I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
896
partial deacetylation. For instance, the solubility of a homogeneously deacetylated sample (deacetylated in solution) is different from a heterogeneously deacetylated sample (deacetylated in the solid state to the same acetyl content). This fact is illustrated in Table VI where the same commercial cellulose acetate was: deacetylated homogeneously (dissolved in 95 % aqueous acetic acid) in the presence of small amounts of sulfuric acid a t low teniperatures, and deacetylated heterogeneously in 5% aqueous ammonium hydroxide (a nonsolvent). CONVERSION CURVES. I n the case of the acetone-water ( 9 6 4 ) solvent, a linear relationship was observed between the intrinsic viscosities of homogeneous cellulose acetates having acetyl contents in the range of 38 to 41% and the corresponding basic degrees of polymerization as measured by direct solution in the cupramrnonium hydroxide solvent. This is illustrated in Figure 1,and the 1 elationship is expressed by the equation basic degree of polymerization
=
228 [ s ]
+ 24
(11
where 171 is measured using acetone-water (96--4) as the solvent. I n routine work, it usually is advantageous and far more convenient to make a single viscosity measurement for characterizing the chain length of a sample of cellulose acetate. A conversion curve is given in Figure 2 which makes this possible. This curve permits the conversion of the specific viscosity, measured a t 0.5% concentration, of soluble cellulose acetates containing acetyl contents between 38 and 41% into basic degree of polymerization values-Le., equivalent t o values obtained by direct solution in cuprammonium solvent. In lieu of the conversion curve, the following equation may be used basic degree of polymerization
=
+ 340
398 log 0.5% sap
(2)
where sapis measured using acetone-water (96-4) a8 the solvent. CONCLUSIONS
A practical procedure for measuring the basic degree of polymerization of cellulose acetates, irrespective of acetyl content, by simultaneous solution and complete deacetylation in cuprammonium hydroxide is shown t o have important advantag- and much wider
VoI. 44, No. 4
application than organic solvents. For example, cupramnioniuni solvent may be used t o dissolve mixtures of cellulose and cellulose acetates, or cellulose acetates having very wide sprea&q in acetyl composition. Such mixtures are insoluble or only partially soluble in organic solvents. The use of the same solvent for cellulose and the acetate derivatives of the cellulose permits direct comparison of chain length variations, because a polymer-homologous system is always maintained. The uncertainties and inconvenience of memuring IHEM., 30, 1200-3 (1938). (12) Malm, C. J., Mench, J. W., Iiendall, D. L., and Hiatt, G. D., Ibid., 43, 684-8 (1951). (13) Malm, C. J., Tanghe, 1,. J., and Laird, B. C., Zbid., 38, 77-82 (1946). (14) Staudinger, H., and Daumiller, G., Ann., 529, 219-65 (1937). RECEIVED for review June 8, 1961. ACCEPTED November 21, 1951. Presented before the Divieion of Cellulose Chemistry at the 119th Meeting of the AMERICAN CHEMICAL SOCIETY, Boston, Mass.
Propane Deasphalted Gas Oil as Catalytic Cracking Feed Stock CRACKING OVER A MIXTURE OF NATURAL AND SYNTHETIC CATALYSTS E. CLARENCE ODEN AND THOMAS S . GRANBERRY The Tutwiler Refinery, Cities Service Refining Corp., Lake Charles, La.
