Samuel Goroshin et al- Optical Pyrometry of Fireballs of Metalized Explosives

Full Paper

Optical Pyrometry of Fireballs of Metalized Explosives

Samuel Goroshin, David L. Frost*, Jeffrey Levine

McGill University, Mechanical Engineering, 817 Sherbrooke St. W., Montreal, Quebec, H3A 2K6 (Canada)

Akio Yoshinaka, Fan Zhang

Defence R&D Canada – Suffield, Box 4000, Stn. Main, Medicine Hat, Alberta, T1A 8K6 (Canada)

DOI: 10.1002/prep.200600024

Abstract

Fast-response optical diagnostics (a time-integrated spectrom-eter and two separate fast-response three-color pyrometers) areused to record the transient visible radiation emitted by a fireballproduced when a condensed explosive is detonated. Measurementof the radiant intensity, in several narrow wavelength bands, isused to estimate the temperature of the condensed productswithin the fireball. For kg-scale conventional oxygen-deficienthomogeneous TNT and nitromethane explosive charges, theradiant intensity reaches a maximum typically after tens ofmilliseconds, but the measured fireball temperature remainslargely constant for more than 100 ms, at a value of about2,000 K, consistent with predictions using equilibrium thermody-namics codes. When combustible metal particles (aluminum,magnesium or zirconium) are added to the explosive, reaction ofthe particles enhances the radiant energy and the fireball temper-ature is increased. In this case the fireball temperatures are lowerthan equilibrium predictions, but are consistent with measure-ments of particle temperature in single particle ignition experi-ments.

Keywords: Optical Pyrometry, Spectrometry, Fireball, MetalizedExplosive

1 Introduction

The addition of metal powders is a common technique formodifying the performance of conventional high explosives.Sincemost high explosives are oxygen deficient, a priori it isnot clear to which extent the overall energy release or therate of energy release will be increased whenmetal particlesare added. Very fine particles will react rapidly, but mayremain within the combustion products, limiting the overalldegree of reaction due to the lack of sufficient oxidizers.Large metallic particles will be dispersed widely followingthe detonation of the charge, but may not ignite within thecombustion products if the residence time is too short toheat the particles sufficiently. Hence there will be an

optimum range of particle sizes for which themetal additivewill react, not only within the explosive products, but alsowith the surrounding air, leading to a large increase inspecific energy release. Conversion of the chemical energystored in a metalized explosive to energy available tosupport the blast wave therefore depends not only on thereaction rate of the metal particles with various oxidizers,but also on the particle dynamics that determine the mixingof the particles with the detonation products and thesurrounding air and the temperature history of eachparticle.Themultiphase fluid dynamic and chemical processes that

occur within the fireball from a metalized explosive arecomplex.Due to the high temperatures and pressures withinthe combustion products, direct in situ measurement of thethermodynamic properties of the multiphase productswithin the fireball is difficult. Hence, it is convenient touse nonintrusive optical techniques to record the radiantemissions from the fireball. In the present paper, bothspectrometry and time-resolved optical pyrometry are usedto record the emissions from heterogeneous explosivesconsisting of a packed bed of metallic particles saturatedwith a liquid explosive. The optical techniques provideinformation regarding the presenceof gaseous or condensedspecies in the fireball as well as the temperature of thecondensed species. Therefore, the optical techniques com-plement the use of other diagnostics such as high-speedphotography and in situ measurement of gas pressure andtemperature.Heterogeneous explosives consisting of a packed bed of

metallic particles saturated with a liquid explosive havebeen studied in many experimental investigations. Proper-ties that have been investigated include the detonationproperties of the explosive itself, such as the critical chargediameter, detonation velocity, detonation pressure, anddetonation temperature [1 – 7]. Previous studies of the blastwave from detonation of these heterogeneous explosives inthe free field have investigated the nonideal explosiveperformance characteristics, explosive particle dispersal,particle momentum effects, and the critical charge diameter* Corresponding author; e-mail: [email protected]

169Propellants, Explosives, Pyrotechnics 31, No. 3 (2006)

G 2006 WILEY-VCH Verlag GmbH&Co. KGaA, Weinheim

for rapid initiation of particle combustion [8 – 13].While theparticle dynamics and combustion immediately after thecharge detonation and in the near field have recently beenan intensive area of research, radiant emission measure-ments and analysis from themultiphase fireball also providea rich source of information on the particle dynamics andcombustion processes [14 – 15]. In an attempt to character-ize the local radiant emission and temperature history, Pahland Kaneshige [16] developed a local pyrometer probe torecord the light emission from a limited measurementvolume within the combustion products of an aluminizedexplosive. Their results show that the pyrometry techniquealone may not be sufficient to quantitatively describe thealuminum combustion within the combustion fireball.Thepresent paper describes the results of an experimental

study of the radiant emission from fireballs generated fromthe detonation of homogeneous explosives including trini-trotoluene (TNT) and sensitized nitromethane (NM)aswellas heterogeneous explosives (sensitized NM with metallicparticles). Time-integrated spectrometry is first used torecord the emission spectrum from the fireball to determinethe nature of the spectrum and location of any atomic ormolecular emission lines. The spectra are then used toprovide guidance in choosing the appropriate spectral filtersfor use with two separate three-color optical pyrometers.The pyrometers are used to infer the time history of thefireball temperature from both homogeneous and hetero-geneous explosives.

