Quantitative imaging of OH concentrations in a
swirling methane jet flame via single-pulse
laser-induced predissociative fluorescence
Yei-Chin Chao Der-Chyun Wu
National Cheng Kung University
Institute of Aeronautics and Astronautics Tainan, Taiwan, 701
E-mail: ycchao@mail.iaa.ncku.edu.tw
Tsarng-Sheng Cheng Chung Hua University
Department of Mechanical Engineering Hsin-Chu, Taiwan, 300
Abstract. A quantitative imaging method using 2-D single-pulse laser-induced predissociative fluorescence (LIPF) of OH concentrations is de-veloped to study the flame structure in a swirling methane jet flame. A narrowband tunable KrF excimer laser is used to excite theP2(8)
rota-tional line of the A2⌺←X2⌸(3,0) transition at ⫽248.46 nm. Though
this transition produces a relatively weak signal, LIPF is much less sen-sitive to collisional quenching in atmospheric flames. Therefore, it is suit-able for generating quantitative data. OH concentration data are ob-tained by careful calibration against flat flame burner data of known fuel-air equivalence ratios using an identical optical setup. Because the distribution of OH concentration has a good correspondence with the flame, the measured 2-D imaging of OH indicates the instantaneous shape of the reaction zone. In the upstream section of the swirling flame, combustion is found to take place in three regions: the shear layer of the fuel jet, the recirculation vortex inside the recirculation bubble, and the thin layer between the recirculation zone and the ambient air. High OH concentration is found in the upstream central portion inside the fuel region. The temperature and radicals of the recirculated hot products appear to accelerate the initial decomposition and radical-generating processes after strong turbulent mixing with fresh fuel and air. This is believed to be superequilibrium OH, because its intensity is higher than that from the recirculated burnt gas and the measured local temperature is low (less than 1200 K). The measured OH concentration structure is strongly influenced by the characteristic swirling flow and flame struc-tures and is also closely related to the NOxformation and flame stabili-zation in the swirling flame. © 2000 Society of Photo-Optical Instrumentation Engi-neers. [S0091-3286(00)00206-3]
Subject terms: swirling jet flames; OH concentration; single pulse; laser-induced predissociative fluorescence; imaging.
Paper 990255 received June 25, 1999; revised manuscript received Nov. 10, 1999; accepted for publication Nov. 16, 1999.
1 Introduction
As modern technology advances, an adequate supply of high-quality energy is of vital importance to most cutting-edge industries. Electricity is considered the most suitable form of high-quality energy. International industrial com-petitiveness heavily depends on their ability to efficiently and reliably secure electric energy resources. To date, most electric energy is obtained through combustion of fossil fuels. Stringent environmental regulations imposed by most industrialized countries drive advanced combustion re-search for clean fuel use. The design of advanced combus-tion systems and processes for the clean use of fossil fuels can be greatly enhanced by the utilization of accurate com-bustion measurements and system parameter control during system operation. Control of various combustion param-eters is essential in determining the intrinsic balance be-tween combustion efficiency and pollution emissions. Laser-based combustion diagnostic techniques can provide nonintrusive, in situ, high-spatial- and temporal-resolution measurements of important chemical parameters. In
gen-eral, there are three techniques that have received substan-tial attention in gas-phase combustion diagnostics, i.e., laser-induced fluorescence 共LIF兲 spectroscopy, coherent anti-Stokes Raman spectroscopy共CARS兲, and spontaneous Raman scattering. The LIF techniques are used to measure temperature and trace species with number densities as low as 1011/cm3. The main disadvantage of LIF is the cence quenching by numerous collisions during the fluores-cence lifetime共varied between 0.1 ns and 10s兲. The prob-lem of measuring the collisional quenching rate required to correct the fluorescence signal can be circumvented by us-ing, for example, saturated LIF 共see Eckbreth1兲 or laser-induced predissociative fluorescence共LIPF兲, where the ab-sorption and emission rates or the predissociation rate are much greater than the collisional quenching rate. CARS techniques are best suited for temperature and major spe-cies concentration measurements in particle-laden combus-tion environments. CARS signals discriminate well against background laser-induced interference such as fluorescence and incandescence and can be orders of magnitude stronger
than the spontaneous Raman signal, because it is coherent and upshifted in frequency to the anti-Stokes side of the pump laser frequency. In the two described techniques, only one species can be monitored at a time unless several beams are introduced. Spontaneous Raman scattering has the unique advantage of being able to provide instanta-neous, spatially resolved measurements of temperature and all the major species simultaneously using only a single laser beam. Such simultaneous measurements are particu-larly important in nonpremixed combustion. Spontaneous vibrational Raman scattering can provide quantitative si-multaneous measurements even with a single laser pulse. In his monograph, Eckbreth1 describes details of these ad-vanced laser techniques applicable to combustion environ-ments. Recently, a combined diagnostic technique of Ra-man, Rayleigh and LIF of the OH radical is proposed by Dibble et al.2and is extended to include NO concentration by Barlow et al.3 and Nguyen et al.4 With the recent ad-vances of high power lasers, a Raman technique in the UV range, where the signal is stronger, has been developed by Cheng et al.5
