Ind. Eng. Chem. Res. 1993,32, 1003-1006
1003
KINETICS, CATALYSIS, AND REACTION ENGINEERING Conversion Of Light Naphthas over Sulfided Nickel Erionite Roland H. Heck’ and Nai Y. Chen Central Research Laboratory, Mobil Research and Development Corporation, Princeton, New Jersey 08540
A natural erionite ore has been exchanged with ammonium and nickel salts to yield a Ni/H erionite catalyst that is active and stable for selectively hydrocracking only the n-paraffins from light straightrun naphthas. The primary product is a C5+ liquid that is 15-20 octane numbers higher than the feed and a propane- and butane-rich gas by-product. Results from a 110-day pilot plant run demonstrated that a catalyst life of more than 1year should be possible. Naphthenes, aromatics, and isoparaffins are neither produced nor consumed in this process, resulting in a C5+liquid product that is lower in benzene and total aromatics than attainable by catalytic reforming of these feeds. Although no further work is planned with this catalyst, a naphtha-upgrading process based on shape-selectivezeolitic hydrocracking could provide an attractive alternative to catalytic reforming or isomerization for these hard to upgrade naphthas. It should be particularly attractive in areas where the by-product propane and butane have good value. Table I. Composition of Ni/H Erionite Catalyet
Introduction
It has been shown that erionite can selectively crack n-paraffins from a mixture of other hydrocarbons (Chen et al., 1969: Chen and Garwood, 1973). Full-range reformates from catalytic reformers usually contain about 1015 wt % C5+ n-paraffins depending on their crude source and reforming severity. By selectively converting these n-paraffms to lower molecular weight paraffins, the octane rating of a reformate can be increased by about 3-7 numbers. A post-reforming process known as the Selectoforming was commercialized in the 1960s for raising the octane rating of reformates while producing propane as a by-product (Chen et al., 1968;Burd and Maziuk, 1972). The need to reduce benzene and other aromatics in gasoline as mandated by recent legislation in the U.S. suggests that alternative processes to conventional catalytic reforming of naphthas could become attractive (Corbett, 1990; Crow, 1992). A process based on shapeselective cracking provides an alternative to catalytic reforming or isomerization for boosting the octane of naphthas without generating additional benzene or other aromatics. A process of this type should be particularly attractive for processing light naphthas. These streams have very low octane, contain as much as 45 wt 7% n-pentane and n-hexane, and are not easily upgraded by catalytic reforming. Presented in this paper are the results from a study in which light straight-run naphthas were hydrocracked over a Ni/H erionite catalyst. Catalyst stability, activity, and selectivity were determined along with final product yields and properties.
Experimental Section The catalyst (Heck and Chen, 1992)was prepared from natural erionite obtained from Jersey Valley, NV. It was exchanged with N&+, washed, dried, exchanged with Ni2+, and then dried again. It was then pressed into 118-h. pellets and calcined to yield the final catalyst containing 3.8 wt 7% NiO. The nickel component was added to overcome the relatively rapid activity decline experienced osas-ssss193/2632-1003$04.00/0
wt%
Si02
12.0
A1203
16.4
NiO K20 CaO Na2O MgO Fez03 total Si/Al
3.8 3.8 0.5 0.1 0.3 3.1 100.0 3.7
equiv/Al 0.32 0.25 0.06
0.01 0.05 0.69
with H erionite with similar feeds and conditions. Table I summarizes the chemical composition of the finished catalyst. The catalyst was placed in a 3-cm-diameter,26-cm-long cylindrical reactor where it was presulfided with H2S before charging the feed over it in a down-flow mode. The unit was started up with a flow of 3-6 mol of pure hydrogen gas per mole of naphtha feed. After start-up, the off gas was recycled from a high-pressure separator and mixed with make up hydrogen to maintain a total gas to feed ratio of 3 to 6. Hydrogen purity of the total gas stream ranged from 80 to 92 mol % for the run, with an average of about 88 mol %. A total pressure of 2.86 MPa was maintained. Composition and properties of the feeds for this study are given in Table 11. Naphtha A is a CS-80 OC refinery derived naphtha to which pentanes were added to restore its natural crude oil composition. Naphtha B is a dehexanizer overhead as received from the refinery. It has a more typical front-end distribution, including pentanes and some butanes. This feed is of higher density and contains less isopentane and more n-heptane than naphtha A. It is also higher in sulfur content and somewhat lower in octane number than naphtha A. Both hydrogen and naphtha were fed into the top of the reactor. The reaction was exothermic, and approximate adiabatic reaction conditionswere maintained by matching the reactor internal temperatures with the reactor wall temperatures using four external electrical heaters dong the length of the reactor. 0 1993 American Chemical Society
1004 Ind. Eng. Chem. Res., Vol. 32, No. 6, 1993 Table 11. Composition and Properties of Feedstocks Used in This Study naphtha A naphtha B composition, wt % isobutane 0.0 0.5 0.0 7.0 n-butane isopentane 22.2 13.6 23.4 21.9 n-pentane isohexanes 16.8 15.5 23.1 13.0 n-hexane isoheptanes 5.4 5.8 n-heptane 0.2 4.5 1.3 1.0 benzene 7.6 9.8 naphthenes 0.0 7.4 other total 100.0 100.0
I
I
I
Cs+RON
67.4 0.656 78 5.2
density, g/cm3 av mol wt sulfur, ppm
I
I
I
0.8 -
-
0.6 -
0.4
-
0.2
-
0 0
I
I
I
I
I
I
I
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
1msv e n-Hexane
properties
I
n-Pentane
Figure 2. Half-order reaction plot.
