Kinetics of catalytic cracking with short contact times

Abstract A novel isothermal pulse reactor was used to study the kinetics of gas oil cracking on a FCC equilibrium catalyst with short contact times. The feed was a lighter gas oil than typically used in FCC-units. Experiments were carried out by varying the catalyst-to-oil ratio, volume of the oil pulse, temperature and residence time. After each hydrocarbon pulse the catalyst was regenerated by introducing several oxygen/nitrogen pulses through the catalyst bed. The amounts of carbon monoxide and carbon dioxide formed were measured and the amount of coke on the catalyst was calculated. The reproducibility of the experiments was excellent. A kinetic model that included five lumps, namely, gas oil, gasoline, liquefied petroleum gas (LPG), dry gas and coke with five cracking reactions was developed first and its kinetic parameters were determined from the experimental results. The data could be best described by the model wherein the rate of cracking of gas oil to gasoline and to LPG were both approximated as second order dependency and the rate of cracking of gas oil to dry gas and to coke as first order dependency on the gas oil concentration. The five-lump model was further enlarged by dividing the gasoline fraction into paraffins, olefins, naphthenes and aromatics resulting in an eight-lump model with eight reactions. In addition, changes in the activity of the catalyst during one experiment was accounted for by using two exponential activity functions, one for catalytic cracking reactions and the other for coke formation. The formation of dry gas was considered to be the product of a thermal reaction only. The kinetic parameters of the Arrhenius’ law and the deactivation parameters were estimated by a non-linear regression program. In the five-lump model 12 parameters and in the eight-lump model 18 parameters (rate coefficients, activation energies and deactivation parameters) were obtained. The kinetic parameters of the Arrhenius’ law were statistically significant in both models.