Energy & Fuels 1999, 13, 655-666
655
Relating Feedstock Composition to Product Slate and Composition in Catalytic Cracking. 5. Feedstocks Derived from Lagomedio, a Venezuelan Crude† J. B. Green,* E. J. Zagula, R. D. Grigsby, J. W. Reynolds, L. L. Young, T. B. McWilliams, J. A. Green, and H. Chew BDM Petroleum Technologies, P.O. Box 2543, Bartlesville, Oklahoma 74005 Received September 25, 1998
The fluid catalytic cracking (FCC) behavior of compound types present in the >650 °F resid from Lagomedio crude was investigated. Distillation and liquid chromatography were employed for separation of selected compound type fractions from the resid; the resulting fractions were then cracked using a bench-scale FCC unit. The FCC behavior for each compound type was defined in terms of the resulting product distribution (yields of gas, gasoline, etc.); sulfur, nitrogen, nickel, and vanadium partitioning; and in selected cases, gasoline composition. Results obtained from Lagomedio fractions were compared to those obtained from earlier FCC studies of compound types from Wilmington, CA, Maya (Mexican), and Brass River (Nigerian) >650 °F resids. An equation is presented for prediction of gasoline yield for feedstocks derived from Lagomedio as well as other crudes. Gasoline yield (wt % of feed) is calculated from these feedstock parameters: the atomic ratio of hydrogen divided by carbon plus sulfur (H/(C + S)), the fraction of the feed volatilized at the cracking temperature fcT, effective metal content (Meff ) Ni + V/4) expressed in µequiv/g, basic nitrogen (NB, wt %) and amide-type nitrogen (NAm, wt %). The relation is G ) 10.25[H/(C + S) + log(fcT)] - 1.5Meff - 29(NB + NAm) + 30.2. Calculated and experimental gasoline yields typically agreed within 1.5 wt %, which is the pooled standard deviation for the experimental data. An infrared spectrophotometric method is provided for determination of NAm; the other parameters are measured using standard methods or previously published procedures.
Introduction This paper is the fifth in a series resulting from a program aimed at relating feedstock composition to product slate and composition from fluid catalytic cracking (FCC) processes. This information is being sought in an effort to improve FCC processing of conventional feedstocks and to extend FCC to lower quality feedstocks, such as atmospheric resids from heavy oils. The first three papers presented results from feedstocks derived from Wilmington (CA), Brass River (Nigeria), and Maya (Mexico) crudes; the fourth dealt with a model for calculating nine product subclasses from mass spectral analysis of hydrocarbon and sulfur types in the feed.1-4 Several equations describing simple feed/product correlations have been introduced in these earlier papers. For continuity, they are listed below. Maximum gasoline yield may be estimated from the proportions of neutral (largely hydrocarbons and sulfur compounds) components in the feed (fN) times the anticipated gasoline yield from pure neutrals (GN): (1) Green, J. B.; Zagula, E. J.; Reynolds, J. W.; Wandke, H. H.; Young, L. L.; Chew, H. Energy Fuels 1994, 8, 856-867. (2) Green, J. B.; Zagula, E. J.; Reynolds, J. W.; Young, L. L.; Chew, H.; McWilliams, T. B.; Grigsby, R. D. Energy Fuels 1996, 10, 450462. (3) Green, J. B.; Zagula, E. J.; Reynolds, J. W.; Young, L. L.; McWilliams, T. B.; Green, J. A. Energy Fuels 1997, 11, 46-60. (4) Sheppard, C. M.; Green, J. B.; Vanderveen, J. W. Energy Fuels 1998, 12, 320-328.
