a monochrome CCD camera and digitized withhelp of an 8 bit frame grabber board, which was installed oncomputer. A commercial image processing software was usedevaluation of statistical values such as the mean injectant contration and the standard deviation by averaging 10 successivetures. The accuracy of this method is estimated to be better thpercent.

The velocity measurements have been carried out with a tcomponent LDA system, operated in backscatter mode. Two cponents of the velocity are measured simultaneously and the tone is measured separately by rotating the probe 90 degrespect to first measurement. The measurement probe wasversed with a computer controlled three-axis traverse system

Fig. 2 SEV combustor as viewed from the downstream end

JANUARY 2001

mb-tioan

nce

iaick-ofin-

eenthe

afor

cen-pic-n 5

o-m-

hirdithtra-

Reference CaseAs a starting point, the mixing characteristics of a fuel injec

without any vortex generators are of interest. A number of diffent injection geometries, including transverse jets injected frinner and outer walls, in-stream tube injectors and central sinpoint injection have been tested. Due to practical considerat~e.g., number of parts, cooling, suitability for liquid fuel injectiothermoacoustic instabilities! an injector geometry as depicted iFig. 3 has been selected. This injector consists of a singlewhich is inserted into the flow through outer liner and bent 90 din the flow direction. Four injection holes are located at the tpointing slightly away from the horizontal plane in orderachieve the best possible distribution over the channel cross

Fig. 3 Injection geometry for the reference case without vor-tex generators

Transactions of the ASME

h-

en-seed

inerencedryft.send

ntsen-lts.tri-

ofti-s.

owinguti-ted.fol-

hee

instexanyer-ge-

d athisers

orstheres-lexi-

onthe

etheur

Fig. 4 The LIF pictures of fuel concentration at four succes-sive stations downstream of a central lance without vortex gen-erators

Fig. 5 The LIF pictures of fuel concentration in streamwiseplanes downstream of a central lance without vortex genera-tors

Journal of Engineering for Gas Turbines and Power

tion without fuel engulfment in the wake of the injector and witout impingement of the fuel jets on the liner walls.

The LIF pictures of injectant concentration over planes perpdicular to mainflow direction are given in Fig. 4, at streamwidistances ofz/H50.1, 1, 2, and 3. As can be clearly observfrom these pictures, there exists zones with a wide variationfuel concentration next to each other in addition to regions whno fuel exists, even at three channel-height streamwise distafrom the injection. Additionally, with the light sheet positioneparallel to flow direction, radial and circumferential symmetplanes are given in Fig. 5. The flow direction is from right to leThe injector is partly visible at the right side of the picture. Theconcentration pictures indicate that the quality of both large afine-scale mixing is far from being acceptable. Additional variaof the same injection geometry with increased injection momtum or other injection hole arrangements deliver similar resuNamely, only incremental modifications in the large-scale disbution pattern can be achieved when relied on the momentumthe injection alone. A major improvement in mixing requires ulizing the momentum of the main stream via vortex generator

Single Vortex GeneratorUpon recognition of the fact that the mass and momentum fl

rate of the injection are not high enough to achieve the mixquality needed within an acceptable axial length, methods oflizing the momentum of the main stream have been investigaThe main requirements from a successful technique are aslows:

1 no recirculation zones or regions of low velocity along tmixing section where self-ignition may occur or the flamcan be attached

2 low pressure drop3 applicable both for gaseous and liquid fuel injection4 suitable geometry for cooling in case it is necessary5 simple and robust design

The requirements of low pressure drop and high safety agaflame in the mixing zone led exclusively to delta-wing type vortgenerators which can generate streamwise vortices withoutrecirculation zones. In order to satisfy other practical considations such as mechanical integrity and cooling, a tetrahedralometry, which consists of two half delta-wing side surfaces anfull-delta-wing upper surface has been chosen. A sketch ofdevice is given in Fig. 6, labeled with the principal parametwhich define the geometry.