1.1 Optical Pyrometry

The determination of detonation-associated tempera-tures by optical means dates back to the late 1950Js and thework of Gibson et al. [17]. More recently Baudin and co-workers [5, 18] and Gogulya et al. [19] used six- and two-color pyrometers, respectively, to optically determine thetemperature of a condensedmaterialJs detonation products.Ogura et al. [20] used two four-color pyrometers operatingin the near-infrared region (wavelengths ranging from 600 –2000 nm) to study temperature evolution of explosionfireballs by recording the radiant emission from 1 – 100 kgTNT charges. Optical pyrometry uses continuous spectraemitted by the condensed reaction products to determinetemperature. Themajority of the condensed products of thedetonation are usually gray emitters (i.e. the emissivity is aweak function ofwavelength) over a relatively narrow rangeof wavelengths. For a gray body, the emission spectrum is acontinuous curve that depends only on temperature andfollows PlanckJs law. Thus, measuring the ratio of radiantfluxes at two differentwavelengths is sufficient to determinethe temperature of the gray emitter. This feature forms thebasis for two-color pyrometry. Bymeasuring the radiance atmore than two wavelengths, it is possible to estimate thedependence of emissivity on wavelength (assuming it is arelatively smooth function of the wavelength) and alsodetermine the temperature of non-gray emitters. However,instrumentation system errors have been shown to increase

strongly as the number of spectral measurements increase.As a result, it is of limited use to record the radiance at morethan three spectral locations [21].The current work uses three-color pyrometry supple-

mented by prior investigation of the emission spectra. Thewavelengths are chosen so as to not coincide with theemission or absorption spectra of the gaseous productspecies. The acquisition of the spectral intensities at threedifferent narrow-wavelength bands allows verification ofthe assumption of the gray spectra for continuous radiation.If the dependence of emissivity onwavelength is sufficientlyweak and the gray body approximation is valid, then thetemperatures calculated based on each of the three binaryemissivity ratios should be of similar magnitude. Thisprocedure is demonstrated in the results section below(see Figure 10) for a charge containing nitromethane andmagnesium particles. In fact, the two-color temperaturesgenerated using the three different wavelength ratioscoincide within the scatter of the measurements (�100 K).The error in the temperature measurements from thecalibration procedure is estimated also to be �100 K, andhence the gray body assumption is confirmed, within anaccuracy of this amount. Interpreting the pyrometer signalsfor both homogeneous and heterogeneous explosive fire-balls is difficult due to the complexity of the thermo/fluiddynamics within the fireball. The fireball is a multi-temper-ature medium, even without combustible particles, as theproducts of the oxygen-deficient explosive start to mix withthe surrounding air forming diffusive combustion fronts. Ifvolatilemetal particles are present, microflames formwith acombustion temperature at the particle surface or in a zoneclose to the particle surface, which is considerably higherthan the temperature of the surrounding gases (by severalhundred or even a thousand K). Emission from the gasspecies within the high-temperature microflames generatesstrong atomic emission lines and molecular bands with anintensity that may surpass that of the condensed productsover narrow wavelength regions. For particles that undergoheterogeneous combustion, the particle surface temper-ature is usually higher than the temperature of thesurrounding gas, and strong absorption lines may appearsuperimposed on the continuous spectrum. Erroneousresults and conclusions may be derived if one (or more) ofthe pyrometerJs working wavelengths coincides with suchemission or absorption lines of the gaseous spectra.To assist in the choice of the working wavelengths of the

pyrometer, the time integrated spectra of the fireballs werefirst obtained both for the pure explosives (TNT, NM) andexplosives with metal additives (NMþMg, Al, or Zr). Thistechnique is described in the following section.

2 Experimental

2.1 Time-Integrated Emission Spectrometer

The spectrometric study of the explosion fireballs wascarried out using a miniature fiber-optic spectrometer

170 S. Goroshin, D. L. Frost, J. Levine, A. Yoshinaka, F. Zhang

Prop., Explos., Pyrotech. 31, No. 3, 169 – 181 www.pep.wiley-vch.de G 2006 WILEY-VCH Verlag GmbH&Co. KGaA, Weinheim

www.pep.wiley-vch.de

(USB-2000, OceanOptics Inc.), which acquired spectra inthe wavelength range of l¼ 350 – 1000 nm using a 2048-element linear CCD array. The image of the explosionfireball was obtained using a telephoto lens and a conven-tional 35-mm SLR camera positioned about 60 m from theexplosive site, which is shown schematically in Figure 1. Thelight from the central part of the image was transmitted tothe spectrometer in the protective bunker via a 50 mmdiameter, 5-meter-long single-strand optical fiberwhichwasinserted through a hole directly onto the film plane of thecamera. The fiber aperture also played the role of thespectrometer entrance slit. The estimated spectral resolu-tion of the spectrometer was about 1 nm. The factorypresetintegration time in the external triggering mode was about50 ms. In order to avoid collection of the intense lightemitted by the detonation itself, a delay time of 1 ms wasintroduced into the triggering circuit of the spectrometer. Toaccount for the nonuniform spectral sensitivity of the CCDsensor, the sensor was calibrated using a tungsten halogenlight source (LS-1-CAL, OceanOptics Inc.) with a knowndistribution of spectral emissivity.

