324
Ind. Eng. C h e m Res 1990,29, 324-329
were calculated and are shown in Table IV. The distribution coefficient of dibenzothiophene in the unoxidized oil extraction was the highest (5.1). After oxidation, hexyl sulfide was not present in the oxidized oil. The distribution coefficients of dibenzothiophene and phenyl disulfide in the extraction of oxidized oil were increased: from 5.1 to 11.2 for dibenzothiophene and from 1.8 to 4.5 for phenyl disulfide. As for the phenyl sulfide, the distribution coefficient in both extractions was relatively low. These results are consistent with the results described previously. Conclusions Oxidation of the Arabian AGO by nitrogen dioxide has converted the sulfur compounds of the oil to forms that are more polar in nature. These sulfur compounds are more readily extracted by polar solvents, such as lactone, that have been demonstrated to be aromatic/olefinic dissolving, thus resulting in higher sulfur removal, lower solvent-to-oil requirement, and hence, higher extraction oil yield for the same sulfur removal. Oxidized oils from other petroleum stocks which have higher contents of olefins and aromatics do not show an oil yield increase in the extraction by lactone. The distribution coefficients
of dibenzothiophene and phenyl disulfide in lactone extraction of AGO are greatly increased for the oxidized oil as compared to the untreated oil. Registry No. NOz, 7697-37-2; nitrogen dioxide, 10102-44-0; ybutyrolactone, 96-48-0; hexyl disulfide, 10496-15-8;phenyl sulfide, 139-66-2; dibenzothiophene, 132-65-0;phenyl disulfide, 882-33-7.
Literature Cited Deal, C. H.; Derr, E. L., Selectivity and Solvency in Aromatic Recovery. Prep--Am. Chem. Soc., Diu.Pet. Chem. 1962, 7 (4), 214. Gerster, J. A.; Gorton, J. A,; Eklund, R. B. Selective Solvents for Separation of n-Pentane from 1-Penteneby extractive Distillation. J . Chem. Eng. Data 1960, 5, 423. Guth, E. D.; Diaz, A. F. Method of Removing Sulfur and Nitrogen in Petroleum Oils. US.Patent 3,847,800, 1974. Guth, E. D.; Helgeson, N. L.; Arledge, K. W.; Brienza, A. R. Petroleum Oil Desulfurization Process. US.Patent 3,919,402, 1975. Tam, P. S.; Kittrell, J. R. Process for Purifying Hydrocarbonaceous Oils. U.S. Patent 4,485,007, 1984. Thompson, R. B.; Druge, L. W.; Chenicek, J. A. Stability of Fuel Oils in Storage--Effect of Sulfur Compounds. Ind. Eng. Chem. 1949, ;I
I
121, 2713.
Received for review July 18, 1989 Accepted November 8, 1989
Desulfurization of Fuel Oil by Oxidation and Extraction. 2. Kinetic Modeling of Oxidation Reaction Patrick S. Tam,tJil J a m e s R. Kittrell,' and John W. Eldridge*J Department of Chemical Engineering, University of Massachusetts, Amherst, Massachusetts 01003, and KSE, Inc., P.O. Box 368, Amherst, Massachusetts 01004
T h e oxidation of Arabian atmospheric gas oil (AGO) with nitric acid t o remove sulfur from the oil is similar to the gradual processes involved in the storage instability of petroleum distillates and synfuels, except that in this process the instability process is accelerated by adding a strong oxidizing agent, nitric acid. The elemental composition of these sediments is compared. A mathematical kinetic model is presented to describe the kinetics of sulfur removal in the oxidation of AGO using a CSTR. This model employs lumping of the sulfur compounds in the oil into four groups, S1, S2, and S3, according t o their retention times (hence, boiling points) in the gas chromatograph, and residue (R) containing some of the other three sulfur groups. The first group (Sl) had a very fast reaction rate. T h e second group (Sa) reacts with second-order kinetics with a rate constant of 17 g of oil/(g of S2.min) at 25 "C. The third group (S3)was not present in the unoxidized oil, and as it was formed, it equilibrated between the oil and the residue phases. The principle of desulfurizing Arabian atmospheric gas oil (AGO) by oxidative desulfurization (ODS) is to form a coproduct (residue) that has high sulfur content and can be separated from the oxidized oil. It has long been recognized that petroleum distillates and synfuels (both crude and refined) are unstable upon long-term storage in the presence of air at ambient conditions, with the formation of sediment (deposit, residue, sludge, and insoluble gum). The sediment formation from the oil is due to the gradual oxidation and polymerization of the reactive compounds present. The ODS process is, in many ways, very similar to the instability process of fuel oil, except that the oxidation rate in ODS is greatly accelerated by using a strong oxidizing agent, nitric acid or nitrogen oxides in this case. The gum formation from jet turbine and diesel fuel is clearly associated with oxidation and cooxidation of hyKSE, Inc. University of Massachusetts. §Present address: The Pritchard Corp.. Overland Park, KS 8
66210.
