Influence o f Phgsieul Fuctors in
-.
Igniting Pulverized Coal BIMALENDU GHOSH' AND A. A. ORNING Coal Research Laboratory, Carnegie Institute of Technology, Pittsburgh, Pa.
THE
metrically by lifting the feed system and burner tube from the furnace and collecting the coal for l minute by aspiration into a filter. Loss of coal in circulation, by adhesion to the tube walls, was avoided by use of sonic energy. A 30-watt University driver unit, D of Figure 1, was used for this purpose. Power was supplied t o thc driver unit by a n oscillator, the circuit diagram for which is shown in Figure 2. The frequency could be varied from 500 to 5000 cycles per second and the wave form was that of a square wave within a sinusoidal, 60-cycle envelope. This wave form, with frequency variation in each 60-cycle wave, eliminated fixed nodal points where coal might adhere.
+ design of pulverized coal burners has developed largely as an industrial art. Coal characteristics, fineness of grinding, distribution of coal in the air stream, heat loss from the flame, and the aerodynamic pattern are factors that influence ignition and flame stability in pulverized coal burners. The interactions of these factors, and even the manner in which they individually influence the flame, are not fully understood. This arises in part because of opposing influences upon energy release in the flame and upon energy transfer processes t h a t are responsible for ignition of the coal as it enters the flame. The purpose of the investigation was t o study the influence of the different factors upon the ignition process independent of the influence of the same factors upon the resulting flame. For this purpose small flames were produced in an electrically heated furnace, so that most of the heat needed for ignition came from the furnace a t an independently controlled temperature rather than from the flame.
TABLE I. PROXIMATE ANALYSES OF COALSUSED (As-received basis) Seam Pittsburgh Pocahontas No. 3 Illinois No. 6
County Fayette, P a . McDowell, W . Va. Franklin, Ill
Moisture 1.6 0.8 6.7
Volatile Matter 33.8
15.3
33.9
Ash 7 5 3.6 Q.1
B t u. 13,910 14,760 12,270
EXPERIXIENTAL
The apparatus used (Figure 1) consisted of a n electrically The furnace was heated to the desired temperature ( 5 1 0 ' C.) heated furnace, F : and a circulatory system for suspending coal and 2 t o 3 grams of pulverized coal was put into the circulatory in air and feedinn it to the furnace. T h e furnace was made of system. A metered supply of dry air refractory insulaiing brick, Armstrong was then used t o expel the coal suspenA-26, set in a tight steel case with a n sion through the burner tube and into open bottom. T h e furnace cavity was the furnace. T h e coal flow was periodi4 x 4 x 1.5 inches high. Heat was supcallv redetermined with an air flow rate plied by two 0.5-inch Globars placed horizontally on 2.5-inch centers in the such that the flame was kept 0.iB inch below the tip of the burner tube. mid-plane of the furnace cavity. A thermocouple was placed with its junction a t the mid-point between the GloThe ignition time was calculated as bare, 0.75 inch from the ceiling of the the transit time from the burner tip t o furnace cavity, and 1.0 inch from the the flame front, assuming that the transit axis of the coal feed tube. The couple velocity was equal to the average vewas made of 16-gage Chrome1 and Alume1 xires with an exposed junction exlocity within the burner tube. Ignition tended 0.125 inch from its supporting times were determined for coals of diftube. The burner tube was a stainless fering ranks, for which proximate steel tube 12 inches long and 0.120 inch analyses are given in Table I. The igin inside diameter. The tip of the tube extended 0.5 inch beyond the end of a nition times were determined as a funcwater-cooled jacket and, in the operattion of coal concentration, furnace teming position, the tip was flush with the perature, intensity of sonic energy, fineceiling of the furnace cavity. A peepness of grinding, and furnace atmoshole in the front wall of the furnace was used for observing the flame. Vertical phere. The data obtained are presented displacement of the flame could be estias plots of ignition time against ratio of mated within 0.05 inch by sighting coal t o air in Figures 3 to 10. against the t,herniocouple tip. The circulatory system was composed of a fan housing 6 inches in diameter and F DISCUSSIOS a circuit of 1.25- and 2.0-inch streamline tubing. -4 vacuum sweeper fan and For each of the coals it, was found motor were totally enclosed. The inlet that the appearance of the flame deend of the burner tube was placed on the axis of the 2.0-inch tube, seven pipe dipended upon the concentration of coal ameters below a 1.25- to 2.0-inch couin air. At l o ~ rconcentrations, individpling. The circulatory system mas airual particles ignited a t random and tight, so that air flow through the burner Iiigure 1. Schematic Diagram of contributed little to the heat transfer tube, B , could be metered a: the air inlet, Apparatus A . Coal ivas added batchwise. Coal flow needed for ignition of their neighbors. through the burner wasdetermined graviA. .4ir inlet With intermediate concentrations, the Present address, Chemical Engineering Department, Jadavpur College, Calcutta 32, India. 1
B.
