Application of Optical Diagnostics Techniques to a Laboratory-Scale Turbulent Pulverized Coal Flame

Application of Optical Diagnostics Techniques to aLaboratory-Scale Turbulent Pulverized Coal Flame

Seung min Hwang,† Ryoichi Kurose,‡ Fumiteru Akamatsu,*,† Hirofumi Tsuji,‡Hisao Makino,‡ and Masashi Katsuki†

Department of Mechanical Engineering, Osaka University, 2-1 Yamada-oka, Suita, Osaka565-0871, Japan, and Energy Engineering Research Laboratory, Central Research Institute of

Electric Power Industry (CRIEPI), 2-6-1 Nagasaka, Yokosuka, Kanagawa 240-0196, Japan

Received June 1, 2004. Revised Manuscript Received November 6, 2004

Combustion measurements based on optical diagnostics techniques, which allow noninvasivemeasurements of velocity, density, temperature, pressure, and species concentration, have recentlybecome of major interest as tools not only for clarifying the combustion mechanism but also forvalidating the computational results for the combustion fields. In this study, the combustioncharacteristics of a pulverized coal flame are investigated using advanced optical diagnostics. Alaboratory-scale pulverized coal combustion burner is specially fabricated. Velocity and shape ofnonspherical pulverized coal particles, light emissions from a local point, and temperature inthe flame are measured by shadow Doppler particle analyzer (SDPA), a specially designedreceiving optics (multicolor integrated receiving optics, MICRO), and a two-color radiationpyrometer, respectively. The simultaneous measurement of OH planar laser-induced fluorescence(OH-PLIF) and Mie scattering image of pulverized coal particles is performed to examine spatialrelation of combustion reaction zone and pulverized coal particle. The results show that the size-classified diameter and velocity of the pulverized coal particles in the flame can be measuredwell by SDPA. The measurements of the OH chemiluminescence and CH band light emissionfrom a local point in the flame using MICRO and the simultaneous measurement of theinstantaneous OH-PLIF and Mie scattering image of pulverized coal are effective for evaluatingthe pulverized coal flames and investigating their detailed flame structure.

Introduction

Energy is the base that supports our economic andother activities in the present advanced civilization. Theoil crisis, which took place more than 20 years ago, wasa great historical event in that the whole world becameaware of the significance of energy.1 Worldwide demandfor energy has continued to increase as populations ofdeveloping nations increase along with their industri-alization.

In this situation, coal is an important and promisingenergy resource for electricity supply because its reserveis far more abundant than those of the other fossil fuels.At present, the major utilization method of coal inthermal power plants is pulverized coal combustion, inwhich coal is pulverized into fine particles of severaltens of micrometers in diameter. For pulverized coalcombustion, it is desired to operate boilers with higherefficiency and lower pollutant emission, and recently toutilize various coals (i.e., low-rank coals) for diversifyingfuel sources and lowering the cost in Japan and Korea.To efficiently develop new technologies for these re-quirements, it is of great importance to understand the

pulverized coal combustion mechanism in detail. How-ever, the combustion process of the pulverized coal isnot well clarified so far, since it is a complex phenom-enon, in which dispersion of coal particles, their devola-tilization, chemical reactions of volatilized fuel, andresidual char with air take place interactively at thesame time and the simultaneous and instantaneousmeasurements for them are very difficult.2,3 Further-more, though numerical simulation of the pulverizedcoal combustion has been performed as a supporting toolfor the developments of the new technologies in thesedays,4-7 its reliability has not been sufficiently con-firmed yet because of the lack of the reliable detailedmeasured data with which to be compared.

Combustion measurements based on optical diagnos-tics techniques, which allow noninvasive measurementsof velocity, density, temperature, pressure, and speciesconcentration, have recently become of major interest.In particular, particle velocity has been measured in thepulverized coal flame using laser Doppler velocimetry(LDV).8 However, LDV cannot provide detailed inves-

* Corresponding author. Phone: 81-6-6879-7253. Fax: 81-6-6879-7247. E-mail: [email protected].

