A

ds(Bmcs©

K

1

taiaccsrgtab[

t

0d

Available online at www.sciencedirect.com

Chemical Engineering and Processing 47 (2008) 2209–2217

Effects of co- and counter-swirl on the droplet characteristicsin a spray flame

R. Hadef a,∗, B. Lenze b

a BP 297, Institut de Genie Mecanique, Universite Larbi Ben M’Hidi, 04000 Oum El Bouaghi, Algeriab Engler-Bunte-Institut, Lehrstuhl fur Verbrennungstechnik, Universitat Karlsruhe (TH), Kaiserstreet 12, 76128 Karlsruhe, Germany

Received 2 May 2007; received in revised form 29 November 2007; accepted 30 November 2007Available online 23 December 2007

bstract

This paper reports measurements of droplet characteristics and flow field in a spray flame with inner and outer swirling air streams. The spatialistribution of droplet characteristics produced by the burner’s airblast atomizer was measured using dual-phase Doppler anemometry (PDA). Thepray flame was operated near the lean blow-out limit at two flow conditions: co-swirling (flow rotation in the same direction) and counter-swirlingflow rotation in opposite directions). In both cases, the flame exhibited a U-shaped form and was marked by a large central recirculation zone.ased on the measurements of the droplet velocity components, differences between both configurations appeared for the counter-rotational setup

ainly in the near burner region, where the decrease of total swirl causes deeper penetration of the droplets from the inner duct into the combustion

hamber, resulting in a much more homogeneous distribution than the other one. The droplet size in terms of the Sauter mean diameter (SMD)hows little variation in the change of the direction swirl condition. Application of counter-swirl results in more turbulent droplet motion.

2008 Published by Elsevier B.V.

aami

ietwaiotis

eywords: Spray; Phase Doppler anemometry; Swirl; Flame; Airblast

. Introduction

Combustion of liquid fuel is encountered in many applica-ions such as gas turbines and diesel engines, industrial furnacesnd liquid-fuelled rocket engines. Swirl flows are widely usedn these applications. Because of performance requirementsnd their effect on the design of these devices, there is aonsiderable interest in identifying optimal swirl, geometri-al conditions and the details of the fuel injector to achievepecific practical goals. Numerous experiments in swirlingeacting flows have been carried out and have established theeneral characteristics of swirl flows. They revealed the impor-ant influence of swirl on decreasing emissions by offering

means to control the stability and intensity of the com-ustion as well as the size and shape of the flame region

1–3].

Most aero-engine and gas turbine applications in serviceoday use a prefilming airblast atomizer because of its good

∗ Corresponding author.E-mail address: rhadef@rocketmail.com (R. Hadef).

twtDOi[

255-2701/$ – see front matter © 2008 Published by Elsevier B.V.oi:10.1016/j.cep.2007.11.017

tomization characteristics, low liquid pressure requirementsnd potential for achieving significant reductions in soot for-ation and exhaust smoke [4]. In this atomizer the fuel supply

s sandwiched between two swirling air streams.The effect of the swirl combinations on spray character-

stics has been studied by several researchers. Through theirxperimental work, Aigner and Wittig [5] were able to showhat with a proper selection of swirl elements, particularlyith counter-swirl, the effectiveness of shear stresses on the

tomization edge can be increased, resulting in a finer atom-zation. Chin et.al. [6] explored the effect of the air swirlingrientation on the Sauter mean diameter (SMD) experimen-ally, and show that the least efficient atomizer configurations that of two counter-rotating air streams to atomize the liquidheet. Moreover, the lowest SMD values were achieved whenhe inner air stream was co-swirling and the outer air streamas counter-swirling with respect to the fuel swirling orienta-

ion. However, they used the atomization model proposed by

ombrowski and Johns [7] for two-dimensional liquid sheets.n the other hand, the application of counter-swirling stream

s better at stabilizing the flame than the co-swirling stream8–10]. Time-resolved measurements of temperature fluctu-

