Nonhydrogenative Demetalation of Residua Using Manganese Nodules Ronald H. Flscher“ and Wllllam E. Garwood Mobil Research and Development Corporatlon. Process Research and Technical Service Division, Paulsboro, New Jersey 08066
Helnz Helnemann Mobil Research and Development Corporation, Central Research Division, Princeton, New Jersey 08540
The shortage of crude oil supply has placed an increased emphasis on methods of upgrading petroleum residua, heavy oils, and oil-bearing sands. However, the high metal content of these stocks has made them very difficult to process. We report here the use of naturally occurring manganese npdules as a potentially inexpensive porous “seeding” surface for the thermal noncatalytic removal of the metallic components of residua. High levels of demetalation (70-90%) were obtained with relatively low coke and gas make. The kinetics of nickel and vanadium removal were found to be first order with respect to their concentrations. The rate of demetalation depends on a stockdependent inhibition term proportional to the asphaltene content of the charge stock. In the temperature range studied (800-920 O F ) the rates of nickel and vanadium removal were about the same. The rate constants and activation energies reported here for these petroleum stocks were in good agreement with pure compound values found by previous workers for the thermal decomposition of vanadium and nickel porphyrins.
Introduction The shortage of crude oil supply has placed an increased emphasis on methods of upgrading petroleum residua and heavy oils and oil-bearing sands. However, the high metal content of many of these stocks will result in poisoning of the catalysts used in processes such as catalytic cracking, hydrocracking, or hydrotreating. For example, one method of removing metals, hydrotreating of residua, is limited to stock of low to moderate metals content (Hildebrand et al., 1973f For high metal feedstocks, processing routes include coking (Jakob, 1971; Stuckey et al., 1969),deasphalting (Selvidge and Torrea, 1973), use of a hydrogen donor solvent (Vanvenrooy, 1974), extraction with acid (Adams et al., 1966; Kemp et al., 1969)and precipitation with a polar solvent (Deschamps and Renault, 1969). The present report deals with the use of manganese nodules as a potentially inexpensive porous “seeding” &face for the thermal noncatalytic decomposition of the metal components of residua. Naturally occurring manganese nodules have been previously reported to be catalytically active for the demetaIation of topped crude (Chang and Silvestri, 1974), oxidation of carbon monoxide and hydrocarbons (Weisz, 1968), and for the reduction of nitric oxide to ammonia (Wu and Chu, 1972). Experimental Section This study was performed in a 1-in. diameter stainless steel reactor fixed-bed unit. The manganese nodules were meshed -8/+30 (U.S. Standard Sieve) and washed and dried before use. Properties of the nodules are shown in Table I. The nodules were sulfided prior to contacting with oil in order to prevent excessive temperature excursions during operation caused by metals present. For this purpose a 2% HzS-98% Hz gas mixture was passed through the reactor at 320 O F and atmospheric pressure for 16 h. Hydrogen gas was not used a t any other point during the operation. Reactor outlet pressure was maintained a t 500 psig by means of a Grove Loader. Initial pressure was attained usirig
Address correspondence to this author at the Energy Research and Development Administration, Washington, D.C. 20545. 570
Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 4, 1976
nitrogen. The temperatures and pressure were such that the feedstock and product were predominantly in the liquid phase. Although we attempted to maintain the reactor iso: thermal, there was an appreciable temperature gradient (20-30 O F ) along the bed, and the temperature reported in Tables 11-IV is the maximum temperature of the bed. In all cases, a liquid hourly space velocity of 1was used. Liquid product was collected in a sight glass and pressure vessel and drained off a t the end of run. Liquid remaining in the reactor section at the end of run was withdrawn by means of a ball valve. After removal from the reactor, the nodules were extracted with benzene to determine the amount of oil still adhering to them. Coke analysis was performed on the benzene-washed nodules. In some cases, finely divided coke was present in the liquid; however, the weight of this coke was not significant. Because of the need to conserve the liquid product, no samples were obtained during the course of the 3-5 h run. Properties of the charge stocks are given in Tables 11-IV. The Wilmington resid was 650 O F f atmospheric resid, while the Agha Jari and San Joaquin were 10oO O F + vacuum residua. These residua exhibit a considerable difference in metals content and Conradson carbon residue (CCR). The San Joaquin resid was a solid a t room temperature and considerable difficulty was experienced in charging the resid to the reactor. Results Operating conditions, yields, feedstock, and product properties are listed in Tables 11-IV. The results are summarized in Figure 1, Which shows the dependence of demed a t i o n , liquid product yield, and coke-make on temperature. Note that demetalation of Agha Jari vacuum resid and Wilmington atmospheric resid shows the same dependence on temperature. Ninety percent demetalation can be obtained at 875 OF for both of these stocks with about 83% liquid product recovery and 11%coke formation. The coke and liquid product yield data from San Joaquin vacuum resid fall on the same curves as the other two; however, the percent demetalation is considerably reduced.
