INDUSTRIAL AND ENGINEERING CHEMISTRY
August 1949
1717
and the densities shown in column 4, Table I, of this comment TABLE I. IDENTIFICATION OF CORROSION PRODUCTS BY DENSITY were computed. The compounds identified as corrosion products MEASUREJIER’TS~
Average N a t u r e of Thickness y’ - zj, G. Density of Surface per 100 of Film. Film, X-Ray Test Layers, p Sq. Cm. G. per 1\11, Analysis H u n t Stewart 4.2 0.162 3.86 FeO, FeCOa Hunt Stexartb 6.7 0.244 3.64 FeCOa 7 6 0.294 3.87 FeO, FeCOa Grant Grantb 9.9 0.378 3.82 FeCOa Cartilage Upper Pettit 6.2 0.234 3.77 PeCOa Carthag: Lower Pettit 3.3 0.126 3.82 FeO, FesOl Tullos h o . 1 4.6 0.174 3.78 FeCOs. FepOj .Jones No. 1 7 0 0 260 3 71 33-6 b 3 0 0 111 3 70 FeCOa, FeO, FesOa a Density.from literature: FeC03, 3.7 t o 3.9 g./ml.; FeO, 5.7 g./ml.; F?30i, 4.96 t o 6.40 &,’nil.: and FeS, 4.84 n./nil. b Inhibited.
where h d y
= thickness of the surface layer, p = density of corrosion products, grams per ml.
y’
=
=
weight loss of coupons as removed, grams per 100 sq. em. weight loss of coupons after descaling, grams PPI 100 sq. em.
Equation 1 was simplified as follows: d
=
100(,~’- y ) / h
Cracking of
by x-ray analysis are shown in column 5. Footnotes in Table I show the range of density values given in the literature for these products. Except for data obtained in the Carthage Lower Pettit tests, the data of Table I indicate close agreement between the measured values of corrosion product density and the literature values for products identified by x-ray analysis. Whether or not such close agreement can always be expected could best be determined by further testing. Assuming that film thickness nieasurements made by means of the microscope would not be more accurate than =tO.3,uL,the uncertainty reflected in most of the density computations would be 3 to 7 7 5 , The differences in thc densities of ferrous carbonatc, ferrous oxide, and ferrosoferric oxide range from 6 to 25%. Computed values of density would not, of course, show the respective amounts of each of three 01 more compounds present in a corrosion product but might distinguish between the oxides and ferrous carbonate. LITERATURE CITED
IKD. ENG.CHEM., 41,1712 (1949). (2) Hackerman, Norman, and Shock, D. A, Ibid., 39, 863-7 (1947). (1) Hackerman, Norman, and Schmidt, H. R.,
RECEIVED February 17, 1949.
igh Sulfur
USE OF STEAM WITH NATURAL CATALYST A. L. CONN AND C. W. BRACKIN, Standard
In the catalytic cracking of high sulfur stocks with natural catalyst, the catalyst rapidly becomes poisoned, resulting in excessively high coke and gas yields at the expense of gasoline yield. Data are presented demonstrating the successful use of steam in solving this probIsm. In one set of experiments, carried out in a 2-barrelper-day fluid pilot plant, the introduction of steam to the regenerated catalyst standpipe serves to hydrate the catal3st prior to contact with oil, thus preventing loss of catalyst selectivity. In another set of experiments, including one full-scale plant test, the use of large quantities of stripping steam in addition to the hydration steam malies it possible to restore poisoned catalyst to normal selectivit,. I t is concluded that the rate of sulfur poisoning is dependent upon the extent to which the catalqst is hydrated at the time it contacts the oil, upon the concentration of sulfur in the feed, and upon the extent to which the poison is eliminated by replacement with steam as the cataljst passes through the stripper.
