Ind. Eng. Chem. Res. 1992,31, 2051-2056
2061
Kinetic Modeling of Direct Liquefaction of Wyodak Coal Catalyzed by Sulfated Iron Oxides The kinetics of the direct liquefaction of Wyodak coal using tetralin and 1000 psig (cold) H2were experimentally measured and mathematically modeled. Experiments were carried out in a 27-cm3 horizontally shaken autoclave microreactor a t 648-698 K for reaction times of 5-120 min. Some experiments were carried out a t lower temperatures to determine if the model would be accurate a t such conditions. Three catalyst combinations were used, Mo/Fe203/S04,Fe203/S04,and Fe203, each with elemental sulfur. Four reaction pathways were considered coal to asphaltenes, coal to gas, coal to oils, and asphaltenes to oils. In each pathway (except for the conversion of coal to gas), parallel thermal and catalytic reactions were included. The conversion of coal to asphaltenes was found to be primarily thermal and conversion of asphaltenes to oils was primarily catalytic, while conversion of coal to oils had significant thermal and catalytic contributions. The frequency factor and activation energy of each significant rate constant were estimated by correlating data for all temperatures (16 parameters). The estimated standard deviation of measured product fractions was 7.6%. Catalytic contributions were most significant a t the highest temperature (698 K). The sulfated iron oxides were significantly more active than the unsulfated oxide. The model successfully predicted conversion during nonisothermal heatup in a stirred autoclave and predicted conversions at temperatures other than the temperatures used to determine the model parameters. Introduction The production of liquid transportation fuels by direct liquefaction of coal is an active research area. While different direct conversion methods exist, most use moderately high temperatures, a solvent, and high pressure hydrogen. It is generally accepted that reactions include cleavage and ring opening of large molecules with addition of hydrogen to stabilize the products as well as retrogressive reactions in which high molecular weight hydrogen-deficient moieties are formed. Reactions involve both noncatalytic and catalytic pathways, and achieving a balance in which desired liquid products are obtained is an important goal in direct coal liquefaction studies (Whitehurst et al., 1980). The present work was undertaken to quantitatively m s s the effect of novel catalysts which are being developed in our laboratory. The catalysts were finely divided iron oxide and two highly acidic sulfated iron oxides, each with elemental sulfur. The sulfated oxides are environmentally benign, relatively inexpensive, and disposable if used in small quantities. We have previously reported that such catalysts are very active for coal liquefaction at low concentrations of metal (Pradhan et al., 1991a,b). A model for the reaction of a complex material such as coal must use lumped components, and the selection of a suitable set is importanL The lumping criteria are dictated by analytical capability and by product value. Oils are usually defiied as those components soluble in a solvent such as pentane or hexane and can be further subdivided by boiling range. Asphaltenes are those materials soluble in a stronger solvent such as methylene chloride or tetrahydrofuran and not soluble in pentane or hexane. Gases, water, and ash can also be measured. We have chosen to characterize liquefaction producta by amounts of oils, asphaltenes, and gases. This is a simple description and has been used by many investigators (Angelova et al., 1989; Cronauer et al., 1978; Curran et al., 1967; Furlong et al., 1982),although more extensive product descriptions have also been used (Singh et al., 1982; Singh and Carr, 1987). The lumped components are then treated as pseudocomponents in the reaction modeL A good summary of relative advantages and disadvantages of lumping is given by Szladow and Given (1981,1982). A valid model also provides a sound basis for scaleup to larger units in which reaction conditions may be significantly different than those used in laboratory autoclaves (Abichandani et al.,
Table I. Prowrties of Iron Oxide Catalysts surface av particle catalvst area, m2/a size, mm Fe203 28.1 50 Fe20dSO4 81.7 20 Mo/Fe203/S04 81.5 20
wt %S
wt % Mo
0.0 3.4 3.1
0.0 0.0 0.5
