J. A. BROWNING,
T. L. TYLER,
and W. G. KRALL'
Dartmouth College, Hanover, N. H.
Effect of Particle Size on Combustion of Uniform Suspensions
Better particle size control may make it possible to extend the lean limits of stable spray combustion. The work reported bridges a g a p left by previous investigations, and provides interesting experimental data in a new and difficult field
IN
THE past several years it has become increasingly evident that the combustion stability of liquid sprays is in part dependent on the average particle diameter of the spray. In turbojet engine practice it has often been observed that a spray composed of droplets covering a wide range of diameters is more desirable from the standpoint of stability than a high pressure, finely atomized spray. In a Dreyhaupt report written in 1942 ( 9 ) it was suggested that a certain droplet size is best suited for combustion. If the droplets in a lean mixture are too small they evaporate completely prior to combustion and may not reach ignition temperature. In large drops there is an insufficient amount of air readily available for combustion, and the drops are cooled to below the ignition temperature. Thus, proportionately more fuel must be supplied for the burning of lean mixtures composed of droplets either larger or smaller than a critical size. Rivikre and Mayorcas (77) have shown that the greatest change of flame condition is brought about by changes in momentum of the liquid spray. This effect will often overshadow the influence of particle size on flame stability. Thus. only those systems exhibiting low momentum and turbulence levels are effective in showing the importance of par-
Present address, General Electric co.,
Pittsfield, Mass.
1 42
INDUSTRIAL AND ENGINEERING
ticle size. Conclusive evidence of this particle size effect has been obtained by Burgoyne and Cohen (5) who have made a study of the lean limit of flammability of Tetralin aerosols. Their experimental results show an increase in the air-fuel ratio from 26: 1 for 10-p diameter drops to 64:l for 55-p drops. This represents an increase in the airfuel ratio of nearly two and one-half times for stable burning of lean mixtures of suspended liquid droplets! B u r g o p e and Cohen have added a substantial contribution to the combustion theory of fuel particles. However, their work has been limited, because of aerosol generator characteristics, to studies of lean mixtures and of drop sizes below 55 1. Additional evidence of the effect of particle diameter on the lean limit can be obrained from the results of tests conducted on sprays of heterogeneous distribution. Anson (7), using kerosine fuel injected a t various pressures, has obtained Sauter mean diameter sprays from 60 to 150 p. Variation of mixture strength a t the lean limit ranges from 15 : 1 for the largest size to 29: 1 for the 60-p spray. A comparison of the results of Burgoyne and Cohen using small drops and those of Anson using larger ones leads to an anomaly. It becomes increasingly evident that a critical drop size will produce the leanest combustible mixtures. I t is unfortunate that their
CHEMISTRY
results cannot be correlated directly. Different fuels were used and the conditions for combustion varied greatly. Whittle (20) has sketched a curve showing the variation of the weak extinction limit as a function of particle size. No values are given. but it clearly shows the strong influence of the degree of atomization. In his experimental work FVhittle studied the effect of fuel volatility on the lean limit of combustion. He summarizes his results: (1) FVith poor atomization. improved weak extinction limit can be obtained by decreasing the particle size; (2) with a high level of atomization, improved weak extinction limit can be obtained by increasing the particle size. hlumerous references on the combustion of solid suspensions discuss the effect of particle size on the lean limit, maximum flame speed, and maximum pressure rise in a n enclosed volume. Hartmann and coworkers (72-74) have caused explosions to occur in suspensions composed of particles as large as 700 g. Lean limits are tabulated for a variety of compounds and pure substances with a typical air-fuel ratio of 48 :1 obtained for polyethylene particles. In all cases the most favorable lean limit was obtained for the smallest particles tested (92y0 through 200 mesh). However, particle sizes considerably smaller than 200 mesh were not tested.
