Automated Molecular Level Composition Modeling for Complex Hydrocarbon Mixtures

We developed a generic approach to model complex hydrocarbon mixtures in molecular level. This new approach was applied to both traditional petroleum oils and unconventional resources such as coal, lignin, and bitumen.

A general complex hydrocarbon conversion was qualitatively described by three essential structural attributes: Core, Side Chain (SC) and Inter-core Linkage (IL). SC and IL were described by a single attribute probability density function (pdf). Core is used to describe the ring and aggregated multi-ring structure, which is too complicated to be determined directly. A program called CoreGen was developed to obtain the identities of cores in a feedstock via the detailed measurement data (FTICR-MS, HPLC etc), literature, and scientific expertise. Based on the identities of cores, core was further decomposed to a set of user-defined structural attribute pdfs. By a juxtaposition sampling of those attribute pdfs, the structures of the molecular compositions of a feedstock were determined. In addition, the lattice statistics and the degree of polymerization were applied to describe the archipelago structures.

Consequentially, the quantitative information of a feedstock (mole fraction or mass fraction) can be calculated via applying an optimization loop. An objective function was given in terms of available analytical measurements. Through adjusting the parameters of those attribute pdfs, the objective function was minimized and thus the optimal mole fraction of the feedstock was obtained

This new approach was fulfilled automatically by an in-house software: Composition Model Editor (CME). Selected complex feedstocks including petroleum resid, coal, lignin were modeled.