Swirl is widely used in gas turbine combustors as well as industrial burners to increase fuel-air mixing and to im-prove flame stabilization. As swirl is introduced to the flow, the tangential component of the swirling flow enhances the turbulent mixing of fuel and air and the swirl-induced re-circulation stabilizes the flame. In general, if the swirl strength is strong enough, as characterized by a nondimen-sional swirl number S, an internal recirculation zone could be established that recirculates hot product gases from downstream and serves as a reservoir of heat and chemical radicals to enhance flame stabilization. The existence and shape of this recirculation zone strongly influences flame stability. Chigier et al.6showed that the blow-off velocity is increased by more than a factor of 7 relative to an un-swirled flame when the circulation is increased to 1.5 m2/s. Furthermore, for strongly recirculating flows, the oscilla-tion of the forward stagnaoscilla-tion point of the recirculaoscilla-tion zone causes unsteady motion of the swirling flow structure as shown, for example, by Gouldin et al.7and Chao et al.8 In general, the tangential flow component and the oscilla-tory recirculation zone characterize the complicated un-steady 3-D swirling flow structure. In the swirl flame, com-bustion will further complicate the dynamic behavior and the resultant flame structure will be hard to observe and measure using conventional probes. Laser diagnostic tech-niques become the major tools for combustion measure-ments, especially to resolve complicated combustion phe-nomena. Although swirling flows have been extensively used in combustor design, the swirling flame structure and its stabilization characteristics still remain a subject of cur-rent research. A series of research efforts have been at-tempted to characterize the swirl flame structure by Chen and Driscoll9,10and Chen et al.11 The formation of NOxis
known to strongly relate to the combustion temperature and the mixing level of cold fresh combustible mixtures with hot gas products, especially for nonpremixed flames. Clay-pole and Syred12 and Chen et al.11 found that the swirl could significantly reduce NOxemissions relative to a non-swirl simple jet flame. The recent laser measurement results by Tacke et al.,13 Fru¨chtel et al.,14 and Holza¨pfel et al.15 clearly point out that swirl-induced rapid mixing,
intermit-tency and finite-rate chemistry effects result in a tempera-ture well below the adiabatic value at the upstream end of the recirculation zone. Pitz et al.16 simultaneously mea-sured the major and minor共OH兲 species using laser Raman and LIPF techniques in turbulent swirl flames. While the mentioned pointwise laser techniques provide useful insight into the flame characteristics, these pointwise results rely heavily on probability density function 共pdf兲 and scatter plots to draw useful conclusions and the conclusions are usually qualitative in nature for turbulent flames.
Imaging of reactive scalars in turbulent flames is helpful to resolve some of the important questions in combustion. The spatial structure of reaction zones and the existence of either thin flamelet zones or broad distributed zones are of primary importance in modeling turbulent flames. Planar laser-induced fluorescence 共PLIF兲 of OH has been widely used to temporally and spatially resolve the flame structure of turbulent flames.17PLIF is a very versatile technique for flame visualization. It can be performed with single-pulse measurements in the nanosecond regime, effectively freez-ing both flow and chemistry in most atmospheric pressure combustion environments. Using either normal18 LIF 关A2⌺(v
⬘
⫽0)←X2⌸(v⬙
⫽0), at ⫽308 nm兴 orlaser-induced predissociative fluorescence19 关LIPF, A2⌺(v
⬘
⫽3)←X2⌸(v⬙
⫽0), at ⫽248 nm兴 imaging of OH,quali-tative information of the flame structure can be obtained. Moreover, the LIPF technique can be extended to measure quantitative OH concentrations in atmospheric and higher pressure flames, because the upper excited state is highly predissociative and the fluorescence signal is free of colli-sional quenching.
Variations in the reaction zone structure with mixing rate or with axial location in the flame can best be under-stood from imaging. However, accurate modeling of scalar dissipation rate and molecular mixing requires quantitative images of reactive and passive scalars. Much work has been performed recently in applying various combustion diagnostic techniques, such as Rayleigh and Raman scatter-ing and LIF, to image the distributions of various scalars in the flame. These are especially valuable in flame regions where the finite chemical reaction rate is significant and the flame structure is very complex due to local extinction and reignition processes. The simultaneous PLIF images of OH and CH radicals have been mapped in a piloted turbulent diffusion flame by Sta˚rner et al.20 They found that the im-ages reveal thin, tendril-like structure of CH closely wrap-ping around the rich side of the OH profiles, and the air dilution of the flame substantially thickens the OH reaction zone to the lean side of the stoichiometric. Kelman and Masri21,22 used the Quad-YAG laser source, a dye laser pumped by a doubled Nd:YAG laser beam generating 562.2 nm output and then frequency doubled to 281.1 nm, to excite the R1(6) 共1,0兲 transition of the OH molecules
and to measure the joint temperature and quantitative OH image with a selection of different hydrocarbon fuels of turbulent diffusion flames. They used a stoichiometric CH4-air flame, produced on a premixed flat-flame burner,
to provide a uniform field of OH for calibration of the OH-LIF and UV optical throughput. However, due to the operating laser wavelength at 562.2 nm, laminar flame cal-culations and temperature measurements are required in or-der to correct for the quenching effect on the LIF image. Chao, Wu, and Cheng: Quantitative imaging of OH concentrations . . .