1982; Shalabi et al., 1979; Shah et al., 1981; Prasad et al., 1986). Kinetic models that describe the rate of conversion of coal to products require that reaction pathways be identified. Weller et al. (1951) proposed that the conversion of coal to oil proceeds through the formation of asphaltenes, and this route continues to be accepted today (Keogh et al., 1991). However, direct conversion of coal to oil has also been reported by Radomyski and Szczygiel(1984)and Suzuki et al. (1987). We have analyzed our experimental data using a lumped parameter model with both series and parallel conversion of coal to oils. We have also incorporated parallel catalytic pathways for each step. Experimental Section Wyodak-Anderson subbituminous coal obtained from the Argonne Premium Coal Sample Bank was used. Two shea of coal, -20 and -100 mesh, were available. Both were used since we found that initial coal particle size did not have a significant effect on the final product distribution at the reaction temperatures employed in our study. Negligible effects of particle size were also reported by Guin et al. (1976). The coal was used as received without drying and had the following composition (on an maf basis): C, 75.0%; H, 5.4%; 0, 18.0%; s,0.5%; N, 1.1%.We chose Wyodak coal for our kinetic studies because of its low pyritic sulfur content (0.17%). Pyrite is a known catalyst for coal liquefaction and would obscure the effects of the catalyst being studied. Tetralin (99+ % ) and elemental sulfur (99% resublimed) were obtained from the Fisher Scientific Co. The solvents used in product fractionation, methylene chloride (99.9%) and n-pentane (99.9+%), were obtained from the Aldrich Chemical Co. Catalysts Employed. Three iron and iron-molybdenum catalysts which have a very f i e particulate size were employed in low concentrations (50ppm Mo and 3500 ppm Fe, both with respect to coal) in the presence of elemental sulfur, added in excess of that needed to form pyrrhotite. Properties of these iron-based catalysts are listed in Table
0888-5885/92/2631-2051$03.O0/0 0 1992 American Chemical Society
2052 Ind. Eng. Chem. Res., Vol. 31, NO. 8,1992
I. A detailed description of the synthesis and the characterization of the sulfated iron oxides is given elsewhere (Pradhan et al., 1991a). Experimental Setup. The coal liquefaction experiments were carried out in 27-cm3tubing bomb microauh l a v e s shaken horizontally at about 160 cyclea/min. The microreador consists of a horizontal reactor tube (1-in.0.d. X 4.625 in.), a vertical reactor stem (1/2-in.0.d. X 10 in.), and a multiport valve connected on top of the reador stem. The horizontal tube is sealed at one end, and the other end is fitted with a cap to allow charging and discharging of the reactants and products. The reador stem isolates the multiport valve from the heat source (a hot fluidized sand bath) and accommodates a pressure transducer, a thermocouple, and a quick connect to allow monitoring of the reaction variables and charging and discharging of gases. The reactor is shaken with a horizontal wrist-action motion. Reaction Procedure. In a typical experiment, 3 g of Wyodak (-20 or -100 meah) are mixed with 12 g of tetralin. Predetermined amounts of sulfur (0.02 g) and iron catalyst (0.01 g), preheated in air at 673 K for 1 h), are added, manually mixed, and transferred to a predried tubing bomb microreador. Amounts are accurate to 1mg. After being purged and pressure-tested, the microreactor is charged with 6.9 MPa of hydrogen. The reactor is then shaken at room temperature at 100 cycles/min to allow for prereaction feed mixing and hydrogen dissolution in the slurry. At time t = 0, the reactor is immersed in the sand bath which is at reaction temperature. Typically the reactor heats to the reaction temperature in 3 min while being shaken at approximately 160 cycles/min. After the predetermined reaction time, the reactor is taken out of the sand bath and quenched to room temperature in approximately 5 min using air. The reactor g a m are sampled and analyzed using a Varian 3300 gas chromatograph. The remaining gases are vented, and the reactor contents are washed with methylene chloride. The product gases consisted of hydrocarbons, C1-C4, C02, CO, H2S, and H20. The total conversion of coal and the conversions to product fractions were defined on an ash-free basis as follows: fractional coal conversion YT = (1- Y) unreacted coal fraction Y = ( Wi - Wc - w&)/ W, asphaltenes fraction Y A = (WA/ Wm,) gas fraction Yc = YT - YM, where YM = WM/ Wmaf oil fraction
YO = YM - YA
where Wi, Wc, Wash, Wmd, WA,and WM are masses of methylene chloride insoluble products, catalyst, ash, moisture- and ash-free coal, pentane insolubles, and methylene chloride solubles. Runs were made for the three catalysts at five times (5, 15,30,60, and 120 min) and three temperatures (648,673, and 698 K)for a total of 45 experiments. An additional five runs were made a t five times at 673 K using half the amount of catalyst 11.
