1266
Ind. Eng. Chem. Process Des. Dev. 1985, 2 4 , 1266-1275
charge show that combined processing yields a net oil increase of 23.3% for 90-min reaction while tetralin provides a net oil increase of 17.7% for 30 min of reaction. Combined processing adds greater flexibility to the liquefaction of coal by eliminating the need for solvent sufficiency and for large recycle streams. Additionally, coprocessing provides an avenue by which difficult to refine heavy petroleum crudes and residua with large asphaltene contents can be upgraded. Combined processing should be considered as a transitional concept, leading to the coal liquefaction processing plants of the future while utilizing and upgrading difficult to refine heavy petroleum materials of today. Acknowledgment We gratefully acknowledge the support of the US Department of Energy and Cities Service Research and Development Co. under Contract DEFG2282PC50793 for this work. We also gratefully acknowledge the coal provided by the Wilsonville Advanced Coal Liquefaction Research and Development Facility and the petroleum crudes and
residua provided by Cities Service Research and Development Company. Registry No. Ni, 7440-02-0; Mo, 7439-98-7.
Literature Cited Curtis, C. W.; Guin, J. A.; Tarrer, A. R.; Huang, W. J. Fuel Proc. Techno/. 1983, 7 , 277. Curtis, C. W.; Pass, M. C.; Guin, J. A. submitted to Ind. Eng . Chem . Process Des. Dev. 1985. Goliakota, S. V.; Guin, J. A,; Curtis, C. W. Ind. Eng. Chem. Process Des. Dev., in press. Green, D. C.; Broderlck, D. H. Chem. Eng. Prog. 1981, 33. Guin, J. A.; Curtis, C. W.; Kwon. K. C. Fuel 1983, 62, 1412. Ho, P. N.; Weller, S . W. Fuel Proc. Techno/. 1981, 4 . 21. Kriz, J. F. Can. J . Chem. Eng. 1983, 61, 68. Moschopedis, S. E.; Hawkins, R. W.; Speight, J. G. Fuel Proc. Techno/. 1982, 5 , 213. Prasher, 8. D.; Gabriel, G. A,; Ma, Y. H. Ind. Eng. Chem. Process D e s . Dev. 1978, 17, 266. Tamm, P. W.; Harnsberger, H. F.: Bridge, A. G. Ind. Eng. Chem. Process D e s . Dev. 1981, 20, 262. Ternan, M. Can. J. Chem. Eng. 1983, 6 1 , 689.
Received for review August 6 , 1984 Revised manuscript received January 9, 1985 Accepted January 21, 1985
Two-Stage Non-Hydrogenative Processing of Residua Ashok S. KrIshna"+and David J. Bott$ Gulf Research and Development Company, Pittsburgh, Pennsylvania 15230
A circulating fluid catalytic cracking pilot plant was used to demonstrate the concept of a two-stage approach (pretreatment over inert substrate followed by conventional catalytic cracking of pretreated oil) for upgrading petroleum residua. A microactivity test unit was used to delineate and distinguish the effects of nickel and vanadium on yields during pretreatment of resid and cracking of pretreated oil. For resids from three different crude sources (West Texas, North Slope, and Kuwait), the yield distributions from the twsstage process are compared with those from direct, single-stage catalytic cracking, and the two-stageapproach is shown to be an excellent way to maximize product selectivity toward middle distillates.
Over the next decade, worldwide prospects indicate that more heavy crude will be refined while product demand will shift toward more middle distillate and less residual oil. Thus, reduction of the bottom of the barrel is the refining industry's main target for the next several years. Among the resid conversion schemes of interest, processing of heavy oils in a fluid catalytic cracking (FCC) unit is receiving considerable attention (Murphy, 1980; Zandona et al., 1982; Campagna et al., 1983; Dean et al., 1983). Resids relatively low in metals and coke precursors can be blended with conventional gas oil feeds and cracked in existing FCC units with a minimum of modifications (Krishna et al., 1984). For poorer quality residua, often two or more stages of processing are required to economically produce prime liquid products (Yanik et al., 1977; Murphy and Treese, 1979; Teichman et al., 1982; Vermillion and Gearhart, 1983). Among the more attractive multistage process options is a two-stage combination involving non-hydrogenative thermal treatment of residua to effect metals and asphaltenes removal using solid substrates, followed by conventional catalytic cracking of the pretreated oils. The concept of using an FCC unit to +Presently with Chevron Research Co., Richmond, CA. with W.R. Grace & Co., Columbia, MD.
