Quantitative infrared imaging of impinging turbulent ...

Available online at www.sciencedirect.comProceedings

ScienceDirect

Proceedings of the Combustion Institute xxx (2014) xxx–xxx

www.elsevier.com/locate/proci

of the

CombustionInstitute

Quantitative infrared imaging of impingingturbulent buoyant diffusion flames

Ashish S. Newale a,c,⇑, Brent A. Rankin a,d, Harshad U. Lalit b,Jay P. Gore a, Randall J. McDermott e

a School of Mechanical Engineering, Purdue University, West Lafayette, IN 47907, USAb School of Aeronautics and Astronautics, Purdue University, West Lafayette, IN 47907, USA

c CD-adapco, Houston, TX 77042, USAd Innovative Scientific Solutions Inc., Dayton, OH 45459, USA

e Fire Research Division, National Institute of Standards and Technology, Gaithersburg, MD 20899, USA

Abstract

A buoyant fire impinging on a horizontal ceiling at a certain distance from the fuel source occurs inmany practical fire scenarios. Motivated by this application, infrared radiation from buoyant diffusionflames with and without impingement on a flat plate is studied using a quantitative comparison ofmeasured and simulated images. The measured quantitative images of the radiation intensity are acquiredusing a calibrated high speed camera. Simulated radiation intensities are rendered in the form of imagesand compared quantitatively with the measured images. The simulated radiation intensities are obtainedusing the radiative transfer equation with local absorption coefficients evaluated using a narrowbandradiation model. The instantaneous local species concentrations, soot volume fractions, and temperaturesnecessary for these simulations are calculated using the Fire Dynamics Simulator (FDS) version 6. Themeasured images reveal that the characteristic necking and bulging (7 Hz� 1 Hz) of the free buoyant flameis suppressed to a large extent by impingement on the plate. The roll-up vortices in the impinging flame aremuch smaller than those in the free flame. The stagnation point boundary layer inferred from the computedimages is thicker at some instances than that inferred from the measurements. Qualitative and quantitativecomparisons between the measured and computed infrared images for both the free and the impinging firesreveal many similarities as well as differences useful for model evaluation.� 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Turbulent buoyant flames; Impinging flames; Flame radiation; Fire Dynamics Simulator; Infrared imaging

http://dx.doi.org/10.1016/j.proci.2014.05.1151540-7489/� 2014 The Combustion Institute. Published by El

⇑ Corresponding author. Address: CD-adapco 11000Richmond Avenue, Suite 110, Houston, TX 77042,USA.

E-mail address: [email protected](A.S. Newale).

Please cite this article in press as: A.S. Newale et al.10.1016/j.proci.2014.05.115

1. Introduction

Fire structure interaction studies explorethe complex interactions between the gas phaseflow field and heat release rate and the structuraland thermal properties of load bearing elements.In particular, a buoyant fire impinging on a

sevier Inc. All rights reserved.

, Proc. Combust. Inst. (2014), http://dx.doi.org/

Nomenclature

n unit normal vector on the cell faces0 direction vector of the radiation

intensitydV cell volume_q000 heat release rate per unit volume_q00c convective heat fluxIbk spectral black body radiation intensityk thermal conductivityL characteristic lengthNu Nusselt numberPr Prandtl numberRe Reynolds numbers pathlength through flameT temperaturet timeU integrated radiation intensity

Greek symbolsvr radiative heat loss fraction_q00r radiative heat flux

Cc empirical coefficient for naturalconvection

Cr corrective factor for radiative lossterm

h heat transfer coefficientI radiation intensityj absorption coefficientjk spectral absorption coefficientk1; k2 filter bandwidth limitsX solid angler Stefan–Boltzman constantsk spectral optical thicknesse emissivity

Subscriptsg gasijk gas phase cell indicess solidw wall

2 A.S. Newale et al. / Proceedings of the Combustion Institute xxx (2014) xxx–xxx

horizontal ceiling at a certain distance from thefuel source occurs in many practical fire scenarios.Heat transfer from a fire to a structure has beenthe subject of many investigations [1–3]. Radia-tion and convection heat transfer have been estab-lished as important modes of energy transferbetween the gas phase and the solid [4]. Trouveand Wang [5] suggested that comparing higherorder statistics resulting from measurements andsimulations of fires is an important direction foradvancing progress within the fire modeling field.The current work involves a comparison of time-dependent images and higher order statistics ofinfrared radiation intensity from non-impingingand impinging buoyant diffusion flames.

