Tools for modeling of heavy hydrocarbon conversion

Due to economic considerations and environmental concerns modern refinery units are subject to ever-increasing scrutiny regarding the detailed molecular composition of their product and effluent streams. This has motivated the development of detailed kinetic modeling at the mechanistic level.; A pre-requisite for mechanistic modeling is a molecularly explicit characterization of the feedstock. Such a detailed analytical characterization is seldom available especially for the heavier feedstocks. Thus, the starting point for the development of mechanistic models is the development of a molecularly explicit description of the feedstock. The development of this molecularly explicit description of the feedstock and its use in developing mechanistic reaction models have been described to the example of gasoil.; This mechanistic modeling is accompanied by a tremendous increase in the number of species and the number of reactions in the governing network. Additionally, the set of ordinary differential equations used to mathematically describe the reaction network is stiff. This, together with the large number of species involved, makes the numerical solution of the model very CPU-intensive. This can be a severe limitation to the development and deployment of mechanistic models. The development of tools for detailed kinetic modeling for heavy hydrocarbon conversion is the focus of this work.; The solve time for mechanistic models can be reduced by reducing the size of the network and/or reducing the numerical stiffness of the involved ordinary differential equations. Use of 'reaction rules' and 'stochastic rules' help reduce the size of the network and modeling of the chemistry at the pathways level eliminates the numerical stiffness of the set of ordinary differential equations. These ideas have been developed and demonstrated for the example of gasoil Fluid Catalytic Cracking.; Another useful strategy for speeding up the solution of mechanistic models involves the splitting of a large reaction network into smaller reaction networks for the purposes of numerical solution. Instead of solving one large N-dimensional model N smaller 1-dimensional models are solved and their results combined. This strategy is embodied in the Attribute based Reaction Model (ARM) developed for gasoil pyrolysis.