I n each suspension Hartmann found that the maximum pressure rise occurred at fuel concentrations well above stoichiometric. For urea resin, a typical compound used, the peak pressure was 80 pounds per square inch at an airfuel ratio of 0.58: 1. Also, higher pressures are obtained as the particle size is diminished. Craig ( 7 ) discusses the results of tests conducted on coal dusts at the Taffanel Experimental Galleries, Lieven, France. Peak flame velocity for 200-mesh coal dust was obtained at an air-fuel ratio of 5.1 : I . Cassel and coworkers ( 6 ) ,using aluminum powders, found an increase in flame velocity with a decrease in particle size. A major difference between the combustion of gaseous fuel mixtures and suspensions of particles is the much greater rich limit of the suspensions. T h e combustion of submicron kerosine mist might he expected to be similar to that of propane vapor. The lean limits and flame velocities are nearly identical, but the rich limit of the mist is significantly greater ( 4 ) . I t is believed that an increase in particle size tends toward higher rich limits. Much richer mixtures of sprays of heavier distillate fuels can be burned than of lighter fuels ( 7 7 , 75). The purpose of the work reported here is to furnish a bridge for the gap between the results obtained by Burgoyne and Cohen on the one hand and Anson on the other. A single fuel should be used over the entire particle size range in question. The problem of producing such uniform mixtures has been a major difficulty. Several attempts were made to use liquid fuel suspensions, but in each case a deterrent became evident. The use of a spinning disk to produce uniform droplets is well understood and has been used to advantage by such investigators as Pilcher (76), Bolt and
Size ranges:
A.
10-20 microns
Figure 2.
Figure 1.
Equipment used for elutriation of Armowax particles
coworkers ( 3 ) ,and Walton and Prewett (19). The major difficulty of the spinning disk technique is that the droplets are thrown off in a single plane making local mixture air-fuel ratio determinations impossible. Several rather ingenious methods were investigated but later discarded. They include the electrical atomization of liquids as discovered by Vonnegut and Neubauer
B.
(78), the vibrating capillary tube used by Dimmock (8),and the collapse of air bubbles breaking through a liquid surface to produce an upward moving jet which breaks into uniform small droplets ( 2 ) . Fenn has produced uniform droplet distributions by means of passing small wires through the surface of the liquid. Each wire picks up the same amount of fluid and is later “flicked” to throw off a
40-50 microns
C.
Photomicrographs of graded Armowax particles-1
100-1 10 microns
00 X
VOL. 49, NO. 1
JANUARY 1957
143
Figure 3.
Combustion system
droplet. '4n aerosol generator follo~ving the design of the LahIer-Sinclair q s t e m was constructed Its use \vas discontinued when the parallel. but prior, work of Burgojne and Cohen was discovered. A paper by Ghosh and Orning (70) has shown that very small solid particles could be continuously circulated about a loop to form uniform suspensions. By use of a sonic driver a sufficient amount of energy was supplied to the fluid system to keep particles from agglomerating and sticking to the walls of the tube forming the loop. I t was decided to use a system similar to that of Ghosh with the exception that the mixture would be ignited within the loop itself. To simulate the conditions of a liquid spray as nearly as possible. the solid particles to be used should have a low melting point, be a hydrocarbon cvhich would be expected to have combustion characLeristics similar to the more common liquid fuels, and to have physical properties of such a nature that the material can be easily ground, graded into size fractions. and stored Armowax, meth) lene bisstearamide, is a product of the Armour Chemical Division and has fulfilled these requirements. Chemical and physical properties are : Free fatty acids, % Color (NPA) Ash, % Specific heat at 60' C. Flash point, O C.
144
5.0 6 0.03 0.532 250
Figure 4. Combustion accessories Fire point, C. Molecular wt. Solidification point,
258,
579 C.
132
I t may be purchased in powder form which is composed of particles ranging from less than 1 to over 150 p .