T
HE increase in demand for petroleum products has led to the
utilization of increasing amounts of high-sulfur crudes. Furthermore, in order to produce the required amounts of distillate fuels and a t the same time maximize gasoline production, it has been necessary to charge higher boiling gas oils to catalytic cracking units. Because the higher boiling fractions generally contain larger proportions of sulfur ( 1 ) both the crude source and the boiling range have combined to increase the amount of sulfur in catalytic crncking charge stocks. Under certain conditions, natural catalyst has an advantage over synthetic silica-alumina catalyst in the production of motor gasoline by catalytic cracking; hence, a number of catalytic crack-
Oil Company (Indiana), Whiting, Ind.
ing units have employed natural catalyst. I t has been found, however ( 2 , S ) , that the presence of large amounts of sulfur in tl e charge stock has a poisoning effect on natural catalyst. causing rapid deactivation and loss of selectivity. As an example of catalyst poisoning it was observed in one fluid cracking unit that, after a few weeks of operation, a progressive increase in coke and gas yields occurred at the expense of gasoline vield. Tests on the catalyst, using standard fixed-bed activity-testing methods ( 5 ) )indicated that the changes in yields weie the result of a change in the selectivity of the catalyst. For example, the carbon factor-the ratio of the carbon yield obtained with the catalyst tested to the carbon yield obtained with a standard catalyst at the same conversion of gas oil-was greater than 2.0. Similarly, the gas factor, which is derived in the same manner, was greater than 1.6. I n order to permit satisfactory operation with natural catalyst and high-sulfur stocks, tests were carried out t o determine means of preventing catalyst poisoning and methods LTere sought t o rejuvenate the catalyst after I t had become poisoned. I n small-scale test work reported bv Davidson (9), in which the operating conditions approximated those in catalytic cracking, the ability of natural catalyst to sorb moisture a t temperatures between 800 and 1060’ F. was studied. This moisture is believed to become associated with the cr)stal structure of the catalvst. Cracking tests, made in a small fixed-bcd unit with a high-sulfur stock ( g ) , demonstrated that hydration of the natural catalyst prior to contact with the oil feed (pichydration) greatlv reduced the rate of catalyst dctcrioration. In the current Tvork, preliminary results of which were reported bv Davidson, experiments confirming this observation were carried out in a small fluid catalytic pilot plant. In these experiments, the steam wae injectrd a t the base of the regenerated catalyst standpipe. Since that time, additional experiments have been carried out on the use of steam
INDUSTRIAL AND ENGINEERING CHEMISTRY
1718
Vol. 41, No. 8 USE OF HYDRATION
HYDRATION TESTWITH COKE-STILL GAS OIL^ TABLE I. STEAM
STEAM
( R u n A, feed 1)
The results of a pilot pIant experiment demonstrating that. hydration steam alone is effective in retarding catalyst deterioration are presented in Table I and Figure 3.
Test No. A-1 A-2 A-3 A-4 A-5 9-6 Hours on stream 0 4-16 16-28 28-40 40-52 52-64 92-104 Operating conditions Hydration steam, u-t. % on catalyst ... 0.54 0.51 0.52 0.63 0.54 0 Dispersion steam, wt. % on catalyst 0 0 0 0 0 0 0.54 0.59 0.52 0.53 Stri ping steam, wt. % on catalyst ., 0.54 0.51 C a r f o n on regenerated catalyst, ivt. % .. .. ., 0.13 0.12 0.12 0.13 0.14 0.19 1.21 1.28 1.30 1.42 1.20 1.19 Carbon on suent catalyst, wt. % Product yields (output basis) Coke, wt. 70 . . 5.91 6.35 6.12 6.53 6.42 7.11 Gas oil product, vol. .., 47.9 47.8 48.9 47.8 49.7 55.1 Gasoline (10-lb. R . v . ~ . ,400O F. e . p . ) , vdl. Slob .,, 42.2 42.3 42.4 4?,! 41.0 34.4 Exccss butanes, vol. yo ... 5.2 4.9 3.5 3.3 2.9 2.8 D r y gas, wt. c;C ... 6.1 5.9 5.9 6.3 6.7 5.9 D r y gas, s.c.f./bbl. chargeb ... 300 313 342 365 406 486 Coke a t 45% conversion, wt. % ., , 5.00 5,34 5.42 5.50 5.97 8.82 Conversion (naphtha-free basis), voI. % ... 48.4 48.6 47.3 48.5 46.4 40.6 Catalyst inspections 21.4 16.8 24.6 23.2 ... ... .. . Activity (Indiana relative) 1.19 1.33 ... ... ... 1.58 2.92 Carbon factor 1.35 1.71 1.22 1.26 . ... Gas factor a Reaction conditions: 900’ F., 10 Ib.jscl. inch gage, 1.9 weight space velocity, a n d about 5 catalyst-to-oil ratio; fitripping temperature, 90Oo F., no dispersion steam, a n d regenerator temperature, l050O F. b R.v.p. = Reid vapor pressure, e.],. = end point, s.c.f. = standard cubic feet, bbl. = barrel.