2.2 Three-Color Optical Pyrometer

Figure 2 shows a schematic and a photograph of the three-color pyrometry system used. As with the spectrometer, a35-mm camera was used to collect light from the radiantevent and focus it on an image plane. An optical fibermounted at the film plane transmitted a spatially resolvedportion of the image light to a photomultiplier block. At the

typical distance of acquisition of about 60 mand diameter ofthe fiber of about 2 mm, the spatial field of view of thesystem corresponded to a circular region with a diameter ofabout 70 cm. The light entered the pyrometer housingthrough an iris, was split twice and passed through threeseparate narrow-band pass filters, each with an effectivebandwidth of about 10 nm. The signals generated with thefast-response photomultiplier tubes were amplified andrecorded using a digital oscilloscope. Analog photomulti-plier detection assemblies used in the presentwork integratea photomultiplier, high-voltage power supply and a fast-response amplifier in one housing (Electron Tubes Inc.,UK). The frequency bandwidth of the detection assemblywas about 25 MHz and the range of the spectral responsewas from 300 to 800 nm.Since radiant emissions from a source are a strong

function of temperature (~T4) and the temperaturesgenerated during the explosion event may span the rangeof 1500 to 6000 K, it is convenient to use two pyrometerswith different sensitivities, in order to increase the dynamicrange of themeasurements. Thus pyrometerAwas setwith ahigher gain (i.e. higher sensitivity) for longer data-capturingtimes that typically corresponded to the measurement oftemperatures below 3000 K, whereas pyrometer B was setwith a lower sensitivity to be able to resolve temperatures upto 6000 K at early times after detonation of the charge. Thepyrometers were calibrated using a factory standardizedtungsten strip lamp (The Pyrometer Instrument Company,NJ) and precision power supply. Three calibration curveswere generated for each pyrometer, using the ratios ofradiation from the red and green filters (R/G), the red andblue filters (R/B), and the green and blue filters (G/B),respectively.

2.3 Experimental Procedure

Liquid nitromethane (NM) experiments were carried outusing 12.3 cm diameter spherical glass-cased charges, con-taining 1,000 g of NM sensitized with 10% by weighttriethylamine (TEA). Globe light bulb casings (with thefilament removed) were used for the 12.3 cm (and smaller)charges, whereas larger charges (5 and 12 liters) utilizedspherical glass boiling flasks. A small booster charge,comprised of about 10 g of C4, was placed in a small glasssphere blown on the end of a glass tube and placed in thecenter of the charge. Charges were mounted on the end of awooden post approximately 150 cm above the ground. TheTNT trials utilized 3.6 kg, 5 cm� 2.5 cm�20 cm rectangularblocks of TNT similarly mounted. Heterogeneous chargeswere prepared using identical spherical charges and filledwith a packed bed of metallic particles saturated withsensitized NM. Spherical atomized magnesium particles(Reade Manufacturing Co.) were tested with four differentparticle diameters, i.e. 60� 15 mm, 85� 10 mm, 240� 60 mm,and 520� 90 mm. Themass fraction of themagnesium in thecharges was 73� 1%. Spherical aluminum powder wasobtained from Valimet Inc. (Stockton, CA) with average

Figure 1. Top view of the experimental test site at DRDC-Suffield.

Optical Pyrometry of Fireballs of Metalized Explosives 171

G 2006 WILEY-VCH Verlag GmbH&Co. KGaA, Weinheim www.pep.wiley-vch.de Prop., Explos., Pyrotech. 31, No. 3, 169 – 181


particle sizes ranging from 2 – 115 mm. For the largerparticles (i.e. particles designated H-50 and H-95), themass fraction of aluminum was 79� 1%, while the massfraction decreased for smaller particles (e.g. 78� 1% for H-10 powder and 73� 1% for H-5 powder). From the particlesize analysis provided by Valimet for the batch of powder

used, the average particle size was 56� 21 mm for H-50 and114� 40 mmforH-95.Zirconiumpowderwas obtained fromAtlanticEquipmentEngineers (Bergenfield,NJ) and sievedto the size range 600 – 850 mm. The mass fraction ofzirconium in the charges was about 76� 1%. Two PhantomVII high-speed video cameras (operated between 25,000 –

Figure 2. Schematic of the three-color pyrometer system and a photograph of the pyrometer with the case open. In the photograph, thelight enters from the bottom and is split and directed into the three photomultiplier tubes.




40,000 frames/s) were used to view the explosive dispersalevent. Further experimental details can be found in [10].

3 Results and Discussion

3.1 Fireball Spectra

Emission spectra were recorded (integrated over 50 ms)for each of the types of charges tested. It was found that thegeneral features of the spectra were quite reproducible anddependedonly on the chemical composition of the explosiveand were independent of the charge size and shape (bothspherical and cylindrical charges were used). Figure 3 showstypical time-integrated spectra of the fireballs from variouscharges, including TNT, pure sensitized nitromethane andpacked beds of Al, Mg, and Zr powder saturated withsensitized nitromethane.As can be seen fromFigure 3, in the spectral regionwhere

pyrometry was performed (400 – 800 nm), fireballs fromcharges containing TNT or sensitized nitromethane aloneexhibit nearly grey spectra. Only the prominent sodiumdoublet lines are visible and protrude about 30% inamplitude above the continuum radiation (the individuallines at 589 nm and 590 nm are unresolved and appear as asingle peak). In contrast, charges containing NM withmagnesium particles produce a particularly rich spectrumwith a wide variety of Mg atomic lines and MgO molecularbands. This is consistent with the mechanism of magnesiumcombustion, which is known to burn in the vapor phase dueto the very low boiling point of the metal (1373 K) incomparison to the adiabatic flame temperature (~3090 K).The temperature of the gaseous species including Mg andMgO in the microflame zones, which envelop each particle,is higher than the temperature of the condensed MgOformed in the periphery of the same flame zones [22]. Thus,the intensity of the gaseous lines and bands in themagnesium combustion spectrum considerably exceedsthe intensity of the background continuum radiation fromthe condensed products and only narrow regions betweenthe lines are available for continuous spectrum pyrometry.Unlikemagnesium, the spectra of theNM-Zr fireballs are

continuous and free of atomic and molecular lines. Thisagain is consistent with the classification of zirconiumcombustion by Glassman [23]. Since the boiling temper-ature of both the metal and zirconium oxide considerablyexceeds the adiabatic flame temperature, zirconium isknown to burn in air via a gas-solid surface reaction withoutthe formation of a gaseous flame. Thus, the temperature ofthe zirconium particle is higher than the temperature of thesurrounding gas, which leads to the appearance of theabsorption lines of some gaseous species (Na).

