Recent advances in the measurement of strongly radiating, turbulent reacting flows

lable at ScienceDirect

Progress in Energy and Combustion Science 38 (2012) 41e61

Contents lists avai

Progress in Energy and Combustion Science

journal homepage: www.elsevier .com/locate/pecs

Review

Recent advances in the measurement of strongly radiating, turbulent reactingflows

G.J. Nathan a,c,*, P.A.M. Kalt a,c, Z.T. Alwahabi b,c, B.B. Dally a,c, P.R. Medwell a,c, Q.N. Chan b,c

a School of Mechanical Engineering, The University of Adelaide, S.A. 5005, Australiab School of Chemical Engineering, The University of Adelaide, S.A. 5005, AustraliacCentre for Energy Technology, The University of Adelaide, S.A. 5005, Australia

a r t i c l e i n f o

Article history:Received 28 September 2010Accepted 4 April 2011

Keywords:Radiation heat transferLaser diagnosticsSootTurbulent reacting flows

* Corresponding author. School of Mechanical EnAdelaide, S.A. 5005, Australia. Fax: þ61 8 8303 4367.

E-mail address: [email protected] (

0360-1285/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.pecs.2011.04.001

a b s t r a c t

Recent advances in diagnostic methods are providing new capacity for detailed measurement ofturbulent, reacting flows that are strongly radiating. Radiation becomes increasingly significant in flamescontaining soot and/or fine particles, and also increases with physical size. Therefore many flames ofpractical significance are strongly radiating. Under these conditions, the coupling between the turbu-lence, chemistry and radiative heat transfer processes is significant, making it necessary to obtainsimultaneous measurement of controlling parameters. These environments are also particularly chal-lenging for laser-based measurements, since soot and other particles increase the interferences to thesignal and the attenuation of the beam. The paper reviews the influence of physical scale and of theproperties of the medium on approaches to perform measurements in such strongly radiating flows. Itthen reviews the recent advances in techniques to measure temperature, mixture fraction, soot volumefraction, velocity, particle number density and the scattered, absorbed and transmitted components ofradiation propagation through particle laden systems. Finally it also considers remaining challenges todiagnostic techniques under such conditions.

� 2011 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .412. Measurement requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423. Need for, and challenges to, measurements in large scale flames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434. Measurement of temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455. Measurement of mixture fraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496. Measurement of soot volume fraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507. Measurements of droplet size, number density and velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 538. Measurements of particle number density and size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539. Measurement of velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

10. Measurement of radiation and its propagation through a scatteringeabsorbing medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5611. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .57

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

gineering, The University of

G.J. Nathan).

All rights reserved.

1. Introduction

Radiation is the dominant mode of heat transfer in most prac-tical high temperature thermal processes, owing to the fourthpower dependence of radiant heat transfer on the temperaturedifferential. For this reason, considerable effort has been expended

mailto:[email protected]

www.sciencedirect.com/science/journal/03601285

http://www.elsevier.com/locate/pecs

http://dx.doi.org/10.1016/j.pecs.2011.04.001



G.J. Nathan et al. / Progress in Energy and Combustion Science 38 (2012) 41e6142

to understand and predict radiation heat transfer and great prog-ress has been made [1e4]. It is well established that the turbulentprocesses are directly coupled to the radiation heat transfer inreacting flows, and these influences can be very significant [4,5].Substantial progress has beenmade in capacity to account for thesecoupling processes, but mostly under conditions where scatteringcan be ignored [6]. However, scattering cannot be ignored in thepresence of soot and/or particulate fuels. Furthermore, the pres-ence of particles such as soot adds greatly to the complexity ofreacting flows, and hence further to the computational require-ments of simulations since it increases the strength of the couplingbetween turbulence, radiation and reactions. Even without thepresence of particles, the development of reliable models requiresdetailed, well-resolved and simultaneous measurements of the keycontrolling parameters. To meet this need, laser diagnostics havebeen developed to provide unrivalled edge over intrusive probes interms of spatial and temporal resolution and species selectivity [7].However, despite their substantial contribution to combustionscience, many of these laser diagnostic tools are limited to appli-cation in environments that avoid the presence of soot d alongwith most of the other complexities of practical systems, whichinclude large physical scale, multiple phases, multi-componentfuels, inorganic species, secondary turbulent motions and pres-sures higher or lower than atmospheric. Nevertheless, a number oflaser-based diagnostic methods are emerging which promise toopen newwindows into these complex processes onwhich radiantheat transfer depends. The aim of the paper is to review thesemethods, predominantly at atmospheric pressure. The influences ofpressure are considered to be beyond the scope of the review, sinceit constitutes a major expansion of the subject and influences eachtechnique in different ways.

The need for improved capacity to understand and predict radi-ative heat transfer in and from flames is driven by the challenge tosupplyever-cleanerenergy.Radiation is typically thedominantmodeof heat transfer in flames containing soot and/or particulate fuels.Hence it influences both energy efficiency and pollutant formation,through temperature [8]. The need to optimise combustion systemsis increasing, owing to ever-increasing need for higher efficiency andlower pollutant emissions. The Intergovernmental Panel on ClimateChange (IPCC) predicts a growth in the utilization of all fuels to 2030by 52%. Also, while a significant drop in world-wide coal utilizationfor power generation from23.9 to 17%over theperiod to 2030, owingto the switch to natural gas fired integrated gas turbine combinedcycles (GTCC), the absolute utilization of coal is expected to increasebyabout 1% [9], and these expectations have been exceeded in recentyears owing to thedemand forenergy inChina. Agrowth in theuseofbiomass and alternative fuels is also expected to increase [10]. Itshould also be noted that, while pyrolysis and gasification mayreduce the proportion of fuels consumed by combustion, this doesnot obviate the need for capacity to provide detailedmeasurements.The use of gasification and pyrolysis offers some advantages inallowing the ash to be retained in the char. However, gasifiers andpyrolysis systems are typically even more complex to investigate indetail than combustion systems, since they operate with very denseparticle loadings, which makes optical access even more difficult.Furthermore, new technologies are emerging such as solar gasifica-tion [11,12] in which the need to obtain detailed understanding andmodelling capacity of radiation in complex media is paramount fordesign optimisation. Hence there is an ever-growing need to over-come the challenges for spatially and temporally resolved measure-ments of the parameters controlling radiation heat transfer incomplex environments.

Computational capability has advanced to the stage where thedirect numerical simulation (DNS) of turbulent reacting systems isnow possible [13]. Nevertheless, DNS of turbulent reacting systems

is unlikely to become a tool of choice for predictive purposes as it islimited to very small physical domains, short flow time-scales,relatively simple flows and relatively simple chemistry [14]. Thisis especially true in the investigation of radiation heat transfer inrealistic systems [6]. Instead, DNS is used to provide complex anddependent correlations for combustion models: essentially smallscale numerical experiments [14]. Likewise, the understanding ofchemical kinetics has advanced to the extent where reliablereduced mechanisms are now available for a significant number ofhydrocarbon fuels [15] and surrogates for practical liquid fuels[16,17]. Nevertheless, a substantial challenge remains to expandthis research to the stage where such schemes are available formany practical fuels [15,16]. These issues mean that the develop-ment of models for practical combustion systems, necessarily witha limited range of validity, will continue to rely on experimentaldata for the foreseeable future for model development and vali-dation. Hence the measurement of temperature, along withmixture fraction and particle volume fraction, on which emissivitydepends, are critical to determining radiation. Furthermore,because radiation propagates along rays, there is a significantadvantage in planar measurements, from which line-of-sight datacan be obtained. Simultaneous planar measurement of theseparameters is therefore highly desirable.

The key challenges to the extension of laser diagnostics instrongly radiating turbulent, reacting flows, which typically involvesoot or other particulate fuels, are listed below. These challengesincrease further with larger physical scale:

� Increased background interference: Optical interferenceincreases with the presence of any particles, which radiateincandescently and/or scatter laser light. The natural back-ground radiation from a flame also increases with the size ofthe flame and the use of furnace walls.

� Optical attenuation: The attenuation of both the incominglaser beam and the outgoing laser-generated signal (termed“signal trapping”) increases with the introduction of particlesand/or the size of a flame, as do the effects of beam steering.

� Reduction in spatial fidelity: Both beam steering due todensity gradients, and optical diffraction due to particles, leadto an increase in the uncertainty in spatial location over theissues of beam quality present in any flow.

� Increased number of parameters to measure: The number ofparameters to measure increases dramatically with the use ofparticulate, multi-component solid fuel particles that containimpurities (i.e. inorganic components). These include particlesize, shape, number density, composition and temperature.

� Reduced optical access: Investigations of large scale and/orpressured systems entail reduced optical access into the flameand may require optical probes to be inserted into the flame toachieve sufficient optical penetration.

Given these challenges, the aim of the present paper is to reviewprogress and ongoing challenges in present ability to performspatially and temporally resolved measurements of increasingrelevance topractical reactingflows. Particularemphasis isplacedonthe use of planar measurement techniques, since these offer manyadvantages in theunderstandingof turbulent reactingflowsandalsoallow for ray tracing of radiation. Despite their practical significance,the investigationofpressurisedcombustionand themeasurementofinorganic species are outside the scope of the present review.

2. Measurement requirements

The requirements for measurements to be suitable for thedevelopment and validation of detailed models in turbulent

Table 1Typical operating conditions for industrial-scale flame and a 1/10th scale model.

d (m) U (m/s) dp (mm) LF (m)

Reference 1 30 100 30Model (1/10th scale) 0.1 See Table 2 100 3

G.J. Nathan et al. / Progress in Energy and Combustion Science 38 (2012) 41e61 43

reacting flows have been reviewed previously [18]. Suchmeasurements should:

� employ well defined initial and boundary conditions,� measure multiple parameters simultaneously,� measure multiple dimensions d planar or 3D,� resolve spatial and temporal gradients of interest.

Even apparently small changes to the in-flow conditions, such asdifferences in a boundary layer profile, can influence a flow or flameas they propagate into the far field [19]. This issue has driven theneed for well-characterised in-flow and boundary conditions forflames to be suitable for model development and validation d aprocess well demonstrated by the International Turbulent Non-premixed FlameWorkshop [20].

The challenge to achieve well-resolved and simultaneousmeasurement of key controlling parameters in turbulent systemshas been driving research for many years [18]. This need arisesbecause of the unsteady and non-linear relationship between thecontrolling parameters of even isothermal turbulent flows, whichincrease with reactions. Considerable effort has thus been investedto define and resolve the smallest scale of turbulent flows, theKolmogorov scale, lK, and of turbulent mixing, the Bachelor scale,lB, along with scalar gradients and the scalar dissipation [21e23].The challenge to resolve scalar dissipation in reacting flows hasonly recently been met with the breakthrough work of Wang et al.[24]. Such methods could, in principle, be extended to larger scale,which would be valuable for systems such as lean pre-mixed gasturbines, where localised extinction under high strain is a particu-larly important issue.

The controlling parameters for turbulent flows transportingreacting particles are strongly coupled, providing increased needfor simultaneous measurements. A notable example is the coupleddependence of soot volume fraction on temperature [25,26]. Boththe formation and oxidation of soot depend upon fuel type, mixturefraction, x (the local mass fraction of species originating from thefuel stream), temperature, T [27], and on the residence time (andhence on strain rate) in the reaction zone [28]. At the same time, Talso depends on soot concentration, which dominates radiant heattransfer, and on turbulent mixing through fluid properties. Like-wise, the rates of heating and reaction of fuel particles depend onthe local T, and the volatile gases and energy they release influenceT. The processes of droplet evaporation and gasification or devo-latilisation are also endothermic.

The many large-scale coherent motions inherent in turbulentsystems make planar measurements desirable. Planar imagespermit the acquisition of spatially correlated measurements andthe measurement of scalar gradients, which are useful both inscientific research and in the study of practical combustion systems[29]. They also provide physical insight that is not possible withsingle-point measurements.

3. Need for, and challenges to, measurements in large scaleflames

Thermal processes become increasingly cost-effective and effi-cient with increased physical size. There is therefore an ongoingneed to develop models that can be applied at ever-increasingscale. At the same time, model development and validation areexpensive, and are inevitably undertaken predominantly at smallerscale where more detailed access is possible. This raises the chal-lenge that flames of sufficiently different physical scale can exhibitquite different behaviour. This point is illustrated below bycomparing four of the key dimensionless parameters that charac-terise some of the most important physical processes in a typical

full-scale pulverized fuel flame and a geometrically similar modelat 1/10th scale. It is particularly the case with radiation heattransfer in the presence of soot and fine particles, where sufficientlylarge scale can result in the attenuation of radiation. These differ-ences mean that reliable prediction of the processes on whichradiation depends at industrial-scale requires experimental data atcomparable scale, and cannot rely solely on data at low Reynoldsnumber and/or small physical scale. The optical thickness of a flamealso increases with physical scale [6] and the radiative properties offires become fundamentally different at sufficiently large scale [30].

Table 1 presents the key physical parameters of the burner andflame. Here d and U represent a reference diameter and bulk meanvelocity characterising the burner, while LF is the length of theflame. In practice all industrial burners are complex, with multiplestreams of fuel and oxidizer, often containing swirl, etc. However,since the reference and model systems are geometrically similar,any single reference diameter can be used, so long as it is consis-tent. Likewise, the fuel particles inevitably comprise a wide distri-bution, but provided that same relative size distribution is used, anyreasonable reference size can be used. In principle it is possible tooperate with reduced size particles, but in practice this is rarelydone owing to the added cost and difficulty in reducing the sizesufficiently to maintain similarity. The dimensionless parameterscompared are the Reynolds Number, Re ¼ rUd/m, the characteristicresidence time (e.g. of a coal particle) in a flame, sres ¼ LF/U. Theratio of flow speed to laminar flame speed, U/SL and the StokesNumber, Stk ¼ rpUd

2p=18mLt , which describes the ratio of particle

response time to a characteristic eddy time. Here Lt is the lengthscale of the large turbulent eddies, which also scale with LF and theother symbols have their usual meanings, U is the characteristicvelocity, m is the absolute viscosity and r is the fluid density.

From the data in Table 2 it is clearly not possible to selecta velocity for the pilot-scale model that does not result in at least anorder of magnitude change in at least two dimensionless parame-ters, and this list of dimensionless groups is not intended to beexhaustive. Of course, the influence of many dimensionlessparameters is typically asymptotic. For example, a change in Re willnot have a large influence when it is sufficiently high to be in thefully turbulent regime. Hence, a pilot-scale program should bedesigned so dimensionless groups with the greatest influence areanalogous to real conditions [31]. Without this, a smaller scalefacility could operate in an entirely different regime (or mode) ofbehaviour. Nevertheless, some compromise is inevitable andmodelvalidation at the laboratory scale will not automatically result inreliable prediction at the industrial scale.

Further evidence of the influence of physical scale on perfor-mance of flames is found in the scaling assessment of Weber [32].Fig. 1 presents the NOx emissions measured from a series ofgeometrically similar burners that are scaled by constant velocityscaling. In this figure, all burners were geometrically scaled, andwere operated at either reduced velocity to achieve constant resi-dence time-scaling or at constant velocity to achieve constantvelocity scaling as per the principles illustrated in Tables 1 and 2. Itis evident that the emissions depend on physical scale, and thatthere is no simple way to scale from one flame to another.

