A computational chemical reaction engineering analysis of complex heavy hydrocarbon reaction systems. (Volumes I and II)

The elucidation of the fundamental chemistry governing the processing of complex petroleum feedstocks poses a significant problem in addressing many of the current environmental and processing concerns. The complex structural features and reactivity patterns of these systems obscure fundamental reaction analyses with the actual feedstocks. Thus, traditional reaction models for these systems, which are based on global reactant and product lumps, e.g., solubility or distillate fractions, are clearly limited. The goal of this work was the development of a molecule-based reaction simulation which would accommodate both the structural and kinetic complexities of these feedstocks.; A flexible Monte Carlo reaction simulation was developed to explore the effects of different feedstocks, processing conditions and catalyst formulations for various upgrading strategies (chemistries) on the resulting molecular and global products and their associated properties. In general the simulation mapped a complex heavy hydrocarbon reaction system into a molecular representation, reacted the ensemble of representative molecules, and finally averaged the molecular products to provide a measure with actual experiments. More specifically, the simulation was comprised of six basic elements: (1) the stochastic feedstock construction, (2) the deduction of appropriate level of reaction modelling, (3) the formulation of intrinsic kinetics data base, (4) the organization of governing kinetics into linear free-energy relationships, (5) the explicit accounting of extrinsic rate effects, and (6) the identification of reaction products and their properties.; The simulation was primarily used as a "learning" model. Thermolysis, hydrotreating, and catalytic cracking were three relevant upgrading chemistries used to analyze asphaltenes, heavy oils, and gas oils, respectively. The results for each of these simulations qualitatively compare with experimental results. In the asphaltene example, a near quantitative match between experiment and simulation was found for both detailed thermodynamic descriptions of the reaction products and "fit" molecular/property correlations. The latter result suggests that one or two adjustable parameters can effectively be used to turn the detailed simulation into a quantitative "process" model. Finally, the role of molecular interactions was found to be considerable and an accurate Monte Carlo methodology for incorporating kinetic coupling was developed.