As discussed, swirl induces high turbulence, rapid mix-ing and finite-rate chemistry in the flame and swirl is di-rectly related to the flame stabilization and pollutant emis-sion processes in swirl flames. Control of the swirl flame characteristics can usually be achieved by modifying the swirl strength and the fuel-to-swirl-air momentum flux ratio 共MR兲 that vary the initial fuel-air mixing and the recircula-tion zone. In this paper, the OH-LIPF technique is extended to 2-D LIPF-OH quantitative imaging for instantaneous measurements of the unsteady flame structure and flow-chemistry interaction characteristics in the swirling jet flames of different flame configurations. The OH concen-tration data can be obtained by carefully calibrating the imaging system against a CH4-air flat-flame burner
oper-ated with known flame temperature and fuel-air equiva-lence ratios from lean to rich using identical optical setup. The quantitative swirl flame OH images can be used for the detailed comparison of the complicated flame-flow interac-tion process and to delineate the NOxformation mechanism
for different flame configurations. 2 Experimental Systems
The experimental setup is schematically shown in Fig. 1. The swirling component is generated by a swirler with six guide vanes at an angle of 45 deg, which is placed coaxially with the central fuel tube, corresponding to a geometrical swirl number S of 0.7, as defined according to Bee´r and Chigier23 S⫽2 3
冋
1⫺共Rh/R兲3 1⫺共Rh/R兲2册
tan␣,where Rhis the hub radius, R is the swirler radius, and␣is
the guide vane angle. The diameter of the swirler is 30 mm. Methane is supplied from the central fuel tube through an annular injector with four 2.5 mm holes inclined by 45 deg to the tube axis. The inclination of the injector holes is
employed to investigate the effect of fuel injection as com-pared with the straight central injector. Honeycomb units and fine mesh screens are installed in the settling chamber of fuel and air to control the flow quality. Details of the swirl burner were reported previously.24The swirl burner is mounted on a 2-D traversing table while the optical system remains fixed. A dimensionless parameter of fuel to swirl-air MR, defined as MR⫽FUF
2
/AUA 2
, whereF,A, UF,
and UA are the density and velocity of fuel and air,
respec-tively, is used to distinguish different combustion modes of the swirling flame. For general qualitative characterization of the swirl flames, conventional instruments are also em-ployed to observe the flame and to measure some mean flame quantities. A two-color, four-beam laser Doppler ve-locimetry 共LDV兲 system, operating in the backward scat-tering mode is used for measurements of the velocity and turbulence characteristics. Details of the LDV system was described in a previous paper by Chao et al.8 An R type ( Pt / Pt⫺13Rh), 125m diameter thermocouple is used to measure the temperature. BeO and Y2O3 coatings are
ap-plied to eliminate catalytic reaction to platinum in the flame. The maximum error due to radiation heat loss is estimated to be 5% in the temperature measurement and the results are corrected accordingly. Due to the slow response of the thermocouple, only the mean temperature is calcu-lated based on measurement data at 250 ms increment. Flow/flame visualization techniques offer a powerful tool to establish an overall picture of the flame/flow interaction. Images are captured by a CCD camera and recorded on a professional recorder through a frame code generator. Dif-ferent swirling combustion modes can be identified by varying suitable operation parameters of MR and S. Gas analyzers are used to detect the major pollution emissions in the combustion products, such as NOx,CO,CO2, etc.
The analyzer system is first calibrated against standard span gases of 20 ppm NO in N2 and 250 ppm CO. A stainless steel sampling probe located more than two flame lengths Fig. 1 Schematic diagram of the flow system and experimental setup.
downstream from the jet exit is used to measure the post-flame emission levels, as suggested by Drake et al.25 The emission indices, EINOX and EICO, as suggested by Turns and Lovett,26 are used here to correct the measurement emission data of NOx and CO.
In either normal LIF or LIPF, a laser beam passes through a medium and excites molecules from a state i into a higher excited state i*, for example, the P2(8) rotational
line of the A2⌺←X2⌸(3,0) transition of OH radical of the current LIPF experiment. The number populating the i*
excited state is proportional to ni, the number density in state i. State i*can be deactivated by fluorescence at a rate
F, predissociation at a rate P or collisional quenching at a
rate Q. The rates F and P depend only on the nature of the excited molecules, while the rate Q also depends on gas composition, density, temperature and the nature of the col-lision partners. For OH共3,0兲 transition Gray and Farrow27 measured the predissociative rate constant on the order of 1010s⫺1. The collisional quenching and fluorescence rate constants are estimated to be of the order of 0.5⫻109s⫺1 and 106⫻ s⫺1, respectively.1 Therefore the collisional quenching correction is unnecessary if fluorescence is mea-sured from thev
⬘
⫽3 state. The excited OH molecules can fluoresce down to the ground electronic state X2⌸ by sev-eral paths such as the 共3,0兲 band near 250 nm, the 共3,1兲 band in the 272 nm range, the共3,2兲 lines at 297.5 nm and 共3,3兲 lines near 330 nm. The 共3,2兲 emission lines at 297.5 nm are used for the OH fluorescence measurements as the other lines overlap with other signals or are outside the spectrometer spectral range. The upper excited electronic state (v⬘
⫽3) can depopulate to the lower vibrational states by vibrational/rotational energy transfer. As a result emis-sions from the lower vibrational states in the excited A2⌺ state are collisionally quenched and are not considered here for OH concentration measurements. For predissociative fluorescence measurement, the measured OH intensity can be related to the OH number density as:IF⫽C关F/共F⫹Q⫹P兲兴IL关NOH兴 fB共T兲,
where C is a constant dependent on optical and detector efficiency, F is the fluorescence rate, Q is the collisional quenching rate, P is the predissociating rate, IL is the
inci-dent laser intensity,关NOH兴 is the OH number density, and
fB(T) is the temperature dependent Boltzmann population
fraction. Since P is greater than F plus Q, the equation can be simplified as:
IF⫽C共F/P兲IL关NOH兴 fB共T兲.