The requirement for data at realistic physical scales is wellknown, and extensive measurements have been performed atpilot-scale, to contribute to model development and validation.

Table 2The influence on key dimensionless parameters of the model relative to full-scale,caused by the choice of characteristic velocity (from Table 1).

Scalingmethod

Umodel (m/s) Remod/Refull U=SL mod=U=SL full smod/sfull Stkmod/Stkfull

Const Re 300 �1 �10 �0.01 �100Const U 30 �0.1 �1 �0.1 �10Const s 3 �0.01 � 0.1 �1 �1


An example of such a program in boiler flames is the extensivedata-sets from International Flame Research Foundation [33,34].Despite their value, such data were, for many years, typicallylimited to mean values of temperature and stable species, asobtained by suction probes, and to two-component velocityobtained by laser Doppler anemometry (LDA). Pilot-scale investi-gations assessing heat flux and total emissions have also playeda vital role in technology development [34e36]. More recently, theextensive investigations by Sandia National Laboratories of largerscale pool fires have included single-point measurements of sootvolume fraction [37] and temperature [38] and planar measure-ments of soot volume fraction [39]. These measurements wereobtained using semi-intrusive laser diagnostics in which opticalcomponents are inserted into the flame in a water-cooled housingto reduce the path-length through the flame beyond themeasurement region, either between the laser and the measure-ment volume and/or of the signal between the probe volume andthe detector. These approaches have provided significant new dataand insights while also illustrating how optical tools can be appliedin large-scale, harsh environments. These have provided significantnew insights and demonstrate the benefits that are possible withfurther developments in laser diagnostic measurement capability.Nevertheless, such data are uncommon and there remainsa significant need for reliable data in flames of significant scale forthe development and validation of models of radiation heattransfer [6].

Fig. 1. The influence of physical scale on the NOx emissions from burners scaled usingconstant velocity scaling [32].

Turbulent mixing occurs at a wide range of scales. The smallestscale of turbulence is characterised by the Kolmogorov scale, lK,which can be estimated for a jet from the well-known relation ofAntonia et al. [40] to be:

lK ¼ LC1 Re�0:75L : (1)

Here L is the local width of the jet, which in turn scales with thenozzle diameter, d, while C1 is a constant (for a non-reactingsystem), which for a cold jet is estimated to be 2.4. Hence thesmallest scales of turbulence, lK are typically about the same atindustrial scale as at the laboratory scale, while the largest energy-containing turbulent eddies that typically control mixing, i.e. L, aremuch greater at the industrial scale. That lK is typically similar atsmall and large scales results from the practical limitation on theflow velocity to avoid an excessive increase in strain rates. Forexample, a flame becomes more stable with an increase in burnerdiameter due to the decrease in strain, resulting in increased blow-off velocity and reduced lift-off height [41] (note that lift-off andblow-off are also influenced by the nozzle profile through exit strainprofile [42]). Increasing the strain, which occurs even at constantvelocity by reducing the diameter, also leads to a reduction in sootvolume fraction [43] and luminosity [44]. To mitigate these influ-ences, the exit velocity is typically either held constant or reducedwith reduced scale. Even for the case of constant velocity scaling atconstant temperature, lK decreases only with the reduction in scaleto the power of 0.25. Likewise at constant velocity and temperature

L=lKf�dfull=dmod

�0:75: (2)

A similar ratio will apply for the Batchelor scale [21]. This meansthat the difficulty of simultaneously resolving both large and smallscales increases with the scale of the facility. Planar measurementsare best obtained with collimated light sheets, the maximumphysical dimension of which is typically limited to around 50 mmfor readily available optics, with larger optics being very muchmore expensive, although they are technically possible. Ability toincrease the sheet size is also limited by available laser power andby the reduction in spatial resolution associated with a largermeasurement area and the size of a given pixel array on a CCDdetector. Nevertheless, these challenges can be expected todiminish with advances in optical technology, as high-poweredlasers become cheaper and as imaging detectors are developedwith increasing resolution (spatial and dynamic) and frame rate.

While the challenge of increasing the scale of the facility onmeasurement resolution is common to reacting and non-reactingfacilities, the challenge of increased background interference isunique to combustion. The accuracy of a laser-based measurementdepends on the strength of the signal relative to the backgroundoptical noise. The background interference, Ib, comprises compo-nents both from the flame and the furnace walls, also for constantvelocity scaling, and can be estimated to scale as follows:

IbzeFsAF

�T4F � T4

amb

�þ eWsAW

�T4W � T4amb

�: (3)

Here eF and eW are the emissivities of the flame and furnace walls,respectively, s is the StefaneBoltzmann constant, AF is the pro-jected area of the flame, AW the area of the furnace wall viewed bythe imaging device, while TF and TW are the temperatures of theflame and furnace walls.

Mostmeasurements in a laboratoryareperformed inopenflames,avoiding the interference fromawall entirely. This cannot be avoidedin many realistic flames. The squared dependence of interferencefrom the natural flame radiation with the diameter of the flameresults from constant velocity scaling, inwhich case the flame power

Fig. 2. Broadening of the Rayleigh line-shape as a function of temperature, in relationto the transmission of an iodine filter, as used for Filtered Rayleigh Scattering (FRS)thermometry [53].


scales with the square of burner diameter (see the earlier discussionof Equation (2) for an explanation for the relevance of constantvelocity scaling). Thebackground interference fromawall scaleswiththe cross-sectional area of the wall imaged by a fixed optical accep-tance angle, i.e.with the square of the path-length from the detectorto the wall, with the additional assumption of geometric similarity.The equation ignores both the attenuation of radiation by the flameand the reflected radiation, but also accounts for only the emittedradiation,while the reflected radiationwill also scale similarlywhichcompensate for each other somewhat. Nevertheless, the equation issufficient to show that the natural flame radiation will thereforebecomes much more significant in larger scale flames than it is inlaboratory-scale flames. This highlights the increased difficulty inproviding reliable measurements at realistic scale. Finally, the back-ground interference from the flame depends on both its temperatureandemissivity.Mostpracticalflamesoperate atelevated temperaturerelative to laboratory flames, owing to the combustion air beingpreheated. Furthermore, theemissivityof aflame isdominatedby thepresence of particles, and inparticular of soot. The volume fraction ofsoot within a flame (fv) scales inversely with the intensity of strain ina flame [45]. While local strain rates fluctuate with both space andtime, they can be characterised by the exit flame strain rate, u/d or itsinverse the characteristicflame time [43] or by the globalmixing rate[46]. It is readily apparent that this corresponds to 1/sres. Returning toTable 2, it is evident that global strain rates decrease dramatically(typically by from one to two orders of magnitude, depending on thescaling approach)with increased scale. The reducedglobal strain ratesignificantly increases the presence of soot inflames of realistic scale,and so also the background interference.

The combination of the above challenges associated withincreased scale means that complete detailed and simultaneousmeasurements have only been performed in small-scale laboratoryflames with ambient temperature air, no soot and without thepresence of furnace walls [47,48]. Nevertheless, significant progresshas been made in the development of measurement approaches ofkey parameters within larger scale flames.

A range of optical measurements has been performed infurnaces. In some cases the optical access is provided through viewports in the walls of the furnace. However another approach is toinsert optical components into the flame, using water-cooledjackets to protect them. This has been termed “semi-intrusive”devices. The International Flame Research Foundation playsa leading role in developing semi-intrusive systems first for single-point Laser Doppler Velocimetry (LDV) systems [49] and then forplanar imaging [50]. More recently Sandia National Laboratory hasdeveloped a water-cooled probe to transfer an image to an ICCDcamera via a fibre-optical coupling for measurements of sootvolume fraction in a metre-scale pool fire [51].

Another ongoing challenge related to the investigation of prac-tical reacting flows is the interaction of the reacting flows with thewalls of a furnace or combustion chamber. Depending on the type ofdevice, the vessel walls may be substantially hotter (e.g. a refractoryquarl) or colder (e.g. boiler tubes and cylinder walls) than the adja-cent reacting flow. The boundary layer therefore exhibits highgradients in both temperature and velocity. Hence these regionshave a significant influence onheat loss or gain, local extinction and/or ignition,flamestabilityandpollutant formation.Most laser-basedtechniques areunable to resolve theboundary layer close to thewall,owing to the strong interference by scattering that occurs close toa surface. This issue is still unresolvedand requires further attention.

4. Measurement of temperature

Temperature is a dominant parameter in combustion processesand the key parameter in radiation heat transfer owing to its fourth

power influence on heat transfer. It characterises the enthalpy ofreaction and controls many of the important chemical and physicalprocesses,which also influence composition. Avariety of laser-basedthermometry techniqueshavebeendeveloped [7]. However,mostofthese are limited to clean combustion environments and are notapplicable in the presence of particles, such as dust, coal, biomassand soot. Absorption, scatter and other interferences due to thepresence of soot and its precursors prevent many laser diagnostictechniques from being applied reliably. This limits the capacity toinvestigate and understand many systems of practical significance.

A relatively simple, yet useful, laser-based thermometry tech-nique is Rayleigh scattering. The elastic scatter of light frommolecules gives a measure of the total number density which,when coupled with ideal gas law and knowledge of the Rayleighcross section, allows the temperature to be deduced. However, todetermine the Rayleigh cross section requires the concentration ofall major species to be known, which in turn requires goodknowledge of the flame chemistry. This presently limits its appli-cation to relatively simple or tailored fuels that react withoutforming soot [52]. Another constraint of particular relevance to thepresent paper is that Rayleigh scattering is highly susceptible tointerference from the elastically scattered light from particles (Mie-scattering) and surfaces. It can therefore only be employed undervery clean, particle-free environments, which limits its applicabilityin practical combustion systems [53]. Filtered Rayleigh scattering(FRS) is an extension of the conventional Rayleigh scattering thatwas first proposed by Miles et al. [54] and widely investigatedsubsequently [53,55e58]. The FRS technique utilises a narrow-band filter at the centre of the frequency of a single mode laser toreject interference that is spectrally identical to the incident light.The FRS technique has been primarily used with molecular-iodinefilters paired with frequency doubled single mode Nd:YAG lasers.The unfiltered component of the pressure- and/or temperature-broadened signal of the scattered light is subsequently used forthermography. The broadening of the Rayleigh line-shape in rela-tion to an iodine filter is apparent in Fig. 2, reproduced fromHoffman et al. [53]. The filter allows temperature imaging to beperformed in the presence of sufficient elastic scatter to otherwiseobscure the Rayleigh scattered signal. However, this comes at theexpense of reduced signal-to-noise ratio, which limits its applica-bility to moderately radiating flames. An example of the impact ofthe filter on the Rayleigh scatter from a mildly sooty, slightly pre-mixed methane-air flame may be seen in Fig. 3, reproduced fromHofmann et al. [56]. It is apparent in the bottom image (collectedwithout a filter) that the scattering from the soot sheets dominates

Fig. 3. Two-dimensional image of temperature obtained using Filtered RayleighScattering (top) and corresponding scattering image (bottom) collected withoutfiltering [56].


the image. When the filter is used (top figure) the temperature maybe determined from FRS without the interference from the soot.

Despite the improved capability of FRS over unfiltered Rayleighscattering, it remains susceptible to variation in the Rayleigh crosssection across the reaction zone, which can be difficult to accountfor in turbulent imaging applications since it requires detailedknowledge of all components in the measurement volume. Zhaoand Hiroyasu [59] provide more information of the application ofRayleigh scattering in reacting environments.

One approach to circumvent problems associated with elasticscatter is to employ the inelastic Raman scattering for thermometry[52]. However, the inherently low signal from spontaneous Ramandoes restrict its application. It is typically limited to point or linemeasurements. Nonetheless, point-wise temperature measure-ments have been collected in practical devices in soot-free envi-ronments, such as a pre-mixed gas turbine swirling flame [60].

Offering stronger signal due to its coherent nature, Coherentanti-Stokes Raman Spectroscopy (CARS) is more applicable to

Fig. 4. CARS excitation processes for nanosecond and f

thermometry in luminous and particle laden flows [61]. IndeedCARS has been the most widely used thermometry technique forharsh combustion environments, such as gas turbine combustors[62], liquid fuel combustors [63] and sooty turbulent pool fires [38].

The basic operating principle of CARS [7,52] involves the use ofthree laser beams (typically two of the same wavelength) thatinteract in the measurement volume to generate the signal (asa fourth beam) as a result of the third-order non-linear suscepti-bility [7,52]. A ‘pump’ beam at frequency upu (non-resonant tomolecular transitions) excites a virtual level. By tuning a secondStokes beam to frequency uS, where upu � uS corresponds toa vibrational or rotational transition of the molecule, Raman reso-nance occurs. A third, ‘probe’ beam (upr) is then used to generatea coherent CARS signal at uCARS ¼ upu � uS þ upr. By scanning theStokes beam across the Raman transitions of the molecule, thespectral shape is used to determine the temperature. Scanning ofthe Stokes wavelength may be performed on a shot-by-shot basis,or alternatively a broadband laser enables single-shot spectra to begenerated. A particular advantage of CARS is that it can be appliedto the N2 molecule, which is ubiquitous throughout the flame,allowing measurement everywhere and avoiding any need forseeding, with its many complications. The advent of femto-secondlasers is particularly important for the continued development ofsingle-shot CARS for turbulent environments. The inherentfrequency spread of femto-second laser pulses can be used to excitemultiple pump-Stokes pairs, as shown in Fig. 4 [64].

A key advantage of the CARS technique for assessing practicalsystems results from the frequency shift in the signal relative to theexcitation beams. This allows filtering to separate out the influenceof scattered interference from soot. Careful selection of the exci-tation scheme is necessary to minimise soot interferences [65].Approaches include a dual-pump CARS technique with an annularphase matching geometry (USED CARS) [66] and shifted vibrationalCARS [67]. In exciting N2, O2 and CO2, this approach has the furtheradvantage of avoiding seeding, and accessing species that arewidely present in the flame. In addition the signal is coherent,allowing the signal-to-noise be increased by moving the collectionoptics further from the flame (since the background is incoherent, itattenuates with r2). The use of short pulse (sub-nanosecond) laserexcitation has the further advantage of allowing the signal to bedelayed relative to the non-resonant interference, providing highsignal quality and effectively eliminating interference from soot[68].

Despite these advantages, CARS suffers from a number ofdisadvantages for application in turbulent flames. The necessity forline-of-sight optical access and the experimental complexity arepractical limitations, although the line-of-sight limitation is prob-ably no less than the normal viewing for planar imaging. However,more fundamentally, the beamconfiguration results in an elongated

emto-second laser pulses. Adapted from Ref. [64].


probe volume of typical dimensions 1.5 mm� 60 mm [63]. This lackof high spatial fidelity in comparison to planar techniques haspreviously been recognised as restricting the generic application ofCARS [7]. In addition, its reliance on three beams with differentoptical paths means that the method is subject to differential beamsteering and differential attenuation in a turbulent environment,especially in the presence of soot. Beam steering is the process bywhich the path of a beam wanders with the propagation of thedensitygradients througha turbulentmedium. It iswell known that,in a multi-beam experiment such as CARS or LDV, this differentialwandering of the beams reduces the probability of their coincidentalignment, and hence of the data rate. Whilst this does not neces-sarily reduce measurement accuracy, it can do. To avoid biasing themeasurement it is necessary firstly, that the differential beamsteering is a truly random process. However, where differentialbeam steering is correlatedwith the passage of coherent motions, itwill introduce an unknown bias in the measurement, since thetemperature distribution is also typically correlated with thesemotions. Further, any differential behaviour of the beams, throughdifferential focussing, will also introduce errors. In addition,a reduction in data rate will, in practice, typically lead to a reduceddata size and thereby increase the measurement uncertainty. Moredetails of the CARS technique can be found in the recent review byRoyet al. [69]. Finally, themethodhas hitherto been limited to singlepoint and nomethod has yet beenproposed to allow its extension toplanar measurement, as is desirable for investigation of turbulentsystems and compatibility with other planar measurements.

