448
Energy & Fuels 1990,4, 448-452
Drying Kinetics of Lignite, Subbituminous Coals, and High-Volatile Bituminous Coals Ramin Abhari and Leslie L. Isaacs* Department of Chemical Engineering, City College of New York, 140th Street and Convent Avenue, New York, New York 10031 Received May 28, 1990. Revised Manuscript Received June 20, 1990
Room temperature drying kinetics of Beulah-Zap Lignite and Wyodak subbituminous, Illinois No. 6, Utah Blind Canyon, West Virginia Lewiston-Stockton Seam, and Pittsburgh No. 8 HV bituminous coals from the Argonne coal sample bank were investigated by use of gravimetric and derivative gravimetric analysis under nitrogen. A bulk moisture/pore moisture model is proposed that s u c m y explains the observed kinetics. The pore moisture loss is found to exhibit a -dX/dt = kXn type kinetics where n increases with rank from 1.3 to 2.4. The kinetics data support the applicability of a desorption/adsorption mechanism for the drying process.
Moisture removal from coal is always an important part of coal processing. Other than the obvious problems associated with handling wet coal (such as extra weight during transport and higher total activation energy for burning), it has been shown' that the dewatering of coal reduces emission of nitrogen oxides during combustion. Empirical rate equations are used for designing dryers. Furthermore, the mechanism for coal dehydration may provide information about coal reactivity and structure. Other reactions that occur along with dehydration (such as oxidation) can be more accurately characterized when the rate of dehydration is known. In addition, since most of the moisture in coal is believed to reside in the pore structure, a drying mechanism will lead to a better understanding of the interaction between water and the pore surface. Recently, Vorres and c o - ~ o r k e r studied s ~ ~ ~ the kinetics of dehydration under vacuum and rehydration under nitrogen, around room temperature, for three coal samples: Illinois No. 6 (HV bituminous), Wyodak-Anderson (subbituminous), and Utah Blind Canyon (HV bituminous). Among their findings are the following: (1)the particle size and degree of oxidation of the samples are variables; (2) for -100-mesh samples, dehydration and rehydration can be explained as processes controlled by desorption and adsorption kinetics, respectively. However, the study was qualitative and did not include any coefficients for the rate expressions. In the studies reported here the emphasis was on determining rate coefficients and dryer design parameters. We investigated all the coal samples in the Argonne Premium Sample (APS) bank. The kinetics results obtained support the mechanistic model suggested by Vorres et al. Thermogravimetry is a suitable technique for the study of dehydration. Kinetics of the dehydration process is obtained from isothermal thermogravimetry. Thermogravimetry as a function of temperature, in conjunction with calorimetry, yields dryer design parameters, e.g., Rebinder numbers. Thermogravimetric data were ob(1)h a y , Blaine W.;Smoot, L. D o u g h Hedman, Paul 0. Combust. Sci. Technol. 1983,35, 15-31. (2) Vorres, Karl S.;Kolman,Roger; Griswold, Timothy Prepr. Pap.Am. Chem. SOC.,Diu.Fuel Chem. 1988,33(2),333-342. (3)Vorres,Karl S.;Kolman,Roger Prepr. Pap.-Am. Chem. ~ o c .Diu. , Fuel Chem. 1988,33(3),7-16.
0887-0624/90/2504-0448$02.50/0
tained for the Argonne Premium Coals by using both modes of operation.
