144
Energy & Fuels 1989,3, 144-147
Hydrocracking of Athabasca Coker Feed Bitumen Using Molten Halide Catalysts in a Bubbling Microautoclavet A. Chakma, E. Chornet,* and R. P. Overend Universitd de Sherbrooke, Sherbrooke, Qudbec, Canada J l K 2Rl
W. H. Dawson Energy Research Laboratories, CANMET, Ottawa, Ontario, Canada KIA OG1 Received August 2, 1988. Revised Manuscript Received January 24, 1989
Athabasca coker feed bitumen has been treated with halide catalysts under a continuous flow of H2 in a 15-mL microautoclave. The H2 was bubbled through the liquid by using a microporous steel grid. ZnC12, CuCl, and ZnC12/CuC1mixtures, with and without tetralin, were used as catalysts. The experimental conditions were as follows: 13.8-MPa operating pressure, 1L/min H2 flow rate, 425-450 OC, and 30-min reaction time. ZnC12has been found effective for converting asphaltenes into maltenes while lowering the coke formation with respect to the uncatalyzed reaction. The addition of tetralin to the reaction mixture minimized coke and gas formation.
Introduction Bitumen and heavy oil upgrading has been the subject of many investigations since the early 193Os.l Present industrial technologies such as delayed coking or fluid coking result in the formation of large amounts of coke (up to 22% of the feed).2 This decreases the yield of valuable liquid products. Whereas coke derived from petroleum in general may find a variety of uses, the high sulfur content of the coke produced from Canadian oil sands bitumen, such as Athabasca bitumen, precludes its use.3 Hydrocracking of Athabasca bitumen yields 10-15% higher liquid distillate product than the conventional coking processes. Although coke may also form during hydrocracking, the amount is significantly less than in coking. This, however, results in reactor fouling. When conventional hydrocracking and desulfurization catalysts are used, catalyst poisoning occurs. Coke formation on external and internal surfaces and deposition of poisoning metals are believed to be the main causes of catalyst d e a ~ t i v a t i o n . ~Coke formation in the catalyst pores decreases the pore size and limits mass transfer by decreasing the flow of the reactant. Coke formation increases with t e m ~ e r a t u r eand ~ , ~decreases with hydrogen partial pressure.5 Traditional approaches to deal with coke formation therefore involved operation at relatively lower temperature and higher H, partial pressure. Since pitch (524 "C+ fraction) conversion decreases with temperature, hydrocracking operations cannot be carried out at temperatures low enough to prevent coke formation. Usually temperatures above 440 OC are employed for oil sands bitumen hydrocracking. Furthermore, H2 solubility in Athabasca bitumen is extremely low, about 0.0008 g of H2/g of bitumen at 13.8 MPa of H, partial pressure and 300 OC.' Therefore, to make H2 available for reaction in the liquid phase, high H2 partial pressures (up to 13.8 MPa) are usually employed. Addition of coal and coal-based catalysts has been claimed to reduce coke formation.8 An alternative approach would be to ensure rapid stabilization of the coke-forming free radicals by enhancing 'Presented at the Symposium on Cod-Derived Fuels-Catalytic Upgrading, 195th National Meeting of the American Chemical Society and 3rd Chemical Congress of North America, Toronto, Ontario, Canada, June 5-10,1988.