...
.
.
..
FLUID
CATALYTIC
CRACKING
PILOT PLANT
FLUE C A S
R E GENE RAT0 R
Figure 1
a t this and other points in the fluid process. Some data have been obtained on the effect of steam added with the oil, termed “dispersion steam,” and other work has demonstrated that the use of large quantities of steam in the spent catalyst stripper, termed “st,ripping st,eam,” is effective in rejuvenating poisoned catalyst. This paper presents the important details of these experiments. The expression ”hydrat,ion steam” as employed herein refers to prehydration (as distinguished from any hydration that niight be effected with dispersion steam and stripping steam) and is used to designate the steam which is introduced into the regenerated catalyst standpipe.
This expcriment was carried out with a high sulfur gas oiY and natural catalyst v-hich had previously been used for cracking low sulfur stocks. The gas oil, referred t o as “coke-still gas oil,” was produccd by delayed coking of a reduced crude of. predominantly West Texas ... origin, and contained 1.47 weight yo sulfur (Table 11). The delayed coking process is one in which the stock to be coked is raaidlv heated to a high t’emperatire and then charged t o a coke drum where coking is effect,ed by means of the contained heat. The catalyst, which was removed from a commercial catalytic cracking unit after approximately 2200 hours of operation, had a n Indiana relative activity (relative weight of reference catalyst which, by comparative runs, gives the same degree of cracking as n-ith a fixed weight’ of the test catalyst, a t t,he same flow rate and otherwise identical conditions) of 24.6 ( 5 ) ,a carbon factor of 1.10, and a gas factor of 1.22. During the pilot plant tests, approsimateJg 0.5 neight % of hydration steam (based on catalyst’) was admitted near the bottom of the regenerated catalyst standpipe. The results in Figure 3 shou that, during 64 hours of operation ‘iTith hydration steam, the catalyst activity and selectivity factors did not change marliedly. At the end of 64 hours, hydration steam n-as stopped and operations were continued for an additional 28 hours, at u-hicli time catalyst inspections and yield data were again obtained. The data show that the catalyst activity had declined sharplv and that the carbon factor had practically doubled. Also, as indicated in Table I, a marked decline in conversion had occurred, although the coke ?-ield had increased. These results are equivalent to a large increase in coke yield a t a givcn conversion. This experiment inriica,ted that hydration steam could be used to retard sulfur poisoning of natural catalyst, and that the removal of hydration steam resultcd in rapid poisoning. I n addition to retarding sulfur poisoning of natural catalyst, hydration steani also has an immediate beneficial effect on the yields ohained in catalytic cracking. This benefit has been emphasized in work previously reported by Hansford (4). T h e magnitude of this effect is shoivn in Table 111, in which data on coke-st,ill gas oil are presented showing that under otherwise
COMMERCIAL FLUID CATALYST UNIT
DESCRIPTIOR OF PILOT PLANT
With the exception of a n experiment carried out in a commercial fluid catalytic cracking unit and a number of catalyst tests performed in standard fixed-bed testing equipment ( 6 ) , all the cracking data presented in this paper were obtained in a P-barrelper-day fluid catalytic cracking pilot plant. Simplified diagrams of the pilot plant and of a typical commercial fluid catalytic cracking unit are presented in Figures 1 and 2, respectively. The pilot plant duplicates the commercial unit in most essential respects. In both units catalyst from the regenerator passes downward through a standpipe where it is maintained in aerated condition by the entrained air or by addition of air or steam. I n the case of the pilot plant, the steam is produced by pumping distilled water to a vaporizer, and then through a superheater; thus, accurately measured amounts of steam, entirely free of foreign material, are provided.