One of the most challenging aspects of coupledfire structure interaction simulations is the signifi-cant range in the time and length scales. The differ-ent numerical methods used for fire, thermal, andstructural simulations lead to difficulties in deter-mining an efficient method of transferring infor-mation between these simulations [4]. Weng andHasemi [1] developed an analytical model to deter-mine the thermal response of an unconfined non-burning ceiling to an impinging buoyant diffusionflame. The model showed good agreement withexperimental heat flux data in regions far fromthe stagnation point. You [6] developed an integralmodel for impinging turbulent buoyant diffusionflames to determine mean velocities, temperatureand major species concentrations in the plumeregion. The integral model was extended to theceiling jet region and provided good agreementwith measured velocities and heat fluxes [2]. Tuttle


et al. [7,8] reported time-dependent and time-aver-aged measurements of axial and radial profiles oflocal heat fluxes from partially premixed meth-ane-air flames impinging on a water-cooled plate.The coupling between flame structure and heattransfer was explored by qualitatively correlatingvisible photographs of the flame to measured heatflux profiles. Premixed flames typically producethe highest heat flux and are used in industrialapplications to provide localized heat transfer.Baukal and Gebhart [9] and Viskanta [10] havereported comprehensive reviews on flame impinge-ment heat transfer for manufacturing applications.

A quantitative comparison of measured andcomputed images of radiation intensity is an effec-tive method for prompting improvements in com-bustion and radiation models [11]. Rankin et al.[12,13] measured and computed quantitative imagesof infrared radiation intensity from a representativeturbulent non-premixed jet flame from the Interna-tional Workshop on Measurement and Computa-tion of Turbulent Nonpremixed Flames. Thecomputed images utilized scalar results from large-eddy simulations (LES) of the flame performed byIhme et al. [14]. The measured and computed radia-tion intensities agreed to within approximately 20%for regions upstream of the stoichiometric flamelength. Differences between the measured and com-puted radiation intensities near and downstream ofthe stoichiometric flame length suggested thatincluding radiation heat loss effects in the simula-tions is important even for weakly radiating flames.

Motivated by a review of existing literature onfire-structure interactions and the potential of


A.S. Newale et al. / Proceedings of the Combustion Institute xxx (2014) xxx–xxx 3

computational and experimental quantitativeinfrared imaging techniques, the specific objec-tives of the current study are:

(1) Measure quantitative images of the narrow-band infrared radiation intensity from arepresentative buoyant diffusion flame withand without impingement on a stagnationplate.

(2) Compute images of the narrowband infra-red radiation intensity using time dependentfire simulations, a narrowband radiationmodel, and the radiative transfer equation.

(3) Study the effects of flame impingement onflame radiation by comparing the measuredand computed images.

(4) Compare measured and computed mean,root mean square (RMS), and power spec-tral densities (PSDs) of the radiationintensity.

The measurements reveal a less prominentpeak at a characteristic frequency of 7 Hz forthe impinging buoyant diffusion flame comparedto the non-impinging flame. The roll-up vorticesin the impinging flame are much smaller thanthose in the free flame. The stagnation pointboundary layer inferred from the computedimages is thicker at some instances than thatinferred from the measurements. The quantitativecomparative study of measured and computedimages suggests the need for radiation and com-bustion model improvements.

2. Experimental methods

2.1. Turbulent buoyant diffusion flame

A turbulent buoyant diffusion flame is estab-lished on a diffuser burner [15] with an exit diam-eter of 7.1 cm. The diverging angle of the burner is7� such that the gaseous fuel (methane) is deceler-ated and forms close to a uniform velocity distri-bution at the burner exit. The methane (CH4)mass flow rate (84.3 mg/s ± 3%) is calibratedusing a dry test turbine meter and controlled bysetting the pressure upstream of a choked orificeplate. Measured and computed vertical and hori-zontal velocity, mixture fraction, and temperaturevalues for this flame have been reported by Zhouand Gore [15] and Xin et al. [16].

A schematic of the experimental arrangementand coordinate system definitions for the radia-tion measurements are illustrated in Fig. 1. Asquare steel plate (25.4 cm � 25.4 cm � 0.254 cm)is oriented perpendicular to the flame centerlineand suspended at an axial location of 14.2 cm(Z=D ¼ 2) downstream of the burner exit. Theplate is supported at each of the four corners on


the bottom surface using thin brackets to mini-mize flame disturbance.