1 Presently
process heavy oils in two stages is not new: the initial operation of the heavy oil cracker at the Phillips' Borger, TX Refinery appears to have been based on such a scheme in the early 1960s (Rush, 1981). A similar concept was proposed and pilot plant tested at Gulf Research & Development Co. (GR&DC) in the mid-1950s; since this effort predated the advent of zeolitic catalysts and short contact time riser cracking, the studies were carried out by using an amorphous cracking catalyst and a fluidized bed cracking reactor. The results obtained in this early investigation, summarized in Table I, showed that essentially complete removal of metals and substantial reduction of carbon residue of the feed can be accomplished in the first, mild cracking stage. The pretreated stocks produced had vanadium contents less than 1.0 ppm and carbon residues of 2.0-3.0 w t % compared to 45.0 ppm and 9.8% for the feed (Kuwait reduced crude). Due to the poor product distribution obtained in the fluidized bed reactor, and the detrimental effect of high metals contents on the amorphous catalyst, the concept proved to be uneconomical at that time. Since the 1960s, several changes in the refining environment have taken place that justify renewed interest in such a two-stage processing scheme: economics for resid conversion have become much more favorable; although there have been very significant improvements in FCC
0196-4305/85/1124-1266$01.50/00 1985 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
Table I. Two-Stage Processing of Kuwait Residua stage OAPI
s, wt %
C residue, w t % distillation OF, at 10% 30 % 50 % 70% 90 % Ni, PPm v, PPm
I Feedstock Properties 15.1 4.12 9.8
I1
45.0 13.4
930 22.0 10.0 same
midcontinent gas oil 27.9 0.59 0.0946 12.72 190.2 0.33 0.3 0.3
West Texas resid 21.7 1.56 0.17 12.00 174.2 4.6 10 33
North Slope resid 15.8 1.58 0.35 11.19 181.8 7.79 17 35
596 685 765 845 934
510 628 780
713 824 908 966
693 803 933
32.2 11.8 10.9 9.5
37.2 18.9 12.5 5.8
61.7 25.2 36.5 0.8 5.3
45.9 19.8 26.1 7.4 2.5 7.0
43.0 21.3 42.6 6.5 0.7 1.8 34.0 12.3 21.8 19.8 3.2
48.1 17.5 16.9 10.4 2.5 0.8 30.3 17.9 12.4 15.6 6.0
Kuwait resid 15.7 3.96 0.22 11.13 187.7 9.76 15 47
%
Yields, vol % of Feed propane-propene 2.6 butane-butene 2.7 gasoline 17.9 furnace oil 16.5 slurry oil 52.9 C2and lighter, wt % 2.2 coke, w t % 12.5 30.6 conversion, vol %' 10 RVP gasoline yield, vol % 19.2 res octane no. 86.5 motor octane no. 75.0 furnace oil - (650+
description gravity, OAPI s,wt % N, wt 70 H, wt 70 aniline pt, O F C residue, w t % Ni, P P ~ v, PPm distillation, D1160, O F 10% 30 % 50 % 70% 90 % hydrocarbon type,
707 740 791 851 950 0.14 0.05
Operating Conditions reactor temp, O F 930 reactor pressure, psig 10.0 catalyst/oil ratio 10.0 catalyst type amorphous silica-alumina
' [lo0 - (430 - 650 O F )
Table 11. Charge Stock Properties total
18.4 3.31 2.29
707 805 915 cracked
1287
OF)
6.6 7.4 34.8 17.6 30.6 4.25 8.5 51.8
6.1 6.6 36.3 25.8 16.2 4.4 10.9 58.0
37.4 90.1 78.1
39.0 88.3 76.6
slurry oil].
catalyst and metals passivation technology, direct cracking of poor quality resids remains infeasible; short contact time riser cracking can minimize conversion during the pretreatment step and improve selectivity to desired products in the cracking step; product demand is shifting toward higher distillate/gasoline ratios; and, finally, techniques for minimizing dehydrogenation reactions catalyzed by metals, developed for the FCC process, can be applied to the pretreatment step. In 1976, a process for pretreating resid by contacting with a porous, refractory oxide in the absence of hydrogen was patented (Rosynek et al., 1976); a recent paper discussed the effects of various process parameters (including oil/catalyst ratio, temperature, and steam) and different substrates for this pretreatment process using a trickle bed reactor, but no information was provided on the secondstage catalytic cracking of the pretreated oils (Yan, 1984). Englehard has developed the Asphalt Residual Treatment (ART) process which involves thermal treatment of heavy oils in a short residence time contactor using a proprietary substrate (Bartholic and Haseltine, 1981). Itoh et al. (1984) recently presented yield data for a two-stage method for catalytic cracking of residual oils and tar-sand bitumen (where the first stage involved pyrolysis of the resid in the absence of a substrate), but detailed product quality information was not included. In principle, all the two-stage non-hydrogenative processes deescribed thus far are similar in that they involve substantially noncatalytic treatment of the heavy oil, followed by conventional catalytic cracking of the pretreated oils. The purpose of the present work is to evaluate the non-hydrogenative two-stage scheme for application in existing FCC units with minimum or no modifications using state-of-the-art catalytic cracking technology. Experiments were conducted in microactivity reactors to study the effects of contaminant metals on the first- and
aromatics mono di tri+ thiophenes unidentified saturates alkanes cycloalkanes polar compds insolubles volatiles
second-stage processes and in a large FCC pilot plant to study the effects of process parameters, feedstock type, and quality and to generate sufficient quantities of products to observe quality variations. Finally, overall yield and product quality comparisons between direct (single stage) cracking and two-stage processing of resids were investigated.
Experimental Section Materials. A midcontinent gas oil and residual feedstocks from three different crude sources, West Texas, North Slope, and Kuwait crudes, were used as charge stocks for the study; properties of these oils are given in Table 11. Properties of substrate used for the pretreatment stage and catalyst used for the second stage are shown in Table 111. Crushed, calcined C-2 alumina, obtained from the Kaiser Corp., was chosen as the substrate for the pretreatment step based on its low surface area and low cost; although the alumina's particle size distribution is coarser than that of conventional, microspherical FCC catalyst, the material was found to have adequate fluidization characteristics. Comparison of microactivity test data using quartz chips and C-2 alumina (Table 111) indicates that the alumina is catalytically inactive. The commercial, high activity, fresh zeolitic cracking catalyst (designated as catalyst Y) was obtained from HarshawFiltrol and the equilibrium version of the same catalyst from a commercial FCC unit. Apparatus. The microactivity test (MAT) unit used for this study is similar to the original MAT unit (Henderson and Ciapetta, 1967) except that feed is injected through a hypodermic needle permanently mounted on the reactor head, and the liquid recovery is established by weighing instead of volumetric measurements. The experimental apparatus used to conduct the pilot plant scale studies is a continuous, adiabatic, 1.0 B/D, transfer line cracking reactor and fluidized bed regenerator setup. This pilot plant was designed specifically to process heavy oils and features a fully integrated process control, data ac-
1268
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
Table 111. ProDerties of Substrate and Catalvsts Tested
descrbtion surface area, m2/g app bulk density, g/cm3 compacted bulk density, g/cm3 particle size distribution,
quartz chips
C-2 alumina 4.3
equilib catalyst
fresh catalyst
Y
Y
145.4 0.844
207.2 0.758
0.955
0.841
IO
1.13
H, C2-and lighter coke
920
840
880
980
1000
820
940
880
980
1000
4 -t------r---r900 920 840
880
880
1000
840
880
980
1000
SEO
lo00
a
wt%
106 attrition index MAT testn conversion, vol %b yields, vol 70 total C3's propylene total C i s isobutane butenes gasoline (C6-430 "F) furnace oil (430-650 O F ) slurry oil (650+ O F ) yields, wt 70
+-r----,--
900
0 3.0 15.6 33.9 47.5 23.7
1.7 10.4 57.3 30.6 4.4
0.3 5.9 38.4 28.3 27.1 7.0
d
28-
' - / 54 800 15
12.74
13.08
72.67
78.79
1.12 0.90 0.48 0.05 0.39 6.70
1.09 0.78 0.46 0.02 0.42 8.59
8.28 6.48 14.56 6.65 1.56 58.50
8.83 6.39 15.13 7.94 1.97 64.64
27.85
27.29
19.00
15.87
59.42
59.63
8.33
5.34
0.01 0.82 0.30
0.01 0.96 0.59
0.08 2.18 2.83
0.05 1.99 4.61
Typical conditions: feed: midcontinent gas oil (properties in Table 11); initial reactor temperature: 960 OF;weight hourly space velocity: -16. b[lOO - (430 - 650 O F ) furnace oil - (650+ O F ) slurry oil].