The signal collected from laser-induced fluorescence (LIF) istypically spectrally-shifted from that of the excitation, allowing itto be separated from the scattered interference that plaguesRayleigh scattering. The fluorescence signal is usually quite strong,making LIF better suited to two-dimensional imaging than Ramanand Rayleigh techniques [29]. Various strategies exist for ther-mometry, all of which are based upon the species’ populationaccording to the Boltzmann distribution. The species of excitationcan be naturally occurring or artificially seeded.

The use of in-situ species (e.g.OH, CN, CH) for LIF thermometry islimited because of their lowconcentrations, and the narrow regionswithin the flame in which they typically exist. Of the naturallyoccurring species within a flame, the OH radical has been mostcommonly used [70e74]. In non-premixed flames, OH only existsover a small range of temperature and mixture fraction, and so isnot well suited for general measurements [75,76]. Furthermore, infuel-rich flames the OH concentration is low, making it poorlysuited to simultaneous measurements with soot, which is typicallyfound on the fuel-rich side of the reaction zone [77]. Nitric oxide(NO), which is formed natively during the combustion process,could potentially be used for thermometry, though additional NO istypically added to the inlet streams to give a much improved signalquality [70,78]. However, in the presence of soot, backgroundinterferences can lead to a significant error in two-line NO-LIFthermometry [79]. Multi-line approaches for NO thermometryalleviate issues with background [80], but at the cost of beingrestricted to time-averaged results. The excitation wavelengthsrequired for NO (around 226 nm) can also cause problems withinterference and attenuation, and pressure broadening of theexcitation lines leads to spectral overlap [81]. Furthermore, in fuel-rich flames, NO is consumed due to reburn reactions, leading toa loss of signal [78,82,83]. As an alternative to naturally occurringspecies, a wide range of seeded species can be introduced, of whichorganic molecules are most common [84].

The accuracy of single-wavelength LIF thermometry relies on theknowledge of the quenching species densities and quenching rateconstant, since the spectra and intensity of LIF are influenced by thevariation in collisional processes across reaction zone. However, this

problem can be avoided by taking the ratio of two excitation/detection wavelengths [29,85]. The number density, absorbingspecies concentration and the collisional quenching dependenciesthat are often associated with single line LIF techniques are alsonullified by taking the ratio of the fluorescence signals. Fig. 5 pres-ents an example of two-line LIF using 3-pentanone as a fuel tracer ina two-stroke internal combustion engine [86]. It is also possible toemploy single laser excitation and detection at two differentwavelengths, e.g. with toluene as the tracer [87]. However, as withmany fuel tracers, 3-pentanone and toluene are consumed in thereaction zone, and so measurements are only possible in the lowtemperature, unburned regions.

Alternative two-line fluorescent techniques have also beenexplored. For aqueous systems, the use of two different laser dyeshas been proposed [88]. For simultaneous thermometry andvelocimetry, the use of liquid-crystals has also been proposed [89].However, their relevance to turbulent flames is not clear, since theyhave a limited temporal response and do not survive the combus-tion environment. Two-line phosphorescence using ZnO:Zn andZnO:Ga has also been reported, with claims of short lifetimesenabling elimination of background interference [90]. Temperaturemeasurements using naphthalene have also recently been reported[91]. However, the feasibility of these techniques in practical flameenvironments is yet to be adequately demonstrated.

Of the laser-based thermometry techniques, two-line atomicfluorescence (TLAF) is emerging as a particularly useful techniquefor many applications in sooting environments, especially wheretwo-dimensional imaging is desired. Themethodwasfirst proposedby Alkemade [92], and shortly after by Omenetto et al. [93,94] andHaraguchi et al. [95,96], who demonstrated its viability in the linearexcitation regime. Unfortunately, a linear relationship between thelaser fluence and fluorescence signal only extends to very low laserfluxes, so that sufficient signal-to-noise ratio for useful measure-ments can only be obtained from an average image. This limits itsoperation in the linear regime to application in laminar flames.Subsequentwork demonstrated good sensitivity over a temperaturerange relevant to combustion, insensitivity to collisional quenchingeffects and capacity to separate the inelastic fluorescence frominterference by spurious scattering [75,97e101]. The developmentof suitable diode lasers has also lead to interest in theTLAF technique[102e104]. More recent work with operation in the non-linearregime, described below, and the correct solvent, has achievedsingle-shot imaging [105e108].

Of the atomic species available, indium seeded into the flame hasbeen identified as a suitable thermometry species for TLAF [95].Indiumhasgoodsensitivityover the temperature range800e2800K[100] and both of the wavelengths are in the visible spectrum (viz.410 nm and 450 nm), where interferences are less pronounced thanin the UV range typically employed for LIF excitation. Feasibilitystudies [75,97] have shown that LIF excitation with indium holdspromise for temperature measurement in a moderately sootingenvironment. Application under more strongly sooting environ-ments appears to be achievable, but requires demonstration ofamethod to correct for interferences from soot precursors [109]. Theindium is typically introduced into the fuel as a fine mist of indiumchloridedissolved ina solvent (methanol orwater).Within theflamefront, neutral indium atoms are generated, which are spectroscopi-cally probedusing theTLAF technique. Sinceneutral indiumis highlyreactive, it is readily oxidised, so that signal is only available from thefuel-rich through to the weakly oxidising side of the reaction zone.This also explainswhy there is little advantage fromseeding into theoxidiser stream.

The strong transitions of atoms are easily saturated. To remainin the linear excitation regime, which is required for insensitivityto collisional quenching effects, the laser energy must be limited

Fig. 5. Instantaneous temperature images from two-line LIF of 3-pentanone in a two-stroke internal combustion engine. Black colouration indicates burned areas. Repro-duced from Einecke et al. [86].


to low fluences. This leads to weak signal strength, and a lowsignal-to-noise ratio, so that time averaging is typically necessary.Medwell et al. [105] extended the technique into the non-linearregime, which allows five orders of magnitude greater fluenceand achieves sufficient signal-to-noise ratio for single-shotimaging. While application in the non-linear regime requires theuse of three empirical constants, these have been shown to beinsensitive to fuel type or operating conditions. Single-shotimaging of temperature has now been demonstrated with andwithout the presence of soot [105,106], and performed simulta-neously with laser-induced incandescence (LII) of soot volumefraction, as shown in Fig. 6 [107].

The strong signal strength of the non-linear excitation regimeTLAF (NTLAF) results in good signal-to-noise ratios and measure-ment uncertainty of about 60 K within its operating range.However, the highly reactive nature of the indium ions means thatthey are readily consumed under oxidising environments. Thislimits the range of local stoichiometry for whichmeasurements canbe performed toF> 0.9. Note that the consumption of indium doesnot directly influence the accuracy of the temperature

measurements, which is determined from the ratio of the intensityof the two colours (Stokes and anti-Stokes). However it mayincrease the measurement uncertainty by reducing the signalrelative to noise. This is a fundamental limitation of the method.However, further advances in seeding approaches are potentiallyachievable, which may allow the method to be extended furtherinto the lean regime beyond its present limitation of F > 0.9 and totemperatures below its present limit of 800 K [105,107,108].Measurements to date have only been demonstrated underconditions of relatively low soot volume fraction, owing to theincreased interference with higher soot loading that occurs due tocondensed species and/or polycyclic aromatic hydrocarbons (PAH)on the fuel-rich side of the soot particles [109]. These interferencesoccur either with off-wavelength excitation and on-wavelengthcollection, or with off-wavelength collection and on-wavelengthexcitation. On this basis it is possible to significantly reduce inter-ference by separately imaging the interference and subtracting itfrom the signal, although this is yet to be demonstrated and to do sowould require an additional measurement, preferably with both anadditional laser and camera.

Since the surface temperature of particles is generally not thesame as that of the surrounding gas, it is also desirable to measuresurface temperature of the soot particles. The leading method toachieve this is via optical pyrometry, which measures the naturalradiation emitted from hot objects, whose intensity varies withwavelength as described by Plank’s law:

Ib;lðTÞ ¼ C1�pl5

�eC2=lT � 1

��1; (4)

where C1 and C2 are the Planck’s constants.This relatively simple and inexpensive method requires the

detection of the light intensity at more than one wavelength, withtwo-colour pyrometry being the longest and best-establishedapproach [110,111]. The choice of wavelengths is very important asthey need to be separated sufficiently to yield a significant intensitychange, but not so far as to introduce a significant change inemissivity, which is also a function of wavelength. To achievea meaningful measurement the detector must be calibrated againsta black-body emitter. In addition, careful selection of the detectionwavelength is needed to avoid interference from the emission frommolecules, atoms or radicals. The method has been successfullyapplied to a wide range of stationary reacting particles [112e115].However, its main limitation in the context of seeking well-resolvedmeasurements in a turbulent environment is that, becausethe signal is integrated along a line of sight, it has relatively poorspatial resolution.

Lee andNa [116] reported soot temperaturemeasurements usingon two-colour pyrometry, employing the Abel inversion to extractradial profiles from line-of-sight measurements through anaxisymmetric diffusion flame at pressures of up to 0.4 MPa. Othermeasurements in laminar diffusion flames have been undertakenusing planar imaging to assess the effect of pressure [117] andbuoyancy in n-heptane fuels [118]. These two-dimensional soottemperature profiles have been used to aid in the analysis andunderstanding of soot production at different pressure, but are oflimited value in turbulent environments. While deconvolutionmethods such as the Abel transformation are appropriate for steady,laminar flames, they can only provide an “averaged” profile ina turbulent environment. However, such line-of-sight measure-ments donot record a truemeanmeasurement since theyare biasedto the hotter regions of soot along the path [119]. Unfortunately, it isnot yet realistic to perform measurements over path-lengths shortenough to resolve the steep gradients in a flame. Also, just as theinstantaneous soot distribution is very different from the mean

Fig. 6. Instantaneous planar measurement of temperature in the presence of soot,measured using Non-linear excitation regime Two-Line Atomic Fluorescence (NTLAF),recorded simultaneously with soot volume fraction by Laser-Induced Incandescence(LII) in a wrinkled laminar flame: a) Stokes fluorescence; b) Anti-Stokes fluorescence;c) Temperature; d) Soot volume fraction; e) Temperature and soot volume fraction.Reproduced from Chan et al. [107].


[46,120], so too the local, instantaneous temperatures differ greatlyfrom themean. Jenkins andHanson [121] reported a variation of theoptical pyrometry method that combines emission and absorptionat the same wavelength to reduce the uncertainty in the soottemperature measurements. However it is does not overcome thechallenge of resolving over long path-lengths. A multicolourimaging-based pyrometric systemwas applied in a pulverized coal-fired flame to monitor the temperature distribution [122]. In thissystem a high-resolution charge-coupled device camera wasemployed to collect three beams of the light of flame. An averagetemperature distribution was reported from the three resultingimages. Fu et al. [123] applied optical pyrometry in a dieselcombustion flame to provide ameasure of the soot temperature. Anoptical fibre spectrometer and a CCD camera were used to providea planar image, but the out of plane measurement was integratedover the path-length of the combustion chamber. Hence it remainsanongoing challenge toextendoptical pyrometry to thepointwhere

it can provide planar measurements that match the spatial resolu-tion of laser-based measurements.

5. Measurement of mixture fraction

Mixing plays a key role in radiation heat transfer through itsinfluence on composition, and so on emissivity, which dependssignificantly on the presence of specular gases such as CO2 and H2O,and of soot [6]. Mixing also influences the control and emission ofpollutants. Both large scale and molecular mixing are important indetermining flame shape, stability and burning characteristics. Thisrole is less important in pre-mixed flames, although turbulence canstill play a role through strain and residence time.

Mixture fraction, x, is a conserved scalar which represents themass fraction of the local gas that originated in the fuel stream. Thisscalar is used in many combustion models to reduce the number ofvariables needed to represent the mixing process and, through it,the local flame structure. Models such as Conditional MomentClosure, CMC, and the Laminar Flamelets Model are based on themixture fraction, which is calculated from species concentration[124e126].

The process of model development and validation requiresreliable experimental measurements of mixture fraction in reactingflows that have been performed under conditions where experi-mental and computational data can be directly compared. Suchdata allow researchers to focus on the compositional structurewithout the complication of flame shape and size. Thesemeasurements are usually conducted on one ormore of the key fuelcomponents that are spectroscopically accessible and whosekinetics are well understood. Measurements are done eitherthrough intrusive sampling probes and gas chromatography orthrough temporally and spatially resolved laser-based techniques.The advantages of laser diagnostics over probe-based measure-ments are well documented. However, even with the currentmethods of laser diagnostics, some challenges remain to accuratelymeasure the mixture fraction in all parts of the flame.

It is possible to measure all of the major species and deduce themixture fraction based on elemental conservation. However, suchmeasurements are restricted to point measurements and 1-D linemeasurements of late. Planar imaging of the fuel has two mainchallenges, which are the early dissociation of the fuel (such asCH4), on the rich side of stoichiometric, and the effect of differentialdiffusion. In addition imaging of the fuel alone restricts themeasurements to the rich side of the flame. Fuel dissociation resultsin very low concentration of the probed species and results in poorquality signal around stoichiometric conditions. Differential diffu-sion arises when molecular diffusion is more significant in onecomponent of the mixture than the others, and especially signifi-cant for multi-component fuels containing H2. This leads toa misrepresentation of the mixing when only one part of the fuelcomposition is measured.

An alternative to the direct measurement of all of the fuelspecies is the use of a fuel tracer. The most widely employed tracerspecies employed to date for this purpose are fuels such as acetone,3-pentanone or toluene, although more recently krypton hasemerged as an exciting alternative, at least for gaseous fuels [127].

The fuel-based tracers are most suitable for liquid fuels, but alsosuffer from a range of systematic errors that limit their applicabilityto the measurement of fuel mass fraction. These limitations includethe difference between spray formation and evaporation of the fueland the tracer (noting that these tracers are typically liquid),different ignition chemistry and their breakdown at temperatureswell below that of the reaction zone. Also, their quantificationdepends on understanding the effect of molecular quenching,correction for interferences and the need to quantify the

Fig. 7. Experimental arrangement for the simultaneous imaging of methane Ramanscattering, Rayleigh scattering, and OH-LIF [129].


temperature to determine the population distribution, known asthe Boltzmann distribution. Higher pressures add furthercomplexity to resolve.