Figure 3. Time-integrated spectra obtained from the radiationfrom fireballs for various homogeneous and heterogeneouscharges. Figures (from top to bottom) correspond to chargescontaining TNT, NM, NM with Mg particles, NM with Alparticles, and NM with Zr particles, respectively.




In accordance with the classification of metal combustion[23], aluminum is placed somewhere between heterogene-ously burning zirconium and vapor-phase burning volatilemagnesium. The boiling point of aluminum is quite high(~2792 K), and is relatively close to the adiabatic flametemperature in air (~3540K). In addition, aluminum oxide(Al2O3) does not exist in the vapor phase. Thus it is generallyaccepted that aluminum burns through a series of steps inwhichgaseousAlO is akey intermediate that forms very closeto the particle surface. It is quickly oxidized to other productssuch as AlO2 and finally forms the condensed product Al2O3

through the process of chemical condensation. A multi-stepreaction mechanism leads to a wider reaction zone incomparisonwithmagnesium,with a less prominentmaximumin gas flame temperature. Thus, in aluminum flames, themaximum temperature of the gaseous species in the micro-flames surrounding the particles is closer to the temperatureof the condensed phase material and only AlO molecularbands (B2Sþ!X2Sþ) are clearly visible in the spectra.Based on the analysis of the spectra, the three spectral

regions for the color temperature measurements of thecontinuum radiation were chosen to avoid gaseous atomiclines andmolecular bands in all explosive compositions usedin the current work, and were centered at 450 nm (violetblue), 568 nm (green) and 690 nm (red). For some tests withthe homogeneous explosives, a 488 nm (blue) filter was usedin place of the violet filter.

3.2 Optical Pyrometry of Homogeneous ExplosiveFireballs

Figure 4 shows the light intensity recorded for twowavelength bands together with the inferred temperaturefor a charge containing only sensitized nitromethane(pyrometer B, short time scale). For voltages greater than1.6 V, the photomultiplier amplifiers become saturated andhence during this time (20 – 45 ms) the temperature datahave nomeaning. At early times (t< 20 ms), an initial rise intemperature is clearly visible, reaching at least 5500 K(corresponding to the temperature of the air shocked by theintense blast wave emerging from the surface of the charge).Since the liquid is transparent, it is possible to observe thesteady propagation of the detonation through the liquid as itpropagates outwards from the center of the charge. Thetimescale of the data was expanded to focus on thedetonation front within the liquid explosive, which is shownin Figure 5. This plot shows an approximately linear regionat 3600� 200 K. This value is slightly higher than thatcalculated using the CHEETAH 2.0 code [24] for asensitized NM mixture (using the equation of state libraryBKWC), which gave aC-J-detonation temperature of about3228 K.Thedetonation temperature observed for sensitizedNM is comparable to the results of Leal et al. [18], whomeasured detonation temperatures for pure (or “neat”)NMof ~3500� 200 K.The behavior of the radiant emissions from the fireball

over a longer timescale is shown in Figure 6. The variation in

the radiant intensity with time illustrates three particularevents during the detonation of sensitized NM and expan-sionof the fireball. First, the large initial spike represents thesum of the light from the detonation front, the shocked airand the hot combustion products. As the combustionproducts expand and cool, the radiant intensity dropsrapidly after about 0.5 ms. Following that, a smaller increasein light intensity after several milliseconds is due to thereflected shock from the ground that induces mixing andintensifies the light emission. Finally, the large increase inthe radiant intensity from 10 – 20 ms is due to afterburningof the fuel-rich products with the surrounding air. Assensitized NM is an oxygen-deficient explosive, the mixingof the combustible products with the surrounding air leadsto considerable afterburningmanifested by the formation ofdiffusive flamelets at the interface of the combustionproducts with the air. Even though the emission intensityvaries by several orders of magnitude in accordance withchanges in the diffusive flame surface area, the temperatureof the individual diffusive flame fronts remains essentially

Figure 4. Pyrometer radiant intensity signals and inferred tem-perature for a charge containing 1,000 g of NM sensitized with10% triethlyamine.

Figure 5. Radiant intensity and inferred temperature for sensi-tized NM, showing early times corresponding to the motion of thedetonation front within the liquid.




constant and close to the adiabatic flame temperature of thenitromethane detonation products and air. As can be seenfrom Figure 6, the temperature of the condensed productsremains approximately constant at a value of 1900 K�150 K during the period 10 – 80 ms after detonation of thecharge.The events described above are consistent with the high-

speed video records. In particular, Figure 7 shows framesfrom the high-speed video record for a trial with 1,000 g ofsensitized NM alone. The intense luminosity from theshock-heated air as well as the combustion products isevident in the first few hundred microseconds. At the timet¼ 1 ms, the blast wave is clearly visible just outside thecombustion product interface. At t¼ 2.5 ms, the initial blastwave is just visible at the periphery of the field of view. Atthis time, the shock wave reflected from the ground hasmoved back into the combustion product cloud and inducedthe luminosity visible near the middle of the photograph.The afterburning intensity continues to increase as theadditional mixing with the surrounding air occurs on atimescale of tens of milliseconds.The measured flame temperature of the diffusive flame

fronts of the nitromethane detonation products and air canbe compared with an estimate made using the equilibriumthermo-chemical codes CHEETAH 2.0 [24] and CEA [25]to compute the following thermodynamic processes. First,CHEETAH 2.0 is used to compute the properties of theproducts following a CJ detonation, which are then allowedto expand adiabatically to atmospheric pressure. All theproducts at this state (only species with concentrationslarger than 10�10 mol/kg are considered) are then used asreactants for the code CEA, which are then allowed to reactwith a stoichiometric mixture of air at constant pressure.This procedure produced a predicted temperature of1967 K, which is within the range of the experimentalmeasurements of 1900� 150 K.TNT, unlike NM, is opaque. Therefore, no pyrometric

data can be derived regarding its detonation temperature.

