This article was downloaded by: [University of Tennessee, Knoxville] On: 07 November 2012, At: 17:12 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Combustion Science and Technology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/gcst20 Flame-spreading Process over Thin Aluminum Sheets in Oxygen-enriched Environments a a a C. L. YEH , D. K. JOHNSON , K. K. KUO & M. M. MENCH a a Department of Mechanical Engineering, The Pennsylvania State University, 234 Research Building East, University Park, PA, 16802, U.S.A Version of record first published: 20 Jan 2011. To cite this article: C. L. YEH, D. K. JOHNSON, K. K. KUO & M. M. MENCH (1998): Flame-spreading Process over Thin Aluminum Sheets in Oxygen-enriched Environments, Combustion Science and Technology, 137:1-6, 195-216 To link to this article: http://dx.doi.org/10.1080/00102209808952051 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. © Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 Combwt. Sci. and Tech.. 1998. VoL 137. pp. 195-216 Reprints available directly from the publisher Photocopying permitted by license only 1998 OPA (Overseas Publishers Association) N.V. Published by license under the Gordon and Breach Science Publishers imprint. Printed in Malaysia. Flame-spreading Process over Thin Aluminum Sheets in Oxygen-enriched Environments C. L. YEH*, D. K. JOHNSON, K. K. KUO and M. M. MENCH Department of Mechanical Engineering, The Pennsylvania State University, 234 Research Building East, University Park, PA 16802, U.S.A. (Received 14 January 199B; In final/arm 24 April 199B) An experimental study of flame-spreading process over thin aluminum (99 % Al and 1 % Mn) sheets was investigated in oxygen-enriched environments. The objective of this study was to determine the dependency of flame-spreading rate over aluminum sheets as a function of initial chamber pressure, sample thickness, oxygen purity, oxygen flow condition, and sample orientation. The reaction mechanism of aluminum in oxygen was also studied by examining the recovered partially-burned sample using a scanning electron microscope (SEM) coupled with an energy dispersive spectrometer (EDS). The flame-spreading rate over aluminum sheets was measured by an array of fast-response lead-selenide (Pb-Se) IR photodetectors. The initial chamber pressure was varied from 0.1 to 6.3MPa. Two grades of oxygen gas were used with purities of 99.996 % and 99.75 %. In terms of the effect of pressure on the flame-spreading rate, as the initial chamber pressure was increased, the flame-spreading rate was found to increase to a maximum, decrease to a minimum, and then increase again. Based upon the comparison of flame-spreading rates in horizontal, upward, and downward orientation, the flame-spreading process over aluminum sheets was found to be dominated by the solid-phase heat conduction mechanism. The continuous oxygen flow showed a strong influence on the flame-spreading behavior, and it was demonstrated that the flame can be blown off when the counter-current flow velocity exceeds a critical value. The flame-spreading rates under high-purity (~99.996 %) oxygen environments were found to be significantly greater than those in commercial grade (~99. 75 %) oxygen. In addition, the oxygen content in the white ceramic-type nodules formed on the burned edge of the recovered partially-burned sample is much higher than that on the unburned surface. These imply that there exist heterogeneous reactions between aluminum and either oxygen or gaseous aluminum sub-oxides on the burning surface. Keywords: Aluminum; flame-spreading rate; heterogeneous reaction; photodetector • Corresponding author. e-mail: [email protected] 195 196 C. L. YEH et al. Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 INTRODUCTION Aluminum is a one of the most widely used materials. Much of the initial interest in aluminum combustion was associated with propulsion applications. Aluminum particles are commonly used as a fuel ingredient in solid propellants to increase the specific impulse and to suppress the oscillatory combustion instability in rocket motors. In recent years, high-surface-area structured aluminum pac kings have been used extensively as the cryogenic distillation columns in air separation plants because it can result in systems that consume less energy than conventional distillation tray technology (Dunbobbin et al., 1991). Among various aluminum alloys, the 3003 aluminum alloy (99 % AI and 1 % Mn) is the major material used in the cryogenic air separation plants for producing oxygen; therefore, the flammability of 3003 aluminum alloy in oxygen is of great interest to the society of compressed gas industry. Increased applications, both in industry and aerospace, have prompted the continuous study on the flame propagation over aluminum rods or sheets and fundamental combustion of isolated aluminum particles. In the flame-spreading study, the upward flame-spreading process along aluminum wires or rods (I, 2, and 3 mm in diameter) has been investigated by Kirschfeld (1961); Long and Sebald (1965); Sa to and Hirano (1986); Benz et al. (1986) and Sato (1989) in oxygen with pressures ranged from 0.1 to 69 MPa. A peculiar variation of the flame-spreading rate with initial oxygen pressures was observed; as the oxygen pressure was increased, the flamespreading rate was found to increase to a maximum, decrease to a minimum, and then increase again. However, this