Apparatus and Procedure Air-elutriation was chosen for the grading of the Armowax particles. Other techniques proved unsatisfactory because the wax has a tendency to adhere LO foreign objects. LVhen suspended in air the particles can be handled easily. Figure 1 is a n illustration of the elutriation equipment. A controlled flow of air passes through the suspension of particles contained above the fine-mesh screens. Particles below a certain size are carried through the vertical rube and are collected. To separate the poivder into the desired size ranges the air flow is first adjusted to remove particles u p to 10 ,u. After the time required to remove an acceptable proportion of this first size range from the sample. the air flow is increased slightly to produce 10- to 20-,u the next size range-the fraction. Each size range produced contains particles of smaller size. A prohibitive amount of time would be required for complete separation. However. the quantity of stray particle sizes in each fraction is so small that its influence on combustion characteristics is considered negligible.
INDUSTRIAL AND ENGINEERING CHEMISTRY
equipment
showing
loop
and
Eleven size ranges have been separated by this technique. Each range includes those particles contained in 10-,u increments. Figure 2 shows photomicrographs of three typical separations. The large out-of-focus regions of Figure 2A are clumps of many small particles which have not been resolved because of an inadequare depth of focus. The presence of undesired small particles is evidenr in Figures 2B and 2C. A major disadvantage of the elutriation method of size separation is the amount of time required for the collection of each sample. A preliminary grading made by the Universal Road Machinery Co., Kingston, N. Y.: saved a great deal of time. Even so, nearly 16 hours has been required to produce the 100 grams of powder contained in each size range. I n its basic operation the apparatus constructed for the combustion tests of the Armowax powder follows the design given by Ghosh (70). Several refinements have been included. Particles less than about 50 ,u have a great tendency to stick to the walls of the tube even in the presence of an intense sound field. Electrical grounding of the combustion bomb coupled with continuous hammering of the tube walls reduces the quantity of adhered particles to an acceptable minimum. Great effort has been made to eliminate regions of low air velocity and recirculation. The propeller motor has been placed outside the
tube, while all re-entry corners have been rounded. The use of stator vanes directly above the propeller was necessary to eliminate vortex flow. ~i~~~~ 3 is a sketch of the combustion bomb. T h e sound field is produced by means of a 30-watt driver connected in series with a n oscillator and amplifier. Figure 4 is a photograph of this equipment. A resonant frequency of just under 500 cycles per second was used in all tests. T h e bomb contains 5 grams of air at atmospheric conditions. T h e powder is added batchwise after weighing on a n analytical balance. T h e airpowder mixture is circulated around the loop by the propeller which is directly connected to a 17,000 r.p.m. electric motor. After several seconds a uniform mixture is ensured. T h e propeller is turned off. A delay period from 2 to 4 seconds allows the mixture velocity to be reduced to a low enough value for easy ignition. Several runs with different delay periods may be required to determine the maximum possible pressure rise in the bomb. T h e pressure is recorded on a motorized Crosby indicator gage. Several different methods of ignition were investigated. T h e ignition energies required by suspensions of solids are much higher than those required by gaseous mixtures, Automotive-type spark ignition was inadequate to give reproducible results. T h e intense arc formed after the vaporization of a small metal wire proved most satisfactory. Steel wire, O.Ol2-inch diameter and '/4 inch between electrodes, will produce such an arc when short-circuited 120 volts a.c. When lean mixtures of larger particle sizes are used, a single ignition source may produce burning in only one branch of the combustion loop. Ignition sources in each of the two main branches obviates this difficulty.