Gcalcd ) GNfN
(1)
GN can be determined experimentally or estimated via the following relation:
GN ) 10.25[H/(C + S) + log(fcT)] + 28.8
(2)
where H/(C + S) is the atomic ratio of those elements and fcT is the fraction of neutrals boiling below the cracking temperature. Gasoline yield from nonneutral types (i.e., acidic or basic) may be compared to that of the neutrals by means of the NEGY (neutral equivalent gasoline yield) parameter defined in eq 3
NEGY ) (Gmeasd - Gcalcd)/GNfA/B
(3)
where Gmeasd is the actual gasoline yield and fA/B is the weight fraction of acids/bases in the feed. An early correlation for total coke yield, CΣ, was based on feed microcarbon residue (MCR, ASTM D 4530), the catalytic coke formed from a nitrogen-free feedstock (C0), the nitrogen content of the feed (N, wt %) and the fraction of feed nitrogen incorporated into coke (χCN).
CΣ ) MCR + C0(1 + NχCN)
(4)
Subsequently, an improved coke correlation was developed based on MCR, basic nitrogen (NB, wt %), hydrogen
10.1021/ef980190n CCC: $18.00 © 1999 American Chemical Society Published on Web 03/19/1999
656 Energy & Fuels, Vol. 13, No. 3, 1999
(HW, wt %), sulfur (SW, wt %), and effective metals (Meff, µequiv/g) in the feed.5 In that case, the expression for total coke (CT) is shown below.
Green et al.
known behavior to obtain an estimate of its overall impact on the yield of gasoline and other light products. Experimental Section
CT ) MCR + 0.66HWe-NB + Meff + 25NBe-(NB/(NB+0.437SW)) (5) For comparison, total coke (CΣ and CT) is calculated here using both of these equations for Lagomedio-based feedstocks. Hydrocarbon-type data for Lagomedio were incorporated into a previous correlation for calculating FCC product subclasses from mass spectrometric analysis of feeds.4 Limited compositional data for the >950 °F Lagomedio resid were also reported in conjunction with a study of coking tendencies of compound classes present in high-boiling petroleum fractions.6 The crude itself is produced in the Lake Maracaibo region of Venezuela and is listed as having an API gravity of 31.5 and a sulfur content of 1.17 wt %.7 A literature search did not turn up significant additional information on the detailed composition or process characteristics of this crude. A major focus of this paper is on prediction of gasoline yield from Lagomedio atmospheric resid (>650 °F) and other feeds containing appreciable levels of heteroatomic (N, O, S, Ni, V) compounds. Although heteroatomic compounds, particularly basic nitrogen types, are wellknown catalyst poisons, most FCC models treat their effects in only a cursory fashion. Although this approach may be adequate for conventional gas oil feeds with low levels of heteroatoms, it is clearly not for many residual feedstocks which often contain an order of magnitude higher levels of heteroatomic species. The high degree of interaction between heteroatomic species and catalyst surfaces has been previously borne out by analysis of spent catalysts,8 studies with pure compounds,9-13 as well as experiments involving heteroatom-rich petroleumbased feeds.14-20 The problem is in how one relates this (5) Green, J. B.; Green, J. A.; Young, L. L. Prepr.sAm. Chem. Soc., Div. Pet. Chem. 1998, 43, 623-627. (6) Green, J. B.; Shay, J. Y.; Reynolds, J. W.; Green, J. A.; Young, L. L.; White, M. E. Energy Fuels 1992, 6, 836-844. (7) Aaland, L. R. Oil Gas J. 1983, Oct. 24, 88-90. (8) Qian, K.; Tomczak, D. C.; Rakiewicz, E. F.; Harding, R. H.; Yaluris, G.; Cheng, W.-C.; Zhao, X.; Peters, A. W. Energy Fuels 1997, 11, 596-601. (9) Hughes, R.; Hutchings, G. J.; Koon, C. L.; McGhee, B.; Snape, C. E.; Yu, D. Prepr.sAm. Chem. Soc., Div. Pet. Chem. 1995, 40, 413417. (10) Hughes, R.; Hutchings, G. J.; Koon, C. L.; McGhee, B.; Snape, C. E. Prepr.sAm. Chem. Soc., Div. Pet. Chem. 1994, 39, 379-383. (11) Corma, A.; Fornes, V.; Monton, J. B.; Orchilles, A. V. Ind. Eng. Chem. Res. 1987, 26, 882-886. (12) Ho, T. C.; Katritzky, A. R.; Cato, S. J. Ind. Eng. Chem. Res. 1992, 31, 1589-1597. (13) Fu, C.-M.; Schaffer, A. M. Ind. Eng. Chem. Prod. Res. Dev. 1985, 24, 68-75. (14) Cimbalo, R. N.; Foster, R. L.; Wachtel, S. J. Oil Gas J. 1972, May 15, 112-122. (15) Scherzer, J.; McArthur, D. P. Ind. Eng. Chem. Res. 1988, 27, 1571-1576. (16) Ng, S. H.; Rahimi, P. M. Energy Fuels 1991, 5, 595-601. (17) Harding, R. H.; Zhao, X.; Qian, K.; Rajagopalan, K.; Cheng, W.C. Ind. Eng. Chem. Res. 1996, 35, 2561-2569. (18) Larocca, M.; Farag, H.; Ng, S. H.; deLasa, H. Ind. Eng. Chem. Res. 1990, 29, 2181-2191. (19) Green, J. B.; McWilliams, T. B.; Sturm, G. P., Jr. Prepr.sAm. Chem. Soc., Div. Pet. Chem. 1995, 40, 681-684. (20) Green, J. B.; Zagula, E. J.; Young, L. L.; Sturm, G. P., Jr. Prepr.sAm. Chem. Soc., Div. Pet. Chem. 1995, 40, 691-694.