observation has not been fully explained. Concerning the combustion mechanism of aluminum, two contradicting theories have existed. Glassman (1959) offered a concept which predicts that aluminum combustion takes place in the vapor phase because the boiling point of aluminum oxide is greater than that of aluminum. Earlier studies (Drew et al., 1964; Drew, 1965; Prentice, 1965 and 1970) on ignition and combustion of aluminum particles (with diameters from 10 to 200 um) appear to support the vapor-phase burning mechanism. Several theoretical models (Brzustowski and Glassman, 1964; Law, 1973; Brooks and Beckstead, 1995) were developed to describe the vapor-phase combustion of aluminum particles. Steinberg et al. (1992), however, argued that Glassman's hypothesis (1959) is not suitable to justify the mode of metal combustion, since the metal oxide could decompose before reaching its boiling point. A f1ame- Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 FLAME SPREADING OVER ALUMINUM 197 spreading study by Sircar et al. (1991) using cylindrical aluminum rods in oxygen suggested that aluminum burns in both gas and liquid phases, according to measurement of the sample weight change during combustion. Recently, Dreizin (1996) based upon his experimental results and proposed a combustion mechanism of aluminum particles in air following three distinct stages: a vapor-phase combustion during the first stage, a second stage combustion with an increase of size and density of the smoke cloud when the dissolved oxygen content in the aluminum droplet reaches the limit needed for liquid Al z0 3 formation, and a third stage associated with oxide cape formation and growth in a non-symmetric manner. Dreizin (1996) believes that the suboxides (AIO and AIO z) formed in the gaseous reaction zone could diffuse to the particle, and then a fraction of these suboxides dissolves into the molten aluminum. The radial profile of gaseous AIO concentration was measured by Bucher et al. (1996) using PLIF during the burning of isolated aluminum particles in air. AlO was found at the particle surface and has been proposed to produce Al-O through the surface heterogeneous reaction with molten aluminum (Bucher et al., 1996). Above studies have shown that heterogeneous reactions should play an important role in the aluminum combustion. The objective of this research was to investigate the flame-spreading characteristics of thin aluminum sheets (3003 aluminum alloy with the composition of 99 % Al and 1 % Mn) in oxygen-enriched environments using a non-intrusive optical diagnostic technique. This investigation includes a study of the effects of initial chamber pressure, sample geometry, oxygen purity, oxygen flow condition, and sample oreintation on the flamespreading rate of aluminum sheets. The data of flame-spreading rates and threshold pressures obtained in this study represent the first experimental measurements for the 3003 aluminum alloy. In addition, a composition analysis was conducted with recovered partially-burned samples to facilitate a fundamental understanding of the combustion mechanism of aluminum. METHOD OF APPROACH Measurement of Flame-spreading Rate A schematic diagram of the experimental setup to measure the flamespreading rate over thin aluminum sheets is shown in Figure I. The experimental setup and diagnostic technique have been described in detail previously (Yeh et al., 1997a and 1997b). Briefly, the flame-spreading rate c. L. YEH ct 198 al. Combustion Product Exhaust OrificePlate Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 Stainless Steel Windowed Test Chamber ,,, __,""S""' " Mirrors t --,- ,, ,, --~-\: ,, ,, , , ,,, , , ,, ,, , ,, ,, , """',,*===F=="-'i*~=9P=~P=~~,*=":I : I --l-r- ..,. Test Celt :"",,,ss,,ss"SSSSS,,ss"SS"SSSSSssss"SS,,ss"SSSSss,,,,,Sj I Camrot Area High-Speed Semi-lransparent mirror ~ U FIGURE I Movie Camc:I11 CCDV'k. Camera Schematic diagram of experimental setup to study flame-spreading behavior over thin aluminum sheets. was deduced from the flame front trajectory measured by an array of leadselenide (Pb-Se) infrared (IR) photodetectors, which were mounted on the top wall of the test chamber. Typical IR photodetector responses and linear flame-front trajectories, indicating a constant velocity process, were also presented (Yeh et al., 1997a). The overall transient event of ignition and flame-spreading of aluminum sheets was recorded by both a CCO video camera and a high-speed movie camera (HYCAM). In addition, the pressure-time history during the flame-spreading event was recorded by three fast-response pressure transducers mounted in the upstream, midstream, and downstream of the test chamber. FLAME SPREADING OVER ALUMINUM 199 Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 Aluminum Sample and Igniter Aluminum samples used in this study were the 3003 aluminum alloy (99 % Al and I % Mn). Plain aluminum sheets have three different thicknesses of 0.15 mm, 0.20 mm, and 0.40 mm. The thickness of corrugated aluminum samples is 0.18 mm. The dimension of test samples was 25.4 mm wide and 432 mm long; the reason for using long sample strips is to accurately determine the flame-spreading rate. Pyrofuze'P, an aluminum-palladium composite wire, which generates tiny discreet particles through an intermetallic exothermic reaction when resistance hated, was used as the igniter (Yeh et al., 1997a). Test Conditions In this study, the