Energy & Fuels 1990, 4 , 24-27
24
Two-step Gasification of Flash Pyrolysis and Hydropyrolysis Chars from Low-Rank Canadian Coals Kouichi Miura,t Mitsunori Makino,t and P. L. Silveston* Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario, N2L 3G1, Canada Received J u n e 5, 1989. Revised Manuscript Received October 13, 1989
In this study, chars produced by flash pyrolysis or hydropyrolysis of low-rank Canadian coals were gasified in an air atmosphere using the temperature-programmed-reaction (TPR) method. It was observed that some chars gasify in what appears to be two steps. It is possible to fit the two-step data by use of a simple model that fits the continuous weight loss data well by assuming parallel gasification of two chars of different reactivities. It seems likely that two-step gasification indicates chars of nonuniform reactivity or uneven ash distribution.
Introduction If flash pyrolysis and/or flash hydropyrolysis for production of coal liquids and tars are to be brought into commercial operation in the near future, utilization of the large amounts of chars produced is essential. Combustion or gasification are means of utilizing char effectively. For this reason, the gasification reactivity of coal chars has been under intensive study in our laboratory.lP2 Particular attention has been paid to the temperature-programmedreaction technique as a means for collecting data more conveniently.' The two-step-gasification observations discussed in this paper are drawn from these studies. Because temperature-programmed reaction has only recently been applied to gasification, we expect that as further chars are examined, the two-step phenomena will be often seen. If char gasification is performed at a temperature higher than the pyrolysis temperature, it is possible that the properties of the chars will be altered. In order to eliminate such change as a source of the two-step behavior, gasification temperature was always set below the pyrolysis temperature. Consequently only low-temperature-gasification measurements were made in this study. Experimental data were analyzed by the model proposed by Bhatia and Perlmutter-3and modified for TPR data by the authors.' This model appears to correlate char gasification data well' but was used in this study as a convenient means of summarizing gasification rate data. Experimental Section Samples. Flash pyrolysis chars prepared from three different coals and flash hydropyrolysis chars prepared from a single coal were used. The ultimate analyses of the parent coals are listed in Table I, and the char preparation conditions are summarized in Tables I1 and 111. Flash pyrolysis chars4 were prepared by using a bench-scale continuous fluidized-bed pyrolyzer developed by D. S. Scott and J. Piskorz at the University of Waterl00.~Coal particles of 74-149 pm were fed to the sand fluidized bed a t a rate of 20-50 g/h. Vapor residence time was a few seconds. The chars produced were collected by a cyclone maintained at the same temperature as the fluidized bed and therefore were stabilized a t that temperature. Visiting Associate Professor, also Associate Professor, Research Laboratory of Carbonaceous Resources Conversion Technology, Kyoto University, Kyoto 606, Japan. t Visiting Research Professor, also Senior Researcher, Coal and Carbon Department, National Research Institute for Pollution and Resources, Tsukuba 305, Japan. 0887-0624/90/2504-0024$02.50/0
Flash hydropyrolysis chars were supplied by the Alberta Research Council (ARC). The chars were produced by a bench-scale, entrained-bed reactor with a 2 kg/h feed rate.6 In this unit, particles were heated rapidly (>10000 K/s) to the pyrolysis temperature, and the chars produced were collected in a cold vessel. Coal particles of 74-149 pm were used. The proximate and the ultimate analyses of the chars were made by ARC. Experimental Methods. A Perkin-Elmer thermobalance (model TGS2) was used for the gasification measurements. T o examine the effect of the pyrolysis conditions properly, gasification was performed a t temperatures lower than 550 "C so as not to exceed the lowest pyrolysis temperature (Table 11). Therefore, we were obliged to use air as the gasifying agent. In our version of the temperature-programmed-reaction (TPR) technique about 2 mg of char particles in a 6-mm-i.d. by 2-mm-high pan were heated linearly in air a t ambient pressure from 300 "C, a temperature a t which the reaction rate is practically zero, to 550 "C a t rates of 2,5, and 10 K/min. The weight change was recorded continuously. Char conversion X as a function of temperature T was obtained in this way. The conversion, X, was defined to exclude ash and any volatile matter evolved up to 300 "C. Analysis of Data. The random pore model proposed by Bhatia and Perlmutte3 was used to correlate the data. This model is represented in differential form by d X / d t = koe-E/RT(l- X)[1 - $ In (1 - X)]'/*
(1)
where t is time, ko is the frequency factor incorporating any pressure dependency, E is the activation energy, R is the gas constant, and $ is a parameter related to the pore structure of the unreacted sample ( X = 0) as discussed by Bhatia and Perlmutter. Equation 1 relates the change in conversion to the progress of reaction by taking into account the pore surface area development. This equation incorporates the volume reaction model by setting $ equal to zero, or the grain model, at least approximately, by setting $ to 1. The temperature T and time t are related by T = To+ at, where a is the heating rate and Tois the temperature a t which heating (1) Miura, K.; Silveston,P. L. Analysis of Gas-Solid Reactions by Use of a Temperature-ProgrammedReaction Technique. Energy Fuels 1989, 3, 243. (2) Miura, K.; Makino, M.; Silveston,P. L. Correlation of Gasification
Reactivities with Char Properties and Pyrolysis Conditions Using LowRank Canadian Coals. Submitted for publication in Fuel. (3) Bhatia, S. K.; Perlmutter, D. D. A Random Pore Model for Fluid-Solid Reactions: I. Isothermal Kinetic Control. AIChE J. 1980, 26, 379.