Burner tube
I.".
Furnace
BL. Blower D . Audio driver unit
117
traces of individual particles were visible throughout the flame, but burn-
INDUSTRIAL A N D ENGINEERING CHEMISTRY
118
m.
Diagram of Oscillator Circuit
Figure 2.
I, 11, 111,IV. 6LGG A. Radio-frequency ammeter C. 0.02-mfd. D. University PA 30 driver unit P. 100-ohm 26-watt 100,000-ohm Helipot, 2 units ganged 7. R. 15000-ohm VCI' T1. Tgordarson T22R07 power transformer, 110,350-0-350,6.3 T 2 . Varimatoh output transformer, C V P2, tapped f o r 5000-ohm primary and 16-ohm secondary
ing particles were close enough t o assist in ignition of their neighbors. At higher concentrations, a continuous flame front, probably of burning volatile matter, assured immediate ignition of particles entering the flame front. Another limiting condition was found a t stil.1 higher concentrations, where recirculation filled the furnace cavity with flame so that the flame front could no longer be observed. The transitions mere too gradual for precise estimation of critical concentrations. T h e curvea of ignition time us. coal concentration (Figures 3 to 10) cover the approximate range within which well-defined flame fronts could be observed. Various uncertainties entered into estimation of ignition times. The jet of air and coal flowed down from the cooled burner tube through the hot atmosphere in the furnace. Acceleration, due to differences in density, was opposed by drag between the jet
VOl. 47, No. 1
and the atmosphere of the furnace. The observed flame size indicated little expansion of the jet or mixing with the surroundings. It was, therefore, assumed that t'he average velocity, between the tip of the tube and the flame front, was equal to the average velocity within the feed tube. Ignition times, calculated on this basis, may be subject to absolute error approaching a factor of 2, but, the relative values are probably good within a few per cent. Init,ial studies showed a ret'arding effect of sonic energy on the ignition process. As shown in Figure 3, ignit,ion of -200-mesh Pittsburgh Seam coal was retarded when po\Ter input to t,he driver unit was 18 rTatts at 480 cycles per second, while experiments at, 7-a-att input shoved slight ret,ardation as compared with experiments a t 1-watt input. Similar experiments a t 1200 and 3000 cycles per second showed no effect. Attenuation of sonic energy in a channel decreases with decreasing frequency: so that, with the low frequency and higher poxer levels, there was sufficient sonic energy in the jet leaving the coal feed tube to influence the ignition process. I n subsequent experiments the unit was operated with 15-1Tatt power input a t 1200 cycles per second wherc the sonic intensity was not a significant variablr. Figures 4 t o 6 show the influence of the rank of coal upon the igriit#iorit,ime. The Illinois coal, Figure 4: was most readily ignited. Flames could be maint,ained with furnace t,emperatures as low as 900" C. and with SOYo excess air. The Pittsburgh Seam coal (Figure 5 ) also gave a good flame a t 900" C. furnace temperature, but, no excess air over the theoretical amount needed for complete combustion could be used. The curves for the Pocahontas coal (Figure 6 ) were shaped like those for lhc Pittsburgh Seam coal, but the temperatures had to be increased about 100" C. and the air supply had to be limited to not more than 70y0ol theoretical. An effect of particle size is shown in Figure 7 for 100- to 200mesh Pitt,sluurgh Seam coal. The curves show significant' shifta toward higher ignition times and higher coal concentrations needed for good flames, when compared with the curves in Figure 5 for -200-mesh samples of the same coal. Similar cffecta were found for the Pocahontas and t,he Illinois coals, but the curve a t 900" C. was not, so nearly independent of coal concentration. The effect of particle size was studied further by using three size fractions prepared from - 200-mesh Pit,tsburgh Seam coal. Size distributioris \-,we determined by microscopic count and calculated t,o a weight basis. The coarse fraction was 50% over BO and 7L5% over 46-micron size. The intermediate fraction waB about 50% over 50 and 75% over 35 microns. The fine fraction
PERCENT OF THEORETICAL AIR I50
100 I
I
r--x\
80
A0
50
60
1
X
I
I
I
I
t
I
I
I
I
I ' lo 6
I 8
I
I
12
14
16
18
COAL CONCENTRATION, (LB./CU.FT.)X
Figure 4.