The scheme and methodology for crude fractionation and blending of the fractions to obtain FCC feedstocks were the same as those used earlier.1 Briefly, the 650-950 and >950 °F boiling ranges of the crude were fractionated into nine fractions using liquid chromatography (LC). These fractions were used singly or in combination as feedstocks to a benchscale (approximately 4 g of oil and 35 g of catalyst per charge) fluidized unit. Feedstocks were cracked at 521 °C (970 °F) using a Davison XP series equilibrium catalyst at a cat./oil ratio of (8.5 ( 0.5)/1. The feedstocks were charged to the unit over a 30 s time period. The sweep gas (N2) velocity was set such that the catalyst volume was displaced approximately every 20 s. The FCC unit, catalyst, and procedures for carrying out cracking experiments were unchanged from the prior work,1 except for refinements added later.2 Good correlation of results obtained using this methodology with those from commercial-scale FCC units has been reported.1 Infrared spectrophotometry was used to determine amide forms of nitrogen in acid fractions. Whole acid fractions were fractionated on basic alumina prior to infrared analysis. Thus, a known weight of the acid fraction, near 25 mg, was dissolved in approximately 1 mL of dichloromethane and charged to a glass column (1 (i.d.) × 10 cm) containing about 6 g of basic alumina (Aldrich) which had been preequilibrated with dichloromethane. Weak acids (predominately carbazole types) were eluted with 15 mL of dichloromethane. Amides were subsequently eluted with 15 mL of 1:1 dichloromethane:methanol. The amide fraction was evaporated to dryness and diluted with a known volume of dichloromethane (usually 1 or 2 mL), and its infrared spectrum was recorded using a conventional 0.5 mm path length, KBr liquid sampling cells. The absorbances of the five-membered ring lactams (ca. 1710 cm-1) and sixmembered ring lactams (ca. 1660 cm-1) were summed and the wt % of amide nitrogen (NAm) was calculated as follows
NAm ) (14 × Abs × 100%)/abc
(6)
where 14 ) atomic weight of nitrogen, a ) molar absorptivity (400 L mole-1 cm-1), b ) path length (ca. 0.05 cm), c ) concentration (mg of acid fraction/mL). The above separation procedure was based on the work of Copelin21 as well as Snyder and Buell.22 The assumed value for molar absorptivity was based on the absorbance of amides isolated from Maya 650930 °F acids.3 This method could also be applied to base fractions as well as whole distillates/resids with minor adjustments in quantites charged to the Al2O3 column and, particularly for base fractions, wavenumber for determination of absorbance. Basic nitrogen (NB) was determined via nonaqueous titration as described previously;5 concentrations of basic and amide forms of nitrogen as well as of Ni and V were used to predict gasoline yields, as discussed below. The procedure for determining Ni and V in feeds and products was described in the third paper of this series.3 A recent change in the software for calculating material balances caused data reported here for gaseous components to generally be biased toward higher levels of C3 and C4 components compared to prior results. Earlier, any proportion of liquid product boiling below 80 °F (26.7 °C) (typically 0-2% of the whole liquid) was mathematically distributed over all gaseous product yields. However, in this work, any detectable proportions of liquids boiling below 80°F were distributed over C3 and C4 hydrocarbon yields exclusively. This change was (21) Copelin, E. C. Anal. Chem. 1964, 36, 2274-2277. (22) Snyder, L. R.; Buell, B. E. Anal. Chem. 1968, 40, 1295-1302.
