1294
INDUSTRIAL A N D ENGINEERING CHEMISTRY
= diffusion rate constant, (min.)-I, defined by Equation 5 = weight of adsorbe:t, grams n = an integer c, = radius in spherical coordinates R = equivalent spherical radius of an adsorbent particle total surface area of an adsorbent particle, sq. meters per gram t = time after initial contact of feed and adsorbent, minutes 2 : = volume of a binary feed mixture, ml. average volume of an adsorbent particle, ml. = specific pore volume of an adsorbent, ml. per gram x = volume fraction of component B in liquid phase a t equilibrium xc = volume fraction of component B in liquid phase at equilibrium zo = volume fraction of component B in a feed mixture phase after t .Tt = volume fraction of component B in liquid . minutes ? / =volume fract>ion of component, B in pore liquid at equilibrium k
??Z
s = v = v,
LITERATURE CITED (1) Arnold, J. E., J . Am. Chem. SOC.,52, 3943 (1930). (2) Bartell, P. E., and Sloan, C. K., Ihid., 51, 1637, 1643 (1929). 13) Brunauer. 8.. “Adsorotion of Gases and VaDors.” I). 287. Princeton, Princeton-UniversityPress, 1943. (4) Ihid., Chap, XI. (5) Ibid., p. 465. j
I
Vol. 42, No. 7
(6) Brunauer, S., Emmett, P. H., and Teller, E., J . AWLChem. SOC., 60,309 (1938). ( i )Drake, L. C., and Ritter, H. L., IKD. ENG.CHEM.,ANAL.ED., 17, 787 (1946). (8) Eagle, S., and Scott, J. W., Petroleum Processing, 4, No. 8 , 8S1 (1949). (9) Emmett, P. H., and Demitt, T., J . Am. Chem. Soc., 65, 1257 (1943). (10) Geddes, R. L., Trans. Am. Inst. Chenz. Engrs., 42, 88 (1946). (11) Harris, B. L., IND. ENG.CHEM.,41, 15 (1949). (12) Hartman, R. J., Kern, R. A., and Bobalek, E. G., J . Collozd Sci., 1, 271 (1946). (13) Hibbard, R. R., IND. ENG.CHEX..41, 197 (1949). (14) Jones, D. C., J . Phys. Chem., 29,326 (1925). (15) LIair, B. J., J . Research Natl. Bur. Standards, 34,435 (1945). (16) Mair, B. J., and Forziati, A. F., Ibid., 32,151 (1944). (17) Ibid., p. 165. (18) Patrick, W.A., and Jones, D. C., J . Phys. Chem., 29, 1 (1926). (19) Rao, B. J., Ihid., 36,616 (1932). (20) Rescorla, A. A., Ottenmeller, J. H., and Freeman, R. S., A m ! . Chem., 20, 196 (1948). (21) Shearon, W.H., Jr., arid Gee, 0. F., IND. EXG.CHEM.,41, 218
(1949). (22) Strain, H. H., Anal. Chem., 21, 7 5 (1949). (23) Wilke, C. R., Chem. Eng. Progress, 45,218 (1949). (24)Williams, R., Jr., and Hightower, J. l‘., Chem. Eng., 55, S o . 11, 133 (1948). RECEIVED October 21, 1949.
esulf urization aphtha
tion and
of Cracke
Application of Cyclic Adsorption P r o c e s s SAM EAGLE
AND
CHARLES E. RUDY, JR.,
California Research Corporation, Richmond, Calif.
A new processing scheme for hydrodesulfurizing thermally cracked naphtha without appreciable octane number loss is presented. Preliminary separation of the cracked naphtha into a paraffin-naphthene-olefinraffinate and a high sulfur aromatic extract is accomplished in a cyclic adsorption plant using pentane as a stripping liquid. The adsorption extract is hydrodesulfurised and reblended with adsorption raffinate to yield a premium low sulfur gasoline blending component. Operating features of a n eightcolumn cyclic adsorption pilot plant are discussed. Experimental data are also presented for several runs in t h e pilot adsorption unit and for hydrodesulfurization of the high sulfur extract. Octane number studies were made using t h e desulfurized-cracked naphtha as a gasoline blending stock.