(4) Royce, A. The Influence of Inorganic Matter on the Flash Pyrolysis of Some Canadian Coals. M.A.Sc. Thesis, University of Waterloo, Waterloo, Ontario, Canada, 1986. (5) Scott, D. D.; Piskorz, J. Can. J. Chem. Eng. 1982, 60, 666. (6) Chambers, A. K.; Knill, K. J.; Mendiuk, J.; Ungarian, D. Preprint, 35th Canadian Chemical Engineering Conference.
1990 American Chemical Society
Two-step Gasification of Chars
Energy & Fuels, Vol. 4, No. 1, 1990 25
Table I. Analyses of Coals proximate anal., w t %, dry ultimate anal., wt %, daf basis VM FC ash C H N S
coal Forestburg Estevan Coronach Highvale a By difference.
43.3 44.6 46.1 35.4
50.6 48.7 42.9 52.6
6.1 6.7 11.0 12.0
69.9 67.6 70.9 74.0
3.7 4.3 6.0 4.6
F-650
Estevan
E-550
F-Dem-COz E-H20 E-Dem 3.5 E-Dem 2.5 E-Dem 1.5
Coronach
(2-650 C-Dem
650 650 550 650 650 650 650 650 650 650
steam N2 N2 N2
N2 N2
Table 111. Flash Hydropyrolysis Conditions for Chars Used' frbm Highvale Coal tR," char yield, T,."C ;%I S wt %. daf samde 3.7 6.7 3.7 10.0 6.7 6.7
4.0 5.0 0.5 0.5 2.6 2.7
~
68.2 66.7 64.7 65.8 48.5 47.8
-[
1 2
E-HZO
E-Dem 3.5 E-Dem 2.5
is started. Tocan be set equal to 0, since the initial temperature is chosen to be low enough so that the gasificationrate is negligible. Then eq 1can be integrated with respect to temperature to give' 1 - exp(
52.6 50.5 52.0 50.6 50.8 44.8 50.5
+
koe-E/R.)l)
(2)
We have discussed the use of the Bhatia-Perlmutter Model in an earlier paper.' Equation 2 successfully correlates the weight-loss data subject to the restriction below; however, it is not our intention in this paper or our earlier one1 to test gasification models. The Bhatia-Perlmutter model serves here only as a means of representing the rate of char gasification. X vs T relationships obtained experimentallyat three different heating rates were fitted to this equation by use of a nonlinear least-squares procedure to obtain ko,E, and $. We have shown that at least three data sets obtained at different heating rates are essential to obtain reliable model parameters.'
Two-step Gasifications Figure l a shows the TPR data with single S-shaped curves. The char employed was prepared by flash pyrolysis a t 650 "C in a COz atmosphere from a demineralized Forestburg subbituminous coal. Measurements were made at three heating rates. The solid lines show the excellent representation of this data by the Bhatia-Perlmutter model using a single set of k,, E, and $ parameters given in Table IV. The sample is F-Dem-COP Two steps or stages can be seen in Figure lb,c. With these chars, gasification started at rather low temperatures but appeared to cease a t a conversion of about 85% . As temperature increased, gasification started once again and proceeded to convert most of the remaining carbon. In-
dem at pH = 1.5 dem + Nab
C-Dem+Na H-600-2 H-600-3 H-700-1 H-700-2 H-700-5 H-700-6
3.20 X 8.92 X 2.06 x 9.07 X 9.86 x
lo8"
2.61 x 2.90 x 4.22 X 3.54 x 3.37 x 1.51 x 2.37 x 6.09 x 9.69 X 1.42 X 2.58 x 1.02 x
107 10'0 10' 10'0 10" 1013 109 1013 lo8
lollb
106
10s 105
1.12 x 105
E-Dem 1.5 C-650 C-Dem
1
2
kOe-E/RT(1
dem at pH = 3.5 dem at pH = 2.5 dem at pH = 1.5
Table IV. Calculated Parameters for the Bhatia-Perlmutter Model char ko,s-l E, kJ/mol # 1.21 x 106 122 3.9 F-650 3.79 x 105 132 27 F-Dem-COz E-550
Residence or contact time of solid.
