EXPERIMENTAL STUDY OF THE EXPLOSION CHARACTERISTICS OF METAL DUST CLOUDS WATARU ISHIHAMA A N D H E I J I ENOMOTO
Department of Mining and Mineral Engineering, Tohoku University, Sendal, Japan An experimental investigation of the explosion characteristics of dust clouds of ahtminum and magnesium was carried out for a wide range of dust concentrations up to 7000 g / m ~, using a new closed experimental apparatus. The explosion characteristics of aluminum and magnesium did not show much difference. The explosion pressure for alumhmm and magnesium tended toward maximlan values at concentrations roughly 3 to 5 times higher than the stoichiometric concentrations in the reactions of A1 to Al~O8 and Mg to MgO. The explosiou pressure for altuniuum decreased rapidly whereas that of magnesium decreased fairly slowly with increasing dust concentration. The optimum dust concentration, at which the explosion pressure was highest, increased for aluminum and decrus.~ed for magnesium as the amo,mt of oxygen available for the reaction decreased. The time-to-peak pressure/dust concentration curve for aluminmn was convex downwards with the minimum value near the optimmm dust concentration, whereas the curve for magnesium was constant after it reached tim mhfimum. In the relation between tim amount of residual oxygen and the concentration of the dust cloud, the trend was the same for both aluminum and magnesium as that of the time-to-peak pressure/dust concentration curve for m~gnesium. The amount of oxygen available for the reaction was changed by diluting the air with nitrogen [case (I)] and by decreasing the initial pressure [case (II)]. I t was found that the explosion region of case (II) was wider with respect to both oxygen and dust concentrations than that of case (I), although the explosion pressure was higher in case (I) than in case(II), except near the explosion limit.
1. Introduction F r o m the vicwpnint of safcty enginecring, H a r t m a n n , N a g y and their co-workers ~--4 have conducted systematic experlmcutal investigations of the explosion characteristics of dust clouds, including metal powders. T h e y reported the ignition temperature, the m i n i m u m energy required for ignition and thc m i n i m u m explusive concentration as factors affecting the ignition sensitivity, and reported the maximum explosion pressure and the rate of pressure rise as factors affecting the explosion severity. In the experimental system they used, dusts wcrc disperscd into the explosion chamber by air sucked into it? W i t h this system, reliable data would not be obtalnable at high concentrations. T h e authors 6 h a v e dcveloped a new expcrlmental apparatus t h a t can generate sufficiently uniform dust clouds over a wide range of dust concentration up to 7000 g/'m 3 or more. The aim of the present study is to investigate,
mainly from the point of view of safety engineering, the effects of the oxygen in the surrounding atmosphcrc and of the dust concentration on the explosion characteristics, and to observe t h e explosion behavior of metal dust clouds, using t h e apparatus mentioned above.
2. Experimental 2.1.
Apparatus and Procedure
The idea of the method for generating a uniform dust cloud came from the ball mill. The apparatus consists essentially of two parts. One is an explosion chambcr and the other is a device to rotate the explosion chamber in order to generate a uniform (lust cloud in it. T h e length and the inner diameter of the chamber are 170 m m and 270 men, respectively, so that the volume is approximately 10 liters. Both ends of the cylindrical chamber have covers that are fastened
6000 o,.I (.9
o U 7 2000 o
DUST CHARGED (g ) FIo. 1. The relation 1)etween the amount of dust charged into the explosion chamber and the dust concentration. (The rotation speed of the chamber is 53.5 r.p.m.)
with bolts to make it air4ight with 0-rings. One cover is a transparent plate through which the explosion phenomena can be observed. The other cover has a pair of cocks on it to introduce gas into the chamber, to measure the pressure in the chamber before and after the explosion, and to take samples of gases for analysis. The pressure generated by the explosion is measured electrically through a slip ring, with a small strain guage type pressure transducer at the center of the cover. Guncotton with an electric detonator fusehead was used as the ignition source. The preliminary experiment showed that to have the same effect as ignition source, the amount of guncotton had to be increased as the amount of oxygen in the surrounding atmosphere decreased. For example, the amount of guncotton used was 0.05 g in air at 1 atm initial pressure, and 0.25 g when the amount of oxygen in the chamber was reduced to 1/3 of the air at 1 arm pressure. The guncotton was placed at the center of the explosion chamber. The concentration of the dust cloud was ob-
uJ rr 4.0 tn 3.0 u3 b~ ~r 2.0 n 1-0
Fro. 2. A pressure-time record with photographs showing flame propagation at an ahmdnum concentration of 1000 g/m~ in air at 1 arm initial pressure.