A series of water model tests with this type of vortex generathave been carried out in order to optimize the strength ofvortices produced, the flow velocities downstream, and the psure drop. These tests have been conducted in a rectangular pglas channel. The ratio of channel height to VG heightH/h wasabove two in order to minimize the influence of channel wallsthe flow field. The measurement planes are perpendicular toflow direction and located atz/H50.1, 0.5, 1.0, 2.0, and 3.0. Thheight to width ratio of the vortex generator was fixed aslength to width ratio of the vortex generator is varied in fo

Fig. 6 The geometry of a single vortex generator element

JANUARY 2001, Vol. 123 Õ 43

e

d

stant

-r ofInof

discrete steps. These four steps are labeled as versions A: leless than width, B: length equal to width, C: length 50 percgreater than width, and D: length equal to twice the width.

Axial Velocity Distribution Along the Mixing Section. Allthree components of the flow velocity have been measure

Fig. 7 Normalized axial velocity „UÕU`… distributions at fivesuccessive planes downstream of vortex generator version A

44 Õ Vol. 123, JANUARY 2001

ngthnt

at

successive stations downstream of a vortex generator for convortex generator height~h! and width ~b!, as the length of thevortex generator~l! has been varied. Among all velocity components, the mean axial velocity component is a good indicatothe potential flame stabilization danger in the mixing zone.Figs. 7–10, a series of contour plots showing the distribution

Fig. 8 Normalized axial velocity „UÕU`… distributions at fivesuccessive planes downstream of vortex generator version B


f

oTt

cir-texnot

the mean axial velocity over transverse planes downstream ovortex generator are presented for all four geometries tested.horizontal axis length of the plots correspond to the centralpercent of the total channel width. The vertical axis of the plcovers the bottom 62.5 percent of the total channel height.velocity values shown are nondimensional, normalized withmean axial velocity.

Fig. 9 Normalized axial velocity „UÕU`… distributions at fivesuccessive planes downstream of vortex generator version C


theThe50tshehe

The data from the first measurement plane atz/H50.1 revealthat the first two geometries, namely A and B give rise to a reculation zone at this plane immediately downstream of the vorgenerator. The other two geometries, namely C and D doexhibit any negative or zero velocity zones.

Fig. 10 Normalized axial velocity „UÕU`… distributions at fivesuccessive planes downstream of vortex generator version D


n

x

e

t

eg

e

g

s ine asof

lanesveith

tio.en-annelnnelFig.ingle

Further downstream atz/H50.5, the recirculation zone downstream of version B is already closed, as the negative velocitieversion A are replaced by zero velocities. Evidently, all variatested exhibit positive axial velocities starting fromz/H51.0. Thegrowth rate of the region which is influenced by the vortex geerator is the largest for the version A. Atz/H53.0, the influenceof the vortex generator is visible over almost the entire chancross-section for this shortest variant tested as the region inenced by the longest one is still relatively small at this station

These results demonstrate that, for a givenh/b ratio, a mini-mum l /b ratio is necessary in order to prevent the formation orecirculation zone downstream of the vortex generator. The evalue of this critical parameter can be best determined after csidering the remaining two aspects of the design. The first onthe circulation, which is a measure of the vortex strength or,directly, the quality of mixing that can be achieved by a vortgenerator. The second one is the pressure drop caused bydevice, which influences the efficiency and power output ofwhole cycle.

Circulation Versus Pressure Drop. In addition to providingthe basis for the assessment of the potential danger of flamebilization in the mixing section, the velocity measurements mtioned above served to determine the strength of the vorticeserated by each vortex generator. Based on the velocity distribuover transverse planes downstream of the vortex generator, itpossible to determine the vorticity distribution, which is definedthe curl of the velocity vector, namely

vW 5¹W 3VW 5S d

dxeW x1

d

dyeW y1

d

dzeW zD3~UeW x1VeW y1WeW z!. (1)

As far as the strength of the longitudinal vortices is concernthe streamwise component of the vorticity vector is of main intest, which is defined as

vz5S ]V

]x2

]U

]y D . (2)

The partial derivatives in Eq.~2! can be approximated by usina central differencing scheme

]V

]x5

Vi 11,j2Vi 21,j

2Dx(3)

and

]U

]y5

Ui , j 112Ui , j 21

2Dy. (4)

Substituting~3! and ~4! in ~2! yields,

vz5S Vi 11,j2Vi 21,j

2Dx2

Ui , j 112Vi , j 21

2Dy D , (6)

wherei and j correspond to the coordinates of the grid point inxandy directions, respectively.