Discussion of Results h c h size fraction of Armowax POWder has been tested for its lean limit of flammability and pressure-concentration Curve. Figure 5 dmws the Pressure curve obtained for the 60- to 70-p range. T h e degree of scatter of the Plotted Points is relatively large. However, curves in each case have been determined by the maximum Pressures. Lower values of Pressure have been disregarded. Optimum conditions for CO~~-~bustion are believed to be Present when the upward Stream velocity Past one ignition Source balances the terminal settling velocity of the Particles. This simple Picture is complicated by random air motion, and several trial tests may have to be de to determine the maximum Possible Pressure rise. While the air Passes upward past one ignition wire, it Passes downward in relation to the other wire. T h e velocity of the Particks Past this ignition ~OUrCeis higher than optimum. Thus, for mixtures m a r the lower limit of only one wire effective in igniting the mixture* The pressure curves for all size fractions successfully tested are included in Figure 6 for comparison. T h e maximum Pressure is inversely ProPOrtional to Particle size. Peak pressure is, in each case, attained a t a mixture well above stoichiometric. T h e curves indicate that the larger particle sizes will sustain high pressures over a greater fuel concentration range than will the small partides. I n fact, the curves for the two smallest size fractions cross those for the larger sizes. I t is doubtful that a distinct rich limit is obtained for these suspensions as in a gaseous fuel. Lack of a sufficient quantity of graded powder limited tests to fuel concentrations below 5570 by weight in the mixture. However, this represents an air-fuel ratio
of 0.82: 1. Comparison of the Armowax results with those of gaseous propane shows the wide range of flammability limits of solid (or liquid) suspensions. The maximum pressure for a propaneair mixture attained in the apparatus is 80 pounds per square inch gage. This value is nearly 20 pounds per square inch less than that which would be produced in a conventional bomb. Two factors contribute to this lower value. T h e combustion loop has a much larger surface area enclosed volume ratio. Radiation effects are larger. Also, the check valve which protects the sonic driver against high pressure rises in the loop by-passes a portion of the combustible mixture to the atmosphere. I t could be assumed that the pressure for the solid suspensions would actually be 25% greater if ignited in the more conventional type apparatus. A plot of maximum pressure rise as a function of particle diameter is shown in Figure 7. Results obtained for combustion in the loop are shown by the solid line. T h e dashed line represenrs the pressures to be expected in the type of apparatus used by Hartmann and others. overthe size range tested the maximum pressure falls nearly linearly with increase of particle diameter. Extrapolation of this curve would indicate that particles above 200 p would experience pressure rise' This does not contradict the ObHartmann who created extained plosions in suspensions of particles as great as 700 p in diameter (12). I t may well be that a point is reached where radiation from the bomb is about equal to the heat energy added by combustion. For such large particles pressure rise is a poor indication of the presence of combustion. T h e higher pressures obtained by the
STOICHKIMETRIC
STOICHIOMETRIC 0
PERCENT FUEL
BY WEIGHT
Figure 5. Pressure-concentration curve obtained for Armowax particles between 60 and 70 microns in diameter
P E R C E N T FUEL B Y WEIGHT Figure 6. Pressure-concentration curves for all size fractions tested VOL. 49, NO. 1
JANUARY 1957
145
Figure
7. Maximum pressure-particle diameter curves
Solid line for results obtained in combustion loop; dashed line for results to be expected using conventional combustion bomb
combustion of the smaller particle sizes can be explained on a basis of high surface area volume ratios. Imagine a uniform flame being propagated into a mixture of solid particles in air. For a compound such as Armowax the solid fuel must be vaporized prior to combustion. A small portion of this heat for vaporization is transferred by radiation ahead of the flame to the particles. However, a greater portion of the required heat is probably supplied by conduction into the mixture ahead of the flame. I n both cases the amount of heat which can be transferred to the particles per unit time is a function of the available surface area of the particles. As the smaller particles present a larger surface per
pound of fuel, it would be expected that vaporization would be more rapid. T h e flame can travel faster with a resulting greater pressure rise. I t is quite possible that factors other than radiation and burning velocity are of importance in causing the larger size suspensions to have lower peak pressures. Different product distributions would produce varying heat capacities. Also. the combustion of the larger particles leads to the formation of considerable quantities of unburned residue. T h e completion of reaction of these suspensions is inversely proportional to the size of particles. T h e high surface area of small particles explains their narrower combustion