However, the temperature of the fireball from the 3.6 kgTNT charges was found to be 2150� 150 K for t~10 – 50 ms(see Figure 8). During the time period 10 – 30 ms, theradiant intensity increases significantly due to afterburningof the combustion products, in a similar fashion as the NMcharges (cf. Figs. 6, 7). For the TNT charges, a maximumtemperature of ~5600 K is observed at very early times(microseconds after detonation of the charge), due to thehigh temperature of the shocked air immediately surround-ing the charge, followed by a gradual decrease to a temper-ature of 3800� 200 K after 300 ms (see Figure 9). Theseresults may be compared to those of Ogura et al. [20], whocarried out tests with larger (10 kg, 32 kg and 100 kg) TNTcharges. They observed maximum temperatures rangingfrom 7000 – 10000 K that later decayed to temperaturesranging from 3500 – 4000 K after 300 ms (from Figure 7 of[20]). For a 32 kg TNT charge, the inferred temperaturecontinued to decay to a minimum after about 5 ms andvaried from 1700 – 1900 K for several hundred milliseconds(Figure 10 of [20]). They also noted that the maximumtemperature results agreed well with the estimate of theshocked air temperature based on the blast strength [20].CHEETAH 2.0 [24] and CEA [25] analyses for the TNT

products (similar to that described earlier for the after-burning of NM combustion products) of the maximumtemperature in the diffusive flamelets produced an adia-batic flame temperature of 2139 K, which is again consistentwith the longer-timescale pyrometer results from thepresent investigation of 2150� 150 K.

3.3 Fireballs from Heterogeneous Explosives

Tests were carried out with spherical charges containingsensitized NM with magnesium, aluminum, or zirconiumparticles. Figure 10 shows the radiant intensity and inferredtemperature history for a 12.3 cm diameter charge contain-ing 240 mmmagnesiumparticles. The temperatures basedoneach of the three wavelength ratios are plotted simulta-neously. The scatter in the temperature values, from a givenwavelength ratio, is the same as the scatter in the temper-ature values between the different wavelength ratios, andhence, to this degree of accuracy, the gray body assumptionis confirmed.The radiant intensity rapidly increases over thefirst several hundredmicroseconds due to the emission fromthe nitromethane combustion products. During these earlytimes, the rapidly accelerated magnesium particles arepredicted to penetrate the combustion productsJ interfaceand first penetrate the leading blast wave (calculations ofthe particle trajectories are given in [9]). For this particlesize, the magnesium particles typically only ignite after adelay of about 1 – 4 ms (see [13]). However, for thisparticular experiment, a vertical plate was located 60 cmfrom the charge. The particles ignited upon impact with theplate (after about 400 ms), generating intense luminositythat saturated the photomultiplier amplifiers. After onemillisecond, the intensity decreases, with a small increase at1.5 ms due to the interaction with a reflected shock. Unlike

Figure 6. Pyrometer radiant emission signals and inferred tem-perature over longer times for a 12.3 cm diameter chargecontaining 1,000 g of sensitized NM.




for the homogeneous explosives, the radiant intensityremains roughly constant during the time 5 – 20 ms. Duringthis time, the inferred fireball temperature is about 2600�

100 K, i.e. about 700 K higher than that of the fireballfrom the homogeneous explosive charge. Two points canalso be noted: first, similar to TNT, the heterogeneous

Figure 7. High-speed video record for the detonation of a 12.3 cm diameter charge containing 1,000 g of sensitized nitromethane only.The vertical plate visible to the right of the charge is located 90 cm away from the charge location.




explosive is opaque and therefore zero time represents thepoint at which the detonation reaches the surface of thecharge. Secondly, the heterogeneous charge generatedradiant intensities several orders of magnitude larger thanthat of the homogeneous charges, hence very small cameraapertures were required to prevent saturation of the signals.As noted above, for charges containing sufficiently large

magnesium particles that are detonated in the free field, theparticles ignite after a certain delay time [13]. In this case,the pyrometer measurements can be used to estimate thedelay time.An example of delayed particle ignition is shownin Figure 11 for a 12.3 cm diameter charge containing240 mm diameter magnesium particles (the pyrometersignals corresponding to this trial are shown in Figure 12).After several hundred ms, the radiant emission from thefireball is decreasing and the fireball temperature reaches aminimum of ~1900 K. At that point, the intensity begins toincrease as the magnesium particles ignite near the top andbottom of the particle cloud and the corresponding fireballtemperature increases to 2500 – 2600 K, which is represen-

tative of the temperature of the burning magnesiumparticles. At 1 ms, the majority of the magnesium remainsunburnt and the radiant intensity is very low so that theinferred temperature data is very inaccurate at this time (seeFigure 12). In the next several milliseconds, a deflagrationfront propagates through the magnesium particle cloud.During this time, the overall intensity recorded by thepyrometer increases, but themaximum fireball temperatureremains approximately constant.For smaller magnesium particles (60 mm and 85 mm

diameter magnesium particles were also tested), the par-ticles ignited promptly within the fireball [13]. For a 12.3 cmdiameter charge containing 85 mm magnesium particles, ascan be seen from the high-speed video records, the fireballluminosity increases significantly after several hundredmicroseconds, indicative of metal particle combustion. Inthis case, the dip in inferred fireball temperature (seeFigure 12) is not observed and the temperature monotoni-cally decays to the asymptotic value of ~2600 K.Figure 13 shows the radiant emissions and inferred