0887-0624/89/2503-0144$01.50/0
mass-transfer rates. In conventional catalytic hydrocracking operations the overall resistance to mass transfer is usually due to a gas-liquid resistance between H2 and bitumen and a liquid-solid resistance between bitumen and catalyst. The former can be enhanced by employing high-performance mass-transfer devices that allow intimate mixing between the respective phases and provide a high specific interfacial area. The latter can be eliminated altogether by using homogeneous catalysts, which would also circumvent the problems associated with catalyst poisoning. This may also allow operation under less severe operating conditions due to better temperature control on the catalytic sites. Molten halide salts are excellent catalysts for hydrocracking polynucleargJOand alkyl-substituted polynuclear1' aromatic hydrocarbons. Zielke et aL9have shown the superiority of molten ZnC12 catalysts over conventional hydrocracking catalysts for pyrene, coal, and coal extract and subsequently carried out continuous hydroliquefaction of subbituminous coal in molten ZnC12.12 The catalytic action proceeds via an ionic mechanism capable of inducing cracking of polyaromatic compounds but unable, under the severity conditions used in coal liquefaction, to open the ring of monoaromatic compound^.'^ (1)Pruden, B. B. Can. J . Chem. Eng. 1978,56,277-280. (2)Nomura, M.; Terao, K.; Kikkawa, S. Fuel 1981,60,699-702. (3)Speight, J. G.; Moschopedis, S. E. Fuel Process. Technol. 1979,2, 295-302. (4)Hardin, A. H.; Ternan, M.; Packwood, R. H. CANMET Report 81-4E;Energy Mines Resources Canada: Ottawa, 1981. (5)Pruden, B.B.;hgie, R. B.; Denis, J. M.; Merrill, W. H. CANMET Report 76-33;Energy Mines Resources Canada: Ottawa, 1976. (6)Khulbe, C. P.; Pruden, B. B.; Denis, J. M. Laboratory Report ERP/ERL 77-79(J);Energy Mines Resources Canada: Ottawa, 1977. (7) Lal, D.; Mather, A. E.; Otto, I. D. In Chemical Engineering Thermodynamics; Newman, S. A., Ed.; Ann Arbor Science: Ann Arbor, MI, 1983;pp 75-84. (8)Ranganathan, R.; Pruden, B. B.; Denis, J. M. CANMET Report 77-62;Energy Mines Resources Canada: Ottawa, 1977. Struck, R. T.; Evans, J. M.; Costanza, C. P.; Gorin, (9)Zielke, C. W.; E. Ind. Eng. Chem. Process Des. Deu. 1966,5,158-164. (IO) Nakatsuji, Y.;Kubo, T.; Nomura, M.; Kikkawa, S. Bull. Chem. SOC.Jpn. 1978,51,618-624. (11)Nakatsuji, Y.;Ikkaku, Y.; Nomura, M.; Kikkawa, S. Bull. Chem. SOC.Jpn. 1978,51,3631-3634. Klunder, E. B.; Maskew, J. T.; Struck, R. T. Ind. (12)Zielke, C. W.; Eng. Chem. Process Des. Deu. 1980,19, 85-91.
0 1989 American Chemical Society
Energy & Fuels, Vol. 3, No. 2, 1989 145
Hydrocracking of Athabasca Bitumen
1
Table I. Selected Properties of Athabasca Bitumen 10.1 density API gravity ash viscosity (25 "C) 42000 CP 83.77 wt % hydrogen carbon 0.37 wt % sulfur nitrogen 0.88 w t % vanadium oxygen 75.5 ppm by w t asphaltene nickel (pentane) pitch (524 "C + 53.7 wt % coke fraction) maltene 83.33 w t %
Coker Feed 0.999 kg/L 0.48 w t % 10.51 wt % 4.75 wt % 200 ppm by wt 16.1 wt %
7 G.C.
1
&
n-rWA CONDENSER
OUT
MICRO-
0.57 wt %
Coke formation from oil sands bitumen is believed to be related to its high asphaltene content.2 Dickie and Yen14have studied structures of asphaltenes from various sources and postulated that asphaltene aggregates are made of planar sheets of condensed aromatic rings, saturated carbon chains, and a loose net of naphthenic rings. Therefore, polynuclear aromatic hydrocarbons may be considered as possible constituents of asphaltene, and catalysts that are effective in hydrocracking polynuclear aromatic compounds should also be effective in asphaltene hydrocracking. Given the above situation, the choice of ZnC12 as our prime homogeneous catalyst corresponds to the need of investigating the activity of this well-known acid catalyst to hydrocrack the polyaromatic and heteropolar structures characteristic of heavy oil. Since the latter is not constituted by particles (as coal is) but rather has a micellar-like structure, it was felt that ZnC1, would readily dissolve under reaction conditions, thus impregnating the micellar microstructures. The extent of solubilization of ZnC1, by the lighter fractions of the maltenes is unknown but our reasoning was that it would be simpler to penetrate and impregnate the micelles (heavy oil case) than particles (coal case). Herrmann et al.15 have carried out a comparative study of Fe, ZnC12, and ZnC1,-promoted Fe catalysts for the hydrocracking of Athabasca bitumen. They found ZnC1, to be the most active and to produce less hydrocarbon gas and significantly less sulfur in the liquid product. Nomura et aL2studied hydrocracking of Athabasca asphaltene over molten ZnC12-KC1-NaC1 mixtures with or without an additive of some transition-metal chlorides and found a mixture of ZnCl,-KC1-NaC1-MoC15 to give the highest asphaltene conversion (60% pentane soluble) and the smallest amount of coke (3.7% benzene insoluble). Since ZnC12 is highly corrosive