Schulz and Sick [84] reviewed the techniques used to measurefuel mass fraction and temperature by LIF, with a focus on appli-cation in internal combustion engines and at high pressure. Theyconcluded that LIF is one of the more robust techniques formeasuring fuel concentration in such environments, and also in thepresence of particles. They noted that the selectivity of the LIFtechnique allows the probing of the specific species, either added astracers or found naturally in the flame. Schulz and Sick have alsohighlighted the difficulties associated with the quantification of themeasured species and their relevance to the actual fuel mass frac-tion. They further identified the lack of discrimination betweenliquid and gas phase signals, as a remaining challenge for tracer-LIFin many applications. Indeed the measurement of mixture fractionin liquid and solid fuels in reacting system remains a challenge forall techniques. The phase transition hampers the accuratemeasurement of the fuel in the gas phase and the presence ofdroplets and particles interferes with most laser diagnostics tech-niques [128]. Finally the presence of soot in the flame adds bothfurther complexity due to laser beam scatter and attenuation due tothe dense medium of soot particle and fluorescence and chem-iluminescence from the hot soot particles, as noted above.

Despite the aforementioned limitations and complexities, manyvaluable measurements of mixture fraction and temperature in“clean” gaseous turbulent non-premixed flames have been reportedin the last two decades [47,60,129]. Such measurements involve anelaborate experimental arrangement and extensive experimentalprocedures for calibration and to correct for the various interfer-ences. Figs. 7 and 8 are examples of such arrangements for two-dimensional and 1-D line configurations, respectively. These datahave contributed significantly to the understanding of the turbulentmixing characteristics (through mixture fraction) and their depen-dence on controlling parameters such as Reynolds number, stoi-chiometric mixture fraction and the type of fuel. Fig. 9 showsdetailed planar structure of the mixing field, OH and temperature,using the Raman, LIF and Rayleigh techniques respectively [48].These data were taken at different heights in a bluff body stabilisedturbulent non-premixed CH4/H2 flame and show the impact on thestructure of the flame downstream of the vortical structure shedfrom the edge of the bluff body.

The capacity to measure mass fractions of major species,temperature and minor species has formed the basis of an inter-national workshop on turbulent non-premixed flame, the TNFworkshop, which capitalises on the wealth of data available formodel validation and development [20]. Invariably, these majorspecies data have been collected using the Raman scatteringtechnique.

The Raman scattering technique is used to measure speciesconcentration in gaseous flows. This technique can probe a specificspecies but suffers from weak signal and cannot be used in anyenvironment where particles or droplets are present. The Ramansignal can also suffer from other interferences such as those fromhigh hydrocarbons and overlap between the Raman lines. Single-point [47], 1-D line [130] and 2-D planar imaging [48] measure-ments have been reported in the past. These data provided valuableinsight into the fate of the fuel in a variety of turbulent jet flamesand fuels and its impact on the reaction zone structure.

Of late, two comparatively similar systems have been developedat Sandia National Laboratories [130] and Darmstadt University[131] to overcome the limitations of the Raman technique. Thesesystems contain four combined Nd:YAG lasers to maximize thefluence in the probe volume and stretch the pulse over 40 ns.Another innovative feature is a spinning wheel systemwith a small

slit to gate the signal and minimise interference from the flame,while synchronising the exposure with the laser pulse. Thisapproach has helped improve the signal-to-noise substantially andreduce the noise associated with an electronic intensifier. Anincreasingly utilised extension involves themodelling of the Ramanspectra [132] to provide the temperature-dependent calibrationand a correction for the cross talk between the different Ramanresponses for the different species. These new systems haveunderpinned capacity to provide resolved measurements of scalardissipation [24] and offer potential to extend the Raman capabil-ities to flames with small amounts of soot or to larger flames thanthe tens of kW scale assessed to date.

The measurement of mixture fraction will remain a keyrequirement for future research to characterise non-premixedflames. The existing laser diagnostic techniques are still limited intheir capacity to deal with many of the issues associated withpractical flames, especially those that contain soot and/or otherparticles.

The method with the greatest promise for conserved scalarmeasurement in strongly radiating flows is the recently proposedmeasurement of krypton by two-photon fluorescence. As a noblegas, krypton is well suited to conserved scalar measurements, beingvery stable in a flame, although it has the disadvantage of beingexpensive. It is also optically accessible, albeitwith some challenges.The two-photon technique means that the collectionwavelength at760 nm is very different from the excitation wavelength of 215 nm,making it well suited to separation of the signal from laser-inducedinterference by filtering. The signal is subject to quenching, so aniterative approach combinedwith themeasurement of temperatureis required to determine mixture fraction. The method has recentlybeen validated against the established RamaneRayleighmethods ina non-premixed flame [127], but has yet to be assessed in flameswith soot. It is also well suited to planar imaging.

6. Measurement of soot volume fraction

The broadband incandescent radiation from soot, where it ispresent, dominates over the narrow-band energy from gaseousspecies, making its measurement of paramount importance. The LIItechnique is currently the most versatile technique for quantitative

Fig. 8. Experimental system for simultaneous line imaging of Raman scattering, Rayleigh scattering, and CO LIF [18].


measurement of fv. Laser energy is introduced at sufficient fluxes toheat the soot particles to temperatures far above the flametemperature. Hence these laser heated soot particles will emitradiation at shorter wavelengths than the flame, allowing itsseparation from the background [7,133]. One attraction of thistechnique is that the detected LII signal intensity, ILII, can closelyapproach linear proportionality to the fv if properly implemented.This enables spatial and temporal quantitative soot measurementswith the use of a high-energy pulsed laser coupled with suitablephoto detection equipment. The method is typically applied in the“plateau” regime, in which the response of the signal is approxi-mately independent of the laser intensity. This has importantadvantages for application in a turbulent environment since itmakes the measurement relatively insensitive to both the attenu-ation of the excitation beam by the presence of soot between thelaser and the measurement volume and to intensity variations dueto beam steering. However, the technique is not strictly a saturationmethod, so that the plateau regime is not entirely independent ofintensity. Qamar et al. [120] assessed the influence of the imperfectplateau by traversing a laminar flame through the light sheet andfound it to increase measurement uncertainty by w8%, but thisvalue is apparatus specific.

The detected ILII counts need to be calibrated, which may beachieved using either in-situ, or ex-situ methods such as laserextinction (LE) [134,135] or by gravimetric sampling [136]. The LEtechnique, preformed in well-controlled laminar pre-mixed flames(as shown in Fig. 10) using continuous-wave lasers, is used widelyto obtain the calibration factor. The value of the measured laserextinction, dimensionless soot extinction coefficient (Kext), isdirectly related to the soot volume fraction, which can be describedby Bouguer’s Law:

Kext ¼ �6p2

lIm

�m2 � 1m2 þ 2

�Nd3p; (5)

where N, dp, l and m are defined here as the number density,particle diameter, laser wavelength and the complex refractiveindex, respectively.

Equation (5) requires knowledge of Kext, the determination ofwhich has been a topic of considerable investigation that hasincorporated both in-situ extinction measurements and ex-situmicroscopic assessments of morphology from samples extractedfrom the flame [137e141]. Strictly, Equation (5) also relies on theassumption of negligible scattering, which is almost never true, butis reasonable in the Rayleigh regime [142] where the soot particlediameters are expected to be in the range of dp < 0.3l/p. This limitsthe range of soot diameters to <50 nm, <60 nm and <101 nm, forlaser wavelengths of 532 nm, 632 nm and 1064 nm, respectively.The size of primary soot particles is typically such that they fall intothe Rayleigh regime, but the influence of the aggregate shapebecomes increasingly significant with the size of the aggregate.Sorensen [143] reviewed the various approaches to calculateoptical properties of aggregates using fractal approaches. He foundthe RayleigheDebyeeGans polydisperse fractal aggregate (RDG/PFA) theory, which ignores internal multiple scattering, to accu-rately describe the angle-dependent light scattering from aggre-gated soot particles for many conditions. Smyth and Shaddix [142]also note that the degree to which scattering affects extinctionmeasurements is typically unknown and will likely vary withina flame as the soot field evolves. Despite the dependence of opticalproperties on morphology already noted, some recent measure-ments suggest that the optical absorptivity of soot is relativelyinsensitive to fuel composition and positions in a flame [140]. Thisis taken to be a reasonable justification for calibration of LII withethylene fuel in a laminar flat flame burner. Nevertheless, othermeasurements show a dependence on soot age [144]. Also therange of values reported for the extinction coefficient and thequestions of the impact of extractive sampling on soot propertiesmean that the evaluation of the refractive index of soot particlesremains a topic of ongoing research, especially for fuels and/orconditions that result in large soot aggregates. The presence of PAHinfluences the line of sight LE and may cause an overestimation ofthe soot volume fraction profiles by an order of magnitude, espe-cially at short wavelengths [145]. Bengtsson and Aldén [146]concluded that, at low burner heights in a laminar flame, the

Fig. 9. Instantaneous image triplets of mixture fraction, temperature, and OH mass fraction collected in flame BB2 (90% of blow-off) [48]. Three triplets are presented for each axiallocation with images covering an 8-mm region centered at 22, 34, 46, and 62 mm. The flow is from bottom to top. Overlaid on the mixture fraction image are contours marking thelean (x ¼ 0.01, yellow) and rich (x ¼ 0.1, white) reactive limits of H2/CH4 fuel mixture. (For interpretation of the references to colour in this figure legend, the reader is referred to theweb version of this article.)


presence of PAH emits strong fluorescence signals that increasewith C/O ratio. Zerbs et al. [145] report on the influence of wave-length on the measured extinction for calibration of LII. Theyrecommend calibration at long wavelengths (e.g. 1064 nm) for thelaser extinction measurements to reduce the laser absorption bymolecular species (e.g. PAH) in the flame.

LII has been used to obtain fv in three different flames toinvestigate the effect of global mixing [46]. More recently, Qamaret al. [120] reported experimental measurements of fv in pilotedturbulent diffusion flames. Henriksen et al. [147] have also reportedLII measurements in a highly sooty pool fire. They also reportedsimultaneous images of fv and OH. Lee et al. [148] extended themeasurements of soot and OH in turbulent ethylene jet flames toalso include PAH-LIF distribution and temperature from CARS.

An important area of interest to soot is the measurement of sootprecursors. The PAH and condensed species (CS) form readily underfuel-rich conditions and are deduced to play important roles in thesoot formation process [149,150]. Whilst fluorescence from PAH iscommonly regarded as an interference to the measurement ofother species, specially designed experiments to measure PAH areparticularly useful. Qualitative imaging of PAH may be readily

Fig. 10. A photo of the flat flame produced by a McKenna-style burner used for LIIcalibration.

performed using major species LIF systems at off-wavelengthconditions [148,151], though more detailed investigations arepossible with specific experiments [150,152]. These measurementsare especially important in the context of soot model validation.

The application of LII in turbulent flames requires carefulconsideration. These flames are generally sufficiently large for thetemperature gradients to cause significant steering, both of thebeam and the signal. In addition, the soot distributionwill influencethe propagation of both the laser and signal through diffraction.These influences lead to an unavoidable degradation in spatialresolution, typically in the range of 500% for a flame width of order0.2 m [153]. Zerbs et al. [145] note that beam steering, due to thecombined effects of turbulence, thermal gradients, pre-heat andpressure, implies that a spatial resolution below 1 mm in thedimension of the laser sheet thickness is not realistic under tech-nical conditions. In addition to the spatial resolution degrading, themagnitude of computed fv values will also be affected by theseissues. Another complexity of turbulent flames is that soot ispresent at different diameters and morphologies, associated withdifferent ages that can cause diameters to exceed the restricted sootdiameter values addressed in Equation (5). To minimise this effect,a longer wavelength for the laser extinction measurements isdesirable, which also reduces the laser absorption by molecularspecies [145]. Finally, the issue of how to address signal trapping isnot yet fully resolved for turbulent flames. Choi and Jensen [154]evaluated its effect in laminar flames, and found it be significantonly for volume fractions >5 ppm for their flames. They devised aniterative correction that is effective for higher volume fractions, butonly for laminar flames. Strictly, it also requires knowledge of therefractive index. More work on signal trapping is addressed forlarger particles in Section 8.

The vulnerability of LII excitation in the visible spectrum tointerference from PAH and other soot precursors is perhaps themain challenge to the novel and otherwise exciting RAYLIX method


developed by Geitlinger et al. [155] and Bockhorn et al. [156]. Theyemployed 532 nm excitation to perform simultaneous 2-D imagingof LII and Rayleigh/Mie-scattering, plus total extinction. Becausesoot volume fraction depends approximately on the cube of particlediameter, while the Rayleigh scattering depends on the square, thismethod offers the possibility for simultaneous measurement ofnumber density and size. Due to the absorption at 532 nm by PAH,it may be more accurate to apply RAYLIX at 1064 nm.

7. Measurements of droplet size, number density and velocity

Turbulent flows transporting reacting droplets are not only ofgreat practical significance in their own right, they represent anintermediate step in complexity between gaseous and particulatefuels. This is as true for the radiative heat transfer processes as theother process of combustion. Droplets have a number of propertiesthat can be exploited in their measurement within reacting envi-ronments. Firstly, being spherical allows the exploitation of a rangeof methodologies that require sphericity, such as phase Dopplerparticle anemometry, PDPA, and particle sizing from scattering[157]. PDPA is a single-point technique that will also provide, foreach droplet, all three components of velocity, as well as diameter.Awide range of single-point measurements have been employed ofturbulent reacting flows with particles, including the series of well-characterised flames by Masri and co-workers [128].

However, multiple droplets within the measurement volumeresult in spurious results, or measurement inaccuracy. This sets thecritical mass loading, above which the number of bad measure-ments exceeds a useful threshold and also results in a significantbias. By counting, the size distribution and mass loading of thedroplets can be obtained. Laser sheet drop-sizing, LSD, measuressimultaneously the scattered signal from one laser, which dependson d2p , and another on the fluorescence from another laser, whichdepends on d3p . Hence the ratio of these two signals can be used tomeasure diameter. While LSD has been used to provide measure-ment of mean diameter, Kalt et al. [158] have found that LSD is notuseful in providing detailed statistics. This is because the cubicdependence of the fluorescence signal on droplet diameter meansthat it is not possible to obtain good resolution of both large andsmall droplets simultaneously. That is, avoiding saturation from thelarge droplets results in too low a signal-to-noise ratio from thesmall droplets for reliable measurement. This method is alsolimited to isothermal environments, since the fluorescent specieswill also persist in the vapour phase, rendering the ratio irrelevant.

A range of liquid fuels is also suitable for fluorescencemeasurements. This has made acetone the fuel of choice for someinvestigations [128]. Alternatively, fluorescent materials such astoluene can be dissolved into other fuels for measurement oftemperature and/or mixture fraction, as described above [84]. Atthe same time, suchmeasurements cannot distinguish between thefluorescent species in the liquid and vapour phases [84]. Henceplanar measurements of droplet size and number density are stillnot realistic under reacting conditions.