temperature from a 12.3 cm diameter charge containingNM and H-95 aluminum powder. The ignition of theparticles occurs after several hundred microseconds due toparticle impact with a vertical plate in close proximity to thecharge. Following particle ignition, a deflagration movesthrough the particle cloud and results in significantly moreintensive light emission than that from the baseline homo-geneous explosives. Themost intense radiation occurs in thefirst 5 ms of the expansion of the fireball. Two intermediatepeaks occur at about 2 ms and 3.5 ms, which are due to theinteraction of a shock wave reflected from the vertical steelplate (1.83 m2) located 0.6 m away from the charge aswell asthe shock reflected from the ground (the height of thecharge is 1.5 m).During this period, the fireball temperaturedecreases fromabout 3000 K to a value that remains roughlyconstant at 2700� 200 K after 5 – 10 ms. Figure 14 shows theearly time behavior of the radiant intensity and temperaturefrom a fireball from a 5 liter charge containing H-50aluminum powder. In this trial, the distance from the charge

Figure 8. Radiant intensity and inferred temperature from fire-ball from a 3.6 kg TNT charge.

Figure 9. Early time behavior of the radiant emission andinferred temperature for a 3.6 kg TNT charge.

Figure 10. Radiant intensity and inferred fireball temperaturefor a 12.3 cm diameter charge containing 240 mm magnesiumparticles.




to the vertical steel plate was the same as the height of thecharge (1.5 m). In the first few tens of microseconds,temperatures up to 5500 K are observed due to the shockedair surrounding the charge. The radiant intensity increasesover the first several hundred microseconds as the fireballexpands, but the temperature remains largely constant after

100 ms at a valueof about 3000 K.From thehigh-speed videorecords for this trial, we observed that the luminosity of thefireball continues to increase during the first millisecond ofexpansion as a result of particle burning, enhanced presum-ably with mixing of the particles with the surroundingatmosphere.

Figure 11. High-speed video record of detonation of 12.3 cm diameter charge containing NM and 240 mm diameter magnesiumparticles. Majority of the particles are dispersed without igniting and then a deflagration propagates through the particle cloud on amillisecond timescale. The charge height is 2 m and the checkerboard plate is located more than 5 m behind the charge.




The experiments with charges containing zirconiumpowder with relatively large particle sizes (600 – 850 mm)exhibit prompt ignition of the particles, similar to thatobserved with the fine magnesium particles. The fireballsare intensely luminous at early times with the intensitydecreasing monotonically with time. Figure 15 shows thatthe fireball temperature reaches an asymptotic value of2700� 200 K after about 10 ms. The early time behavior offireball emissions for a 12.3 cm charge is shown in Figure 16.A peak temperature of 4800 K is observed, which decreasesto about 4000 K at the point when the radiant intensity is atmaximum (~300 ms). The inferred temperature continues toslowly decrease over the next few milliseconds as theparticles expand and penetrate the leading shock wave andcontinue to burn in the surrounding ambient air (thetrajectories of the burning particles are readily evident inthe video records).In each case with charges containing NM and metallic

particles, the asymptotic color temperature (observed onthe timescale of tens of ms) of the explosion fireballs was700 – 800 K higher than for pure nitromethane charges. This

Figure 12. Early time behavior of radiant intensity and inferredtemperature for a 12.3 cm diameter charge containing 240 mmmagnesium particles corresponding to trial shown in Figure 11.

Figure 13. Radiant intensity and inferred temperature for a12.3 cm diameter charge containing H-95 aluminum particles.

Figure 14. Early time behavior of radiant intensity and inferredtemperature for a 5 liter charge containing H-50 aluminumparticles.

Figure 15. Radiant intensity and inferred temperature from a12.3 cm diameter charge containing 600 – 850 mm zirconium par-ticles.

Figure 16. Early time behavior of radiant intensity and inferredtemperature for a 12.3 cm diameter charge containing zirconiumparticles (same trial as in Figure 15).




clearly indicates that the recorded temperatures correspondto individual burning metal particles rather than the after-burning soot from the NM combustion products. In theabsence of heat losses, the maximum temperature in theindividual burning particles should be close to the stoichio-metric flame temperature. Using the equilibrium codeTHERMO (developed by Dr. A. Shiryaev at the Instituteof Structural Macrokinetics, Moscow, in 1994) to computethe stoichiometric constant pressure combustion temper-ature in air gives 3090 K, 3541 K, and 3859 K for magne-sium, aluminum and zirconium particle combustion, re-spectively. The temperatures measured in the presentinvestigation are somewhat lower than these equilibriumpredictions. However, the temperatures from the presentinvestigation correspond closely with the values measuredby other investigators for individual particles burning in air[22, 26 – 27].Florko et al. [22] measured temperatures of 2600� 200 K

for burning magnesium particles (0.5 – 2 mm diameter)using the two-color pyrometer method, which is consistentwith the temperatures recorded for charges containingmagnesiumparticles in the present investigation (i.e. 2600�100 K).For explosively dispersed aluminumparticles, the temper-

ature history observed (i.e. from Figure 13, T~3000 K forthe first millisecond followed by a decrease in temperatureto ~2700 K after 5 ms) closely resembles the behaviorobserved by Dreizin [26] in an experimental study of thecombustion of single aluminum particles (85 – 190 mmdiameter). He found that the aluminum combustion pro-ceeded in three distinct stages. In the first stage, temper-atures of 2900 – 3000 K were observed, which lie betweenthe boiling points of aluminum (2792 K) and Al2O3

