- Email: [email protected]
Fragmentation behavior of single coal particles in a fluidized bed
Twentieth Symposium (International) on Combustion/The Combustion Institute, 1984/pp. 1495-1503
FRAGMENTATION
BEHAVIOR OF SINGLE IN A FLUIDIZED BED
C. A. SUNDBACK, J. M. BEIs
COAL
PARTICLES
A. F. SAROFIM
Department of Chemical Engineering Massachusetts Institute of Technology Cambridge, Massachusetts 02139 The fragmentation hehavinr of single coal particles has been studied in a small scale fluidized bed for a Kentucky No. 9 bitumim)us coal. A typical coal particle fragments and swells slightly during dcvolatilization, producing particles in two distinct size regimes. Although the fragmentation pattern varies from particle to particle, the bimCxtal mass-based char size distribution obtained for the pyrolysis of a collection of coal particles was found to be little influenced by changes in bed temperature (1023-1123 K), oxygen concentration (0-10%), and co',d particle size (2.2-6.2 mm). l)uring oxidation of single coal particles, the dramatic increases in burning rate following particle breakup made it possible to infer fragmentation from the carbon dioxide concentration variation with time of the gas effluent. The number and size of the coal fragments were deter,nined by fitting the CO, data with the predictions of a char combustion model. Swelling and fiagmentation contribute to an approximately twofold reduction in the average, hurning time at the experimental conditions studied.
Introduction Predictions of'the carbon content in fluidized beds are important becanse carbon losses from the bed are proportional to the carbon loading, and because of carbon's role in the reduction of nitrogen oxides formed in the bed. The carbon load is proportional to the carbon particle burnout time and it is therelure important that processes which change the burning time be understood and quantitatively characterized. Since burning times are proportional to the particle diameter, raised to a power between one and two, l even modest particle fragmentation will significantly decrease the burning time, and hence the bed carbon content. Fragmentation and attrition produce different sizes of particles. Fragmented particles are large enough to burn in the bed while attrited particles arc generally elutriated as combustible solids in ordinary fluidized combustors. Fragmentation of carl)on particles under fluidized bed combustion conditions have been observed in several small scale studies, z : The observations, with the exception of the work of Massimilla and coworkers, have been qualitative. Chirone et al. :~7 inferred fragntentatiun from the size measurement of char particles retriew,~d from a fluidized bed fed with narrowly sized coal particles, providing evidence of the formation of many smaller particles. The objective of this study was to determine the impact of fragmentation on char burnout time and
individual particle behavior by examining the rate of combustion of single particles fed to a bed. Bv use of single particles, the major increases in bunling rate that occur during fragmentation can be measured and interpreted whereas these increases are less evident in multiparticle studies. Separate studies carried out on swelling and fragmentation during devolatilization provided information on the initial particle size distribution for the process of char oxidation.
Experimental Appartus and Procedure The fluidized bed consists of a 64 mm Stainless tube with a 40 Ixm sintered inconel distributor plate. The bed could be operated in two modes: for studies of swelling and fragmentation during pyrolysis a nichrome wire mesh basket with 1.16 mm spacings was immersed, in the bed for retrieval of char particles during a rnn; for the time resolved oxidation studies a funnel for collecting the combustion products was bolted to the top of the bed. Silica particles of 180 to 212 I~m, fine enough to pass through the wire mesh basket, were used for bed solids. The measured minimum fluidizing velocity was 31.8 mm/s at 1073 K. All runs were made with a velocity five times minimum fluidization. The bed temperature, measured with a chromel-alumel thermocouple, was controlled by varying the input voltage to a three-zone furnace.
1495
1496
COAL COMBUSTION TABLE I Properties of Kentucky no. 9 coal char Coal'
Apparent ASTM rank Proximate Analysis (kg/kg, as received) Moisture Ash Volatile matter Fixed Carbon Ultimate Analysis (kg/kg, dry, ash-free) Carbon Hydrogen " Nitrogen Sulfur Oxygen Gross Heating Value (J/kg, as received) Particle Density (kg/m s) Free Swelling Index Volatile Yield" (%) (kg/kg, dry basis)
Char" hvAb
N_~ BET area, 77 K(m~/kg) Swelling Ratio Particle Density (kg/in '3)
50000 1.08 555
0.033 0.067 0.415 0.486 0.783 0.056 0.017 0.033 0.111 31 x 106 1320 2.5 47
'Char produced under flnidized bed conditions: T = 1073 K, O2 5%, with 6.2 mm coal.
Devolatilization In order to quantify swelling and fragmentation during devolatilization, single particles of a Kentucky bituminous coal (see Table 1 for properties) were fed to the bed and the char particles retrieved at predetermined times, selected to exceed the maximum volatile flame duration for the condition when particles were burned in air (the times varied from 6 seconds to 37 seconds, depending upon temperature and particle size). At the end of the run, the oxygen was replaced by nitrogen and the char particles extracted from the bed and quenched in a cold nitrogen stream. The resulting char partide dimensions (major and minor axes) were determined using an image analyzer and effective partide diameters calculated assuming the particles to be prolate ellipsoids. Thedimensions on the image analyzer were calibrated to provide the correct mass of coal particles of known density. Combustion The progress of combustion of single coal particles was followed by continuously monitoring the CO2 concentration in the exit gas. One percent water was added to the fluidizing gas to catalyze CO oxidation. The CO2 measurements were made by nondispersive infrared (NDIR) detectors and the gas
sample was filtered and dried to eliminate interference from water vapor with the CO2 signal. The CO concentration determined by spot chromatographic measurements throughout a run was always less than 30 ppm. The conditions used in the pyrolysis and combustion experiments are summarized in Table II.
TABLE II Experimental conditions for single particle studies Static Bed Height (ram) Superficial Velocity (mm/s at 1073 K) Devolatilization Bed Temperatnre (K) Coal Diameter (mm) Oxygen ill Inlet Gas (%) Coal particles/run Combustion Bed Temperature (K) Coal Diameter (mm) Oxygen in Inlet Gas (%) Water in Inlet Gas (%) Coal particles/run
64 159 1023, 1073, 1123 -6.7 + 5.66 -3.35 + 2.80 -2.36 + 2.00 0,5, 10 36 1073 -6.7 + 5.66 5 1
31
FRA(;MENTATION BEHAVIOR OF COAL PARTICLES
Results and Discussion
2000
.
.
.
1497
.
---~
Devolatilization One to ten char particles were produced during the devolatilization of individual coal particles. The average number of char particles produced by a single coal particle varied from two to five at the experimental conditions studied. The apparent volatile yield was a credible 47 percent indicating that particles too small to be collected by the wire mesh basket did not constitute a significant mass fraction of the total. The cumulative mass size distribution of the feed coal andthe product char particles from 34 experiments are shown in Fig. 1 on a lognormal probability plot. The particle size distribution for the coal is well represented by a unimodal lognormal distribution with a mean diameter of 6.25 mm. In contrast, the char shows a bimodal distribution. The smaller particles are the fragments produced during devolatilization with a mean diameter of 2.69 mm. The size of the larger particles is influenced by both swelling and loss of mass by fragmentation; the mean value is 6.79 mm corresponding to an increase in diameter of 8 percent. The statistical fits8 of the data are shown in Fig. 1. Results obtained for the range of bed temperatures, oxygen concentrations, and coal particle sizes listed in Table II were very similar when normalized with the mean initial coal particle diameter, so that Fig. 1 may be used to characterize the frag99.99 99.8
~
COAL: P=50+5Oerf [ 7.7,/n (d-l.8)]
951-
80 90 i
>. m _--J " '~ II1
70 6O 50 40
a.