0888-5885/90/2629-0324$02.50/0
drocarbon mixtures (Mayo and Lan, 1983). Walters et al. (1949) studied the gum formation in cracked gasoline and found that the gum and peroxide formation is autocatalytic in nature. It was proportional to the amount of air available for reaction, indicating that oxidation played the major role in this process. Not all fuel oils show the same rate of sediment formation. For thermally cracked gasoline, heavier fractions are less stable than light fractions (Walters et al., 1949). In general, paraffins, aromatics, monoolefins, and diolefins are increasingly unstable toward oxidation in the order named. Offenhauer et al. (1957) also found the same trend. Straight run distillates which normally contain mostly paraffinic compounds are more stable, whereas catalytically cracked distillate fuels containing more unsaturated hydrocarbons are unstable (Brooks, 1926; Cassar, 1931; Flood et al., 1933). For shale oil and shale-derived fuels, the instability is even more pronounced (Inneen and Bickel, 1951; Cooney et al., 1984; Frankenfeld et al., 1982). Certain heterocyclic sulfur and nitrogen compounds in addition to those responsible for the gum in cracked petroleum distillates are involved in 0 1990 American Chemical Society
Ind. Eng. Chem. Res., Vol. 29, No. 3, 1990 325 Table I. High Acid-to-Oil Ratio (0.10) Oxidation of AGO by 90% HNO, (Acetic Acid/Feed Oil = 0.03) wt of reactants (W) in CSTR," g wt of oil ( WOiJ in CSTR, g
oil feed rate (Ff),g/min space time (01, min oxidized oil yield, wt % sulfur content, 70
s1
52 s3 residue yield, wt % sulfur content, 70 aqueous phase yield, wt % sulfur content, 70 sulfur balance closure, 70
530 476 135 3.5
483 410 48.2 8.5
479 410 45.7 9.0
578 506 24 21.1
556 479 18.1 26.5
458 376 8.9 42.2
474 383 8.5 45.1
100.2 0.75 0.045 0.576 0.129
98.7 0.63 0.008 0.452 0.171
98.4 0.64 0.033 0.460 0.147
98.9 0.54 0.050 0.358 0.132
97.7 0.50 0.037 0.328 0.135
96.1 0.46 0 0.309 0.151
96.1 0.45 0.009 0.278 0.163
6.1 3.10
10.3 3.36
10.0 3.44
9.5 3.21
11.3 3.20
12.7 3.32
11.9 3.27
7.00 0.64 92.2
5.6 0.48 92.9
6.0 0.48 92.9
5.6 0.46 81.0
5.1 0.38 81.3
7.3 0.50 99.6
6.2 0.39 79.0
"Oil plus residue and acids. Yields were based on oil feed. Oxidized oil had about 2% acetic acid.
the formation of sediment in shale oil naphtha. Nitrogen compounds are also known to be deleterious to the stability of hydrocarbon fuels (Nixon, 1962; Thompson et al., 1951). Pyrrole-type compounds are found to be significant sediment precursors in fuel oil. Dominant compounds are alkyl-substituted pyrroles, especially 2,5-dimethylpyrrole (DMP). Indoles also play an important role in promoting sediment formation, but not as significant as the pyrroles. Mathematical models are widely used for the description of the kinetics of petroleum reactions. For such complex reactions, lumping the components into pseudospecies is commonly necessary. Blanding (1953) treated the kinetics of catalytic cracking by using a one-lump conversion model and found first-order kinetics. Weekman (1979) used a three-lump model for the kinetics of cracking of gas oil. Smith (1959) described catalytic reforming reactions with a four-lump model. In the present effort, the kinetics of sulfur removal in oxidative-desulfurization (ODS) reactions of Arabian atmospheric gas oil (AGO) are modeled by lumping the sulfur compounds into four groups. The predicted sulfur removal compares favorably with the measured values. The yield of sediment formation can also be estimated.
Experimental Section A continuous stirred tank reactor (CSTR) was used for the oxidation of a middle distillate (Arabian atmospheric gas oil) with a mixture of 90% nitric acid and glacial acetic acid. Use of acetic acid was to disperse the residue coproduct uniformly in the reactor. The extent of oxidation was controlled by the acid-to-oil (A/O) ratio and the spacc time. The A/O ratio is defined as the ratio of the flow rate of HNO, (calculated based on 100% acid) to the flow rate of feed oil, by weight. The space time used was also based on weight. The CSTR was a two-piece, 1-L, cylindrical vessel constructed of Klimax glass with the top and bottom halves held together by clamps. Oil was pumped into the reactor from the top, and mixed acids were added to the reactor from the top by a fine stopcock buret. Reaction temperature was measured by a thermocouple connected to a digital temperature controller. The oxidation is exothermic. Reaction thermperature was maintained at 25 "C by immersing the bottom half of the reactor into a water bath. Agitation of the reaction mixture at 300 rpm provided uniform mixing of reactants. Products emerged from the exit tube of the reactor by overflow. The total sulfur content of the oil was determined by using a Gamma-Tech Model 100 chemical analyzer. A
Table 11. T h r e e Groups sulfur group elution time, min column temp, "C bp, "C
of S u l f u r Compounds
s1
0-10 80-170 180-290
s2
10-15 170-220 290-380
s3 15-18 220-250 380+
Table 111. Comparison of Residue from Oxidation of AGO w i t h Sediment Formed in Cycle Oil deposit original oil (residue) S, % N, m m S, % N, % AGO, 0.07 A/O HNO, oxidation 1.19 60 3.82 3.46 AGO, 0.11 A/O HNO, oxidation 1.19 60 3.39 3.97 catalytic cycle stock la 1.28 100 3.38 3.58 1.72 0.80 catalytic cycle stock 2" 0.96 200 nSource: Thompson et al. (1951).