initial chamber pressure was varied from O. I to 6.3 MPa. Selection of the pressure range was based upon the operation pressure in the cryogenic air separation plants. Two different grades of gaseous oxygen were used with purities of 99.75 % and 99.996 %. For stagnant tests (no continuous oxygen flow), flame-spreading rates in horizontal, upward, and downward orientations were measured in order to determine the dominant heat transfer mechanism. For flowing tests (with a continuous oxygen flow), two modes of flame spread were studied. One is the flame-spreading process occurring with the external oxygen gas flow moving in the same direction of flame propagation (i.e., concurrent flow). The other is external oxygen flow directed against the spreading flames (i.e., counter-current flow). In this study, external oxygen flows were set to have a velocity which varied from 5 to 13.5 tn]«. DISCUSSION OF RESULTS Recorded Flame-front Trajectory and Pressure Trace The flame-spreading rates (VFS) in this study were determined from measured flame-front trajectories by both the photodetector output and high-speed movie film. Figure 2 shows a plot of flame-front trajectories deduced from the photodetector output in different horizontal flamespreading tests. The linearity of the time derivative of trajectories indicates that the flame-spreading process can be treated as a constant-velocity event. The accuracy of the flame-spreading measurement can be demonstrated by a comparison of measured flame-front trajectories using photodetectors and 200 C. L. YEH et al. 500 --0-- I ------.- 400 Pressure Al Thickness 4.1 MPa 0.20 mm 99.75% 1.8 MPa O.J5mm 99.996% 5.1 MPa 0.40 mm 99.75% COX Purity Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 ><: '" ~ 1P... 300 .z8'" s: E 200 0 .I: ...u ~ is 100 o-t-~~-r-"-r-~~~---r~~-,-~--,-~~~-r-+- 0.5 1.0 1.5 2.0 2.5 Time (s) FIGURE 2 Typical flame-fronltrajectories and deduced flame-spreading rates in horizontal flame-spreading tests. high-speed movie pictures, as shown in Figure 3. In this particular test, the HYCAM framing rate was set as 10,000 pictures per second (pps). The pictures recorded by the high-speed camera can be found in Johnson (1997). It should be noted that the deduced flame-spreading rates by these two different methods are in very close agreement. A typical pressure-time trace recorded from a test with stagnant oxygen is shown in Figure 4. The observed increase of chamber pressure was the result of gaseous mass and heat generation due to the combustion of the aluminum sample. The decrease of pressure beyond the peak value at the end of combustion was caused by the cooling of combustion products by the ambient gases and chamber walls. Figure 5 shows a typical pressure-time trace obtained from a test which maintained a continuous oxygen flow. As can be seen in Figure 5, the chamber pressure reaches a maximum value at the completion of combustion, decreases to the initial value due to heat loss, and then drops rapidly after the oxygen supply is shut off. It is useful to note that measured pressure traces from three pressure transducers mounted in the upstream, midstream, and downstream of the chamber were almost Flame-Front Trajectories Determined/rom - - - HYCAM Picture; VfS = 464 rnm/s 400 - - Cl - • Photodetector: V FS = 466 mmls Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 300 200 100 o 200 400 600 800 1000 Time (ms) FIGURE 3 A comparison of flame-spreading rates deduced from high-speed camera films and photodetectors, 1000 Test #39: 0.15 mm AI Sheet, 99.996% Oxygen andInitial Chamber Pressure = 3.1 MPa(430 psig) 6.5 900 Completion of Combustion .ee, 6.0 / 'CO .;;; 800 ". 5.5 1f "1:! ~ ~ 5.0 700 ~ ~ " .c "3 [ -e il ~ ~ _"il 0.. E () 4.5 ." ~ -e 600 U 4.0 500 e 3.5 3.0 400 2.0 3.0 4.0 5.0 6.0 7.0 Time (5) FIGURE 4 A typical pressure-time trace recorded from a test with stagnant oxygen. 202 C. L. YEH et al. 8.0 Completion of Combustion 7.2 TeSl#51: 0.15 nun AI Sheer,99.75%Oxygen 1000 Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 Initial Pressure = 4.2 MPa(600 psig) GOX Flow Velocity =5.4 m/s 6.4 800 5.6 4.8 oe- 3 ~ ""Cl '" e '" !l 600 4.0 Onsetof Ignition 3.2 i::""-e .::. 400 2.4 200 +~~-,-.....-r-r-,-~-r--,-,~~""~--,--,,,,~--,--1 1.6 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Time (s) FIGURE 5 A typical pressure-time trace recorded from a test with flowing oxygen. identical, showing that there was no pressure wave generated during the flame-spreading event. Stagnant Tests Horizontal Flame Spreading Figure 6 shows the measured flame-spreading rate (VFS) versus initial chamber pressure (P;) for different samples in the horizontal orientation. As the initial pressure is increased, VFS increases to a maximum, decreases to a minimum, and then increases again. At a constant initial pressure, the V FS is higher for the thinner aluminum sheets. The threshold pressure for selfsustained flame propagation increases with increasing sample thickness. Limited data for the corrugated sheet (0.18 mm) were presented and the V FS of corrugated sheets were between those of 0.15 mm and 0.20 mm thick plain sheets. The observed reverse trend of VFS with Pi can be explained by three possible mechanisms. First, as Pi increases there is more accumulation of 203 FLAME SPREADING OVER ALUMINUM 700 Al Sample Thickness o 0.15 rnm (plainsheet) • 0.20 mm (plain sheet) Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 