20
~
5
.J
2
0 nr
30 9
~__
I u
.5
CHAR:
.I .01
- -
/
2
50+6.6 erf [0.74-0.822n16.5/d-l) + 43.4 erf [ 6.5 (~n d- 1.9)] 5 I0 20 PARTICLE SIZE (mml
50
I I00
FIG. 1. Mass-based coal and initial ' char particle size distributions. The solid lines are the fit of the data to cumulative probability equations, given on the figure.
fE ca.
15001
1000 ~J
5OO
0
10
20
30
TIME (min) FIG. 2. Trace of effluent CO2 concentration as a fimction of time produced by burning a single coal particle, no. 118, in the fluidized bed. mentation during devolatilization for all conditions studied.
Combustion A representative CO2 trace for the combustion of a single coal particle is shown in Fig. 2. In general, the CO 2 concentration rises during devolatilization to a level of about three percent. Towards the tail end of devolatilization, the CO2 rapidly drops to a level of 600 to 1850 ppm, the level being related to the number and sizes of fragments formed during devolatilization. For the 31 particles studied, the devolatilization time varied between 28.3 and 45.9 seconds as estimated from the time at which the steep devolatilization curve leveled off. The area under the COz curve beyond the devolatilization spike provides a measure of the mass of the initial char particles. The mass of the char was found to vary from 46 to 58 percent of the initial coal mass, consistent with an average volatile content of 47 percent. The ability to measure volatile yield from the COz suggests that mass loss of carbon in unreacted attrited particles was negligible for our conditions, although it is known to be significant at higher fluidization velocities. 9A~ One would normally expect the COz profile to decrease monotonically during combustion. The high frequency fluctuations during combustion may be related to the circulation of char particles in the bed and decrease as combustion proceeds due to the increase in r~umber of char fragments; n the present paper is concerned with the variations in the mean COz concentration obtained by averaging the signals over seven second intervals. There are,
COAL COMBUSTION
1498
by Smith 21 from Sergeant and Smith's bituminous coal data zz and our measured surface area of 50 • 103 mZ/kg. The penetration depth 23'24 of oxygen was estimated to be 17 I~m so that a shrinking pore model is appropriate for most of the life of the particle. The above equations were solved to obtain first the d - t and then the yco~ - t relationships as described in Ref. 11. The particle temperature overshoot was less than 14 K for our conditions.
however, periodic step increases in the CO2 level resulting from the acceleration in the burning rate following fragmentation. At least one and up to ten such step increases have been observed for the different single particle experiments. The char size distribution, including the effects of fragmentation, can be estimated from the shape of the smoothed CO2 9
Theoretical Considerations
Determination of Kinetic Parameters
From considerations of the model of char combustion by Smith, ~.2 the mass transfer correlation to large particles in fluidized beds of fine particles by LaNauze and Jung, 13'14 and the relationship between Nusselt and Sherwood numbers by Ross et al., 15 the following relationships for calculating yco2, d, and Tp respectively were derived:
The chemical reaction rate coefficient may be inferred from the COz concentration history of a coal particle that does not fragment during pyrolysis. Such particles can be identified from the CO2 traces because (a) they yield a relatively low CO2 level since their surface area is smaller than that of the
[ QoPo] 12
,rrd2
1 ATp exp [-E~,/(RTp)]
2
AT. exp [-E,,/(RT.)]
"t-
(1)
d
+
3D[2~mf+0.69[~]l/2scl/3] 8
3 - D 2emf + 0.69
1/2
=
r '? dt'
~2)
8
d Cg H~ k 2 + --
Prt/3 (Tp - T) + O'~rd(Tp4 - T 4)
emf 1
d
ATp exp (-Ea/(RTp))
An order of reaction of unity was selected even though the order is known to vary from 0 to 1.12'16'17 The higher order is supported by low temperature, high pressure flow reactor experiments 18 where diffusional limitations are absent, and by the pressure dependence of the burning rate in a fluidized bed.19 The coefficient of the apparent chemical rate fitted to our data at bed temperatures of 1073 and 1123 K and oxygen concentrations of 5 and 10% was Rc = 0.152 rip exp (-75.7/(R Tp)).u At a reaction temperature of 1087 K, Re = 0,038 (kg C)(m)/(kg Oz)(s), intermediate to a value of 0.016 derived from the correlation of Field et al. 2~ and a value of 0.109 determined using the intrinsic kinetic rate derived
1'2sc1'318
pieces of a fragmented particle, and (b) the slope of the COz curve is fairly fiat in the absence of rapidly reacting small particles. The CO2 traces for three particles, which satisfy these constraints, are shown in Fig. 3. The initial particle size was estimated from the initial char mass obtained by integrating the CO2 trace, and found to be in good agreement with the char volume, deduced from the coal size adjusted for the small swelling factor. The initial values of R~ were used to derive the rate equation used in the kinetic model. For the relatively large initial particle diameter used in this study, the external mass transfer resistance accounted for half of the total; the fractional resistance due to mass transfer
FRAGMENTATION BEHAVIOR OF COAL PARTICLES i
i
i
i
1500-~--,---118
i
E cm
H
1000
e--
O
cz~
10
20
30
TIME (min) Flc. 3. Measured (smoothed) and calculated COs
concentration curves for three single coal particles, no. 106, 118, and 123. Solid lines are the measured curves and dashed lines are based on model calculations. would be even greater if it were not for the low surface area 50 • 103 m2/kg of the present char.
