The essential challenge of building detailed kinetic models for heavy hydrocarbons is due to the staggering complexity of not only the reaction mixtures but the complexity of each molecule within the mixture. There will often be thousands of “multi-functional” component species. The sheer size of the thus-implied modeling problem engenders a con¬flict between the need for molecular detail and the formulation and solution of the model.

Matt Neurock's early work in the use of quantum chemical calculations in the solution of engineering problems addressed these issues head on. His development of atom-based computational representations of complex molecules coupled with his approaches to the Monte Carlo simulation of both the structure and reactions of these molecules allowed the very practical problem of resid upgrading to be phrased in molecular terms.

Our recent work has explored automated strategies to represent heavy hydrocarbon structure and reaction both in terms of discrete molecules and probability density functions (pdf's) for molecular attributes. Monte Carlo simulation of feedstock structure is one method to cast the modeling problem in molecular terms. This technique samples pdf's for the attributes of the heavy hydrocarbon molecular structures to construct a representative molecular sample whose properties are compared against measured properties. Optimization methods are used to minimize the weighted sum of squares difference, and the final set of pdf parameters are the mathematical representation of heavy hydrocarbon structure. Subsequent molecular or mechanistic reaction models can be based on discrete molecules or the molecular attributes of the pdf's. The latter “Attribute Reaction Model” provides a large reduction in the number of reaction equations and thus solution time.

The relative merits of these strategies are discussed.