HEN average mixed base Gulf Coastal crudes as processed in the Tutwiler Refinery of the Cities Service Refining Corp., Lake Charles, La., are fractionated in an atmospheric tower at coil outlet temperatures ranging from 650'to 740' F., a yield of approximately 15 to 30% crude residuum is obtained, depending upon the particular crude being processed. The residuum may be suitable for lube plant feed stock, depending upon the type of crude from which it was produced, for coking operations, visbreaking, or it can be blended to heavy residual fuel oil. With the price t h a t must be paid for crude oil a t the present time, economics more or less dictate that the residuum yield must be processed for something other than heavy residual fuel oil. If it is not utilized in one of the processes listed above, it can be reduced by installing vacuum flash equipment or propane de-
asphalting equipment to render the major part of this mntriial suitable for catalytic cracking feed stock. At this refinery a settler-type propane deasphalting unit, as previously described by Oden and Foret ( 3 ) ,is used to reduce the Conradson carbon residue content and contaminating effect of the residuum to a minimum. This is necessary since Conradson o weight to coke in the carbon residue apparently goes 1 0 0 ~ by cracking process and fluid catalytic cracking units are generally limited by coke burning capacity. The catalytic cracking of deasphalted gas oil-the valuable heavy gas oil extracted from asphaltic petroleum residues-is a relatively new commercial petroleum process. There are articles in the journals describing commercial propane deasphalting equipment of the tower type used t o prepare catalytic cracking feed stock a t other refineries
April 1952
INDUSTRIAL AND ENGINEERING CHEMISTRY
(4, 6). Economic data are presented comparing methods for catalytic cracker feed stock preparation (I), but no individual yield data are known to have been presented for either the deasphalting or the cracking operation. The purpose of this paper is to present data obtained from cracking propane deasphalted gas oil and a method by which pilot plant correlations on individual feed stocks can be used to predict yields from commercially cracking a mixture of the feed stocks.
TABLE I. TYPICAL INSPECTIONS OF FEEDSTOCKS AND CATALYST Deasphalted Gas
Oil
22.3 6455 685 735 836 866 900 956 65% At 970'
PROCESS EQUIPMENT AND OPERATING PROCEDURES
A simplified process flow diagram of the commercial fluid catalytic cracking unit in which all the deasphalted gas oil produced at this refinery is being cracked a t the present time is shown in Figure 1. It is a downflow-type unit and was first brought ('on stream" in 1944. The unit was designed to give 70% by volume conversion when charging 15,500 barrels per day of virgin gas oil and a slurry return of 1550 barrels per day from the bottom of the fractionator. The regenerator measures 42.5 feet inside diameter and 35 feet between knuckle rings. Combustion air a t a rate of approximately 285,000 pounds per hour is utilized in regenerating the catalyst at approximately l l O O o F. to give a regenerated catalyst normally having a carbon content of approximately 0.7% by weight. A very simple method haa been developed for determining whether the carbon on the spent catalyst is increasing or decreasing b y observing the temperature differential between the reactor bed and a point in the spent catalyst transfer line about 30 feet below where it enters the regenerator. The regenerator bed normally contains approximately 400 tons of catalyst and there are approximately 70 tons in the transfer lines and standpipes. With the 30 tons in the reactor this gives a total catalyst inventory in the unit of about 500 tons. Catalyst stack losses average about 3 tons per day under normal operating conditions. The excess oxygen in the flue gas analyzes between 0.1 and 0.5% by volume and averages about 0.3% by volume. Deasphalted gas oil and virgin light gas oil streams from the deasphalting and crude topping units are pumped into the ct acker's fresh feed tanks at rates which normally give a mixture in the tank containing between 40 and 50% by volume deasphalted gas oil. I n addition, separate lines have been installed through which deasphalted gas oil a t a rate of'approximately 2300 barrels per day is put into the bottom of the fractionator and a varying amount of virgin light gas oil is put into the fresh feed pump suction, The object of both of these lines is to increase the amount of deasphalted gas oil which can be cracked in the unit. I n effect, charging the deasphalted gas oil to the bottom of the
897
...
...
Vir in Lig%t Gas Oil
Rerun Bottoms
34.4
F.
499 520 526 530 533 537 544 550 560 574 618 99,o
0.5
Regenerated Catalyst
28.5 445 454 456 459 463
... ...
... ...
471 523 99.0 0.5 32.5
,..
... ...
...
Viscosity, SUS at 100' F. ... 38.8 ... Viscosity, SUS a t 210° F. 71,3 ... ... ... Flash, F. 330 260 220 ... Pour point, F. 90 0 30 ... Carbon residue, wt. yo 2.15 0.00 0.00 ... Aniline point, F. 224 160 88 ... Diesel index 50 55 25 ... Salt, lb./1000 bbl.