EXPLOSION OF METAL DUST CLOUDS talned from the relation between the amount of a0 the dust introduced into the explosion chamber and the dust concentration. This relation is ~ 7 0 shown in Fig. 1. The detailed description of the apparatus and the procedure, and the d i s - ~ 6 . 0 cussion of the validity and accuracy of the experimental system, have been reported elsewhere. ~ ~ 5 0 The amount of oxygen available for the roan- ~ " I tion was changed by the dilution of air with nitrogen ['case (I)'], and by the decrease of the ~ &0 initial pressure ['case (II)']. t~ ~a0
Aluminum and magnesium were chosen as 9 14 " 9 ', 10 " testing materials. The aluminum and magnesium ~ 1.0 @ 6 were commercial products and were, respectively, O [ n f l . P r e s s . 510 m m H g i i i i i I i reported as 99.0 and 99.9% pure. The particle W 0 2000 4000 6000 8000 sizes of almninum and magnesium are --400 mesh and --150 mesh, respectively. The aluDUST CONCENTRATION ( g l m =) minum particle was shaped like a flake and the Fro. 4. The relation between the explosion presmagnesium particle was shaped like a sharply snre and the concentration of the magnesium dust edged stone. cloud when the air is diluted with nitrogen or the initial pressure is rednced. 3, Results Figure 2 is a pressure-time record with photographs obtained with a high speed camera at 800 frames per second. The flame propagated
e O2 conc.21~ r ', 16
2000 4000 600 DUST CONCENTRATION ( g / m 3
Fro. 3. The relation between the explosion pressure and the concentration of the aluminum dust cloud when the air is diluted with nitregen, or the initial pressure is reduced in order to change the amount of oxygen available for the reaction.
approximately concentrically. However, strictly speaking, the flame propagation speed was slightly lower in the downward direction than in other directions. Tile pressure-time curve reached the peak at the moment when the flame front in the downward direction reached the chamber wall. The pressure generated by the explosion is plotted against the dust cloud concentration in Figs. 3 and 4. The curves for aluminum and magnesium showed little difference. Both curves were convex upwards near the concentration at which the explosion pressure was highest, that is, the optimum dust concentration, and they were convex downwards at higher concentrations. For aluminum, thc inflection point was near the optimum dust concentration. The explosion pressure decreased sharply as the dust concentration increased. At higher concentrations, the curve became almost fiat and then suddenly went to zero near the upper limit concentration. On the other hand, for magnesimn the inflection point was far from the optimum dust concentration and the explosion pressure decreased fairly slowly as the concentration of the dust cloud increased from the optimum dust concentration to the inflection point. At higher concentrations, the curve became almost flat like that for aluminum. The highest explosion pressure obtained was 6.2 kg/em~ for ahmfinum and 7.4 kg/cm2 for magnesium. Figures 5 and 6 show the relation between
~ 02 w 0,I
DUST CONCENTRATION (g/m3) Fro. 5. The relation between the time to peak pressure ~nd the dust cloud concentration for almninum. 08 nile
r O2conc.21% ( a i r )
: ',:: 14,, ,o::
~ 0-6~ t
o Inth Press. 510mmHg
centration increased. However, the time to the peak pressure for magnesium became constant after it reached the minimum. Another interesting result was that when the amount of oxygen available for reaction was reduced by reducing the initial pressure, the time to the peak pressure was the same as that in air at 1 arm initial pressure. In other words, the time to the peak pressure varied with the molar ratio of oxygen and nitrogen, not with the amount of oxygen available for reaction. This result was obtained for both aluminum and magnesium. I~ Fig. 7, the reciprocal of the time to peak pressure was plotted against the amount of oxygen consmned. The figure shows that for case (II) the reciprocal of the time to the peak pressure was independent of the amount of oxygen consumed and that for case (I) the relation between the reciprocal of the time to the peak pressure and the amount of oxygeu consumed could be approximated by a straight line. The amount of residual oxygen after the explosion was plotted against the dust cloud concentration in Fig. 8. The figure shows that the amount of residual oxygen was coustant at concentrations higher than approximately 1500 g / m s. Moreover, it did not vary with b~tiM conditions such as oxygen concentration. The amount of residual oxygen for magnesium was wJ @
d" -(,.i, I.iJ
rr 0 0 the dust cloud concentration and the time from ignition at which the explosion pressure reached maximum (time to peak pressure). The time to peak pressure decreased as the dnst cloud concentration increase*l, until it reached the minimum. After that, for aluminum, the time to peak pressure increased as the dust cloud con-
DUST CONCENTRATION (g/m a) Fio. 6. The relation betweelx the time to peak pr~stLre and the dust cloud concentration for magnesimn.