Results and Discussion Mathematical Model. The kinetic model used in our work is shown below and is based on both series and parallel conversion of coal into oils. It is similar to one employed by Radomyski and Szczygiel(1984) who found that a series-parallel model was superior to models based on either only-series or only-parallel pathways. Suzuki et
-
al. (1987) have reported that, in a catalytic process, oils are formed both directly via asphaltenes and by decomposition of the organic fraction of coal.
gases
- coal
Kcc
asphaltenes
KCA
-
oils
KAO
The model includes the following assumptions. (1) The products of coal liquefaction are unreaded coal, asphaltenes, oils, and gases. Unreaded coal includes char and is defined as methylene chloride insolubles. Asphaltenes are methylene chloride soluble and pentane insoluble. Oils are pentane solubles. Material unaccounted for is assumed to be gas and includes any losses. (2) All reactions are irreversible with a pseudo-fmt-order reaction rate constant
Ki = Aiexp(-Ei/RT)
(1)
(3) Thermal and catalytic pathways are allowed for coal to oils, asphaltenes to oils, and coal to asphaltenes. Only statistically significant reaction pathways were included in the final model. (4) The rate constant for catalytic conversion is proportional to the amount of catalyst. (5) Effects of particle size and mixing intensity are neglected. In separate experiments, conversions were found to be independent of the coal particle size and the mixing intensity. (6) The effect of hydrogen pressure and solvent are not included since hydrogen pressure was essentially constant and hydrogen (either from the gas phase or from the solvent) was present in excess. (7) Not all the carbon in the coal can be converted to liquids regardless of the reaction times. The organic portion of coal contains some inert macerals such as fusinite and semifusinite, and the ultimate conversion is less than 100% (4. We assume that the ultimate conversion (C*)depends only on the reaction temperature for a particular coal and solvent combination. We carried out thermal, noncatalytic runs at three different temperatures (648, 673, and 698 K) for 3 h to determine the ultimate conversion. A reaction time of 3 h was chosen since it was significantly longer than the reaction times employed for the catalytic runs used in the kinetic analysis. Total coal conversions of 76%, 87%, and 93% were obtained (all on an maf basis) at 648,673, and 698 K, respectively. The rate constant for the conversion of asphaltenes to oils (KAO)was assumed to have parallel thermal and catalytic reaction pathways:
KAO= KAOT+ ~ J A + ~O Z K AC O+ CI I~I ~ K A O C I I I(2) where KAoT,K+I, K*OCII, and KAocII are the rate constants for conversion of asphaltenes to oil by the thermal pathway, by catalytic pathway I (Fez03 + S),by catalytic pathway 11(Fe203/S04+ S),and by catalytic pathway 111 (Mo/Fe203/S04+ S),respectively, and ml, q, and m3are the masses of catalysts 1-111, respectively. Similar equations were used to represent the conversion of coal to oil and coal to asphaltenes. Since only one catalyst was used in any given experiment, two of the masses in eq 2 are zero for each run. Determination of Kinetic Parameters. A total of 50 experimental runs were made and included in the correlation. For each run, three measurements were made: fraction of coal remaining, fraction converted to asphaltenes, and fraction converted to oils. The fraction convertad
Ind. Eng. Chem. Res., Vol. 31, No. 8,1992 2063 Table 11. Rate Constants and Asymptotic Standard Deviations for All Reaction Pathways Included in the Final Model rate 673 K 648 K 698 K constant," U estimate U min-' estimate estimate U A, min-' E , cal/mol 0.0034 0.007 (0.010) 0.0029 KCG 0.0084 (0.ooSs) 0.012 (0.013) 1.6 X l@ 0.0038 1.3 x 104 0.0031 0.034 (0.031) 0.0036 1.1 x 10s 0.036 (0.045) 0.0042 1.4 X 10' KCAT 0.017 (0.020) 0.0025 0.016 (0.021) 0.0042 KCOT 0.021 (0.019) 0.34 0.030 (0.024) 0.0071 3.7 x 103 0.008 0.011 (0.019) 0.0086 1.9 0.020 (0.019) 0.011 m&wII 0.029 (0.019) 0 0.008 0.025 (0.023) 0.01 0.047 (0.034) 1.7 X lo6 0.015 m&cwm 0.023 (0.015) 1.5 x 104 0.004 (0.004) 0.004 0.0021 0.0025 (0.0045) 0.0025 5.6 3.5 x 109 mlKAocI 0.005 (0.004) 0.005 0.011 (0.011) 0.0095 (0.005) 5.8 X log 3.1 X 10' 0.0047 m&AWII 0.00 (0.002) 0.0099 (0.0059) 0.01 0.017 0.009 (0.009) 94 x 103 1.6 X 10' 0.0047 m&AwnI 0.004 (0.004) a Values in the parentheses are the rate constanta calculated using frequency factors and activation energies from the last two columns.