2.2. Radiation intensity measurements

Time dependent images of the radiation inten-sity from the flame and the plate are measuredusing a calibrated infrared camera with a 25 mmlens (f =2:3) and an indium antimonide (InSb)detector. A narrowband filter (2.77 lm ± 0.12lm) is used to measure the radiation intensityoriginating from water vapor, carbon dioxide,and soot. The spatial resolution of the radiationintensity measurements is approximately0.97 mm for each pixel at the center of the flame.The infrared camera sampling frequency is 330 Hzand the exposure time is 10 s. The experimentaluncertainty in the mean and RMS radiation inten-sity for diametric paths is estimated to be less than�15% based on three repeated measurements of aturbulent jet diffusion flame using the sameinfrared camera and optical arrangement [12].

3. Computational methods

3.1. Scalar computations

Fire Dynamics Simulator (FDS) version 6 isused to simulate the turbulent buoyant diffusionflame with and without impingement. FDS is alarge eddy simulation (LES) code based on a setof governing equations for low-Mach numberflows [17]. For the calculations presented in thispaper, the LES subgrid closures are based onthe dynamic Smagorinsky model [18] with con-stant turbulent Schmidt and Prandtl numbers.The mean chemical source term is closed usingthe eddy dissipation concept [19].

The computational domain for the flamewithout impingement is 20 cm � 20 cm�40 cm.The computational domain used for the imping-ing flame simulation is 30 cm � 30 cm � 22 cm.A uniform grid with 4 mm spatial resolutionis used for both the non-impinging and imping-ing flame simulations. A radiative heat lossfraction of 10% is specified for both flamesbased on past estimates [16]. The radiative lossterm is computed in each cell with the gray gasassumption as a product of blackbody radiationintensity and a corrective factor, Cr [17] asdefined below:

Cr ¼

P_q000ijk>0 vr _q000ijk þ jijk U ijk

� �dV

P_q000ijk>0 4jijk rT 4

ijk

� �dV

: ð1Þ

The corrective factor is set to dynamicallyensure that the global integral of the product ofradiative heat loss fraction and heat release rateis satisfied.


Fig. 1. A schematic of the experimental arrangement for radiation intensity imaging measurements of the turbulentbuoyant diffusion flames without (left) and with (right) impingement on a stagnation plate. Note that D = 7.1 cm andd = 80 cm.


The boundary conditions at the burner exitplane are defined to be a uniform axial velocityand fuel mass fraction profiles (w ¼ 3:14 cm/s,Y F ¼ 1). Open boundary conditions are specifiedfor the remaining bounding surfaces of the com-putational domain. The plate is specified as con-ducting and heat transfer occurs from both theflame and wake sides of the plate. The initial tem-perature of the plate is specified to be equal toambient conditions. The boundary conditions onboth the front and back surfaces of the plate areimplemented [17] as follows:

�ks@T s

@xð0; tÞ ¼ _q00c þ _q00r : ð2Þ

The convective heat flux and convection heattransfer coefficients are computed as,

_q00c ¼ hðT g � T wÞ; h ¼ max CcjT g � T wj13;

kL

Nu� �

;

ð3Þ

where Cc is specified as 1:52 for a horizontal plateand L is specified as 1 m. The Nusselt number iscomputed as,

Nu ¼ 0:037Re0:8Pr0:33: ð4ÞThe radiative heat flux is computed as follows:

_q00r ¼ eZ

s0 �nw<0

Iwðs0Þjs0 � nwjdX� erT 4w: ð5Þ

3.2. Radiation intensity computations

Computed images of the radiation intensityfrom the flames are rendered using the computedscalar values, a narrowband radiation model andthe radiative transfer equation with a customizedin-house code. The narrowband radiation


intensity (I) emitted from parallel lines-of-sightthrough the flame is computed using the solutionof the radiative transfer equation for non-scattering media [20], with the addition [11–13]of a spectral coefficient (ak) associated with thetransmission optics and spectral integration overthe band-pass filter,

I ¼Z k2

k1

akIkð0Þe�sk dk

þZ k2

k1

Z sk

0

akIbkðs�kÞe�ðskþs�kÞ ds�k dk: ð6Þ

The spectral optical thickness is defined as:

sk ¼Z s

0

jk ds: ð7Þ

The spectral absorption coefficient (jk) is cal-culated using the computed scalar values and anarrowband radiation model [21]. The intensityis calculated by integrating along lines-of-sightthrough the flame, applying the spectral responseof the optics and camera, and integrating overthe spectral range of the filter to allow for com-parison with the measured values. An image ofthe radiation intensity from the flames isobtained by performing the radiation intensitycomputations along multiple parallel lines-of-sight through the flame consistent with thespatial and temporal resolution of the computedscalar values.

