Energy & Fuels 1988,2,423-430
423
Spectral Emittance Measurements of Coal Particles?*$ L. L. Baxter,*vs T.H. Fletcher, and D. K. Ottesen Sandia National Laboratories, Livermore, California 94550 Received February 3, 1988. Revised Manuscript Received M a y 2, 1988
The spectral emission characteristics of coal are examined by using Fourier transform infrared (FTIR) emission spectroscopy. The data are collected from narrowly size-classified samples of coal and graphite placed on a heated NaCl window with temperatures ranging from 120 to 200 OC. The useful wavenumber interval of the data is approximately 600-4000 cm-'. Spectral and total emittances for particles with mean diameters of 40 and 115 pm and ranks ranging from lignite to bituminous coal are included. The focus of this work is to determine the effect of nongray emissions on particle temperature measurements by pyrometry and on radiative-heat-transfer rates for particles. Both rapid spectral variations of the particle emittance arising from chemical functional group absorption and slower variations due to changing background emittance are analyzed to this end. The dependences on temperature, particle size, rank, and extent of devolatilization are also characterized, with the emittance generally increasing with size and rank and becoming more gray with increasing size and extent of devolatilization. Emittance spectra of coal particles were found to be nongray at particle sizes of interest to pulverized coal research and applications. The errors in pyrometry measurements induced by the nongray emittance were found to vary from a few hundred to thousands of kelvins, depending on the particle temperature and wavelength regions in which the pyrometer operates. The spectral emittances showed only slight dependence on temperature. Total emittances for radiative heat transfer varied from 0.58 to 0.97, depending on the radiative environment of the particles. Planck-weighted total emittances ranged from 0.77 to 0.95, depending on the size and rank of the particles and the effective temperature of their environment. This range of emittances does not differ greatly from commonly used values. The total emittance generally decreased with increasing temperature of the environment.
Introduction The determination of kinetic rate constants from coal devolatilization data depends strongly on the particle temperature history. The rapid heating and reaction rates that typify devolatilization reactions in typical burners complicate both experimental and theoretical determinations of the particle temperature. This paper presents an investigation into the spectral emittance characteristics of coal particles and evaluates the impacts of these findings on pyrometric temperature measurements and emittance coefficients for use in radiative-heat-transfer calculations. Spectral and total emittances of coal have been reported in the past to range from 0.1 to near unity. Brewster and Kunitomo,l Huntjens and van Krevelen,2 and most recently Solomon and c o - ~ o r k e r s , have ~ - ~ published results which indicate that coal is not a strong absorber of radiation in infrared regions. Values of the imaginary part of the complex index of refraction on the order of 0.05 are reported by these authors. Other authors report much higher imaginary coefficients, on the order of 0.30, indicative of higher absorbance and emittance of radiation.G8 A strong and irregular dependence of spectral emission would be expected of an organic compound containing a variety of chemical functional groups, such as coal, if the material is either generally transparent or very thin. Such 'Presented at the Symposium on Coal Pyrolysis: Mechanisms and Modeline. 194th National Meeting of the American Chemical Society, New-Orleans, LA, August 31-%eptember 4, 1987. *Researchconducted at the Combustion Research Facility, Sandia National Laboratories, Livermore, CA, and sponsored by the U.S. Department of Energy through the Pittsburgh Energy Technology Center's Direct Utilization Advanced Research and Technology Development Program. Work performed while a student at Brigham Young University on leave at Sandia National Laboratories.
0887-0624/88/2502-0423$01.50/0
results are reported by Solomon and c o - ~ o r k e r s ~for -~ particles less than 40 pm in diameter. Commonly available infrared absorption and diffuse reflectance spectra of coal samples are consistent with the spectral features reported by Solomon. As the particle size increases, these features should become indistinguishable from the background absorption. The degree of nongray behavior in particle sizes of interest to pulverized coal combustion is critical to the proper analysis of pyrometers and radiative heat transfer. Large or highly absorbing particles should show less variation of emittance with wavenumber; they should become approximate graybodies. The overall emittance spectrum of coal includes broadband absorption in addition to the peaks discussed in the preceding paragraph. The broad-band absorption of coal probably arises from electronic excitations of 7r electrons in the graphitic, aromatic bonds in the coal matrix.+" These electrons are loosely bound by the nuclei and can absorb radiation over a continuous wavenumber region that extends far into the infrared. The electrons are not (1) Brewster, M. Q.; Kunitomo, T. ASME Trans. 1984, 106. ( 2 ) Huntjens, F.J.; van Krevelen, D. W. Fuel 1954, 33. (3) Solomon, P. R.; Carangelo, R. M.; Best, P. E.; Markham, J. R.; Hamblen, D. G. Presented at the Twenty-First Symposium (International) on Combustion,Munich West Germany, 1986. (4) Solomon, P.R.; Carangelo, R. M.; Best, P. E. Fuel 1987, 66. ( 5 ) Best, P. E.; Carangelo, R. M.; Markham, J. R.; Solomon, P. R. Combust. Flame 1986, 66. (6) Foster, F.J.;Howarth, C. R. Carbon 1968, 6. ( 7 ) Blokh, A. The Problem of Flame as a Disperse System; Halsted 1974. (8) Cannon, C. R.; George, W. H. Proceedings of the Conference on Ultrafine Structure of Coals and Cokes; BCURA London, 1944. (9) Berry, R.S.;Rice, S. A.; Ross, J. Physical Chemistry; Wiley: New York, 1980. (10)Berkowitz, N. The Chemistry of Coal; Elsevier: Amsterdam, 1985. (11) van Krevelen, D.W. Coal; Elsevier: Amsterdam, 1981.