In the current measurements, the laser is not tuned to a single resonant line. The OH fluorescence is excited by the residual broadband output 共⬃10 mJ兲 of the KrF excimer laser which is low and nonlinearities in the fluorescence signal are relatively small. It is argued by Cheng et al.5 based on their calculation that the sum of the weighted OH molecular ground state population fractions from the rota-tional lines within the laser tuning range is temperature insensitive 共⬍10%兲 from 1500 to 3000 K. Therefore the OH fluorescence signal is linearly proportional to the OH concentration.
In Fig. 2, for 2-D laser-induced predissociative fluores-cence imaging of OH, a narrow-band tunable KrF excimer laser 共Lambda Physik LPX-250兲 is tuned to excite the
P2(8) rotational line of the A2⌺←X2⌸(3,0) transition at
⫽248.46 nm. The laser is tunable from 247.9 to 248.9 nm with a bandwidth of 0.003 nm. The maximum pulse energy is 450 mJ with a pulse duration of 30 ns. For this work a pulse energy of 200 mJ is used. A thin, 34 mm height, 0.2 mm thick laser sheet is generated by passing the laser beam through a single cylindrical lens ( f⫽1000 mm). The laser sheet is aligned to intersect the flame axis vertically. Only the 25 mm high central portion of the laser sheet, where the laser intensity is high and uniform, is used for imaging. The OH fluorescence signal is collected by a UV camera lens 共Nikkon, f ⫽105 mm, f /4.5兲 and imaged onto an intensified CCD camera 共Princeton Instruments, 576⫻384 array, 22 ⫻22m pixels兲 located at a direction perpendicular to the laser sheet. A 10 mm thick butyl acetate liquid-filter is placed in front of the camera to absorb the Rayleigh scat-tering. Although the Stokes Raman signals may also enter the camera, the influence of the Raman signal on the OH fluorescence signal is insignificant due to the high fluorescence-to-Raman signal ratio.
The linear relationship of OH LIPF signals and OH con-centration already derived can be obtained by calibration against a standard burner of known flame characteristics. A careful calibration procedure of OH concentration is carried out on the Hencken burner, which is operated at 1 atm to produce stabilized laminar methane flames for known equivalence ratios from lean to rich. The Hencken burner consists of 100 capillary fuel tubes, each surrounded by six air passages within a honeycomb metal matrix. The Hencken burner generates rapid mixing and diffusion above the burner rendering a uniform stream of equilibrium products. Please refer to Barlow et al.28 for detailed de-scription of the Hencken burner. An identical optical setup is used as for the OH imaging. Measurements are taken in the Hencken burner in an area of 20⫻ 20 mm located 5 cm downstream in the central portion. This is where the OH radicals have recombined to arrive at an equilibrium con-centration, which is verified by the uniform, close-to-equilibrium flame temperature measured by a
thermo-Fig. 2 Schematic diagram of the LIPF diagnostic system.
couple. The intensified charge-coupled device 共ICCD兲 output did show the uniformity of fluorescence in this area. The OH recombination time was estimated to be about⬃3 ms for a stoichiometric methane-air flame, thus the 5 cm downstream distance is estimated to be adequate, see Cheng et al.5 for a detailed argument. The OH mole frac-tion of the known equivalence-ratio methane flame can be obtained from an equilibrium program 共for example STANJAN program, by Reynolds29兲 or a standard flame code by means of the measured flame temperature. The calibration curve for the OH-LIPF intensity and OH mole fraction can then be obtained by best fit of the data from the Hencken burner calibration, as shown in Fig. 3. The root mean square共rms兲 error in the curve fitting is less than 3%. The quantitative OH concentration distribution in the swirl-ing flames can, therefore, be derived from the calibration curve.
3 Results and Discussion
3.1 Flame Visualization and Qualitative OH Images
The swirling flow field is characterized by the large toroidal recirculation zone and the high turbulence at the edge of the recirculation zone, if the swirl strength is high enough, usu-ally when S⬎0.6. When fuel jet is introduced to the swirl flow, the interaction between the fuel stream and the swirl-ing recirculation results in complicated flow and mixswirl-ing structure. The mixing and the consequent flame structure can be described by including an additional parameter of MR, as already defined. As depicted previously by Cheng et al.,24 the interaction and flow structure can be described by the sketch in Fig. 4. The fuel jet stream is seen to im-pinge on the head end of the recirculation zone obliquely and split into two streams. Most of the fluid moves along the layer in the close vicinity of the edge of the recircula-tion zone where turbulence is high and mixing is intensive. A very small portion of the fluid moves against the rotation of the recirculation vortex toward the centerline. As the MR is decreased, the recirculation zone moves upstream against
the fuel jet. The impingement point moves toward center that significantly enhances mixing as the mixing path along the edge of the recirculation zone is lengthened. The flame configurations in terms of flame length, color and shape are changed as MR is varied. In general, for the current S ⫽0.7 flames three types of characteristic swirling flames, represented by the nozzle exit MR⫽1.5, 0.67 and 0.14, respectively, are generated. The photographs of the swirl-ing jet flames are shown in the left column of Fig. 5. As swirl is introduced and for the high MR共⫽1.50兲, the flame 共Fig. 5, left top兲 is long, mostly yellow in color, quiet and with a highly luminous flame tip, indicating improper mix-ing and poor combustion as evidenced by soot accumula-tion on the surface of the fuel injector. When the flow rate of the swirling air is increased to MR⫽0.67 共Fig. 5, left middle兲, the flame becomes shorter and broader with minor oscillations. The broad flame base indicates the existence of the recirculation zone near the fuel injector. As the flow rate of the swirling air is further increased (MR⫽0.14, Fig. 5 left bottom兲, the flame is greatly shortened almost by a factor of 5 relative to that of the MR⫽1.50 case. The flame becomes open bowl-shaped without a flame tip, blue in color, noisy and oscillatory indicating violent turbulent mixing and combustion. Neither further shortening of the flame nor change in shape can be achieved by further in-creasing the flow rate of swirling air. This type of flame can be classified as the strongly recirculating flame.