quisition, and display system. Oil feed is continuously drawn from a constant temperature reservoir and discharged to the low-temperature section of the feed oil preheater. The oil flows through the low- and high-temperature preheater sections into the oil-catalyst section to meet hot regenerated catalyst and dispersion steam. Oil, steam, and catalyst enter the riser reactor together with bleed nitrogen flowing concurrently through the reactor where the cracking reactions take place. The reaction products are carried into a small disengaging chamber and a filter where the entrained catalyst is separated from the product fluid. Hydrocarbons which have been absorbed on the catalyst are removed in the spent catalyst stripper by steam and N2. The stripped hydrocarbons, N2, and steam pass into the reactor disengaging space where they follow the same path as the reactor vapors. The vapors from the reactor cyclone pass to the product recovery train which consists of a quench tower, pressurized stabilizer, slurry oil, furnace oil, and gasoline towers. Spent catalyst from the reactor/stripper flows through a standpipe and slide valve into a lift unit which effectively permits control of the catalyst circulation rate. In the lift unit, the catalyst is picked up by air or nitrogen and carried up the transfer line into the regenerator. Here, gases, excess air, and steam leave the top of the regenerator and pass in series through a cyclone, filter, surge bomb, double-pipe steam heater, a pressure control valve condenser, water accumulator, and a wet test meter. Regenerated catalyst flows by gravity through a slide valve into the oil-catalyst transfer line, where hot regenerated catalyst and dispersion steam meet with the oil to be cracked.
-..
,
I
1 900
920
840
980
-_
980
RISER OUTLET TEMP, F
900
820
-
1000
-----
2
2 5
900
820
940
980
RISER OUTLET TEMP, F
Figure 1. Resid pretreatment with C-2 alumina-riser outlet temperature effects.
Procedure. The fresh catalyst samples were pretreated prior to use in experiments. This pretreatment consisted of a heat shock treatment at 1100 or 1200 O F for 1 h, followed by a 14-h steam purge at 1350 OF at atmospheric pressure with 100% steam. Equilibrium samples of catalyst were calcined for 10 h at lo00 O F prior to use. Nickel and vanadium impregnations on catalysts were carried out by using the procedure described by Mitchell (1980). Prior to charging feed to the pilot plant, the catalysts or substrates were loaded and circulated for 10-20 h in order to establish desired flows, pressures, and nitrogen/steam settings. Following this initial period of catalyst circulation, a midcontinent gas oil was charged and the unit operated for 14-18 h to establish temperatures and levels in the product recovery system. Once stable operation with the gas oil feed was achieved, the feed was switched to the residual oil under study and the unit operated for 16-20 h to reach a steady state. Following this, 8 h of data and product were collected and analyzed to arrive at average yields and product quality information reported herein. Results and Discussion West Texas Resid Data. First- and second-stagepilot plant resulta for West Texas resid are summarized in Table IV. Although virgin, atmospheric resids and 650+ O F pretreated oils were used as feedstocks, conversion is defined as volume percent 100 - (430+ O F product) in keeping with the convention adopted in fluid catalytic cracking. The substrate used during the pretreatment step was Kaiser alumina, while equilibrium catalyst Y was used in the cracking step. The pretreatment experiments involved pilot plant runs at three different riser outlet temperatures and a fourth run at a higher catalyst inlet temperature to study the effect of lowering the substrate/oil ratio. The effect of increasing temperature during pretreatment, shown graphically in Figure 1, is typical of thermal cracking reactions: small increases in conversion and lower yields of the 650+ OF product, reductions in coke yield due to increased vaporization of the feed, and overcracking of the gasoline at the highest temperature, resulting in a decrease in gasoline yield and a corresponding
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985 1269 Table IV. West Texas Resid Processing-Pilot Plant Data Summary P1 P2 run no. catalyst/ substrate feedstock riser outlet, OF riser inlet, OF feed preheat, O F conversion, vol %' yields, vol % total C3's
C3= total C,'s i-Ca L4T
gasoline (C5-430OF) furnace oil (430-650 OF) slurry oil (650+ OF) total C3+ liq vields, wt %
H2 C2 and lighter coke product properties gasoline gravity, OAPI
s, wt %
hydrocarbon type, vol % aromatics olefins saturates maleic anhydride value oxidation stability, D525, min research octane no. motor octane no. furnace oil gravity, "API
s,wt
%
hydrocarbon type, vol % aromatics olefins saturates cetane no. slurry oil gravity, "API
s, wt %
C residue, concn, w t % Ni, ppm v, PPm
P3b same same 973 1200 530 21.6
P4b same same 910 1250 530 14.2
F1 zeolitic catalyst 650+ OF slurry oil from run P I 982 1200 530 75.7
12.6 25.7 55.5 99.3
5.0 3.7 2.2 0.1 1.9 11.7 29.0 49.4 97.7
1.8 1.3 0.8 0.03 0.7 7.8 24.9 60.9 96.4
12.5 10.3 17.8 5.0 11.5 54.6 14.0 10.3 109.1
0.03 1.3 4.7
0.10 1.7 4.3
0.11 2.6 4.1
0.03 1.0 4.5
0.06 2.5 6.8
53.4 0.176
54.6 0.364
49.3 0.212
22.4 17.0 0.0 47.0 120.0a 75.8 62.9
24.4 75.3 0.0
28.1 44.3 27.7
76.4 64.5
92.6 78.4
34.1 0.47
33.9 0.90
18.9 1.94
25.8 18.7 55.5 52.2
28.2 18.6 53.2
64.8 0.0 35.2
21.4 1.49 2.06 1.9 5.4
19.6 1.56 2.64 1.6 4.1
10.3 4.30 3.01
C-2 alumina West Texas Resid 911 1200 530 13.5
same same 949 1200 530 18.8
2.7 1.9 1.3 0.1 1.0 7.2 24.5 62.0 97.7
3.8 2.7 1.7 0.1
aGulf specification is 360 min. bProduct quality data not obtained.