Interpretation of CO2 Profile For a Single Particle In this section we wish to infer the number and size of the particles from the shape of the CO2 profiles. The CO2 level is a measure of burning rate and the curvature is a measure of particle size (in the limits of diffusion and kinetic control the yco~ - t curve is convex and concave respectively). The size distribution is easiest to infer by starting at the time of burnout tb. The time to burnout, tl, tfi, from the time tfi at the last fragmentation point (step increase in CO2) provides a measure of the largest particle size at tf~. This can be obtained using the integral form of the "shrinking sphere" burning law, Eq. 2. For particle 118 in Fig: 3, t2 = tb = 31.3 min and tl = tA = 15.1 min. The difference in time corresponds to the burning time of a particle 2.4 mm in diameter. The size of the next largest particle is obtained from the time at which the measured yco2 curve deviates significantly from that calculated from Eqs. 1-3 for a 2.4 mm particle burning over the time interval tf~ to tt,. A new Yco2 curve can be calculated from tf~ to the burnout time Of the second particle and the process repeated to match the yco2 curve down to tf,. This procedure may be used to introduce more than one particle of a given size at a given time. This inversion procedure yielded 22 particles for particle 118 at tfi. At a fragmentation point the constraints can be -
1499
imposed that (a) mass is conserved, (b) the number and size of particles prior to fragmentation yield the correct lower value of Yco2, (c) the number and size are so selected as to match the ycoz - t curve to next (prior) fragmentation point. The simplifying assumption was made that only one particle fragments at each step increase in ycoz. The number and size of fragments recombined, stepping back in time, were determined by trial and error using the match of the measured and calculated ycoz curve between fragmentation times as a check. Particles which burn out between two fragmentation points are detected, as in the previous discussion, by a systematic positive deviation of the measured Ycoz from the calculated values. An illustration of the fragmentation behavior between the end of devolatilization, (t = 0) and tf~, the last fragmentation point, is shown for particle 118 in Fig. 4. The time from the end of devolatilization is shown along the top of the figure. The shrinkage of particles between points is shown by horizontal lines. For each time marked at the top of the figure, one fragmentation occurs; the sizes of the parent particles and fragments are shown interconnected. For example, at 13.3 rain, a 2.7 mm particle fragments to yield six particles 1.8, 1.5, 1.4, 1.4, 1.4, and 1.2 mm in diameter. The matches between the measured and calculated Yco~'S are shown in Fig. 3. For each particle, Ea was assumed to be 75.7 kJ/mole which is consistent with both the literature values 12 for apparent activation energy and the temperature dependence of our data. The pre-exponential factor, A, was estimated from the Arrhenius equation using the measured Rc value of the particle. The estimated A values were 0.174, 0.136, and 0.185 (kg C) (m)/(kg O2)(s)(K) for particles 106, 118, and 123 respectively. The good match of the overall features of the CO2 profile suggest that the fragmentation behavior postulated is roughly correct, although we recognize that slightly different combinations of particle number and size would give equally good results. No attempt was made to match t h e fine structure which we believed to be accounted for by very small fragments.
Implications The measured normalized mass curve in Fig. 5 was obtained by summing the normalized masses of the 31 individual particles studied. For a single particle, fragmentation appears as small changes in the slope of the normalized mass curve, n But with a number of particles, the inflection points are no longer evident as seen in the measured normalized mass curve. These results underline the importance of using single particlestudies for determining the effect of fragmentation. The impact of fragmentation on the burning time
1500
COAL COMBUSTION
TIME (rain) 0.0
3.i
5
6.9
8.2
:1:1.:1
3.6
3.2
3.0
2.6
3.9
3.5
3.3
2.9
:]3.3
:13.9
2.4
2.3
27
2.2
~5.:1
2.~
L8
L7
t.6
t.5
1~
L4
L3
t.2
1.4
1.3
1.2
1.2
L4
L3
t.2
t2
1.4
L3
~.2
LI
l.t
1.0
0.9
3.0
2.B
2.5
L2
2.~
LO
1.9
3.0
2.9
2.5
2.2
2.~
2.0
LO
3.0
LB
L4
2.~
2.1
LO
19~[~
LB
2.4
2A
2.0
~.B
I.B
2.B
2.4
2.!
2.1
2.0
1.9
1.2
[~ - - ~ 5
:14.7
--~-~6
Flc. 4. Char particle diameters (mm), estimated from model calculations, at fragmeutation points tbr particle 118. The fragmentation times are relative to the start of char combustion. The shrinkage of particles btween fragmentations is shown by horizontal lines. The sizes of the parent particles and fi'agments are interconnected.
is examined by comparing in Fig. 5 the measured normalized char mass of the 31 particles studied with those calculated assuming that the initial char particles (a) have the same dimensions as the parent coal particles--an assumption often made in modeling studies--and (b) undergo swelling and fragmentation during devolatilization only, using as input the empirical data from Fig. 1. The results show how the burnout time, and
therefore carbon loading, would be overestimated using tile two approximate formulations. The times required to achieve 99 percent burnout are 27.4, 49.2, and 55.3 minutes determined respectively from COz measurements, calculation using the devolatilized char size distribution, and calculation using the initial coal size distribution. Fragmentation of char during combustion, often overlooked in calculations, is of major importance for the set of condi-
FRAGMENTATION BEHAVIOR OF COAL PARTICLES ~.0
,
9
t
i
i
~
Greek symbols
i
( emf
0.8 \
0.6
(r ,r crC
"
\
Bed voidage at superficial velocity Bed voidage at minimum fluidization velocity Emissivity of carbon particle Stephan-Boltzmann constant, J/s m 2 K4 Carbon particle density, kgC/m 3
Acknowledgments
al
0.4
1501
The experimental measurements were supported by the Tennessee Valley Authority, Contract No. 55783A. Support of C.A.S. by the Joseph R. Mares Chair and TVA is gratefully acknowledged. 0
lO
20
30
40
50
60
70
REFERENCES
TI~E (man)
FIG. 5. Measured normalized cbar mass and calculated normalized char masses based on coal and initial char particle size distributions. The empirical data given on Fig. 1 were used as input in the calculation of the coal and initial char normalized masses, tions studied. Further studies, with different coals, particle sizes, and combustion conditions are needed to fully establish its significance.
Nomenclature A Cg d D E~ Hc k Qo P Pr Po R Rc Re Sc T To T~, t tt, tf Yco2
Pre-exponential factor, kgC m/kgO2 s K Oxygen concentration in bulk gas, kgO2/m 3 Particle diameter, mm Bulk gas diffusion coefficient, mZ/s Apparent activation energy, kJ/mole Heat of combustion, kJ/kgC Effective thermal conductivity, J/s m K Flowrate at standard pressure and temperature, sL/s Cumulative probability Prandtl number Standard pressure, pascal Gas constant = 8.314 J/kmole K, 8.31 kPa m3/kmole K Chemical rate coefficient kgC m/kgO2 s Reynolds number based on superficial velocity Schmidt number Bed temperature, K Standard temperature, K Particle temperature, K Time, s Burnout time, s Fragmentation time, s Effluent CO2 concentration, ppm
1. AVEDESIAN,M. M. AND DAVIDSON,J. F.: Trans. Inst. Chem. Engrs. 51, 121 (1973). 2. ANDREI, M.: "Time Resolved Burnout in the Combustion of Coal Particles in a Fluidized Bed." S. M. Thesis in Chem. Eng., M.I.T., Cambridge, MA, 1978. 3. CHIaONE, R.: Thesis in Chem. Eng., University of Naples, 1980. 4. CAMPBELL, E. K. AND DAVIDSON, J, F.: Fluidised Combustion, p. A3-1, Institute of Fuel Symposium Series 1, 1975. 5. PILLM, K. K.: J. of Inst. of Energy 54, 142 (1981). 6. JUNG, K. AND LANAUZE, R. D.: Can. J. of Chem. Eng. 61, 262 (1983). 7. CHmONE, R., CAMMAROTA,A., D'AMOBE, M. AND MASSIMILLA, L.: Seventh International Conference on Fluidized Bed Combustion, Vol. 2, p. 1023, Avail. NTIS, DOE/METC/83-48, 1983. 8. IRANI, R. R., AND CALLIS, C. F.: Particle Size: Measurement, Interpretation, and Application, p. 39, John Wiley and Sons, 1963. 9. DONSI, G., MASSIMILLA, L. AND MICClO, M.: Combustion and Flame 41, 57 (1981). 10. WALSH, P. M., DUTI'A, A. AND BEER, J. M.: "Char Combustion in the Freeboard above a Flnidized Bed Burning a High Volatile Bituminous Coal." This Symposium. 11. SUNDBACK,C. A.: Sc.D. Thesis in Chem. Eng., M.I.T., Cambridge, MA, 1984. 12. SMITU, I. W.: Nineteenth Symposium (International) on Combustion, p. 1045, The Combustion Institute, 1982. 13. LANAUZE, R. ]). AND JUNG,-K.: Nineteenth Symposium (International) on Combustion, p. 1087, The Combustion Institute, 1982. 14. LANAUZE,R. D., JUNG, K., AND KASTL,J.: "Mass Transfer to Large Particles in Fluidized Beds of Smaller Particles." Submitted to Chem. Eng. Sci,, 1983.