8. Measurements of particle number density and size

Most solid fuels, both of fossil or biomass origin, are burned asa suspension flow, since the combustion intensity can be increasedas particle size is reduced. The size distribution of the milledpulverised coal is typically bi-modal, with the majority by massbeing in the range 10 mm<dp< 100 mmbut themajority by numberbeing of order 1 mm. The smaller of these particles interact withlight in the Mie regime, where the wavelength is the same order asthe light, while the larger ones scatter by nephelometry. Biomass istypically shredded, rather than pulverised, since the more fibrous

nature of the material makes it unrealistic to achieve the samedegree of comminution. Hence biomass is burned at larger sizesthan pulverised coal, typically in the millimetre range. Hence thesefuels require greater residence time and exhibit different aero-dynamic behaviour than pulverised coal and is one of the factorsthat limit the extent to which biomass can substitute coal inexisting processes. In addition, for both classes of fuel, nanometre-sized ash particles are formed within flames by a range of subli-mation, condensation and nucleation processes. Although ofa similar size to soot, these will typically form in cooler, post-reaction regions of the flame. The distribution and numberdensity of fuel particles as they disperse into a combustion chamberinfluences the location of the ignition plane, the distributions oftemperature andmixture fraction and hence also the radiation heattransfer and pollutant emissions [159].

The large numbers of particles, combined with the presence ofsoot they produce, make radiation from these flames even moredominant than in gas flames. Knowledge of these parameters isthus central to the optimisation of such systems.

In pulverised fuel (PF) combustion systems, the fuel is typicallycrushed to produce particles with a log-normal size distributionthat spans two to three orders of magnitude. The mean size istypically around 60 mm, the largest particles are typically around300 mm and the smallest particles are <1 mm. The smaller of theseparticles interact with light in Mie regime, where the wavelength isthe same order as the light, while the larger ones scatter bynephelometry. This size also allows them to be conveyed pneu-matically and also provides a high surface area-to-volume ratio forrapid combustion. Such particles are typically conveyed in the“dilute” phase, i.e. at velocities sufficient to prevent particles fromsettling to the floor of the duct, with mass loadings of less than10 kgparticles/kgair but more commonly at around 1 kgparticles/kgair[160]. Under these conditions the volume fraction of particles istypically about 1% and 0.1% respectively. However, the particlesdisperse as they move from the conveying system into, andthrough, the flame, resulting in significantly lower number densi-ties in much of the flame. Importantly, the cubic relationshipbetween particle numbers and mass causes the fine particles todominate particle numbers. For example, a typical coal of density1300 kg/m3 and amass loading ratio of 10would have 1013 particlesper m3, based on 10 mm diameter. However, in addition,nanometre-sized ash particles are formed within flames by a rangeof condensation and nucleation processes. Although of a similarsize to soot, these will typically form in cooler, post-reactionregions of the flame.

It is difficult to measure particle numbers accurately with suchawidesizedistribution.This isbecause intensityof thescatteredsignalfrom a laser scales approximately with d2p when the particle is muchlarger that thewavelength of light. This scaling causes scattered signalto be dominated by the large particles. The effects of optical attenua-tion are also significant in investigations of conditions of relevance topulverised fuel flames. These effects have typically limited previouslaser-scattering images from pulverised coal particles to providingqualitative insight, as in the 2 MW, pilot-scale combustion investiga-tions of Smith et al. [161]. The combined effects of attenuation andsignal trapping have also limited most previous investigations ofinstantaneous particle number densities to either single-pointmethods [162] or to planar investigations under dilute conditionswhere the effects of attenuation can be neglected [163].

Nevertheless, the attenuation in PF environments is typicallynot so great as to prevent sufficient signal from being collected toallow the effects of attenuation to be determined and corrected for[161]. Hence it is realistic to consider that reliable measurementsare possible under conditions of relevance to practical pulverisedfuel combustion.

Fig. 11. Schematic representation of the cylindrical (and potentially overlapping)shadows cast by spherical particles for the case of collimated light. Ray tracing can beused to correct for these attenuations [164,165].


One important step in the development of reliable planarmeasurements of particle distributions is the shot-by-shot correc-tion for in-plane attenuation. Kalt et al. [164] have developed a ray-tracing approach in non-reacting environments in which no beamsteering effects arise. Ignoring diffraction, BeereLambert absorp-tion accurately describes the attenuation of incident light byparticles along the history of the incoming ray. The integratedextinction history is derived for each pixel in the image, backtowards the illumination source. From any given pixel volume, thesignal scattered out the volume by particles can be expressed as:

4P ¼ CK�pr2P

�nPI

0: (6)

Here 4P is the detected signal from the particles, CK is a constant ofcorrection that can be applied for the entire optical arrangement,pr2P is the average particle cross-sectional area available to scattersignal, nP is the number of particles and I0 is the laser intensityentering the measurement volume. The appropriate constant ofcorrection is determined by calibration. The effective mean area ofthe scattering bodies in the pixel is known for mono-dispersespherical particles, or may often be measured for droplets andspheres using Phase Doppler Analysis. The incident light enteringthe pixel volume is calculated from the integrated extinction/transmittance history of earlier pixels back along the 1-D ray to thelight source, illustrated in Fig. 11, can then be corrected to givea measurement of particle concentration. It can also be applied inmodified form to divergent light sheets [165].

Diverging laser sheets cause some additional complications.Firstly, the sheet divergence causes both a drop in the local laserfluence [J/mm2] and an increase in the variation of fluence over theentire image plane, particularly when a large region is imaged or theimaging region is located near to the sheet forming optics. Thisincreases the required dynamic range of the detector, and typicallyrequiresa trade-off betweenacceptinga lower signal-to-noise ratio inregions of lower fluence and a larger amount of saturation in theregions of high fluence. Secondly, the particles cast conical shadowsrather than cylindrical shadows. The influence of particles on theextinction of the diverging sheet depends on the particle positionwith respect to the virtual origin of the laser sheet (Fig. 12). Theextrusion of square image volumes allows ray tracing through thelaser sheet. Fig.13 depicts themethodology of constructing a 1-D rayfor each pixel in the diverging sheet. The virtual origin (and lightsource) isdepictedas the crosshairsonthe left handsideof the image.Fromany pixel [i,j] a 1-D ray is constructed back to the light source byinterpolation throughknownpixels, computedonprevious iterationsthrough the image. By iterating from away from the light source, allpixels can be determined by 1-D analogue without recursion.

Importantly, Kalt et al. [164] also assessed the effect of signaltrapping, i.e. the attenuation of signal between the measurementvolume and the detection optics. They found that, using sufficientlylarge collection optics, a correction for signal trapping determinedfrom the ensemble and applied to the shot-to-shot images,provided an accuracy of single-shot images of typically 3%, or up to10% for the case where transmittance was only 50%, i.e. where halfof the energy was attenuated. However, to date, no method toaccount for non-spherical particles has been developed.

Furthermore, many biomass particles tend to be fibrous inshape, which makes their aerodynamic behaviour even morecomplex. The drag depends on orientation, aspect ratio and rota-tional velocity. Their motions are now becoming to be betterunderstood under settling [166,167], but are only beginning to beinvestigated in detail in non-reacting turbulent environments[168]. The measurement of fibres is also significantly more difficultthan of spheres, since the intensity of the scattered signal alsodepends on the orientation of the fibre within the light sheet. These

challenges have limited the conditions in which the most detaileddata-sets are available to relatively simple systems. Hence thechallenge to provide detailed measurements under conditions ofdirect relevance to turbulent combustion remains significant.

Another exciting recent development for the planar measure-ment of densely laden flows of particles is Structured Laser Illu-mination Planar Imaging (SLIPI) [169,170]. This method allowsremoval of the influence of secondary scattering as the laserpropagates through a particle-laden flow. It employs three lasersheets, each slightly separated in time and each with a sinusoidalspatial variation in the distribution in intensity, slightly offset fromeach other. The sinusoidal variation in spatial intensity, generatedby the use of a grating, is illustrated in Fig.14. Each image comprisesthe superposition of the primary signal, generated only by thedirect scattering of the laser sheet, with the secondary scattering,generated by multiple scattering events. The primary signal isidentical in each sheet, except for the variation in intensity, whilethe secondary scattering is random. The appropriate recombinationof the three images allows the random scattering to be removed, asillustrated in Fig. 15. It should be noted that, whilst this method hasthus far been demonstrated only in liquid droplets, it is equallyapplicable to spherical solid particles. The method is also expectedto have application for non-spherical solid particles, albeit witha degradation in performance. However, its performance with non-spherical particles is yet to be evaluated.

9. Measurement of velocity

It is well established that strain rate is a major controllingparameter for the formation of soot, while also influencingtemperature andmixture fraction. Particle image velocimetry (PIV),which is the leading method for planar measurement of velocity, istherefore the preferred technique. A number of reviews of themethod are available [171,172], and the technique is also well

Fig. 12. The dependence of the intensity at each pixel in a diverging light sheet can be determined from its position in the light sheet by tracing back to the virtual origin [164].


established for the measurement of velocity and of strain in turbu-lent flames [173]. However, laser-based velocity measurements inturbulent flames with soot suffer from large interferences. Whilemeasurements in such environments have been reported (e.g.[174],), the effect of such interferences has not been assesseddirectly. Given the focus of the present review on measurements instrongly radiating flames, the present section is limited to addressthe issue of ensuring that the soot (which is not a flow tracer, beingboth produced and consumed) does not bias measurements by PIVor LDA. Likewise, the simultaneous measurements of velocity andsoot volume fraction have been hindered by the need to demon-strate that the PIV seed particles do not interfere with the LIImeasurement.

To confirm that reliable optical discrimination between the twomeasurements is possible, the authors have performed a prelimi-nary investigation in a wrinkled, pseudo-laminar, low Reynoldsnumber (Re¼ 2200) ethylene diffusion flame issuing from a 10mmnozzle at 1 m/s, illustrated in Fig. 16 a. Both the fuel and air wereseeded with 0.5 mm aluminium oxide particles at a seeding densitysuitable for high quality PIV. Illumination was provided at 532 nmwith a Quantel Twins B PIV double pulsed Nd:YAG laser, operated at300 mJ/pulse, and detection was with a Kodak ES 1.0 Megapluscamera with 1018 � 1008 pixel array with 10-bit resolution. Theimages were collected with a camera gate of 0.25 ms for the firstimage and w200 ms for the second image. The LII was undertakenusing a Nd:YAG laser frequency doubled to 532 nm, collimated toprovide a parallel sheet of width 14 mm, thickness w0.5 mm anda high fluence ofw1.8 J/cm2 to provide a conservative assessment ofthe potential for interference. For both techniques, the in-planeresolution was w120 mm/pixel. Detection was undertaken with

Fig. 13. The simplified 1-D model of collimated attenuation used to correct fordivergence by ray tracing from a planar image [164].

a 12-bit intensified CCD (ICCD) camera. A 430 nm interference filterwas found tobe sufficient to suppress any interference fromthe seedparticles on the LII image to below detectable limits. This isdemonstrated in Fig.16b,where signal from the LII detection systemis found only in the thin sheet of the flame, consistent withnumerous other investigations employing LII alone (e.g. [175]). Anyinterferencewill be further reduced with the use of 1064 nm for theLII excitation. The peak, local instantaneous soot volume fraction isestimated here to be about 4 ppm from our ownmeasurements andfromthese earlier investigations [175]. This verifies that it is possibleto undertake PIV in a wide range of flames of interest that containsoot using methods that will not interfere with LII measurements.

These investigations also found that the use of a linear polarisingfilter was sufficient to suppress the scattering from the soot tobackground levels, toavoiddetectable interferenceof sooton thePIVmeasurement. An interference filter at 532 nm on the PIV cameraexcludes any luminosity fromtheflamebrush fromappearingon theMie-signal, particularly when the second frame of the PIV correla-tion pair is very long. It should be noted that the Mie-scatteringefficiency of the aluminium oxide particles used in these prelimi-nary investigations is reduced by a factor of approximately 8 at thetemperatures in the flame front relative to that at ambient temper-ature. This results in a low signal intensity of the Mie-scattering in

Fig. 14. The planar nephelometry image of a spray generated by one (of the three)laser sheet with a sinusoidal distribution of laser intensity, in turn imposed by the useof an optical grating [170].

Fig. 15. The difference between a raw planar image generated by a laser sheet througha spray (left) with that obtained using the SLIPI method to remove secondary scat-tering [170].


the combustionproducts, as shown in Fig.16c,which corresponds tothe second image in the pair, with the longest gate and hencepoorest discrimination. Nevertheless, despite the drawbacks ofaluminium oxide, Fig. 16d demonstrates that it provides sufficientMie-scattering signal to conduct PIV under laminar conditions.Discrimination could be significantly enhanced by replacing thealuminium oxide with particles of higher scattering efficiency atelevated temperature, such as magnesium oxide, and/or by the useof a camera with a shorter gate. Hence simultaneous PIV and LIImeasurements are possible for turbulent flames with soot.

Fig. 16. Demonstration of conditions that avoid mutual interference during simulta-neous LII and PIV in a laminar diffusion flame: a) Illustration of flame condition; b) LIImeasurement; c) PIV measurement; d) Soot volume fraction and velocity. Each imagemeasures 50 � 30 mm.

10. Measurement of radiation and its propagation througha scatteringeabsorbing medium

The direct measurement of the total radiant transfer froma flame has long been a research objective, since it is the primaryperformance characteristic of many combustion systems. In addi-tion, the radiation from a flame is directly coupled to anotherprimary performance characteristic, the thermal NOx emissionsfrom a flame, through the influence of radiation on flame temper-ature [19,32,176]. Hence many measurements of radiation havebeen performed in lab-scale open flame investigations [42,44,177],pilot-scale burner development programs [32,35,176,178], andinvestigations of fires [30]. However, although these techniques areuseful for providing mean measurements, they lack the capacity toprovide well-resolved measurements spatially and temporally. Thegrowing advancement in radiation modelling techniques bringswith it the need for detailed in-flame measurements to becombined with measurements of radiation heat flux from a flame,as noted in the recent review by Coelho [6]. Hence, while the focusof the present paper is to review detailed, laser-based measure-ment techniques, this link behoves some comment on the tech-niques used to measure total flame radiation.

The natural radiation from a flame contrasts that employed inthe laser-based measurements described above in several ways.Natural radiation is broadband, incoherent, originates frommultiple sources and the measurement flux depends on both thetemperature and emissivity of the detector. For this reason,detectors should account for the receiving angle, emissivity andtemperature of the detector. The temperature of the receiver istypically controlled by water cooling and measured. The depen-dence of the measurement on receiver angle and emissivity can bereduced by the use of detectors with hollow ellipsoidal total radi-ation pyrometers, with cavity receivers to provide multiple internalreflections [179]. However, this does not eliminate the need forcalibration. More common is the use of compact and commerciallyavailable calibrated thermopiles (e.g. [180]), shielded by a window(typically sapphire), which is transparent to most of the infra-redand visible spectrum. These have been employed to investigatelab-scale, open flames (e.g. [44,42]) and also the flames in

pilot-scale furnaces (e.g. [178]). Nevertheless, the use of a windowintroduces dependence on acceptance angle and on its spectralcharacteristics. Likewise, the spectral characteristics of the receiverretain an influence. Hence, the development of improved model-ling capability is bringing with it the need for further refinement inradiation detector measurements.

A method to calculate the total heat flux from a measured meanaxial profile has been established by Sivathanu and Gore [181], andthe shape of the profile is also of direct importance. The mean heatflux profile can also be measured from the heat extracted throughcooling tubes or a segmented water-cooled shell. However, theselatter methods measure both the radiant and convective compo-nents of heat transfer and require knowledge of the convective heattransfer coefficient and flow field to isolate radiation.