-
to gas was obtained by difference and not used in the correlation. The following equation was minimized
f =N c[wC(yC.i -
?C,i)2
i=l
+ wA(yA,i -
+ W O ( y 0 , i - ?O,i)'] (3)
The wi in eq 3 are weighting factors and were used to compensate for the fact that the accuracy of the measurements for the three products was not equal. Values of 1,0.9, and 0.8 were used for coal, asphaltenes, and oils, respectively, and were selected based on the reproducibility in duplicate runs (f2.0% for total coal conversion, f2.290 for asphaltenes, and f2.5% for oils). Since the four reactions in our kinetic model are assumed first order, the differential equations describing the change of composition with time can be readily solved to give ?c(t) = (1 - c*) + c * eXp[-(KcA Kco +KAo)] (4)
nr
(&A
(6)
+ Kco + KCC)(KAO - KCA- Kco - KCC) ?G(t) = 1 - [p7'(t)+ YA(t) + ?O(t)] (7)
If values for the rate constants are known,eqs 2 and 4-6 can be used to calculate the predicted values for product fractions. These and the experimentally measured values are then used in eq 3 to determine f. The desired set of rate constants is that which gives the lowest value for f and can be found by a systematic search using one of the standard statistical routines which are available. We used a derivative free nonlinear regression program AR from the BMDP library (BMDP, 1990). Tests for significance were made using an F test for the ratio of variance explained by the regression to estimated error variance. This test is valid for linear regression, and when used for nonlinear regression, as in the present example, must be considered as an approximation. An F value equivalent to a 95% likelihood that the result was significant was used to distinguish between alternative regression equations. A maximum of 13 rate constants could be included in the regression equation. There are four thermal rate constants, KAm, Kcm, KCG, and KCAT; three catalytic rate constants for catalyst I, KAmI, KcmI, and KCACI; three for catalyst II,KAmn, KCWE,and KcAcn; and three for catalyst
111, KAOCIII, KcocIII, and Kcpc111. In order to determine which catalytic effects were important, the data were examined separately at each of the three temperatures used in the experiments. A regression was first made using only the four thermal rate constants. Next, the coal to oil catalytic constants were added as a group, and the confidence level for addition of these three coefficients varied from 95% for the lowest temperature to >99.9% at the highest temperature. It is clear that addition of catalytic pathways for coal to oil significantly improves the regression at all temperatures. Three more rate constants were then added for catalytic conversion of asphaltenes to oil. There is a significant improvement only at the highest temperature. If the order of adding catalytic coefficients is reversed-that is, asphaltenes to oils added first-the results are almost identical. The addition of a catalytic step for coal to asphaltenes did not give a signifcant improvement over the use of the thermal coefficients alone. In the procedure outlined above, the catalytic rates were added in groups. However, it is not likely that all 10 rates are equally significant, and the next step was to determine if selected rate constants could be removed without significantly reducing the fit. The rate with the largest coefficient of variation (ratio of asymptotic standard deviation to estimated value) was KAOT, and it was determined that removing it from the regression did not significantly reduce the fit at any temperature. KcmI had the next largest coefficient and also could be removed. The third parameter to be examined, K A ~ was I , significant at the 95% level for 673 K and was retained. The values of the eight rate constants at each of the three temperatures are given in Table 11together with the estimated standard deviation of each. Three reaction rates are given in Table I1 for each rate constant-one for each temperature studied. In order to extend the correlation to other temperatures, the rate constants were put in the form of eq 1. This could be done by fitting the three values in Table I1 to eq 1. We chose, instead, to correlate all the data (150 measurements) using 16 parameters (two for each rate constant). This should give a better fit to the experimental data. The results are shown in the last two columns of Table 11. It should be emphasized that the frequency factor and activation energy are positively correlated at a very high level, and almost as good a fit can be obtained by increasing a frequency factor and the corresponding activation energy in the correct proportion. Estimated standard deviations are not presented in Table I1 for the frequency factor and activation energy because the individual values should be treated with caution, although the calculated rates should be more reliable. The rate constants obtained by using the frequency factor and activation energy are shown in Table II for comparison. The estimated standard deviation for calculated product fractions is 0.076.