† Osaka University.‡ Central Research Institute of Electric Power Industry.(1) Harada, M. Proceedings of The International Conference on

Clean Coal Technologies for our Future (CCT2002), Sardinia, Italy,2002; pp 1-11.

(2) Hurt, R. H.; Mitchell, R. E. Proc. Combust. Inst. 1992, 24, 1233-1241.

(3) Hurt, R. H.; Mitchell, R. E. Proc. Combust. Inst. 1992, 24, 1243-1250.

(4) Kurose, R.; Makino, H.; Suzuki, A. Fuel 2004, 83, 1447-1455.(5) Kurose, R.; Ikeda, M.; Makino, H. Fuel 2001, 80, 1457-1465.(6) Kurose, R.; Makino, H.; Suzuki, A. Fuel 2004, 83, 693-703.(7) Kurose, R.; Ikeda, M.; Makino, H.; Kimoto, M.; Miyazaki, T.

Fuel 2004, 83, 1177-1785.

382 Energy & Fuels 2005, 19, 382-392

10.1021/ef049867z CCC: $30.25 © 2005 American Chemical SocietyPublished on Web 01/19/2005

tigation on particle diameter change and size-classifiedparticle velocity, because LDV measures overall velocityof pulverized coal particles with diameter distribution.For measuring particle velocity and size simultaneously,phase Doppler anemometer (PDA) is widely used nowa-days. However, the application is limited to sphericalparticles. Hardalupas et al.9 and Maeda et al.10 devel-oped shadow Doppler particle analyzer (SDPA) byextending the LDV system as a method for simulta-neous measuring of velocity and particle shape ofnonspherical particles such as pulverized coal. TheSDPA has the advantage over other sizing methods forirregular particles in that it is tolerant of the opticalmisalignment and fouling which are inevitable whenpassing laser beams through optical access windows ina confined flame measurement.11 On the other hand,light emissions and radicals such as OH, CH, NO, andso on in the combustion fields have been monitored bya specially designed receiving optics (multicolor inte-grated receiving optics, MICRO) and a laser-inducedfluorescence (LIF).12-15 Presently, LIF is extended to theplanar LIF (PLIF) to visualize two-dimensional distri-bution with high temporal and spatial resolution.

The purpose of this study is to apply advanced opticaldiagnostics to a pulverized coal flame and investigatethe combustion characteristics in detail. For this pur-pose, a laboratory-scale pulverized coal combustionburner is specially fabricated. Velocity and shape ofnonspherical pulverized coal particles, temperature, gasspecies concentration, and light emissions from a local

point in the flame are measured by SDPA, a two-colorradiation pyrometer, sampling probe, and MICRO,respectively. In addition, to examine spatial relation ofcombustion reaction zone and pulverized coal particles,the simultaneous measurement of OH-PLIF and Miescattering image of pulverized coal particles is per-formed.

Experimental Apparatus and Methods

A turbulent pulverized coal combustion burner and thesupplying system are illustrated schematically in Figure 1. Themain air for combustion is supplied from a compressor andthe flow rate is regulated by a mass-flow controller. Pulverizedcoal particles, supplied and regulated by a screw feeder, issucked by the main air flow and mixed with the air in aninjector to form solid-gas two-phase jet issued from the mainburner port (inner diameter: 6 mm). Methane is supplied toannular slit burner (width: 0.5 mm) installed outside of thecentral port to ignite the two-phase jet because the coalinjection rate is very small and the flame stabilization isimpossible for pure pulverized coal flame. The burner is setup on a three-dimensional traverser for precise positioning inmeasurement and is confined in an acrylic duct with octagonsection for isolating the flame from external disturbance andpreventing scattering of pulverized coal. Eight quartz windowsare equipped on the sidewall of the duct for optical access forlaser insertion, light detection, and flame observation. Theorigin for measurement is located at the center of the burnerport, and z and r denote the axial and radial distances fromthe origin, respectively.