Feedstocks Derived from Lagomedio
Energy & Fuels, Vol. 13, No. 3, 1999 657
high quality crude, data for its >950 °F acid/base fractions are in fact quite comparable to those of Maya3 or other poor quality crudes. The main difference is that Lagomedio contains substantially lower levels of compounds categorized as strong or weak acids by the LC separation technique. In addition, Lagomedio >950 °F neutrals contain significantly lower levels of sulfur and metals compared to Maya neutrals, and they exhibit a lower MCR. Table 3 lists the hydrocarbon-type distribution for Lagomedio neutrals determined by high-resolution mass spectrometry (MS).23 The predicted product slate from cracking pure neutrals is also shown in the table. The algorithm originally described for prediction of products3 was modified slightly in this case to compensate for the change in material balance calculations described in the Experimental Section. Specifically, the assumed relative product distribution for dibenzothiophene types was changed from 50% heavy cycle oil (HCO)/50% coke + light gas to 50% HCO/25% coke + light gas/25% C3/C4 gas + gasoline. With this slight modification, good agreement in predicted versus actual (see bottom of table) product slate was achieved. Table 4 lists the proportions of Lagomedio fractions blended and the corresponding elemental/MCR data for feedstocks used in FCC experiments. As noted in the table, two of the feeds were comprised of acid/base fractions from Wilmington 650-1000 °F distillate blended with Lagomedio >650 °F neutrals. Properties of feeds were calculated from those of the individual LC fractions listed in Table 2. Whole >650 °F neutrals and >650 °F polar/sulfide-free neutrals were blended from the respective 650-950 and >950 °F fractions according to their proportions in the whole >650 °F resid. FCC Product Distributions. Table 5 details the FCC results for each feedstock. In general, yields of gasoline approached those of the corresponding feeds derived from Brass River >650 °F resid and were significantly above those from Wilmington or Maya feeds. In addition, there was good agreement between yield data from Lagomedio 650-950 °F acids versus Wilmington 650-1000 °F acids blended with Lagomedio neutrals (feeds 6 and 7), as well as between corresponding blends of Lagomedio versus Wilmington distillate bases (feeds 10 and 11). The feeds exhibiting the lowest conversion and gasoline yield were those containing the distillate base fractions. Sulfur partitioning to H2S was relatively high for most feeds, as observed previously.1-3 Similarly, nitrogen partitioning to liquid products was appreciably higher for feeds enriched in distillate acids
Table 1. Compound-Type Distribution in Lagomedio as a Function of Boiling Range (wt %)a sample no. 3293
3064
boiling range, °F 650-950 >950 acids strong 8.1 ( 0.3 weak 6.3 ( 0.4 total 3.5 ( 0.1c 14.4 ( 0.5 bases strong 5.2 ( 0.1 weak 6.5 ( 0.7 total 1.0 ( 0.1 11.7 ( 0.7 neutrals 95.5 73.5 ( 0.3 polar-neutral 0.6 ( 0.1 2.8 ( 1.0 sulfide 6.3 ( 0.3 8.9 ( 0.2 nonsulfide 84.7 ( 0.3 60.0 ( 1.2 total 91.6 ( 0.4 71.7 ( 1.6 total 100.0 ( 0.5 99.6 ( 1.8
3293/3064 >650b 5.2d 4.0d 10.5e 3.3d 4.1d 7.8e 81.5 2.0 8.0 68.9 78.9 99.8
a From LC mass balances. b Calculated from data for 650-950 (36.2 wt % of >650 °F) and >950 °F (63.8 wt % of >650 °F) boiling ranges. c Uncertainties given are average deviations from 2 to 3 separations. d Includes >950 °F portion only. e Total >650 °F acids or bases.