and has a relatively high melting point and since a decrease in acid strength (by KC1 addition) of both ZnC12 and SnC12 has resulted in higher coal conversion and an improved selectivity toward hexane-soluble products,13we considered CuCl as an alternate choice since its acidity is lower than that of ZnC1, and its low melting point and better corrosion properties could, in principle, make it more suitable than ZnC12 itself for heavy oil cracking. Work on CuCl mixed with ZnC1, has been reported by Nomura et al.16717 On the basis of this information, we decided to carry out systematic experiments to investigate the hydrocracking of Athabasca bitumen with molten ZnC12, molten CuCl, (13) Derbyshire, F. J. Catalysis in Coal Liquefaction: New Directions for Research; IEACR/O8; International Energy Agency Coal Research London. 1988. (14) Dickie, J. P.; Yen, T. F. Anal. Chem. 1967, 39, 1847-1852. (15) Herrmann, W. A. 0.; Mysak, L. P.; Belinko, K. CANMET Report 77-50; Energy Mines Resources Canada: Ottawa, 1977. (16) Nomura, M.; Mikaye, M.; Sakpshita, H.; Kikkawa, S. Fuel 1982, 61, 18-20. (17) Nomura, M.; Sakashita, H.; Mikaye, M.; Kikkawa, J. Fuel 1983, 62, 73-77.
TO
GAS
COLLECTION TANK
MOLTEN SALT
F i g u r e 1. Schematic diagram of the bubbling microautoclave system.
and their mixtures. Tetralin, a good hydrogen donor compound, was added to some of the reactions.
Experimental Section Feed Stock a n d Reagents. Athabasca coker feed bitumen (ABCF), produced by Suncor Inc. using hot water extraction of mined oil sands of the McMurray stratigraphic unit, was obtained from the Alberta Research Council. Table I #summarizes the relevant properties of the bitumen. All reagents used were ACS reagent grade. Tetralin and ZnC12 were purchased from Aldrich Chemicals, Milwaukee, WI, while CuCl was purchased from Anachemia Ltd., Montreal, Canada. ZnClz and CuCl were oven-dried at 100 "C for 3 h to eliminate moisture. Equipment. Traditionally, laboratory-scale hydrocracking experiments are usually carried out in stirred autoclave^'^ and rocking autoclaves.2 However, the fluidodynamic conditions in these traditional autoclaves are considerably different from the continuous-flow conditions of industrial hydrocrackers. T o more closely simulate industrial conditions, we used a bubbling microautoclave system of our own design. A schematic diagram of the experimental setup is shown in Figure 1. It consists of a 15-mL bubbling microautoclave made of 316 stainless steel, a water-cooled condenser, a gas-collection assembly, a molten salt heating bath and a cooling water bath. The gas is dispersed inside the bubbling microautoclave by a microporous steel grid attached to the end of the inlet tube. The gas-collection assembly was specially designed to collect all the product gases. It consists of two 75-L tanks, capable of withstanding pressures up t o 20 psig. T h e contents inside the tanks can be mixed thoroughly by means of two built-in stirrers. In addition a pump can transfer the gas back and forth between tanks to facilitate intermixing. Thorough mixing of the gases in this way was essential t o get reproducible analyses. Experimental Procedure. The microautoclave was loaded with 7 g of bitumen plus the desired amount of catalyst when used. Hydrogen was then bubbled through the microporous grid, and the system pressure was raised to the desired pressure (13.8 m a ) . The hydrogen flow rate (1LsTp/min) was established by a precalibrated mass flow meter (Brooks Instruments, Model 5850). The microautoclave was then plunged into the preheated molten salt bath and the gas collection into the previously evacuated gas-collection assembly started. The temperature of the microautoclave content was monitored constantly by an electronic temperature indicator (Analogic, Model AN2572). I t took 5-7 min to reach the desired reaction temperature (425-450 "C). Experiments lasted 30 min after the reaction temperature was reached. The reaction temperature was carefully controlled to within 3 "C by partially withdrawing the autoclave from the molten salt bath to reduce the temperature as necessary. At the end of the experiment, the microautoclave was quickly immersed into the cooling water bath while the hydrogen flow continued. When room temperature was reached (25 "C), which took about 1-2 min, the gas flow was stopped. The gas remaining
146 Energy & Fuels, Vol. 3, No. 2, 1989
-
100 I
';I
0
0
I
Chakma et al. 1
I
n
tu
U W
Rn
LA
0
25 w t %
$
+
5
3
W
z
Y
60
0
I
I
425
430
440 TEMPERATURE ( " C )
450
Figure 2. Maltene content as weight percent of feed, as functions of temperature and ZnClz concentration during hydrocracking of Athabasca bitumen (H2partial pressure 13.8 MPa; reaction time 30 min). I
I
I
I
01 425
I 430
1
I 440 TEMPERATURE ("C)
450
Figure 4. Coke formed as weight percent of feed, as functions
of temperature and ZnClz concentration during hydrocracking of Athabasca bitumen (H2partial pressure 13.8 MPa; reaction time 30 mid.