0 1988 American Chemical Society
Baxter et al.
424 Energy & Fuels, Vol. 2, No. 4, 1988
entirely free from nuclear attractions, however, and their emission spectra would not necessarily be expected to follow Planck's law. Therefore, even the background radiation from coal may not be gray in its characteristics. The literature cited above indicates that the spectral emission of coal particles a t sizes of importance to pulverized coal combustors (50-150pm) is potentially nongray and could depend in complicated ways on particle size, coal rank, temperature, and extent of reaction. The emission would be expected to behave more like a gray body with increasing particle size, coal rank, temperature, and extent of reaction. Although all of the literature suggests that there is some size at which the coal particle emission is nongray, no study has been sufficiently definitive to quantify such a size. The purpose of this work is to evaluate the extent of nongray emittance spectra for particles in the size ranges, ranks, and extents of devolatilization of interest in coal combustion applications and experiments and to evaluate the impact of these nongray emissions on pyrometry and heat-transfer calculations. Emission Fourier transform infrared (FTIR) spectroscopy was used to measure the emittance spectra. The specific experimental technique combined with the data analysis used in this study is not the most common approach and, to the knowledge of the authors, is unique to this experiment among the attempts to measure particle properties with emission FTIR. Emission FTIR spectroscopy avoids many of the limitations of alternative techniques, such as issues with surface roughness, radiative reflection, and maximum particle size. The major limitations in emission spectroscopy are associated with signal strength, data analysis, and atmospheric interference through absorption by H 2 0 and COz. The experimental work reported in this paper was designed to make best use of the advantages of emission FTIR and to minimize the impact of these limitations. In both the pyrometry and heat-transfer investigations, the results are analyzed as a function of particle size, coal rank, and extent of reaction. Particle sizes of 40 and 115 pm were chosen to represent a typical range of sizes in pulverized coal applications and experiments. Bituminous and subbituminous coals and a lignite were examined to represent the range of coal ranks of interest in coal combustion applications and experiments. Both raw coals and partially devolatilized coals were used to reflect the dependence of the results on extent of reaction. The effect of variations in particle temperature is also illustrated, although the particle temperatures are all less than 500 K. Specific meanings of several terms that are not always consistently used in the literature are intended in this discussion. The emissivity of a sample is taken to represent an inherent optical property of a material and is determined by its physical and chemical structure. It is a spectral quantity and should be independent of the shape or size of the material. It can be quite different from the emittance of the sample due to effects of surface area with very small particles, self absorption with large particles, and interference between the particles and the radiation when the particle size parameter is comparable to the wavelength of the radiation. The emittance refers to an effective emissivity of a material when its size, shape, inherent emissivity, temperature, and other properties are specified. Emittances can be either spectral or total. Both are discussed in this paper. Both the emissivity and the emittance are dimensionless properties. The radiance is a measure of the actual energy emission from a material and is not dimensionless. Radiances can also be either
TOP VIEW COAL
NDOW MOUNl
POWD
HgCaTe DETECTOR
SAMPI IN HOR
0
T
IRIS
Figure 1. Schematic diagram of experimental apparatus used to determine spectral emittance of particles.