600 I 500 > ~ 400 ee 300 '"E 200 Oxygen Purity = 99.75% 0.40 mm (plain sheet) 0.18 mm (corrugated sheet) ~ '"~.. e0. .." ti: 100 o -j-,-.oro<>r~-,--p-p-r-~<J,---r-r<O;-T-J~~-,--r~-,,-~~-I 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Initial Chamber Pressure, Pi (MPa) FIGURE 6 Effects of initial chamber pressure and sample thickness on measured flamespreading rates of horizontal orientation. gaseous aluminum sub-oxides (e.g., AI20 and AlO) or small AI203 droplets near the flame front. These combustion products could form a diffusion barrier which reduces the collision frequency between aluminum vapor and oxygen, thereby resulting in a lower V FS' Second, a transition of the ratelimiting step occurs from one gas-phase reaction to another, which shows a lower reaction rate in spite of the higher pressure (Kirschfeld, 1961). Third, as initial pressure increases the size of the enveloping diffusion flame around the burned edge of the aluminum sheet decreases; therefore, it results in a smaller view factor of the flame by the unburned aluminum surface, causing a decrease in radiation heat fluxes and flame-spreading rates. As shown in Figure 6, further increases in pressure lead to a higher VFS' This increase is believed to be caused by the increase of convective heat-transfer flux from the flame zone to the unburned material. Since the gas density increases when pressure is higher, it leads to larger Reynolds and Nusselt numbers of the induced flow near the combustion zone. Therefore, the convective heat transfer coefficient is increased with the increase of the Nusselt number. In comparison with the previous study (Sato, 1989) using aluminum rods or wires, the flame-spreading rates of thin aluminum sheets were C. L. YEH et 01. Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 204 substantially higher than those of rods, even though a similar trend of flamespreading rate versus pressure was observed. This is believed to be caused by the larger surface area-to-volume ratio (A/V) of thin aluminum sheets than that of small rods or wires. For thin sheets, A/V"" 2/ I where I is the thickness of sheets. For rods or wires, A/V = 4/d where d is the diameter. Table I shows a comparison of measured flame-spreading rates between aluminum sheets and rods at initial chamber pressures of about 5 MPa of commercialgrade oxygen. The effect of oxygen purity on the flame-spreading rate is presented in Figure 7. For 0.15 mm thick aluminum sheets, results indicated that the threshold pressure (0.65 MPa) for self-sustained flame propagation in highpurity (99.96 %) oxygen is much lower than that (2.86 MPa) in commercial grade oxygen (99.75 % purity). The substantial increase of threshold pressure in commercial grade oxygen is because of the presence of argon (Ar), the major impurity in gaseous oxygen cylinders. The accumulation of argon near the combustion zone could create a diffusion barrier to the inward diffusion of oxygen, resulting in a higher threshold pressure. The strong influence caused by a small amount of Ar in oxygen on the flammability of aluminum was also observed by Benning et al. (1988). The difference in threshold pressures between 3003 (99 % Al and 1% Mn) and 6061 (99 % Al and I % Mg) aluminum alloys can be seen in Table II. Due to the different alloy composition, it was found that the threshold pressures of 3003 aluminum alloy are much higher than those of 6061 aluminum alloy in similar oxygen-purity environments. This result indicates that a slight difference in the alloy composition can lead to a significant difference in the flammability. Figure 7 also shows that, under the same initial chamber pressure, the flame-spreading rate measured in high purity oxygen (99.996 %) environments is substantially higher than that in oxygen with a purity of 99.75 %. TABLE I Comparison of measured flame-spreading rates between aluminum sheets and rods at initial chamber pressures of about 5 MPa Data source sample geometry The present study SOia, 1989 Sheet (25.4 mm wide and 432 mm long) Rod (120mm long) Sheet thickness (I, mm) or rod diameter (d, mm) 0.15 0.20 0.40 Surface area-tovolume ratio (A/V, rnm"') 13.33 10 5 Flame-spreading rate (VFS, mm/s) 441 360 236 2 3 4 2 1.33 68 60 48 FLAME SPREADING OVER ALUMINUM 205 Oxygen Purity • 99.996% Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 99.7S% o 800 600 400 200 AISampleThickness: 0.15 mm (Plain Sheet) o +r--,<>;l"«'r~-,-p;>r-~""~--r-r~~--,r~.,,~~+ 0.0 1.0 2.0 4.0 3.0 6.0 5.0 7.0 Initial Chamber Pressure, Pi (MPa) FIGURE 7 Effects of oxygen purity and initial chamber pressure on measured flame-spread- ing rates of horizontal orientation. TABLE II alloys Comparison of measured threshold pressures between 3003 and 6061 aluminum Data source Aluminum alloy Oxygen purity Threshold pressure (MPa) The present study I 3003 AI 99.75% 2.86 3003 AI 99.996% 0.65 Benning et al., 1988' 6061 AI 99.793 % 0.83 6061 AI 99.993% 0.21 1 The lest sample geometry is an aluminum sheet with a thickness of O. )5 mm, a width of 25.4 mm, and a length of 432mm. 2 The test sample geometry is an aluminum rod with a diameter of 6.4 mm, and a length of 76 mm. The flame-spreading rate changes by several tens of a percent, while the oxygen concentration is only changing by a couple thousandths of a percent. The high sensitivity of aluminum consumption rate to the oxygen purity indicates that heterogeneous reactions on the aluminum sample surface could play an important role in the combustion of aluminum. It is known that the mass consumption rate by heterogeneous reactions on the burning metal surface is directly proportional to the local oxygen concentration, which is relatively small (Glassman and Law, 1991). Any small, or even C. L. YEH et al. 206 trace, quantities of impurities (such as argon) could effectively reduce the oxygen concentration at the surface, thus significantly decreasing the aluminum consumption rate and flame-spreading rate. Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 Upward Flame Spreading In order to obtain a better understanding of the heat transfer mechanism in flame-spreading processes, the influence of sample orientation on the flamespreading rate has received considerable attention. Excluding external sources, the three principal mechanisms by which heat may be transferred to the unburned material are: radiation from the flame, conduction or convection through the gas from the flame, and conduction through the solid (Fernandez-Pello and Hirano, 1983). As shown in Figure 8, the vertically upward flame-spreading rates are nearly the same as those in horizontal orientation at similar initial chamber pressures. This implies that the heat conduction through the aluminum sheet is the dominant heat transfer mode. I 1000.0 I Sample Orientation . I 800.0 'V'V Horizontal Harne Spreading Upward FlameSpreading DownwardFlame Spreading 0 'V ~ >~ ~ l>: AI Sample: 0, J5 mm Plain Sheet Oxygen Purity= 99.75% 600.0 I- bll ~" ~ . . ..'to--. 400.0 - "'~" 0 ~ til f- § fi: 200.0 - 0.0 I- , , , 4.0 5.0 -\-,-,o-r--.<rr-~-<>?~-,--j>"~~,,...~,,~-r-,--r~~+ 0.0 1.0 2.0 3.0 6.0 7.0 Initial Chamber Pressure, Pi (MPa) FIGURE 8 Effects of sample orienlation on measured flame-spreading rates. FLAME SPREADING OVER ALUMINUM 207 Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 Downward Flame Spreading As also shown in Figure 8, vertically downward flame-spreading rates were much higher than those measured from horizontal and upward orientations under the same initial chamber pressures. It should be noted that, unlike horizontal and upward flame spread, the downward flame-spreading rate is not constant and it accelerates along the length of the sample. As shown in Figure 9, the downward flame-spreading rate increases from 670mm/s (which is based upon the output of first three photodetectors) to 1120mm/s from the last three photodetectors. The data of downward flame spread plotted in Figure 8 are based upon the linear fit of all six photodetector outputs. This substantially increase in downward flame-spreading rate is believed to be caused by the fact that the dripping of molten aluminum or oxide from the reaction zone to the unburned sample enhances the rate of heat transfer and then the flame-spreading rate. The dripping and flowing of molten aluminum along the sample sheet could increase the reacting surface area of aluminum, thereby resulting in the acceleration of flame-spreading rates. It is useful to note that a significant amount of molten materials was Downward FlameSpreading Initial Chamber Pressure = 5.06 MPa Al SampleThickness = 0.15 nun(PlainSheel) Oxygen Purity = 99.75% 300 250 E V FS = 873.4 mmls 200 ~" .§. c u s "« = 1120.7 mrnIs 150 ~ is 100 v" = 669.4 mrnIs 50 0 3.4 3.5 3.6 3.7 3.8 3.9 Time (sec) FIGURE 9 A typical flame-front trajectory of downward flame-spreading tests. C. L. YEH et al. 208 found at the bottom of the test chamber after downward flame-spreading tests, when compared to upward and horizontal flame-spreading tests. Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 Flowing Tests Concurrent Oxygen Flow All of the tests with continuous oxygen flow were conducted under the horizontal orientation in this study. Effects of both concurrent and countercurrent oxygen flows on the flame-spreading rate were shown in Figure 10. Under concurrent oxygen flows, the V FS increases with the flow velocity, reaches a maximum value and then starts to decrease with a further increase in the oxygen velocity. It is anticipated that for a given initial chamber pressure there exists an optimum oxygen velocity beyond which the V FS decreases with the increase of oxygen velocity. For the series of tests at pressures of 4.24 MPa shown in Figure 10, the optimum oxygen velocity is Oxygen Flow Rate (SCFM) 0.0 800 5.0 10.0 15.0 20.0 , 25.0 30.0 35.0 , I +~-'--.l~~-LL~J...L~~'--'-~'-'-~........J'--'-'~"""",.- - - .. - . Concurrent oxygen flow - 9 - - Counter-current oxygen flow • Stagnant oxygen 700 - I- 600 --'-- .•... ~ /---l-" 500 - 400 300 200 Initial Chamber Pressure = 4.24 MPa Oxygen Purity = 99.75% AI Sample Thickness e 0.15 nun (Plain Sheet) 100 ....~~rr~..,....,~ ....I ~~rr~-,-~ ...I ..,....,~tI I I o-h~ 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 Oxygen Flow Velocity (mls) FIGURE 10 Effect of oxygen flow condition on measured flame-spreading rates of horizontal orientation. Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 FLAME SPREADING OVER ALUMINUM 209 about 5 tn]«, which corresponds to a Reynolds number of 4.43 x 105 in a 2.7 ern by 4.5 em flow channel (the cavity dimension of the test chamber). In the concurrent mode of flame spread, the hot, still reacting or post combustion gases generated at the burning region of the material are driven ahead of the pyrolysis front close to the unburned combustible surface (Fernandez-Pelle and Hirano, 1983). The proximity of the hot gases or combustion products (e.g., Alz0 3 fine particles) to the unburned surface favors the heat transfer (convection and radiation) to the unburned material, thus enhancing the flame-spreading rate. When the oxygen flow rate is larger than the optimum value, the heat loss to the cold oxygen gas stream dominates the enhanced heat transfer by the flow; therefore, a decrease in V FS is observed. Counter-current Oxygen Flow For experiments with counter-current oxygen flow (also shown in Figure 10), there exists an initial region at flow velocities less than about 10 mls where the flame-spreading rate remains practically independent of the oxygen velocity. After this initial region, the flame-spreading rate increases with the oxygen flow velocity, reaches a maximum and then decreases drastically as the opposed flow velocity increases. Once the opposed oxygen velocity reaches a threshold value, the flame plume is blown off by the flow and the propagation rate drops abruptly from a measurable value to zero. It was found that extinction or non-flame propagation occurs in a very narrow range of flow velocities. Above flame-spreading phenomena are the characteristic of thermally-thick fuels, which are dominated by conduction heat transfer through the solid phase during the flame-spreading process (Fernandez-Pello et al., 1981). The existence of an initial region at low opposed velocities where the flame-spreading rate remains almost independent of the oxygen velocity is because the opposed flow obviously affects the gas-phase heat transfer more than the solid phase by convective cooling. Therefore, the net amount of heat transferred through the solid-phase conduction still remains nearly constant, resulting in a constant flame-spreading rate. The second regime shows a near linear increase of V FS with the opposed flow velocity and is mainly dominated by thermal processes (Fernandez-Pello et al., 1981). As the opposed oxygen velocity is increased, the diffusion flame moves closer to the aluminum surface and the rate of mass burning and the rate of heat transfer through the solid increases. It is useful to note that the character of Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 210 C. L. YEH el at. the leading edge flame can be described by the ratio of the residence time (time required for a fuel particle to travel to the gas preheated region) to the chemical time (time required for a fuel particle to react with the oxidizer), i.e., the Damkohler number (Fernandez-Pello et af., 1981 and FernandezPello and Hirano, 1983). For large Damkohler numbers (low opposed flow velocities), the reaction rate is very large. Gas phase chemistry becomes unimportant and the flame-spreading process is controlled primarily by heat transfer to the unburned material. For small Damkohler numbers (high opposed flow velocities), the reaction rate is slow. Finite rate chemical kinetics becomes important and flame-spreading rate depends strongly on the rate at which the fuel is consumed. If the Darnkohler number becomes smaller than a critical value, the extinction of the reaction occurs at the flame leading edge (FernandezPello and Hirano, 1983). Therefore, the sudden decline in VFS shown in the last regime of Figure 10 is mainly caused by the gas-phase chemical kinetic effect (Fernandez-Pelle and Hirano, 1983 and Ray and Glassman, 1983). As the opposed velocity is increased further, the flame front must propagate against a velocity greater than the lean limit flame speed. Consequently, the name is blown off downstream. Figure II shows the relative velocity (V'e1) between oxygen gas velocity (Vgas) and flame-spreading rate (V FS) versus the flame-spreading rate. The relative velocities for different test conditions are defined below. For concurrent flow tests, V,cl = V gas- V FS' For counter-current flow tests, V,cl = - Vgas- VFS' For stagnant tests, V,cl = - VFS' As shown in the Figure II, the flame was blown off in a counter-current flow test when the relative velocity was - 13.5 m/s. However, when the relative velocity was about 13 mls in the concurrent flow test, the flame was not extinguished by the flow. This difference may be due to the effect of concurrent flow on the shape of molten region and thus its reactive surface. Even though there is more heat loss to the external flow when the concurrent flow rate is higher, the flow-induced drag force can increase the reacting surface of the molten aluminum to continue the flame propagation. It is still possible for the blown-off phenomenon to occur in the concurrent flow test, as long as the heat loss from the combustion zone to the external flow exceeds a critical value, in which the heat flux transferred to the unburned material is not sufficient to pyrolyze the aluminum and to maintain the flame propagation process. FLAME SPREADING OVER ALUMINUM , 800 Initial Chamber Pressure = 4.24 MPa Oxygen Purity = 99.75% AI SampleThickness = 0.15 nun (PlainSheet) 700 Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 ~ g :2 > " :a" "ea. fI- 600 -. 