20
lo3
Ignition Time us. Coal Concentration -200-mesh Illinois No. 6 Seam coal Furnace temperature 900"
x
11000
c. c.
INDUSTRIAL AND ENGINEERING CHEMISTRY
January 1955
was 50% over 6 with none over 20 microns. Figure 8 shows the ignition curves a t 1000' C. furnace temperature for each size fraction. The finest showed no greater ease of ignition than that for the intermediate size fraction. This may be due to the dependence upon particle size of the balance between the rates of heat transfer t o the coal by radiation and rates of heat loss from the coal particle to surrounding air by conduction (4,I O ) It is also possible that the coal particles were not thoroughly dispersed in the air, as some evidence of clustering was found when a sample of the finest coal was collected on a microscope slide by passing it through the jet of coal in air issuing from the feed tube.
80
100
PERCENT OF THEORETICAL AIR 60 50 40 35 \
30 I
,
A,'
119
the temperature of a coal particle, t, will rise at first according to the relation.
where c is the heat capacity, p is the density, d is the diameter of the coal particle, 0 is the time, and u is the radiation constant. As the particle temperature rises, heat is lost t o the surrounding atmosphere and the rate of rise falls off. The modes of heat transfer reach a quasi equilibrium after a period of time, 8,, given by the relation
where k is the thermal conductivity of the atmosphere and N is the ratio of heat capacities of the associated quantities of atmosphere and coal. Thereafter the particle temperature continues to rise according t o the approximate relation
Nusselt's estimate differed from that of Traustel in t h a t a limited amount of excess air was not recognized. Under these conditions, N = m and the last relation becomes
t
= t o + - - *
OF
PERCENT
I
6I
8I
12 I
10 I
1I4
1I6
II0
C O A L CONCENTRATION, ( L B . / c u . F T . ) x
Figure 5.
d
20
103
ad
(4)
2k
TkEORETICAL AIR
:
Ignition Time us. Coal Concentration -200-mesh Pittsburgh Seam coal Furnace temperature
9000 c. 1000" c. 11000 c. Dashed curves calculated relative to 1000° C. curve, A A and D D on basis of heat transfer from furnace only, BB and CC on basis of heat transfer from furnace and from flame
2x
The influence of furnace atmosphere upon ignition was studied by replacing the air in the furnace cavity by oxygen and by nitrogen. Figure 9 shows resulting data on ignition time us. coal concentration for the different furnace atmospheres with -200-mesh Pittsburgh Seam coal at 1000 O C. Figure 10 shows similar data with the same coal and furnace temperature and with air as the furnace atmosphere, but with oxygen enrichment in the primary stream of coal and air. Oxygen enrichment, in the primary coal and air stream, accelerated ignition, while enrichment of the furnace atmosphere had little effect except at low coal concentrations. This was interpreted as evidence for little mixing of furnace atmosphere with t h e jet of coal and air prior t o ignition. The transfer of thermal energy b y mixing was correspondingly small as compared with tl'ansfer by radiation. Interpretation of these experimental observations requires an analysis of the energy transfer processes involved in ignition of the coal. Nusselt (4)and Traustel ( I O ) have made theoretieal estimates of the rate of rise of temperature of coal particles prior to ignition. It was assumed that heat was transferred t o the coal by radiation and lost from the coal to the immediately surrounding atmosphere by conduction. The relations between time and temperature are simple in two stages of t h e process (6, 6): first, immediately after the coal suspended in cool air enters the heated space, and again after the two heat-transfer processes have reached a quasi equilibrium. If the coal and air enter a t the same temperature, to, a t zero time, and are exposed t o radiation from all directions a t the absolute temperature, T,,
30'
Io
;1
Ib
I't3
;1
210
2;
2'4
C O A L CONCENTRATION, (LB./CU.FT.) x i 0 3
Figure 6.