implemented in response to cumulative experience in detailed liquid analyses which indicated that the bulk of the carryover of gases into liquids consisted primarily of C3/C4 components.
Results Feedstock Composition. Table 1 lists the distributions of compound types determined from LC separation of the 650-950 and >950 °F boiling ranges of Lagomedio crude. Those data were used, in turn, to calculate the distribution of compound types for the >650 °F resid indicated in the table. The 950 °F cut point cited for the two boiling ranges was the crossover point for their GC-simulated distillation profiles. In fact, there was considerable overlap in their distillation curves. As noted in the table, the relative proportion of 650-950 °F material in the >650 °F resid was only 36.2 wt %. For comparison, the proportions of distillate in >650 °F resids from Brass River, Wilmington, and Maya crudes were 93.0, 34.3, and 22.3 wt %, respectively. The proportion of neutrals in >650 °F Lagomedio resid is intermediate between the extremes of Brass River (93.1 wt %) and Maya (63.5 wt %). The greater proportion of total acidic compounds compared to basic types evident for Lagomedio also held true for >650 °F resids from the other three crudes. Table 2 shows elemental and MCR data for LC fractions and the corresponding whole materials. Although Lagomedio is generally perceived to be a fairly
Table 2. MCR and Elemental Data for Lagomedio Fractions wt % boiling range, °F 650-950
>950
ppm (w/w)
fraction
C
H
N
S
Total
MCR
Ni
V
whole distillate acids bases neutrals polar/sulfide-free neutrals whole resid strong acids weak acids strong bases weak bases neutrals polar/sulfide-free neutrals
85.6 76.5 82.3 85.8 86.4 84.5 82.1 83.5 84.1 83.3 85.3 85.6
12.55 8.72 9.76 12.76 12.85 11.43 8.16 9.16 8.60 9.99 11.94 12.42
0.13 2.38 2.86 0.007 0.006 0.48 1.26 1.26 1.78 1.04 0.142 0.061
1.85 0.56 1.57 1.88 1.26 2.64 3.31 2.94 3.66 3.21 2.26 1.61
100.1 88.2 96.5 100.4 100.5 99.1 94.8 96.9 98.1 97.5 99.6 99.7
0.27 7.5 7.2 650 °F neutrals >650 °F neutrals >650 °F neutrals >650 °F neutrals >650 °F neutrals >650 °F neutrals >650 °F neutrals
wt % in feed
wt % in feed
component B
100 100 100 85.0 85.0 85.0 84.7 85.0 85.0 85.0 84.2
>950 °F strong acids >950 °F weak acids 650-950 °F acids 650-1000 °F acids from Wilmington crude oila >950 °F strong bases >950 °F weak bases 650-950 °F bases 650-1000 °F bases from Wilmington crude oila
elemental composition, wt % feed
metal content, ppm (w/w)
total
MCR wt %
Ni
V
85.5 11.91 0.35 2.35 100.1 85.5 12.36 0.106 2.10 100.1 85.9 12.61 0.046 1.59 100.1
9.4 3.9 2.9
23 3.3 2.3
230 29 14
C
H
N
S
15.0 15.0 15.0 15.3
85.0 85.2 84.2 85.0
11.73 11.88 11.81 11.91
0.28 0.28 0.45 0.36
2.28 2.23 1.87 1.93
99.3 99.6 98.3 99.2
10.1 8.4 4.4 4.0
28 10 3.4 2.9
266 121 27 25
15.0 15.0 15.0 15.8
85.3 85.2 85.1 85.4
11.80 12.00 11.97 11.95
0.36 0.25 0.52 0.51
2.33 2.27 2.02 2.01
99.8 99.7 99.6 99.9
10.3 7.5 4.4 3.6
28 18 3.5 3.1
208 127 25 25
See ref 1.