spectral or total, and both are used in the data reductions discussed below. These definitions are not necessarily standard but are useful in clarifying the following discussion. Experimental Procedure Figure 1 is a schematic illustration of the experimental equipment and optical design employed in this study. A single layer of sized particles of coal and graphite was placed on a horizontal window of NaC1, which in turn was placed in an aluminum mount. The NaCl window was used because it can withstand more heat without degradation and has a higher thermal conductivity than commonly available alternative windows. An electric strip heater was wrapped around the aluminum mount and insulated from the outside. This arrangement produced particle temperatures as high as 400 "C without substantial degradation of either the NaCl window or the optical signal. However, the stability of the coal samples was uncertain a t the high temperatures, and the temperatures used for the spectra in this report seldom exceeded 200 "C.The heated sample, optics, interferometer, and detector were enclosed in a nitrogen-purged container. The strip heater and all other warm surfaces other than the window and the particles were kept out of the optical path both in front of and behind the window to avoid ambiguities in the collected interferograms. Standard coal samples were selected from The Pennsylvania State University Coal Bank. The coal samples were ground in a nitrogen atmosphere and size classified by sieve trays. The free moisture was removed in a nominally pure nitrogen atmosphere at 305 K. Graphite samples were prepared similarly. All samples were visually examined, and some were photographed after they were placed on the window to ensure they were well size-classified and formed a single layer on the NaCl window. The coal and graphite particles were placed in separate areas on the NaCl window in large enough quantities to provide homogeneous coverage of two 6-mm diagnostic areas. The particles were prevented from entering a third diagnostic area of the same dimension to provide measurements of the radiance from the window. The emitting areas of the graphite and coal and the transmissivity of the window were determined by measuring the extinction of a HeNe laser beam as it passed through the samples and the window. The HeNe beam was expanded, and a central portion of the beam with the same diameter as the diagnostic area (6 mm) was used to minimize errors from gradients in beam intensity. A power meter with a large detection area was used to measure the beam intensities and minimize errors due to scattered laser light. The particles typically covered 50-65 percent of the 6-mm diagnostic area. The illustrated optical train introduced the collected radiance from the particle diagnostic area into the emission port of a Digilab FTS-20 interferometer. The heated sample disk was movable in the horizontal plane, allowing measurement of emission from the coal and graphite through the heated window and from the heated
Energy & Fuels, Vol. 2, No. 4,1988 425
Spectral Emittance of Coal Particles window itself. It was observed that a portion of the signal from the edge of the collecting lens was lost, probably due to overfilling the detector, which effectively reduced the nominal dimensions of the diagnostic area. This had no effect on the results if the particles in the diagnostic area were homogeneously dispersed. The spectra in this paper were produced by averaging 800 interferograms,each with a resolution of 4 em-' (data point every 2 em-'). Boxcar apodization and no zero-fillingwere used in this experiment. Approximately 5 min were required to obtain one spectrum under these conditions. The high number of scans and moderate resolution contributed to the production of a high signal-to-noise ratio over most of the infrared region and peaks with sufficient resolution to allow comparison with literature results. The emittance spectrum of the coal was determined from the following equation
R d A , - Rw/Aw - Rw/Aw
en=( Rr/4
1.0
0.6
1
1
. . a b
: :
/ :
c
be
I :
A
-
..,
.., .,
..
:
h i
jki
L f
.
.
. :. : : m
;:: .q:
i
n o p r s
,
g
(1)
where R is the measured emission intensity (radiance and system response), A is the emitting area, e is emittance, and subscripts s, r, and w refer to the sample (coal or char), reference (graphite), and window, respectively. The symbols R and e represent spectral quantities. Since the measurements were obtained by looking through the window from below, radiance measurements from the sample and reference include radiance from the window as well. Therefore, the numerator in eq 1represents the energy flux emitted by the sample, having corrected for background emission from the window, heated beamsplitter, and other sources. The fraction in parentheses represents the emittance of the sample relative to the emittance of the graphite. Therefore, the right side of the equation is multiplied by e, to obtain the absolute emittance of the sample. The spectral emittance of the graphite was experimentally determined by more common emission FTIR data acquisition and reduction techniques'2 and was found to be essentially gray. In general, the data manipulation indicated by eq 1 should be performed with the interferograms rather than the radiances to avoid improper phase correction of the resulting emittance spe~trum.'~In this application, the magnitude of the radiance from the window was small enough that no benefit was derived from this practice. Emission FTIR is often subject to intereference from atmospheric absorption and emission. This effect was minimized in this experiment by measuring the emittance of a reference body (graphite) under the same conditions as the coal. This experimental procedure is roughly equivalent to individually calibrating the system responsivit for each measurement. This technique, combined with eq 1, yie ded spectra of superior quality to the more common approach of determining a single fixed system responsivity for several or all of the spectra. The particle temperatures were measured with a type K thermocouple placed on the window in the vicinity of the sample. The temperature at the center of the window was typically 5-10 K lower than that at the edge. However, the samples were located equidistant from the window edge to minimize errors due to temperature gradients. The data analysis can be completed without specifying the actual temperature so long as the window, coal, and graphite temperatures in each of the three 6 mm diameter diagnostic areas are equal. The major source of uncertainty in the calculations of the spectral emittance was the determination of the actual emitting surface area. Although the HeNe laser could accurately measure the cross-sectionalarea of the samples within the nominal 6-mm sample, there were indications that a fraction of the transmitted radiation from the diagnostic area was either blocked or overfilled the detector. This would have occurred for both the sample and reference measurement. If the particles were homogeneously scattered on the window, the effect in the numerator and denominator of eq l would cancel, leaving the measured results
i
(12)Griffiths, P. R.; de Haseth, J. A. Fourier Transform Infrared Spectrometry, Wiley: New York, 1986. (13) Kember, D.; Chenery, D. H.; Sheppard, N.; Fell, J. Spectrochim Acta, Part A 1979, 35A.
Table I. Identification of Chemical Functional Groups Resoonsible for SDectral Peaks Found in Coal Samdes' wavenum- alphabetber, em-' ic code(s) chemical functional groups 3700-3300 a, b -OH and -NH stretching 3030 C aromatic C-H stretching 2940 d aliphatic C-H stretching 2925-2860 e aliphatic C-Hi stretching 2370 f atmospheric COz interference 1865 g unknown 1700 h C=O stretching 1600 i "coal" peak (probable aromatic ring structures) 1500-1400 j aromatic C=C stretching aliphatic -CH3 asymmetric deformation aliphatic -CH2 scissor deformation 1370 k -CHS symmetric deformation, cyclic -CH2 1250-1020 1-0 phenolic and alcoholic C-0 stretching aromatic and aliphatic C-0-C stretching Si-0 850-750 p-r polycyclic aromatic skeleton 700 8 aromatic -CH3
'Compare with Figure 2. Most assignments were adopted from Berkowitzloor van Krevelen." unaltered. However, if the particles were inhomogeneouslyspread on the window, some error could be introduced in the calculations. The estimated accuracy of the emittances is h0.05.