Experimental Details The list of the samples investigated are presented in Table I. In referring to each sample, only the letter designation next to the name shall be used. All samples were -100 mesh and were purchased from the Argonne National Laboratory Premium Coal Sample program. These coal samples along with two other medium and low-volatile bituminous samples, Upper Freeport and Pocahontas (drying kinetics not analyzed here because of very low moisture), constitute an eight-sample set. All the samples were kept in sealed ampules until use. A Du Pont thermogravimetric analyzer (TGA) with the capability to perform derivativethermogravimetry (DTG) was used for the experiments. The experimental accuracy was 1% in mass-loss determination. Traditionally, samples for coal analysis are stored in a moisture-rich atmosphere. This is done to avoid moisture loss prior to analysis. However, in order to start the analysis under a moisture-free initial condition, we chose to permit preanalysis moisture loss. The TGA was placed in a plastic enclosure where a continuous flow of nitrogen was maintained. Sample ampules were kept in the same enclosure for at least 24 h prior to their use. The coal samples in the ampules were homogenized by vigorous shaking and tuning of the ampules prior to use. The seal of the well-mixed ampule was broken inside the enclosure, and a microspatula full of sample (15-30 mg) was placed into the TGA pan. The sample formed a loose layer of particles in the pan. This experimental proviso assured that the observed rates were due to the diffusion and evaporation of water from the coal particles. The balancebeam unit (with the sample pan and thermocouple) was placed into the furnace unit where nitrogen flow provided a purge. The TGA was set to collect data automatically at 32 "C for the isothermal runs. For the temperature runs the TGA heating rate was set to a heating rate of 10 OC/min. The total time between the breaking of the seal and the start of the data recording did not exceed 5 min for any of the samples. During this time interval we have no knowledge of the moisture-loss dynamics. The nitrogen used to provide the inert atmosphere and the purge in the TGA came from room temperature evaporation of liquid nitrogen. This resulted in a flow rate of 0.5 L/min. The isothermal runs were continued for 160 min for the IL, WY, UT, ND, and PIT samples and 130 min for the WV sample. At the end of these time periods no further weight changes were observed; hence it was assumed that the coal-water system had reached steady-state conditions. The accuracy of the isothermal operation was checked by plotting the temperature of the sample 0 1990 American Chemical Society
Energy & Fuels, Vol. 4, No. 5, 1990 449
Drying Kinetics of Lignite and Coals coal Beulah Zap Wyodak Anderson Illinois No. 6 Blind Canyon Lewiston Stockton Pittsburgh No.8 Upper Freeport Pocahontas No. 3
Table I. The Argonne Premium Coals” designation MAF C,b wt % X,(300 K) Xo X, ND 72.94 0.476 0.3934 0.106 WY IL UT
75.01 77.67
WV
82.58 89.20 85.50 91.05
0.391 0.087 0.049 0.025 0.017 0.011 0.007
80.69
PIT PA
VA
0.4295 0.0607 0.0360 0.0153 0.0110
0.127 0.022
k, min-’
n
0.408 0.133 1.422 2.389 10.23 92.17
1.5 1.3 1.6 1.8 1.9 2.4
b, m i d 0.1091 0.0607 0.0832
c, m i d 0.0215 0.0271 0.0052
’For a compilation of Argonne Premium Coal properties refer to the Users Handbook ANL/PCSP-89/1. Prepared by K. S. Vorres. *wt % carbon content on a moisture- and ash-free basis is used as the ranking parameter. 1.0 3
295,
1 - {040
2
0
7 20 40
0 1 0I 60 80 100 120 140 160 180 Time ( m i n )
Figure 1. TGA/DTG output of isothermal dehydration data for Illinois No. 6, which is typical of all coals studied. 0 004
.-C
X
0.003-
X
E \
X
c 0.002-
X
X
c
U \ x
p
xx
0.001-
0
”
“X
0004
0008 X
0012
0016
Figure 2. Drying rate curve of Stockton coal, typical of lowmoisture coals. as a function of time. For most samples there was an increase from 28-30 to 31-32 O C in the first 5-10 min and fluctuations around 31-32 “C for the rest of the run. This deviation from ideal isothermal behavior was regarded as insignificant, and no heattransfer effects were considered in the kinetic analysis. The collected derivative of weight data were smoothed by using
the General Analysis software of the TGA. Where the smoothed curve did not approximate the collected data accurately, a manual smoothing was performed. For each sample the time, weight, and smoothed derivative of weight data were entered (0.5-min interval) into a Lotus 1-2-3worksheet for data analysis.