Tables 1 and 2 show the properties of used coal and theexperimental condition. The coal we used is Newlands ofbituminous coal when it classifies from a proximate analysisvalue and heating value, and this is pulverized on the samecondition for actual pulverized coal combustors in thermalpower plants. The mass base median diameter measured bya laser diffraction particle size analyzer is 33 µm (the numberbase average diameter is 25 µm). The methane flow rate isthe minimum amount to form stable flame. The pulverized coalis supplied after the methane diffusion flame became stable,and the measurement is started when the pulverized coalinjection rate is regulated. Unburned pulverized coal particlesin exhaust gas are collected by bag filters and exhaust gas isreleased into atmosphere by a ventilator after cooling by a heatexchanger.

(8) Schnell, U. Phys. Chem. 1993, 97, 1582-1589.(9) Hardalupas, Y.; Hishida, K.; Maeda, M.; Morikita, H.; Taylor,

A. M. K. P.; Whitelaw, H. Appl. Opt. 1994, 33 (36), 8417-8426.(10) Maeda, M.; Morikita, H.; Prassas, I.; Taylor, A. M. K. P.;

Whitelaw, H. Part. Syst. Charact. 1997, 14, 79-87.(11) Tsuji, H.; Kurose, R.; Makino, H. Trans. JSME B 2002, 68,

306-312 (in Jpn.).(12) Rothe, E. W.; Andresen, P. Appl. Opt. 1997, 36 (18), 3971.(13) Deguchi, Y.; Iwasaki, S. Proceedings of The 1st Asia-Pacific

Conference on Combustion, Osaka, Japan, 1997; pp 22-25.(14) Tsuji, H.; Miyazaki, T. Visualization of Radicals in Flames by

Laser-Induced Fluorescence (LIF); CRIEPI Report W97007; CentralResearch Institute of Electric Power Industry, 1998 (in Jpn.).

(15) Akamatsu, F.; Wakabayashi, T.; Tsushima, S.; Katsuki, M.;Mizutani, Y.; Ikeda, Y.; Kawahara, N.; Nakajima, T. Meas. Sci.Technol. 1999, 10, 1240-1246.

Figure 1. Schematic diagram of a pulverized coal burner and supply system.

Application of Optical Diagnostics Techniques Energy & Fuels, Vol. 19, No. 2, 2005 383

Figure 2 shows the direct photograph of the pulverized coalflame (exposure time: 1/8000 s). The pulverized coal particlesare visualized by an Ar-ion laser sheet inserted above theburner port. The photograph shows that the ignition ofpulverized coal particles starts in the vicinity of the burnerport of z ) 30 mm and combustion proceeds downstream toform a luminous flame.

Figure 3 shows the schematic diagram of SDPA. The SDPAis based on the imaging of a conventional LDV probe volumeonto a linear photodiode array. The new aspect of this

instrumentation is that the shape of the particles is obtainedfrom the temporal output from elements of a linear photodiodearray positioned at the focal plane of the collection optics ofLDV to record the shadow of the passage of the particle. Thedetailed measurement principle is described in refs 9 and 10.The shape and two components of velocity of burning pulver-ized coal particles are measured in the turbulent pulverizedcoal flame. The equivalent circle diameter is calculated fromthe section area of each particle for size measurement.

The two-color radiation pyrometer is used for temperaturemeasurement of a pulverized coal particle. It is a system whichreceives optic by the sensor and asks the ratio of the outputcorresponding to each wavelength for temperature, afterdividing the synchrotron orbital radiation from the measuredobject into two different wavelengths (λ1 ) 0.9 µm 0 λ2 ) 1.0µm). However, it is difficult to define the temperature obtainedby two-color pyrometer measurements, since this value isstrongly affected by coal particles and soot. Moreover, strictlyspeaking, this is not the value at local point. It can be saidthat this temperature is overall temperature containing coalparticle, soot, and gaseous species in a relatively large region.

The OH chemiluminescence (306 nm), CH band lightemission (431.5 nm), and Mie scattering (514.5 nm) from alocal point of pulverized coal flame are measured by MICRO.The detailed optical system for local point measurement isdescribed in ref 15.