A
CYCLIC liquid-phase adsorption process for refining various
petroleum fractions has been described recently by Eagle and Scott ( 4 ) . Briefly, the process permits the separation of aromatics and olefins from paraffins and naphthenes in wide boiling range distillates by adsorption on silica gel or other adsorbents in a multiple fixed-bed plant. D a t a were presented in this paper on typical separations achieved in a pilot unit. These included the removal of aromatics from a hgdroformed naphtha, a kerosene distillate, and a gas-oil, and the separation of both aromatics and olefins from a thermally cracked naphtha. The present paper describes in a more detailed manner the adsorption separation of the aromatics and sulfur compounds from a high sulfur-thermally cracked naphtha and the hydrodesulfurization of the aromatic extract obtained therefrom for the purpose of preparing low sulfur gasoline blending components. It has been recognized for some time that hydrogenation over sulfur-active catalysts is an effective way of eliminating sulfur from petroleum stocks. Numerous recent publications attest t o the efficacy of these methods (1-3, 5-8) for a variety of petroleum stocks. However, such a process has not been applied
commercially to the desulfurization of thermally cracked naphthas, primarily because the olefins present in these naphthas are more or less completely hydrogenated while reducing the sulfur content to a sufficiently low value to be attractive. Xot only does this result in an excessive consumption of hydrogen but also in a decrease in the unleaded octane number of the desulfurized product. The octane number loss has been observed t o be as much as ten to twelve units. A decrease in octane number. however, can largely be avoided by making a preliminary separation of the naphtha by silica gel adsorption into a paraffinnaphthene-olefin raffinate and an aromatic-sulfur compound extract. This high-sulfur aromatic extract can then be hydrodesulfurized over a suitable sulfur-active catalyst. The purpose of this paper is to describe the operation of a cyclic adsorption pilot plant in making such a separation of thermally cracked naphtha, to present data on the hydrodesulfurization of the aromatic extract, and to discuss octane number results on blends of the desulfurized product x ith the low-sulfur adsorption raffinate. Preliminary adsorption studies made in batch columns demon-
July 1950
INDUSTRIAL AND ENGINEERING CHEMISTRY
1295 OPERATION OF CYCLIC ADSORPTION PLANT
tt
t
The schematic flow diagram of the cyclic adsorption plant (Figure 1) shows the separation process as applied to desulfurization of cracked naphtha in an eightcolumn pilot unit. The eight adsorption columns ( A t o H ) are packed with silica gel; the stripping liquid used is pentane. Each packed column is shifted in counterdirection to the liquid flow on a regular cycle through the successive zones of raffinate refining, extract enriching, hot stripping, and cooling. The number of columns in each zone may be one, two, or more, depending upon the prescribed purity of products and the residence times required in heating and cooling to accomplish satisfactory regeneration of the adsorbent for re-use. The general sequence of operation is as follows:
Figure 1.
Flow Diagram for Cyclic Adsorption Process
strated that a t a moderate feed-to-gel ratio, the sulfur compounds were almost completely confined to the aromatic portion of the gasoline. At a lower feed-to-gel ratio, separation of the paraffinnaphthene fraction from olefins was reasonably complete. About 4070 of the original dehexanized naphtha charged was recovered as a paraffin-naphthene fraction free of olefins, 35% as an olefin concentrate, and 25y0 as a high sulfur, unsaturated, aromatic extract. Suitable processing methods and uses for the three separated fractions were studied but will not be discussed in this paper. Desulfurization of the aromatic extract was of greatest interest and was the primary objective in making pilot plant studies of the separation of cracked naphtha in the cyclic adsorption plant. The larger scale studies were confined to the dehexanized naphtha, although consideration of the whole refining scheme for gasoline manufacture indicates that it is preferable to use a depentanized naphtha as the adsorption feed stock.