Results and Discussion Coal is a complex mixture of organic matter, inorganic matter, and moisture. Each of these components contributes to the overall response when coal is subjected to external stimulation, heating, for example. Mraw and Naas-O’Rourke4 have studied the low-temperature heat capacity behavior of moisture in Wyodak coal. They concluded that the moisture associated with the coal can be classified as to two types: freezable water, which behaves in the fashion of bulk water, suggesting that it exists as a distinct macroscopic phase; and nonfreezable water, which resides in the form of small molecular clusters in the microscopic-size pores of the coal. The partitioning of moisture into freezable and nonfreezable portions is not unique to the Wyodak coal but occurs in general for all the lower rank coals5 and in porous media.4’ As will
X
Figure 3. Drying rate curve of Wyodak coal, typical of highmoisture coals. become apparent, the kinetics of coal drying is strongly influenced by this partitioning of the moisture into pore moisture and bulk moisture. Isothermal Experiments. The measured TGA and DTG data were obtained as direct outputs in the form of weight and derivative of weight as functions of time. A typical result is presented as Figure 1. Note that the quantity “derivative of weight” as recorded by the TGA is the rate of loss of sample weight per unit time. The weight loss is due to the evolution of moisture from the coal. In order to make valid comparisons, all the weight data were normalized to a “dry” basis. The word dry is put in quotations because the sample is not completely moisture-free a t the end of the run, but rather it has reached a characteristic weight (2’ = 32 “C)at which no further dehydration occurs. Using this weight as the dry basis, the variable X is defined as the “concentration” of water in coal. The expression is
The totalmoisture content (ASTM values per dry basis) for each coal is presented in Table I. This represents an upper limit for the amount of moisture each sample contains and is referred to as X,, in Table I. Thus X,,, minus Xo (the moisture content at the beginning of the run) is an approximation to the moisture loss before data collection began. It is to be noted that Xofor the WY coal is greater than Xm,. This anomaly might be the result of a systematic instrumental malfunction during this particular experimental run. Hence the numerical results for this particular coal may be in error. From eq 1,an analogous rate variable is obtained that is directly proportional to the measured “derivative of weight” quantity: -dX/dt = [-dW/dtl-l/ Wdrycoal (2) (4)Mraw, S. C.;Naas-O’Rourke, D. F. Science 1979, 205, 901-902. (5)Tsafantakis, E.;Isaacs, L. L. Prepr. Pap.-Am. Chem. Soc., Diu. Fuel Chem. 1987,32, 243. (6)Hoeve, C.A. J.; Tata, A. S. J . Phys. Chem. 1978,82, 1660. ( 7 ) Cruz, M. I.; Letellier, M.; Fripiat, J. J. J. Chem. Phys. 1978, 69,
2018.
Abhari and Isaacs
450 Energy & Fuels, Vol. 4 , No. 5, 1990
Graphs of -dX/dt vs X demonstrate two basic types of behavior. For low-moisture coals the data points form a concave up increasing curve as illustrated in Figure 2 for the WV coal. For the high-moisture coals such as the WY coal, the data points follow a sigmoidal curve, as illustrated in Figure 3. The sigmoidal behavior implies the existence of an inflection point in the function describing the data. The inflection point value of X is designated Xcr. It is a characteristic parameter for a given coal. The kinetic observations must be interpreted in terms of the two types of moisture present in the coals: (1)Due to pore moisture. This type of moisture, which resides in coal's micropore structure, solely accounts for the concave up drying rate. In other words, the drying rate for this type of moisture may be approximated as kX" ( n > 1). The nature of the moisture/pore surface interaction must be consistent with the mechanism of pore dehydration kinetics. (2) Due to bulk moisture. This type, which exists as a thin film around the coal particle, blocks some of the pores and alters the pore water dehydration kinetics for high-moisture coals. Bulk moisture evaporates completely by the time the moisture content is reduced to the characteristic moisture content Xcr. We use the word "evaporate" to refer to the bulk moisture removal because it seems that the drying rate curve approaches a constant value (zero-order kinetics) with increasing X. Although this may not be completely accurate, it is consistent with our observations. For low-moisture coals, where sigmoidal functionality of the data points is not observed, X,, may have been reached very early in the dehydration process, i.e., during the initial experimental time interval where we have no knowledge of the loss dynamics. Alternatively, it is also possible that low-moisture coals contain only pore water. Values of X, are listed for the high-moisture coals in Table I. Pore Moisture Loss Kinetics. A rate expression for pore water loss is obtained by fitting the data to -dX/dt = kX"
(3)
In the above expression, k is the rate constant and n the order of the dehydration. For high-moisture coals only the data in the region X = 0 to X = X, were used. The values of k and n that give the best fit are included in Table I. Throughout the work, "best fit" refers to a correlation of 0.99 or higher as determined by the Lotus 1-2-3 regression program.