The circulation,G, as the value of the net vorticity over a regioof the flow, is defined by

G52 R VW •dsW. (7)

This line integral can be converted to a surface integralemploying the theorem of Stokes@6#, namely

G52E ES

~¹W 3VW !•dSW (8)

or


-s ofts

n-

nelflu-.

f aacton-e isin-xthis

he

sta-n-en-

tionwasas

ed,r-

n

by

G52E ES

vW •dSW . (9)

This surface integral can now be calculated in discrete steporder to calculate the circulation for each measurement planthe sum of streamwise vorticity multiplied with the surface areathe element

G52(j 51

M

(i 51

N

v i , jDxDy. (10)

This procedure has been repeated over all measurement pfor all four variants tested. The results are given in Fig. 11 for fisuccessive planes, in the form of the circulation calculated wthis method as a function of vortex generator length to width ra

Additionally, static pressure difference across the vortex gerators, as measured between a plane located at one chheight upstream and a second plane located at eight chaheight downstream of the vortex generator, are presented in12. These measurements have been carried out both with a selement and four identical elements positioned side-by-side.

Fig. 11 Circulation values calculated from the velocity mea-surements at five successive planes as a function of length towidth ratio of the vortex generator

Fig. 12 The pressure drop coefficient of a single and four vor-tex generators as the length to width ratio is varied


gr

ms

eFt

ie

ne

u

tn

t

r

g-for

ctedn-ting

vertialsted.g

fine-

ed.

orsn inghry

The dependence of pressure drop coefficient on the vortexerator length is monotonic, as one would expect. In other wothe pressure drop caused by a vortex generator increases stewith decreasing length. This can be attributed to the fact that wdecreasing vortex generator length, the portion of the dynapressure which can be recovered downstream is reduced asteeper angle of attack leads to higher transverse velocity comnents and eventually to a recirculation zone with zero and netive axial velocities.

On the other hand, the circulation distribution reaches a paround version C as the vortex generator length is reduced.ther reduction in the length after this point causes the circulato decrease. This also can be explained in the light of the velomeasurements which are in agreement with the previous stuon swirling flows and vortex breakdown. Studies on delta-wtype vortex generators have shown that for a fixed angle of swincreasing the angle of attack leads to breakdown position ofvortices to move upstream. In the extreme case of very high aof attack, the breakdown position is located at the leading edgthe vortex generator. After breakdown, part of the streamwvorticity is transformed into spanwise vorticity within the recirclation zone.

Based on this information, it can be concluded that fortetrahedral vortex generator geometry selected, an optimum leexists which generates the strongest vortices for a given widthheight. By selecting the geometry of the vortex generator atoptimum value, the most important requirement on the wayachieve the best mixing quality at the minimum pressure dprice is fulfilled.

Multiple Vortex Generators in a Duct SegmentOnce the geometry of a single vortex generator element is

fined for maximum vortex strength and for minimum pressudrop, the next step is the arrangement of these elements in asegment in a way which is compatible with the fuel injectio

Fig. 13 The geometry of two opposing vortex generators withcentral injection


en-ds,adilyithicthe

po-ga-

akur-

ioncitydiesngep,thegleof

ise-

hegth

andhistoop

de-reductn.

Ideally, the fuel is injected from a single location for each sement such that the same injection nozzle could be utilizedgaseous and liquid fuels.

The arrangement of the vortex generators has to be selewith this consideration in mind. Namely, the vortex pattern geerated by a specific arrangement should be capable of distribuboth gaseous and liquid fuel from a single injection location othe entire cross-section of the segment. A number of potencandidates for the arrangement have been considered and teA relatively quick and inexpensive method of determininwhether a special arrangement is suitable for the large andscale mixing purposes is the laser induced fluorescence~LIF! testsin water model. Here two example arrangements are present

A Pair of Opposing Vortex Generators with a Central FuelLance. In this arrangement, a pair of identical vortex generatare mounted on the opposing walls of the segment, as showFig. 13. The fuel lance is located at the center, injecting throufour holes which are pointing slightly away from the symmetaxis.