2 ring and Processing 47 (2008) 2209–2217

awficTatomrwopsc

hittadv

2

2

babcaed

psbfisaafttstaefiwct[a

Fa

dhacs

2

tattitwa

210 R. Hadef, B. Lenze / Chemical Enginee

tions and subsequent calculated thermal micro-time scalesithin an unconfined premixed flame in a double-swirl con-guration indicate a stronger radial exchange in case of theo-swirling burner comparing to the counter-swirling one [11].hree-dimensional laser dual-velocimetry measurements withinn isothermal flow of an airblast atomizer nozzle showed thathe flow field of the counter-swirl arrangement, compared to thatf the co-swirl configuration, exhibits a marked increase of theass flow recirculated in the internal recirculation zone and a

eduction of its length in the axial direction [12]. Following thisork, several studies [13–17] continued to investigate the effectsf the gas swirling orientation on the spray dispersion pattern ofrefilming airblast atomizers. However, the mechanism of thepray dispersion in such atomizer configurations has not stillompletely understood and thus further research is necessary.

Therefore, the present research focuses on investigatingow the swirl configuration (co- and counter-swirling) keep-ng all other parameters fixed, affects the spray dispersion inhe present geometry. Measurements of droplets characteris-ics were obtained non-intrusively using the dual-phase Dopplernemometry (PDA) technique. Data are presented for the meanroplet size, velocity components, number density and axialolume flux of fuel droplets within the spray flame.

. Experimental setup

.1. Atomizer

The experiments were carried out using an atmospheric air-last nozzle in a cylindrical combustion chamber with opticalccess for laser diagnostics. A sketch of the experimental com-ustion burner under investigation is given in Fig. 1a. The nozzleonsisted of a modular arrangement of two radial swirl gener-tors, an atomizer lip which separates the two air streams fromach other within the nozzle, and an air diffuser with a throatiameter of D0 = 25 mm.

The liquid fuel was pressurized to 8 bar and injected by a sim-lex atomizer, which spread a thin wavy film onto the prefilmingurface, interfacing the inner flow. This fuel film was transportedy shear forces to the atomization edge. Upon arriving at the pre-lmer lip, the liquid exited the airblast atomizer as an annularheet which was then subjected to high-speed swirling innernd outer air streams. The liquid sheet–air interaction gener-lly produces waves that become unstable and disintegrate intoragments (Fig. 1b). Under favourable conditions the oscilla-ions are amplified, these fragments become ligaments and, inurn, break-down into droplets by the shearing effect of the high-peed swirling air streams on both sides. This process is referredo as primary atomization and involves the action of pressure,erodynamic, centrifugal, surface tension forces, and internalffects such as turbulence and those arising from velocity pro-le relaxation. The droplets were then transported and mixedith air by the swirl, and delivered to the primary zone of the

ombustion chamber. The presence of the atomizing air leadso shorter break-up length [18] and enhances liquid–air mixing19]. It is well known that the initial thickness of the dischargednnular liquid sheet as it leaves the atomizer is the most critical

mtet

ig. 1. (a) Presentation of the burner. (b) Double-swirler-Fuel atomizationssembly. (c) Optical arrangement of a dual-mode phase Doppler system.

imension; it determines mean drop size in the fuel spray andas direct impact on combustion efficiency, pollutant emissionsnd combustion instability [4]. Theory predicts, and experimentonfirms, that the mean drop size is roughly proportional to thequare root of the sheet thickness [4].

.2. Operating conditions

Co- and counter-swirling nozzles were created by replacinghe secondary swirl generator by one which offers the samemount of swirl, but in the opposite direction. The design ofhe nozzle fitting allows the mass flow rate and the preheatemperature of primary and secondary airflow to be adjustedndependently of each other. The fuel, liquid kerosene, was fedhrough an atomizer located on the burner centreline. The sprayas ignited by a high voltage spark and the flame was confined incylindrical combustion chamber (i.d. 10 cm). Limitation of the

ain reaction zone was provided by an orifice with 40% diame-

er reduction placed at the exit of the combustion chamber. Thentire burner was fixed and the optical setup was mounted on ahree-axis stepper motor traverse system.

ring and Processing 47 (2008) 2209–2217 2211

uoiflvwcλ

ecnk

Sn

S

wtubtst[aasS

2

uc(oldtbt

Fig. 3. Spatial distribution of droplets number con centration [droplets/cm3].