Ignition Time us. Coal Concentration -200-mesh Pocahontas No. 3 Beam coal Furnace temperature
2x
;12000 E% g: c.
This analysis shows that the rate of rise of temperature is lower
if the thermal conductivity of the atmosphere is increased. It was found experimentally that ignition was retarded when sonic vibrations increased the rate of thermal conduction into air surrounding the coal particles. If Nusselt's assumptions were correct, the time required for ignition should be inversely proportional to the fourth power of the absolute temperature of the source of radiation responsible for ignition. This hypothesis was tested for the data of Figure 5. Assuming the effective radiant temperature to be that indicated by the furnace thermocouple, and basing the proportionality on the 1000' C. curve, curves for 900' and 1100' C. were found by calculation t o be as indicated by the dashed curves A4 and DD, respectively. If it is assumed that the coal received radiant energy through half the total solid angle from the furnace and through the other half from the flame at 1500' C., these curves
INDUSTRIAL A N D E N G I N E E R I N G C H E M I S T R Y
120
PERCENT OF THEORETICAL AIR 60
80
cn n z
55.
50
PERCENT OF THEORETICAL AIR 30
35
40
W
,
50
60
35
40
1
I
45
I
--
Lo
80
'00
1
8
8 i
Vol. 47, No. 1
50-
2 W'
I
45-
z
0 i-
40-
w
2
6 MICRON
I
12
IO
14
I
I
16
,
b
20
I8
22
I8 , ,
24
I
COAL CONCENTRATION, ( L B / C U F T ) X I 0 3
Ignition Time ws. Coal Concentration
Figure 7.
x
c. c.
would fall in the position of curves B R and CC. This shows t h a t the desired condition of having the walls of the furnace supplj- all the heat needed for ignition was not ach~eved. A substantial part of the heat transfer for ignition came from the flame. Radiation froin the furnace, nevertheless, was important. Below a critical tenipeiature, which could be determined approximately for each coal, individual particles ignited a t random without a definite flame front The curves of ignition time us coal concentration depended in p u t upon furnacr temperature and in part upon a flame tempei ature that inust have been i elativcly constant n i t b changing furnace temperature. The rate of heat tranqfer from the flaine depends upon the solid angle from the coal puticle to the flame If a coal particle moves n i t h constant velocity, u, along the axis toward a flame front JT-ith diameter, a,the total heat, q, transferred to thc particle as it inovet from a n infinite distance to the flame front is given by the relation.
,
(5)
C(U/ZL)
PERCENT OF THEORETICAL 150
Bp
I?
6f
AIR
5p
I
I
FT.) x
io3
Air-elutriated Eractions of Pittsburgh Beam coal, l o O O ' C furnace temperature, mean particle sizes a8 indicated
9000 1000° C. 11000
q =
I
Ignition Time z's. Coal Concentration
Figure 8.
100- to ZOO-mesh Pittsburgh Seam coal Furnace temperature
0 A
I
COAL CONCENTRATION,(LB./CU.