1502
COAL COMBUSTION
15. Ross, I. B., PATEL, M. S. AND DAVIDSON, J. F.: Trans. Chem. Engrs. 59, 83 (1981). 16. ESSENHIGH, R. H.: Sixteenth Symposium (International) on Combustion, p. 353, The Combustion Institute, 1977. 17. YOVNG, B. C. AND SMITH, I. W.: Eighteenth
Symposium (International) on Combustion, p. 1249, The Combustion Institute, 1981. 18. RnNISH, J. M. AND WALKER, P. L. JB.: Sixteenth Conference on Carbon, p. 158, American Carbon Society, 1983. 19. TURNBULL, E., KOSSAKOWSKI, E. R., DAVlDSON, J. F., HOPES, R. B., BLAEKSHAW, H. W. aND GOODYER, P. T. Y.: "The Effect of Pressure on
20.
21. 22. 23. 24.
the Combustion of Char in Fluidised Beds." To be published. FIELD, M. A., GILL, D. W., MORGAN, B. B. AND HAWKSLE'/, P. G. W. : Combustion of Pulverized Coal, p. 187, The British Coal Utilization Research Association, Leatherhead, 1967. SMITH, I. W.: Fuel 57, 409 (1978). SERGEANT, G. D. AND SMITH, I. W.: Fuel 52, 52 (1973). MVLCXHY, M. F. R. AND SMITH, I. W.: Rev. Pure and Appl. Chem. 19, 81 (1969). LAURENDEAU, N. M.: Prog. Energy Comb. Sci. 4, 221 (1978).
COMMENTS I. W. Smith, CSIRO Division of Fossil Fuels, Australia. To what extent is the assumption that the particles burned as shrinking spheres in "conflict with mechanisms of fragmentation which require oxygen attack within the particle? Did you take the opportunity to freeze the bed after a fragmentation and count the n u m b e r of particles?
Authors" Reply. Predictions of fragmentation using percolation theory have been conducted for the case of chemically controlled reactionJ For this limiting case a particle breaks up into many fragments when the porosity reaches 70 percent. Our studies, however, were under conditions in which the penetration depth, which was on the order of 20 Ixm, was much smaller than the particle diameter, But we o b s e r v e d cracks in the char particles, with widths of 50-250 ~m, which would be penetrated by oxygen. Therefore, although the char particles burned as shrinking spheres, the oxidation within the cracks could explain the periodic particle breakup we inferred from our CO2 profiles. The bed was not frozen after a fragmentation in this study. However similar, single particle experiments were conducted with 5.5 mm Kentucky no. 9 coal particles at the University of Naples. 2 The char particles were retrieved from the bed at predetermined times. For the 6 coal particles studied, 2-4 fragments were produced during a fragmentation and 2 - 7 fragments were produced from a single coal particle.
REFERENCES 1. KERSTEIN, A. R. AND NIKSA, S.: 1983 International Conference on Coal Science, p. 743, International Energy Agency, 1983.
2. MASSIMILLA, L.: Discussion at this Symposium, 1984.
K. W. Ragland, University of Wisconsin, USA. Your work suggests extensive breakup of mm size, high volatile bituminous coal during combustion in a fluidized bed. I believe you stated that up to 22 fragments are obtained from one coal particle. This is quite different from the behavior that we have observed (Ref. 1) with bituminous, sub-bituminous and lignite coals in an electrically heated fluidized bed. With 10% oxygen 50% of the bituminous particles broke up, while only 25% of the sub-bituminous and lignite particles broke up. When breakup occurred, only a few pieces were formed, typically. We observed fragmentation directly through the quartz wall, and by quenching and dumping the bed. Did you make any direct observations of fragmentation to supplement the gas sampling traces? Did you try various coal types? Did your coal samples contain a lot of surface fractures?
REFERENCE 1. RXGLXND, K. W., JEHN, T. C. AND YANG, J. T., "Coal Combustion at High Reynolds Number," Eighteenth Symp. (Int.) on Combustion, 12951303, The Combustion Institute, 1981.
Authors" Reply. As stated in our paper, we believe that fragmentation behavior will be dependent upon coal type and combustion conditions. In our study, 2 - 6 fragments were produced during a single fragmentation and up to 22 particles were produced from a single coal particle. Each coal particle fragmented at least once and up to 10 times during
FRAGMENTATION BEHAVIOR O F COAL PARTICLES char combustion. Fragmentation was visually observed during char combustion but char particles were not retrieved from the bed. We did not make quantitative visual observations about fragmentation in the present study, but a previous study in our laboratory ~ using a Montana lignite and recent studies at the University of Naples z using a Kentucky No. 9 bituminous coal provide direct measurements of particle breakup. In the latter study, two to four fragments were produced during a single fragmentation and up to 7 particles were produced from a single coal particle. As in our study, each coal particle fragmented at least once. The difference in the mass of the coal particles used in the present study and the University of Naples study accounts for the difference in the total number of char particles. The Kentucky No. 9 coal used in this study had submerged cracks which propagated to the particle surface during devolatilization. The char particles had surface cracks of 50-250 txm width at spacings of about 700-4000 txm which we believe account for the fragmentation pattern.
A. Kerstein, Sandia National Laboratories, USA. Your data for the time sequence of fragmentation events may yield information concerning the evolving macrostructure of char particle and fragments. For instance, it would be i n t e r e s t i n g to check whether the subsequent time evolution of an Nth generation fragment (N = O signifying an unfragmented particle) is N-dependent.
Authors" Reply. Detailed time-resolved char size distribution analysis has only been done on the CO2 profiles from t h r e e coal particles. Because the amount of data was limited, only the time between fragmentations and the n u m b e r of particles produced during a fragmentation were determined; the statistics of these two variables are given in the accompanying table. There was no trend in the value of either variable from one generation to the next. The average number of particles produced during a fragmentation was approximately the same for all three particles. However the average time between fragmentations did vary from particle to particle.
Coal Particle
Time to Next Fragmentation (min)
Number of Particles Produced During a Fragmentation
106 118 123
7.9 -+ 4.6 4.5 -+ 1.7 2.8 -+ 1.5
3.1 -+ 1.2 3.5 -+ 0.9 4.1 -+ 1.4
REFERENCES 1. ANDREI, M. A.: S.M. Thesis, D e p a r t m e n t of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA, 1978. 2. MASSIMILLA, L.: Discussion at this Symposium, 1984.