R. Hadef, B. Lenze / Chemical Enginee

The air stream through the nozzle was electrically preheatedp to 200 ◦C and split between the inner circular passage and theuter annular canal. The splitting of the air stream is character-zed by the parameter μ, defined as the ratio of the inner air massow to the outer one: μ = Min/Mout. For the current study, thealue of μ was 0.6. The total mass flow rates of air and keroseneere MA = 85.17 kg/h and MF = 2.5 kg/h, respectively. This

orresponds to a thermal load of 30 kW and an air-to-fuel ratio= 2.5 which is very close to the lean blow-out limit, measured

xperimentally as λ = 2.7 [20]. The Reynolds number is cal-ulated as the product of the axial average air velocity at theozzle exit and the throat diameter of the diffuser divided by theinematic viscosity of air and yields approximately 46,200.

The swirl strength is characterized by the outlet swirl number0, defined as the ratio of angular to axial momentum flux in theozzle divided by the outside nozzle radius R0 as given by:

0 =∫ R0

0 ρ(U Wr + u′w′r)r dr

R0∫ R0

0 ρ(U2 + u′2)r dr

here ρ is the fluid density, r is the radial coordinate, U and W arehe axial and tangential velocity components respectively, and′and w′ are their velocity fluctuations. The value of S0 shoulde calculated from measured values of velocity profiles. Dueo the lack of detailed velocity measurements, the theoreticalwirl number Sth calculated from only the geometrical data ofhe swirl generator is used since it is approximately equal to S021]. In the present work it is of the inner airflow Sth,in = 0.46nd of the outer airflow Sth,out = 1.0 in the case of the co-swirlnd Sth,out = 0.85 at the counter-swirl. The value of resulting totalwirl number is Sth,out = 0.80 for the co-swirl configuration andth,total = −0.40 for the counter-swirl configuration.

.3. Measurement technique

Measurements of droplet size and velocity were conductedsing a commercial DANTEC dual-PDA system. The systemombines a conventional and a planar particle dynamics analyzerFig. 1c) and was developed to improve measurement accuracyf mass flux and concentration [22]. Due to its higher reso-ution, the size measurement was performed using the phase

ifference of the conventional PDA. The planar PDA is usedo resolve the 2π ambiguity and for validation. The latter isased on comparison of the droplet diameters measured byhe planar and conventional PDA. As both configurations are

Fig. 2. Flame form (a) and streamlines (b).Fig. 4. Radial and axial droplet size distribution for both swirl configurations(©) co-swirl (�) counter-swirl.

2 ring

an[

iiosbavptwrw

m2tvocbasauaw

Fd(

212 R. Hadef, B. Lenze / Chemical Enginee

ffected differently by the slit effect, the trajectory effect, andon-spherical droplets, erroneous measurements are rejected23].

The system used an argon-ion laser (INNOVA 70) operat-ng at an output power of 200 mW. The laser beam was steerednto the fiber drive which separates the beam into two beamsf equal intensity. The beams are subsequently directed intoingle-mode fibers, which steer the light to the transmitter. Oneeam of each colour was frequency-shifted by 40 MHz by usingBragg cell for the purpose of direction recognition of the

elocities. The polarization direction was adjusted to be per-endicular to the scattering plane. In the present experiment,wo components of the velocity were measured simultaneously

ith two beam pairs with wavelengths of 514.5 and 488 nm,

espectively. A 30◦ off-axis forward scattering configurationith a 400 mm focal length receiving lens was selected. The

tar

ig. 5. (a) Axial droplets mean velocity distribution at both swirl configurations ( ) 1istribution at both swirl configurations ( ) 10–20 �m ( ) 30–40 �m ( ) total droplet