| At (~) o MQ(I)
FIG. 7. Variation of the reciprocal of the time to pe~k pressure with the amount of oxygen consmned by the reaction at a constant dust cloud concentration of 1100 g/m 8 for aluminum and 1500 g/m ~ for magnesium
EXPLOSION OF METAL DUST CLOUDS less than that for alumiunm. When the air was used at 1 atm initial pressure, the amount of oxygen in the chamber was approximately 2000 em 8. In this case, the residual oxygen was 5% of the initial amount of oxygen for magnesium and 7.3% for alunfinum at dust cloud concentrations above 1500 g/'m~.
reduction. The pressure generated by the quantity of guncotton used in the experiment was less than 0.1 kg,/cm2, so that the error would be very small.
4.2. OptimumCondition and Explosion PressureDust Concentration Relation
If the system is adiabatic and the gas is ideal, the explosion pressure is proportional to the amount of heat released. In the system used in the present study, part of the beat produced by 4.1. Influence of Guncotton the reaction is lost to the chamber wall. Therefore the explosion pressure may be considered proSince guncotton does not have enough oxygen portional to thc rate of heat release. The rate of in itself to complete the reaction to produce heat release is a function of the reaction rate of a CO~, NO~, and H~O, it consumes some amount single particle and the propagation speed of the of oxygen iu the explosion chamber. The effective flame front. Flame propagation is due to heat amount of guncotton used in the present study is transfer between burning particles and ambient 0.05 g. The calculation shows that 0.05 g gun- particles, that is to say, the flame propagation cotton requires approximately 24 em ~ additional speed is influenced by the distance between the oxygen to produce CO~, NO2, and H~O, This particles. amount of oxygen corresponds to 4.9% of the The effect of the concentration of the dust initial oxygen when this initial amount was the cloud on the explosion pressure is such that the least used in the present study. Since combustion increase of dust concentration incrcasos the of the guncotton is rapid and localized and the specific heat of the gas solid mixture. The produces gases such as CO2, NO2, and H20 the distance between two adjacent particles changes influence may be greater. However, particles are proportionately to the reciprocal of the cube root ignited and the flame front propagates before the of the dust concentration, assuming unkform guncotton burns up. Therefore, it can be said that distribution of the particles, and so on. Some the influence would not be much greater than particles have positive effects and some have indicated above. negative effects. The explosion pressure-dust The pressure generated by the burning of the concentration curve represents their total effect guncotton itself was not considered in data and the explosion pressure reaches a maximum value at a certain dust concentration that is the optimum dust concentration. 100(3 The optimum dust concentrations is shown in o AI 02 conr (air) Figs. 3 and 4 as thc concentration at which the @ 14,, explosion pressure is maxlnmm. The optimum 80O e ,, Intl. Press. 510ram Hg dust concentrations of aluminum and magnesium 9 Mg 02 cone.2T% (ai r) in air at 1 atm pressure are about 1000 and I) ', 1/-* ,, 1800 g/'ma, respectively. The stoichiometric 60(3 dust concentrations in ideal reactions in air at 1 @ 10 " e, 6" atm pressure at room temperature for Al to @ Intl. Press. 510 mmHg form A120a and Mg to form MgO are 320 and 400 420 g / m a, respectively. The optimum dust @ concentration is about 3 to 5 times higher than the steichiometric dust concentration. This ~U~200 i ' e ~ ~ ' ~ o , n ~ means that the m a ~ m u m explosion pressure is ngenerated at the condition for which about 1/3 to 1/'5 of the mass of each particle participates J O 0 '~o~o ~;oo 8ooo in the reaction if the sizes of all particles are the same. BUST CONCENTRATION ( g / m s) Furthermore, the optimum dust concentraFro. 8. Amount of residual oxygen for aluminum tion seems to vary with the initial amount of and magnesium. oxygen, the opthnum dust concentration of 4. D i s c u s s i o n
INITIAL PRESSURE(rnrnHg) 200 400 600 760
MAX. RATE OF PRESS. RISE
7.0 nr u~ bJ
(~ A I ( I ) 9 ,, (lI) o Mg(I)
,, ( I I )
9 01 CONC. 21% (~,ir) 9 " 14(~ Intl. PRESS, 510mmHg
AVE. RATE OF PRESS* RISE (~ O= CONC, 21% ((2it) O ,, 14 -
O IntL PRESS. 510mmHg
~ 4.0 ffl
h1120 U3 n" 10~ LU nt"
(II Dilution of Air (11) Intl, Press. Red.