500 B 0:: ea 211 400 -----. , , , ,','~ I- I- 300 Vl "" e [i: 200 - Counter-current Flow 100 Concurrent Flow Tests Tests - -< o -15.0 -10.0 -5.0 0.0 5.0 10.0 15.0 Relative Velocity, V~I (mls) FIGURE II Effect of relative velocity on flame-spreading rates of thin aluminum sheets. Another interesting point is that with continuous oxygen flow (both concurrent and counter-current flow), fresh oxygen is constantly available in the reaction zone and takes part in combustion during the flame-spreading event. Therefore, the diffusion is not the only gas transport mechanism in flowing tests. As shown in the figure, when the relative velocity (Vrel ) is zero, a minimum value of the flame-spreading rate is expected. This is because the diffusion is the only mechanism to transport oxygen to the reaction zone. Analysis of Recovered Samples The partially-burned aluminum sheets were quenched by a rapid depressurization of the test chamber to interrupt the combustion event. The burned edge shows nearly one-dimensional flame propagation over the sample. The molten portion of the sample (from the edge to the unburned material) is about 0.8 mm. It was found that a number of ceramic-type white nodules was sporadically deposited on the edge. The recovered partially burned aluminum sheets were examined under a scanning electron microscope (SEM) coupled with an energy dispersive Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 212 C. L. YEH et al. spectrometer (EDS). The EDS spectrum measured from the unburned sample is shown in Figure 12. The atomic oxygen concentration on the unburned sample is about 2.2 %, which is believed to be associated with an oxide film formed on a clean metal surface exposed to room air. The SEM image, shown in Figure 13, represents a typical structure of the burned edge (perpendicular to the sample surface) of a recovered aluminum sheet. White ceramic-type nod ules were formed on the burned edge of the recovered sample. As shown in Figure 14, the atomic oxygen concentration (about 25.5 %) on those white nodules "is much higher than that on the surface of unburned samples. It is speculated that the formation of white nodules could be caused by either growth of oxide caps or products of heterogeneous reactions on the molten aluminum surface. The substantial increase in oxygen content on the burned edge of recovered samples provides an evidence that the heterogeneous reactions of molten aluminum with either oxygen or gaseous aluminum sub-oxides are important. It is believed that the gas-phase intermediate product, AIO, Coun'ts (XIO' ) , I 16 14 - 12 Al - 10 B- - 6 - 4 - 2- - 0 \ 0 ~ ) \ I I 1 2 Ran •• (keV) FIGURE 12 EDS spectrum measured from the surface of an unburned aluminum sample. Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 FLAME SPREADING OVER ALUMINUM 213 FIGURE 13 Scanning electron micrograph showing white ceramic-type nodules formed on the burned edge of a recovered aluminum sample. Count .. (XIO') I I 3- Al I- 2- I - 0 }, I I I \. I 2 Ran •• (k.V) FIGURE 14 EDS spectrum measured from white ceramic-type nodules formed on the burned edge of a recovered aluminum sample. Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 214 C. L. YEH et al. diffuses from the gas phase to the sample surface, resulting in the increase in the oxygen content on the burning edge (Bucher et al., 1996 and Dreizin, 1996). The heterogeneous reaction of A10 and molten Al on the surface has been proposed as a means to produce AlzO (Bucher et al., 1996 and Yuasa et al., 1997). Based upon the measured AIO profile of isolated burning aluminum particles (Bucher et al., 1996), the non zero value of AIO at the burning surface suggests that the rate of this heterogeneous reaction is kinetically and not diffusively limited. This also explains that the higher flame-spreading rates measured in 99.996 %-purity oxygen than those in 99.75 %-purity oxygen (see Fig. 7). CONCLUSIONS Based upon the experimental measurements and observations of this study, several major findings are listed below. I. In terms of pressure effect on the flame-spreading rates, as the initial chamber pressure was increased the flame-spreading rate was found to increase to a maximum, decrease to a minimum, and then increase again. 2. The flame-spreading rates under high-purity (~99.996 %) oxygen environments were found to be significantly greater than those in commercial grade (~99.75 %) oxygen. This is believed to be caused by higher exothermic heat release from heterogeneous reactions on the molten sample surface. Moreover, the threshold pressure for selfsustained flame propagation of aluminum sheets was considerably lower under a high-purity oxygen environment. 3. For vertically oriented samples in upward flame-spreading tests, the flame-spreading rates were found to be nearly the same as those of horizontal tests with the same initial chamber pressure. In downward flame-spreading tests, flame acceleration along the length of the sample was observed. This is believed to be caused by the dripping and flowing of molten aluminum and oxide onto the unburned sample, resulting in an increase of reacting surface area and heat transfer rate. 4. Under the concurrent oxygen flow test conditions, the oxygen flow enhanced the flame-spreading rate. It is anticipated that there exists an optimum oxygen velocity (about 5 mls at 4.24 MPa) beyond which the flame-spreading rate decreases with the increase of oxygen velocity due to convective heat loss. Under the counter-current oxygen flow test conditions, it was demonstrated that the flame can be blown off when Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 FLAME SPREADING OVER ALUMINUM 215 the counter-current flow velocity exceeds a critical value of about 13.5m/s. 5. White ceramic-type nodules were