The single-pulse images of OH corresponding to each MR case in the upstream region are shown in the right column of Fig. 5 for comparison. Each image is taken at 2 mm above the burner exit to reduce Mie scattering from the Fig. 3 Calibration curve of OH intensity and mole fraction.
Fig. 4 Sketch describing the interaction of the fuel jet stream and
Fig. 5 Photographs and single-pulse 2-D LIPF qualitative imaging of OH inS⫽0.7 swirling methane jet flames for MR⫽1.50 (top), 0.67 (middle), and 0.14 (bottom). Image intensity range, low→high: blue→green→yellow→red→white.
burner surface. The height of each image is 25 mm. In addition to the calibration procedure already described, sev-eral types of noise and corrections had to be estimated and corrected before meaningful data could be obtained. In gen-eral, the nonuniform response of the camera window, the nonuniform laser intensity distribution in the laser sheet, and the possible misalignment of the camera window with respect to the imaging plane, etc., all generate distortions of the resultant images. These distortions can only be cor-rected in situ with the experimental setup. The above Hencken burner calibration of OH fluorescence provides such in situ corrections. Each single-pulse 2-D image is also corrected for the laser power drifting as measured by the output of the monitoring photomultiplier tube 共PMT兲. There are several types of noise in the image that can be eliminated by proper image correction and processing tech-niques. Background noise is one of them. Background noise, such as the dark noise, Mie noise from floating dust, stray light, Raman scattering, etc., in an image must be eliminated or sufficiently reduced to distinguish the data and to provide a clear basis for quantitative analysis. Using optimal thresholding and high-emphasis filtering methods during initial image processing 共see Hansen et al.30 and Gonzales and Wintz31兲 can eliminate the background noise. Before the measurement, 30 laser shots through the air without combustion and fuel flow are performed and the images are recorded and averaged as the background noise for later subtraction from the fluorescence image. Another type of noise of importance is the shot noise. According to Eckbreth,1the relative error共NSR, noise-to-signal ratio兲 of the current fluorescence shot noise is estimated to be much less than 1%. The maximum OH intensity, occurring in the MR⫽1.50 flame, is used to normalize all the OH intensities so that direct comparison of the OH intensity among the flame images can be made. Because the OH intensity reaches a maximum value in the flame front, the measured 2-D imaging of OH may indicate the instantaneous shape of the reaction zone. As indicated by OH images in Fig. 5, for the flames studied, combustion may take place in three regions: the mixing layer between the fuel jet and the re-circulation zone, inside the rere-circulation zone, and the thin layer between the recirculation zone and the ambient air. The swirl-induced recirculation zone acts as a toroidal vor-tex that stretches the fuel jet and rolls up thin layers of fuel with cold fresh air or hot products. For the MR⫽1.5 case 共right top of Fig. 5兲, it is interesting to note that the highest OH intensity occurs in the upstream central portion inside the fuel region. The OH radicals are brought back from the downstream region with hot gases by the recirculating vor-tex to mix with the fresh fuel and air and initiate reaction there. The fuel-air-hot-product mixing induced by the com-plicated interaction of the fuel jet with the recirculating vortex initiates strong initially fast reactions that accumu-late a large amount of OH radicals locally. In other words, the temperature and radicals of the recirculated hot prod-ucts appear to accelerate the initial decomposition and radical-generating processes after strong turbulent mixing with fresh fuel and air. The high OH intensity region cor-responds to the upstream dark region in the luminous flame image of the top photograph in the left column in Fig. 5. This process is more evident as the swirling air is increased in the MR⫽0.67 case 共right middle in Fig. 5兲. The structure
of the OH image shows the characteristic precession of the recirculation vortex in the swirling flames. Most of the high OH intensities, appearing near the recirculation zone, could be due to superequilibrium OH, because the intensity here is higher than that from the recirculated burnt gas and the measured temperature in that region is low 共shown later兲. For the strongly recirculating flame (MR⫽0.14), the strong swirl induces an adverse pressure gradient and brings the recirculation vortex closer to the fuel injector. Low OH intensity, much lower than that in MR⫽1.50 and 0.67 flames, distributes along a thin layer between the re-circulation edge and the ambient air in the OH imaging. In contrast, the flame photograph shows the light blue pre-mixed flame characteristics without a flame tip implying strong premixing due to intense mixing of fuel and air in the upstream region. Because the flame is open without a luminous flame tip, unlike the other two cases there are almost no OH radicals recirculated in the recirculation zone to accelerate the OH generation and accumulation. Further-more, the flow time may be as short as, or even shorter than, the reaction time. Therefore finite chemistry occurs, especially at the upstream end of the recirculation zone as reported earlier.13,14 These OH images demonstrate the variation of the flame structure with increasing swirling air in the swirling jet flames at fixed swirl number.