X
dem" at pH = 1.5
N2
C-Dem+Na N2 a Demineralized. * Na was loaded by ion exchange onto demineralized coal.
600 600 700 700 700 700
remarks
57.9 53.3
N2 CO2
0" 24.5 26.9 22.6 20.2
0.2
Table 11. Flash Pyrolysis Conditions for Chars Used' sample pyrolysis char yield, abbrev T,,,O C atm wt %, daf
coal Forestburg
HI600-2 H-600-3 H-700-1 H-700-2 H-700-5 H-700-6
1.9 1.2 0.5 1.0
1 2 1 2
108 107 10'2
142 213 122 135 133 124 132 198 134 161 175 229 148 239 144 146 138 216
0 7.6 10 1.9 29 63 0 0.57 1.6 0 0 0 0 0.02 0 3.6 9.0 0
Value for the first-stage reaction. *Valuefor the second-stage reaction. deed, the residual shown at 550 "C corresponds to the ash content of the char. The chars used were prepared from an Estevan lignite by flash pyrolysis in N, a t 550 "C and from a Highvale subbituminous coal by flash hydropyrolysis at 700 "C under a hydrogen partial pressure of 6.7 MPa. With several chars, the second step is difficult to distinguish. This is the case for a char prepared by flash pyrolysis of a Coronach lignite in N2at 650 "C. The TPR data appear in Figure Id. It is possible to represent the two-step process using the Bhatia-Perlmutter model by assuming that gasification proceeds via parallel reactions. The first of these involves the reactive fraction of the char, while the second applies to the less reactive fraction that requires temperature approaching 550 "C for complete gasification. The model represents the twestep observations quite well as the solid lines in Figure lb,c attest. With the Coronach char, Figure Id, either the one-step or twestep assumption can be used. Parameters for the two-step model for samples E-550 and
Miura et al.
26 Energy & Fuels, Vol. 4, No. 1, 1990 1.o 380°C
c
n
600°C
\
-
E [
I V
0.5
-
lo.. I
I
0
A
EXPERIMENTAL
S VI
50
0
0 350
400
450 500 TEMPERATURE ("C)
550
100 TIME (min)
150
200
Figure 2. Weight loss as a function of time for sample H-700-6 from a Highvale subbituminous coal using a two-stage constant-temperature TGA measurement. (Axis on right shows correction for ash content.)
1 .o
n
I
v
0.5
l ' O r
I
h
2-
0.51 L
J
-
-
IO0 0
TEMPERATURE ("C)
300
T p = 650'C a = 5 K/min
1
I
350
400
450
500
550
TEMPERATURE ( " C )
1 .o
Figure 3. Weight loss as a function of temperature and extent of demineralization for samples E-H20,E-Dem 3.5, E-Dem 2.5,
and E-Dem 1.5 from an Estevan lignite.
h
I V
0.5 I 7
0 300
300
350
350
400 450 TEMPERATURE ("C)
400 450 TEMPERATURE ( " C )
500
500
550
550
Figure 1. Weight loss as a function of temperature in TPR
measurements (a) sample F-Dem-CO, from a demineralized Forestburg subbituminous coal; (b) sample E-550 from Estevan lignite; (c) sample H-700-6 from a Highvale subbituminous coal; (d) sample C-650 from a Coronach lignite. H-700-6 are given in Table IV.
Discussion Both data and modeling suggest that at least two distinct processes are occurring during gasification. All chars contain residual volatile matter (RVM) which is driven off during gasification. In the chars studied, RVM ranged from 3 to 14% by weight for the flash pyrolysis chars and
up to 25% for flash hydropyrolysis ones. Nevertheless, the RVM content is much too low to account for the behavior in Figure lb-d. Could the two-step observation result from the TPR procedure? This was tested by using the Highvale char that gave the TPR data shown in Figure ICby a constant-temperature TGA experiment performed in stages. The first two stages exposed the sample to air at 110 OC for 10 min, then heated rapidly to 380 "C at 100 OC/min, and held the char sample at this temperature for 180 min. Thereafter, the sample was heated at 100 OC/min to 600 "C and held for 10 min. The weight-loss curve vs time is shown in Figure 2. A plateau is attained a t about X = 0.85. On increasing the temperature to 600 "C, gasification starts again consuming the remainder of the carbon until just ash remains. This conversion at the plateau for 380 "C measurement agrees well with the end of the first stage indicated in Figure IC. Repeating the experiment a t 420 OC gave the same result. This experiment indicates that two-stage gasification is not primarily an artifact of the TPR method. Clearly, refractory carbon detected in constant-temperature-gasification runs is itself evidence of the phenomenon. Other investigators have observed the results just des~ribed.'~~ Demineralization influence has at least two explanations. Mineral matter is not uniformly distributed in the original coal, and pyrolysis probably does not increase the distribution in the char that results. The less reactive portion of the coal may have a lower mineral content (or a mineral composition that is less active for gasification) so that higher temperatures are needed for its gasification. An alternate variation on this explanation is that there is a (7) Takeda, S.; Kitano, K.; Kubota, J.; Kawabata, J.; Sato, E.; Chiba, T. J . Fuel Soe. Jpn. 1985, 64, 409. (8) Takarada, T.; Tamai, Y.; Tomita, A. Fuel 1986, 65, 679.