58.0 0.662 80 76
-
number (RON)of 83 for naphtha A and -76 for naphtha B. The required temperature rose sharply for the first few weeks and then leveled out for the remainder of the run. Catalyst stability is better illustrated by the decline in the reaction rate constant for the reactions of interest. It was shown in earlier work that n-heptane cracking over this catalyst can be modeled using a half-order rate expression (Heck and Chen, 1992). n-Pentane and nhexane conversion for these naphthas can also be modeled using a half-order rate equation and an activation energy of 34 kcal/mol (Figure 2). Using these kinetics, the rate constants for conversion of both n-pentane and n-hexane were calculated from the temperature requirements shown in Figure 1. These rate constants are plotted versus time on stream in Figure 3. This log-log plot shows a catalyst deactivation rate equivalent to that which has been previously reported for reformate cracking over a similar catalyst (Burd and Maziuk, 1972). The curves in Figure 3 can be expressed mathematically in a Voorhies (1945) type equation:
4 430
37 350 0
20
40
60
80
100
120
Days on Stream
Figure 1. Temperature required for 75% n-pentane conversion.
The study began by processing naphtha A for 34 days at about 1.6liquid hourly space velocity (LHSV). Although the reactor temperature ranged from 377 to 426 "C, most of the aging data (3-34 days on stream) were obtained between 377 and 389 "C. For the last 66 days naphtha B was processed at a constant temperature of 388 "C except for a short process variable study near the end of the catalyst aging test. Results and Discussion A. Catalyst Activity and Stability. The reactor effluent compositions and octane ratings of the cS+liquid obtained at selected temperatures and space velocities for the two feedstocks are given in Table 111,together with the composition of the feeds and the reaction conditions. Although olefins are primary products of the cracking reactions they were almost completely saturated at the conditions of this study, and only trace amounts of olefins were found in the reaction products. The measured hydrogen consumption agrees well with that stoichiometrically required to saturate the olefinic products of n-paraffin cracking. The hydrogen consumption ranged from about 0.4 wt % feed at the least severe conditions with naphtha B to almost 2 wt % of feed for the most severe conditions with naphtha A. However, for consistency, all yields in this work are reported on a weight percent hydrogen free product basis. The temperature requirement to achieve 75 % conversion of n-pentane for the length of the run is shown in Figure 1. This severity is equivalent to a research octane
k, = kot" or ln(kJko) = n In t where kt, ko = reaction rate constant at t and 0 days on stream, n = catalyst deactivation constant, and t = days on stream. The value of the deactivation constant, n, determined from the slope of the curve in Figure 3 is about -0.16 for both n-pentane and n-hexane. This translates to a 61% decline in the reaction rate constant for the first year of operation. This catalyst deactivation would require a process temperature increase of -26 "C, an acceptable operating window for a projected 1-year cycle length. Although naphtha B contains 15 times the sulfur of naphtha A (76 vs 5.2 ppm), the activity and stability of the catalyst for these two feeds are equivalent. Furthermore, the sulfur in the liquid product from naphtha B was less than 5 ppm, indicating substantial desulfurization as a result of processing over this Ni/H erionite catalyst. B. Octane Boost. The composition of the reactor effluent shown in Table I11 has been normalized to zero conversion of isopentane. In all cases,the correction factor was less than 3 % , indicative of the low reactivity of isopentane and all branched paraffins, aromatics, or naphthenes over this catalyst. The octane rating of the c5+liquid increased as its yield decreased by the selective cracking of Cg+ n-paraffii~to CL products. Conversion of n-hexane ranged from 67 to 99% for naphtha A and from 59 to 90% for naphtha B. n-Pentane conversion was consistently lower than n-hexane conversion,ranging from 44to 88 7% for naphtha A and from 45 to 895% for naphtha B.