The challenge of performing and interpreting a reliable heat fluxmeasurement increases both with the physical scale of a flame and


with the degree of confinement. For a confined flame, the radiationcomes both from the flame itself and from the walls. Together theyform a coupled system since the temperature of the flame, andhence its radiation flux (as well as NOx emissions), depends on theheat losses to the walls, i.e. on wall temperature distribution andemissivity, so that both should be reported. With a measurement oftotal radiation combined with knowledge of the wall temperaturedistribution and emissivity, it is potentially possible to estimate theseparate contributions from the wall and flame. However, in prac-tice this is not easy to do reliably, since the wall temperature israrely uniform, the wall surface may become covered by ash orsubjected to oxidising or reducing conditions, which can changethe surface emissivity, because the radiation calculations arecomplicated by multiple internal reflections and because the flameis not truly optically thin, especially for strongly radiating flames.Hence it is desirable to isolate the radiation from the wall and theflame. One method to remove the background radiation fromthe wall is to insert a small cold target into the wall and collect theradiation through a narrow-angle aperture, with the detectorplaced diametrically opposite to, and aligned with, the cold target.This approach has the advantage of totally removing the influenceof the background, but the disadvantage that it only collects theradiation from a small part of the flame. Also the measurement issensitive to the acceptance angle of the detector and to goodalignment with the target, which in turn is subject to beam steeringinfluences. The other option is to utilise the different shaped energyspectrum of the radiation from the wall and flame to separate theirrelative contributions. This can be done, for example by assumingthat both spectra follow a black-body distribution, provided thatthe flame temperature is significantly greater than the walltemperature. Nevertheless, it is evident that there is also need forfurther development of the methods to separate the radiation fromthe flame and the wall.

The method developed to correct for the attenuation of laserlight described in Section 8 can also be extended to provide directplanar measurement of the three components of radiation propa-gation through a scatteringeabsorbing medium, namely the scat-tered, absorbed and transmitted components, on a shot-to-shotbasis. Previous measurements have only been performed underconditions where absorption and attenuation are negligible. Incontrast, no measurements have previously been reported underparticle loadings of relevance to reactors and combustors whereattenuation and absorption are parameters of primary interest, as isrequired for model development and validation. The attenuationcorrections can be used to quantify particle concentration if theappropriate scaling constant, CK, can be found. This scaling constant,which represents the detector sensitivity to scattered radiation can

Fig. 17. The instantaneous planar measurement of all components of radiation propagation tlight sheet [182]: a) Scattered Signal; b) Divergence correction; c) Transmitted fraction; d)

be determined for a given optical arrangementd either in advanceby testing against a known concentration of known-diameterparticles, or by measuring the intensity of the transmitted lasersheet and iteratively determining a CK value that yields the sameoverall transmission. The quantitative processing of laser nephe-lometry data will then provide laser sheet fluence, particleconcentration and corrections for local laser intensity due toabsorption of the laser sheet and divergence of the beam. Nephe-lometry is the generalised term used to describe the measurementof particle concentration inferred from scattered light. Strictly, Mie-scattering applies for the regime where the particle size isapproximately the same as the wavelength of light, i.e. of micron-sized particles for visible light. Nephelometry extends also toparticles of larger diameter where the scattering is outside the trueMie regime (generalised Lorentz/Mie-scattering), so is the relevantrange for particles of tens of micron in diameter.

Fig. 17 [182] shows how the scattered and transmitted compo-nents of radiation can be used to determine the component absor-bed by the particulate medium. The diffusely scattered component,which depends on the surface properties and diameter, is typicallydominant (around96%)while theabsorbed component is verymuchweaker. For sufficiently largeparticles the scattered light is no longercoherent, becomes depolarised, and contributes to the multiplescattering signal and background radiation levels. The influence ofthe absorbed radiation depends on its composition and environ-mentandcan result in an increaseof its temperature and/ora changein phase or composition. However, the wavelength of the emittedradiation from a particle is typically broadband and temperaturedependant, and can be separated from the narrow-band scatteredradiation by use of a sufficiently long wavelength laser.

It is a relatively easy proposition to measure the laser powershot-to-shot, as the data is collected. If the total power and lasersheet energy profile are known then the scattered and absorbedradiative components can be quantified on a shot-to-shot basis, asillustrated in Fig. 17.

11. Conclusions

A number of complementary planar laser diagnostic techniqueshave recently emerged that together offer the potential fora breakthrough in capability to investigate strongly radiatingturbulent reacting flows. They could, in principle, be appliedsimultaneously, although this is yet to be done. Together thesemethods, summarised in turn below, offer the potential to enabledetailed measurement of the coupled processes of turbulence,combustion and radiation, which has not previously been possiblein strongly radiating flows.

hrough a turbulent scattering, absorbing medium of spherical particles with a divergingParticle concentration; e) Radiation absorbed by particles.


For in-situ laser-based thermometry, perhaps the best-estab-lished method is Rayleigh scattering, whose simplicity continues tomake it the method of choice for clean environments. However itsvulnerability to interference from scattering makes it poorly suitedto investigate strongly radiating flows, such as those involving soot.Of the established techniques, the Coherent anti-Stokes RamanSpectroscopy (CARS), offers the greatest capability in suchenvironments. The use of pico-second lasers has recently beenshown to enable good separation between signal and noise in thepresence of soot. Nevertheless, CARS has the disadvantages ofa relatively large probe volume (i.e. moderate spatial resolution),a vulnerability to differential beam steering and to being poorlysuited to planar measurement. The thermometry technique withgreatest promise for planar and well-resolved measurement oftemperature in the presence of soot and other fine particles isnon-linear excitation regime two-line atomic fluorescence (NTLAF).This method has recently been demonstrated to be suitable forsimultaneous application with laser-induced incandescence (LII) toprovide joint measurement of both temperature and soot volumefraction. While, to date, it has only been demonstrated underconditions of moderate soot loading, it is anticipated that correc-tions for interference will allow the range to be extended to highsoot loadings also. Similarly the presently accessible temperaturerange is limited to 800K < T < 2800 K, and that of mixture fractiongreater than 0.9. These ranges are of primary immediate interestbecause they span the high temperature reaction zone, and withfurther development, it is anticipated that the operating range willbe extended.

For the measurement of mixture fraction in turbulent flames,RamaneRayleigh techniques have proved to bewell suited for cleanflames. However, these methods are highly vulnerable to interfer-ence from the presence of soot (or other particles) and its precur-sors. Such measurements have presently also been limited tolaboratory-scale flames, although incremental refinement offerspotential to extend them to larger physical scale. Tracer laser-induced fluorescence (LIF) is presently the only realistic techniqueby which to measure mixture fraction in the presence of particles,and is also well suited to planar measurements. Until recently, thetracers identified for measurement are themselves combustible,limiting their application to the relatively low temperature regionon the fuel-rich side of the reaction zone. The use of krypton asa tracer gas offers potential to overcome these disadvantages. Beinginert, this tracer survives the reaction zone. It is also accessible witha two-photon technique, which allows the signal to be separatedfrom spurious scattering. Its recent emergence means that thismethod is yet to be demonstrated for turbulent flames with soot.Nevertheless, it has potential to allow the planar measurement ofmixture fraction in strongly radiating turbulent flames.

Laser-induced incandescence has advanced to the stagewhere ithas become the method of choice for the measurement of thevolume fraction of soot and other nano-sized particles. It is wellsuited to planarmeasurements andhas also beenperformed at largephysical scale using water-cooled collection optics coupled to theICCD with a fibre-optical probe. Nevertheless, the technique is stillunder development owing to the complex nature of soot and theenvironment under which it is produced and consumed. Forexample, calibration remains a challenge in turbulent flames espe-cially in those environments where some of the soot grows to sizesthat take it outside of the Rayleigh limit. These issues can be mini-misedby theuseof longerwavelength radiation. Theuseof 1064nm,corresponding to the fundamental of a Nd:YAG laser, is thereforewell suited both to minimising such losses and also avoiding inter-ference frompolycyclic aromatic hydrocarbons (PAH) andother sootprecursors. Also, even in moderate scale turbulent flames, say of200 mm diameter, the presence of soot, combined with other beam

steering issues in turbulent flames, causes broadening of the lasersheet, and hence a loss in spatial resolution of order 500%.

A series of recent measurements have confirmed that LII can beperformed simultaneously both with temperature (by NTLAF) andwith the seed particles required for velocity measurements in thepresence of soot, at least under laminar conditions. This providesconfidence that simultaneous measurement of soot volume frac-tion, temperature and velocity is possible, although they are yet tobe demonstrated all together in a turbulent environment.

Measurement in the presence of soot or other particles alsoimposes the additional challenges of optical diffraction, attenuationand signal trapping. A number of approaches are being devised toincrementally advance capability to correct for these effects.However, such corrections retain some constraints and are pres-ently limited to spherical particles. Also, the simultaneous planarmeasurement of particle size and particle number density is pres-ently not realistic under reacting conditions, even for liquid fuels.Hence substantial development is required before it is possible toprovide detailedmeasurement of reacting flows under awide rangeof conditions of practical relevance.

The simultaneous measurement of the scattered, absorbed andtransmittedcomponentsof radiationhas recentlybeendemonstratedfor non-reacting turbulent flows. This represents an importantadvance in capacity to develop detailed understanding of radiationpropagation in more realistic conditions. This method could, inprinciple, be extended to reacting flows, but further work is requiredto account for non-spherical particle sizes and for non-uniformparticle size distribution. It could also be combined with theStructured Laser Illumination Planar Imaging (SLIPI) method, whichallows secondary scattering effects to be greatly reduced. Not only dothesemethods offer the advantage of spatially correlated informationthat is desirable for all turbulent systems, they also enable instanta-neous ray-tracing, which is particularly valuable for radiationmodelling. Such data, while not presently available in strongly radi-ating and turbulent environments owing to the challenges whichparticles impose on the measurements of key parameters such astemperature and mixture fraction, are now becoming possible.

These recent advances inmeasurement capability suggest that itis now possible to achieve significant advances in the under-standing and modelling capability of strongly radiating turbulentreacting flows. Such data are a prerequisite for the progress recentlyachieved in soot-free turbulent flames. Nevertheless, significantchallenges remain to be overcome before such measurements arepossible under many of the other conditions of practical signifi-cance, including non-spherical particles, polydisperse particle sizedistributions and/or larger physical scale of the combustors.

Acknowledgements

This research has been supported by the Australian ResearchCouncil (ARC) and by the Centre for Energy Technology (CET). Weare also grateful to the staff and postgraduate students of theCentre, whose work has contributed to the understanding reportedherein. Finally, we gratefully acknowledge the many detailed andconstructive comments by the anonymous reviewers of this paper,and the comments on an earlier version of the document by Dr A.Klimenko. Addressing their comments has strengthened thedocument considerably.

References

[1] Sarofim AF. Radiative heat transfer in combustion: friend or foe. ProcCombust Inst 1986;21:1e23.

[2] Lindstedt RP, Louloudi SA. Joint-scalar transported PDF modeling of sootformation and oxidation. Proc Combust Inst 2005;30:775e83.


[3] Watanabe H, Kurose R, Komori S, Pitsch H. Effects of radiation on spray flamecharacteristics and soot formation. Combust Flame 2008;152:2e13.

[4] Mehta RS, Haworth DC, Modest MF. Composition PDF/photon Monte Carlomodelling of moderately sooting turbulent jet flames. Combust Flame 2010;157:982e94.

[5] Song TH, Viskanta R. Interaction of radiation with turbulence: application toa combustion system. J Thermophys Heat Tr 1987;1:56e62.

[6] Coelho PJ. Numerical simulation of the interaction between turbulence andradiation in reactive flows. Prog Energy Combust Sci 2007;33:311e83.

[7] Kohse-Höinghaus K, Jeffries JB. Applied combustion diagnostics. London, UK:Taylor & Francis; 2002.

[8] Viskanta R, Mengüc MP. Radiation heat transfer in combustion systems. ProgEnergy Combust Sci 1987;13:37e160.

[9] Sims R, Schock R, Adegbululgbe A, Fenhann J, Konstantinaviciute I,Moomaw W, et al. In: Metz B, Davidson OR, Bosch PR, Dave R, Meyer LA,editors. Energy supply in climate change 2007: mitigation contribution ofworking group III fourth assessment report of the IPCC. Cambridge, UK:Cambridge University Press; 2007.

[10] Demibras A. Combustion characteristics of different biomass fuels. ProgEnergy Combust Sci 2004;30:219e30.

[11] Z’Graggen A, Haueter P, Trommer D, Romero M, de Jesus JC, Steinfeld A.Hydrogen production by steam-gasification of petroleum coke using concen-trated solar power e II. Reactor design, testing, and modeling. Int J HydrogenEnergy 2006;31:797e811.

[12] Z’Graggen A, Steinfeld A. Hydrogen production by steam-gasification ofcarbonaceousmaterialsusingconcentratedsolar energyeV.Reactormodeling,optimization, and scale-up. Int J Hydrogen Energy 2008;33:5484e92.

[13] Hawkes ER, Sankaran R, Sutherland JC, Chen JH. Scalar mixing in directnumerical simulations of temporally evolving plane jet flames with skeletalCO/H2 kinetics. Proc Combust Inst 2007;31:1633e40.

[14] Lignell DO, Chen JH, Smith PJ. Three-dimensional direct numerical simula-tion of soot formation and transport in a temporally-evolving, nonpremixedethylene jet flame. Combust Flame 2008;155:316e33.

[15] Lu T, Law CK. Toward accommodating realistic fuel chemistry in large-scalecomputations. Prog Energy Combust Sci 2009;35:192e215.

[16] Battin-Leclerc F. Detailed chemical kinetic models for the low-temperaturecombustion of hydrocarbons with application to gasoline and diesel fuelsurrogates. Prog Energy Combust Sci 2008;34:440e98.

[17] Dooley S, Won SH, Chaos M, Heyne J, Ju Y, Dryer FL, et al. A jet fuel surrogateformulated by real fuel properties. Combust Flame 2010;157:2333e9.

[18] Barlow RS. Laser diagnostics and their interplay with computations tounderstand turbulent combustion. Proc Combust Inst 2007;31:49e75.

[19] Nathan GJ, Mi J, Alwahabi ZT, Newbold JR, Nobes DS. Impacts of a jet’s exitflow pattern on mixing and combustion performance. Prog Energy CombustSci 2006;32:496e538.

[20] Turbulent Nonpremixed Flames Workshop. Available at: http://public.ca.sandia.gov/TNF/abstract.html; 2010.

[21] Mi J, Nathan GJ. The influence of probe resolution on the measurement ofa passive scalar and its derivatives. Exp Fluids 2003;34:687e96.

[22] Bilger RW. Some aspects of scalar dissipation. Flow Turb Comb 2004;72:93e114.

[23] Pitts WM, Richards CD, Levenson MS. Large and small-scale structures andtheir interactions in an axisymmetric jet. Report No. NISTIR 6393. NationalInstitute of Standards and Technology, United States Department ofCommerce Technology Administration; 1999.

[24] Wang GH, Barlow RS, Clemens NT. Quantification of resolution and noiseeffects on thermal dissipation measurements in turbulent non-premixed jetflames. Proc Combust Inst 2007;31:1525e32.

[25] Zimberg MJ, Frankel SH, Gore JP, Sivathanu YR. A study of coupled turbulentmixing, soot chemistry, and radiation effects using the linear eddy model.Combust Flame 1998;113:454e69.

