Structurally explicit composition model of petroleum vacuum residue

Abstract

The molecular modeling of fuels requires comprehensive composition data: names and molar fractions of the components. For fuels like petroleum residues, two problems arise: first, state-of-the-art analytical techniques can not detect the full list of molecule names, and second, this list would lead to a number of components that is unmanageable. A comprehensive composition can be modeled by the use of stochastic models that generate a combination of digital molecules whose global properties are close to the properties of the fuel under study. The current models only offer a set of representative molecules, from which an enormous number of different structures may be inferred. Different structures result in multiple initial conditions of a detailed kinetic model or conflicting results in a computational chemistry software. For an end user, it would be easy to build impossible structures that violate the rules of chemistry while interpreting the representative structures. To address these difficulties, a computer program was created which is able to model composition of any vacuum residue with unambiguous molecules. A custom-made molecule core library and a "molecular sieve" have been introduced, which consents generation of species that are proven to exist experimentally. Data analysis of the experimental results of three vacuum residues was performed and compared with simulation results. SARA separation simulation was executed for model validation. Detailed molecular class fractions of the simulation were provided; many distributions that are challenging to obtain by the use of experimentation were presented to better understand the molecular structures inside a vacuum residue.