3.2 Temperature and Quantitative OH Distribution
One of the objectives of this research is to study the effect of fuel-air mixing on NOx emissions and to identify the
NOxformation mechanism in swirling methane jet flames.
Results of NOx emission measurements on identical flame
conditions were reported previously in Ref. 24. The mea-sured emission index of NOx 共EINOX兲 of the flames is
3.03, 1.42 and 0.42 g/kg-fuel for MR⫽1.50, 0.67 and 0.14 cases, respectively. The reduction is more than 85% as MR is decreased from 1.5 to 0.14. The strong swirl and low momentum flux ratio cause strong and rapid mixing of the fuel and air that increases mixture homogeneity, shortens the flame, and eliminates the long flame tip. These effects greatly shorten the residence time for radicals in high tem-perature regions, reduce the slow thermal NOx generation
and result in a lower NOx emission index. To identify the
NOx formation mechanism in these flames, the calibrated
OH mole fraction images and temperatures are employed. For most flames NO is the most important element in NOx
output. Miller and Bowman32 provided a comprehensive review of NO formation pathways in flames. Summarized in their review are the three most important pathways for NO formation in hydrocarbon flames:共1兲 thermal NO, 共2兲 formation of NO involving intermediate N2O, and 共3兲
prompt NO. For the swirling flames operated at atmo-spheric conditions, the second pathway is insignificant be-cause the intermediate N2O is important only for
high-pressure combustion.32 Hence, only the first and third pathways must be considered. The calibrated OH mole fractions for different MR cases, expressed as a bar graph, together with the measured mean temperatures, expressed as the curve with solid circles, are plotted versus physical radial distance for five typical downstream locations (X ⫽5, 10, 15, 20, and 25 mm兲 in Fig. 6. In general, the OH mole fraction decreases with decreasing MR. For MR
⫽1.50 and 0.67 flames, high OH mole fractions occur up-stream (X⬍20 mm) near the centerline, close to the head end of the circulation zone, where the average temperature is lower than 1200 K. As discussed, radicals and hot gases recirculated from downstream mix with fresh fuel and air and accelerate the initial fuel decomposition and radical-generation reactions here. At these locations OH radicals accumulate. Also noted from the high temperature regions in the temperature profile in Fig. 6 that reaction continues as the reacting mixture moves with the stream along the thin layer between the edge of the recirculation zone and the ambient air and reaches the downstream flame tip. At the local temperature of 1200 K the corresponding pre-dicted equilibrium OH mole fraction in a methane flame is almost negligible. Since the OH concentration at these lo-cations is much higher than the known value of the calibra-tion flat flame under equilibrium condicalibra-tion, the关OH兴 in this region appears to be superequilibrium. This superequilib-rium phenomenon is closely related to the recirculated radi-cals from the downstream flame tip. This phenomenon is completely different from the diffusion jet flame where high关OH兴 corresponds to the high temperature regions of the flame layer and the flame tip. For the MR⫽0.14 flame, the OH intensity at each downstream location is much lower than that in the MR⫽1.50 and 0.67 flames. The low OH results of the MR⫽0.14 flame and the absence of a flame tip provide supporting evidence for the preceding
dis-cussion. However, its corresponding temperature is higher. In a study of a lifted hydrogen diffusion flame, Cheng et al.5found that most superequilibrium OH radicals occur at X/D⬉100 and toward equilibrium values near the flame end at X/D⫽175. They also found that the average tem-peratures at X/D⬉100 are below the adiabatic equilibrium temperatures, due to the slow three-body recombination re-actions of OH radicals. The location at X⫽25 mm in the MR⫽0.14 flame is close to the flame height 共see Fig. 5兲 where the OH has recombined resulting in a lower OH intensity and higher temperature distributions in this region. Also, the measured maximum flame temperatures in these swirling flames are lower than 1600 K and the OH resi-dence time is greatly reduced as MR is reduced. Thermal NO mechanism is not prevailing in these flames, especially the MR⫽0.14 flame. Thus, it is tempting to conclude that the measured low temperatures and high 关OH兴 due to su-perequilibrium OH pathway in the reaction zone and burned gas may be one of the major sources of NO forma-tion in swirling methane jet flames. However, more re-search is needed in the future to make firm this notion, especially in a H2flame where there is no CH molecules to
lead to prompt NO formation. 4 Conclusions
Using single-pulse 2-D LIPF imaging of OH radicals, we demonstrate images of the flame structure in the S⫽0.7 swirling methane jet flames operated with three character-istic fuel-air momentum-flux ratios. The LIPF technique, virtually free from collisional quenching in the fluorescence signals, is extended to obtain quantitative 2-D OH concen-trations in flames by means of careful calibration against the flat flame burner. The change of combustion and recir-culation zone characteristics with a change in momentum flux ratio is resolved by imaging the quantitative OH inten-sity distribution in the swirling flames. The recirculated radicals and hot gases from the downstream flame tip re-gion generate the characteristic superequilibrium OH phe-nomenon near the head-end region of the recirculation zone. Superequilibrium OH may lead to one of the NOx
formation pathways in these flames. No substantial thermal NO formation occurs in the strongly recirculated swirl flame. More research in a H2flame where there is no CH
molecules to lead to prompt NO formation is required in the future.