increase in light gases. The effect of lowering the substrate/oil ratio, as shown in Table IV, is to decrease coke and light gas yields without significant changes in liquid product yields. Properties of products from two pretreating runs are also summarized in Table IV. As expected, the thermally cracked gasoline is highly olefinic and is comprised of over 75 vol % olefins. The maleic anhydride value and the oxidation stability test result for the gasoline product from run P1 indicate a potential problem with stability of the gasoline. The octane numbers are typical of thermally cracked gasolines and are significantly lower than those of FCC products. The cetane number of the furnace oil is relatively high, at 52.2; the furnace oil is also high in olefinic content compared to catalytically cracked product. The carbon residue and metals levels in the slurry oil product indicate that extensive removal of these contaminants had been achieved. The slurry oil product from pretreatment run P1 was riser cracked at typical FCC conditions using catalyst Y. The results, shown in Table IV, indicate that the pretreated oil from the West Texas resid will constitute good quality feed to existing FCC units. Feedstock Effects. First- and second-step pilot plant experiments, similar to those described above for West
[lo0 - (430 - 650 OF) furnace oil - (650+ OF) slurry oil].
Texas resid, were conducted with a midcontinent gas oil and residua from North Slope and Kuwait crudes. Results from the pretreating pilot plant runs are summarized in Table V; the West Texas data are repeated for comparison. At similar run conditions, conversion ranged from 18.8 vol % with the West Texas resid to 26.8 vol % with the Kuwait feed. The furnace oil-to-slurry oil ratio varied considerably, ranging from a high of 0.46 for the West Texas to a low of 0.18 for the gas oil. The higher value for the West Texas resid relative to the other feeds is probably due to greater amounts of easily crackable saturates (Table 11). The gas oil likely produced less distillate because the molecules in the gas oil compared to the resid fraction crack with more difficulty. The relationship between pilot plant coke yield and feedstock carbon residue is shown in Figure 2. It is evident from the plot that the coke yield is roughly 95% of the feed carbon residue for the set of experimental data reported in Table V. Effectiveness of contaminant removal during pretreatment for the three different residua is summarized in Table VI. Higher than 88% removal of metals is achieved at the conditions tested for all feedstocks, while sulfur and carbon residue removal was in the 27-33% and 61-78% ranges. The positive effect of reaction temperature on metals removal for the West Texas resid is in
1270
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
Table V. Pilot Plant Pretreating with C-2 Alumina run no. P5* charge stock conditions riser outlet riser inlet conversion, vol 70" yields, vol ?& total C,'s C,= t o t i (2,'s
i-C, C,= gasoline (C5-430 O F ) furnace oil (430-650 slurry oil (650+ O F ) total C3+ liq yields, wt %
OF)
H* C2 and lighter coke product properties gasoline gravity, OAPI w t 90 hydrocarbon type, vol 70 aromatics olefins saturates research octane no. motor octane no. furnace oil gravity, OAPI s, wt 70 hydrocarbon type, vol 70 aromatics olefins saturates cetane index (no.) slurry oil gravity, OAPI s, wt 70
midcontinent gas oil
P2 West Texas resid
P6 North Slope resid
P7 Kuwait resid
956 1200 20.0
949 1200 18.8
946 1200 19.9
944 1200 26.8
5.1 4.1 2.6 0.1 2.0 12.1 12.3 67.7 99.8
3.8 2.7 1.7 0.1
3.8 2.7 1.8 0.1
12.6 25.7 55.5 99.3
1.5 10.8 15.5 64.6 96.6
4.3 3.0 2.4 0.1 1.9 13.4 11.7 61.5 93.2
0.02 1.5 1.2
0.10 1.7 4.3
0.20 2.0 8.2
0.13 3.1 10.3
54.6 0.364
56.4 0.210
55.9 0.509
24.4 75.3 0.0 76.4 64.5
20.7 61.5 17.8 79.2 67.6
19.7 68.2 12.1 75.2 64.9
33.9 0.90
28.2 0.89
27.2 2.19
28.2 18.6 53.2 48.4
32.3 27.6 40.1 41.8 (40.3)
38.9 21.8 39.3 37.1
19.6 1.5
17.5 1.44
16.9 3.91
s,
a
[lo0 - (430 - 650 O F ) furnace oil - (650+
OF)
slurry oil]. *Product quality data not obtained.
Table VI. Effectiveness of Contaminant Removal" during Pretreatment run no. P1 P2 P6 P7 feedstock riser outlet temp, O F Ni removal, 70 V removal, 70 S removal, 70 C residue removal, 70
West Texas same North Slope Kuwait resid resid resid 911 949 946 944 88 90 32 72
91 93 27 67
88 89 33 61
I
95 96 32 78
Calculated as (contaminant level in feed - contaminant level in total liquid product) X 100 divided by contaminant level in feed.