Inferences about Mechanism of Pore Moisture Loss. From Table I we notice a correlation between rank and the order of drying, n. As rank increases, so does n. This relation is expressed graphically in Figure 4. Again, note that the rate coefficients for the WY coal are out of line with the rest of the results. In retrospect we conclude that the X values calculated for this coal have a systematic error. The actual instantaneous weights must be smaller than the recorded weights (scale offset). Since 0 < X < 1,this implies that as X approaches 0, a low-rank coal loses pore water faster than a high-rank coal of the same moisture content. Coal pore surface area decreases with rank.8 A mechanism traditionally used to describe kinetics of dehydration from a porous media is the adsorption/ desorption m e c h a n i ~ m . ~ The simplest case for this mechanism is the Langmuir model. Applied to our case, the desorption process of eq 4 has the rate given by eq 5: (8)Mahajan, 0. P. In Coal Structure; Meyers, R. A., Ed.; Academic: New York, 1982. (9) Forment, Gilbert F.; Biechoff, Kenneth B. Chemical Reactor Analysis and Design; Wiley: New York, 1979; Chapter 2.
2.41
I
c 224
20
72
74
76
80
76
82
84
W t %C ( M A F )
Figure 4. Rank dependence of drying order n in rate = k X n . 307
I
,008
Temperoture ("C)
Figure 5. TGA/DTG output of constant heating rate dehydration data for Illinois No. 6, which is typical of all coals studied.
pore water = water vapor
+ pore surface
(4)
rate = &[pore water] (5) Similarly, for the adsorption process (reverse of eq 4)we have rate = k,[water vapor][pore surface] (6) Vorres and c o - w o r k e r ~have ~ ~ ~shown that the adsorption step is much slower than the desorption step. Thus for a process involving simultaneous desorption and adsorption, as is the case with dehydration under nitrogen, the overall rate is controlled by eq 6. As a result, a larger pore surface area (i.e., lower rank) implies a larger overall drying rate. This corresponds to our observations. The drying stops after the moisture content of the space surrounding the coal sample becomes negligible due to the removal of the moisture by the purge gas. It is of interest to understand the nature of the interaction of the water with the pore surface. Is the porewater interaction physical or does it involve the formation of a chemical bond? The magnitude of the activation energy gives a clue to this. By the Arrhenius relation one may estimate the activation energy E , from the expression E, = RT In ( k o / k ) (7) where ko is the preexponential constant and k is the rate constant. Thus the ratio of k o / k determines whether the Langmuir physisorption model, assumed here, is valid or not. For E , to be positive, k o / k must be greater than 1. For an assumed value of k o / k = 100 and T = 300 K the calculated activation energy is approximately 11.5 kJ/mol H,O. This is an order of magnitude less than the usual value of an activation energy for a chemical reaction. The value of ko/ k would have to be of the order of lom,to have an E, of chemical reaction magnitude.