The simultaneous measurement system for OH-PLIF andMie scattering image of pulverized coal particles is shown inFigure 4. To illuminate the pulverized coal particles, a second-harmonic wave of Nd:YAG laser (Spectra-Physics, GCR-270-10) is used. For the OH-PLIF, P1 (8) absorption line (285.586nm) of (1, 0) band is excited by using a wavelength tunablelaser (Spectra-Physics, MOPO-730) pumped by a third-harmonic wave of the same Nd:YAG laser. The two laserbeams of the different wavelengths are aligned in the samepassage using a dichroic mirror and are formed into a thinlaser light sheet, approximately 1.5-mm thickness, illuminat-ing a vertical plane including the central axis of the flameabove the burner port. Mie scattering of pulverized coalparticles is passed through the optical band-pass interference

Figure 2. Direct photograph of pulverized coal.

Figure 3. Optical arrangement of shadow Doppler particle analyzer (SDPA).

Table 1. Properties of Coal

parameter value

high heating valuea 29.1 MJ/kglow heating valuea 28.1 MJ/kg

proximate analysismoistureb 2.60 wt %asha 15.20 wt %volatile mattera 26.90 wt %fixed carbona 57.90 wt %

ultimate analysiscarbona 71.90 wt %hydrogena 4.40 wt %nitrogena 1.5 wt %oxygena 6.53 wt %total sulfura 0.44 wt %combustable sulfura 0.39 wt %

a Dry basis. b As received.

Table 2. Experimental Conditions

parameter value

pulverized coal feed rate 1.49 × 10-4 kg/sthermal input of coala 4.19 kWthermal input of CH4

a 0.83 kWair flow rate 1.80 × 10-4 m3/sCH4 flow rate 2.33 × 10-5 m3/sbulk equivalence ratio φ ) 6.09Reynolds number, Re 2544

a Based on the low heating value.

384 Energy & Fuels, Vol. 19, No. 2, 2005 Hwang et al.

filter with transmission peak wavelength of 532 nm and half-value width of 1.0 nm and is captured by a CCD camera (VisionResearch systems, Phantom Ver. 5.0). Laser-induced OHfluorescence, on the other hand, is passed through the opticalband-pass interference filter with transmission peak wave-length of 320 nm and half-value width of 20 nm and iscaptured by a CCD camera (Roper, Model EEV 02-06-202)coupled with an image intensifier. The imaging area is asquare of 30 mm by 30 mm located at around the central axisof the flame. TTL signals generated by two pulse delaygenerators (PDG) (Stanford Research Systems, Model DG535)are used for timing control of the Nd:YAG laser and imagecapture of OH-PLIF and Mie scattering image of pulverizedcoal particles.

Results and Discussion

Simultaneous Measurement of Velocity andShape of Pulverized Coal Particles by SDPA. InSDPA measurement, velocity is obtained only when thediameter is measured correctly, because the diameterdetected by SDPA is confined because of the particlesize. Hence, the validity of velocity data obtained bySDPA should be assessed by comparing with LDVmeasurement. Figure 5 shows the axial distributionsof mean and rms (root mean square) velocities ofpulverized coal particles by SDPA and LDV. Combus-ting and noncombusting cases are shown. Concerningthe noncombusting case, however, the SDPA measure-ment was impossible because the number density ofparticle was very high, so that only the LDV measure-ment was conducted. For the combusting case, thevelocity data of SDPA is well consistent with that byLDV. This verifies that accuracy of velocity measure-ment by SDPA is the same level as LDV. The compari-son between the results in combusting and noncombus-ting cases shows that the mean velocity U of theparticles in the noncombusting case is markedly de-creased with increasing axial distance, whereas thetrend in the combusting case becomes gentle. This isdue to the heat expansion of the gaseous phase causedby combustion reaction. RMS velocity in combusting

case is small compared with the noncombusting case.It is because that the high gaseous temperature in-creases the kinematic viscosity, which suppresses theturbulence.