A
MID CONTINENT GAS
OIL
W E S T TEXAS RESID NORTH SLOPE RESID
general agreement with observations made by Yan (1984). Slight decreases in sulfur and carbon residue removal were observed at the higher temperature, probably due to increased vaporization of heavier, more sulfur-containing portions of the feed, which also results in lower coke yield. It should be noted that the effectiveness of contaminant removal over metal-loaded alumina was not determined in our study. However, recent commerical data on the ART process (Busch et al., 1984) indicate that the effectiveness of metals removal is relatively independent of metals on substrate in the 7000-15,000 ppm nickel and vanadium range and comparable to that observed in our study with metals-free alumina. Experiments were also conducted on the MAT unit (at conditions shown in Table 111) by using the same feedstocks and substrate that were used in the pilot plant. The
I
KUWAIT RESID
0
1
I
I
2
3
I 4
I 5
I
I
1
I +
6
7
8
9
10
11
F E E D CARBON RESIDUE, WT.%
Figure 2. Resid pretreatment with C-2 alumina-pilot yield vs. feed carbon residue.
plant coke
results are summarized graphically in Figure 3. The MAT results also show a linear carbon residue-coke relationship. The high distillate yield from the West Texas stock, relative to the other two feeds, is also evident. The correspondence between MAT and pilot plant data is shown in Figure 4. It is apparent that the sets of yield data can be correlated fairly well; catalyst-to-oil contact time differences are likely responsible for the different slopes of
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
1271
-- I
WEST TEXAS RESID
0 NORTH SLOPE RESID KUWAIT RESID
"
A
CONV., VOL.%
COKE, WT.xlO
GASOLINE, VOL.%
Table VII. Pretreated S l u r r y Oil Properties feed descript slurry oil from slurry oil from slurry oil West Texas North Slope from resid resid Kuwait resid pretreatment run no. P1 P6 P7 properties 17.5 17.1 gravity "API 21.4 1.44 3.82 s, wt % 1.49 0.32 0.16 Ni, wt % 0.15 11.19 10.80 H, wt % 11.75 aniline point, O F 176.4 164.3 163.4 4.68 3.61 C residue, wt % 2.06 3.3 1.3 1.9 Ni, ppm 6.0 3.3 v, PPm 5.4 distillation, D1160,
CONVERSION, V 0 L . X C O K E , WT. 5
40 -
FURNACE OIL, VOL.%
GASOLINE. VOL. % FURNACE OIL. VOL. 5
/
"F I
0
10
20
I
I
30
40
50
MAT YIELD, %
Figure 4. Resid pretreatment with C-2 alumina-relationship pilot plant and MAT unit yields.
of
lines relating riser pilot plant and fixed-bed MAT data. Thus, while the MAT unit cannot provide information on the effectiveness of contaminant removal during pretreatment, the MAT yield distribution appears to be a good indicator of potential pilot plant performance. The properties of liquid products from the pretreating experiments are summarized in Table V. The gasolines, again, are highly olefinic, containing greater than 60 vol % olefins. The furnace oils have relatively high cetane indexes, in the range of 37-48. Detailed inspections of the pretreated slurry oils are presented in Table VII. These pretreated oils were riser cracked at conventional FCC conditions using equilibrium catalyst Y, and the results are summarized in Table VIII. At roughly 10 O F higher reaction temperature, the conversion of North Slope-derived oil is significantly lower than that of West Texas and Kuwait oils. As shown in Table VI, carbon residue removal during pretreatment was also the lowest for this resid, resulting in high coke yields from the pretreated oil during the cracking step. This poor susceptibility of the North Slope stock to cracking, relative to West Texas and Kuwait oils, is likely due to its more refractory molecular composition and higher nitrogen (polar) content; nitrogen compounds are well-known cracking catalyst poisons and
10% 30 % 50 % 70 % 90 % hydrocarbon type,
639 718 787 882 1019
718 797 856 94 1 1056
676 761 845 933 1104
43.7 14.1 15.4 8.9 4.8 0.3 49.8 21.5 28.3 6.5 0.0
54.3 21.6 17.9 11.0 3.0 0.8 37.9 11.9 25.9 7.8 0.0
55.2 22.4 16.2 7.5 8.8 0.3 37.1 16.9 20.2 7.2 0.5
%
aromatics mono di tri+ thiophenes unidentified saturates alkanes cycloalkanes polar compds insolubles
contribute to poor coke selectivity (coke/conversion) in the FCC process (Mills et al., 1950). FCC of pretreated oils was also demonstrated on the MAT unit. The conversion and coke data are summarized in Figure 5. The North Slope stock again yielded lower conversion and higher coke relative to the other stocks, confirming the pilot plant results. The virgin residua were also cracked in the MAT unit, and the results, shown in Figure 5, indicate that the North Slope resid is also the poorest of the three feeds for direct, single-stage processing. Since the nitrogen content of the virgin and pretreated North Slope stock were roughly the same, the high nitrogen level could explain poor susceptibility of this feed to both single- and two-stage processing.
1272
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
Table VIII. FCC of Pretreated Slurry Oil (Catalyst Y) run no. feed riser outlet, O F riser inlet, O F feed preheat, O F conversion vol %" yields, vol % total C3's C3= total C4's IC,= C,' gasoline (C,-430 OF) furnace oil (430-650 O F ) slurry oil (650+ O F ) total C3+ liq yields, wt %
HZ C2 and lighter coke prod properties gasoline gravity, OAPI
s, w t %
hydrocarbon type, vol % aromatics olefins saturates research octane no. motor octane no. furnace oil gravity, OAPI s, wt Yo hydrocarbon type, vol % aromatics olefins saturates cetane index slurry oil gravity, OAPI s, wt % [lo0 - (430 - 650 100
OF)
F1 slurry oil from West Texas resid 982 1200 520 75.7
F2 slurry oil from North Slope resid 993 1200 520 65.0
F3 slurry oil from Kuwait resid 983 1200 520 70.2
12.5 10.3 17.8 5.0 11.5 54.6 14.0 10.3 109.1
10.1 7.9 11.3 3.4 7.0 48.5 19.7 15.4 104.9
10.5 9.0 15.5 3.7 10.9 53.8 15.8 14.0 109.6
0.06 2.5 6.8
0.08 3.4 9.8
0.06 3.1 6.1
49.3 0.212
57.3 0.115
57.4 0.295
28.1 44.3 27.7 92.6 78.4
28.7 38.9 32.8 92.1 79.9
24.6 48.4 27.0 91.9 78.1
18.9 1.94
22.1 0.94
15.6 4.71
64.8 0.0 35.2 26.0
65.3 0.0 34.7 30.0
76.0 0.0 24.0 20.0
4.3 3.01
6.8 2.02
-1.1 8.38
furnace oil - (650+ O F ) slurry oil]
-
UNTREATED
--
PRETREATED
-
80
60
40
WEST TEXAS RESID
0 NORTH SLOPE RESID
20
KUWAIT RESID
0 CONV., VOL.%
COKE, WT.% x 1 0
CONV., VOL.%
COKE, WT.% x 10
Figure 5. MAT yields with untreated and pretreated stocks-catalyst
Y.