Figure 2
INDUSTRIAL AND ENGINEERING CHEMISTRY
August 1949
1719
I
TABLE 11. CHARGESTOCK INSPECTIONS
W
CokeStill Gas Oil Feed No. Used in pilot plant run N O . Gravity A.P.I. A.S.T.,M. d:istillation O F. Initial boiling poi&
16
HYDRATION
STEAM
")^Ia
EI.2 u 150 ,+8= 0
Figure 3.
CARBON '
50% 60% 70% 80% 90%
FACTOR-^-^'
I
I
20
40
HOURS ON
YTREAM
A
E2
I
2.5
I.Ol
p/
29.6
29.8
31.6
377 4 83 519 552 583 614 645 675 708
250 429 486 529 568 602 633 668 698 746
339 419 476 512 559 583 619 655 692 722
410 500 532 562 585 612 642 676 710
1.47 0.11
1.46 0.09
1.39 0.04
0.35 0.11
7.2
10.2
13.5
3.8
.
.
...
,..
by
4
D
27.2
I
Max. Sulfur wt. % (bomb) Carbo; residue, wt. % Gasoline (400" F,. e.p.1 vol. lab. frsctionation
Mixed Gas Oil 2 3 B C
1
10% 20%
NO HYDRATION---. STEAM
Low Sulfur Gas Oil
...
...
...
100
Steam Hydration Test with Coke-Still Gas Oil
Effect of hydration steam on rate of change of catalyst activity and selectivity
similar operating conditions, the addition of 0.5% hydration steam effected a 4% increase in conversion a t 6.5% coke yield. The increase in conversion, or gas-oil disappearance, was reflected primarily as added gasoline yield. As a result of the above experiment, plans are being made to test the beneficial effects of hydration steam in a commercial unit employing natural catalyst. USE O F DISPERSION STEAM
Dispersion or process steam has been reported by Davidson ( 9 ) to protect natural catalyst against sulfur poisoning in a manner similar t o that observed for hydration steam. However, the quantity of dispersion steam required for a given degree of protection was estimated ( 8 )t o be in the neighborhood of three times as great as the amount of hydration steam required, the exact ratio probably varying with feed stock. Although the benefits of dispersion steam in maintaining catalyst selectivity have not been definitely confirmed in the pilot plant, the use of steam in this role has been found to cause a definite improvement in product yields when cracking mixed gas oil feed of high sulfur content. As shown in Table 111, when 0.4 weight % dispersion steam was employed in cracking this feed at a coke yield of 5.6 weight %, conversion was increased more than 5%. This higher conversion was reflected as a 1.5% increase in gasoline yield and 5% of additional butane. The catalyst was essentially the same in the two tests, except for a small difference in carbon factor which, in this range, is not considered significant.