Results Figure 2 is typical of the spectra collected in this study. Peaks from various functional groups are coded in this figure a n d identified in Table I. T h e peaks identified in Table I agree closely with those reported in other literature sources.loJ1 One peak, at about 1865 cm-l, is n o t typical of coal spectra collected by means of KBr pellet transmission a n d has n o t yet been identified. T h e peak is sometimes seen in diffuse reflectance measurements of some coals. T h e remaining peaks agree precisely with published spectra of coal collected with a variety of instruments and techniques, although the specific molecular origin of some of the peaks is still debated in the literature. Similar spectral features were found in t h e coals of other ranks. T h e useful wavenumber range over which spectra were collected is approximately 4000-600 cm-'. Signal-to-noise ratios became increasingly worse at high wavenumbers because of t h e low radiance of the particles in this region. T h e low-wavenumber limit is a property of t h e NaCl window used t o support t h e coal particles. T h e spectra obtained show maximum emittances close t o unity in regions of functional group absorption. T h e
Baxter et al.
426 Energy & Fuels, Vol. 2, No. 4, 1988 1.1
1.0
t-
I
--
mprratur. ametrr
1.0
Temperature = 182 'C Diameter =
o.6
3000
2600
1600
2000
1
121 'C
(PSOC 14611
3600
1
40 pm
8icumlnous Coal
4000
1
P ;I
0 . 5 ' , ' , 4000
600
1000
Wavenumber (cm-' )
"
" '
3500
115 p m
Bituminous C o a l (PSOC 1451)
"
"
3000
"
'
2500
"
'
I
"
2000
Wavenumber
'
'
I
'
"
1500
'
I
'
1000
"
" 500
(a")
Figure 4. Spectral emittance of 115-pm bituminous coal particles (PSOC 1451) at 182 "C. Compare with Figure 2.
o.a
t
E
Bitumlnour Coal (PSOC 14511
0 . 5 1 ' ' ' ~ ' ' ' ' " ' ' ~ ~ ' ' ~ ~ ~ ' ' ' ~ ' ' ' " ~ ' 4000 3 5 0 0 3000 2500 2000 1500 1000 600
Wavenumber (cm-')
0.0 OI
2
(cl
0.8
=:
'E w
Tamperatura
Oiamrtar
0.7
Xme
-
-
-
Y1
201 'C
40 p m
2 htc
Bituminous Coal (PSOC 14511
0.6
o.s""""'""""""'"''''~ 4000
3500
2500
3000
ZOO0
1500
1000
600
Wavenumber (cm-',
Figure 3. Spectral emittance of 40 pm hvA bituminous coals at (a) 121 "C, t = 0 min; (b) 171 "C, t = 45 min; and (c) 201 "C, t = 120 min.
absorption of many of these peaks is typically high for submicrometer particles and should not decrease with increasing diameter. It is not impossible that the emittance of these peaks would exceed unity in a single particle measurement where the particle size and shape may interact with the wavelength of the radiation.14 The large number (5oo(t1oooO) and random shapes and orientations of particles involved in this experiment suggest that it is unlikely that such a phenomenon would have affected these measurements. This and other spurious effects of light scattering were further minimized by using a reference that had been ground under the same conditions as ~
~
~
~~
~
(14)Bohren, C. F.;Huffman, D. R. Absorption and Scattering of Light by Small Particles; Wiley, New York, 1983.
the sample and, therefore, would have produced results that canceled in the data analysis. Therefore, the absolute values of these emittances seem to be properly indicated in the spectrum. If, however, the particle emittance did exceed unity, this technique may not have been sensitive to the effect because of the similarity of the coal and graphite particles. Some reaction of the coal was observed when the temperature was held above 150 OC for 2 h or longer. A typical result is the evolution of a C=O peak over a 2-h period, as illustrated in the sequence of spectra in Figure 3. The peak at 1850 cm-' becomes obscured by the C=O peak at ' the ' ' end ' u of the sequence. Other coal reactions that were observed include loss of the hydrogen-bonded hydroxyl peak and a small decrease in the aliphatic C-H stretching band. However, consecutive spectra taken within 1h were reproducible at temperatures below 170 OC and individual spectra did not change, in general, during the 5 min required to collect them. The three spectra shown in Figure 3 also illustrate a weak temperature dependence. The location of the midpoint of some peaks moves to slightly higher wavenumbers as temperature increases, as expected. Higher temperatures also allowed measurements at somewhat higher wavenumbers because of the increased strength of the radiance from the particles. Both of these effects are minimal in terms of the total effective emittance. The spectral emittance for 115-pm particles of a Pittsburgh seam (high-volatile A) bituminous coal is shown in Figure 4. The coal sample was obtained from The Pennsylvania State University coal bank (PSOC-1451) and ground at Sandia National Laboratories in a nitrogen atmosphere. A nominally 40-pm size fraction of the same coal was used to collect the data shown in Figures 2 and 3.