El W
5
5
01 425
430 440 TEMPERATURE ("C)
I 450
Figure 3. Asphaltene content as weight percent of feed, as
functions of temperature and ZnClz concentration during hypartial pressure 13.8 MPa; drocracking of Athabasca bitumen (Ha reaction time 30 min). in the microautoclave was then evacuated into the gas-collection assembly. Analytical Procedure. The gases were well mixed in the gas collection assembly for about 1 h and then analyzed by a gas chromatograph (Hewlett Packard, Model 5890) equipped with Porapak Q and molecular sieve columns in series. The microautoclave was weighed before and after the experiment to give the quantity of gas formed plus any liquid entrainment. Usually, entrainment was negligible as the condenser was purposely overdesigned and most of the condensable gases were returned to the microautoclave. The microautoclave was then cleaned thoroughly with dichloromethane. Dichloromethane-insolubleproducts were fdtered, dried in an oven at 100 OC, and weighed. When catalysts were used, the dichloromethane-insoluble material was washed with hot water in an ultrasonic bath to remove any ZnC12/CuC1(except those chemically combined). The solvent was then removed in a rotary evaporator (Buchi, Rotavapor RE121). n-Pentane in a 1:50 ratio was added to the liquid products and the mixture placed in an ultrasonic bath for 15 min. It was left to settle for 24 h before n-pentane-insolubles,termed asphaltenes, were filtered off. The asphaltenes were then dried in a vacuum desiccator, weighed, and collected,n-pentane was removed from the filtrate in the rotary evaporator, and the remaining yellowish material, termed maltenes, was collected.
Results Effect of ZnC1, Concentration. The effects of ZnCl, concentration on the maltene, asphaltene, coke, and gas fractions as a function of temperature are shown in Figures 2-5.
The maltene fraction increases with ZnC12concentration and decreases with temperature as can be seen from Figure 2. The asphaltene fraction, on the other hand, decreases
425
430
440
450
TEMPERATURE ("C)
Figure 5. Gas formed as weight percent of feed, as functions of
temperature and ZnClz concentration during hydrocracking of Athabasca bitumen (H, partial pressure 13.8 MPa; reaction time 30 min). with ZnC12concentration as well as with temperature as shown in Figure 3. However, ZnCl, concentration has a more pronounced effect than temperature on asphaltene cracking. As shown in Figure 4, the coke formation increases with temperature. ZnC12not only decreases coke formation with respect to the uncatalyzed reaction but also suppresses the effect of temperature, as can be seen from the reduced dopes of the curves above 440 "C. The gas formation, like the coke, also increases with temperature and decreases with ZnCl, concentration, as can be seen from Figure 5. The results presented in Figures 2-5 indicate that ZnC1, converts asphaltenes mostly into maltenes while suppressing the gas and coke formation reactions. A t high temperatures both the gas and coke fractions increase somewhat at the expense of the maltenes while there is no substantial change in the conversion of the asphaltenes. All these suggest that, in the presence of ZnC12, there is not much to be gained by operating at high temperatures. Effect of CuCl Concentration. The effect of increasing CuCl concentration on product distribution at 440 "C is shown in Figure 6: the maltene and coke fractions both increase, and the asphaltene fraction decreases, while the gas fraction remains virtually unchanged. Effect of Tetralin Addition. Figure 7 shows the effect of tetralin on product distribution in the absence of any molten salt catalysts. The maltene fraction increases asymptotically. The gas, asphaltene, and coke fractions all decrease. A substantial quantity of tetralin is required to reduce coke formation by an appreciable amount.