Model and Rate Expression for Bulk Moisture Effects. For coals of high moisture, the mixed kinetics (i.e., bulk moisture effects on pore moisture loss) for X
X,,follows the expression
>
Energy & Fuels, Vol. 4, NO.5, 1990 451
Drying Kinetics of Lignite and Coals
-dX/dt = bXn-' - c
(8)
A model that explains the above kinetic expression can be developed if we regard the high moisture drying rate as a "corrected" form of strictly pore moisture loss kinetics. Our assumption is that the bulk water surrounds the coal particle such that most of the pores are covered. As the bulk water evaporates, the pores become uncovered and pore dehydration occurs. The driving force for this process is the developing pressure gradient. The mixed kinetics should be expressed as a function of an exposure factor, a, defined as a = (pore surface exposed)/(total pore surface) (9)
The rate may be expressed by drying rate = (a)(pore moisture drying rate) (10) ncnwwbwbww
5 0 0 0 0 0 0 0 0
2
Equation 8, which is the empirical best fit curve, and eq 10, which is based on a model expressing the interaction of bulk and pore moisture, are equivalent if the exposure factor is inversely proportional to X. Note that the constants b and c have units of [l/min] and are associated with the zero-order rate constant of bulk water evaporation. In the limiting case of X = 1 the rate as given by eq 8 becomes a constant b - c. This zero-order rate constant should be rank independent; we observe it to be 0.0876/min for ND and 0.0780/min for IL. The values of b and c are presented in Table I. Constant Heating Rate Experiment. From the perspective of the process designer, the isothermal drying kinetics data alone are insufficient. For the design of coal dryers and for the estimation of heat loads, information is required on the dependence of the moisture content with temperature and on the dependence of the rate of change of the moisture content with temperature. Such data were obtained a t constant heating rates of 10 "C/min. Figure 5 is typical of the data collected between 25 and 440 "C. On heating the coal sample, we observed two temperature regions of mass loss. Moisture was evaporated between room temperature and 160 "C. Above 320 "C the mass loss was due to the decomposition of the mineral matter and to the evolution of reaction products from the organic matter in the coals.1° The upper limit of the temperature was chosen to be 440 "C so that pyrolytic decomposition of the coal organic matter was avoided. The TGA data are summarized for all the coals using a modified form of the previously defined X parameters, X', in Table 11. X ' is given by X ' = [W(T "C) - W(160 "C)]/W(160 "C) (11)
b? 2
$c a= wmbommwmmmmm t - b e m m m m ~ e
" "1909999888 0 0 0 0 0 0 0 0 0 0 0 0
I2
3
8 .d
When drying a wet material, energy must be supplied for moisture evaporation and for raising the temperature of the residual material. For a given heating rate, dT/dt, and drying rate, dX'/dt, the required heat flux, q, per unit material surface can be estimatedL1as q = [RdH(dX'/dt)][l + (C/H)(dT/dX?] (12) Here R is the volume to surface ratio of the heated material, d is the density of the dry material, H is the latent heat of evaporation, and C is the heat capacity of the wet material. C may be estimated as c = c, + CIX' (13) where C, and C1are the heat capacities of the dry material and of the liquid moisture respectively. (10)Isaacs, L. L.; Tsafantakis, E. To be published. (11) Strumillo, Czeslaw; Kudva, Tadeasz. Drying: Principles, Applications and Design; Gordon and Breach: London, 1986; Chapter 3.
Energy & Fuels 1990,4, 452-455
452
The quantity (C/H)(dT/dX? defines the Rebinder number, Rb, as a dimensionless group. It determines the ratio of heat used to raise the temperature of the wet material to the heat used to evaporate the moisture present in the material. The Rebinder numbers calculated for these coals are also tabulated in Table 11. Previously obtained5 heat capacity data were used to calculate Rb.
Acknowledgment. The material in this paper was part of the data presented at the Pacifichem89 conference.1° The work was supported by a grant from CUNY and by the E. I. du Pont de Nemours and Co. R.A. also received support from the School of Engineering of City College to perform this work in partial fulfillment of the requirements of the Masters of Chemical Engineering degree.
Extraction of Argonne Premium Coal Samples with CS2-N-Methyl-2-pyrrolidinoneMixed Solvent at Room Temperature and ESR Parameters of Their Extracts and Residues Toshimasa Takanohashi and Masashi Iino* Chemical Research Institute of Non- Aqueous Solutions, Tohoku University, Katahira, Aoba-ku, Sendai 980, Japan Received May 2, 1990. Revised Manuscript Received July 23, 1990
The eight Argonne Premium Coal Samples were extracted a t room temperature with CS2-Nmethyl-2-pyrrolidinone(NMP) mixed solvent. Upper Freeport, Pittsburgh No. 8, Blind Canyon, Illinois No. 6, and Lewiston-Stockton coals gave high yields, i.e., 59.4, 39.0, 33.6, 33.1, and 27.1 wt 70 (daf), respectively. On the other hand, Pocahontas No. 3, Wyodak-Anderson, and Beulah-Zap coals gave only low yields, i.e., 2.8,9.9, and 2.3 wt % (daf), rsepectively. The extracts obtained were fractionated with acetone and pyridine, respectively. For the extract from the Upper Freeport coal, a considerable amount of pyridine-insoluble fraction, i.e., a heavier extract fraction than preasphaltene, was obtained. ESR parameters of spin concentration, line width, and g value for the raw coals and their extracts and residues were determined. The spin concentration was the order of extract fractions < raw coal < residue for all coals used. The extract fractions also had larger line widths than those of the residues.