limits. I n the combustion of large particles vaporization is relatively slow. It is a simple matter for a favorable airfuel ratio to be formed ahead of the flame even in the richest mixtures of solid fuel in air. As the particles become smaller vaporization is more rapid. For the same mass concentration of solid particles a richer vapor-air mixture is preseni ahead of the flame. and the flame mav well be adversely affected. For kerosine particles of submicron size the rich portion of the pressure-concentration curve is more similar to that of a vapor than to the results presented in Figure 6 ( 4 ) . Maximum flame velocities and pressure rises occur in mixtures well above stoichiometric. Even for the smallest sizes tested vaporization rates are not sufficiently high to produce uniform mixtures of vapor and air ahead of the flame. Burgoyne and Cohen (5) have found that liquid droplets must be smaller than 10 p to support combustion similar to that of a gaseous fuel system. Great difficulty \vas experienced in the ignition of the 0- to 10-pfraction. In fact, out of nearly 20 trials only three ignited. These three runs gave such low pressure rises that the results are considered inconclusive. A possible explanation of such poor ignitability may lie in the fact that a considerable random motion of the air exists even after the fan is turned off. These small particles easily follow such air currents. An additional disturbance arises from the vaporization of the ignition wire which displaces the particles beyond the region of most intense heat. Alternate methods of ignition such as sparks and hot wires were tried. but to no avail. A limitation became evident as to the maximum particle size which could be successfully tested. Above 90 p a homogeneous suspension could not be formed. A more powerful fan system could extend this upper range. Figure 8 is a detailed graph of the results obtained near the lean limits of combustion of the different suspensions. Particle size has a pronounced effect on the lean limit. T h e lean limits found for the various size fractions are: Particle Diam.,
Figure 8. Detail graph of pressure-concentration curves near lean limit of combustion
146
INDUSTRIAL AND ENGINEERING CHEMISTRY
Fuel,
P
70
BIF
10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90
3.6 3.3 2.7 2.4 1.9 1.7 3.0 3.3
27 28 36 41 57 58 32 29
Particle diameters of about 60 p produce the most stable systems in relation to the lean limit of flammability. Both smaller and larger sizes are less favorable. An explanation of this reversal may be attributed to a balance between two opposing trends. In suspensions composed of
large particles the flame progresses from one droplet to another across a relatively vapor-free space. Each droplet burns to form a diffusion flame, and a sufficient amount of heat must be transferred from particle to particle to allow the flame to progress through the entire mixture. At constant air-fcel ratio a reduction in the size of large particles produces a more closely spaced suspension. Above 60 1 this closer spacing gives conditions more favorable for this type of diffusion flame propagation. As the particle size is reduced higher vaporization rates arise, and a larger amount of the fuel is transformed into vapor. This vapor in turn is diffused into the spaces between the droplets. T h e vapor tends to burn in the same fashion as a gaseous fuel, while the particles which are not completely vaporized burn as diffusion flames. A mixed type of combustion results even though ignition might well be sustained by droplet-to-droplet propagation alone. As long as the flame is predominantly of diffusional character the production of excess vapor prior to combustion is a loss. Thus, more fuel must be available for combustion to proceed. As the droplet size approaches 10 1 the vapor concentration becomes greater between the droplets until a point is reached where successful combustion of the mixture is completely dependent on its gaseous state. The ability of the 60-h size suspension to burn a t the leanest air-fuel ratios is thus brought about by a balance between the rate of vapor production and the ability of the flame to progress from droplet to droplet. T h e curves in Figure 8 show that the rate of pressure rise as a function of fuel concentration is more sensitive for the smaller size fractions. For these small particles the flame velocity is dependent on the concentration of fuel vapor, in which case a slight increase in vapor content can greatly increase the reaction velocity. T h e curves for sizes larger
f
than 60 /I are nearly linear, indicating that particle spacing is the predominant factor. T h e solid curve of Figure 9 is a plot of the lean limit of flammability as a function of particle size. This curve rises gradually to a maximum air-fuel ratio 58: 1 for 60-p particles and falls quite rapidly for larger sizes. T h e dashed curve representing the results of Burgoyne and Cohen (5) shows a n even wider spread of the lean limit. This may be due to the fart that their use of an aerosol generator produced a nearly quiescent suspension. Thus, there would be little influence of air motion on the ignition of the mixture. Also, the ignition energy for liquid suspensions is lower than that for solids. Anson’s results, which were obtained for a rapidly moving heterogeneous spray, are also shown. T h e high turbulence level in his experiments considerably reduces the influence of particle size on the lean limit. Conclusions Although turbulent effects produce the major influence on the lean limit of combustion of sprays and solid suspensions, the effect of particle size is important. I t might well be possible to extend the lean limits of stable spray combustion by means of better particle size control. Evidence of this effect is shown by the greater stability of turbojet systems containing a portion of large droplets. T h e use of carefully graded wax, or similar low melting point compounds, produces suspensions whose characteristics are quite similar to those of liquid suspensions. These solids are easily handled and a wider range of experimental knowledge can be obtained by their use. Further investigation of these solid suspensions should lead to important conclusions. Several fields of interest are evident, and the following
BURGOYNE 8 COHEN /---
.&
, PROPANE-A I R
-----20
- - - ----A . ANSON
\
PARTICLE DIAMETER, MICRON Figure
9.