To investigate the behavior of a pulverized coalparticle in more detail, the mean diameter (equivalentcircle diameter), the particle size distribution, and thecircularity are considered using the result of simulta-neous measurement of the particle velocity and shapemeasured by SDPA. The axial and radial distributionsof the mean diameter of the pulverized coal particle are

Figure 4. Schematic apparatus of simultaneous measurement system for OH-PLIF and Mie scattering.

Figure 5. Axial distributions of (a) mean velocity and (b) rootmean square (rms) velocity of pulverized coal particles bySDPA and LDV.


shown in Figure 6. Also, the PDF distributions at somelocations are shown in Figures 7 and 8. The meandiameter gradually increases as the distance fromburner increases, whereas it is found to decrease asgoing to edge. In addition, as the streamwise distanceincreases and the radial distance decreases, the PDFof the small particle decreases and the particles diam-eter giving the maximum probability density increases.Hence, it is speculated that the increase of the meanparticle diameter is caused by the disappearance ofsmall particles and the particle swelling is due todevolatilization of volatile matter in the combustionprocess. Moreover, the reason that the peak of a particlediameter distribution is shifted to the small diameterside with increasing the radial distance is also consid-

ered so that the particles with small diameters, whichare easy to follow an air current, tend to move to theoutside of a flame by the influence of turbulent orga-nized motions and heat expansion of the gaseous phase.The probability density of the particle smaller than 20µm becomes large as the radial distance increases.Saitoh et al.16 applied the PDA measurement to thepolydisperse premixed-spray flames in a stagnation flowand investigated the followability to the flow of a fueldroplet. According to the result, it is reported that if itis a droplet of the particle diameter to 20 µm, thefollowability to a gaseous phase flow is concluded good.Hence, it is considered that the boundary of the fol-lowability to the gaseous phase is before and after 20mm also in pulverized coal from this result like a dropletin spray combustion.

The circularity11 which is one of the most importantindices to characterize the particle shape is examined.The circularity C (0 < C < 1) is an index, which gives 1for a true circle and 0 for a needle shape, and iscalculated as follows:

where A is the shadow area of the particle and c is theperimeter. The circularity was calculated at three pointsof z ) 60, 120, and 180 mm on a burner center axis inthe combusting case. For the noncombusting case, theSDPA measurement could not be performed because ofthe high-number density of particles. Hence, the mea-surement for the noncombusting case was done withlower number density than that for the combusting caseonly at one point of z ) 60 mm. Figure 9 shows thecircularity against the particle size at three streamwiselocations. Although for both the combusting and non-combusting cases the circularity decreases with increas-ing the particle size, the trend is marked for thecombusting case. It is considered that for the combus-ting case, particles’ crack and breakup due to theswelling increase the large particles’ circularity, and thechar reaction and ash melting decrease the smallparticles’ circularity.17

Figure 6. Mean diameter by SDPA: (a) axial distribution and(b) radial distribution.

Figure 7. Particle size distributions measured at z ) 30-180 mm with every 30 mm on the central axis of the burner (thenumber base).

C )x4πA

c(1)


To further investigate the behavior of the pulverizedcoal particle, the pulverized coal particle was classified

into the class of every 5 µm, and the velocity field ofthe particle for every particle diameter class wasexamined. Figures 10 and 11 show the size-classifiedmean and rms axial velocities of pulverized coal par-ticles at the distances from the burner of z ) 60, 120,and 180 mm. The measurements are done on the centerand edge of the flame jet. The mean and rms velocitiesof pulverized coal particles on the center (r ) 0 mm)are independent of the particle size, whereas those onthe edge highly depend on the particle size. The reasonfor this is considered as follows. The particles in thecentral high-speed region tend to be transferred tooutside in radial direction owing to thermal expansionof gas phase, and organized motions appeared in jet.

(16) Saitoh, H.; Akamatsu, F.; Katsuki, M. Trans. JSME B 2003,69, 207-214 (in Jpn.).

(17) Wells, W. F.; Kramer, S. K.; Smoot, L. D. Proc. Combust. Inst.1984, 20, 1539-1546.

Figure 8. Particle size distributions on the radial direction (z ) 60, 120, and 180 mm).