Ind. Eng. Chem. Res., Vol. 31, No. 8, 1992 1.ooy
o.goi
1
0.80
0.70 0.80
-.-
~
- - - - - - -1
-
1
0.0oA 0
~
10
:
20
;
30
:
40
:
50
:
80
:
70
:
80
:
I
:
./
,
0.10 90 100 110 120
-,,
0 . 0 0 l : 0 10
:
20
: 30
:
40
:
60
:
80
:
70
: 80
:
:
~
I
90 100 110 120
Reaction Time in minutea
Reaction Time in minutea
Figure 1. Plots of observed vs predicted product fractions for the Fe203+ S catalytic system at 673 K. Observed product fractions are unreacted coal (a), asphaltenes (A), and oils (W; and predicted product fractions are unreacted coal (-), asphaltenes (-4,and oils (-*-).
1
1.OOV
_
_
-
-
----__
-
-c----
-
-
-
I
*
l
L
Reaction Time in minutes
Figure 2. Plots of observed vs predicted product fractions for the Fen03/S04+ S catalytic system at 673 K. Observed product fractions are unreacted coal (a),asphaltenes (A), and oils (m);and predicted product fractions are unreacted coal (-), asphaltenes (--), and oils (- -).
-
Effect of Temperature and Type of Catalyst Employed. One of the main objectives of this study was to obtain quantitative measures of the effectiveness of the three catalytic systems at 648, 673, and 698 K. As discussed above, the results indicate that the catalyst affects the conversion of asphaltenes to oils and the conversion of coal to oils without an asphaltene intermediate. Thermally, oils are produced directly from coal, but not from asphaltenes. The catalytic reactions are more significant at higher temperatures, and the uncertainty in the catalytic rate constants at 648 K (Table 11) reflects the lack of catalytic activity at this temperature. At higher temperatures, the sulfated catalysts were much more active than iron oxide plus sulfur, and the addition of small amounts of molybdenum to the sulfated catalyst doubled the rate at which coal wm converted to oil. Figures 1-3 compare predicted and experimental results at 673 K for all three catalytic systems. The predominantly catalytic nature of the conversion of intermediate asphaltenes to lighter oils (n-pentane solubles) supports the studies reported by Suzuki et al. (1987) for the direct liquefaction of low-rank and bituminous coals using soluble organometallic complexes as catalysts. The model also agrees with Han and Wen (1979) who proposed that the reactions taking place during the initial stages (fmt 5 min) are predominantly thermal and produce substantial amounts of reactive preasphaltic fragments which are mainly stabilized by hydrogen present within the coal and hydrogen from the solvent. In this reaction scheme, the initial coal reactions do not involve
Figure 3. Plots of observed vs predicted product fractions for the Mo/Fe203/S04+ S catalytic system at 673 K. Observed product fractions are unreacted coal (01,asphaltenes (A), and oils (m); and predicted product fractions are unreacted coal (-), asphaltenea (--), and oils (-*-).