Introduction The heterogeneous and complex nature of coal often gives us much trouble. The chemical and physical properties of coals depend critically on the sample history of the coal, i.e., place of mining and how it was mined, transported, and stored.'P2 Many researchers in coal chemistry around the world are using the Argonne Premium Coal Samples. These samples donated through this program are chemically and physically as identical and free Oxidation or weathering of coals of oxidation as p~ssible.~ has been reported to decrease extractability with organic solvent^.^^ The Argonne Samples are expected to give the "true" extractability with little effect of oxidation. CS2-N-methyl-2-pyrrolidinone (NMP) mixed solvent gave high extraction yields (40-65 wt 70,daf) a t room temperature for many bituminous No occurrence (1) Klein, R.; Wellek, R. Sample Selection, Aging, and Reactiuity of Coal; Wiley-Interscience: New York, 1989. (2) Monthiony, M.; Landais, P. Fuel 1987,66, 1703-1708. (3) Vorres, K. S. Prepr. Pap.-Am. Chem. SOC.,Diu.Fuel Chem. 1987, 32 (4), 221-226. (4) Buchanan, D. H.; Osborne, K. R.; Warfel, L. C.; Mai, W.; Lucas, D. Energy Fuels 1988,2, 163-170. (5) Seki, H.; Ito, 0.: Iino, M. Fuel 1990, 69, 317-321. (6) Isaacs, J. J.; Liotta, R. Energy Fuels 1987, I , 349-351. (7) Iino, M.; Kumagai, J.; Ito, 0. J . Fuel SOC.Jpn. 1985, 64, 210-212. (8) Iino, M.; Takanohashi, T.; Ohsuga, H.; Toda, K. Fuel 1988, 67, 1639-1647. (9) Iino, M.; Takanohashi, T.; Obara, H.; Tsueta, H. Fuel 1989, 68, 1588-1593.
of significant reaction between a coal and the solvents was reported.* The characterization of the chemical structure of the extracts and residues from five bituminous coals, including Upper Freeport coal, which is one of the Argonne Premium Coal Samples, was carried The measurements of ESR parameters, i.e., spin concentration, line width, g value, spin-spin relaxation time, and spin-lattice relaxation time, for various coals and the extracts and residues after extraction have been performed to obtain information about the coal Yokokawa'O reported a decrease in the spin concentration of coals after solvent treatment. Duber et a1.l' also reported a decrease in the spin concentration for the extracts and the residues after extraction. They suggested that these decreases in spin concentration were a t least partly attributable to the recombination of coal radicals. On the other hand, Seehra and co-workers12 reported that the weight-average spin concentration of the extracts and residues after extraction with NMP at 202 "C are nearly equal compared to that of the raw coals, unlike the cases in the above literature.lOJ1 This may be attributed to the difference in the extraction temperature (202 " C ) com(10)Yokokawa, C. Fuel 1969, 48, 29-40. (11) Duber, S.; Wieckowski, A. B. Fuel 1984, 63, 1641-1644. (12) Seehra, M. S.; Ghosh, B.; Zondlo, J. W.; Mintz, E. A. FuelProcess. Technol. 1988, 18, 279-286. (13) Ito, 0.; Seki, H.; Iino, M. Bull. Chem. SOC.J p n . 1987, 60, 2967-2975.
0887-0624/90/2504-0452$02.50/0 0 1990 American Chemical Society