Overall yields for the two-stage process, shown on Table IX, indicate that this scheme is an excellent way to maximize distillate yield. For the West Texas resid, total two-stage gasoline/furnace oil and furnace oil/slurry oil ratios are 1.24 and 5.20, respectively, compared to 2.9 and
1.5 estimated for a typical single-stage operation. About 17% of the two-stage process gasoline is of poor quality and must be reformed; however, about 74% of the furnace
oil boiling range material is good quality distillate with a high cetane number. Similar conclusions apply for the
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985 1273 Table IX. Two-Stage Cracking of West Texas Resid (A), North Slope Resid (B), and Kuwait Resid (C)-Pilot Plant Data F1 (FCC of estimate for direct run no. (descrbtion) riser temp, O F conversion, vol 70 gasoline, vol % furnace oil, vol % slurry oil, vol % coke, wt % total C3+ liq, vol % product ratios gasoline/ furnace oil furnace oil/slurry oil
run no. (descriDtion)
P 2 (resid DretreatmenV) 911 13.5 7.2 24.5 62.0 4.7 97.7
pretreated 650+ O F Droduct) A 982 75.7 54.6 14.0 10.3 6.8 109.1
total pretreatment and FCC
60.4 41.1 33.2 6.4 9.0 103.3
cracking with metals on equilib catalvstb 980 70 52 18 12 10 104
1.2 5.2 P 6 (resid metreatment)
2.9 1.5
F2 (FCC of pretreated 650+ O F moduct)
total pretreatment and FCC
993 65.0 48.5 19.7 15.4 9.8 104.9
61.9 42.1 28.2 9.9 14.5 99.8
B riser temp, O F conversion, vol % gasoline, vol % furnace oil, vol % slurry oil, vol % coke, wt % total C3+ liq, vol % product ratios gasoline/furnace oil furnace oil/slurry oil
run no. (description)
946 19.9 10.8 15.5 64.6 8.2 96.6
1.5 2.8 P7 (resid pretreatment)
F3 (FCC of pretreated 650+ O F product)
total pretreatment and FCC
983 70.2 53.8 15.8 14.0 6.1 109.6
68.5 46.0 23.1 8.4 13.7 99.5
C riser temp, O F conversion, vol gasoline, vol % furnace oil, vol % slurry oil, vol % coke, wt % total C3+ liq, vol 70 product ratios gasoline/furnace oil furnace oil/slurry oil
942 26.2 13.6 13.6 60.2 10.0 93.7
2.0 2.8
+
OLess than 500 ppm Ni V on substrate. bEstimate assumes a catalyst makeup rate of -2 lbs/bbl resulting in -1500 ppm Ni and 5000 ppm V on the equilibrium catalyst. Antimony and tin use for metals passivation was assumed. Estimate based on pilot plant data with low metals equilibrium catalyst and MAT data on metals effects. [lo0 - (430 - 650 OF) furnace oil - (650+ O F ) slurry oil].
data on North Slope and Kuwait residua; in each case, the gasoline/furnace oil ratio is less than 2.0, and the furnace oil/slurry oil ratio is 2.8. In the following section, this comparison between single- and two-stage processing is reexamined with higher metals levels on substrate and catalyst. Metals Effects. Residual feeds contain relatively high levels of contaminant metals such as nickel and vanadium which are well-known cracking catalyst posions. While both nickel and vanadium act as dehydrogenating agents, contributing to coke and gas make, vanadium can attack the zeolite in the cracking catalyst and destroy its activity (Campagna et al., 1983). In order to quantify the effect of nickel and vanadium on substrates during the pretreatment stage, and to realistically compare the two-stage process (which would involve high levels of metals on the substrate in the first stage and relatively low levels of metals on cracking catalyst in the second stage) with direct, single-stage FCC of residua (which would involve relatively high levels of metals on cracking catalyst), a series of MAT experiments was undertaken by using the three residua described in Table 11. Arbitrary levels of nickel (5OOO ppm) and vanadium (10000 ppm) were chosen for the substrate (Kaiser alumina) used in the pretreatment runs and for the cracking catalyst Y (6000 ppm each of nickel and va-
nadium) used in the second-stage cracking runs. The same substrate and catalyst were used for all three feedstocks. The rationale for the arbitrary selection of metals levels was based on the need to limit the number of experiments since (a) the optimum amount of metals on substrate and catalyst can differ from feed to feed (the absolute and relative levels of nickel and vanadium are different for each of the three feeds, and demetallization levels during pretreatment also differ from feed to feed), (b) the make-up rate of the substrate in the first stage and catalyst in the second stage may be varied to achieve a wide variation in metals levels on substrate and catalyst, and (c) cracking catalysts are being developed with significantly improved tolerance to vanadium (this is the basis for selecting a higher nickel/vanadium ratio on cracking catalyst compared to the substrate). MAT data comparing the two-stage process (with high metals on substrate and low metals on cracking catalyst) and the one-stage process (with high metals level on cracking catalyst) are shown in Table X. Although not included as part of this investigation, the use of antimony compounds is expected to minimize the tendency of nickel and vanadium to catalyze dehydrogenation reactions (Dale and McKay, 1977) during pretreatment and cracking. The MAT data on resid pretreatment using low and high
1274
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
Table X. Comparison of Two-and One-Stage Processing: West Texas Resid (A), North Slope Reside (B), and Kuwait (C) Cracking-MAT Data
MAT run no. (description)
M1 (pretreatment)
M2 (pretreatment with high metalsb)
14.38 8.38 44.66 40.97 1.18 96.35
21.06 12.14 29.45 49.49 3.66 93.57
M3 (FCC of pilot plant pretreated 650+ OF Droduct?