Figure 9. Circularity for each 5-µm particle size class (z )60, 120, and 180 mm, center position).


Although the axial velocity of the small particlestransferred outside is quickly decreased by the ambientfluid with lower velocity, that of the larger particles isnot affected by the ambient fluid because of largerinertia. Therefore, the larger the particle size is, thehigher the velocity is on the edge.

Flame Temperature by Two-Color RadiationPyrometer and Gas-Species Concentration. Figure12 shows the axial profile of mean temperature by two-color radiation pyrometer. Temperature measurementwas impossible in the vicinity of the burner port of z <18 mm because of the weak combustion reaction. Thetemperature shows the maximum value at z ) 30 mmand decreases gradually with increasing the axialdistance, z. The peak temperature is higher than thegeneral ones for the pulverized coal flame (approxi-mately 1600 °C), and the axial distance showing thepeak temperature is not correspondent to that wherethe pulverized coal combustion actively occurs (seeFigure 2; at z ≈ 30 mm, the pulverized coal combustionappears to just start). According to the existing pa-

per,18,19 heating rates of the pulverized coal in furnaceare ∼103 deg/s. However, the present value becomes a

(18) Essenhigh, R. H.; Misra, M. K.; Shaw, D. W. Combust. Flame1989, 77, 3-30.

(19) Farzan, H.; Essenhigh, R. H. Proc. Combust. Inst. 1982, 19,1105-1111.

Figure 10. Mean and rms particle velocity for each 5-µmparticle size class (z ) 60, 120, and 180 mm, center position).

Figure 11. Mean and rms particle velocity for each 5-µmparticle size class in outer portion of the flame (z ) 60, 120,and 180 mm, edge position).

Figure 12. Distribution of particle temperature on the centralaxis by two-color radiation pyrometer.


much bigger value of ∼106 deg/s. This is because themethane flame is used to ignite the pulverized coals inthis study.

The distribution of gas-species concentration on thecentral axis is shown in Figure 13. The change of O2concentration is rapidly consumed with increasing theaxial distance, and CO and CO2 concentrations give thepeak values at z ) 120 mm. The tendency of the COand CO2 concentrations having peaks is in agreementwith the previous experiment.20 It is also observed thatthe NO concentration suddenly increases near theburner exit at z ) 90 mm and approaches a fixed value.

Local Measurement of OH Chemiluminescenceand CH Band Light Emission from Flame and MieScattering of Pulverized Coal Particles byMICRO. Figure 14 shows the axial intensity profilesof OH chemiluminescence, CH band light emission, andMie scattering intensity from a local point of pulverizedcoal particles by MICRO to observe time-averaged flamestructure. The high OH chemiluminescence and CHband light emission are detected near the burner belowz ) 20 mm. In this region, the OH chemiluminescenceintensity is higher than the CH band light emissionintensity, which suggests that it is influenced by themethane pilot flame. It is considered that the ignitionstarts in the region, where the Mie scattering intensitydecreases rapidly from near z ) 40 mm and the OH andCH-light emissions’ intensities begin to rise again.Furthermore, CH band light emission intensity is strongcompared with the OH chemiluminescence because thecontinuous spectrum accompanied by the solid-body

radiation from soot is overlapped on CH band emission.Accordingly, the flame luminescence increases as thecombustion process proceeds, as shown in Figure 2.

Simultaneous Measurement of OH-PLIF andMie Scattering Images of Pulverized Coal Par-ticles. Figure 15 shows the Mie scattering images ofpulverized coal in the noncombusting case and thecombusting case, obtained at four axial locations. In aninstantaneous image, the large-scale eddy structure(coherent structure) of a turbulent flow is observed inthe number density distribution of a pulverized coalparticle. Although there is an opinion that such a large-scale eddy structure cannot be seen since the diameterof pulverized coal particles is several tens of microme-ters and specific gravity is much larger than the gaseousphase, it is very interesting that the coherent structureis observed. In the noncombusting case, the width of theparticle-laden flow spreads and the number density ofthe particles decreases going downstream. In the com-busting case at z ) 135 mm, nonuniformity of the

(20) Azuhata, S.; Narato, K.; Kobayashi, H.; Arashi, N.; Morita, S.;Masai, T. Proc. Combust. Inst. 1986, 21, 1199-1206.