W
0
-20 CATALYST
ACTIVITY
, VOL. %
D+ L
Figure 4. Conversion Correction us. Catalyst Activity
0.5y0by weight carbon on regenerated catalyst as noted thereon. Correction factors for the main operating variables on conversion and yields are presented in Figures 4-6, 8-10) and 12. I n Figures 3, 7, 11, 13, 14, and Table I1 D A G 0 refeis to deasphalted gas oil, VHGO refers t o virgin heavy gas oil, VLGO refers t o virgin light gas oil, HCO refers to heavy aycle oil, and D . I . refers to Diesel index, Diesel index being the product of the API gravity times aniline point, ' F., divided by 100. The Diesel index is used a t this refinery as a n indication of the quality of a cracking feed stock. It is the authors' opinion that the Diesel indexes of cycle oils are much more indicative of their quality as catalytic cracking feed stocks than the Diesel indexes of distillate virgin gas oils. The Diesel index of an oil under consideration as catalytic cracking feed stock should be compared only Kith the Diesel indexes of oils having a similar origin.
899
the cycle stocks vary considerably with the type of feed stock being cracked and the type of catalyst being used. It is believed that the quality of both the light and heavy cycle stocks produced from deasphalted gas oil is inferior to that of the. cycle stocks produced from virgin heavy gas oil at the same conversion. Pilot plant data indicate that natural catalyst produces a higher yield of light cycle stock and a lovi-er yield of heavy cycle stock than synthetic catalyst a t the same conversion. METHODOF CORRELATING COMMERCIAL D a ~ a . The pilot plant correlations just discussed are based on data obtained from cracking each of the feed stocks separately. However, since mixtures of the feed stocks are cracked in the commercial units, it was most desirable to develop a method for predicting yields from cracking various mixtures of the individual feed stocks. In order t o determine whether or not these pilot plant correlations could be used to predict commercial operations, it was necessary t o check them against operations for which special commercial test data had been obtained, inasmuch as the material boiling below 450" F. from all three units is processed in one common recovery unit. The commercial test data used were taken over a 6-hour period of steady operation on the unit. In order to facilitate checking the correlations in this manner the form presented in Table I1 was devised. It will be much easier to understand these calculations if the example worked out in Table I1 is followed closely when reading the calculation procedure outlined below.
03 04 05 06 0 7 CARBON ON REGENERATED CATALYST
.%
Figure 6. Conversion Correction Carbon on Regenerated Catalyst
REACTOR TEMPERATURE, 'F.
Figure 5. Conversion Correction us. Reactor Temperature
It is impossible to present data on the yields of light and heavy cycle stocks as no method has been developed for accurately predicting the split in the cycle stock which is obtained from commercial, operation. The available pilot plant data indicate that the cycle stock yields vary primarily with the conversion level and are not affected to any great extent as to the method employed to obtain the conversion level. Of course, the amount and quality of
VS.
The average values of the primary operating variables which affect the yields are recorded a t the top for convenience, since it will be necessary t o correct from the standard conditions noted on the pilot plant correlations to the actual average operating conditions during the test period. The feed stocks are listed as shown in Table 11. True slurry is considered to be 1.5% by volume of the total feed and t o yield 4001, by weight coke. This assumption on the volume of true slurry is derived from others' experience with slurry settlers. The remainder of the slurry is considered to be of the same quality as the heavy cycle oil stream from that unit. Any straight run gasoline in the feed, or for that matter any material boiling below 430' F. true boiling point, is subtracted from the virgin light gas oil portion of the feed before calculating the conversion and is also subtracted from the gasoline product. Two columns are shon-n for deasphalted gas oil. One is used to record the yields on a carbon residue-free basis on which the pilot plant correlations are made up, and the second column is used for recording the data after correcting to the actual carbon residue content during the test run. In correcting to the actual carbon residue basis it is assumed that 100% of the carbon residue goes directly to coke and that volume per cent and weight pe? cent are interchangeable a t these low values, which are normally below 3%. For example, suppose the carbon residue content was 2%; then the corrected
Vol. 44. No. 4
INDUSTRIAL AND ENGINEERING CHEMISTRY
900
total conversion for deasphalted gas oil on a carbon residue-free basis would be multiplied by 100.0 - 2.0 or 98%, and the 2.0y0 is added in order t o get the corrected total conversion including the carbon residue. The final coke yield is obtained in the same manner. So are all the other yields with the exception t h a t the weight per cent carbon residue is not added to them.