1503
FRAGMENTATION
BEHAVIOR OF SINGLE IN A FLUIDIZED BED
C. A. SUNDBACK, J. M. BEIs
COAL
PARTICLES
A. F. SAROFIM
Department of Chemical Engineering Massachusetts Institute of Technology Cambridge, Massachusetts 02139 The fragmentation hehavinr of single coal particles has been studied in a small scale fluidized bed for a Kentucky No. 9 bitumim)us coal. A typical coal particle fragments and swells slightly during dcvolatilization, producing particles in two distinct size regimes. Although the fragmentation pattern varies from particle to particle, the bimCxtal mass-based char size distribution obtained for the pyrolysis of a collection of coal particles was found to be little influenced by changes in bed temperature (1023-1123 K), oxygen concentration (0-10%), and co',d particle size (2.2-6.2 mm). l)uring oxidation of single coal particles, the dramatic increases in burning rate following particle breakup made it possible to infer fragmentation from the carbon dioxide concentration variation with time of the gas effluent. The number and size of the coal fragments were deter,nined by fitting the CO, data with the predictions of a char combustion model. Swelling and fiagmentation contribute to an approximately twofold reduction in the average, hurning time at the experimental conditions studied.
Introduction Predictions of'the carbon content in fluidized beds are important becanse carbon losses from the bed are proportional to the carbon loading, and because of carbon's role in the reduction of nitrogen oxides formed in the bed. The carbon load is proportional to the carbon particle burnout time and it is therelure important that processes which change the burning time be understood and quantitatively characterized. Since burning times are proportional to the particle diameter, raised to a power between one and two, l even modest particle fragmentation will significantly decrease the burning time, and hence the bed carbon content. Fragmentation and attrition produce different sizes of particles. Fragmented particles are large enough to burn in the bed while attrited particles arc generally elutriated as combustible solids in ordinary fluidized combustors. Fragmentation of carl)on particles under fluidized bed combustion conditions have been observed in several small scale studies, z : The observations, with the exception of the work of Massimilla and coworkers, have been qualitative. Chirone et al. :~7 inferred fragntentatiun from the size measurement of char particles retriew,~d from a fluidized bed fed with narrowly sized coal particles, providing evidence of the formation of many smaller particles. The objective of this study was to determine the impact of fragmentation on char burnout time and
individual particle behavior by examining the rate of combustion of single particles fed to a bed. Bv use of single particles, the major increases in bunling rate that occur during fragmentation can be measured and interpreted whereas these increases are less evident in multiparticle studies. Separate studies carried out on swelling and fragmentation during devolatilization provided information on the initial particle size distribution for the process of char oxidation.
Experimental Appartus and Procedure The fluidized bed consists of a 64 mm Stainless tube with a 40 Ixm sintered inconel distributor plate. The bed could be operated in two modes: for studies of swelling and fragmentation during pyrolysis a nichrome wire mesh basket with 1.16 mm spacings was immersed, in the bed for retrieval of char particles during a rnn; for the time resolved oxidation studies a funnel for collecting the combustion products was bolted to the top of the bed. Silica particles of 180 to 212 I~m, fine enough to pass through the wire mesh basket, were used for bed solids. The measured minimum fluidizing velocity was 31.8 mm/s at 1073 K. All runs were made with a velocity five times minimum fluidization. The bed temperature, measured with a chromel-alumel thermocouple, was controlled by varying the input voltage to a three-zone furnace.
1495
1496
COAL COMBUSTION TABLE I Properties of Kentucky no. 9 coal char Coal'
Apparent ASTM rank Proximate Analysis (kg/kg, as received) Moisture Ash Volatile matter Fixed Carbon Ultimate Analysis (kg/kg, dry, ash-free) Carbon Hydrogen " Nitrogen Sulfur Oxygen Gross Heating Value (J/kg, as received) Particle Density (kg/m s) Free Swelling Index Volatile Yield" (%) (kg/kg, dry basis)
Char" hvAb
N_~ BET area, 77 K(m~/kg) Swelling Ratio Particle Density (kg/in '3)
50000 1.08 555
0.033 0.067 0.415 0.486 0.783 0.056 0.017 0.033 0.111 31 x 106 1320 2.5 47
'Char produced under flnidized bed conditions: T = 1073 K, O2 5%, with 6.2 mm coal.
Devolatilization In order to quantify swelling and fragmentation during devolatilization, single particles of a Kentucky bituminous coal (see Table 1 for properties) were fed to the bed and the char particles retrieved at predetermined times, selected to exceed the maximum volatile flame duration for the condition when particles were burned in air (the times varied from 6 seconds to 37 seconds, depending upon temperature and particle size). At the end of the run, the oxygen was replaced by nitrogen and the char particles extracted from the bed and quenched in a cold nitrogen stream. The resulting char partide dimensions (major and minor axes) were determined using an image analyzer and effective partide diameters calculated assuming the particles to be prolate ellipsoids. Thedimensions on the image analyzer were calibrated to provide the correct mass of coal particles of known density. Combustion The progress of combustion of single coal particles was followed by continuously monitoring the CO2 concentration in the exit gas. One percent water was added to the fluidizing gas to catalyze CO oxidation. The CO2 measurements were made by nondispersive infrared (NDIR) detectors and the gas
sample was filtered and dried to eliminate interference from water vapor with the CO2 signal. The CO concentration determined by spot chromatographic measurements throughout a run was always less than 30 ppm. The conditions used in the pyrolysis and combustion experiments are summarized in Table II.
TABLE II Experimental conditions for single particle studies Static Bed Height (ram) Superficial Velocity (mm/s at 1073 K) Devolatilization Bed Temperatnre (K) Coal Diameter (mm) Oxygen ill Inlet Gas (%) Coal particles/run Combustion Bed Temperature (K) Coal Diameter (mm) Oxygen in Inlet Gas (%) Water in Inlet Gas (%) Coal particles/run
64 159 1023, 1073, 1123 -6.7 + 5.66 -3.35 + 2.80 -2.36 + 2.00 0,5, 10 36 1073 -6.7 + 5.66 5 1
31
FRA(;MENTATION BEHAVIOR OF COAL PARTICLES
Results and Discussion
2000
.
.
.
1497
.
---~
Devolatilization One to ten char particles were produced during the devolatilization of individual coal particles. The average number of char particles produced by a single coal particle varied from two to five at the experimental conditions studied. The apparent volatile yield was a credible 47 percent indicating that particles too small to be collected by the wire mesh basket did not constitute a significant mass fraction of the total. The cumulative mass size distribution of the feed coal andthe product char particles from 34 experiments are shown in Fig. 1 on a lognormal probability plot. The particle size distribution for the coal is well represented by a unimodal lognormal distribution with a mean diameter of 6.25 mm. In contrast, the char shows a bimodal distribution. The smaller particles are the fragments produced during devolatilization with a mean diameter of 2.69 mm. The size of the larger particles is influenced by both swelling and loss of mass by fragmentation; the mean value is 6.79 mm corresponding to an increase in diameter of 8 percent. The statistical fits8 of the data are shown in Fig. 1. Results obtained for the range of bed temperatures, oxygen concentrations, and coal particle sizes listed in Table II were very similar when normalized with the mean initial coal particle diameter, so that Fig. 1 may be used to characterize the frag99.99 99.8
~
COAL: P=50+5Oerf [ 7.7,/n (d-l.8)]
951-
80 90 i
>. m _--J " '~ II1
70 6O 50 40
a.