5 10 15 CONCENTRATION ( % )
02 Fro. 9. The variation of the maximum explosion pressure with the concentration of oxygen and the initial pressure.
aluminum increasing and that of magnesium decreasing as the amount of oxygen decreases. It has been shown that the optimum dust coneentration of a coal dust explosion increases as the volatilc content increases and/or the particle stze decreases. 6 As mentioned at the beginning of this paragraph, the optimum dust concentration is an optimum condition in the semse of the correlation between the reaction of a single particle and the propagation of the reaction. To clarify this, one must understand the detailed mechanism of the reaction and its propagation. The shape of the cxplosion pressure-dust concentration curves of aluminum and magnesium is different from that of coal dust, which is convex upwards in the whole explosion region and is almost symmetrical with the axis at the optimum dust concentration. The shape of the explosion pressure-dust concentration curve of relatively coarse agricultural dusts is almost the same as that of coal dust. 7 Bartlett et al., 8 reported the burning time of aluminum, and Liebman et al., 9 reported the burning time of magnesium particles. Howard and Essenhight~ showed the volatile matter evolution-tlme relation of coal particle~. From these results it can be said that the over-all reaction rate of ahiminum and magnesium partitles is about 10 to 100 times faster than that of coal particles. This may be the main reason why the characteristics of the explosion pressure-dust concentration relation of aluminum and mag-
w 11. u. 04C r
DUST CONCENTRATION(g/n~) FIe. 10. The maximtLmand the averagerate of pressure rise for ahtminum.
200 ~ I 80
O 02 CONe. ~l%(air)
O .... AVE. e [nil, PRESS.510~ , MA.
22oF 2oo0 ~ooo~ooo
DUST CONCENTRATION (glrnP) FIG. 11. The maximum and the average rate of pressure ribs for mag'nssium.
EXPLOSION OF M E T A L DUST CLOUDS
,,4000L%: / / / / g~~ ,ooo
E z 300o o
Fro. 12. The explosion limit of ahtmimml and the curves representing the conditions at which the same explosion presstu'e is generated (in the case of dilution with nitrogen). nesium are different from those of coal dust. The difference between aluminum and magnesium m a y be attributed to tim differences of the particle size and shape, and of their therraodynamic characteristics, such as boiling point and heat of combustion. (The boiling points of aluminum and magnesium are about 2600 and 1400~ and the heats of combustiml arc 8.3 and 9.0 kcal/g of 02, respectively.) 4.3.