Ind. Eng. Chem. Res., Vol. 32, No. 6, 1993 1005 Table 111. Reactor Effluent Analysis and Octane Rating of CS+Liquid
1.6 426 1.6
A. Naphtha A 2.6 424 1.6
67.4
84.4
CS iC4 nC4 iCs nCa nCe other total
22.2 23.4 23.1 31.3 100.0
CSt
100.0
days on stream T,OC LHSV
13.6 377 1.6
25.6 377 1.6
27.6 377 1.6
83.1
products 76.0
76.5
76.2
7.3 31.5 1.4 7.2 22.2 2.8 0.3 27.3 100.0
3.9 25.7 1.3 8.5 22.2 3.9 5.9 28.6 100.0
1.8 18.4 0.4 5.7 22.2 13.1 7.7 30.7 100.0
1.5 18.2 0.5 6.0 22.2 12.9 7.4 31.3 100.0
1.2 18.3 0.5 6.0 22.2 11.7 7.6 32.5 100.0
52.6 47.4
60.7 39.3
73.7 26.3
73.7 26.3
74.0 26.0
feed octane, RON composition, w t %
c1+ c2
c4
~~
B. Naphtha B 77.6 388 1.6
60.6 388 1.6
days on stream T,OC LHSV
100.3 428 1.4
101.0 429 2.3
103.0 424 1.7
104.0 426 3.7
~
feed -
octane. RON
products
58.0
72.5
69.5
76.3
72.9
74.2
70.4
0.5 7.0 13.6 21.9 13.0
1.5 16.3 1.1 8.5 13.6 10.1 4.1 44.8 100.0
1.3 16.6 1.1 8.1 13.6 12.1 5.1 42.1 100.0
5.8 33.0 1.4 6.9 13.6 2.2 1.3 35.8 100.0
3.7 24.4 1.1 7.5 13.6 7.5 2.6 39.6 100.0
4.6 27.6 1.4 7.4 13.6 2.7 1.3 41.4 100.0
2.4 15.5 0.8 8.0 13.6 11.7 5.3 42.7 100.0
92.5 7.5
72.7 27.3
72.9 27.1
52.9 47.1
63.2 36.8
59.0 41.0
73.3 26.7
~~~
~~
composition, w t % c1+ c2
CS iC4 nC4 iC6 nCs nCa other total cs+ C41
0
-
1
-
8-
'1
5RelsUve Reaction
Rate Constant
-.: -.
Naohtha '8'
NaohlhaA'-'
F
I
14 1
Yield, c5+ wt % I
2
5
10
20
40
Bo
1W
150
-
slopes for n-paralfin removal
Days on Stream
Figure 3. Catalyst aging.
Figure 4 shows how the CS+liquid product yield varied with the CS+RON for both feeds. The octane boosting efficiency (octane gain per yield loss) is about 0.38 for naphtha A over the entire range of severity. However, for naphtha B the octane boosting efficiency is 0.65 at the lowest severity and drops to about 0.32 at the highest severities. Also shown on this plot is the approximate slope of the yield/octane curve that should result from simple elimination of CSto C, n-paraffins from the feed. Although the blending octane value of the n-paraffins varies with the octane numbers of the stock into which they are blended (Heck, 1989) for the calculated slopes shown in Figure 4,octane values of 61.7,24.8,and 0.0 were used for n-pentane, n-hexane and n-heptane, respectively. The corresponding calculated octane boost for removal of these paraffins from a 70 RON feed is 0.10,0.50,and 0.78. In the case of the naphtha A, which contained nearly equal
50
-I nC..-
nCR-
I
I
55
60
I
I
I
I
65
70
75
Bo
I
85
Octane Number, R+O Naphtha'A
Naphtha'B'
Figure 4. Yield/octane plot.