FEED STOCK The basic feed stock for these investigations was Dubbs cracked naphtha which had been dehexanized and treated in the refinery with dilute caustic and weak acid for removal of acids and nitrogen bases. This dehexanized naphtha was then rerun in the laboratory t o yield approximately 400' F. end point gasoline. An inert atmosphere of carbon dioxide was maintained over the stock in rerunning, storage, and subsequent use in order t o prevent any oxidation of the stock which might contribute t o activity decline of the silica gel adsorbent. Inspections on the dehexanized naphtha used as a charging stock for adsorption studies are shown in Table I. The composition by hydrocarbon type was read from the adsorptogram obtained in the separation of the feed stock in an analytical silica gel column using ethyl alcohol as desorbent. The feed-to-gel ratio was low enough t o give sharp breaks between paraffin-naphthene, olefin, and aromatic fractions, The midpoints of the breaks occurred a t the same volume per cent of liquid out the bottom of the column when read from the refractive index, aniline point, or density plots. The paraffin-naphthene split was estimated by the refractivity intercept-density method, employing the chart given in the A.S.T.M. Emergency Standard-45a.
The feed stock passes first through a n Excelso dryer for removal of entrained water and then through guard chambers for removal of very strongly adsorbed compounds, such as organic acids and nitrogen bases which are not readily removed from the silica gel by hot pentane stripping later in the process. If not removed, these compounds would enter the main adsorption zone, accumulate on the silica gel, and result in gradual activity decline of the adsorbent. The guard chambers may be packed with any suitable adsorbent-e.g., silica gel. The feed is then introduced into the raffinate-refining zone along with the effluent from the extract-enriching zone. The first effluent from the raffinate-refining zone is cold strippant, which is diverted to cold strippant storage. The balance of the liquid is a mixture of strippant and raffinate which is separated in the raffinate still. In the cooling zone cold strippant is introduced from storage in sufficient quantity t o return the adsorbent t o a favorable temperature for raffinate refining, the next zone on its countercurrent cycle. The first effluent from the cooling zone is hot strippant which is diverted t o hot strippant storage. Hot strippant enters the stripping zone in sufficient quantity to remove substantially all adsorbate from the adsorbent and thus regenerate it for re-use before it passes on to cooling. A recycle of the first liquid effluent from the stripping zone (comprising strippant and the first eluted adsorbate) is introduced into the extract-enriching zone in quantity substantially equivalent to that required for displacing the total liquid contained in one packed column. The remaining effluent from the stripping zone is then fed to the extract still for separation of strippant and extract. Periodically the guard chambers go through a special regeneration cycle during which time the accumulated compounds are desorbed and the adsorbent is reactivated for use in processing
Table I. Dehexanized Thermally Cracked Naphtha Charging Stock Gravity, A.PdI. Aniline point, F. Refractive index, n%O Bromine No., g . / l O O g . Sulfur, wt. % A.S.T.M. distillation, D-86, Start
%% Rnv-
50.2
96 1.4349 71
a
F.
En&point Hydrocarbon-type analysis, vol. % Paraffins Naphthenes Olefins Aromatics plus sulfur compounds
1.22 214 230 272
3.54 ___
427
30 14 36 20
INDUSTRIAL AND ENGINEERING CHEMISTRY
1296
Table 11. Properties of Fresh and Used Adsorbent from Cyclic Adsorption Plant ;lilsorbcnt Colrlirin charge, '1.
28-200 Mesh Silica Gel Fresh Vscd 4780 ... 0.72 2.10.i 1.17
-
48-80 Mesh Silica G e l l'resh Uscd 4369 ...
... ...
0.Gti
...
...
... ... ...
...
sorptive capacity, cc./g.c
h l e s h analysis, Ty1i.r.
Wt. % 20 -30 30 -40
40 -GO
0 24
r)
10
0.22
11.21
4 !I 238 43.8 l ( i .0 4.3 4.0
1.0 1.7.3 .)1.0 7 . (i
0.2
...
0.2 91.7 6.3 0.5
89:3
3.!1 630-80 51.9 2.3 80-1 00 100-200 12.3 0.9 :3 . 3 'l'hroueh 200 0.6 2.8 0.2 1.0 (6 lleasured by iso-octane displacrinent. b hleasured by mercury displacement. 6 Toluene selective adsorptive capacity is t h o voliiiiie in cubic ccntinirti.r, of toluene selectiyely adsorbed per gram of adsorbrnt from a n eqiii\~oliiiiie mixture of toluene and ino-octane. T h o used adsorbent after 180 o i i e r n t ~ r i p cycles is eyaanated to 10-2 ~ n i ni.o i 4 h o ~ i r sa t 200' I'.brfore testing.
