Search

A Ni−Mo/Al2O3 hydroprocessing catalyst that was used for a year in a continuous commercial process for the hydroprocessing of Arabian Light and Arabian Heavy petroleum residua underwent deactivation associated with deposits of nickel and vanadium sulfides and coke. The used (spent) catalyst and a regenerated catalyst formed by burning off the coke deposits in the spent catalyst were tested for hydroprocessing of naphthalene, dibenzothiophene, or quinoline in a batch reactor at 350 °C and 165 atm. The conversion data were analyzed to determine approximate reaction networks, and the pseudo-first-order rate constants in these networks were determined for each reactant, for the fresh, spent, and regenerated catalysts. The spent catalyst, which contained 1.1 wt % Ni and 2.7 wt % V (based on the mass of the fresh catalyst) retained only very little of its initial activity, and most of the activity loss is attributed to coke deposits, as the activity of the regenerated catalyst was only slightly less than that of the fresh catalyst. The performance of the regenerated catalyst was nearly the same as that of a model catalyst with the same Ni and V contents that was made by treating the fresh catalyst with nickel and vanadium naphthenates to form metal sulfide deposits. The data show that the performance of the model catalyst provides a good approximation of the performance of the regenerated catalyst. Thus, such model catalysts are inferred to provide a good basis for developing correlations to predict the effects of deposited nickel and vanadium sulfides on catalyst aging. However, the results are restricted to catalysts with relatively small amounts of metal sulfide deposits and do not account for the effects of pore plugging that result when large amounts of deposits form.

The reactions of acrylonitrile, crotonitrile, propionitrile, and a small set of related reaction products were investigated in high-temperature water. Hydrolysis of the cyano group yielded the corresponding amides, which underwent similar hydrolysis to yield the carboxylic acids. The conversion of acrylonitrile was more rapid than that of its saturated analogue propionitrile, as the unsaturated site of acrylonitrile provided a facile pathway for the formation of alcohols, amines, and ethers. Carbon−carbon bond cleavage pathways involving the hydration of the olefin moiety were also observed for acrylonitrile.

Ni−Mo/γ-Al2O3 hydroprocessing catalysts enriched in nickel and vanadium by contacting with solutions of the respective metal naphthenates were sulfided and tested for hydroprocessing of naphthalene, dibenzothiophene, and quinoline in a batch reactor at 350 °C and 165 atm. Approximate reaction networks were determined for each reactant, and the data showed the dependence of the pseudo-first-order rate constants on the catalyst nickel and vanadium contents. The nickel sulfide deposits only slightly affected the rate constants for hydrogenation, but the vanadium sulfide deposits led to decreases in the rate constants for hydrogenation reactions in the naphthalene network and to increases in those for hydrogenation reactions in the dibenzothiophene network. Nickel sulfide deposits led to almost no change in the rate constants for hydrogenolysis of dibenzothiophene, but vanadium sulfide deposits led to decreased rate constant for this reaction. The nickel sulfide deposits have little activity for reactions giving lower-molecular-weight (“cracking”) products, but the vanadium sulfide deposits have a relatively high activity for cracking, which suggests that they are acidic; the effects are reversed by the presence of the basic quinoline in the reactants. The results indicate a need for representing the nickel and vanadium sulfide deposits separately in process models for heavy oil hydroprocessing.

A mechanistic model for the catalytic cracking of n-heptane was developed using a novel mechanism-based lumping scheme that exploits the chemical similarities within reaction families. The formal application of 13 reaction family matrices, which correspond to the 11 reaction families in the model, to the matrix representations of the reactants and derived products generated 70 species, 235 elementary steps and 70 ordinary differential equations. The reaction family concept was further exploited to constrain the kinetics within each reaction family to follow a quantitative structure/reactivity Polanyi relationship. Ultimately, four Polanyi relationship parameters and one catalyst specific parameter were optimized using experimental data obtained from the cracking of n-heptane at 500 °C over HZSM-5 with a Si/Al ratio of 21.25. The model correlations were excellent, as were the a priori predictions of experimental results at 450 and 550 °C with an HZSM-5 Si/Al ratio of 21.25 and at 500 °C with HZSM-5 Si/Al ratios of 35.25 and 63.5. The thus validated model was then used to probe the controlling elementary steps of n-heptane cracking. Carbonium ion cracking, β-scission, and hydride transfer were the kinetically significant reactions.

A mechanism-derived lumping strategy for modeling the acid-cracking kinetics of hydrocarbon mixtures based on a limited set of pure component experimental data was tested in terms of its ability to predict the kinetics and product spectra from the reaction of a 1-phenyloctane/n-hexadecane mixture. The modeling approach has two main components. First, reactant and product molecules are organized into compound classes, e.g., paraffins, olefins, and aromatics. Second, the elementary steps of the acid cracking of each member of a compound class are constrained by quantitative structure−reactivity relationships (QSRRs) determined from pure component experiments only. This two-dimensional, mechanism-derived lumping approach provided a significant reduction in the number of parameters required to model the cracking reaction. Application of this lumping strategy to the acid cracking of the phenyloctane/hexadecane reacting mixture provided good agreement between experimental data and a kinetic model containing only 14 parameters obtained from separate pure component experiments. The model revealed the applicability of the QSRR-based lumping approach to the cracking kinetics of hydrocarbon mixtures.