any appreciable molecular hydrogen consumption. It is only at longer reaction times that the catalytic conversion of asphaltols and asphaltenes to lighter oils by reactions such as hydrogenolysis begins and involves appreciable amounts of molecular hydrogen participation. Singh and Carr (1987) have also suggested that the conversion of intermediate reactive organic matter from coal (SRC) to lighter distillates is mainly rate controlling and catalytic in nature. An interesting result of this study is that the kinetic data and statistical analysis indicate that there are both thermal and catalytic pathways for the production of oils from coal. The production of oil from coal through noncatalytic reactions supports the argument for the existence of a "mobile" phase in the macromolecular network of coal (Schindler, 1989). The cross-links in coal are believed to contain aUryl and ether linkages; it is also known that some loosely held and easily broken alkyl side chains are attached to the polyaromatic structures in coal. Substantial amounta of light paraffiiic (n-pentane soluble) compounds are known to be formed during the first few minutes in short contact-time coal liquefaction at temperatures in excess of 673 K. Our kinetic studies support these findings but also indicate that this route can be catalytically enhanced for sulfated catalysta,especially when molybdenum is present. While the enhanced efficacy of the sulfated catalysts appears to be related to the smaller catalyst crystal sizes that sulfating produces, it is not clear if the catalytic conversion is related to greater coal/catalyst contact or to some unique catalytic features of the sulfated forms. The sulfated catalysts appear to allow the catalyst to influence the reaction at earlier stages than does the unsdfated forms. Effect of Catalyst Loading. In order to assess the dependence of the catalytic rate constants on the amount of the catalyst, a set of five experiments was carried out at reduced catalyst concentrations (0.005 g versus 0.01 g used for most experiments) and included in the data set used for regression. The results are shown in Figure 4, where experimental and predicted values are plotted. The most obvious effects of decreased catalyst loading are a decrease in oil yields and increase in asphaltene yields. These trends are well predicted by the model (within f4%), and the inclusion of mass catalyst as a factor in the rate equation is justified. Model Applications. We have used our 16-parameter model to extrapolate the product distributions from the direct liquefaction of Wyodak subbituminous coal at lower reaction temperatures (573 and 623 K) and found that observed and predicted product fractions agree within
Ind. Eng. Chem. Res., Vol. 31, No. 8, 1992 2058 Table 111. Experimental and Predicted Data coal catalyst ~~
I I I I I I I I I I I I I I I I1 I1 I1 I1 I1 I1 I1 I1 I1 I1 I1 I1 I1 I1 I1 I11 I11 I11 I11 I11 I11 I11 I11 I11 I11 I11 I11 I11 I11 I11 I1 I1 I1 I1 I1
temp, K 648 648 648 648 648 673 673 673 673 673 698 698 698 698 698 648 648 648 648 648 673 673 673 673 673 698 698 698 698 698 648 648 648 648 648 673 673 673 673 673 698 698 698 698 698 673 673 673 673 673
amount of catalyst, g 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.005 0.005 0.005 0.005 0.005
time, min 5 15 30 60 120 5 15 30 60 120 5 15 30 60 120 5 15 30 60 120 5 15 30 60 120 5 15 30 60 120 5 15 30 60 120 5 15 30 60 120 5 15 30 60 120 5 15 30 60 120
1 39%. Thus,this model is useful in extrapolating product distributions at conditions other than those used to obtain parameters. The model was also found to successfully account for conversions during the 40-min heat-up period in well-stirred batch autoclaves (300cm9. Using an experimental temperature-time profile, the predicted conversion was 56% with 34% oils. The experimentally measured conversion was 54% with 33% oils. Finally this model has also been successful in predicting fractional yields of oils as a function of catalyst loadings. Thus, the ability to explore the effects of changing reaction conditions is a useful feature of our kinetic model. Conclusions The lumped parameter model we have proposed is relatively simple but still capable of distinguishing between catalytic and thermal steps. A statistical analysis of all experimental measurements has allowed significant pathways to be identified. The asphaltenea to oils pathway was found to be exclusively catalytic in nature containing no significant thermal component. This step is strongly in-
expt 0.77 0.59 0.54 0.42 0.35 0.67 0.48 0.37 0.20 0.13 0.62 0.38 0.30 0.16 0.09 0.67 0.47 0.40 0.36 0.32 0.53 0.41 0.34 0.18 0.13 0.50 0.34 0.23 0.09 0.07 0.65 0.51 0.41 0.32 0.24 0.57 0.37 0.27 0.20 0.13 0.45 0.24 0.15 0.12 0.08 0.69 0.52 0.42 0.32 0.18