) 10–20 �m ( ) 30–40 �m ( ) total droplets.

and Processing 47 (2008) 2209–2217

easurement volume was approximately 53 �m in diameter andmm in length. Due to the lens on the receiving probe a por-

ion of approximately 200 �m is cut out of the measurementolume. The light scattered by the droplets was transmitted byptical fibers to photodetectors, where the light signals wereonverted into electronic signals. They were further processedy a real-time signal analyzer and the data were collected byn acquisition board installed in a personal computer. The mea-urements include size, velocity components number densitynd turbulent kinetic energy of the droplets as well as the liq-id volume flux. For each measurement location, 5000 samplesre recorded to allow a determination of the mean propertiesith low statistically error. No measurement was made when

he data rate was below 5 Hz. In addition, measurements belown axial height of 11 mm were not possible due to optical accessestrictions.

0–20 �m ( ) 30–40 �m ( ) total droplets. (b) Radial droplets mean velocitys. (c) Tangential droplets mean velocity distribution at both swirl configurations

R. Hadef, B. Lenze / Chemical Engineering and Processing 47 (2008) 2209–2217 2213

(Cont

3

3

flasoflvsaRobgar

3

ps(tcddo

StR

Fig. 5.

. Results

.1. Flame shape

Fig. 2a displays a photograph of the investigated flame. Theame is U-shaped due to a low value of the parameter μ, i.e. theirflow is mainly provided through the outer duct with strongwirl. Fig. 2b shows pseudo-streamlines in the z–r plane basedn the mean axial and radial velocity components for a stableame. The flow field is in the form of a one-celled great toroidalortex caused by the adverse pressure gradient induced by thewirl and occupying the entire burner volume. Exhaust gas prop-gates radially outwards, setting up a wall jet at the enclosure.ounding and smoothing of wall surfaces enhance the formationf wall jets. This recirculation zone primarily aids in flame sta-

ilization by providing an aerodynamic blockage and reducingas velocities necessary to stabilize a flame. It also transports hotnd chemically active combustion species from the downstreamegion of the flame to the root of the flame.

cF

i

inued )

.2. Droplet characteristics

The spatial distribution of droplet number concentration (pro-ortional to the droplet number density) is shown in Fig. 3. Ithows that the counter-swirl combinations of the swirler pairinner and outer swirlers) result in a larger droplet popula-ion in the inner region of the spray than the correspondingo-swirl case. The counter-swirl case also tends to introduceroplets into the inner region of the spray, especially in theownstream region close to z = 15 mm, compared to the co-swirlne.

The droplet size of interest in combustion flows is theauter mean diameter, which represents the ratio of the

otal volume to the total surface area of the spray droplets.adial and axial distributions of the SMD values for both

onfigurations on four measurement planes are shown inig. 4.

For both cases, the largest droplets are measured roughlyn the wake of the atomizer lip, due to poor secondary atom-

2214 R. Hadef, B. Lenze / Chemical Engineering and Processing 47 (2008) 2209–2217

(Cont

icoag0ormdslfltos[t

lsewrorecdertri

Fig. 5.

zation. With increasing radial distance, as well as near theombustor exit, a considerable increase in droplet diameter isbserved. Only large droplets, due to their high inertia, areble to penetrate the region of hot recirculating combustionases, so that a growth of relative concentration of droplets with< SMD < 50 �m within the radial range from 0 < r < 10 mm isbserved. As will be shown later on, this maximum matchesegions of very low liquid volume flux. Due to the swirlingotion of the gas flow and to resulting centrifugal forces, larger

roplets are separated from the smaller ones and cause a con-iderable increase in droplet size at larger radii. Due to theirarger surface to volume ratio, very small droplets evaporateaster, resulting in a less pronounced dependence of SMD onocal measurement position with increasing axial distance fromhe atomizer. This trend agrees well with the previous result

btained at elevated pressure [24] where the qualitative sheettability behaviour is similar to that at atmospheric pressure25]. However, the most noticeable result is that the SMD ofhe droplets in the counter-swirl configuration is found to be