total twostaged
M4 (FCC with low metals catalyste)
M5 (FCC with high metals catalystf )
63.11 49.58 23.33 13.56 4.43 105.91
52.29 36.68 41.00 6.71 5.85 96.49
72.06 55.72 19.84 8.10 3.82 104.85
53.72 38.31 32.30 13.98 10.82 94.67
0.89 6.11
2.81 2.45
1.19 2.31
A conversion, vol %g gasoline, vol % furnace oil, vol % slurry oil. vol % coke, wt 70 total C,+ liq, vol % product ratios gasoline/furnace oil furnace oil/slurry oil
MAT run no. (description)
M6 (pretreatment)
M7 (pretreatment with high metalsb)
18.42 10.53 29.37 52.22 2.77 95.78
27.77 14.57 16.36 55.86 6.85 90.47
M8 (FCC of pilot plant pretreated 650+ OF productc)
total twostagec
M9 (FCC with low metals catalvste)
M10 (FCC with high metals catalvstf)
56.14 40.81 22.85 21.02 6.56 99.24
59.13 37.37 29.12 11.74 10.51 90.04
65.62 48.88 20.12 14.27 8.12 102.05
56.92 32.45 23.69 19.39 15.66 85.19
1.28 2.48
2.43 1.41
1.37 1.22
B conversion, vol %g gasoline, vol % furnace oil, vol 7'0 slurry oil, vol % coke, wt 70 total C3+ liq, vol 70 product ratios gasoline/furnace oil furnace oil/slurry oil
MAT run no. (description)
M11 (pretreatment)
M12 (pretreatment with high metalsb)
18.36 9.59 32.05 49.59 3.71 95.43
29.87 14.55 16.87 53.26 5.39 88.82
M13 (FCC of pilot plant pretreated 650+ OF productc)
total twostaged
M14 (FCC with low metals catalyst')
M l 5 (FCC with high metals catalystf )
61.40 45.91 23.64 14.96 5.65 102.41
62.57 39.00 29.46 7.97 8.40 90.10
72.73 52.62 18.13 9.14 7.00 105.02
60.25 33.82 24.29 lij.47 15.69 85.39
1.32 3.70
2.90 1.98
1.39 1.57
C conversion, vol %g gasoline, vol 90 furnace oil, vol % slurry oil, vol % coke. wt % total C3+ liq, vol % product ratios gasoline/furnace oil furnace oil/slurry oil
a Kaiser alumina. *Kaiser alumina with 5000 ppm Ni and 10 ppm V. Pretreatment over metals-free alumina. Calculated by summing yields from M, and M,. e Catalyst Y with 240 ppm Ni and 220 ppm V. 'Catalyst Y with 6000 ppm Ni and 6000 ppm V. g [ 100 - (430 - 650 O F ) furnace oil - (650+ O F ) slurry oil].
metals-laden alumina highlight the effects of nickel and vanadium on yields during pretreatment. Since the substrate does not contain any zeolite, the vanadium is not expected to impact on conversion which is primarily due to thermal means. In fact, the presence of nickel and vanadium actually causes an increase in conversion to gasoline and lighter products plus coke because the extent of dehydrogenation reactions is now comparable to that of cracking reactions. The increased conversion results in higher gasoline, coke, and slurry oil yields, and lower furnace oil yield, in each of the cases investigated. Thus, with the use of metals-loaded alumina, the ratio of furnace oil to slurry oil is not as attractive as that observed with metals-free alumina. The effect of metals on the cracking catalyst for the direct, single-stage processing of residua is markedly different. Catalyst activity is severely impaired by the presence of vanadium, and the dehydrogenation reactions are much more pronounced at the higher levels of nickel and vanadium on catalyst. On the basis of the present set of limited experiments, combined two-stage yields are compared to yields from the
single-stageoperation in Table X. In calculating two-stage yields, the conservative assumption is made that the quality of the slurry oil from pretreatment over metals-free alumina is the same as that from metals-loaded alumina. (Actually, metals loading on substrate is likely to improve the quality of the slurry oil since its yield is increased while contaminant removal effectiveness may be unaffected.) Metals effects during pretreatment should be minimized, however, when passivators such as antimony are used to reduce their hydrogenation activity. Another difficulty with making such a comparison is that a wide range of operating conditions can be employed in the FCC step for both the two-stage and single-stage processes. For example, recycle of slurry oil can be practiced in the single-stage operation to increase the ratio of furnace oil to slurry oil, but this will also contribute to a higher yield of coke. Subject to the limitations of the comparison, the following conclusions can be drawn from the MAT data: combined two-stage yields (with metals-contaminated pretreatment substrate) produce more distillate than single-stagecracking (with or without metals on the catalyst). For the same feedstocks, the combined yield of gasoline
Ind. Eng. Chem. Process
Des. Dev. 1085, 24, 1275-1281
and furnace oil is a t least as high in the two-stage cases (with metals-contaminated pretreatment substrate) as it is for the direct cracking cases. In addition, the two-stage process provides a significant reduction in coke yield and a corresponding increase in total liquid recovery. Conclusions