[26] Xin Y, Gore JP, McGrattan KB, Rehm RG, Baum HR. Fire dynamics simulationof a turbulent buoyant flame using a mixture-fraction-based combustionmodel. Combust Flame 2005;141:329e35.

[27] Leung KM, Lindstedt RP, Jones WP. A simplified reaction mechanism for sootformation in non-premixed flame. Combust Flame 1991;87:289e305.

[28] Magnussen BF, Hjertager BH. On mathematical modeling of turbulentcombustion with special emphasis on soot formation and combustion. ProcCombust Inst 1977;16:719e29.

[29] Lee MP, McMillin BK, Hanson RK. Temperature measurements in gases byuse of planar laser-induced fluorescence imaging of NO. Appl Opt 1993;32:5379e96.

[30] Kuhr C, Staus S, Schönbucher A. Modelling of the thermal radiation of poolfires. Prog Comp Fluid Dyn 2003;3:151e6.

[31] Spalding DB. Colloquium on modelling principals: the art of partial model-ling. Proc Combust Inst 1963;9:833e43.

[32] Weber R. Scaling characteristics of aerodynamics, heat transfer, andpollutant emissions in industrial flames. Proc Combust Inst 1996;26:3343e54.

[33] Sayre A, Lallemant N, Dugué J, Weber R. Effect of radiation on Nitrogen Oxideemissions from nonsooty swirling flames of natural gas. Proc Combust Inst1994;25:235e42.

[34] Smart JP, Weber R. Reduction of NOx and optimisation of burnout with anaerodynamically airstaged burner and an air staged pre-combustor burner.J Inst Energy 1989;62:237e45.

[35] Parham JJ, Nathan GJ, Smart JP, Hill SJ, Jenkins BG. The relationship betweenheat flux and NOx emissions in gas fired rotary kilns. J Inst Energy 2000;73:25e34.

[36] Weber R, Orsino S, Verlaan AL, Lallemant N. Combustion of light and heavyfuel oils in high-temperature air. J Inst Energy 2001;74:38e47.

[37] Murphy JJ, Shaddix CR. Soot property measurements in a two-meter diam-eter JP-8 pool fire. Combust Sci Technol 2006;178:865e94.

[38] Kearney SP, Frederickson K, Grasser TW. Dual-pump coherent anti-stokesRaman scattering thermometry in a sooting turbulent pool fire. ProcCombust Inst 2009;32:871e8.

[39] Frederickson K, Kearney SP, Grasser TW. Laser-induced incandescencemeasurements of soot in turbulent pool fires. Appl Opt 2011;50:A49e59.

[40] Antonia RA, Satyaprakash BR, Hussain AKMF. Measurements of dissipationrate and some other characteristics of turbulent plane and circular jets. PhysFluids 1980;23:695e700.

[41] Kalghatgi GT. Blow-out stability of gaseous jet flames. Part I: in still air. CombSci Tech 1981;26:233e9.

[42] Langman AS, Nathan GJ, Mi J, Ashman PJ. The influence of geometric nozzleprofile on the global properties of a turbulent diffusion flame. Proc CombustInst 2007;31:1599e607.

[43] Kent JH, Bastin SJ. Parametric effects on sooting in turbulent acetylenediffusion flames. Combust Flame 1984;56:29e42.

[44] Turns SR, Myhr FH. Oxides of nitrogen emissions from turbulent jetflames: part I e fuel effects and flame radiation. Combust Flame 1991;87:319e35.

[45] Decroix ME, Roberts WL. Transient flow field effects on soot volume fractionin diffusion flames. Comb Sci Tech 2000;160:165e89.

[46] Qamar NH, Nathan GJ, Alwahabi ZT, King KD. The effect of global mixing onsoot volume fraction: measurements in simple jet, processing jet, and bluffbody flames. Proc Combust Inst 2005;30:1493e500.

[47] Masri AR, Dibble RW, Barlow RS. The structure of turbulent nonpremixedflames revealed by RamaneRayleigh-LIF measurements. Prog EnergyCombust Sci 1996;22:307e62.

[48] Dally BB, Masri AR, Barlow RS, Fiechtner GJ. Instantaneous and meancompositional structure of bluff-body stabilised nonpremixed flames.Combust Flame 1998;114:119e48.

[49] Dugué J, Weber R, Horsman H. The effect of combustion on swirling flowsexpanding in a medium confinement furnace. IFRF Document No F 59/y/7.International Flame Research Foundation; 1990.

[50] Dugué J, Ereaut PR. Laser sheet visualisation in cold flows, gas and coalflames. IFRF Document No. K 70/y/241e12. International Flame ResearchFoundation; 1990.

[51] Frederickson K, Kearney SP, Grassner TW, Castaneda JN. Joint temperatureand soot-volume-fraction measurements in turbulent meter-scale pool fires.In: 47th AIAA Aerospace Sciences Meeting. AIAA 2009-1264. Orlando, USA;2009.

[52] Eckbreth AC. Laser diagnostics for combustion temperature and species.Amsterdam: Gordon and Breach Publishers; 1996.

[53] Hoffman D, Leipertz A. Temperature field measurements in a sooting flameby Filtered Rayleigh Scattering (FRS). Proc Combust Inst 1996;26:945e50.

[54] Miles RB, Forkey JN, Lempert WR. Filtered Rayleigh Scattering measurementsin supersonic/hypersonic facilities. In: AIAA 17th Aerospace Ground TestingConference, AIAA-92-3894. Nashville, USA: AIAA; 1992.

[55] Yalin AP, Miles RB. Ultraviolet Filtered Rayleigh Scattering temperaturemeasurements with a mercury filter. Opt Lett 1999;24:590e2.

[56] Hoffman D, Münch KU, Leipertz A. Two-dimensional temperature determi-nation in sooting flames by filtered Rayleigh scattering. Opt Lett 1996;21:525e7.

[57] Elliott GS, Glumac N, Carter CD. Molecular Filtered Rayleigh Scatteringapplied to combustion. Meas Sci Technol 2001;12:452e66.

[58] Kearney SP, Schefer RW, Beresh SJ, Grasser TW. Temperature imaging innonpremixed flames by joint filtered Rayleigh and Raman scattering. ApplOpt 2005;44:1548e58.

[59] Zhao F-Q, Hiroyasu H. The applications of laser Rayleigh scattering tocombustion diagnostics. Prog Energy Combust Sci 1993;19:447e85.

[60] Meier W, Weigand P, Duan XR, Giezendanner-Thoben R. Detailed charac-terization of the dynamics of thermoacoustic pulsations in a lean premixedswirl flame. Combust Flame 2007;150:2e26.

[61] Afzelius M, Bengtsson PE, Bood J, Brackmann C, Kurtz A. Development ofmultipoint vibrational coherent anti-Stokes Raman spectroscopy for flameapplications. Appl Opt 2006;45:1177e86.

[62] Stricker W, Lückerath R, Meier U, Meier W. Temperature measurements incombustion e not only with CARS: a look back at one aspect of the EuropeanCARS Workshop. J Raman Spectrosc 2003;34:922e31.

[63] Roy S, Meyer TR, Lucht RP, Belovich VM, Corporan E, Gord JR. Tempera-ture and CO2 concentration measurements in the exhaust stream ofa liquid-fuelled combustor using dual-pump coherent anti-Stokes Ramanscattering (CARS) spectroscopy. Combust Flame 2004;138:273e84.

[64] Roy S, Kinnius PJ, Lucht RP, Gord JR. Temperature measurements in reactingflows by time-resolved femtosecond coherent anti-Stokes Raman scattering(fs-CARS) spectroscopy. Opt Commun 2008;281:319e25.

[65] Weikl MC, Seeger T, Wendler M, Sommer R, Beyrau F, Leipertz A. Validationexperiments for spatially resolved one-dimensional emission spectroscopytemperature measurements by dual-pump CARS in a sooting flame. ProcCombust Inst 2009;32:745e52.

http://public.ca.sandia.gov/TNF/abstract.html

http://public.ca.sandia.gov/TNF/abstract.html


[66] Malarski A, Beyrau F, Leipertz A. Interference effects of C2-radicals inNitrogen vibrational CARS thermometry using a frequency-doubled Nd:YAGlaser. J Raman Spectrosc 2005;36:102e8.

[67] Geigle KP, Schneider-Kühnle Y, Tsurikov MS, Hadef R, Lückerath R, Krüger R,et al. Investigation of laminar pressurised flames for soot model validationusing SV-CARS and LII. Proc Combust Inst 2005;30:1645e53.

[68] Kliewer CJ, Gao Y, Seeger T, Kiefer J, Patterson BD, Setterson TB. Picosecondtime-resolved pure-rotational coherent anti-Stokes Raman spectroscopy insooting flames. Proc Combust Inst 2011;33:831e8.

[69] Roy S, Gord JR, Patnaik AK. Recent advances in coherent anti-Stokes Ramanscattering spectroscopy: fundamental developments and applications inreacting flows. Prog Energy Combust Sci 2010;36:280e306.

[70] Hartlieb AT, Atakan B, Kohse-Höinghaus K. Temperature measurement infuel-rich non-sooting low-pressure hydrocarbon flames. Appl Phys B 2000;70:435e45.

[71] Seitzman JM, Hanson RK, DeBarber PA, Hess CF. Application of quantitativetwo-line OH planar laser-induced fluorescence for temporally resolvedplanar thermometry in reacting flows. Appl Opt 1994;33:4000e12.

[72] Cattolica R. OH rotational temperature from two-line laser-excited fluores-cence. Appl Opt 1981;20:1156e66.

[73] Giezendanner-Thoben R, Meier U, Meier W, Aigner M. Phase-lockedtemperature measurements by two-line OH PLIF thermometry of a self-excited combustion instability in a gas turbine model combustor. FlowTurbul Combust 2005;75:317e33.

[74] Qin W, Chen YL, Lewis JWL. Time-resolved temperature images of laser-ignition using OH two-line laser-induced fluorescence (LIF) thermometry.IFRF Combust J; 2005. Tech. Rep. Article Number 200508.

[75] Nygren J, Engström J, Walewski J, Kaminski CF, Aldén M. Applications andevaluation of two-line atomic LIF thermometry in sooting combustionenvironments. Meas Sci Technol 2001;12:1294e303.

[76] Meier UE, Wolff-Gabmann D, Stricker W. LIF imaging and 2D temperaturemapping in a model combustor at elevated pressure. Aerosp Sci Technol2000;4:403e14.

[77] Haudiquert M, Cessou A, Stepowski D, Coppalle A. OH and soot measure-ments in a high-temperature laminar diffusion flame. Combust Flame 1997;111:338e49.

[78] Tamura M, Luque J, Harrington JE, Berg PA, Smith GP, Jeffries JB, et al. Laser-induced fluorescence of seeded nitric oxide as a flame thermometer. ApplPhys B 1998;66:503e10.

[79] Bessler WG, Hildenbrand F, Schulz C. Two-line laser-induced fluorescenceimaging of vibrational temperatures of seededNO. Appl Opt 2001;40:748e56.

[80] Kronemayer H, Ifeacho P, Hecht C, Dreier T, Wiggers H, Schulz C. Gas-temperature imaging in a low-pressure flame reactor for nano-particlesynthesis with multi-line NOeLIF thermometry. Appl Phys B 2007;88:373e7.

[81] Lee T, Jeffries JB, Hanson RK. Experimental evaluation of strategies forquantitative laser-induced-fluorescence imaging of nitric oxide in high-pressure flames (1e60 bar). Proc Combust Inst 2007;31:757e64.

[82] Atakan B, Hartlieb AT. Laser diagnostics of NO reburning in fuel-rich propeneflames. Appl Phys B 2000;71:697e702.

[83] McMillin BK, Palmer JL, Hanson RK. Temporally resolved, two-line fluores-cence imaging of NO temperature in a transverse jet in a supersonic crossflow. Appl Opt 1993;32:7532e45.

[84] Schulz C, Sick V. Tracer-LIF diagnostics: quantitative measurement of fuelconcentration, temperature and fuel/air ratio in practical combustionsystems. Prog Energy Combust Sci 2005;31:75e121.

[85] Hanson RK, Seitzman JM, Paul PH. Planar laser-fluorescence imaging ofcombustion gases. Appl Phys B 1990;50:441e54.

[86] Einecke S, Schulz C, Sick V. Measurement of temperature, fuel concentrationand equivalence ratio fields using tracer LIF in IC engine combustion. ApplPhys B 2000;71:717e23.

[87] Luong M, Zhang R, Schulz C, Sick V. Toluene laser-induced fluorescence forin-cylinder temperature imaging in internal combustion engines. Appl PhysB 2008;91:669e75.

[88] Robinson GA, Lucht RP, Laurendeau NM. Two-color planar laser-inducedfluorescence thermometry in aqueous solutions. Appl Opt 2008;47:2852e8.

[89] Dabiri D. Digital particle image thermometry/velocimetry: a review. ExpFluids 2009;46:191e241.

[90] Särner G, Richter M, Aldén M. Two-dimensional thermometry usingtemperature-induced line shifts of ZnO: Zn and ZnO: Ga fluorescence. OptLett 2008;33:1327e9.

[91] Hirasawa T, Kamata Y, Kaneba T, Nakamura Y. Visualization of ambient gastemperature based on two-color LIF. In: 9th Asia-Pacific Conference onCombustion. Taipei, Taiwan: National Taiwan University; 2009.

[92] Alkemade CThJ. A theoretical discussion on some aspects of atomic fluo-rescence spectroscopy in flames. Pure Appl Chem 1970;23:73e98.

[93] Omenetto N, Browner R, Winefordnerz J, Rossi G, Benetti P. “Color temper-ature” atomic fluorescence method for flame temperature measurement.Anal Chem 1972;44:1683e6.

[94] Omenetto N, Benetti P, Rossi G. Flame temperature measurements bymeans of atomic fluorescence spectrometry. Spectrochim Acta B 1972;27:453e61.

[95] Haraguchi H, Smith B, Weeks S, Johnson DJ, Winefordner JD. Measurement ofsmall volume flame temperatures by the two-line atomic fluorescencemethod. Appl Spectrosc 1977;31:156e63.

[96] Haraguchi H, Winefordner JD. Flame diagnostics: local temperature profilesand atomic fluorescence intensity profiles in air-acetylene flames. ApplSpectrosc 1977;31:195e200.

[97] Engström J, Nygren J, Aldén M, Kaminski CF. Two-line atomic fluorescenceas a temperature probe for highly sooting flames. Opt Lett 2000;25:1469e71.

[98] Aldén M, Grafströom P, Lundberg H, Svanberg S. Spatially resolvedtemperature measurements in a flame using laser-excited two-line atomicfluorescence and diode-array detection. Opt Lett 1983;8:241e3.

[99] Joklik RG, Daily JW. Two-line atomic fluorescence temperature measurementin flames: an experimental study. Appl Opt 1982;21:4158e62.

[100] Kaminski CF, Engström J, Aldén M. Quasi-instantaneous two-dimensionaltemperature measurements in a spark ignition engine using 2-line atomicfluorescence. Proc Combust Inst 1998;27:85e93.

[101] Dec JE, Keller JO. High speed thermometry using two-line atomic fluores-cence. Proc Combust Inst 1986;21:1737e45.