20
~
5
.J
2
0 nr
30 9
~__
I u
.5
CHAR:
.I .01
- -
/
2
50+6.6 erf [0.74-0.822n16.5/d-l) + 43.4 erf [ 6.5 (~n d- 1.9)] 5 I0 20 PARTICLE SIZE (mml
50
I I00
FIG. 1. Mass-based coal and initial ' char particle size distributions. The solid lines are the fit of the data to cumulative probability equations, given on the figure.
fE ca.
15001
1000 ~J
5OO
0
10
20
30
TIME (min) FIG. 2. Trace of effluent CO2 concentration as a fimction of time produced by burning a single coal particle, no. 118, in the fluidized bed. mentation during devolatilization for all conditions studied.
Combustion A representative CO2 trace for the combustion of a single coal particle is shown in Fig. 2. In general, the CO 2 concentration rises during devolatilization to a level of about three percent. Towards the tail end of devolatilization, the CO2 rapidly drops to a level of 600 to 1850 ppm, the level being related to the number and sizes of fragments formed during devolatilization. For the 31 particles studied, the devolatilization time varied between 28.3 and 45.9 seconds as estimated from the time at which the steep devolatilization curve leveled off. The area under the COz curve beyond the devolatilization spike provides a measure of the mass of the initial char particles. The mass of the char was found to vary from 46 to 58 percent of the initial coal mass, consistent with an average volatile content of 47 percent. The ability to measure volatile yield from the COz suggests that mass loss of carbon in unreacted attrited particles was negligible for our conditions, although it is known to be significant at higher fluidization velocities. 9A~ One would normally expect the COz profile to decrease monotonically during combustion. The high frequency fluctuations during combustion may be related to the circulation of char particles in the bed and decrease as combustion proceeds due to the increase in r~umber of char fragments; n the present paper is concerned with the variations in the mean COz concentration obtained by averaging the signals over seven second intervals. There are,
COAL COMBUSTION
1498
by Smith 21 from Sergeant and Smith's bituminous coal data zz and our measured surface area of 50 • 103 mZ/kg. The penetration depth 23'24 of oxygen was estimated to be 17 I~m so that a shrinking pore model is appropriate for most of the life of the particle. The above equations were solved to obtain first the d - t and then the yco~ - t relationships as described in Ref. 11. The particle temperature overshoot was less than 14 K for our conditions.
however, periodic step increases in the CO2 level resulting from the acceleration in the burning rate following fragmentation. At least one and up to ten such step increases have been observed for the different single particle experiments. The char size distribution, including the effects of fragmentation, can be estimated from the shape of the smoothed CO2 9
Theoretical Considerations
Determination of Kinetic Parameters
From considerations of the model of char combustion by Smith, ~.2 the mass transfer correlation to large particles in fluidized beds of fine particles by LaNauze and Jung, 13'14 and the relationship between Nusselt and Sherwood numbers by Ross et al., 15 the following relationships for calculating yco2, d, and Tp respectively were derived:
The chemical reaction rate coefficient may be inferred from the COz concentration history of a coal particle that does not fragment during pyrolysis. Such particles can be identified from the CO2 traces because (a) they yield a relatively low CO2 level since their surface area is smaller than that of the
[ QoPo] 12
,rrd2
1 ATp exp [-E~,/(RTp)]
2
AT. exp [-E,,/(RT.)]
"t-
(1)
d
+
3D[2~mf+0.69[~]l/2scl/3] 8
3 - D 2emf + 0.69
1/2
=
r '? dt'
~2)
8
d Cg H~ k 2 + --
Prt/3 (Tp - T) + O'~rd(Tp4 - T 4)
emf 1
d
ATp exp (-Ea/(RTp))
An order of reaction of unity was selected even though the order is known to vary from 0 to 1.12'16'17 The higher order is supported by low temperature, high pressure flow reactor experiments 18 where diffusional limitations are absent, and by the pressure dependence of the burning rate in a fluidized bed.19 The coefficient of the apparent chemical rate fitted to our data at bed temperatures of 1073 and 1123 K and oxygen concentrations of 5 and 10% was Rc = 0.152 rip exp (-75.7/(R Tp)).u At a reaction temperature of 1087 K, Re = 0,038 (kg C)(m)/(kg Oz)(s), intermediate to a value of 0.016 derived from the correlation of Field et al. 2~ and a value of 0.109 determined using the intrinsic kinetic rate derived
1'2sc1'318
pieces of a fragmented particle, and (b) the slope of the COz curve is fairly fiat in the absence of rapidly reacting small particles. The CO2 traces for three particles, which satisfy these constraints, are shown in Fig. 3. The initial particle size was estimated from the initial char mass obtained by integrating the CO2 trace, and found to be in good agreement with the char volume, deduced from the coal size adjusted for the small swelling factor. The initial values of R~ were used to derive the rate equation used in the kinetic model. For the relatively large initial particle diameter used in this study, the external mass transfer resistance accounted for half of the total; the fractional resistance due to mass transfer
FRAGMENTATION BEHAVIOR OF COAL PARTICLES i
i
i
i
1500-~--,---118
i
E cm
H
1000
e--
O
cz~
10
20
30
TIME (min) Flc. 3. Measured (smoothed) and calculated COs
concentration curves for three single coal particles, no. 106, 118, and 123. Solid lines are the measured curves and dashed lines are based on model calculations. would be even greater if it were not for the low surface area 50 • 103 m2/kg of the present char.