5 4
* =
c'
3a
w 1.0 0 Y
0.5 0.4
a3 a2 10
20
40
30
Figure 7. 20
60
50
CONVERSION ,VOL.
Z
Coke Yield lis. Conversion CATALYST 25
ACTIVITY 30
, VOL Z
925
950
35
DtL 40
45
+04
e 9 r02 2
0
w 0
E
8
O
w
8
-02
actual conversion on total feed obtained on a 100% weight balance basis. If the calculated and the actual conversion vary more than *o.5y0 by volume, the net conversions on the individual feed stocks are corrected t o the actual by multiplying by the ratio of the actual over the calculated in order to get the correct conversion level for predicting the product yields. Using the net corrected conversion previously predicted for each individual feed stock, the catalytic coke yield a t standard conditions is obtained from Figure 7. The catalytic coke yield is defined as the actual coke yield obtained in the pilot plant. Then the catalytic coke yield is corrected to the actual test conditions using Figures 8, 9, and 10. The proper way to use Figure 10, which shows the effect of contamination on the coke, is to read the curve a t the average hydrogen production in cubic feet per barrel converted during the test. The numerical value read from Figure 10 is multiplied by the net conversion expressed as a derimal on that particular stock; Le., if the hydrogen production was 500 cubic feet per barrel converted, then the number read from the curve on Figure 10 would be 2.0, and if the conversion was 50%, then 1.0 weight 7 0 would be added to the catalytic coke yield. It is also necessary to correct the coke yield for the stripping, which is poorer when obtained in the commercial unit than in the pilot plant. This is accomplished by multiplying the catalyst to oil ratio by 0.15. If this figure is examined, it will show that the assumption has been made that each pound of catalyst circulated in the commercial unit carried 0.0015 pound or 0.15% by weight of hydrocarbons, which wm stripped off in the pilot plant, to the regenerator. The coke yields are then totaled for each individual feed stock and multiplied by the portion it represents of the total feed. These weight per cents are then added to get the total weight per cent coke on total feed. If the predicted coke yield does not agree with the actual coke yield, it is corrected in the same manner as the conversion. Using the net corrected conversion for each individual feed stock previously predicted, the dry gas yield at standard conditions is read from Figure 11. This yield is then corrected to the actual conditions during the test using Figure 12. Then the dry
-0.4
875
900 REACTOR
TEMPERATURE
975
1000
,OF.
Figure 8. Coke Correction us. Catalyst Activity and Reactor Temperature
100 200 300 HYDROGEN PRODUCTION
400
500
600
,C.F /Bm. CONVERTED
Figure 10. Contamination Correction Factor c s , Hydrogen Production
01
02 03 04 05 0 6 070809IO CARBON ON REGENERATED CATALYST WT %
,
Figure 9.
.