(3253 K). Hence in this stage some superheating or evap-oration of the aluminumdroplets occurs. In the second stageof combustion, which occurred tens of milliseconds later, herecorded intense fluctuating luminosity with inferred tem-peratures in the range 2700� 100 K. He speculated that therelative motion of the parent droplet and the surroundingcloud of condensed Al2O3 particles contributed to theintensity oscillations and that the temperature was anaverage of the temperatures of the aluminum droplets andAl2O3 particle cloud. A third combustion region wasobserved at late times, in which the average temperaturedecreased to a value close to the Al2O3 melting point(2323 K) at which point the combustion terminated.Finally, the late-time temperatures observed in the

present investigation for the fireballs from charges contain-ing zirconium particles (i.e. 2700� 200 K) were comparableto the measured temperatures of ~2500K recorded byMolodetsky et al. [27] in experiments with single burningzirconium particles.

4 Conclusions

The fireball temperature histories corresponding to thedetonation of homogeneous charges (TNT and sensitized

NM) and heterogeneous charges (NM/Mg, NM/Al, andNM/Zr) have been experimentally determined using three-color optical pyrometry assisted byqualitative spectroscopy.For pure sensitized nitromethane, optical measurementsallowed the determination of i) the detonation temperatureof the nitromethane microseconds after detonation of thecharge, ii) the temperature of the ionized air shocked by theblast wave leaving the detonation products and iii) thetemperature of the diffusive flamelets forming at themixinginterface of the combustion products and air during theafterburning that occurs on a millisecond timescale. Theobtained asymptotic temperature values agreed well withthermodynamic estimations and available literature data.For metalized explosives, asymptotic temperature values

were reached (after tens of milliseconds) that were consis-tently 700 – 800 K higher than the corresponding temper-atures from fireballs from the baseline homogeneousexplosives. This temperature difference was attributed tothe higher temperatures associated with the burning met-allic particles. The inferred temperatures were less thancalculated adiabatic flame temperatures for metal combus-tion, but were consistent with previous experimental datafor temperature measurements of single burning metalparticles in air.In some cases, explosively dispersed metallic particles

ignite only after some delay time. In this case, opticalpyrometry can be used to estimate the ignition delay time ofthe metal particles. For example, for charges containing240 mm diameter magnesium particles, the particle ignitionoccurred after a delay typically on the order of millisecondsand clearly coincided with a sudden increase in fireballtemperature. As a deflagration front propagated throughthe dispersed particle cloud, the overall radiant intensityincreased, but the inferred temperature remained largelyconstant. If the ignition delay was less than about 100 ms,then the contribution of the particle combustion to theemitted radiation could not be discerned since the radiantemission at early times is dominated by the high-temper-ature shocked air around the charge. For the charges withthe other particles tested, i.e. 60 mm and 85 mm diametermagnesium particles, various aluminum particles with sizesranging from 2 – 115 mm diameter and zirconium particleswith a size range of 600 – 850 mmdiameter, no ignition delaywas observed and from the high-speed video records, theparticles appeared to be burning at the instant when theyfirst emerge from the combustion products. While themeasurements in the current paper were based on thecollection of light at a distance with a large spatial field ofview, pyrometry utilizing a localized light collection methodmay improve the resolutionof the ignition delay of themetalparticle combustion.

5 References

[1] R. Kh. Kurbangalina, Critical Diameter of Liquid Explosivesas a Function of Powder Content, Zhurnal PrikladnoiMekhaniki I Tekhnicheskoi Fiziki 1969, 10(4), 133.




[2] J. J. Lee, D. L. Frost, J. H. S. Lee, A. Dremin, Propagation ofNitromethane Detonation in Porous Media, Shock Waves1995, 5, 115.

[3] J. J. Lee, M. Brouillette, A. Dremin, J. H. S. Lee, D. L. Frost,Effect of Diethylenetriamine Sensitization on Detonation ofNitromethane in Porous Media, Combust. Flame 1995, 100,292.

[4] P. J. Haskins, M. D. Cook, R. I. Briggs, The Effect ofAdditives on the Detonation Characteristics of a LiquidExplosive, 12th Conference of the American Physical SocietyTopical Group on Shock Compression of Condensed Matter,Atlanta, Georgia (USA), June 24 – 29, 2001, AIP ConferenceProceedings 620, pp. 890.

[5] G. Baudin, A. Lefrancois, D. Bergues, J. Bigot, Y. Champion,Combustion of Nanophase Aluminum in the DetonationProducts of Nitromethane, 11th Symposium (International)on Detonation, Snowmass Village, Colorado, August 31 –September 4, 1998, p. 989.

[6] Y. Kato, C. Brochet, Detonation Temperature of Nitro-methane Aluminum Gels, in: J. R. Bowen, N. Manson, A. K.Oppenheim, R. I. Soloukhin, (Eds.) Dynamics of ShockWaves, Explosions, and Detonations, Progress in Astronauticsand Aeronautics Series, 94, AIAA Press, New York, 1985,p. 416.

[7] F. Zhang, A. Yoshinaka, S. Murray, A. Higgins, ShockInitiation and Detonability of Isopropyl Nitrate, 12th Sympo-sium (International) on Detonation, San Diego, California,August 11 – 16, 2002.

[8] D. L. Frost, H. Kleine, M. Slanik, A. J. Higgins, S. McCahan,F. Zhang, S. B. Murray, Blast Waves from HeterogeneousExplosives, 22nd International Symposium on Shock Waves,London, UK, 182 – 3 July, 1999, p. 947.