TABLE 111. EFFECT O F HYDRATION AND DISPERSION STEAM ON CONVERSION AT CONSTANT COKEYIELD (Reactor temperature, 900° F.) Charge Hydration steam, wt. % on catalyst Dispersion steam, wt. % on catalyst Stripping steam, wt. % on catalyst Conversion (naphtha-free), vol. % Catalyst carbon factor Gasoline (10-lb. R.v.p., 400' F. e . p . ) , vol. Excess butanes, vol. Yo D r y gap, wt. % Coke yield, wt. %
Coke-Still Gas Oil 0 0.5 0 0
%
0.5 0.5 43.3 4 7 . 0 1.6 1.6 37.0 4 1 . 0 4.5 3.7 5.6 6.6 f6.5-
Mixed Gas Oil 0 0 0.4 0
0.5 43.5 2.3 38.5 3.6 5.8
0.5 48.8 1.9 40.0 8.7 6.7
t5.6-+
while deterioration of the catalyst occurred when less than 1% stripping steam was used. During 88 hours of operation at the higher stripping steam rates, the carbon factor decreased from 2.21 to 1.31 and the gas factor from 1.66 to 1.19. The fixed bed rclative activity increased from 20.8 t o 26.1, probably as a result of decreased tendency of the catalyst to produce coke. On the basis of these pilot plant results, an experiment was carried out in the commercial unit, running on a nixed gas oil fecd of high sulfur content, using higher stripping-steam rates arid lower oil rates in order to effect the maximum possible improvement in catalyst selectivity. The results of this test are shown in Table V and are compared in Figure 4 with the corresponding pilot plant data. The rate of improvement in the commercial unit was very nearly the same as in the pilot plant. I n Figure 4, the beginning of the plant test (4561 hours on stream) corresponds to "24 hours on stream" for the pilot plant test, as plotted along the abscissa. (In plotting the data, it was assumed that the properties of the catalyst a t hour 4561 were the same as for the composite sample for hours 4509 to 4533.) The plant stripping steam rates, which varied between 1.15and 2.0 weight % were considerably lo~vcrthan
w
USE OF STRIPPING STEAM
In addition t o the function of stripping steam in maintaining low coke yields, the use of stripping steam in quantities greater than those normally employed can rejuvenate sulfur-poisoned natural catalyst. Experiments demonstrating catalyst rejuvenation, which were carried out in both the pilot plant and in a commercial unit, are described below. The first experiment was carried out in the pilot plant using a mixed gas oil containing 1.39% sulfur and a poisoned catalyst removed from a commercial unit. No hydration or dispersion steam was used. After a 24-hour period in which the strippingsteam rate was somewhat lower than that employed in the commercial unit, this rate was increased beyond the commercial unit value. The results showed a large decrease in coke yield together with a considerable improvement in the catalyst selectivity. The data are presented in Table IV and Figure 4. As shown in Figure 4,the improvement of the catalyst occurred during t h a t portion of the run in which 4 to 6 weight % of stripping steam was used,
2ot
-0-PILOT
PLANT REJUVENATION
OF CATALYST REMOVED FROM COMMERCIAL UNIT
2.2
I
-*-TEST IN COMMERCIAL UNIT 1-2 WT. % STRIPPING STEAM
I I
20
I
I 40
,
I 60
I
I
I
80
100
'
HOURS ON STREAM Figure 4. Improvement of Catalyst Activity and Selectivity with High Stripping-Steam Rate
INDUSTRIAL AND ENGINEERING CHEMISTRY
1120
T - L B I ,IT-, ~ I ~ I ~ R ~ VOF E C.IT$J,YST ~ I E ~ TSELECTIVITY
( R u n C, feed 3) C-1 C-2 C-3 0-12 12-24 28 -40
Teat S o .
1.9 4 9
Carbon on spent catalyst, w t . % k'rodiict yicids [o:ltput basis) Coke. wt. % Gas oil product, vol. 76
HIGH STRIPI'IXG
WITH
C-4 40-52
3.76 47.5
C-5 52-61
C-6 76-88
C-7 88-100
1,5 4 3
1.5 4 4
1.8 3.5
1 9 3.1
1.9
1.9
3.1
4 7
1,6;
1.14
1.11
101
3.04 53.3
2.86 54.6
:;ji ::it t:;; "0; 4 83 50.7
STEAbI k T E "
3,::;
3 42 51 2
C-8
100.112 1 ,j 4
i:!: "01;:. 1 7 118
;:&
1.07
4 33 46.5
4 42 46.0
8-
Vol. 41, No. 8