Acknowledgments
This research is supported by the National Science Council of the Republic of China under Grant No. NSC85-2212-E006-096 共YCC兲 and Grant No. NSC86-2212-E-216-001 共TSC兲. The financial support is gratefully acknowledged. References
1. A. C. Eckbreth, Laser Diagnostics for Combustion Temperature and
Species, 2nd ed., Gordon and Breach Publishers, Amsterdam共1996兲.
2. R. W. Dibble, A. R. Masri, and R. W. Bilger, ‘‘The spontaneous Raman scattering technique applied to nonpremixed flames of meth-ane,’’ Combust. Flame 67, 189–206共1987兲.
3. R. S. Barlow, G. J. Fiechtner, and J.-Y. Chen, ‘‘Oxygen atom con-centrations and NO production rates in a turbulent H2/N2jet flame,’’
in Proc. 26th Symp. (Int.) on Combustion, pp. 2199–2205, The Com-bustion Institute, Pittsburgh共1996兲.
4. Q. V. Nguyen, R. W. Dibble, C. D. Carter, G. J. Fiechtner, and R. S. Fig. 6 OH mole fraction and temperature distributions for MR
⫽1.50, 0.67, and 0.14 swirl flames (from left to right) at five down-stream locations.
Barlow, ‘‘Raman-LIF measurements of temperature, major species, OH, and NO in a methane-air bunsen flame,’’ Combust. Flame 105, 499–510共1996兲.
5. T. S. Cheng, J. A. Wehrmeyer, and R. W. Pitz, ‘‘Simultaneous tem-perature and multispecies measurement in a lifted hydrogen diffusion flame,’’ Combust. Flame 91, 323–345共1992兲.
6. N. A. Chigier, J. M. Bee´r, D. Grecov, and K. Bassindale, ‘‘Jet flames in rotating flow fields,’’ Combust. Flame 14, 171–180共1970兲. 7. F. C. Gouldin, R. N. Halthorne, and B. T. Vu, ‘‘Periodic oscillations
observed in swirling flows with and without combustion,’’ in Proc.
20th Symp. (Int.) on Combustion, pp. 269–276, The Combustion
In-stitute, Pittsburgh共1984兲.
8. Y. C. Chao, J. H. Leu, Y. F. Hung, and C. K. Lin, ‘‘Downstream boundary effects on the spectral characteristics of a swirling flow field,’’ Exp. Fluids 10, 341–348共1991兲.
9. R. H. Chen and J. F. Driscoll, ‘‘Nitric oxide levels of jet diffusion flames: effects of coaxial air and other mixing parameters,’’ in Proc.
23rd Symp. (Int.) on Combustion, pp. 281–288, The Combustion
In-stitute, Pittsburgh共1990兲.
10. R. H. Chen and J. F. Driscoll, ‘‘The role of recirculation in improving fuel-air mixing within swirling flames,’’ in Proc. Symp. (Int.) on
Combustion, pp. 531–540, The Combustion Institute, Pittsburgh
共1988兲.
11. R. H. Chen, J. F. Driscoll, J. Kelly, M. Namazian, and R. W. Scheffer, ‘‘A comparison of bluff-body and swirl-stabilized flames,’’ Combust.
Sci. Technol. 71, 192–217共1990兲.
12. T. C. Claypole and N. Syred, ‘‘The effect of swirler burner aerody-namics on NOx formation,’’ in Proc. 18th Symp. (Int.) on
Combus-tion, pp. 81–89, The Combustion Institute, Pittsburgh共1981兲.
13. M. M. Tacke, T. C. Cheng, E. P. Hassel, and J. Janicka, ‘‘Study of swirling recirculating hydrogen diffusion flame using UV Raman spectroscopy,’’ in Proc. 26th Symp. (Int.) on Combustion, pp. 169– 175, The Combustion Institute, Pittsburgh共1996兲.
14. G. Fru¨chtel, E. P. Hassel, and J. Janicka, ‘‘Turbulent length scales in a swirling flame,’’ in Proc. 26th Symp. (Int.) on Combustion, pp. 195–202, The Combustion Institute, Pittsburgh共1996兲.
15. F. Holza¨pfel, B. Lenze, and W. Leuckel, ‘‘Swirl induced intermittency—a novel effect modifying the turbulence structure of swirling free jets,’’ in Proc. 26th Symp. (Int.) on Combustion, pp. 187–194, The Combustion Institute, Pittsburgh共1996兲.
16. R. W. Pitz, T. S. Cheng, S. R. March, and J. A. Wehrmeyer, ‘‘Effects of swirl on finite-rate chemistry in lifted jet diffusion flames,’’ pre-sented at 29th Aerospace Sciences Meeting, Reno, NV, paper 91-2319, American Inst. of Aeronautics and Astronautics共1991兲. 17. R. K. Hanson, ‘‘Combustion diagnostics: planar imaging techniques,’’
in Proc. 21st Symp. (Int.) on Combustion, pp. 1677–1691, The Com-bustion Institute, Pittsburgh共1986兲.