found on the burned edge of the recovered partially-burned sample. The atomic oxygen concentration (25.5 %) in those white nodules is much higher than that (2.2 %) on the unburned surface. This implies that there may exist heterogeneous reactions between aluminum and either oxygen or gaseous aluminum sub-oxides on the burning surface. Acknowledgements This work presents the results obtained from the research project sponsored by Praxair, Inc. and Air Products and Chemicals, Inc. The authors are grateful to R. A. Van Slooten, R. Zawierucha, and J. F. Million of Praxair, Inc. and J. G. Hansel, P. A. Houghton, and E. C. Rogusky of Air Products and Chemicals, Inc. for their input and support of this project. References Benning, M. A., Zabrenski, J. S. and Le, N. B. (1988) The Flammability of Aluminum Alloys and Aluminum Bronzes as Measured by Pressurized Oxygen Index, Flammability and Sensitivity of Materials in Oxygen-enriched Atmospheres: Third Volume, ASTM STP 986,54-71. Benz, F. J., Shaw, R. C. and Homa, J. M. (1986) Burn Propagation Rates of Metals and Alloys in Gaseous Oxygen, Flammability and Sensitivity of Materials in Oxygen-enriched Atomospheres: Second Volume, ASTM STP 910, 135-152. Brooks, K. P. and Beckstead, M. W. (1995) Dynamics of Aluminum Combustion, Journal of Propulsion and Power, II, 769-780. Brzustowski, T. A. and Glassman, 1. (1964) Vapor-phase Diffusion Flames in the Combustion of Magnesium and Aluminum: I. Analytical Developments. Heterogeneous Combustion, 75-116. Bucher, P., Yetter. R. A., Dryer, F. L., Parr, T. P., Hanson-Parr, D. M. and Vicenzi, E. P. (1996) Flame Structure Measurement of Single, Isolated Aluminum Particles Burning in Air, Twenty-Sixth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1899-1908. Dreizin, E. L. (1996) Experimental Study of Stages in Aluminum Particle Combustion in Air, Combustion and Flame, 105, 541 - 556. Drew, C. M., Gordon, A. S. and Knipe, R. H. (1964) Study of Quenched Aluminum Particle Combustion, Heterogeneous Combustion, 17-39. Drew, C. M. (1965) Some Further Comments on the Paper "Estimating Aluminum Particle Combustions Kinetics" by Bartlett et 01., Combustion and Flame, 9, 205-208. Dunbobbin, B. R., Hansel, J. G. and Werley, B. L. (1991) Oxygen Compatibility of Highsurface-area Materials, Flammability and Sensitivity oj Materials in Oxygen-enriched Atmospheres: Fifth Volume, ASTM SPT 1111, 338-353. Fernandez-Pello, A. c., Ray, S. R. and Glassman, I. (1981) Flame Spread in an Opposed Forced Flow: The Effect of Ambient Oxygen Concentration, Eighteenth Symposium (International) on Combustion, The Combustion Institute, Pittsburgb, 579- 589. Fernandez-Pelle, A. C. and Hirano, T. (1983) Controlling Mechanisms of Flame Spread, Combust. Sci. Tech., 32, 1- 31. Downloaded by [University of Tennessee, Knoxville] at 17:12 07 November 2012 216 C. L. YEH et al. Glassman, I. (1959) Metal Combustion Processes, American Rocket Society, ARS Preprint No. 938-59. Glassman, I. and Law, C. K. (1991) Sensitivity of Metal Reactivity to Gaseous Impurities in Oxygen Environments, Combust. Sci. Tech., 80, 151-157. Johnson, D. K. (1997) "Flame Spreading Phenomena over Thin Aluminum Sheets in an Oxygen-rich Environment", M. S. Thesis, The Pennsylvania State University. Kirschfeld, L. (1961) Combustion Rate of Light Metal Wires in High Pressure Oxygen, Metallurgy, 15, 873-878. Law, C. K. (1973) A Simplified Theoretical Model for the Vapor-phase Combustion of Metal Particles, Combust. Sci. Tech., 7, 197-212. Long, G. H. and Sebald, H. (1965) The Burning Rate of Aluminum and Magnesium Wire in Pure Oxygen and Oxygen-water Vapor Atmospheres at Higher Pressures, Western States Combustion Institute, p. 67. Prentice, J. L. (1965) On the Combustion of Single Aluminum Particles, Combustion and Flame, 9,208-210. Prentice, J. L. (1970) Combustion of Pulsed-heated Single Particles of Aluminum and Beryllium, Combust. Sci. Tech., 1,385-398. Ray, S. R. and Glassman, I. (1983) The Detailed Processes Involved in Flame Spread over Solid Fuels, Combust. Sci. Tech., 32, 33-48. Sato, J. (1989) Fire Spread Rates along Cylindrical Metal Rods in High Pressure Oxygen, Flammability and Sensitivity of Materials in Oxygen-enriched Atmospheres: Fourth Volume, ASTM STP 1040, 162-177. Sato, J. and Hirano, T. (1986) Behavior of Fire Spreading along High-temperature Mild Steel and Aluminum Cylinders in Oxygen, Flammability and Sensitivity of Materials in Oxygenenriched Atmospheres: Second Volume, ASTM STP 910, 118-134. Sircar, S., Gabel, H., Stoltzfus, J. and Benz, F. (1991) The Analysis of Metals Combustion Using a Real-time Gravimetric Technique, Flammability and Sensitivity of Materials in Oxygen-enriched Atmospheres: Fifth Volume, ASTM STP 1111, 313-325. Steinberg, T. A., Wilson, D. B. and Benz, F. (1992) The Combustion Phase of Burning Metals, Combustion and Flame, 91, 200-208. Yeh, C. L., Johnson, D. K. and Kuo, K, K. (1997a) Experimental Study of Flame-spreading Processes over Thin Aluminum Sheets, Flammability and Sensitivity of Materials in Oxygen-enriched Atmospheres: Eighth Volume, ASTM STP 1319,283-296. Yeh, C. L., Mench, M. M. and Kuo, K. K. (l997b) An Investigation on Flame-spreading Process of Thin Film MgjPTFEjMg Pyrotechnics, Combust. Sci. Tech., 126, 271-289. Yuasa, S., Shu, Y. and Sogo, S. (1997) Ignition and Combustion of Aluminum in Oxygenj Nitrogen Mixture Streams, Combustion and Flame, 108,387-396.
© Copyright 2026 Paperzz