ift

inued )

ower than in the co-swirl configuration, particularly down-tream of the atomizer edge. This is due the high shear forcesxerted by the two opposing airstreams on the liquid filmhich support efficient atomization, in addition to the very

apid decay of the air swirl component that reduces the chancesf droplet coalescence. On the other hand, since the growthate of disturbances on the liquid–air interface (which gov-rns the characteristics of the resulting spray) is higher in theo-inner/counter-outer air stream combination [25], it likely pro-uces the finest spray. This behaviour is consistent with thearlier measurements [6]. Also, when the swirl orientation iseversed, the change of annular liquid sheet velocity profile athe atomizer exit contributed to this distinction as it has beeneported in the recent theoretical analyses and experimental val-dations [26–32].

Smaller droplets vaporize faster than larger droplets, result-ng in faster fuel vapour mixing with air, which inducesull vaporization of the spray at the burner exit plane (Dueo lack of droplets, measurements were not made above

ring and Processing 47 (2008) 2209–2217 2215

au

icf(ntfivUe

iiddcdtdvdlla

cfdtc

fiittdittussmtF

iipratt

FC

tdlfivnt

gcabgw

R. Hadef, B. Lenze / Chemical Enginee

n axial height of 35 mm in the counter-swirling config-ration).

In the present study, size-dependent droplet behaviour wasnvestigated to increase the understanding of the spray dispersionharacteristics mentioned above. In the following, the resultsor the droplet velocity components in two different size classes10–20 and 30–40 �m) are also presented (Fig. 5). For conve-ience, they are called 15 and 35 �m size classes, or simplyhe smaller and the larger droplet classes, respectively. The floweld of the droplets is described in Fig. 5 through the meanelocity in axial, radial and tangential directions denoted by, V, and W, respectively, as well as by their turbulent kinetic

nergy, k.Generally, all profiles of the mean axial velocity show the typ-

cal behaviour of jets, i.e., a linear increase with radial distancen the centre region to reach a maximum value, followed by aecrease in the outer periphery of the spray. As the flow divergesownstream, this maximum decreases. Compared to the co-swirlonfiguration the flow field of the counter-swirl arrangement, theroplet mean axial velocity component is higher in the centre ofhe burner. This observation is more pronounced for the smallerroplets because these follow the gas flow more closely. Pre-ious measurements have shown a moderately higher velocityownstream of the nozzle exit region [33]. On the other hand, thearger droplets move in to this region more slowly, leading to aonger residence time in the combustion zone and consequentlymore rapid vaporization.

Examination of the droplet mean radial velocity componentonfirms the large spreading of the jet in the radial directionor this configuration and the inward motion of the droplets justownstream of the nozzle opening within the central recircula-ion zone within the radial range from 0 < r < 10 mm up to theiromplete evaporation, i.e. at z = 30 mm.

The radial profiles of the mean tangential velocity componentor the droplets emanating from the inner duct of the atom-zer are identical, since the geometry of the primary swirlers kept constant in both cases. The radial profiles of tangen-ial velocity caused by the outer swirler are expected to be ofhe same shape, but have opposite sign. With growing axialistance, the tangential mean velocity decreases substantiallyn the central region, without any change in the periphery ofhe spray for the co-swirl configuration. Further downstream,he effect of inlet conditions vanishes and the profile becomesniform with a complete disappearance in the counter-rotatingwirl. The non-zero values recorded near the axis arise fromome instability of the central recirculation zone whose centreoves slowly around the axis [34] as suggested by the higher

urbulent kinetic energy values measured on the axis, visible inig. 6.