The images of the radiation intensity are calcu-lated starting at 10 s to exclude initial transients inthe simulations. Time intervals of 10–20 s with asampling frequency of 33 Hz are sufficient toobtain statistically stationary mean values of theradiation intensity. The higher order statisticsare calculated over a time interval of 10–40 s witha sampling frequency of 200 Hz.


Fig. 2. Measured [15] and computed radial profiles of the mean vertical velocity (left), horizontal velocity (center), andtemperature (right) at select heights above the burner exit.


4. Results and discussion

Figure 2 shows a comparison of the radial pro-files of vertical velocity, horizontal velocity, andtemperature from the current computation ofthe non-impinging flame with previous measure-ments reported by Zhou and Gore [15]. These pastmeasurements extend to the distance of 10 cmdownstream of the burner exit compared toapproximately 22 cm for the current radiationintensity measurements. The radial profiles fortemperature reported by Zhou and Gore [15]and Xin et al. [16] were estimated from speciesconcentration measurements assuming an adia-batic flame. The measured and computed verticalvelocities agree to within approximately 15%along the flame centerline. The locations of thepeak magnitude of the vertical and horizontalvelocities observed in the measurements and com-putations agree to within approximately 1 cm.The results of the computations agree with theexperimentally derived temperature profileswithin 15% beyond 0.5 cm downstream of theburner exit.

Figure 3 compares six instantaneous tempo-rally related realizations of radiation intensityimages obtained from the measurements and thecomputations for the non-impinging buoyant dif-fusion flame. The six realizations encompass onepulsation cycle. The instantaneous realizations oftime-dependent radiation intensity images are sep-arated by an interval of 30 ms. The initial phase inthe pulsation cycle is matched for measurements(top six panels) and computations (bottom sixpanels). The peak magnitude of radiation intensityobserved in the measurements (500 W=m2-sr) iscaptured by the computations in each of the sixpanels comprising one realization of the pulsationcycle. The remarkable dynamic range (0 W=m2-srto 500 W=m2-sr) illustrated by the measurementsis matched by the computations. It is remarkable


that the identical range and increment used inthe rendering color palette leads to many similar-ities in the observed patterns. For example, theleft most measured and computed images bothshow a concentric conical flame attachmentregion followed by necking and formation of abulging structure with two symmetric red circlescorresponding to the high temperature regionassociated with the buoyant vortex ring. Thisagreement approximately repeats itself in everysixth image as shown in the right most images ofthe top and the bottom panels. However, thereare specific differences in the degree of spatialorganization between the measured and the com-puted images in each of the six interim phase pan-els. In particular, the measured intensities showsignificantly higher symmetric organization com-pared to the computed intensities. For example,the highest intensities depict a continuation of asymmetric bulging shape in the measured imageof the second top panel while the highest intensi-ties show a dislocation of the left part of the bulg-ing shape by apparent faster advection andexpansion in the computed image of the secondbottom panel. These processes result in an L-shaped structure with the two highest intensity cir-cles appearing at each end in the computed image.Similar distortions exist in each one of the fourintermediate phase panels in the top and bottomrows of Fig. 3. While similar differences each withtheir own aesthetic but unimportant individualityexist throughout the measured and simulatedimages, the net result is of significantly larger dis-organization in the simulated compared to themeasured cycles. These observations may be ofvalue for continuous improvements in aspects offire dynamics simulations not discussed here forbrevity with interested readers referred to theFDS documentation [17].

Figure 4 shows a comparison of six time-depen-dent radiation intensity images for the impinging


Fig. 3. Measured (top) and computed (bottom) time-dependent images of the radiation intensity from the non-impinging turbulent buoyant diffusion flame (characteristic puffing mode frequency is 7 Hz).