A comparison of Figure 4 with Figures 2 and 3 illustrates the dependence of spectral emittance on coal particle size. All of the functional groups that appeared in the emittance spectra of the smaller particles also appear in the spectrum of the large particles. However, in the case of the larger particles, the peaks are broader, the valleys are higher, and the background emittance has increased. This trend was anticipated, as discussed earlier. The coal emittance spectra illustrated thus far are not well characterized as graybodies. For example, the emittance of 115" particles of this bituminous coal varies from 0.7 to near 1.0 over the range 500-4100 cm-l. The sizes of coal particles used in this study are representative of those used in experimental and applied studies of pulverized coal combustion. These results suggest care should be taken in dealing with this nongray behavior
Spectral Emittance of Coal Particles
8 g .-
Energy & Fuels, Vol. 2, No. 4, 1988 L27
1.1
1.1
1.0
1.0
0.9
0.9
0.8
E w
Temperature
0.6
O"
1
Diameter
-
-
0 7
Subbituminous C o i l (PSOC 14451
0.6
3000
2500
2000
~ 1500
1000
500
Wavenumber (cm-') Figure 5. Spectral emittances of 115-pm subbituminous coal
particles (PSOC 1445) at 180 "C. 1.1
I
t
5 ~ ~ ~ 4000 3500
Diameter
-
115
p m
Bituminous Coal Char (PSOC 14511
~
~ ' 3000
" ~ 2500
"
" ~ 2000
" ~ 1500
~
~ " 1000
"
~
"
500
Wavenumber (cm-')
Figure 7. Emittance spectra of originally 115-rm bituminous
coal particlea (PSOC1451)after 50% m m lass by devolatilization.
The particle temperature during the emittance measurement was 182 "C.
r
I
WSOC 15071
0.6
0.6
4000 3500
t
180
115 p m
0.5
4000 3500
t1
3000
2500
2000
1500
1000
500
Wavenumber (cm-I)
Figure 6. Emittance spectrum of 115-pmlignite particles (PSOC
1507) at 190 "C.
when designing and analyzing experiments. Similar trends have been found for different ranks of coal. The spectral emittances for 115-pm particles of subbituminous coals and lignites are shown in Figures 5 and 6, respectively. Both samples are from The Pennsylvania State University coal sample bank. The subbituminous sample (PSOC1445) is a western coal, blue no. 1, and the lignite sample (PSOC 1507) is a Beulah Zap lignite. The nongray behavior of the lignite is more pronounced than that of either of the higher rank coals, with emittance varying from 0.57 to near 1.0 in the spectral region indicated. However, little indication of aromatic spectral features is present in either of these samples. The dependence of emittance spectra on coal rank can be seen by comparing Figures 5 and 6 to Figure 4. There are several regions in the spectra of both the lignite and subbituminous coal where emissions from water and water vapor are evident. These include the closely spaced peaks near 1500 and 3400 cm-'. These features represent a combination of adsorbed water in these two samples and vaporized water near the particles. The peaks could be eliminated to obtain representative spectra from moisture-free coal. However, they represent the expected real behavior of coal spectra and for the applications considered in this discussion are included in the figures. The interference of atmospheric C 0 2 at about 2400 cm-' is insignificant. Water vapor often purges slower than COz, however, because of stronger surface interactions. The same experimental procedure that was used to collect the spectra from the lower rank coals was used to collect the spectrum shown in Figure 2, which shows an insignificant amount of interference from atmospheric water vapor.
Therefore, the peaks associated with water that appear in the spectra of the lower rank coals arise from inherent moisture of the coal and not atmospheric water vapor. A spectrum of partially devolatilized bituminous (PSOC 1451) coal appears in Figure 7. This sample was prepared by entraining the coal in a 1000 K nitrogen stream in a down-fired,laminar-flowreactor. The approximate volatile yield of the coal was 50% daf under these conditions. The unreacted coal particles used in this analysis were the same as those used to generate Figure 4. The dependence of particle emittance spectra on the extent of coal devolatilization can be seen by comparing Figures 4 and 7. The emittance spectrum of these partially devolatilized coal particles is quite constant at 0.8 for wavenumbers above 1900 cm-l. The aliphatic and hydroxyl groups, which were emitting strongly in the unreacted coal, do not appear in this spectrum. These functional groups typically react early during dev~latilization,'~ which is indicated by this spectrum. However, the more stable aromatic bonds are still evident, as seen by the weak peak at 3000 cm-l and several of the peaks at wavenumbers less than 1900 cm-'. The emittance of the aromatic compounds in the coal, which have peaks between 500 cm-l and 1900 cm-', slightly exceeds that of the parent coal, possibly due to the formation of tar or graphitic bonds during the early stages of devolatilization. Finally, the background emittance did not change appreciably from that of the parent coal. As devolatilization continues, particularly if the remaining char graphitizes, this background emittance would be expected to increase.