Pressure and Rate. of Pressure
In Fig. 9, the maxiraum explosion pressure is plotted agailmt the oxygen coneentratkm or the initial pressure. The maximum explosion pressure decreases almost linearly as the amount of oxygen decreases. However, when the air is diluted with nitrogen to reduce t,he amount of oxygen, case (I), the maximum explosiml pressure decreases more rapidly near the explosion limit. When the initial pressure of thc air is reduced, case (II), the explosion pressure is lower than that of case (I), for the same amount of oxygen. However, the explosion limit in case (I) corresponds to a greater amount of oxygen t h a n in case (Ii): For aluminum the explosiou in case (I) does not occur tit a lower oxygen concentration than 10% whereas the explosion occurs at 8% (equivalent) of oxygen iu case (II). Figure 3 also shows the explosion pressure
as a function of the concentration of aluminum in the dust cloud in case (I) and case (II) when the amount of oxygen available for the reaction is almost the same. The explosion pressure ill case (I), curve (a), is higher than that in case (II), curve (b). However, whereas curve (a) has its upper ILmit near a concentration of 4000 g/'m3, in case (H) the explosion occurs at higher concentrations. That is to say, the explosion region of case (II) is wider than that of case (I) both with respect to the amount of oxygen and the dust concentration. Figures 10 and 11 show the maximum and tile average rate of pressure rise fur aluminum and magnesium against the dust concentration. The shape of these curves is similar to that of the explosion pressure-dust concentration curves, but the pressure rise~lust concentration curves have sharper peaks near thc optimum dust concentrations, particularly for aluminum. The figures also show that for the same amount of oxygen available for reaction, both the m a x i m u m and the avcrage rates of pressm'e rise are higher in case (II) than in case (I). Therefore we cannot, necessarily say that ease (II) is safer than ease (I), although the e~)losion pressure is higher ill case (I) than in case (II) in the explosion region of case (I). The hlghesl, values of the maximum rate of pressure rise obtained in the present study are 220 k g / c m L see for aluminum and 200 k g / c m : , see for magnesium. The fact that the highest value of the maximum rate of pressure rise is greater
for ahmlinun~ than for magesium may be due to the differences of dust particle size and shape, 4.4.
than that of aluminum and the limit oxygen concentration was approximately 4%. REFERENCES
Figure 12 shows explosion limit with lines representing the conditions at which the same pressure is produced by the cxplosiml. No explosion of the aluminum dust cloud occurs at any (lust cloud concentration when the concentration of oxygen is lower than 10%. The upper limit dust cloud concentration decreases as the oxygen concentration decreases and the lower limit concentration decreases as the oxygen concentration hlcreases. In other words, the cnrve of the exl)losinn limit forms an cxplosion peninsula. The shape of the curve rcprcscnting the conditions at which the same pressure is produced is almost tile same as that of thc cxplosion limit curve. The line drawn to connect the tops of each peninsula is straight. This line rcprcscnts the optimum condition for explosion, the explosion pressure being highest at the condition shown by the line, when the concentration of oxygen is fixed. The dust cloud concentration at the optimum condition decreases as the oxygen concentration increases. At a dust cloud concentration lower than that shown by this line, a slight variation of the dnst cloud concentration produces a larger difference of the explosion pressure than at a dust cloud concentration higher than that shown by the line. The figure also shows that the variatinn of the explosion pressure is large near the explosion limit. The explosion limit of magnesium was not obtained. However it was found that the slope of the curve of the upper explosion linfit of magnesium was steepcr
1. IIARq~TANN,I., NAOV, J., AND B]~ow:% tI. R.: Inflammability and Explosivility of MetM Powders. U.S. Bureau of Mines, Report of Investigation 3722, 1943. 2. JIARTM4NN, I., NAtty, J , Asn JACOtlSON, M.: Explosive Characteristics of Titanium, Zirconium, Thorium, Uranium and Their Hydrides. U.S. Bureau of Mines, Report of Investigation 4835, 1951. 3. &~coBsos, M., CooPsu, A. R., AND Nxor, J.: Explosivility of MetM Powders. U.S. Bureau of Mine~, Report of Investigation 6516, 1964. 4. N~Gv, J., CooPnR, A. R., a x n DORS~Tr, H. G., Jr.: Explosivility of Miscellaneous Dusts. U.S. Bureau of ~[ines, Report of Illvestigation 7208, 1968. 5. DonsF.Tq', tl. G., JR., JACOaSON, M., NAOv, J., AND WILLIAMS,R. P.: Laborator7 Equipment and Test Procedures for Evaluating Explosivitity of Dusts. U.S. Bureau of Mines, Report of Investigation 5624, 1960. 6. ISHIHAMA, W. AND ENOMOTO, H.: Combust. Flame 21, 177 (1973). 7. IsaiR~_-aA, W.: Explosion Characteristics of Agricultural Dusts. A Report Submitted to Enviromnental Di~tuption Engineering Exploitation Research Corporation, Tokyo, Japan. 8. B.~_I~TLETT,R. W., ONG, J. N., J1~., ANn PAPV, C. A.: Combust. Flame 7, 227 (1963). 9. LIEBMA-N',I., CORRZ, J., A.~D PERLNE, H. E.: Comb. Sci. Tech., 5, 21 (1972). 10. HOW.~RD,J. B. ANDESSENaIGH,R. H.: Eleventh Symposium (International) on Combustion, p. 399, The Combustion Institute, 1967.