Energy & Fuels, Vol. 3, No. 2, 1989 147
Hydrocracking of Athabasca Bitumen
Table 111. Effect of CuCl Addition on the Reactivity of ZnC1, in Terms of Product Distribution (T= 440 OC) product, wt % of feed catalvst coke amhaltene maltene eas 4.8 10.4 66.7 18.1 none 10% ZnCl, 4.1 3.2 72.1 20.6 10% ZnCl, + 1.5% CuCl 3.6 6.1 70.0 20.3 72.4 19.0 10% ZnC1, + 4.5% CuCl 3.4 5.2 ~~
ASPHALTENE COKE
01 0
I 5
IO
CuCl ( w t % )
Figure 6. Weight percent of various fractions as a function of CuCl added during hydrocracking of Athabasca bitumen at 440 "C (H2 partial pressure 13.8 MPa; reaction time 30 min). 80
I
I
0
25 TETRALIN (WT %)
50
Figure 7. Weight percent of various fractions as a function of tetralin added during hydrocracking of Athabasca bitumen at 450 "C (H2 partial pressure 13.8 MPa; reaction time 30 min). Table 11. Effect of Catalyst Type and Quantity on Product Distribution (T = 440 "C) product, w t % of feed catalyst coke asphaltene maltene gas 66.7 18.1 none 4.8 10.4 69.9 18.0 1.5% ZnClz 3.8 8.3 67.9 18.2 1.5% CuCl 6.7 7.2 10% ZnCl, 4.1 3.2 72.1 20.6 10% CuCl 5.9 6.3 69.8 18.0
Catalyst Reactivity. The reactivity of ZnClpand CuCl catalysts in terms of product distribution is compared in Table 11. ZnClz shows more reactivity toward asphaltene cracking than CuC1. The maltene fraction and to some extent the gas fraction is higher with ZnCl,, while the coke formation is higher with CuC1.
Table IV. Effect of Tetralin Addition on Catalyst Reactivity in Terms of Product Distribution (T = 440 "C) Droduct. wt % of fced catalyst coke asphaltene maltene gas 66.7 18.1 4.8 10.4 none 4.1 3.2 72.1 20.6 10% ZnC1, 10% ZnCl, + 25% tetralin 4.13 7.16 72.71 16.0 10% CuCl 5.90 6.3 69.8 18.0 67.50 15.80 10% CuCl + 25% tetralin 4.47 12.24
Table I11 shows the effect of CuCl addition on the reactivity of ZnC1,. The asphaltene cracking reactivity of ZnCl:, decreases with CuCl addition. Coke formation and to a lesser extent gas formation also decrease with CuCl addition, which can be attributed to the improvement of the hydrogenation activity of ZnC1, in the presence of CuCl as was observed by Nakatsuji e t al.I1 Table IV shows that tetralin addition reduces the asphaltene cracking activity of both ZnC1, and CuC1. Tetralin also reduces the amounts of coke and gas formed. Some observations on the performance of the bubbling microautoclave with respect to conventional stirred autoclaves are worth mentioning. The present results were compared with some of the results of Herrmann et al.,15 carried out under somewhat similar conditions with Athabasca bitumen. A substantial difference in, the quantity of gas formed was noticed. In the present case it ranged from 15 to 23 wt 70 for runs carried out a t 440 and 450 "C. In the case of Herrmann et al.,14it was about 6 wt % under similar conditions. This is probably an indication of improved cracking in the microautoclave. We attribute this to a homogeneous distribution of temperatures and gas throughout the fluidized system and to the absence of hot spots on the walls. Conclusions. Athabasca coker feed bitumen has been hydrocracked in a bubbling microautoclave with molten halide catalysts. ZnC1, has been found to be an effective catalyst for the conversion of asphaltenes into maltenes. CuCl is less reactive than ZnCl, and reduces the latter's reactivity when added to it. Tetralin addition reduces the asphaltene cracking ability of ZnC1, and CuCl. It also reduces gas and coke formation, and a substantial quantity of tetralin is required to have any significant reduction in coke.