pred 0.85 0.63 0.43 0.29 0.24 0.77 0.48 0.27 0.15 0.13 0.69 0.35 0.15 0.08 0.07 0.79 0.53 0.35 0.26 0.24 0.72 0.40 0.21 0.14 0.13 0.64 0.28 0.12 0.07 0.07 0.79 0.53 0.35 0.26 0.24 0.68 0.36 0.19 0.13 0.13 0.57 0.21 0.09 0.07 0.07 0.74 0.44 0.24 0.14 0.13
asphaltene expt pred 0.10 0.07 0.17 0.16 0.10 0.23 0.15 0.27 0.20 0.24 0.22 0.11 0.32 0.25 0.37 0.34 0.43 0.35 0.38 0.29 0.17 0.25 0.29 0.34 0.30 0.42 0.41 0.33 0.32 0.39 0.16 0.06 0.14 0.10 0.13 0.19 0.21 0.21 0.14 0.19 0.33 0.11 0.34 0.22 0.32 0.27 0.26 0.25 0.21 0.18 0.25 0.16 0.29 0.28 0.14 0.29 0.17 0.21 0.18 0.09 0.19 0.06 0.14 0.16 0.18 0.16 0.15 0.18 0.15 0.13 0.26 0.10 0.29 0.20 0.28 0.23 0.19 0.19 0.14 0.11 0.26 0.15 0.24 0.24 0.14 0.24 0.14 0.17 0.12 0.08 0.17 0.11 0.26 0.24 0.24 0.31 0.22 0.32 0.32 0.28
oil expt 0.09 0.21 0.32 0.38 0.41 0.08 0.16 0.22 0.32 0.43 0.08 0.27 0.33 0.44 0.46 0.15 0.41 0.45 0.40 0.50 0.12 0.21 0.31 0.52 0.61 0.18 0.32 0.55 0.67 0.68 0.14 0.31 0.41 0.51 0.59 0.15 0.31 0.42 0.57 0.67 0.22 0.43 0.66 0.68 0.71 0.11 0.18 0.30 0.42 0.43
Prd
0.06 0.16 0.25 0.34 0.41 0.08 0.19 0.28 0.36 0.45 0.09 0.20 0.28 0.36 0.46 0.12 0.28 0.39 0.46 0.50
0.14 0.31 0.43 0.51 0.59 0.16 0.34 0.47 0.49 0.71 0.12 0.29 0.40 0.49 0.55 0.18 0.38 0.50
0.59 0.67 0.24 0.46 0.57 0.66 0.75 0.11 0.25 0.35 0.42 0.48
fluenced by the type of catalyst. The direct conversion of coal to oils was also found to have an important catalytic contribution when sulfated catalysts were used and a thermal contribution in all cases. The addition of trace amounts of molybdenum to the sulfated catalyst substantially enhances the rate of direct conversion of coal to oils. Production of asphaltenes from coal was primarily thermal in nature. Using the pre-exponential factors and the corresponding activation energies as parameters, the model can also describe the effect of temperature, reaction time, catalyst loading, and catalyst type. Specifically, it can predict the conversions obtained at “zero reaction time” that are due to the heat-up period in a stirred autoclave, can predict the conversions at temperatures as low as 573 K, and can predict the conversions obtained at very long reaction times. The catalytic rate constanta (for asphaltenes to oils) for the Mo/Fe203/S04catalyst system were much higher than that for the unsulfated iron oxide catalyst. Thus, model suggests that presence of a dispersed catalyst might be of significance for the first stage of a two-stage coal
2056 Ind. Eng. Chem. Res., Vol. 31, No. 8, 1992 T'l
= fractional coal conversion
Yi= fraction of product i predicted by model
----0
. 0 0 m : : 0 10 20 30 40
: : : : 60 60 '70
:
80
- 2'
: : : : 80 100 110 120
Reaction Time in minutes
Figure 4. Plots of observed vs predicted product fractions for the Fe20s/S04 + S catalytic system at 673 K at reduced catalyst concentration (0.005 g). Observed product fractions are unreacted coal (O),asphaltenes (A),and oils (m); and predicted product fractions are unreacted coal (-), asphaltenes (--), and oils (-.-).
liquefaction process. While the model cannot be used for a given system in the absence of experimental data, it can provide a framework for interpreting the resulta of direct liquefaction studies. Acknowledgment We gratefully acknowledge the funding support from the U S . Department of Energy under Award DE-FC2290PC90029. The contributions of the Argonne Premium Coal Sample Bank for providing the coal samples used in this study are also acknowledged. Nomenclature A = frequency factor, min-' C* = ultimate fractional coal conversion E = activation energy, cal-mol-' f = objective function to be minimized K A ~ IKAocII, , K~oCIII= rate constant for conversion of asphaltenes to oils for catalyst I (Fe203 + s),catalyst I1 (Fe203/S04+ S),and catalyst I11 (Mo/Fe203/S04+ S), min-'-(g of catalyst)-' K A ~K ,w ,Km, KCA= rate of reaction for thermal Conversion of asphaltenes to oils, coal to oils, coal to gas, and coal to asphaltenes, min-I KCOCI, KCWII, Kcwn~= rate of reaction for conversion of coal to oils for catalyst I, catalyst 11, and catalyst 111, min-'.(g of catalyst)-' ml,m2, m3 = mass of catalysts 1-111, g of catalyst N = total number of experimental data pointa used in correlation R = gas constant, 1.987 cal.K-'-mol-' t = time, min T = temperature, K WA, W d ,WC, W,, W M ,W,, = mass of n-pentane insoluble products, ash, catalyst, methylene chloride insolubles, methylene chloride solubles, and maf coal, g wA, wc,w o = weighting factor for asphaltenes, coal, and oils yc, YA,yc, YM,YO = observed fraction of unreacted coal, asphaltenes, gas, methylene chloride solubles, and oils
Appendix The experimental data used to develop the correlation presented in the paper and the predicted values obtained from the correlation are listed in Table 111. Registry No. Mo, 7439-98-7;Fe203,1309-37-1;SO?-,1480879-8; s, 7704-34-9.