Figure 13. Distribution of gas-species concentration on thecentral axis.

Figure 14. Distribution of OH chemiluminescence and CHband light emission from flame and Mie scattering intensityof pulverized coal in the z-direction by MICRO.

Figure 15. Mie scattering of pulverized coal. The monitoringarea of each image is a square of 30 mm × 30 mm thatcorresponds to r ) -15 to 15 mm and z ) 15-45, 45-75, 75-105, and 105-135 mm.


number density is clearly observed because combustionreaction takes place locally and the small pulverized coalparticles with high flammability disappeared preferen-tially. These results will be discussed in the next sessionof the simultaneous measurement of the OH-PLIF andMie scattering images of pulverized coal particles.

To examine the spatial relations of combustion reac-tion zone and pulverized coal particles for the combus-ting case, OH-PLIF and Mie scattering images of theparticles are compared. Figure 16 shows the instanta-neous OH-PLIF and Mie scattering images of theparticles obtained simultaneously. The intensity profileson the central horizontal line (z ) 30, 60, 90, 120, 150,and 180 mm) of both images are shown in Figure 17.In the upstream, OH-LIF intensity is high in the

periphery of the high number density regions of pulver-ized coal particles (clusters). Moving downstream, onthe other hand, OH-LIF intensity becomes high in thecenter of the clusters of the pulverized coal particles.From these results, it is found that, in the upstreamregion, combustion reaction occurs only in the peripheryof the clusters where high-temperature burned gas ofthe methane pilot flame is entrained and oxygen supplyis sufficient. In the downstream region, however, com-bustion reaction can be seen also inside of clusters ofpulverized coal particles because the devolatilizationprocess of the coal particles proceeds accompanied bytemperature rise of the particles, and the mixing processbetween the volatile matter and circumference air ispromoted. If this is applied to the group combustion ofliquid fuel spray,21 the combustion mode in the up-stream region corresponds to the external droplet group

(21) Chiu, H. H.; Kim, H. Y.; Croke, E. J. Proc. Combust. Inst. 1982,19, 971-980.

Figure 16. Simultaneous images of Mie scattering and OH-LIF of pulverized coal flame.

Figure 17. Distribution of brightness intensity of the (a) Miescattering and (b) OH-LIF images of pulverized coal flameon the central cross section.


combustion, and that in the downstream correspondsto the internal droplet group combustion. In the pulver-ized coal combustion, however, mechanism of the groupcombustion is different from that of liquid fuel spraybecause the behavior of the irregular-shaped pulverizedcoal is very complicated accompanied by injection of thevolatile matter from the surface3 and the small holesare formed on the surface of the pulverized coal non-uniformly.20

Since quantitative evaluation of pulverized coal com-bustion cannot be performed only by observing theimages, statistical analysis is applied to the intensityprofile on the central axis of both the images shown inFigure 16. The total number of data used in FFTprocessing is 256 () 28), and a basic wavenumber is 50l/m in this case. The autocorrelation coefficient RMS(τ)of the profile of Mie scattering intensity is shown inFigure 18. Here, τ in the abscissa is a displacement. Itcan be thought that width of zero-crossing points ofautocorrelation coefficient is related to the length scaleof the cluster of pulverized coal particles. The width ofan autocorrelation coefficient becomes narrow goingdownstream. The length scale of the cluster decreasesas the combustion proceeds since width of zero-crossingpoints of the autocorrelation coefficient is 9.68 mm inthe upstream region at z ) 15-45 mm and 5.94 mm inthe downstream region at z ) 165-195 mm.