[102] Hult J, Burns IS, Kaminski CF. Two-line atomic fluorescence flame ther-mometry using diode lasers. Proc Combust Inst 2005;30:1535e43.

[103] Burns IS, Hult J, Kaminski CF. Spectroscopic use of a novel blue diode laser ina wavelength region around 450 nm. Appl Phys B 2004;79:491e5.

[104] Burns IS, Hult J, Hartung G, Kaminski CF. A thermometry technique based onatomic lineshapes using diode laser LIF in flames. Proc Combust Inst 2007;31:775e82.

[105] Medwell PR, Chan QN, Kalt PAM, Alwahabi ZT, Dally BB, Nathan GJ. Devel-opment of temperature imaging using two-line atomic fluorescence. ApplOpt 2009;48:1237e48.

[106] Medwell PR, Chan QN, Kalt PAM, Alwahabi ZT, Dally BB, Nathan GJ. Instan-taneous temperature imaging of diffusion flames using two-line atomicfluorescence. Appl Spectrosc 2010;64:173e6.

[107] Chan QN, Medwell PR, Kalt PAM, Alwahabi ZT, Dally BB, Nathan GJ. Simul-taneous imaging of temperature and soot volume fraction. Proc Combust Inst2011;33:791e8.

[108] Chan QN, Medwell PR, Kalt PAM, Alwahabi ZT, Dally BB, Nathan GJ. Solventeffects on two-line atomic fluorescence (TLAF) of Indium. Appl Opt 2010;49:1257e66.

[109] Chan QN, Medwell PR, Alwahabi ZT, Dally BB, Nathan GJ. Assessment ofinterferences to nonlinear two-line atomic fluorescence (NTLAF) in sootyflames. Appl Phys B; 2011. doi:10.1007/s00340-011-4497-0.

[110] Timothy LD, Sarofim AF, Beer JM. Characteristics of single particle coalcombustion. Proc Combust Inst 1982;19:1123e30.

[111] Tichenor DA, Mitchell RE, Hencken KR, Niksa S. Simultaneous in situmeasurement of the size, temperature and velocity of particles ina combustion environment. Proc Combust Inst 1984;20:1213e21.

[112] Jorgensen FRA, Zuiderwyk M. Two-colour pyrometer measurement of thetemperature of individual combusting particles. J Phys E Sci Instrum 1985;18:486e91.

[113] Huang Y, Yan Y, Riley G. Vision-based measurement of temperature distri-bution in a 500 kW model furnace using the two-color method. Measure-ment 2000;28:175e83.

[114] van Eyk PJ, Ashman PJ, Alwahabi ZT, Nathan GJ. Simultaneous measurementsof the release of atomic sodium, particle diameter and particle temperaturefor a single burning coal particle. Proc Combust Inst 2009;32:2099e106.

[115] Saw WL, Nathan GJ, Alwahabi ZT, Ashman PJ, Hupa M. Simultaneousmeasurement of the surface temperature and the release of atomic sodiumfrom a burning black liquor droplet. Combust Flame 2010;157:769e77.

[116] Lee W, Na YD. Soot study in laminar diffusion flames at elevated pressureusing two-color pyrometry and Abel inversion. JSME Int J Ser B 2000;43:550e5.

[117] Yelverton TLB, Roberts WL. Soot surface temperature measurements in pureand diluted flames at atmospheric and elevated pressures. Exp Therm FluidSci 2008;33:17e22.

[118] Reimann J, KuhlmannSA,Will S. Investigationson soot formation inheptane jetdiffusion flames by optical techniques. Microgravity Sci Tec 2010;22:499e505.

[119] Murphy JJ, Shaddix CR. Influence of scattering and probe-volume hetero-geneity on soot measurements using optical pyrometry. Combust Flame2005;143:1e10.

[120] Qamar NH, Alwahabi ZT, Chan QN, Nathan GJ, Roekaerts D, King KD. Sootvolume fraction in a piloted turbulent jet non-premixed flame of natural gas.Combust Flame 2009;156:1339e47.

[121] Jenkins TP, Hanson RK. Soot pyrometry using modulated absorption/emis-sion. Combust Flame 2001;126:1669e79.

[122] Lu G, Yan Y. Temperature profiling of pulverized coal flames using multicolorpyrometric and digital imaging techniques. IEEE T Instrum Meas 2006;55:1303e8.

[123] Fu TR, Wang Z, Cheng XF. Temperature measurements of diesel fuelcombustion with multicolor pyrometry. J Heat Trans 2010;132:0516021e7.

[124] Klimenko AY. Multicomponent diffusion of various admixtures in turbulentflow. Fluid Dyn 1990;25:327e34.

[125] Peters N. Laminar flamelet concepts in turbulent combustion. Proc CombustInst 1986;21:1231e50.

[126] Bilger RW. Conditional moment closure for turbulent reacting flow. PhysFluids A 1993;5:436e44.

[127] Hsu AG, Narayanaswamy V, Clemens NT, Frank JH. Mixture Fraction Imagingin turbulent non-premixed flames with two-photon LIF of krypton. ProcCombust Inst 2010;33:759e66.


[128] Chen YC, Starner SH, Masri AR. Characteristics of turbulent spray combustionin a piloted jet flame burner. Proc Combust Inst 2002;29:625e32.

[129] Masri AR, Kelman JB, Dally BB. The instantaneous spatial structure of therecirculation zone in bluff-body stabilised flames. Proc Combust Inst 1998;27:1031e8.

[130] Barlow RS, Frank JH, Karpetis AN, Chen JY. Piloted methane/air jet flames:transport effects and aspects of scalar structure. Combust Flame 2005;143:433e49.

[131] Gregor MA, Seffrin F, Fuest F, Geyer D, Dreizler A. Multi-scalar measurementsin a premixed swirl burner using 1D Raman/Rayleigh scattering. ProcCombust Inst 2009;32:1739e46.

[132] Geyer D, Kempf A, Dreizler A, Janicka J. Turbulent opposed-jet flames:a critical benchmark experiment for combustion LES. Combust Flame 2005;143:524e48.

[133] Eckbreth AC. Effects of laser-modulated particulate incandescence on Ramanscattering diagnostics. J Appl Phys 1977;48:4473e9.

[134] Axelsson B, Collin R, Bengtsson PE. Laser-induced incandescence for sootparticle size and volume measurements using on-line extinction calibration.Appl Phys B 2001;72:367e72.

[135] Appel J, Jungfleisch B, Marquardt M, Suntz R, Bockhorn H. Assessment of sootvolume fractions from laser-induced incandescence by comparison withextinction measurements in laminar, premixed, flat flames. Proc CombustInst 1996;26:2387e95.

[136] Vander Wal RL, Zhou Z, Choi MY. Laser-induced incandescence calibrationvia gravimetric sampling. Combust Flame 1996;105:462e70.

[137] Dalzell WH, Sarofim AF. Optical constants of soot and their application toheat-flux calculations. J Heat Trans-T ASME 1969;91:100e4.

[138] Choi MY, Mulholland GW, Hamins A, Kashiwagi T. Comparisons of the sootvolume fraction using gravimetric and light extinction techniques. CombustFlame 1995;102:161e9.

[139] Zhu J, Choi MY, Mulholland GW, Manzello SL, Gritzo LA, Suo-Anttila JM.Measurement of visible and near-IR optical properties of soot produced fromlaminar flames. Proc Combust Inst 2002;29:2367e74.

[140] Shaddix CR, Palotas AB, Megaridis CM, Choi MY, Yang NYC. Soot graphiticorder in laminar diffusion flames and a large-scale JP-8 p.ol fire. Int J HeatMass Tran 2005;48:3604e14.

[141] Williams TC, Shaddix CR, Jensen KA, Suo-Anttila JM. Measurement of thedimensionless extinction coefficient of soot within laminar diffusion flames.Int J Heat Mass Tran 2007;50:1616e30.

[142] Smyth KC, Shaddix CR. The elusive history of ~m ¼ 1.57e0.56i for therefractive index of soot. Combust Flame 1996;107:314e20.

[143] Sorensen CM. Light scattering by fractal aggregates: a review. Aerosol SciTech 2001;35:648e87.

[144] Migliorini F, Thomson KA, Smallwood GJ. Investigation of optical propertiesof aging soot. Appl Phys B; 2011. doi:10.1007/s00340-011-4396-4.

[145] Zerbs J, Geigle KP, Lammel O, Hader J, Stirn R, Hadef R, et al. The influence ofwavelength in extinction measurements and beam steering in laser-inducedincandescence measurements in sooting flames. Appl Phys B 2009;96:683e94.

[146] Bengtsson PE, Aldén M. Application of a pulsed laser for soot measurementsin premixed flames. Appl Phys B 1989;48:155e64.

[147] Henriksen TL, Nathan GJ, Alwahabi ZT, Qamar NH, Ring TA, Eddings EG.Planar measurements of soot volume fraction and OH in a JP-8 pool fire.Combust Flame 2009;156:1480e92.

[148] Lee SY, Turns SR, Santoro RJ. Measurements of soot, OH, and PAH concen-trations in turbulent ethylene/air jet flames. Combust Flame 2009;156:2264e75.

[149] Schoemaecker Moreau C, Therssen E, Mercier X, Pauwels JF, Desgroux P.Two-color laser-induced incandescence and cavity ring-down spectroscopyfor sensitive and quantitative imaging of soot and PAHs in flames. Appl PhysB 2004;78:485e92.

[150] Ciajolo A, Ragucci R, Apicella B, Barbella R, de Joannon M, Tregrossi A.Fluorescence spectroscopy of aromatic species produced in rich premixedethylene flames. Chemosphere 2001;42:835e41.

[151] Joedicke A, Peters N, Mansour M. The stabilization mechanism andstructure of turbulent hydrocarbon lifted flames. Proc Combust Inst 2005;30:901e9.

[152] CiajoloA,TregrossiA,BarbellaR,Ragucci R,ApicellaB,de JoannonM.Therelationbetween ultraviolet-excited fluorescence spectroscopy and aromatic speciesformed in rich laminar ethylene flames. Combust Flame 2001;125:1225e9.

[153] Alwahabi ZT, Qamar NH, Nathan GJ. Unpublished results.[154] Choi MY, Jensen AA. Calibration and correction of laser induced incandes-

cence for soot volume fraction measurement. Combust Flame 1998;112:485e91.

[155] Geitlinger H, Th Streibel, Suntz R, Bockhorn H. Two-dimensional imaging ofsoot volume fraction, particle number densities and particle radii in laminarand turbulent diffusion flames. Proc Combust Inst 1998;27:1613e21.

[156] Bockhorn H, Geitlinger H, Jungfleisch B, Lehre Th, Schön A, Th Streibel, et al.Progress in characterization of soot formation by optical methods. PhysChem Chem Phys 2002;4:3780e93.

[157] Black DL, McQuay MQ, Bonin MP. Laser-based techniques for particle-sizemeasurement: a review of sizing methods and their industrial applica-tions. Prog Energy Combust Sci 1996;22:267e306.

[158] Kalt PAM, Nathan GJ, Kelso RM. Planar laser measurements of the aero-dynamic agglomeration caused by Indigos Emitter. School of MechanicalEngineering, University of Adelaide; 2005. Report No. MECHTEST MT0659-3.

[159] Cassel HM, Liebman I. The cooperative mechanism in the ignition of dustdispersions. Combust Flame 1959;3:467e75.

[160] Dhodapkar S, Jacobs K, Hu S. In: Crowe CT, editor. Multiphase flow hand-book: fluid-solid transport in ducts. Boca Raton, USA: CRC Press; 2005.

[161] Smith NL, Nathan GJ, Zhang DK, Nobes DS. The significance of particleclusters in pulverised coal flames. Proc Combust Inst 2002;29:797e804.

[162] Fan J, Zhao H, Cen K. An experimental study of two-phase turbulent coaxialjets. Exp Fluids 1992;13:279e87.

[163] Longmire EK, Eaton JK. Structure of a particle-laden round jet. J Fluid Mech1992;236:217e57.

[164] Kalt PAM, Birzer CH, Nathan GJ. Corrections to facilitate planar imaging ofparticle concentration in particle-laden flows using Mie-scattering, part 1:collimated laser sheets. Appl Opt 2007;46:5823e34.

[165] Kalt PAM, Nathan GJ. Corrections to facilitate planar-imaging of particleconcentration in particle-laden flow using Mie-scattering part 2: diverginglaser sheets. Appl Opt 2007;46:7227e36.

[166] Herzhaft B, Guazzelli E. Experimental study of the sedimentation of diluteand semi-dilute suspensions of fibres. J Fluid Mech 1999;384:133e58.

[167] Kuusela E, Lahtinen JM, Ala-Nissila T. Collective effects in settling of spher-oids under steady-state sedimentation. Phys Rev Lett 2003;90:094502.

[168] Lin JZ, Shi X, Yu ZS. The motion of fibers in an evolving mixing layer. Int JMultiphase Flow 2003;29:1355e72.

[169] Berrocal E, Kristensson E, Richter M, Linne MA, Aldén M. Application ofstructured illumination for multiple scattering suppression in planar laserimaging of dense sprays. Opt Express 2008;16:17870e81.

[170] Kristensson E, Berrocal E, Richter M, Petterson SG, Aldén M. High-speedstructured planar laser illumination for contrast improvement of two-phaseflow images. Opt Lett 2008;33:2752e4.

[171] Prasad AK. Particle image velocimetry. Curr Sci 2000;79:51e60.[172] Adrian RJ. Twenty years of particle image velocimetry. Exp Fluids 2003;39:

159e69.[173] Han D, Mungal MG. Simultaneous measurements of velocity and CH distri-

butions. Part 1: jet flames in co-flow. Combust Flame 2003;132:565e90.[174] Cetegen BM, Basu S. Soot topography in a planar diffusion flame wrapped by

a line vortex. Combust Flame 2006;146:687e97.[175] Shaddix CR, Smyth KC. Laser-induced incandescence measurements of soot

production in steady and flickering methane, propane and ethylene diffusionflames. Combust Flame 1996;107:418e52.

[176] Delichatsios MA, de Ris J, Orloff LL. An enhanced flame radiation burner. ProcCombust Inst 1992;24:1075e82.

[177] Turns SR. Understanding NOx formation in nonpremixed flames: experi-ments and modelling. Prog Energy Combust Sci 1995;21:361e85.

[178] Smart JP, O’Nions P, Riley GS. Radiation and convective heat transfer andburnout in oxy-coal combustion. Fuel 2010;89:2468e76.

[179] Chedaille J, Braud Y. In: Beer JM, Thring MW, editors. Measurements inflames: industrial flames, vol. 1. London: Edward Arnold; 1972.

[180] Medtherm Corporation. Description and operating instructions MedthermH-201 heat flux meter, model E. Alabama: The Corportation; 1987.

[181] Sivathanu YR, Gore JP. Total radiative heat loss in jet flames fromsingle point radiative flux measurements. Combust Flame 1993;94:265e70.

[182] Kalt PAM, Birzer CH, Nathan GJ. Quantified nephelometry: planar images ofparticle concentration in optically thick flows. In: Int Conf Multiphase Flows,30 Maye4 June, Florida, USA: 2010.

Recent advances in the measurement of strongly radiating, turbulent reacting flows

Documents