In vic:n- of these promising repilot plant test sults, was madc, using all three steam addition points, to determine whether the poisoned catalyst in t h e commercial unit, could be iniproved sufficierill>T t o niakc it) useful for further comnicrcial operations. In this test, a lo^ sulfur chargc stock was used i n ordertoniininiiircf.l~rtherpoisoning iyhilo attempting to improve
4.34 4.5.5
factor n-as rcduccd from 1.57 to 1.11,the gas fact,or dropped from 5 , O 293 4 . 8 288 5 . 0 370 6.2 1.32 t o 0.96, while the activity 417 4115 . 6 323 3 3 63 . 2 363 45,0 11 -L 38.4 36 9 40 1 i 6 8 16 2 47.4 iiicrcasedfroiii20.7to24.6. The stripping-steam rates used m w 24,l 22.6 23.8 ... 25.3. 26.0 1.05 to 1.56 n-eight yo: and 1 21.65 09 11 .. 83 11 . , .. h?-dration steam was introduced 1 . 86 52 11 .. 43 21 in thc regenerated cat,alyst standReact!r, 900" F. a n d 10 lb./:q. inch gag?: stripper, BOOo r,:rrgrneratoi, l000O t o 1 0 2 3 O I'. and no liyrirati~n pipe in thc aimunt of 0.4 t o 0.7 or clisperaian steam. b Lh. oil per li:,?b. catalyst in reactor. \?-eight 70.This cxperiment C Lb. catal3-st Ih. oil. demonstrates that tho use of high stripping steam rates in conjunction n-ith a low sulfur IEXT OF CATALYST SELECTIVITY IX charge stock makes it p o k b l e to rejuvenate pois COJIJILXC~AI~ USIT catalyst. t o selectivity leGcls approximating fresh cat 42.6
10 0 1 6
f.:
38.1 5.4
30 9 4 (5
40.3 2 9
42 2 7.8
4? 9 c
4
4:
fi
6I ,: 4!
,:;
Q
FTouri on Stream from Start of Run 4SOO--i533 4661
4S63-45(i7
43334587
4691-4596 4681-4805
4607
~
-. Catalyst
c ti7.i t y
Comment. Co 117 poii te sainple regcnerat ed c?ta!yst talien Stripping s t ~ a i i i increased 12.G00 113. 'lir. (0.97 \rt. catalrst) to 15,000 lb./hr. n-r. % o n catalyst) 1'e.d reduced froin 842 t o 8 2 5 bbl. 1 hr. (stripping-steam rate increased from 1.16 17 t . 3~i on catalyst to 1.56 wt, ?& o n cataIyst) Feed reduced from 625 t o 588 b h W hr. (stripping-steam rate increased from 1.56 wt. Yc o n catalyst to 1.68 wt. %; on catalyst), StriiJixn$ steam increased froin 15,000 Ib./hr. (1.68m. 470 o n catalyst) t o 18,400 ib. 'hr. (2.04 a-t. %c on catalyst) Composite Paiiiplc resenerated catalyst t a k e n Spot snmide of regenerated catalyst taken
ndinna lalive)
20 8
.,
Incpcrt,iqni Cnrhon Eartor
Gas factor
2 10
1 ,X
. .
. . ,
, . .
...
,..
23.6
.
..
..
...
...
1.e9
1.40
1.53
1 37 I
T H E O R Y OF ACFIOX O F STEAM W'PTH 1UATUR.AL CATALYST
The data presented above hzvc demonstrated that in catalytic cracliing \vitli natural catalyst, the addition of steam a t different points has a retarding influencc on thc dcgrce oE poisoning caused by high sulfur stocks; also that the u.sr of large enough quantities of steam inaltes it possible t o maintain catalyst se1cctivit.v equal to that of I'resh catalyst. In order t o faciiitate the application of these results t o cracking units differing in design and operating in different rc,iinery situations, it appears desirahle t o have a vorliing hypothesi; of a mechanism of this reaction which might explain the observed fact,s. JVth this point in mind the folloJ?-ingtheory of the action of &ani v-it11 natural catalyst is presented. Davidson (2) indicated that the poisoning of natural eatslyst by hydrogen sulfide docs not appear definitely attributahlc to the presence of iron and suggests that it might bc more closely connected n-ith hydration. He postulates that the loss of hydroxyl water from natural catalj-st requires a redistrihution of forces in the dehydrated catalyst. h s a result, t h set, up in the crystal latticc which is rclieved up0
the pilot of 4 to 6 lyeiglit yo, ~ hit appears ~ ~that , t,hc rates used in the pilot plant operation may have been considerably in excess of those actually required, and that lon-er rates can he used effectively in the plant. As a result of these experiments, further TABLE VI. 8 ' r ~ a l rHYDRATIOX TIXT KITH YIIXED Gas OIL