The turbulent kinetic energy radial profiles (Fig. 6) show sim-lar behaviour for both cases. They are characterized a high peakn the centre region of the inlet combustion chamber due to theroduction of turbulence is the shear layer caused by the strong

ecirculation zone where turbulence eddies are highly energeticnd anisotropic. Farther downstream, the turbulence level beginso decay owing to the combination effect of swirling motion andhe wall friction. Firstly, the swirling motion generates a cen-

([tb

ig. 6. Radial and axial distribution of the droplets turbulent kinetic energy (©)o-swirl (�) Counter-swirl.

rifugal force that drives the flow to move outwards in the radialirection; the stronger outward radial flow transports the turbu-ent kinetic energy in the radial direction towards the wall faster;nally, turbulence near the wall is dissipated quickly by the walliscous friction. The production of turbulence near the walls isot significant since the velocity and the velocity gradient nearhe wall are low.

However, the most noticeable finding is that the dropletsenerally exhibit more significant turbulent motion in theounter-swirl case than in the co-swirl case. This improves their-fuel mixing which results in a uniform temperature distri-ution following combustion without significant temperatureradient, therefore reducing the formation of local hot spotshere the thermal formation of harmful oxides of nitrogen (NOx)

exponentially dependent on the temperature) can be severe35,36]. On the other hand, it has been analytically demonstratehat the ratio of NOx produced to the total mass of fuel dropletsurned is an increasing function of the droplet size (i.e., finer

2216 R. Hadef, B. Lenze / Chemical Engineering

Ft

seo

osasopSoh

Fdddsrfvbtcmd

4

stw

pl

temhl

ab

A

tt

R

ig. 7. Spatial distribution of Sauter mean diameter and vector velocity of theotal droplets.

prays produce less NOx) [37]. This finding has been confirmedxperimentally [38–41] and an excellent review has been writtenn this subject by Sirignano [42].

Fig. 7 displays a flow field of droplets as a vector plot overlaidn the SMD and the axial liquid flux. Every vector represents theum of the local axial and radial mean velocity components atll measured positions. This figure clearly shows a radial expan-ion of droplets to the wall for both configurations. The impactf counter-rotating swirl is twofold. Firstly, it causes a moreronounced presence of the droplets in the inner central zone.econdly, the droplets are completely evaporated at the heightf z = 40 mm. This is due to their smaller size and perhaps to theigher gas temperature.

Axial liquid volume flux measurements are also reported inig. 8. As expected, it is closely related to the flow field ofroplets, resulting in an extremely wide cone angle of sprayispersion. On radii R > 20 mm the cone angles of liquid fluxistribution of co- and counter-swirl show good agreement. Atmaller radii, application of counter-swirl exhibits a marked seg-egation of the spray into a second region of smaller cone angleormed mainly by small drops. This is due to the higher gas phaseelocity and an increased penetration depth of the circular nearurner zone of positive axial velocity, which convects preferen-

ially small droplets into the combustion chamber. This finding isonfirmed by the comparison of the measurements of the axialean droplets velocity carried out in both configurations and

isplayed in Fig. 5a.

Fig. 8. Spatial distribution of mean axial liquid flux of the total droplets.

[

[

[

[

[

and Processing 47 (2008) 2209–2217

. Conclusions

The behaviour of droplets in a spray flame formed by twowirling annular jets has been studied using the dual-phase PDAechnique. Two swirl configurations, co- and counter-swirling,ere compared in the same operating conditions.The counter-swirl configuration is characterized by a higher

opulation of droplets near the burner axis. This observation isikely responsible for the good flame stability [33].

The results indicate that atomization is finer and more spa-ially dispersed in the counter-swirl configuration. The firstffect is due to the break-up phenomena of the liquid liga-ents exiting the lip atomizer, and the second results from the

igh additional turbulence level generated in this configuration,eading to enhanced fuel–air mixing.

This advantage leads to easier ignition, a wider burning range,nd lower pollutant emissions, which helps in improving com-ustion efficiency.

cknowledgements

The authors gratefully acknowledge the financial support ofhe BMBF for funding during the KEROMIX program and ofhe DAAD for the award of a fellowship to R. Hadef.

eferences

[1] J.H. Kwark, Y.K. Jeong, C.H. Jeon, Y.J. Chang, Effect of swirl intensityon the flow and combustion of a turbulent non-premixed flat flame, FlowTurbul. Comb. 73 (2004) 231–257.