Figure 19 shows the cross-correlation coefficient,SMS,OH(τ), between Mie scattering and OH-LIF. Here,τ in the abscissa is a displacement of OH-LIF to Miescattering. The cross-correlation coefficient has a deep

follow at τ ≈ 0 mm in the upstream region. In thedownstream region, on the other hand, where thedevolatilization and combustion process proceeds, thecross-correlation coefficient has the positive peak at τ≈ 0 mm. From these results, it can be confirmed thatthe combustion reaction only exists in the periphery ofcluster of pulverized coal particles in the upstreamregion and invades inside the cluster as combustionproceeds in the downstream region.

Conclusions

A laboratory-scale pulverized coal combustion burnerwas specially fabricated, and the precise optical diag-nostics were applied to the pulverized coal flame.

The size-classified diameter and velocity of the pul-verized coal particles in the flame were measured bySDPA. It was observed that the particle mean diameterincreases as the distance from the burner increases, andthis is caused by the decrease of small particles’diameter and the increase of large particles’ diameter,which result in the char reaction and the particleswelling due to devolatilization, respectively. The stream-wise velocities of pulverized coal particle with size-classified diameter in the central region of the jetindicate the same magnitude, whereas those on the edgeare different depending on the particle diameter. Thisbehavior was well explained in terms of particle inertia.

The measurements of the OH chemiluminescence andCH band light emission from a local point in the flameusing MICRO and the simultaneous measurement of

Figure 18. Autocorrelation coefficient of Mie scatteringsignal.

Figure 19. Cross-correlation coefficient between Mie scat-tering and OH-LIF signals.


the instantaneous OH-PLIF and Mie scattering imageof pulverized coal were effective for evaluating thecombustion diagnosis of the pulverized coal flames andfor investigating their detailed flame structure. In theupstream region, combustion reaction occurs only in theperiphery of the clusters where high-temperature burnedgas of the methane pilot flame is entrained and oxygensupply is sufficient, and in the downstream region,however, combustion reaction can be seen also insideof clusters of pulverized coal particles. This is becausein the downstream region, the devolatilization processof the coal particles proceeds accompanied by temper-ature rise of the particles and the mixing processbetween the volatile matter and circumference air ispromoted. From these results, it can be said that thepresent diagnostics techniques are quite effective forevaluating the pulverized coal flames.

Finally, the present measurements and obtained dataare valuable not only for elucidating the detailedstructure of the pulverized coal combustion but also forthe assessment of the numerical simulations of the two-phase combustions such as pulverized coal and spraycombustions. Though numerical simulations of suchtwo-phase combustions have been performed in thesedays, its reliability has not been sufficiently confirmedyet because of the lack of the reliable detailed measured

data with which to be compared. Therefore, recently,the data obtained by the measurements for variousflames themselves have come to be important forvalidation studies.22 The precise data measured bynonintrusive optical techniques presented here wouldbe very useful especially for the development of the highaccurate numerical technologies based on large-eddysimulation (LES)23,24 or direct numerical simulation(DNS).25,26

Acknowledgment. The authors would like to thankDr. Yoshihiro Deguchi of Mitsubishi Heavy Industries,Ltd. for helpful suggestions on this study.

EF049867Z

(22) Masri, A. R.; Kali, P. A. M.; Barlow, R. S. Combust. Flame 2004,137, 1-37.

(23) Kurose, R.; Makino, H. Combust. Flame 2003, 135, 1-16.(24) Kurose, R.; Desjardins, O.; Nakamura, M.; Akamatsu, F.;

Pitsch, H. In Annual Research Briefss2004, Center for TurbulenceResearch, NASA Ames/Stanford University.

(25) Akamatsu, F.; Saitoh, H.; Katsuki, M. Proceedings of The 8thInternational Conference on Liquid Atomization and Spray System(ICLASS-2000), Pasadena, CA, 2000; pp 25-32.

(26) Akamatsu, F.; Saitoh, H.; Katsuki, M. Proceedings of The 4thKorea Society of Mechanical Engineers (KSME)/Japan Society ofMechanical Engineers (JSME) Thermal Engineering Conference, Kobe,Japan, 2000; pp 259-264.


Documents

Application of Optical Diagnostics Techniques to a Laboratory-Scale Turbulent Pulverized Coal Flame