Fig. 14 The LIF pictures of fuel concentration over four suc-cessive planes downstream of a pair of opposing vortex gen-erators with central injection


i

t

h

t

n

l

u

hi s of

histhebe

e,perher to

n inthet thehele,

The LIF pictures associated with this case are presented in14. It is evident from these pictures that the large scale distrtion of the injected fuel is not complete even at three chanheight downstream from the injection. Given the fact thatlength of the mixing section in the machine is less than two chnel heights, it is difficult to achieve the emission targets with tconfiguration. Additionally, due to almost completely fuel-freouter regions, insufficient flame stabilization is to be expecwith such a configuration, since the recirculation zones aftersudden expansion into the combustor could not be efficieused.

Further attempts in the direction of optimization of the mixinquality with this configuration, by modifying the injection angand momentum of the fuel jets or the geometry of vortex genetors, brought only marginal improvements at the cost of increapressure drop and flashback danger.

Additional tests with the similar configuration, but with vortegenerators mounted on the upper and lower walls have prodsimilar results, namely, a good portion of the channel crosection not receiving enough fuel, resulting in large nonuniformties in mixing. As in the reference case mentioned earlier, wthe momentum of the main stream is not used to the extentneeded, problems with increased flashback danger arise du

Fig. 15 The geometry of two pairs of opposing vortex genera-tors with central injection

Fig. 16 The vortex pattern generated by two pairs of opposingvortex generators at the fuel injection plane


Fig.bu-nelhean-iseedthetly

gera-sed

xced

ss-i-ent ise tooverloading of the vortex generators by increasing the angleattack and sweep in order to increase the mixing quality. Tproblem can be avoided only by increasing the number ofvortex generators, namely from a pair to two pairs as willexplained in the next section.

Two Pairs of Opposing Vortex Generators with a CentralFuel Lance. This arrangement is similar to the previous onwith an additional pair of vortex generators mounted on the upand lower walls, as shown in Fig. 15. The axial positions of tends of the pairs are shifted with respect to each other, in ordeprovide the space needed for the central injector.

The vortex pattern generated by this arrangement is showFig. 16, composed of velocity components perpendicular tomean flow direction. This measurement has been carried out aaxial location of the tip of the fuel lance, in the absence of tlance itself. All four pairs of counter-rotating vortices are visib

Fig. 17 The LIF pictures of fuel concentration over four suc-cessive planes downstream of two pairs of opposing vortexgenerators with central injection


aw

i

itnt

i

dinng

n

r

t

s

ia

lta-a

rtexgthal

w orsec-t or

entm-

mix-id

h a

firstirr,avethe

Dr.inardMr.DA

6

990,es:90-

ndder

ion2

96,4/

each generated by one of the vortex generators. The pairs fromside vortex generators are clearly larger, in order to provide bepenetration and distribution of the fuel to both sides, whichdistanced further from the lance with respect to upper and loregions.

The LIF pictures of fuel concentration for this configuratioover four transverse planes downstream of the injection are gin Fig. 17. A dramatic increase in the quality of the mixingobserved from these pictures, when compared to the case wsingle pair of vortex generators. Already at one channel hedownstream from the injection plane, the large-scale distribuof the fuel is practically completed. Further downstream the fiscale mixing progresses, yielding locally more uniform concention distribution.

The LIF pictures from the same configuration in streamwplanes as observed from the side and plan views are given in18. As it is evident from these pictures, the injected fluid is rapitransported away from the injector and entrained into the vortshown in Fig. 16. Moreover, no significant impingement of ijectant to the side or top and bottom walls is visible, indicatithat, in the case of liquid fuel injection, the coke formation dandue to droplets contacting hot surfaces prior to evaporationpractically nonexistent. An additional advantage of this configration is that the fuel is transported away from the injector itially, to come back in the middle of the channel after the recculation zone in the wake of the lance is closed completely, thensuring no flashback danger due to fuel entrainment in the lawake.

