Reaction modeling approaches for complex hydrocarbon mixtures: Synthesis and application

The ability to process the heavy fraction of petroleum plays an important role in the financial viability of petroleum refining. Resid hydrotreating, visbreaking and coking are among the major processing steps in which large molecules are cracked into smaller molecules. The conversion in these processes is primarily thermal, with catalysts added in resid hydrotreating to desulfurize, demetallize and upgrade medium-size molecules. Understanding the mechanisms and kinetics of thermal reaction is a fundamental step in optimizing these processes.; Mechanistic reaction models summarize the current level of chemical knowledge in a form suitable for process optimization. The mechanistic modeling of pyrolysis reactions of petroleum mixtures has traditionally been limited by the complexity of the mixtures and a lack of computational power. A heavy-oil mixture may contain at least {dollar}10sp4{dollar} components and {dollar}10sp5{dollar} radicals and reaction pathways. Reaction models which seek to conserve this detail become very CPU intensive. A mechanism-based kinetic model has been developed with a small CPU requirement which would allow the model to be incorporated into a reactor simulation with fluid dynamics and heat and mass transfer.; The model utilizes a small CPU requirement by lumping radicals with similar reactivity together. A 42-lump subset of the {dollar}10sp5{dollar} radicals is used to describe all the elementary reactions with a high degree of accuracy. Structure/Reactivity relationships are utilized to provide rate constants for these elementary steps. The model predictions are compared to the results of full mechanistic simulations and the reaction of pure and synthetic mixtures of model compounds.; This dissertation also includes the development of useful tools to aid the modeling of thermal reaction systems. A simple modification of the long chain approximation has been developed and utilized in the Mechanism-Based Lumping model just described. Taylor series expansions also provide a way to derive semiempirical rate laws to approximate the mechanistic rate laws in limiting regimens. A simple inverse additive rule, analogous to parallel resistance addition, combines these limiting-case semiempirical rate laws to provide a rate law which is more accurate over the full range of experimental conditions.