Literature Cited Abichandani, J. S.; Shah,Y. T.; Cronauer, D. C.; Ruberto, R.0. Fuel 1982, 61, 276. Angelova, G.; Kameneki, D.; Dimova, N. Fuel 1989,68, 1438. EMDP Statistical Software Manual, Vol. 1; University of California Press: Berkeley, CA, 1990, pp 395-425. Cronauer, D. C.; Shah,Y.T.; Ruberto, R. G. Znd. Eng. Chem. Process Des. Dev. 1978, 17, 281. Currau, G. P.; Struck, R.T.; Gorin, E. Znd. Eng. Chem. Process Des. Dev. 1967, 6, 167. Furlong, M. W.; Baldwin, R. M.; Bain, R. L. Fuel 1982,61,61. Guin, J. A.; Tarrer, A. R.;Taylor, L.; Prather, J.; Green, S. Znd. Eng. Chem. Process Des. Dev. 1976,15,490. Han, K. W.; Wen, C. Y. Fuel 1979,58, 779. Keogh, R. A.; Teai, K.; Xu, L.; Davis, B. H. Energy Fuels 1991,5, 625. Pradhan, V. R.;Tierney, J. W.; Huffman, G. P.; Wender, I. Energy Fuels 1991a, 5,497. Pradhan, V. R.;Herrick, D. E.; Tierney, J. W.; Wender, I. Energy Fuels 1991b, 5, 712. Prasad, G. N.; Agnew, J. B.; Sridhar, T. MChE J. 1986,32 (a), 1288. Ftadomyski, B.; Szczygiel, J. Fuel 1984, 63, 744. Schindler, H. D. Coal Liquefaction-A Research and Development Ne& Assessment; DOE/ER-400, National Technical Information Service; Springfield, VA, March 1989; Vol. 2. Shah, Y. T.; Kriehnamurthy, S.; Ruberto, R. G. Reaction Engineering in Coal Liquefaction; Addison-Wesley Publishing Co.: Reading, MA, 1981; pp 162-210. Shalabi, M. A.; Baldwin, R. M.; Bain, R. L.; Gary, J. H.; Golden, J. 0.Znd. Eng. Chem. Process Des. Dev. 1979,18,474. Singh, C. P. P.; Carr, N. L. Znd. Eng. Chem. Res. 1987, 26, 501. Singh, C. P. P.; Shah,Y. T.; Carr, N. L.; Prudich, M. E. Can. J. Chem. Eng. 1982,60,248. Suzuki, T.; Ando, T.; Watanabe, Y. Energy Fuels 1987, I, 341. Szladow,A. J.; Given, P. H. Znd. Eng. Chem. Process Des. Dev. 1981, 20, 27. Szladow, A. J.; Given, P. H. Chem. Eng. Commun. 1982,19, 115. Weller, S.; Pelipetz, M. G.; Friedman, S. Znd. Eng. Chem. 1961,43, 1572. Whitehurst, D. D.; Mitchell, T. 0.;Farcasiu, M. Coal Liquefaction; Academic Press: New York, 1980; pp 39-67.
* To whom correspondence should be addressed. Vivek R. Pradhan, Gerald D. Holder Irving Wender, John W. Tierney* Chemical and Petroleum Engineering Department University of Pittsburgh Pittsburgh, Pennsylvania 15261 Received for review January 15, 1992 Revised manuscript received May 29, 1992 Accepted June 16, 1992