Petroleum residua from three different crude sources (West Texas, North Slope, and Kuwait crudes) have been upgraded by using a two-stage approach. The first stage involved pretreatment of the residua over a low surface area alumina substrate. At the conditions tested, 88-9670 metals removal, 27-33% sulfur removal, and 61-78% carbon residue removal were achieved. The effect of reaction temperature on pretreatment yields and contaminant removal effectiveness is consistent with typical thermal cracking operations. The second-stage experiments involved conventional catalytic cracking of the pretreated oils. Among the three residua tested, the North Slope resid proved to be the least susceptible to pretreatment and cracking. The effects of nickel and vanadium (deposited on substrate or catalyst) on yields during pretreatment are different from those observed during conventional cracking. The MAT unit is a useful tool for determining yield distributions from residual oil pretreatment and subsequent catalytic cracking of pretreated oils. The two-stage process is an excellent way to maximize the yield of middle distillate boiling range (450-650 O F )
1275
material from the processing of residual oils. Registry No. Ni, 7440-02-0; V, 7440-62-2. L i t e r a t u r e Cited Barthollc, D. 6.; Haseltine, R. P. Oil Gas J. 1981, 79 (22), 89-92. Busch, L. E.; Hetlinger, W. P.; Krock, R. P. OilGas J. 1984, 82(51), 54-56. Campagna, R. J.; Krishna, A. S.; Yanik, S. J. Oil Gas J. 1983, 81 (44), 128-134. Dale, G. H.; McKay, D. L. Hydrocarbon Process, 1977, 9, 97-102. Dean, R. R.; Mauleon, J. L.; Letzsch, W. S.; Legendre, M. Paper presented at the 1983 NPRA Annual Meeting, San Francisco, CA, March 20-22, 1983, AM-83-42. Henderson, D. S.; Ciapetta, F. G. Oil Gas J. 1967, 65 (42), 88-93. Itoh, M.; Suzuki, T.; Tsujlmoto, Y.; Takegami, Y.; Watanabe. Y. Ind. Eng. Chem. Process Des. Dev. 1984, 23, 622-625. Krishna, A. S.;Campagna, R . J.; Engllsh, A. R.; Kowalczyk, D. C. Paper presented at the 1984 NPRA Annual Meeting, San Antonio, TX, March 25-27, 1984, AM-84-51, Mills, G. A.; Boedeker, E. R.; Oblad, A. G. J. Am. Chem. SOC. 1950, 72, 1554. Mitchell, B. R. Ind. Eng. Chem. Prod. Res. Dev. 1980, 19, 209-213. Murphy, J. R.; Treese, S.A. OilGas J. 1979, 77(26), 135-142. Murphy, J. R. Oil Gas J. 1980, 78 (354, 108-110. Rosynek, M. P.; Shipman, G. F.; Yan, T. Y. US. Patent 3 983030, Sept. 28, 1976. Rush, J. 6. Chem. Eng. Prog. 1981, 77(12), 29-32. Teichman, D. P.; Bridge, A. G.; Reed, E. M. Hydrocarbon Process. 1982. 5 , 105-109. Vermillion, W. L.; Gearhart, J. A. Hydrocarbon Process. 1983, 9 , 89-91. Yan, T. Y. Ind. Eng. Chem. Process Des. Dev. 1984, 23, 415-419. Yanik, S. J.; Frayer, T.; Huling, G. P.; Somers, A. E. OilGas J . 1977, 75(20), 139-1 45. Zandona, 0. J.; Busch, L. E.; Hettiger, W. P. Paper presented at the 1982 NPRA Annual Meeting, San Antonio, TX, March 21-23, 1982, AM-82-61.
Received for review August 24, 1984 Revised manuscript receicied January 22, 1985
Coke Formation in the Pyrolysis of n-Hexane Mrltinjoy Pramanlk and Deepak Kunzru’ Department of Chemical Engineering, Indian Institute of Technology, Kanpur, India
Pyrolysis of n-hexane was studied in a jet-stirred mixed reactor in the temperature range of 993-1083 The kinetics of the overall pyrolysis, as well as the coke formation, was investigated. The overall pyrolysis was essentially first order, and the major products were ethylene, methane, propylene, and hydrogen. The effect of temperature, conversion, and molar ratio of steam to hexane on the rates of coke deposition was measured by periodically weighing a small cylinder suspended into the center of the reactor. Coking rates increased with increasing temperature, conversion, and initial partial pressure of n-hexane. The experimental coking rates could be adequately fitted by a model in which ethylene was the coke-forming species.
Pyrolysis of hydrocarbons such as ethane, propane, and naphtha is an important process for the production of olefins. Pyrolysis of any hydrocarbon is always accompanied by coke formation which deposits on the inner walls of the cracking coil. This accumulation of coke with time gradually increases the pressure drop across the tubular reactor and reduces the overall heat-transfer coefficient across the tube wall due to the extra thermal resistance of the deposited coke layer. The normal industrial practice is to periodically shut down the reactor and burn the deposited coke with a mixture of steam and air. Depending upon the feedstock, decoking is done anywhere after 30-60 days. Coke is a complex of carbon and hydrogen which can be formed either from the reactants or products or from both. Coke deposition depends upon several factors, such as the nature of the feedstock, hydrocarbon partial pressure, temperature of the reactor, conversion, presence of 0198-4305/85/1124-1275$01.50/0
sulfur compounds, and the nature of the reactor surface. Studies of the coking phenomena during pyrolysis of pure hydrocarbons should be helpful in understanding the more complex interactions occurring during the pyrolysis of natural feedstocks. However, very meager information is available on the coking kinetics of pure hydrocarbons. Hirt and Palmer (1963) studied the coke formation from methane pyrolysis between 1160 and 1370 K and obtained an activation energy of 432 kJ/mol. Johnson and Anderson (1962) investigated the coke formation from acetylene at temperatures ranging from 770 to 1270 K and observed both the coke and polymer as the products. They suggested that the coke and polymer were formed by two competing parallel reactions in the gas phase. Albright and McConnell(l978) reported that the rate of coke formation in ethane pyrolysis was significantly affected by the material of construction of the pyrolysis tube and also on the pretreatment given to the reactor. Albright and Yu (1978) 0 1985 American
Chemical Society