Interpretation of CO2 Profile For a Single Particle In this section we wish to infer the number and size of the particles from the shape of the CO2 profiles. The CO2 level is a measure of burning rate and the curvature is a measure of particle size (in the limits of diffusion and kinetic control the yco~ - t curve is convex and concave respectively). The size distribution is easiest to infer by starting at the time of burnout tb. The time to burnout, tl, tfi, from the time tfi at the last fragmentation point (step increase in CO2) provides a measure of the largest particle size at tf~. This can be obtained using the integral form of the "shrinking sphere" burning law, Eq. 2. For particle 118 in Fig: 3, t2 = tb = 31.3 min and tl = tA = 15.1 min. The difference in time corresponds to the burning time of a particle 2.4 mm in diameter. The size of the next largest particle is obtained from the time at which the measured yco2 curve deviates significantly from that calculated from Eqs. 1-3 for a 2.4 mm particle burning over the time interval tf~ to tt,. A new Yco2 curve can be calculated from tf~ to the burnout time Of the second particle and the process repeated to match the yco2 curve down to tf,. This procedure may be used to introduce more than one particle of a given size at a given time. This inversion procedure yielded 22 particles for particle 118 at tfi. At a fragmentation point the constraints can be -
1499
imposed that (a) mass is conserved, (b) the number and size of particles prior to fragmentation yield the correct lower value of Yco2, (c) the number and size are so selected as to match the ycoz - t curve to next (prior) fragmentation point. The simplifying assumption was made that only one particle fragments at each step increase in ycoz. The number and size of fragments recombined, stepping back in time, were determined by trial and error using the match of the measured and calculated ycoz curve between fragmentation times as a check. Particles which burn out between two fragmentation points are detected, as in the previous discussion, by a systematic positive deviation of the measured Ycoz from the calculated values. An illustration of the fragmentation behavior between the end of devolatilization, (t = 0) and tf~, the last fragmentation point, is shown for particle 118 in Fig. 4. The time from the end of devolatilization is shown along the top of the figure. The shrinkage of particles between points is shown by horizontal lines. For each time marked at the top of the figure, one fragmentation occurs; the sizes of the parent particles and fragments are shown interconnected. For example, at 13.3 rain, a 2.7 mm particle fragments to yield six particles 1.8, 1.5, 1.4, 1.4, 1.4, and 1.2 mm in diameter. The matches between the measured and calculated Yco~'S are shown in Fig. 3. For each particle, Ea was assumed to be 75.7 kJ/mole which is consistent with both the literature values 12 for apparent activation energy and the temperature dependence of our data. The pre-exponential factor, A, was estimated from the Arrhenius equation using the measured Rc value of the particle. The estimated A values were 0.174, 0.136, and 0.185 (kg C) (m)/(kg O2)(s)(K) for particles 106, 118, and 123 respectively. The good match of the overall features of the CO2 profile suggest that the fragmentation behavior postulated is roughly correct, although we recognize that slightly different combinations of particle number and size would give equally good results. No attempt was made to match t h e fine structure which we believed to be accounted for by very small fragments.
Implications The measured normalized mass curve in Fig. 5 was obtained by summing the normalized masses of the 31 individual particles studied. For a single particle, fragmentation appears as small changes in the slope of the normalized mass curve, n But with a number of particles, the inflection points are no longer evident as seen in the measured normalized mass curve. These results underline the importance of using single particlestudies for determining the effect of fragmentation. The impact of fragmentation on the burning time
1500
COAL COMBUSTION
TIME (rain) 0.0
3.i
5
6.9
8.2
:1:1.:1
3.6
3.2
3.0
2.6
3.9
3.5
3.3
2.9
:]3.3
:13.9
2.4
2.3
27
2.2
~5.:1
2.~
L8
L7
t.6
t.5
1~
L4
L3
t.2
1.4
1.3
1.2
1.2
L4
L3
t.2
t2
1.4
L3
~.2
LI
l.t
1.0
0.9
3.0
2.B
2.5
L2
2.~
LO
1.9
3.0
2.9
2.5
2.2
2.~
2.0
LO
3.0
LB
L4
2.~
2.1
LO
19~[~
LB
2.4
2A
2.0
~.B
I.B
2.B
2.4
2.!
2.1
2.0
1.9
1.2
[~ - - ~ 5
:14.7
--~-~6
Flc. 4. Char particle diameters (mm), estimated from model calculations, at fragmeutation points tbr particle 118. The fragmentation times are relative to the start of char combustion. The shrinkage of particles btween fragmentations is shown by horizontal lines. The sizes of the parent particles and fi'agments are interconnected.
is examined by comparing in Fig. 5 the measured normalized char mass of the 31 particles studied with those calculated assuming that the initial char particles (a) have the same dimensions as the parent coal particles--an assumption often made in modeling studies--and (b) undergo swelling and fragmentation during devolatilization only, using as input the empirical data from Fig. 1. The results show how the burnout time, and
therefore carbon loading, would be overestimated using tile two approximate formulations. The times required to achieve 99 percent burnout are 27.4, 49.2, and 55.3 minutes determined respectively from COz measurements, calculation using the devolatilized char size distribution, and calculation using the initial coal size distribution. Fragmentation of char during combustion, often overlooked in calculations, is of major importance for the set of condi-
FRAGMENTATION BEHAVIOR OF COAL PARTICLES ~.0
,
9
t
i
i
~
Greek symbols
i
( emf
0.8 \
0.6
(r ,r crC
"
\
Bed voidage at superficial velocity Bed voidage at minimum fluidization velocity Emissivity of carbon particle Stephan-Boltzmann constant, J/s m 2 K4 Carbon particle density, kgC/m 3
Acknowledgments
al
0.4
1501
The experimental measurements were supported by the Tennessee Valley Authority, Contract No. 55783A. Support of C.A.S. by the Joseph R. Mares Chair and TVA is gratefully acknowledged. 0
lO
20
30
40
50
60
70
REFERENCES
TI~E (man)
FIG. 5. Measured normalized cbar mass and calculated normalized char masses based on coal and initial char particle size distributions. The empirical data given on Fig. 1 were used as input in the calculation of the coal and initial char normalized masses, tions studied. Further studies, with different coals, particle sizes, and combustion conditions are needed to fully establish its significance.
Nomenclature A Cg d D E~ Hc k Qo P Pr Po R Rc Re Sc T To T~, t tt, tf Yco2
Pre-exponential factor, kgC m/kgO2 s K Oxygen concentration in bulk gas, kgO2/m 3 Particle diameter, mm Bulk gas diffusion coefficient, mZ/s Apparent activation energy, kJ/mole Heat of combustion, kJ/kgC Effective thermal conductivity, J/s m K Flowrate at standard pressure and temperature, sL/s Cumulative probability Prandtl number Standard pressure, pascal Gas constant = 8.314 J/kmole K, 8.31 kPa m3/kmole K Chemical rate coefficient kgC m/kgO2 s Reynolds number based on superficial velocity Schmidt number Bed temperature, K Standard temperature, K Particle temperature, K Time, s Burnout time, s Fragmentation time, s Effluent CO2 concentration, ppm
1. AVEDESIAN,M. M. AND DAVIDSON,J. F.: Trans. Inst. Chem. Engrs. 51, 121 (1973). 2. ANDREI, M.: "Time Resolved Burnout in the Combustion of Coal Particles in a Fluidized Bed." S. M. Thesis in Chem. Eng., M.I.T., Cambridge, MA, 1978. 3. CHIaONE, R.: Thesis in Chem. Eng., University of Naples, 1980. 4. CAMPBELL, E. K. AND DAVIDSON, J, F.: Fluidised Combustion, p. A3-1, Institute of Fuel Symposium Series 1, 1975. 5. PILLM, K. K.: J. of Inst. of Energy 54, 142 (1981). 6. JUNG, K. AND LANAUZE, R. D.: Can. J. of Chem. Eng. 61, 262 (1983). 7. CHmONE, R., CAMMAROTA,A., D'AMOBE, M. AND MASSIMILLA, L.: Seventh International Conference on Fluidized Bed Combustion, Vol. 2, p. 1023, Avail. NTIS, DOE/METC/83-48, 1983. 8. IRANI, R. R., AND CALLIS, C. F.: Particle Size: Measurement, Interpretation, and Application, p. 39, John Wiley and Sons, 1963. 9. DONSI, G., MASSIMILLA, L. AND MICClO, M.: Combustion and Flame 41, 57 (1981). 10. WALSH, P. M., DUTI'A, A. AND BEER, J. M.: "Char Combustion in the Freeboard above a Flnidized Bed Burning a High Volatile Bituminous Coal." This Symposium. 11. SUNDBACK,C. A.: Sc.D. Thesis in Chem. Eng., M.I.T., Cambridge, MA, 1984. 12. SMITU, I. W.: Nineteenth Symposium (International) on Combustion, p. 1045, The Combustion Institute, 1982. 13. LANAUZE, R. ]). AND JUNG,-K.: Nineteenth Symposium (International) on Combustion, p. 1087, The Combustion Institute, 1982. 14. LANAUZE,R. D., JUNG, K., AND KASTL,J.: "Mass Transfer to Large Particles in Fluidized Beds of Smaller Particles." Submitted to Chem. Eng. Sci,, 1983.