Coke Correction us. Carbon on Regenerated Catalyst
The conversion of each individual feed stock is predicted using Figure 3 and t h e actual severity factor during the test run. The conversion is then corrected t o the actual operating conditions during the t a t run, rn shown in Table I1 using Figures 4 , 5 , and 6. Then t h e net conversion obtained on each individual feed stock is multiplied by the volume per cent i t represents of the total feed. To get the over-all donversion on total feed, the conversions calculated on the individual components on total feed are added. The conversion calculated in this manner is compared with the
10
20
30 CONVERSION
40 50 VOL %
,
60
Figure 11. Dry Gas us. Conversion
INDUSTRIAL AND ENGINEERING CHEMISTRY
April 1952
901
OF PREDICTED YIELDSWITH ACTUALYIELDS TABLE111. COMPARISON
Teht Number 127 131 137 138 139 141 142 143 144 145 147 148 150 161 152 Average
Conversion, Vol. % Pred. Actual 37.1 40.3 37.6 38.5 31.8 32.4 32.8 32.8 34.5 34.5 33.4 33.4 32,6 34.1 33.4 33.7 30.0 30.6 29.8 30.2 30.8 29.9 33.0 33.8 33.8 35.4 34.3 35.5 31.9 32.5 33.1 33.8
Coke, Wt. % Pred. Actual 6.1 5.6 6.0 5.0 4.7 5.0
4.8 5.2 5.1 4.9 4.9
D r y Gas, Wt. % Pred. Actual
5.1 4.9 5.0 4.8 4.9
a n d Heavier , Vol.' % Pred. Actual 98.2 97.8 97.6 95.9 97.4 97.4 97.6 97.8 97.9 97.1 97.5 97.8 97.9 97.5 98.0 98.3 97.9 97.3 98.0 99.0 97.1 95.8 97.6 97.0 97.6 97.6 98 1 98.3 98,Z 98.8 97.4 97.8
Gasoline, Vol. yo Pred. Actual
Total C4, Vol. 7 0 Pred. Actual
C4
I
3 2
3 5
gas yield on total feed is obtained and corrected, if necessary, in the same manner as the total coke yield. The total butane-butylene yield is obtained from Figure 13, using the net corrected conversion and multiplying the yield for each individual feed stock by the volume per cent it represents on the total feed, and then totaling these results. It is not necessary to correct the total butane-butylene yield to the actual operating conditions, since it was found in pilot plant data that while reactor temperature and catalyst activity had considerable effect on the butane-butylene split, only conversion level and feed stock type and quality appear to affect the total butane-butylene yield appreciably. The gasoline yield at standard conditions is read from Figure 14 using the net corrected conversions previously obtained on the individual feed stocks. It has been found in commercial data that the increase in strippable hydrocarbons appears to come from the gasoline yield. Therefore, in order to keep a 100% weight balance, the net weight per cent corrections to the actual coke and dry gas yields for each individual feed stock are multiplied by 1.15, which is a factor to change from weight per cent coke and dry gas to volume per cent gasoline on feed, and are then applied to the gasoline yield with the opposite algebraic sign. The CCto end-point yield is obtained by adding the total butane-butylene and gasoline yields. The total cycle oil yield is obtained by subtracting the conversion from 100.0. The liquid recovery is obtained by adding the C4 to end point and total cycle oil yields. The resulta from checking 15 process test runs in this manner are shown in Table 111. The following is a brief summary of the deviations of the predicted yields from the actual yields:
5.5
5.7
25.7
25.9
CATALYST ACTIVITY
20
Ed
,V 0 L . I
OIL
25
30
35
40
900
925
950
975
cO.4
6- t o e c
Y
E o u)
3 g
-0.2
0
875
REACTOR TEMPERATURE,nF.
Figure 12. Dry Gas Correction vs. Catalyst Activity and Reactor Temperature I
'
I
I
I 20
30
I
I
I
50
60
Deviation
Max. Conversion, volume % Yields Coke, weight % D r y as w e i g h t % Totay C;, volume % ' Gasoline volume Ch a n d hLavier, v 6 m e %
3.2
Av. 0.7
1.o 1.0 1.5 2.7 1 6
0.3 0.2 0.2 0 4
0
I
Figure
In general, the large deviations were experienced for only one or two of the runs, and as the average deviations were so small, the agreement between the predicted and the actual yields is considered to be very good. CATALYST CONTAMINATION
The curve presented in Figure 10, showing the effect of contamination on the coke yield, is very important in predicting yields for a feed stock, such as deasphalted gas oil, which contaminates the catalyst bed in which it is being cracked. Recent operations a t this refinery indicate fresh catalyst additions, either natural or synthetic, to be more effective in eontrolling the contaminants than the same amount of reject synthetic catalyst from
I 40 CONVERSION
,V O L %
Total Butane-Butylene Yield us. Conversion
13.