R. H. Essenhigh, Penn State University, USA. You mentioned the teeth in your chamber are desigued to produce a uniform dust cloud. Can you amplify that, for example, how uniform, and over what concentration ranges? You also show different trends between the metal dusts and the carhonaceous dusts (coal etc.) Do you have an explanation for tbese differences or alternatively are you developing any analysis to predict or explain the observed behavior?
Authors' Reply. We could obtain a sufficiently urdform dust cloud without teeth on the inner wall of the chamber. However, the dust cloud generated without teeth is unstable. Moreover, without teeth we cannot obtain a dust cloud of required concentration. Thcrcfore the teeth play a very important role. Although we never obtained tile upper limit of dust concentration tbat the chamber can produce, we can obtain aluminmn and magnesium dust clouds of at least 7000 g / m 3 as shown in Fig. 1, and a coal dust cloud
EXPLOSION OF METAL DUST CLOUDS of 4000 g / m ~. We think that this dust concentration is high enough to study the combustion and the explosion of dust clouds. UniformiLy of tim dust cloud was examined using a photoelectric cell. The result showed that the dust cloud was sufficiently uniform. Dctails of the discussion of the validity and the accuracy of the apparatus are presented in Ref. 1. It is a hard task, at thjs moment, to answer professor Esscnhigh's question about the different trends between the metal dusts and the carbonaceous dusts. The experimental results of gas analysis after coal dust explosions showed that the amount of o.xygen consumed by reaction and the amomlt of heat released by reaction was largest near the optimum dust concentration (o.d.e.). This may mean that the explosion pressure hlcreases as the dust concentration increases, reaches the maximmn at o.d.c, and decreases as the dust concentration increases. --~loreover, the excess amount of dust acts as a heat sink. Therefore thc drop of explosion pressure is rapid as the dust concentration increases. On the other hand, iu metal dust explosion, there is some heat recovery due to the condensation of gas phase metal oxidc to solid or liquid oxide after reaction. Furthermore, as shown in Fig. 8, the amo~mt of oxygcn co~sumed by reaction is constant at higher concentrations than o.d.c. Thesc may bc the reasons why the explosion pressure of metal dust decreases more slowly as the dust concentration increases at higher concentrations than o.d.c. In the explanation mentioned in the text and
above, we assumed the gas phase reaction in the coal dust explosion, though you found that the surface reaction was dominant for coals less than 15 microns in size.2 Because thc smallest particle size of coals used in thc experiment was 37 microns. 1 We know that the over-all reaction rate may increase as the particle size decreases and/or the volatile content increases. Therefore the same trend m~y be expected for both metal dusts and carbonaceous dusts if the particle size is small and the volatile content is high. For example, the explosion pressure~lust concentration curve of Ponbetsu Coal (volatilc content: 42V~o, particle size: 45 ~) shown in Fig. 6(a) in Ref. 1 could hc said to show the same trend as that of metal dusts. Also results of a recent study on the explosion characteristics of agricultural dusts show that the trend of the explosion pressure dust concentration curves of fine dusts is the same as that of metal dusts, a To give a more explicit answer, we have to know what condition the o.d.c, is. As mentioned at the presentation, all phenomena are not reason~ ably explained yet. This gives us future work. REFER.ENCE 1. ISaI~A~L~, W. AND E~OMOTO, H.: Combust. Flame 21, 177 (1973). 2. HOWARD,J. B. AND ESSENHIGH, R. H.: Eleventh Symposium (Inter-national) on Combustion, p. 399, The Combustion Institute, 1967. 3. IsmnA~., W. A-'~DENOMOTO,ll.: J. Japan Soc. for Safety Eng. 1~, 4 (1975).