[9] D. L. Frost, F. Zhang, Effect of Scale on the Blast WaveGenerated by a Metallized Heterogeneous Explosive, Ener-getic Materials: Reactions of Propellants, Explosives andPyrotechnics, 34th Int. Annual Conference of ICT, Karlsruhe,Germany, June 24 – 27, 2003, p. 7/1.

[10] F. Zhang, D. L. Frost, P. Thibault, S. B. Murray, ExplosiveDispersal of Solid Particles, Shock Waves 2001, 10, 431.

[11] D. L. Frost, F. Zhang, S. McCahan, A. Higgins, S. Murray, M.Slanik, M. Casas-Cordero, C. Orthanalai, Near-Field ImpulseEffects from Detonation of Heterogeneous Explosives, 12thConference of the American Physical Society Topical Groupon Shock Compression of Condensed Matter, Atlanta, Geor-gia (USA), June 24 – 29, 2001, AIP Conference Proceedings620, p. 946.

[12] G. E. B. Tan, T. K. Lam, Momentum Loads from CylindricalExplosive Charges with Different Casing Thicknesses, 17thInternational Symposium on the Military Aspects of Blast andShock (MABS 17), Las Vegas, Nevada, 10 – 14 June 2002.

[13] D. L. Frost, F. Zhang, S. B. Murray, S. McCahan, CriticalConditions for Ignition of Metal Particles in a CondensedExplosive, 12th Symposium (International) on Detonation,San Diego, California, August 11 – 16, 2002.

[14] S. Goroshin, D. L. Frost, J. Levine, F. Zhang, OpticalPyrometry of Fireballs of Metalized Explosives, 18th Interna-tional Symposium on the Military Aspects of Blast and Shock(MABS 18), Bad Reichenhall, Germany, 27 Sept.–1 Oct.,2004.

[15] A. Yoshinaka, F. Zhang, J. Anderson, L. Legare, Near-FieldReflected Temperature in Fireballs of Heterogeneous Ex-plosives, 18th International Symposium on the MilitaryAspects of Blast and Shock (MABS 18), Bad Reichenhall,Germany, 27 Sept.–1 Oct., 2004.

[16] R. J. Pahl, M. J. Kaneshige, Post-Detonation AluminumParticle Temperature Measurement for Thermobaric Explo-sives using Two-Color Pyrometry, 31st International Pyro-technics Seminar, Fort Collins, Colorado, 11 – 16 July, 2004.

[17] F. Gibson, M. Bowser, C. Summers, F. Scott, C. Mason, Use ofElectro-optical Method to Determine Detonation Temper-atures in High Explosives, J. Appl. Phys. 1958, 29(4), 628.

[18] B. Leal, G. Baudin, J. C. Goutelle, H.-N. Presles, An OpticalPyrometer for Time Resolved Temperature Measurements ina Detonation Wave, 11th Symposium (International) onDetonation, Snowmass Village, Colorado, August 31 – Sep-tember 4, 1998, 353.

[19] M. F. Gogulya, A. Y. Dolgoborodov, M. A. Brazhikov, Inves-tigation of Shock and Detonation Waves by Optical Pyro-metry, Int. J. Impact Eng. 1999, 23, 283.

[20] T. Ogura, K. Okada, T. Abe, K. Wakabayashi, K. Ishikawa, E.Kuroda, T. Matsumura, Y. Nakayama, M. Yoshida, PyrometryStudy on Fireballs Generated upon the Explosion of TNT,Energetic Materials: Reactions of Propellants, Explosives andPyrotechnics, 34th Int. Annual Conference of ICT, Karlsruhe,Germany, 24 – 27 June, 2003, p. 18/1.

[21] V. N. Snopko, Methods of Optical Polychromatic Pyrometry,Teplofizika Vysokikh Temperatur 1987, 25(5), 980.

[22] A. V. Florko, A. N. Zolotko, N. V. Kaminskaya, V. G. Shev-chuk, Spectral Investigation of the Combustion of Magnesi-um Particles, Fiz. Goreniya Vzryva 1982, 18(1), 17.

[23] I. Glassman, Combustion, 3rd Ed., Academic Press, SanDiego, 1996, p. 436.

[24] L. E. Fried, CHEETAH 2.0 User8s Manual, Lawrence Liver-more National Laboratory, Energetic Materials Center, 2000.

[25] B. McBride, S. Gordon, Chemical Equilibrium Analysis,NASA–GLENN Research Center, Cleveland, Ohio, 2000.

[26] E. L. Dreizin, Experimental Study of Stages of AluminumParticle Combustion in Air, Combust. Flame 1996, 105, 541.

[27] I. E. Molodetsky, E. L. Dreizin, C. K. Law, Evolution ofParticle Temperature and Internal Composition for Zirconi-um Burning in Air, 26th International Symposium onCombustion, Naples, Italy, 1996, p. 1919.

Acknowledgements

The authors would like to thank C. Ornthanalai, S.Janidlo, J. Pryszlak, S. Jacobs, V. Tanguay, M. Cairns, and F.Jouot, for their invaluable assistance in the performance ofthe experimental trials. The authors also gratefully acknowl-edge the field support of A. Nickel, assistance with dataacquisition instrumentation from D. Boechler, high-speedphotography support from R. Lynde, and S. Trebble as wellas theMunitionTeam for their contributions during the fieldtrials at theDefenceR&DCanada – SuffieldMultiburst testsite. The authors also acknowledge the useful comments onthe manuscript provided by A. Higgins. This work wasfunded partially by the US Technical Support WorkingGroup.

(Received: March 26, 2005; Ms 2005/111)




Samuel Goroshin et al- Optical Pyrometry of Fireballs of Metalized Explosives

Documents

speed video

vertical steel

18th international

adiabatic

gray body

speed video

heterogeneous

metallic particles