Perkin-Elmer Sigma 2000 gas chromatograph with flame photometric detector (FPD) was used to analyze for specific sulfur compounds, using the conditions summarized in Table I of part 1 of this series.
Oxidation of AGO Atmospheric gas oil with an initial sulfur content of 1.07% was treated with 90% HNO, in a CSTR at 25 "C and 1 atm, at 0.10 A/O ratio. The weight ratio of acetic acid to feed oil was 0.03. The results are shown in Table I. The sulfur removal increased with increasing space time until at around 45 min it leveled off. Figure 1 shows the yields of oxidized oil and residue as a function of sulfur removal. As the sulfur removal was increased, the oil yield decreased and the residue yield increased. Specific sulfur chromatograms of the oxidized oil are shown in Figure 2. A certain group of lower boiling sulfur compounds was completely removed, even at 3-min space time. These lower boiling sulfur compounds eluted a t a maximum retention time of 10 min. Formation of higher molecular weight product sulfur compounds with retention times greater than 15 min was also observed. These compounds were not present initially in the feed oil and they constitute one group of product sulfur compounds believed to have been produced during oxidation. The elution times, column temperatures, and boiling points corresponding to the three sulfur groups are listed in Table 11. Mechanism of Oxidation Sulfur removal by oxidation is due to the formation of high-sulfur-containing residue, which results from the accelerated sedimentation of the oil from instability induced by nitric acid. Table I11 shows the comparison of
326 Ind. Eng. Chem. Res., Vol. 29, No. 3, 1990 Table IV. Comparison of Elemental Compositions of Residue a n d Deposits from Unstable Fuels elemental compositions of deposits, % oil description C H N O S empirical formula HNO, oxidation of AGO: 0.07 A / O 60.5 3.7 14.9 3.8 6.3 HNO, oxidation of AGO: 0.11 A i 0 64.5 3.8 12.9 3.4 6.6 fuel oil blends with cat. cracked components from West Texas 76.9 6.7 1.8 8.7 1.9 California 78.5 6.9 3.1 8.5 1.3 6.9 2.1 8.5 4.1 Middle East 77.5 cracked components-Middle East 76.1 6.3 3.4 7.7 2.6 5.0 diesel fuel marine (DFM) 62.0 3.0 25.0 2.0 crude shale oil 74.8 5.8 10.5 0.9 8.0 7.4 6.4 9.6 1.0 mildly hydrotreated shale-derived middle distillates 76.6 61.3 11.3 21.5 NA" 5.7 diesel fuel with 2,5-DMP as dopant shale-derived fuel with dopants DMP 62.1 5.7 11.4 20.4 0.1 5.7 DMP plus thiophenol 67.7 4.9 12.1 9.1
OXIDIZED OIL
reference
Offenhauer et al., 1957
Offenhauer et al., 1957 Jones et al., 1984b Frankenfeld et al., 1982 Frankenfeld et al., 1982 Frankenfeld et al., 1982 Jones et al., 1984a
FEED OIL
< 07 % s ATTENUATION = 256
0
A/O=0.10
A
A/O:0.03 3 5 MIN SPACE TIME
0 75 %
s
ATTENUATION
RESIDUE
:64
9 0 MIN SPACE TIME 0 64 % S ATTENUATION = 32
SULFUR REMOVAL (%) Figure 1. Relation between sulfur removal and yields of oxidation with 90% HN03.
residue in this process with the sediments formed in cycle oils due to storage. Although different compounds can be present in different feedstock, this does show the similarity of the residue obtained from AGO oxidation and the precipitate from unstable cycle oils of similar sulfur content. These similarities are further illustrated in Table IV, which shows the elemental compositions (C/H/N/O/S). Various feedstocks-diesel, shale-derived distillate, and cracked distillates-with and without dopants are reported. The empirical formulas for the deposits were also determined. Adding dopants (dimethylpyrrole (DMP) or thiophenol) to the fuel also resulted in deposits of similar elemental composition. They are based on one atom of nitrogen. Residue from the oxidation of AGO has a formula of C20H24N04S0.4. Deposits from unstable fuels have carbon atoms ranging from 14 to 50, depending on the properties of the oil. However, the H/C ratio of the deposits from unstable fuels was from l to 1.3, which was
45 1 MIN SPACE TIME 045% S ATTENUATION = 8
J
1
80
i
4 (10
l
7
l
1 140
1
I
0