Energy & Fuels, Vol. 4, No. 1, 1990 27
T w o - s t e p Gasification of Chars Table V. Metal Contents of Raw and Acid-Washed Coals (g eauiv/(ka of coal)) sample Ca Mg Na K 0.53 0.02 0.53 0.21 Estevan 0.01 0.12 0.28 E-Dem 3.5 0.40 0.01 0.03 0.20 E-Dem 2.5 0.16 0.01 0.02 0.19 E-Dem 1.5 0.07 0.008 0.09 0.23 Forestburg 0.37 0.02 0.18 0.003 F-Dem 0
~~
separation of mineral matter as gasification proceeds. Eventually, a t high conversions, the catalytic effect of the char ash is heavily reduced or disappears. The much higher activation energies found for the less reactive char supports either of the above explanations. The explanations were tested by using chars made from demineralized Estevan lignites (Figure 3). Demineralization of coal greatly reduces the gasification reactivity of the char. This was observed for all the coals tested. In Figure 3, the X vs T relationship shifts to higher temperatures as the acid strength used to demineralize the coal increases. It is well-known that metals such as Ca, Na, K, and Mg bound to carboxyl groups accelerate the gasification reaction. The metal contents for Estevan and Forestburg coal chars are given in Table V. Decreases of Ca, Mg, and Na contents with the increase of acid strength explains the difference in the reactivity between the raw coal chars and the acid-washed coal chars. Demineralization eliminates the two-step gasification, but the extent of demineralization does not seem important (ash removal is varied by washing with solutions of different pH as Table V demonstrates). A gradual disappearance of the two-step behavior should have been observed if the mineral matter distribution explanation is correct. Furthermore, Table IV shows that ko and E for the demineralized coals are much lower than those obtained for the less reactive fraction of the parallel reaction model. This suggests that the demineralized char is a different material than the unreacted fraction. An alternate explanation for the two-stage effect is that mineral matter catalyzes volatiles cracking during pyrolysis. Cations such as Ca and Na in coals are said to increase the char yield on pyrolysis? This is because volatile matter produced during pyrolysis is polymerized to form coke in the presence of such metals. The coke formed in this way is expected to be less reactive, because the ordering of graphite-like structure starts to develop.' I t was also reported that the gasification of coal chars supporting Ni proceeds in two stages.* Thus, Ni also seems (9) Franklin, H.D.;Cosway,R. G.; Peters, W. A.; Howard, J. B. Ind. Eng. Chem. Process Des. Deu. 1983, 22, 39.
"
Tp = 650°C O L a = 5 K/min
~
n
y\c,
\\
C-Dem
v
0.5
I
0
300
350
400 450 TEMPERATURE ("C)
500
550
F i g u r e 4. Weight loss as a function of temperature and coal pretreatment for samples C-650, C-Dem, and C-Dem+Na from Coronach lignite.
to act as the catalyst for both gasification and coke formation. Figure 4 shows an examination of this explanation. Char prepared from demineralized coal (C-Dem) begins to gasify a t about 450 "C, roughly the temperature where the less reactive fractions also starts to react. However, removing mineral matter should stop the secondary volatile cracking process, so this observation would support an earlier explanation and not the catalyzed cracking hypothesis. Although the C-Dem char does not exhibit the two-stage behavior as the hypothesis predicts, the char prepared from a coal that was demineralized and then ion exchanged to add sodium exhibited a reactivity close to the char from the untreated coal and also failed to show two-stage behavior. This does not agree with the catalytic char formation hypothesis. A further explanation connecting two-stage gasification with the maceral composition of the original coal is unsatisfactory because it does not predict the effect of demineralization on the behavior.
Conclusion Although two-stage gasification in a TPR measurement is clearly associated with heterogeneity in the char as demonstrated by the strong influence of mineral matter, is has not been possible to determine the source of the heterogeneity on the basis of the experiments falling within the scope of our study. Acknowledgment. The research discussed in this paper was performed as part of the Canada-Japan Joint Academic Research Program and was supported by a Strategic Grant to P.L.S. from the Natural Sciences and Engineering Research Council. K.M. and M.M. acknowledge paid leaves from their home institutions to participate in the research program. The supply of char is also gratefully acknowledged.