40
3,5
,
where variables independent of angle factors have been included in the constant of proportionality, c. T h k expression is valid within the range in which the particle temperature is low as compared with the temperature of t,he radiant source, so that backradiation can be neglected. The total heat transfer to the part,icle remains constant, provided the flame size is increased in proportion t o the approach velocity. As the portion of the heat t>ransferredby conduction from the particle t o the surrounding air, and hence not available for ignition, is dependent only on local conditions and not on the flame size, TCqua,tion5 est,ablislles a condition of similitude. inch in diameter In the present work the flame was about and the approach velocity was on the order of 2 feet per second. On the basis of Equation 5 , this would correspond to a ve1ocit)yof 96 feet per second for a flame 6 inches in diameter. With possibly half the heat supplied by the furnace rather than by the small flame, t,he velocity for the 6-inch flame would be on the order of 50 feet per second. This estimat,e is in accord wit.h industrial practice, but there are no published data that permit a sound comparison. A set of curves showing flame velocit,y as a function of the ratio of coal to air and the proximate analysis of the coal was prepared by DeGrey ( 2 ) and has been reproduced
I
X
I
4
I 6
I
8
I
I
I
I
10
12
14
16
COAL CONCENTRATION, (LE. /CU. FT. 1 X
Figure 9.
I I8
I
20
'
1
6
-200-mesh Pittsburgh Beam coal, 1000° C. furnace temperature 0 Oxygen X Air
I
8
IO
12
I4
COAL CONCENTRATION
IO'
Ignition Time in Various Furnace -4tmospheres
Nitrogen
PERCENT OF THEORETICAL OXYGEN I
Figure 10.
6
cie./cu.
8
22
20
FT.) x
'
to3
Effect of Oxygen Enrichment in Coal Stream on Ignition Time
-200-mea11 Pittsburgh Seam coal, 1000° C. furnace tempeiaiiire 0 30% oxygen X Ai;, 0 40,c oxygen
January 1955
INDUSTRIAL AND ENGINEERING CHEMISTRY
( I , 3 ) a t times with transformed coordinates. The curves were apparently based on experience with rock dusting as a means for controlling coal dust explosions in mines and upon experiments by Taffanel and associates (7-9) w t h coal dust flames in a refractorylined conical tube. The experimental data were not sufficient to establish quantitative relations with proximate analyses. Nevertheless, DeGrey presented curves for coals of various volatile matter and ash contents. The curves showed maximum flame velocities a t dispereions ranging from about 40 to 60 cubic feet of air per pound of coal. Curves from the present investigation lie entirely on the lean side of the maxima shon-n by DeGrey. Furthermore, they do not indicate the high ratio of flame velocities, nearly 3 to 1, for a coal of 30% compared with 15% volatile matter content, that was shown in the earlier curves. The strong influence of volatile matter content, in the earlier estimates of flame velocities, was probably based upon experience with explosions in mine galleries, where concentration of energy in the moving flame front depends more upon rapid combustion of volatile matter than upon sloTver and probably incomplete combustion of fixed carbon. The present investigation indicates that a higher furnace temperature is needed to establish a flame with the coal of lower volatile matter content, but that the flame velocity is not so markedly dependent upon volatile matter content. Limitation of primary air is a commonly practiced means for getting good ignition with coals of lower volatile matter content. According to Traustel’s analysis of the heattransfer system, the improvements result, a t least in part, from the reduced thermal load resulting from the restricted amount of air associated with the coal prior to the point of ignition. It may also result, as is inherently recognized in Traustel’s as well as Nusselt’s analyses, from an increased radiant intensity of the flame. A study of the ignition of individual particles in a pressured atmosphere ( 5 ) showed that convective heat transfer and the “time-dependent” rather than the “steady-state” form of the equations for the heat-transfer system from the coal to the surrounding atmosphere, must be considered. iz hindering effect of pressure was observed which was not readily overcome by oxygen enrichment. It could be explained on the basis of the idea that, within the short time prior to ignition, the thermal or the convective yave advanced only into a limited portion of the at-