[2] S. Hoffmann, B. Lenze, H. Eickhoff, Results of experiments and models forpredicting stability limits of turbulent swirling flames, J. Eng. Gas Turbines120 (1998) 311–316.

[3] A.K. Gupta, D.G. Lilley, N. Syred, Swirl Flows, Abacus Press, TurnbridgeWells, 1984.

[4] A.H. Lefebvre, Atomization and Sprays, Hemisphere, New York, 1989.[5] M. Aigner, S. Wittig, Swirl and counter-swirl effects in prefilming airblast

atomizers, J. Eng. Gas Turbine Power 110 (1988) 105–110.[6] J.S. Chin, N.K. Rizk, M.K. Razdan, Effect of inner and outer Airflow

characteristics on high liquid pressure prefilming airblast atomization, J.Propul. Power 16 (2000) 297–301.

[7] N. Dombrowski, W.R. Johns, The Aerodynamic instability and disintegra-tion of vicious liquid sheets, Chem. Eng. Sci. 18 (1963) 203–214.

[8] A. Ateshkadi, V.G. McDonell, S.G. Samuelsen, Effect of hardware geom-etry on gas and drop behaviour in a radial mixer spray, Proc. Comb. Inst.27 (1998) 1985–1992.

[9] K. Merkle, H. Buchner, N. Zarzalis, N.O. Sara, Influence of co- and counterswirl on lean stability limits of an airblast nozzle, in: Proceedings of ASMETurbo Expo 2003, GT2003-38004, 2003.

10] A.K. Gupta, M.J. Lewis, M. Daurer, Swirl effects on combustion char-acteristics of premixed flames, J. Eng. Gas Turbine Power 123 (2001)619–626.

11] M.D. Durbin, D.R. Ballal, Studies of lean blow-out in a step swirl combus-tor, J. Eng. Gas Turbine Power 118 (1996) 72–77.

12] K. Merkle, H. Haessler, H. Buchner, N. Zarzalis, Effect of co- and counter-swirl on the isothermal flow- and mixture-field of an airblast atomizernozzle, Int. J. Heat Fluid Flow 24 (2003) 529–537.

13] K. Matsuura, Y. Kurosawa, Effect of swirl combinations on spray dispersion

characteristics of a multi-swirler airblast atomizer, Proc. ILASS (2005).

14] J. Colby, S. Menon, J. Jagoda, Flow field measurements ina counter-swirling spraycombustor, in: Proceeding of the 41stAIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, PaperNo.AIAA 2005-4143, 2005.

ring

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

R. Hadef, B. Lenze / Chemical Enginee

15] R. Kumara Gurubaran, R.I. Sujith, S.R. Chakravarthy, Flow field mea-surements of aprefilming airblast atomizer, in: Proceeding of the 41stAIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, PaperNo. AIAA 2005-4148, 2005.

16] M.R. Soltani, K. Ghorbanian, M. Ashjaee, M.R. Morad, Spray character-istics of a liquid–liquid coaxial swirl atomizer at different mass flow rates,Aerospace Sci.Technol. 9 (2005) 592–604.

17] R. Hadef, B. Lenze, Measurements of droplets characteristics in a swirl-stabilized spray flame, Exp. Therm. Fluid Sci. 30 (2005) 117–130.

18] I.S. Carvalho, M.V. Heitor, Liquid film break-up in a model of a prefilmingairblast nozzle, Exp. Fluids 24 (1998) 408–415.

19] G. Lavergne, P. Trichet, P. Hebrard, Y. Biscos, Liquid sheet disintegra-tion and atomization process on a simplifed airblast atomizer, J. Eng. GasTurbine Power 115 (1993) 461–466.

20] R. Hadef, K. Merkle, B. Lenze, W. Leuckel, An Experimental study ofairblast atomizer spray flames, J. Inst. Energy 74 (2000) 50–55.

21] W. Leuckel, Swirl intensities, Swirl types and energy losses of differentswirl generating devices. IFRF-Doc. No G02/a/16(1967), Ijmuiden.