or bases. As with the other crudes, relative carryover of nitrogen into liquids was highest for feeds enriched in distillate acids. Carryover of metals (Ni, V) into product liquids has been determined previously only for Maya-based feeds. However, in that case, as with Lagomedio feeds, the HCO from the neutrals + weak base feedstock (no. 9) contained the highest metal concentrations. This effect has tentatively been attributed to the prevalence of porphyrinic forms of metals in the >950 °F weak base fraction.3 Comparison of calculated (eq 1) versus measured gasoline yields reveals that acid/base components produce measurable levels of gasoline, except perhaps for Wilmington 650-1000 °F bases. The related NEGY parameter (eq 3) expresses the gasoline production of acid, base, or neutral sulfide fractions relative to that (23) Teeter, R. M. Mass Spectrom. Rev. 1985, 4, 123-143.
of neutrals. As shown in Table 5, several fractions are capable of providing up to 80% (NEGY ) 0.7-0.8) of the gasoline yield from neutrals. Coke production calculated using the older correlation (eq 4) tends to run below the measured values, particularly for the whole resid, neutral + strong acid, and neutral + distillate base feeds (nos. 1, 4, 10, and 11). The newer correlation (eq 5) provides coke values within 1.5 wt % absolute in all cases and within 1 wt % for most feeds. Coke data from eq 5 also tend to be more randomly distributed around the experimental data than those from eq 4. Gasoline Composition. Table 6 provides a summary of the gasoline composition for seven feedstocks. Detailed gasoline compositions are supplied as supplemental data to this paper. Gasoline produced from feeds containing distillate (650-1000 or 650-950 °F) acids or bases contained lower proportions of isoparaffins and higher (except for feed no. 6, neutrals + Lagomedio
Feedstocks Derived from Lagomedio
Energy & Fuels, Vol. 13, No. 3, 1999 659
Table 5. Overall Product Distributions from Lagomedio Feedstocks Obtained at 521 ( 1 °C (970 °F) and a Cat/Oil Ratio of 8.5 ( 0.5a nitrogen partitioningi
wt % feed feedb 1. WR 2. N 3. N - P/S 4. N + SA 5. N + WA 6. N + DA 7. LagN + Wil DA 8. N + SB 9. N + WB 10. N + DB 11. LagN + Wil DB
gasc gasolined LCOe HCOf coke
total
conv.g Geffh liquid
HCO metal content ppm (w/w)
sulfur partitioningj
coke
gas
liquid coke
total
Ni
V
17.8 16.1 20.0 15.9 16.3 15.2 16.9
40.0 43.4 44.8 38.2 41.9 41.1 41.5
11.9 14.7 12.7 13.3 14.7 15.1 13.2
8.3 14.0 10.6 12.3 11.0 15.3 14.0
21.2 99.2 11.8 100.0 11.8 99.9 19.8 99.5 16.2 100.1 13.2 99.9 13.5 99.1
79.0 71.0 76.6 73.8 74.5 69.5 71.9
50.7 61.2 58.5 51.9 56.2 59.1 57.7
5.5 8.0 6.6 5.1 11.7 31.9 26.4
94.5 92.0 93.4 94.9 88.3 68.1 73.6
64.2 62.8 63.5 55.3 61.8 57.2 55.8
19.6 30.5 29.8 21.8 27.2 32.5 32.5
13.9 97.7 6.7 100.0 5.1 98.4 18.9 96.0 9.1 98.1 7.5 97.2 10.1 98.4