Discussion The foregoing figures are representative of the observed effects of particle size, coal rank, and extent of devolatilization on the spectral emittance of coal particles that are of interest for studies of pulverized coal combustion. The implications of these results on both pyrometry measurements and on heat-transfer calculations are discussed in this section. 1. Pyrometry. The emittance spectra of coal particles impact coal combustion research in at least two significant areas. The first is the pyrometric measurement of coal or char temperatures. Such measurements typically are done with multicolor pyrometry. This diagnostic technique is based on the concept that the radiating source is a gray(15)Suuberg, E.M.;Peters, W. A.; Howard,J. B.Proceedings of the Seventeenth Symposium (International)on Combustion; The Combustion Institute: Pittsburgh, PA, 1979.
"
~
"
~
428 Energy & Fuels, Vol. 2, No. 4, 1988 2400
E
L
Barter et a1.
/
c
f
2000 1600
VI
2
1200
E
800
H c
s
a,
400
Q
5
0
.&
-400
I-
L
e
-800 800
1000
1200
1400
1800
1800
Actual Particle Temperature
2000
2200
(K)
Figure 8. Errors in measured temperature that would result from pyrometry measurements if emittances differed at the two wavelengths by a ratio of 8:9.
body. This study investigates the magnitude of the probable errors associated with this assumption and uses the measured emittance spectra to evaluate the impact of this error. The sensitivity of two-color pyrometry to nongrey emissions is illustrated in Figure 8 for three sets of wavelengths. Wavelength is the most common unit used in pyrometric discussions and is, therefore, used in the labels in this figure. The ratio of emittances at A, and X2 was assumed to be 8:9, with the positive errors corresponding to the high emittance being associated with the low-wavelength radiance, and vice versa. Pyrometry measurements at lower wavenumbers (higher wavelengths) are more sensitive to nongray spectral emittances, as illustrated in this figure. The wavelength region measured in this study varied from approximately 2.5 to 16 pm. The magnitude of the potential errors, especially as the wavenumbers used in the pyrometer move further into the infrared region, provides practical motivation for measuring the emittances. Studies have been reported recently that discuss the spectral variation in coal emittance.'^^-^ The most significant conclusion from these is that the assumption of gray behavior is not valid for small coal particles. Our findings agree with this part of the conclusion, at least for emittance between approximately 600 and 4000 cm-'. Much of the work in the previous studies was based on small particles. The same effects are shown here to be important for particles in the size range of interest to pulverized coal combustion. The previous work suggested that particle emittances could be very low, sometimes less than 0.1 in some regions of the spectrum. The lowest emittance observed in this study was approximately 0.55, and although the trends are consistent between this and previous work, the lowest emittances observed in this study for similar coal particle sizes, ranks, and spectral regions are somewhat higher than those previously reported. In any case, the implications for pyrometry measurements are nearly the same; pyrometry measurements in the infrared region which was investigated in this study should not assume that the coal emittance is independent of wavenumber, rank, degree of reaction, or particle size. The radiance from coal particles at low temperatures (prior to and during devolatilization) is weak, and the strength of the signal collected by pyrometers can be increased by operating in the mid-infrared region. Figure 8 illustrates the increased sensitivity of the pyrometer to either nongray emissions or random error in the measurements as the effective operating region of the pyrometer moves further into the infrared region. The benefit
of the improved strength of the radiative signal realized by operating the pyrometer in the mid-infrared region may well be negated or even overwhelmed by this corresponding increase in the sensitivity of the results to both noise and nongray emissions. The spectral emittance shown in Figure 4 will be used to evaluate quantitatively the potential impact of nongray emissions on pyrometry measurements. For example, a two-color pyrometer temperature measurement based on the radiance at 3333 and 2500 cm-l (3 and 4 pm) would overestimate the temperature of a 1500 K particle by 200 K if the particle were assumed to be a gray body. On the basis of radiance measurements at 2900 and 2500 cm-', the error would be about 800 K. Pyrometry measurements based on the radiance at 2000 and 1667 would underestimate the particle temperature by about 700 K. Extreme cases could result in temperature measurement errors of many thousands of kelvins. The variation of these results with particle temperature and wavelength were further illustrated in Figure 8. These errors are not easily accounted for by pyrometry measurements but would substantially impact Arrhenius parameters and similar data determined from experimental studies. For example, adding a third channel to the pyrometer would only add confusion to the temperature measurement of particles with emittance spectra similar to those illustrated above. In typical pulverized coal applications, where particle temperatures change rapidly, an error of several hundred degrees in particle temperature measurement can become an error of orders of magnitude in the devolatilization reaction rate coefficient. Errors in pyrometry measurements due to nongray emittances can be minimized by making one measurement near or in the visible region or by increasing the separation between wavenumbers, the former being more effective than the latter. These trends were illustrated in Figure 8. However, signal strengths at typical combustion temperatures decrease sharply with increasing wavenumber in the near-infrared and visible regions, requiring a judicious balance between acceptable signal-to-noiseratios and sensitivity to this type of error. Random experimental errors enter the error analysis in a mathematically equivalent manner and would have both the same sensitivity and similar means of reduction or elimination. 2. Radiative-Heat-TransferRates. The second area in which coal particle emittance spectra impact coal combustion research and applications is radiative heat transfer. The importance of spectral emittances to radiative heat transfer becomes difficult to evaluate because of the large variety of radiative fields to which particles are subjected in experimental and industrial applications. The issue obviously becomes substantial when the particles are primarily or entirely heated by radiation. Such is the case in a small number of flow reactors and in many experiments involving laser heating of coal particles. If the particles enter hot gases, the radiative contribution to the overall heat-transfer rate typically is quite small. The majority of applications lie between these two extremes. This study reports experimentally determined total emittance spectra for use in radiative-heat-transfer analyses. The results of this study indicate that the total emittances during radiative heat transfer will depend in a complicated way on coal rank, particle size, temperature, and extent of devolatilization. For high-rank coals above 40 p m in size, the emittance could vary between 0.65 and near 1.0. Lignites have a wider variation in emittance. The importance of these variations and the effect they have on
Energy & Fuels, Vol. 2, No. 4, 1988 429
Spectral Emittance of Coal Particles 0.88 I
I '
______._ 110 pm B i t u m l n o u i Cool
_............... i
a
0.88
E
0.86
w -
.