22] C. Tropea, T.-H. Xu, G. Grehan, F. Onofri, G. Grehan, P. Haugen, M.Stieglmeier, Dual-mode phase-Doppler anemometry, Part. Part. Syst. Char.13 (1996) 165–170.

23] H.-E. Albrecht, M. Borys, N. Damaschke, C. Tropea, Laser-Doppler andPhase-Doppler Measurement Techniques, Springer-Verlag, Heidelberg,2003.

24] M. Brandt, M. Rachner, G. Schmitz, An experimental and numerical studyof kerosine spray evaporation in a premix duct for gas turbine combustorsat high pressure, Comb. Sci. Technol. 138 (1998) 313–348.

25] A.A. Ibrahim, M.A. Jog, Effect of liquid and air swirl strength and relativerotational direction on the instability of an annular liquid sheet, Acta Mech.186 (2006) 113–133.

26] J. Shen, X. Li, Instability of an annular viscous liquid jet, Acta Mech. 114

(1996) 167–183.

27] J. Shen, X. Li, Breakup of annular liquid jets in two gas streams, J. Propul.Power 12 (1996) 752–759.

28] Y. Liao, S.M. Jeng, M.A. Jog, The effect of air swirl profile on the instabilityof a viscous liquid jet, J. Fluid Mech. 424 (2000) 1–20.

[

[

and Processing 47 (2008) 2209–2217 2217

29] M. Gavaises, C. Arcoumanis, Modelling of sprays from high pressure swirlatomizers, Int. J. Engine Res. 2 (2001) 95–117.

30] A.T. Sakman, M.A. Jog, S.M. Jeng, M.A. Benjamin, Parametric study ofsimplex fuel nozzle internal flow and performance, AIAA J. 38 (2000)1214–1218.

31] J. Xue, M.A. Jog, S.M. Jeng, Effect of geometric parameters on simplexatomizer performance, AIAA J. 42 (2004) 2408–2415.

32] A.A. Ibrahim, M.A. Jog, Effect of liquid swirl-velocity profile instabil-ity of a swirling annlar liquid sheet, Atomization Sprays 16 (2006) 237–264.

33] M. Merkle, A. Ament, B. Lenze, W. Leuckel, R. Hadef, Untersuchungenzum stabilitatsverhalten einer airblast-zerstauberduse, VDI Berichte 1492(2000) 449–454.

34] P. Wang, X.S. Bai, M. Wessman, J. Klingmann, Large eddy simulation andexperimental studies of a confined turbulent swirling flow, Phys. Fluids 16(2004) 3306–3324.

35] T. Terazaki, S. Hayashi, The effects of fuel-air mixing on NOx forma-tion in non-premixed swirl burners, Proc. Comb. Inst. 26 (1996) 2733–2739.

36] A.H. Lefebvre, The role of fuel preparation in low emission combustion,J. Eng. Gas Turbine Power 117 (1995) 617–654.

37] F.V. Bracco, Nitric Oxide Formation in droplet diffusion flames, Proc.Comb. Inst. 14 (1973) 831–842.

38] A.A. Nizami, N.P. Cernansky, NOx formation in monodisperse fuel spraycombustion, Proc. Comb. Inst. 17 (1978).

39] L.P. Cooper Effect of Degree of Fuel Vaporization Upon Emissions for aPremixed Partially Vaporized Combustion System, NASA Technical Paper1582, 1980.

40] S.A. Drennan, V. Mandayam, Low NOx experiences firing residual oilin industrial boilers, in: Paper presented at the 1997 AFRC InternationalSymposium, Ottawa, 1997.

41] J.A. Nicholls, C.W. Kauffman, D.G. Pelaccio, D.R. Glass, J.F. Driscoll, Theeffect of fuel sprays on emissions from a research gas turbine combustor,Comb. Sci. Tech. 23 (1980) 203–213.

42] W.A. Sirignano, Fuel droplet vaporization and spray combustion theory,Prog. Energy Combust. Sci. 9 (1983) 291–322.