additional feed. Tlir i'cgc'nt ion (,osisiats in desorbing t i l e - e compounds v-ith c t h ~ - alcolio l it1 ill removing the desorl~ent1 ) ~ . pentane vapor strippirig :It :in c1evntc:tl temperature. Pilot adsorption Jtutiies 11-ere carried out in r i unit conip~ihing eight 2-iiich >< IO-foot jncketed steel columns, a 4-inch X 4-foot guard ch:mibcr, ta-o convcntional stills \\-it114-inch X 5-foot fr:lc.tionating sccticm packed with I3crl saddles, auxiliar>- ttiuks, pumps, and control equipmelit. lic,;iting or cooling of adsorbent ivas accomplished by nltematiiig si~twiior water in the coluniri jackets so that, the tenipcratures :I tged TO" F. for adsorption and 200" t o 250" F. for hot stripping. 130th through 28-mesh o i i 200-mesh (28-200) and through .I8-medi on 80-mesh (48-80) silica gel adsorbent g~ivcgood a e p a i x t ion of :tromatics from noliaromatic constitueiit,>iii thc, IO-foot c~olunins:it superficial mlocities as high as 250 gallons per 1 r pcr square foot of coluniii (:i'ossscction. Cycle tinics for sh iig 01 cdumns were normally 45 to GO minutcs, but scparntioiis n ~ r equally e good at a cycle time of 20 miriutc,s xr-ith t\\-o (:oI~i~inis iii the liot stripping zone Feed-to-adsolbcnt ratios in tllc pilot plant, desulfurization studies ranged from 0.078 to 0.0!)1 gdlon ol clchexanized naphtha per pound of silica gel regcncmtcJ 01' 0.45 I O 0.53 volume pel' volume (volume of chary(! stock per empty column volume j. Stripping liquid quantities 11-ei'c 1.8 t o 2.7 voliimes per volume, whereas recycle liquid r:ingcd from 0.6 to 0.7 volume per volume. Small changes in the recycle ratio TVCYC critical in c~ontrollingyield and quality of raffiiiate and extract products. h'ormally the unit was operated tvith liquid don-nfloiv through the adsorbent bed?. Enough back pressure was niaiIit aincd to 1 ~ x 1 the 1 columns full of liquid and to prevent vaporization of the stripping liquid a t wgeneration temperatures. Prcsiure drop limitations indicated use of an atlwohent l a r g c ~ than 80 mesh, while tliflusivitj- c~)usidcratioiisfavored sni:rll article size. 1Iost pilot plant runs \\-ere made with commi91'cially available 28-200 mosh DiLviaon silica gel, although satihfactory results were obtaincd with the 48-80 mesh gel. Inspwtions on silica gcl charged aiid rcnioveJ from the unit after 180 operating cj-des ai'e slioH-n ill Table 11. Commerciltl n-pentane T Y ~ L Sused as the stripping liquiil. In practice the stripping liquid v a s rrcovered as the overliea~1 streanis from the raffimite and estract stills aiid recycled to thr stripping operation. Typical inspertions were: 92.5" A.L'.I. gravitb-, O.Ol"l, sulfur, and 93' to 100" F.boiling range. Carcful control of the product stabilization K X S nccessary t'o prcl-clii contaminating the strippant nit,h i,afiinste or extract and to ellsure it represcntative yield and puritJ- of loiv sulfur raffinate. Typical inspections on raffinutos and extracts produced in the c'>.c.licadsorption plant under different operating conditions a w 4ion-II in Tahle 111. It is to I)(> noted that the sulfur content 01
Vol. 42, No. 7
the rafinate varies from 0.05 to 0.018$4. I n runs 48 and 49 t h : tenipemture of the raffinate still n-as maintained above thc usual control point to reject the lowcbt lioiling ends of the raffinatc overhenti, aud the estract still \vas operated a t a lower control Iempcraturc than usual to prevcnt any contamination of tho strippant and raffinate by high sulfur content extract. I n run 50 Ircsli pentane !vas used a 'jplxint, and product separation was conipleted after the ads ioii scctioii was shut down. The reduction in rafinatc, sulfur coiitent from 0,04894 i i i ruii -13 t o O.O38C