Table 111. Thermal Demetalation of Agha Jari Vacuum Resid (500 psig, 1 LHSV)
Operating conditions: charge Max. temp, O F Hours on stream Yields, wt % 1. Coke 2. Liquid product 3. Gas Liquid product properties Nickel, ppm 52 Vanadium, ppm 170 % Demetalation Sulfur, wt % 3.22 CCR, % 13.2 Molecular weight 855
8 837 4
9 865 4
6 90 3
10
12
84 6
81
28
10
2.2
100
31 82 2.39
64 96 2.16
44 2.70 ,
..
... ...
...
10
885 4
7
...
...
Table IV. Thermal Demetalation of San Joaquin Vacuum Resid (500 psig, 1 LHSV)
Operating conditions: charge Max. t e m a O F Hours on stream Yields, wt % 1. Coke 2. Liquid product 3. Gas Liquid product properties Nickel, ppm 260 Vanadium, ppm 140 % Demetalation 1.04 Sulfur, wt % CCR, wt % 22.6 Molecular weight 1231
Temperature, “ i
Figure 1. Thermal demetalation of Wilmington, Agha Jari, and San Joaquin residua: open spaces, Wilmington atm resid; dark spaces, Agha Jari vac. resid; half dark spaces, San Joaquin vac. resid. I t is evident from Tables 11-IV and Figure 1 that liquid product yield diminishes with increased temperature while gas-make remains about the same. Therefore, as temperature is raised, most of the decreased liquid product shows up as additional coke. However, high demetalation can be obtained with relatively low coke formation. The seventh run in Table I1 shows the results of thermal demetalation of Wilmington atmospheric resid over alundum
Surface area, m2/g Pore volume, cm3/g Particle density Real density Fe, wt % Mn, wt % SiOz, wt % A1203,
wt %
Ni, wt %
cu, wt %
c o o , wt % Na, wt %
Lake Michigan
243 0.415 1.36 3.10 13.6 23.5 18.6 4.36 0.91 0.47 0.18 0.10
248 0.399 1.52 3.862 44 3.5 29.0 1.91
4
13 905 3.5
7 90 3
15 78 7
20 73 7
200
160
65
110
100 35
40
22 1.12
1.12
... ...
73 0.67
...
...
balls instead of manganese n0dL-s. These - h a show that a t the same temperature significantly lower demetalation was achieved with alundum (56%) than with manganese nodules (75%) (Run 4).
Table I. Properties of Manganese Nodules
Pacific Ocean
12 832
11
832 4
Kinetics Equation 1gives the pseudo-first-order rate equation, for nickel or vanadium removal.
where C, is the concentration of nickel or vanadium in the demetalized product, k, is the first-order rate constant, t is the residence time, and C#J is a stock-dependent inhibition term. Writing eq 1 for nickel and for vanadium, integrating, and eliminating time dependence, we obtain the selectivity for