Database-driven and property-constrained inference of molecular composition of petroleum fractions from routine experimental data

Introduction:

Heavy petroleum fractions contain highly complex molecular mixtures, and detailed molecular characterization usually requires specialized high-resolution instruments. This paper proposes a faster and more accessible inference framework based on routine industrial measurements, including gas chromatography and elemental analysis. A database of heavy gas oil samples links routine measurements with high-resolution molecular composition data. For a new sample, the method first finds a weighted combination of reference samples whose chromatographic profiles match the target, then refines the inferred molecular composition under property constraints such as elemental composition, density, and saturated-to-aromatic ratio. The final inferred compositions agree well with high-resolution mass spectrometry results, helping bridge ordinary refinery lab data and research-grade molecular insight.