amounts of n-pentane and n-hexane and very little n-heptane, the octane boosting efficiency is intermediate between those calculated for n-pentane and n-hexane removal. However, for naphtha B which contained 4.59% n-heptane, 13.0% n-hexane, and 21.9% n-pentane, the octane boosting efficiency at low severity is intermediate between those calculated for n-heptane and n-hexane removal. This efficiency drops with increasing severity to a value between those calculated for n-pentane and n-hexane. Although the conversion of n-heptane was not determinedin these experiments,elimination of n-heptane
1006 Ind. Eng. Chem. Res., Vol. 32,No. 6, 1993 Table IV. C4-Product Selectivity A. Naphtha A
T,OC LHSV
c1+cz c3
iC4 nC4
T,OC LHSV c 1 + cz c3
iC4 nC4
377 1.6
Operating Conditions 377 377 1.6 1.6
Normalized Net CC Product, wt 4.8 6.7 5.9 70.5 70.0 69.3 1.9 1.6 1.9 22.9 21.6 22.9 B.Naphtha B Operating Conditions 388 426 429 388 1.6 1.6 3.7 2.3
424 1.6
426 1.6
%
9.9 65.3 3.2 21.6
Normalized Net CC Product, wt % 7.8 12.5 12.6 6.6 82.0 80.7 83.5 84.8 2.9 1.7 2.1 3.0 7.3 5.1 1.9 5.6
15.5 66.5 3.0 15.1
424 1.7
428 1.4
13.6 82.5 2.7 1.3
14.7 83.4 2.2 -0.3
is undoubtedly contributing to the larger octane boosting efficiency achieved for naphtha B in this study at low severity. C. Compositionof C k Cracked Products. Table IV shows the normalized net C p product selectivity. Cracked products were almostexclusively n-paraffins, with propane andn-butaneyieldstotaling8GW%ofthenet Cpproduct. However, n-butane and propane can both crack with this catalyst, particularly at high severities and high concentrations (Heck and Chen, 1992). Thus, for naphtha A that contained 0.2 wt % n-butane, the net n-butane yield is very high. However, for naphtha B that contained 7.5 wt 5% n-butane, the net butane produced is significantly lower. In fact, for the highest severity run at 1.4 LHSV and 428 "C the net yield of n-butane is negative. The crackingof n-butane and a minor amount of propane results in a large increase in methane and ethane. This is particularly evident for the runs at high temperatures and high severities. Comparing the yields from naphthas A and B in Table IV, it can be seen that if n-butane is a desired product it should be eliminated from the feed. However, if propane has a significantly higher value than
n-butane. this catalvst can be used to convert n-butane in the feed additional propane product. The net yield of isobutane is small, but consistently positive and roughly equivalent for both feeds. Since the molecular size of isobutane precludes its entry into the zeolitic pores, its formation is most likely occurring on the exterior of the zeolite.
Conclusions Ni/H erionite can selectively crack n-hexane and npentane from light naphthas. The octane rating of the Cg+ liquid can be increased by 15-20 research octane numbers with the major cracked products being propane and n-butane. This catalyst will both generate and crack n-butane. Thus, if n-butane is a valued product, it can be eliminated from the feed and additional n-butane will be produced. However, if propane is more highly valued than n-butane, n-butane can be included in the feed and additional propane will be produced. Results from a 110day pilot plant run demonstrate that the catalyst should have a projected life of more than 1year. A process based on a small-pore shape-selective zeolite like erionite could provide an alternative to catalytic reforming for upgrading the octane of light naphthas without increasing benzene or total aromatics. It could be particularly attractive if there is a good market for propane and/or n-butane.
Literature Cited Burd,S. D.; Maziuk, J. Hydrocarbon Process. 1972,51, (S), 97. Chen, N. Y.;Garwood, W. E. Adu. Chem. Ser. 1973,121, 575. Chen, N. Y.;Maziuk, J.; Schwartz, A.; Weisz, P. B.Oil Gas J. 1968, 66 (47),154. Chen, N. Y.;Lucki, S. J.; Mower, E. B.J. Catal. 1969,13, 329. Corbett, R.A. Oil Gas J. 1990,88 (25),33. Crow, P. Oil Cas J. 1992, 90 (15),21. Heck, R. H. Energy Fuek 1989,3 (l), 109. Heck, R.H.;Chen, N. Y. Appl. Catal. A 1992,86, 83. Voorhies, A. A.,Jr. Ind. Eng. Chem. 1945,37, 318.
Receiued for reuiew October 1, 1992 Revised manuscript received February 9, 1993 Accepted February 18, 1993