| Abstract |
A new predictive semi-lumped model developed for the cracking of paraffins has been applied to n-heptane. The model is based on a complex network reaction scheme including the ensemble of reactions taking place during the catalytic cracking of paraffins. The reaction scheme is detailed to the carbon atom number level and a large number of reactions are taken into account, ranging from cracking, hydride transfer and chain growth, which eventually leads to coking. It is considered that all the species present can participate in all of the above reactions, creating a complex network where all the species interact with all the others by means of the three sets of reactions included. Since the reaction network involves 1125 reactions, written in terms of elementary steps, an important problem to be solved hereby is to assign values to the corresponding reaction rate constants; For most of the elementary reactions considered experimental data is not directly available in the literature so, for their kinetic rate constants quite simple arbitrary equations were used to estimate them for each set of reactant/products considered. Each kinetic expression was parameterised with three parameters in order to describe the dependence of the corresponding kinetic rate constant on the number of carbon atoms of the adsorbed and gas phase species involved. The estimation of the nine rate constant parameters is discussed, as well as their influence on the evolution with time on stream of the profiles of global gas phase and predicted surface species distributions, coke content, catalytic activities and deactivation curves. The model is applied to the cracking of n-heptane and validated using experimental data. It can be shown that, despite the assumptions that were considered in this approach, very meaningful results are obtained. The model computes the product distributions at the exit of the reactor as well as the global distribution of surface adsorbed species, the coke content at the surface and the global activities, can be predicted as a function of time on stream. (C) 1999 Elsevier Science Ltd. All rights reserved. |