The fuel concentration recordings from LIF pictures have beevaluated with digital image processing techniques in ordequantify the mixing quality. The data from all three cases reporare presented in Fig. 19, where the standard deviation in localconcentration,s, normalized with the mean concentration is ploted as a function of the axial distance from injection. Apparenin the absence of any vortical structures, other than those geated by the injection itself, the mixing progresses very slow. Afthe first channel height axial length, the reduction in the normized standard deviation is very gradual, reaching to a value of oabout 45 percent atz/H53. A single pair of vortex generatorbrings an improvement of about 12 percent in mixing qualityz/H53.0, when compared to the reference case. The case wpair of vortex generators shows the best value of 15 percent vtion coefficient atz/H53.0.

Fig. 18 The LIF pictures of fuel concentration over streamwiseplanes downstream of two pairs of opposing vortex generatorswith central injection


thetterreer

nvenisith aghtione-ra-

seFig.ly

ces-geris

u-i-ir-us,nce

ento

tedfuelt-ly,ner-teral-nly

atth aria-

SummaryA comprehensive investigation has been carried out with de

wing type vortex generators with the aim of fuel-air mixing inreheat combustor. This investigation produced an optimum vogenerator geometry, which provides the maximum vortex strenat the minimum pressure drop price. Owing to the longitudinvortices generated by this geometry, no zones with reverse flounacceptably low velocities are encountered along the mixingtion, thus ensuring reliable operation without flame attachmenflashback prior to complete mixing.

In addition to optimization of a single element, an arrangemof two opposing pairs of vortex generators in a rectangular cobustor segment is presented. This arrangement allows rapiding with a single, central injector, both for gaseous and liqufuels. The experience gained with this arrangement througnumber of high pressure tests@5# verified the feasibility of theconcept presented here. More recently, starting with theGT24 machine in Gilbert, NJ, and the first GT26 machine in BSwitzerland, a number of gas turbines with this technology hbeen commissioned into service, confirming the findings fromdevelopment tests.

AcknowledgmentsThe authors would like to acknowledge the cooperation of

Rolf Althaus, Dr. Jaan Hellat, Dr. Yansong Liu, Dr. Yau-PChou and Dr. Bettina Paikert. We are also grateful to Mr. RichStraessle for invaluable assistance with the experiments andBarend Jenje for his careful measurements and evaluation of Ldata.

References@1# Frutschi, H. U., 1994, ‘‘Advanced Cycle System With New GT24/GT2

Turbines—Historical Background,’’ ABB Rev.,1, pp. 21–25.@2# Sattelmayer, T., Felchlin, M., Haumann, J., Hellat, J., and Steyner, D., 1

‘‘Second Generation Low Emission Combustors for ABB Gas TurbinBurner Development and Tests at Atmospheric Pressure,’’ ASME PaperGT-162.

@3# Aigner, M., Mayer, A., Schiessel, P., and Strittmatter W., 1990, ‘‘SecoGeneration Low Emission Combustors for ABB Gas Turbines: Tests UnFull Engine Conditions,’’ ASME Paper 90-GT-308.

@4# Senior, P., Luturm, E., Polifke, W., and Sattelmayer, T., 1993, ‘‘CombustTechnology of the ABB GT13E2 Annular Combustor,’’ 20th CIMAC G21993, London.

@5# Joos, F., Brunner, P., Schulte-Werning, B., Syed, K., and Eroglu, A., 19‘‘Development of the Sequential Combustion System for the ABB GT2GT26 Gas Turbine Family,’’ ASME Paper 96-GT-315.

@6# Kuethe, A. M., and Chow, C.-Y., 1986,Foundations of Aerodynamics, JohnWiley and Sons, New York.

Fig. 19 Variation coefficient „standard deviation to mean con-centration ratio … as a function of axial distance


Vortex Generators in Lean-Premix Combustion - Thermothgr/gasturbiner/Material_for_generating_slide… · Vortex Generators in Lean-Premix Combustion A novel fuel-air mixing technique

Documents