110 p m Lignlte 46 p m Bituminous Coal 115 pm B l l u m l n o v i Char
0.80
5
Y S .-
__
2
-
0.84
$
c
0 0.82
0.55
1
0 5 0 1
0.80 0
0.78
0.76
065 0.60
L? "
400 600
"
800
"
1000
'
I
1200
'
I
1400
'
'
1600
I
'
I
1800 2000
'
I
2200
'
2400
EffectiveBlackbody Temperature (K) Figure 9. Planck-weighted,total emittances as a function of particle size, type, extent of devolatilization, and temperature of radiative environment.
heating rate depend primarily on the combustor configuration and flow field. The emittance spectra illustrated previously can be used to generate representative total emittances for each type and size of particle studied. The value used for the total emittance will depend on the wavenumber dependence of the radiative intensities. The appropriate emittance for monochromatic radiation, such as would result from lasers, can be found directly from the figures. An arithmetic average emittance, which has little practical application, is also easily derived from the figure. However, the following results will illustrate a Planck-weighted emittance, which is defined by
where P(q) is Planck's function (3) and v1 and v2 represent the lowest and highest wavenumbers at which the results in the figures are free from overwhelming noise. C1and C2are constants with values and units of 5.9544 X W/m2 and 1.4388 X lo4 I.cm K, respectively. The temperature sensitivity of the spectral emittances illustrated in the previous figures is small enough to be neglected in this analysis. The radiative fields of the most reactors do not follow black- or graybody curves ideally. However, these total emittances are a reasonable representation of effective emittances and offer a method to determine quantitatively their dependence on coal rank, particle size, and extent of devolatilization. Figure 9 illustrates these Planck-weighted total emittances as a function of particle size, type, extent of devolatilization, and temperature of the radiative environment. Many of the trends discussed earlier are represented quantitatively in the figure. The two results from the bituminous coals illustrate that the effective total emittance increases with increasing particle size. A comparison of the variation of the effective emittance of the lignite and of the bituminous coal illustrates that the emittance becomes more gray with increasing coal rank. The variation in total emittance is smallest for the partially devolatilized coal sample, labeled a char, illustrating that the emittances
0351
\
040
030
'
'
400 600
'
'
BOO
'
'
1000
1200
'
1400
1600
1600
2000
2200
2400
Temperature (K) Figure 10. Fraction of total radiative emission contained in the wavenumber region studied in this investigation. become more gray with increasing extent of devolatilization. The absolute value of the total effective emittance is not greatly different from those commonly used previous to this study. The value varies from 0.77 to 0.95 for the Planck-weighted results, as illustrated in the figure. However, not all of the radiative energy is contained within the wavenumber region studied in this investigation. The fraction of the total energy represented by the spectra reported above can be determined quantitatively if the radiative field is again typified as a black- or graybody. Under this assumption, the fraction of the total radiative energy contained in this wavenumber region, qfiac,is given by
where u is the Stefan-Boltzman constant. This quantity also varies with temperature and is illustrated in Figure 10. The result illustrated in the figure is typical. Small differences are seen for each spectrum, depending on the exact values of q1 and q2, which were assumed to be 550 and 4500 cm-' in this case. As is seen, between 93 and 35% of the total radiative energy is represented by the spectra reported in this study. As little as 20% of the total blackbody radiative energy was contained in the spectra with the most limited range of ql and q2, such as the range shown for the spectrum illustrated in Figure 3a. The total effective emittance in high-temperature black- or graybody environments could differ from those reported here if the spectral emittances outside of the measured wavenumber region differ greatly from those reported here.