Fig. 4. Measured (top) and computed (bottom) time-dependent images of the radiation intensity from the turbulentbuoyant diffusion flame impinging on the plate suspended at Z ¼ 14:2 cm.


flame. Six images are shown as before to encom-pass one pulsation cycle and the initial phase inthe pulsation cycle is matched between measure-ment and computations. The time-dependentimages show a region of high radiation intensity


at small distances upstream of the plate. Thecomputations capture the peak magnitude ofradiation intensity observed in the measurements(1000 W=m2-sr). The peak magnitude observedfor the impinging flame is twice that for the



non-impinging flame partially because of theincreased path length through the stagnation pointboundary layer. Very strong stabilizing effects ofthe flow stagnation are apparent in the experimen-tal images. The computed intensity images con-tinue to show remnants of the axially symmetricvortex structure in most panels in the bottom halfof Fig. 4. This qualitative observation is confirmedby comparisons of measured and computed powerspectral densities discussed later in the paper.The third and the fourth measured and computedimages in the cycle show impingement of the vor-tex structure on the stagnation plate. The apparenthigh intensity region of the measured structure istwice as thick as that of the computed structure.Measured and computed instantaneous images ofthe infrared radiation intensity from the free andimpinging buoyant diffusion flames are renderedin the form of animations and provided as supple-mentary material. Further quantitative compari-sons of the effects rendered in the instantaneousimages shown in Figs. 3 and 4 are studied in Figs. 5and 6.

Figure 5a shows the measured and computedmean radiation intensities from diametric pathsas function of distance downstream of the burnerexit for the non-impinging and the impingingflames. The measurements and the computationsbeyond approximately 4 cm downstream of theburner exit are within 25% of each other. The fac-tor of two differences between the computed andmeasured mean and fluctuating radiation intensitynear the burner exit exist because of relatively sim-ple models of the burner exit boundary conditionsincluding partial premixing, heat loss, and finiterate chemistry. It is remarkable that the meanradiation intensities are dominated by the buoy-ancy induced advection and mixing overcomingthe effects of boundary conditions. Both the mea-surements and the computations show that the

Fig. 5. Measured and computed mean (a) and normalized rootfrom diametric paths (Y ¼ 0 cm) as a function of distance dow


peak magnitude of the mean radiation intensityfor the impinging flame occurs at approximately0.5 cm upstream of the stagnation point. This isconsistent with the relatively short duration expo-sure of the relatively cold plate to the impingingflame. The differences in the sharp peak value ofthe intensity may be addressed by future enhance-ments in the stagnation point combustion model.The location of the peak magnitude of mean radi-ation intensity for the impinging flame is alsoindicative of the upstream propagation effect ofthe plate on the buoyant diffusion flame. Mea-sured and computed time-averaged images of theinfrared radiation intensity from the free andimpinging buoyant diffusion flames are providedas supplementary material.

The root mean square (RMS) values of radia-tion intensity from diametric paths normalizedby the mean radiation intensity are plotted as afunction of distance downstream of the burnerexit in Fig. 5b. The measured and computed nor-malized RMS increase downstream of the burnerexit region for both the impinging and the non-impinging flames. The measured and computednormalized RMS profiles of radiation intensityfor the impinging flame show a local minimumat approximately 0.5 cm upstream of the stagna-tion point caused by the stabilizing effects of thestagnation point boundary layer.

Figure 6 shows a comparison of the measuredand computed power spectral density (PSD) ofradiation intensity from diametric paths for thenon-impinging and impinging buoyant diffusionflames at a location selected to correspond tothe measured peak mean intensity (Y ¼ 0 cm,Z ¼ 13:6 cm) for the impinging flame. The mea-sured and computed PSDs shown in the left paneldepict the expected peaks at pulsation frequenciesof approximately 7 Hz and the expected decre-ment according to a power law at frequencies

mean square (RMS) (b) of the radiation intensity emittednstream of the burner exit.


Fig. 6. Measured and computed power spectral density (PSD) functions for the non-impinging (left) and impinging(right) buoyant diffusion flames at a select location.


higher than 10 Hz. The measured PSD in the righthand panel of Fig. 6 show that a less prominentpeak is observed at the characteristic frequencyof 7 Hz compared to the non-impinging flame.This can be attributed to the pressure gradientintroduced by the stagnation point boundarylayer. This drop in magnitude of the PSD at thepulsation frequency is consistent with the reducedRMS of intensity and underlying temperature andscalar fluctuations resulting from the effects of thestagnation point. This experimental observation isnot only of significant practical relevance forexample in fire sensing but also may promptfuture improvements in fire simulations.