121
mosphere associated with the coal. If such a limitation exists, the limitation given by Traustel for the heat capacity of the entire amount of associated atmosphere mould seem of secondary importance. Though the analyses of Nusselt and Traustel may be questioned on theoretical grounds, the importance of the effects with which they deal has been substantiated experimentally. The ignition process involves heat transfer to the coal by radiation and heat loss by conduction into the immediately surrounding atmosphere. Anything that changes the rates of these processes will change the rate of flame propagation. Because radiant heat transfer through transparent media depends upon angle factors, but not on distance, the propagation of a pulverized coal flame depends upon the size and geometry of the flame a9 well as upon the fuel distribution and the aerodynamic pattern. The pulverized coal flame differs in this rerpect from gaseous flames. In so far as fuel gases are transparent to thermal radiation, Aame propagation depends upon energy transfer from zones within or immediately adjacent to that portion of the flame front under consideration. While the propagation of a gaseous flame depends upon local conditions, the propagation of a pulverized coal flame depends upon the flow of radiant energy throughout the flame. LITERATURE CITED
Craig, O., Trans. Am,. SOC.Mech. Engrs., 61,369-72 (1939). (2) DeGrey, M. A.,Reo. nadt., 19, 645-55 (1922). (3) Kreisinger, H., Proe. Engrs. SOC.West. Penn., 39, 244-9 (1923). (4) Nusselt, W., Z . V e r . d e u f . Ing., 68, 124-8 (1924). (5) Omori, T. T., and Oming, A. A , Trans. A m . Soc. Mech. Enqrs., 72, 591-7 (1950). (G) Orning, A. A., “Combustion of Pulverized Coal.” pp. 1522-87, Chap. 34 of “Chemistry of Coal Utilization,” H. €I. Lowry, (1)
ed., New York, John Wiley & Sons, 1045. (7) Taffanel, J., Ann. mines, Mirn., SQrie11, 2, 167-205 (1912). (8) Taffanel, J., and Durr, il., Colliery Guardian, 103,227. (9) Taffanel, J., and Durr, A., Cornit6 central des houillbes de France, Paris, 1911. (10) Traustel, S., Feuerungstech., 29, 1-6, 25--31, 49-60 (1941). ACCEPTED October 12, 1951. Based on a D.Sc. thesis a t the Carnegie Institute of Technology entitled “A Study of the Conditions Necessary for Producing Small Flames with Pulverized Coal“ by Bimalendu Ghosh, 1951.
RECEIVED for review June 16, 1954.
Flame-Stability Studies on an Inverted Bunsen Burner d
PHILIP F. KURZ Battelle Memorial Institute, Columbus 1 , Ohio
F
LAME-stability studies have been carried out on an inverted Bunsen burner. The experimental work dealt with a comparison of the stability limits of upright and inverted flames of single and mixed hydrocarbon fuels and with the influences of nitrogen on the stability limits of hydrocarbon-air mixtures. DESCRIPTION O F APPARATUS
Figure 1 shows a schematic diagram of the inverted Bunsen burner and the auxiliaries for metering gas flows, for quenching flash back, and for cooling the burner tube. The burner tube is of brass, 19.1 mm. in inside diameter and about 122 em. long. The water jacket extends upward about 107 cm. from the plane of the burner port. Cooling water is circulated through the jacket by means of a pump. Effluent water is returned to a sump for recirculation. Flash backs were quenched by introducing nitrogen into the fuel-air mixture a t the inlet to the burner tube.
A Smithells tube to exclude ambient air from the environs of the burner port is essential. Unshielded flames curl upward and move about under the influence of random currents in the ambient air. Fuel and air flows were metered by means of critical flow orifices. Compressed air from commercial cylinders was used in the experimental work in order to ensure close control over the air flow a t all times. DESCRIPTION O F IlriVERTED FLAMES
Figure 2 shows sketches of lean and rich hydrocarbon-air flames a t the last stable composition prior to blowoff on the inverted Bunsen burner. The sketches represent flames produced a t moderate air rates-7 to 9 liters per minute. B t lower air rates the flames are shorter and are more rounded at the tip, and a t high air rates the flames are longer and more pointed. Lean ethylene flames are essentially the same as lean flames of other hydrocarbons. However, because shielded rich ethylene