The Kinetics of Hydrocarbon Cracking

Abstract

A general kinetic model which describes the catalytic cracking of pure hydrocarbons is presented. The model includes a monomolecular cracking path based on the Langmuir adsorption isotherm as well as a bimolecular path, following Rideal kinetics, which accounts for the possibility of a chain cracking mechanism being involved. Catalyst decay is accounted for using the time-on-stream decay function. Fitting of experimental data from n-nonane cracking on USHY at 673 K, combined with Monte Carlo simulations indicates that, in that case, the total catalytic activity could include between 0 and 90% of activity due to chain processes. This large margin of error stems from the combined effects of a large decay rate, forcing the experimenter to use average conversion data, and of experimental error. Fitting of the model to previously published cracking data for 2-methylpentane on USHY showed that the model lacks a suitable parameter to account for thermal reactions which were not accounted for in the original data set. This observation supports the impression that the model is sensitive to departures from the postulated mechanism. The above kinetic model has also been fitted to the results of n-nonane cracking at three temperatures as well as to previously published data for various other linear paraffins. In all these systems the parameter A2, which is a function of the bimolecular cracking rate constants, is found to be statistically insignificant, in spite of other experimental evidence which supports the existence of this route of cracking. Parametric analysis of the n-nonane conversion results suggests that catalyst surface composition is very sensitive to temperature due to a large difference in the enthalpy of adsorption between the reactant and the average product of cracking. As temperature is increased, the reactant competes more successfully for active sites, with the result that the relative importance of monomolecular cracking processes increases with temperature at the expense of bimolecular reactions. The rate of cracking per crackable bond was also considered. We present arguments that previous reports of increased cracking rate per bond as chain length of the feed molecule is increased are due to the use of inadequate models of the kinetics involved, rather than constituting a real phenomenon.