Conclusions An emission FTIR experimental technique is demonstrated that provides spectra largely free from atmospheric interference. This technique is used to study emission characteristics of coal particles. Coal particles in the size ranges between 40 and 115 pm show nongray behavior at wavenumbers between 600 and 4000 cm-'. Emittance spectra of high-rank coals vary from 0.7 to 0.98, depending on wavelength, size, and extent of devolatilization. Similar spectra of lignites vary from 0.5 to 0.98. The particles generally are less gray as particle size, rank, and extent of devolatilization decrease. The spectral and total emittances generally increase with increasing rank and particle size. As the extent of devolatilization increases, the particle emittance can either in-
Energy & Fuels 1988,2,430-437
430
crease, decrease, or remain constant, depending on the region of the spectrum being considered. Pyrometry measurements in this wavenumber interval are subject to errors due to nongray effects. The errors in temperature measurement vary from a few hundred degrees to many thousands, depending on the wavenumbers chosen for the channels of the pyrometer. Operating one channel of the pyrometer at a high wavenumber reduces the chance for error.
The effect of nongray emissions on heat-transfer rates will depend on the particular combustor and flow field being used. Total particle emittance ranged from about 0.6 to 0.95 for 115-pm-sized particles, depending on particle temperature and size, coal rank, and radiative environment. Smaller particles from low-rank coals could have lower emittances. Partially devolatilized samples of bituminous coal emitted as gray bodies over a large portion of the infrared, with an emittance of about 0.8.
Intraparticle Nonisothermalities in Coal Pyrolysis+ Mohammad R. Hajaligol, William A. Peters, and Jack B. Howard* Energy Laboratory and Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 Received December 21, 1987. Revised Manuscript Received April 13, 1988
Time and spatial temperature gradients within pyrolyzing coal particles can exert strong effects on devolatilization behavior including apparent pyrolysis kinetics. This paper mathematically models transient and spatial nonisothermalities within a single, spherical coal particle, with temperatureinvariant thermal and physical properties, pyrolyzing by a single first-order reaction. Analyses were performed for two surface heating conditions of practical interest, a constant surface heating rate and a constant surface heat flux density. The treatment provides three distinct indices of heat-transfer effects by quantitatively predicting the extent of agreement between (a) center-line and surface temperature, (b) volume-averaged pyrolysis rate [or (c) volume-averaged pyrolysis weight loss], and the corresponding quantity calculated by using the particle surfac,etemperature for the entire particle volume. Regimes of particle size, surface heating rate or surface heat flux density, and reaction time, where particle "isothermality" according to each criterion (a-c) is met to within prescribed extents, are computed for conditions of interest in entrained gasification and pulverized-coal combustion, including pyrolysis under nonthermally neutral conditions.
Introduction Many coal combustion and gasification processes involve particle sizes and surface heating conditions producing temporal and spatial temperature gradients within the coal particles during pyrolysis. These gradients may strongly influence volatiles yields, compositions, and release rates and can confound attempts to model coal pyrolysis kinetics with purely chemical rate expressions. Mathematical modeling of particle nonisothermalities is needed to predict reaction conditions (viz. particle dimension, surface heating condition, final surface temperature, and reaction time) for which intraparticle heat-transfer limitations do not significantly influence pyrolysis kinetics and to predict pyrolysis behavior when they do. When pyrolysis is not thermally neutral, the analysis is further compounded by the fact that local temperature fields are coupled nonlinearly to corresponding local heat release (or absorption) rates and hence to local pyrolysis kinetics. Much of the pertinent literature has addressed pseudo-steady-state models for spatial temperature gradients within catalyst particles playing host to endothermic or exothermic reactions, including, for some cases, mathe*Towhom correspondence should be addressed at the Department of Chemical Engineering, MIT. Presented at the Symposium on Coal Pyrolysis: Mechanisms and Modeling, 194th National Meeting of the American Chemical Society, New Orleans, LA, August 31-September 4, 1987.
matical treatment of the attendant limitations on intraparticle mass-transfer of reactants or products (see ref 1 and references cited therein). There appear to have been few analyses of nonisothermalities within a condensed phase material simultaneously undergoing nonthermally neutral chemical reaction(s). Previous work includes rather empirical approaches to fitting coal weight loss kinetics (see reviews by Howard2 and Gavalas3 and more refined analyses of spatial nonisothermalities within exploding solid^.^,^ Gavalas3calculated regimes of coal particle size where pyrolysis kinetics should be free of heat-transfer effects, and Simmonse provided similar information for cellulose pyrolysis. Valuable contributions are also emanating from the laboratories of Essenbigh' and Freihaut and Seery.8 (1)Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis & Design; Wiley: New York, 1979. (2) Howard, J. B. In Chemistry of Coal Utilization, Second Supplementary Volume; Elliott, M. A., Ed., Wiley-Interscience: New York, 1981; Chapter 12. (3) Gavalas, G. R. Coal Pyrolysis; Elsevier Scientific: New York, 1982. (4) Boddingtan, T.; Gray, P.; Scott, S. K. J . Chem. SOC.,Faraday Trans. 1982,2, 1721. ( 5 ) Scott, S. K.; Boddington, T.; Gray, P. Chem. Eng. Sci. 1984,39, 1079. (6) Simmons, G. M. Prepr. Pap.-Am. Chem. SOC., Diu. Fuel Chem. 1984. 29(2). 58. (7) Errsenhigh, R. H., Department of Mechanical Engineering, The Ohio State University, Work in Progress, 1987.
0887-0624188l 2 ~ 0 2 - 0 4 ~ 0 $ 0 l .,I5O0 0 1988 American Chemical Society I
,