another catalytic cracking unit. Based on the data presented in this article, the effect of the contaminants contained in deasphalted gas oil on the catalyst bed decreases as the average catalyst make-up rate for the previous 30-day period is increased, as shown in Figure 15, under the conditions prevailing in this plant. The curve shown in Figure 15 is based on a limited amount of data, especially in the range below 17.5 tons per day of catalyst additions, during the previous 30-day period. Most of the deviation of the points from the curve are due to variations in the type of catalyst-fresh or reject-added during the previous 30-day
INDUSTRIAL AND ENGINEERING CHEMISTRY
902
period, variations in the amount of Kest Texas crude being processed, variations in laboratory anallses for hydrogen, and differences in extent of asphalt i cniova1 iesulting in variations in metal contents The line has h e n curved above 20 tons per day
50 .A
P 5Y 40 W
zJ
30
Vol. 44, No. 4
ual fuel oil yields. However, because of the effect of the contaminants contained in propane deasphalted gas oil it is V P I V desirable from an economic standpoint to process the total production of deasphalted gas oil in one cracking unit if a t all PO+ sible. From the data presented in this paper, it can be seen thatlthe hydrogen production in cubic feet per barrel converted can tie used as a paranwter in determining the extent of contamination and its effect on the product distribution; the increase in coke yield appears to come from the gasoline yield on a weight bail,, and tile most effective method of controlling the effect of the contaminantr contained in deasphalted gas oil appears to be increaiing the catalyst make-up or flushing rate through the unit
2 ?
20
ci 0
"
y1
10
I
I
10
20
1
30
40 CONVERSION, VOL "h
I
I
50
60
Figure 14. Gasoline Yield TS. Conversion
of c:i(:tlyst make-up as indicatcd, since the hydrogen produrtioii would not be expected to go to zero at a catalyst make up rate of 50 to 55 tons per day, although a straight line extrapolated through the data would indicate this. The effect of the containinants contained in deasphalted gas oil is very expensive and therefore should be closely controlled. There is the adverse selectivity effect, which is t>heincrease in coke ?-ield a t the expense of the gasoline yield, and the capacity effect. If a catalytic ciarliing unit is limited by the amount of coke that can be burned, as is usually the case, it can be shown that its throughput and, consequently, conversion capacity could easily be reduced by 20 to 25% owing to excessive catalyst contamination, Another effect of contamination is t,hat the large amount of hydrogen produced may cause a decrease in t h e recovery of butanes and lighter hydrocarbons because of the partial pressure effect of hydrogen in the absorber or fractionat,ion equipment. For these reasons el-ery effort is made a t this refinery to hold t'he hydrogen production on the unit cracking deasphalted gas oil between 300 and 400 cubic feet p1.r barrel converted. COhc LU SION s
The method of using pilot plnrit corrclations on individual feed stocks present,ed herein can be succesfully applied to predict conimercial operation in this refin n-hen cracking various niixtures of these feed stocks. The Conradson carbon residue content of the deasphalted gas oil is converted 100% to coke during the cracking process. I t is both economically feasible and desirable to catalj-ticidly crack propane deasphalted gas oil in ortle1, to reduce heavy resid-
200 HYDROGEN
Figure
300 400 500 600 PRODUCTION C.F /BEL. CONVERTED
,
15.
Catalyst Make-rzp Rate c's. Hydrogen Production ACKNOWLEDGMENT
The authors wish to express their appreciation to Cities Servic,+. Refining Corp. for permission to publish this paper and to rec'ognize the cooperation of R. S. Hays and C. R. Grisuoll of thv operat,ing department of this corporation. The indispensakilr. assist,ance of Charles E. Smith, Jr. and Ehner h-.Coultw i n preparing the graphs is also gratefully acknowledged. LITERATURE CITED
(1) Jemell, J. W., Jr., and Connor, J. P., Petrole?tmProcessiiiO, 5, So. 11, 1199-1201 (1950). ( 2 ) Mills, G. A., IND.ENG.CHEM.,42, S o . 1, 182-7 (1950). (3) Oden, E. C., and Foret, E. L., Ibid., 42, No. 10, 2088-95 (19501 (4) Petroleum Refiner, 28, No. 4, 126-9 (1949). ( 5 ) Weber, George, Oil Gas J . , 49, No. 5, 54-6 (19501. RECEIVED for review December 15, 1950. A C C E P T E D Soveinber 19, 1951. Presented at t h o Sixth Southwest Regional Meeting of t h e I s r a a r c . t x CHI:,IICALS o C I E , r Y , San dntonio, T e x . , December 7-9, 1950.