18. H. Becker, A. Arnold, R. Suntz, P. Monkhouse, J. Wolfrum, R. Maly, and W. Pfister, ‘‘Investigation of flame structure and burning behav-iour in an IC engine simulator by 2D-LIF of OH radicals,’’ Appl.
Phys. B: Photophys. Laser Chem. 50, 473–478共1990兲.
19. P. Andersen, A. Bath, W. Gro¨ger, H. W. Lu¨lf, G. Meijer, and J. J. ter Meulen, ‘‘Laser-induced fluorescence with tunable excimer lasers as a possible method for instantaneous temperature field measurements at high pressure: checks with an atmospheric flame,’’ Appl. Opt. 27, 365–378共1988兲.
20. S. H. Sta˚rner, R. W. Bilger, R. W. Dibble, R. S. Barlow, D. C. Four-guette, and M. B. Long, ‘‘Joint planar CH and OH LIF imaging on piloted turbulent jet diffusion flames near extinction,’’ in Proc. 24th
Symp. (Int.) on Combustion, pp. 341–349, The Combustion Institute,
Pittsburgh共1992兲.
21. J. B. Kelman and A. R. Masri, ‘‘Quantitative imaging of temperature and OH in turbulent diffusion flames using a single laser source,’’
Appl. Opt. 33, 3992–3999共1994兲.
22. J. B. Kelman and A. R. Masri, ‘‘Simultaneous imaging of temperature and OH number density in turbulent diffusion flames,’’ Combust. Sci.
Technol. 122, 1–32共1997兲.
23. J. M. Bee´r and N. A. Chigier, Combustion Aerodynamics, Krieger, Malabar, FL共1983兲.
24. T. S. Cheng, Y. C. Chao, D. C. Wu, T. Yuan, C. C. Lu, C. K. Cheng, and J. M. Chang, ‘‘Effects of fuel-air mixing on flame structures and NOxemissions in swirling methane jet flames,’’ in Proc. 27th Symp.
(Int.) on Combustion, pp. 1229–1237, The Combustion Institute,
Pittsburgh共1998兲.
25. M. C. Drake, S. M. Correa, R. W. Pitz, W. Shyy, and C. P.
Fenni-more, ‘‘Superequilibrium and thermal nitric oxide formation in turbu-lent diffusion flames,’’ Combust. Flame 69, 347–365共1987兲. 26. S. R. Turns and J. A. Lovette, ‘‘Measurements of oxides of nitrogen
emissions from turbulent propane jet diffusion flames,’’ Combust. Sci.
Technol. 66, 233–249共1989兲.
27. J. A. Gray and R. L. Farrow, ‘‘Predissociation lifetimes of OH
A2⌺⫹(v⬘⫽3) obtained from optical-optical double-resonance
line-width measurements,’’ J. Chem. Phys. 95, 7054–7060共1991兲. 28. R. S. Barlow, R. W. Dibble, J.-Y. Chen, and R. P. Lucht, ‘‘Effect of
Damko¨hler number on superequilibrium OH concentration in turbu-lent nonpremixed jet flames,’’ Sandia Report SAND89-8402共1989兲. 29. W. C. Reynolds, STANJAN Program Manual, 3rd ed., Stanford
Uni-versity共1986兲.
30. R. K. Hanson, J. M. Seitzman, and H. P. Paul, ‘‘Planar laser fluores-cence imaging of combustion gases,’’ Appl. Phys. B 50, 441–451
共1990兲.
31. R. C. Gonzales and P. Wintz, Digital Imaging Processing, Addison-Wesley, Reading, MA共1977兲.
32. J. A. Miller and C. T. Bowman, ‘‘Mechanism and modeling of nitro-gen chemistry in combustion,’’ Prog. Energy Combust. Sci. 15, 287– 338共1989兲.
Yei-Chin Chao received his PhD degree from the School of Aerospace Engineering, Georgia Institute of Technology, in 1984 and has since been affiliated with the Insti-tute of Aeronautics and Astronautics of Na-tional Cheng Kung University. He became an associate professor in 1984 and has been a full professor since 1991. His re-search has been in the areas of gas turbine combustion, combustion diagnostics, flaflow interaction, propulsion, fluid me-chanics, and acoustics. He has published more than 100 technical articles in well-established journals and conference proceedings. He also serves on the Board of Directors of the Combustion Institute, Chinese Taipei Chapter.
Der-Chyun Wu received his MS degree from the Institute of Aeronautics and Astro-nautics, National Cheng Kung University, in 1995 and is currently studying for his PhD degree. His major research is in flame-flow interaction, fluid mechanics, control of nitric oxides, and the develop-ment of Raman scattering and laser-induced fluorescence techniques to study the nonpremixed swirling methane jet flames.
Tsarng-Sheng Cheng received his MS and PhD degrees in mechanical engineer-ing from Vanderbilt University, in 1987 and 1991, respectively. He became an associ-ate professor with the Department of Me-chanical Engineering, Chung Hua Univer-sity, Taiwan, in 1991 and became a professor in 1999. He headed the depart-ment for 3 years and has been the dean of student affairs since 1996. His major ac-complishment and his current research field are the development and application of laser Raman scattering and laser-induced fluorescence techniques to the study of subsonic and supersonic hydrogen jet flames and to the study of pollutant emissions from nonpremixed and partially premixed swirling meth-ane jet flames.