5. Conclusions

A quantitative comparative study of measuredand computed images of the radiation intensityfrom non-impinging and impinging turbulentbuoyant diffusion flames is utilized to evaluate firedynamic simulations. The computed mean radia-tion intensities agree with the measurements towithin 20 % downstream of the flame stabilizationregion. The computed images of the radiationintensity capture the flame necking and bulgingof the non-impinging flames observed in themeasurements.

The pressure gradient introduced by the stag-nation point boundary layer leads to significantsmoothening of the PSD near the characteristicfrequency (7 Hz) of the buoyant diffusion flame.The roll-up vortices in the impinging flame aremuch smaller than those in the free flame. Thestagnation point boundary layer inferred fromthe computed images is thicker at some instancesthan that inferred from the measurements. Quali-tative and quantitative comparisons between themeasured and computed infrared images for both


the free and the impinging fires reveal many simi-larities as well as differences useful for modelevaluation.

Acknowledgements

The National Institute of Standards and Tech-nology and the National Science Foundation aregratefully acknowledged for supporting previousflame structure interaction work at Purdue Uni-versity. The authors thank Dr. Matthias Ihmefor insightful technical discussions.

Appendix A. Supplementary data

Supplementary data associated with this articlecan be found, in the online version, at http://dx.doi.org/10.1016/j.proci.2014.05.115.

References

[1] W. Weng, Y. Hasemi, Heat Mass Transfer 42(2006) 652–659.

[2] H. You, Fire Mater. 9 (1985) 46–56.[3] H. You, G. Faeth, Fire Mater. 3 (1979) 140–147.[4] H. Baum, Mech. Res. Commun. 38 (2011) 1–11.[5] A. Trouve, Y. Wang, Int. J. Comput. Fluid Dynam.

24 (2010) 449–466.[6] H. You, Fire Mater. 8 (1984) 28–39.[7] S. Tuttle, B. Webb, M. McQuay, Int. J. Heat Mass

Transfer 48 (2005) 1236–1251.[8] S. Tuttle, B. Webb, M. McQuay, Int. J. Heat Mass

Transfer 48 (2005) 1252–1266.[9] C. Baukal, B. Gebhart, Combustion Sci. Technol.

104 (1995) 339–357.[10] R. Viskanta, Exp. Therm. Fluid Sci. 6 (1993) 111–

134.



[11] B. Rankin, D. Blunck, V. Katta, S. Stouffer, J.Gore, Combustion Flame 159 (2012) 2841–2843.

[12] B. Rankin, Qnatitative Experimental and Model-based Imaging of Infrared Radiation Intensity fromTurbulent Reacting Flows, Ph.D. thesis, PurdueUniversity, West Lafayette, IN, USA, 2012.

[13] B. Rankin, D. Blunck, J. Gore, J. Heat Transfer135 (2013).

[14] M. Ihme, H. Pitsch, D. Bodony, Proc. CombustionInst. 32 (2009) 1545–1553.

[15] X. Zhou, J. Gore, Proc. Combustion Inst. 27 (1998)2767–2773.

[16] Y. Xin, J. Gore, K. McGrattan, R. Rehm, H.Baum, Combustion Flame 141 (2005) 329–335.


[17] Fire Dynamics Simulator, Technical ReferenceGuide, National Institute of Standards and Tech-nology, Gaithersburg, Maryland, USA and VTTTechnical Research Centre of Finland, Espoo,Finland, sixth ed., 2013.

[18] P. Moin, K. Squires, W. Cabot, S. Lee, Phys. Fluids3 (1991) 2746–2757.

[19] B. Magnussen, B. Hjertager, Proc. Combustion Inst.16 (1977) 719–729.

[20] M. Modest, Radiative Heat Transfer, McGraw-HillInc., New York, 2003.

[21] W. Grosshandler, RADCAL: A Narrow-bandModel for Radiation Calculations in a CombustionEnvironment, National Institute of Standards andTechnology, 2005.


Quantitative infrared imaging of impinging turbulent ...

Documents

Transcript of Quantitative infrared imaging of impinging turbulent ...