Lean limit air-fuel ratio as function of particle size
questions arise. What is the effect of including a certain portion of 60-/L particles in a suspension mainly composed of larger or of smaller particles? W h a t influence does particle size have on the combustion of solid fuels which burn by means of solid-gas reactions? Is the critical size of 60 p found in these experiments a function of geometry alone? If not, how important a role does fuel volatility play? Is there a maximum size particle above which combustion rates are so low as to prove impractical in commercial devices? T h e answers to these questions should furnish a n important step toward the further understanding of this difficult subject. literature Cited Anson, D., Fuel 32, No. 1, 39-51, (1953). Blanchard, D. S., Woods Hole Oceanographic Inst., Tech. Rept. 8. Ref. No. 54-27. Mav 1953. Bolt, J. A., Boyle, T. A.,’Mirsky, W., Am. SOC.Mech. Engrs. Paper 55SA-67. Browning, J. A , , Krall, W. G., “Effect of Fuel Droplets on Flame Stability, Flame Velocity, and Inflammability Limits,” Fifth Combustion Symposium, Reinhold, New York, in mess. Burgoyne, 3. H., Cohen, L., Proc. Roy. SOC.225A, NO. 1162, 375-92 (1954). Cassel, H. M., Das Gupta, A. K., Guruswamy, S., Third Symposium on, Combustion, pp. 185-90, Williams and Wilkins, Baltimore, Md., 1949. .. . _ .
Craig, O., Trans. Am. SOC.Mech. Engrs. 61, 369-72 (1939). Dimmock, N. A., National Gas Turbine Establishment, M115, May 1951. Dreyhaupt, F., Deutschekraftforschung Tech. Forschungsbericht Zwischenbericht I l l , 75-88 (1942). Ghosh, B., Orning, A. A., IND.ENG. CHEM.47, 117-21 (1955). Gibbons, L. C., Jonash, E. R., Am. SOC.Mech. Engrs. Paper 48-A-104, 1948. Hartmann, I., IND.ENG.CHEM.40, 752-8 (1948). Hartmann, I., Nagy, J., U. S. Bur. Mines, Rept. 3751, 1944. Hartmann, I., Nagy, J., Brown, H. R., U. S. Bur. Mines Rept. Invest. 3722. 1943. Mullen, J. W.; 11, Fenn, J. B., Garmon, R. C., IND.ENG.CHEM.43, 195-211 (1951). Pilcher, J. M., Rodman, C. W., Battelle Memorial Inst., Tech. Rept. 15032-1, August 1953. Rivitre, M., Mayorcas, R., J . Inst. Fuel 26, 211-24 (1953). Vonnegut, B., Neubauer, R. L., General Electric Research Laboratory, Occasional Rept. 36 (Oct. 1,1952). Walton, W. H., Prewett, W. C., Proc. Phys. Soc., Sec. B, 62, part 6 (June 1949). Whittle, J., Fuel 33, No. 4, 192-4 (1954). RECEIVED for review September 18, 1955 ACCEPTED July 2, 1956 This research was conducted under the auspices of Project SQUID, jointly sponsored by the Office of Naval Research and the Office of Air Research under Contract N6ori-105 T.O. 111. VOL. 49, NO. 1
JANUARY 1957
147