1502
COAL COMBUSTION
15. Ross, I. B., PATEL, M. S. AND DAVIDSON, J. F.: Trans. Chem. Engrs. 59, 83 (1981). 16. ESSENHIGH, R. H.: Sixteenth Symposium (International) on Combustion, p. 353, The Combustion Institute, 1977. 17. YOVNG, B. C. AND SMITH, I. W.: Eighteenth
Symposium (International) on Combustion, p. 1249, The Combustion Institute, 1981. 18. RnNISH, J. M. AND WALKER, P. L. JB.: Sixteenth Conference on Carbon, p. 158, American Carbon Society, 1983. 19. TURNBULL, E., KOSSAKOWSKI, E. R., DAVlDSON, J. F., HOPES, R. B., BLAEKSHAW, H. W. aND GOODYER, P. T. Y.: "The Effect of Pressure on
20.
21. 22. 23. 24.
the Combustion of Char in Fluidised Beds." To be published. FIELD, M. A., GILL, D. W., MORGAN, B. B. AND HAWKSLE'/, P. G. W. : Combustion of Pulverized Coal, p. 187, The British Coal Utilization Research Association, Leatherhead, 1967. SMITH, I. W.: Fuel 57, 409 (1978). SERGEANT, G. D. AND SMITH, I. W.: Fuel 52, 52 (1973). MVLCXHY, M. F. R. AND SMITH, I. W.: Rev. Pure and Appl. Chem. 19, 81 (1969). LAURENDEAU, N. M.: Prog. Energy Comb. Sci. 4, 221 (1978).
COMMENTS I. W. Smith, CSIRO Division of Fossil Fuels, Australia. To what extent is the assumption that the particles burned as shrinking spheres in "conflict with mechanisms of fragmentation which require oxygen attack within the particle? Did you take the opportunity to freeze the bed after a fragmentation and count the n u m b e r of particles?
Authors" Reply. Predictions of fragmentation using percolation theory have been conducted for the case of chemically controlled reactionJ For this limiting case a particle breaks up into many fragments when the porosity reaches 70 percent. Our studies, however, were under conditions in which the penetration depth, which was on the order of 20 Ixm, was much smaller than the particle diameter, But we o b s e r v e d cracks in the char particles, with widths of 50-250 ~m, which would be penetrated by oxygen. Therefore, although the char particles burned as shrinking spheres, the oxidation within the cracks could explain the periodic particle breakup we inferred from our CO2 profiles. The bed was not frozen after a fragmentation in this study. However similar, single particle experiments were conducted with 5.5 mm Kentucky no. 9 coal particles at the University of Naples. 2 The char particles were retrieved from the bed at predetermined times. For the 6 coal particles studied, 2-4 fragments were produced during a fragmentation and 2 - 7 fragments were produced from a single coal particle.
REFERENCES 1. KERSTEIN, A. R. AND NIKSA, S.: 1983 International Conference on Coal Science, p. 743, International Energy Agency, 1983.
2. MASSIMILLA, L.: Discussion at this Symposium, 1984.
K. W. Ragland, University of Wisconsin, USA. Your work suggests extensive breakup of mm size, high volatile bituminous coal during combustion in a fluidized bed. I believe you stated that up to 22 fragments are obtained from one coal particle. This is quite different from the behavior that we have observed (Ref. 1) with bituminous, sub-bituminous and lignite coals in an electrically heated fluidized bed. With 10% oxygen 50% of the bituminous particles broke up, while only 25% of the sub-bituminous and lignite particles broke up. When breakup occurred, only a few pieces were formed, typically. We observed fragmentation directly through the quartz wall, and by quenching and dumping the bed. Did you make any direct observations of fragmentation to supplement the gas sampling traces? Did you try various coal types? Did your coal samples contain a lot of surface fractures?
REFERENCE 1. RXGLXND, K. W., JEHN, T. C. AND YANG, J. T., "Coal Combustion at High Reynolds Number," Eighteenth Symp. (Int.) on Combustion, 12951303, The Combustion Institute, 1981.
Authors" Reply. As stated in our paper, we believe that fragmentation behavior will be dependent upon coal type and combustion conditions. In our study, 2 - 6 fragments were produced during a single fragmentation and up to 22 particles were produced from a single coal particle. Each coal particle fragmented at least once and up to 10 times during
FRAGMENTATION BEHAVIOR O F COAL PARTICLES char combustion. Fragmentation was visually observed during char combustion but char particles were not retrieved from the bed. We did not make quantitative visual observations about fragmentation in the present study, but a previous study in our laboratory ~ using a Montana lignite and recent studies at the University of Naples z using a Kentucky No. 9 bituminous coal provide direct measurements of particle breakup. In the latter study, two to four fragments were produced during a single fragmentation and up to 7 particles were produced from a single coal particle. As in our study, each coal particle fragmented at least once. The difference in the mass of the coal particles used in the present study and the University of Naples study accounts for the difference in the total number of char particles. The Kentucky No. 9 coal used in this study had submerged cracks which propagated to the particle surface during devolatilization. The char particles had surface cracks of 50-250 txm width at spacings of about 700-4000 txm which we believe account for the fragmentation pattern.
A. Kerstein, Sandia National Laboratories, USA. Your data for the time sequence of fragmentation events may yield information concerning the evolving macrostructure of char particle and fragments. For instance, it would be i n t e r e s t i n g to check whether the subsequent time evolution of an Nth generation fragment (N = O signifying an unfragmented particle) is N-dependent.
Authors" Reply. Detailed time-resolved char size distribution analysis has only been done on the CO2 profiles from t h r e e coal particles. Because the amount of data was limited, only the time between fragmentations and the n u m b e r of particles produced during a fragmentation were determined; the statistics of these two variables are given in the accompanying table. There was no trend in the value of either variable from one generation to the next. The average number of particles produced during a fragmentation was approximately the same for all three particles. However the average time between fragmentations did vary from particle to particle.
Coal Particle
Time to Next Fragmentation (min)
Number of Particles Produced During a Fragmentation
106 118 123
7.9 -+ 4.6 4.5 -+ 1.7 2.8 -+ 1.5
3.1 -+ 1.2 3.5 -+ 0.9 4.1 -+ 1.4
REFERENCES 1. ANDREI, M. A.: S.M. Thesis, D e p a r t m e n t of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA, 1978. 2. MASSIMILLA, L.: Discussion at this Symposium, 1984.
1503