Crude oil properties predicted by NMR

NMR spectroscopy supported by multivariate analysis has been used to predict the physical and chemical properties of crude oil in a rapid alternative to current time consuming lab methods. In a method reported in the journal Magnetic Resonance in Chemistry, Sonia Puligheddu and colleagues from Saras Ricerche e Tecnologie S.p.A. and the University of Cagliari described how they applied the process to 64 crude oil samples from 28 extraction fields worldwide. They varied from light to heavy crudes with high to low sulphur content and paraffinic to naphthenic compositions.

Using a 300-MHz spectrometer, neat crude samples were analysed directly to give spectra with good resolution and, in the case of light crudes, excellent resolution. The spectral data were processed using partial least squares regression models which were calibrated and tested using the physical and chemical properties of some crudes that were measured using conventional analytical methods. The refined models were then applied to the other crudes to predict properties like the density, sulphur content, total acidity number (TAN), true boiling point (TBP) distillation yields and the UOP characterisation factor (k_UOP) which is a measure of the content of paraffins, naphthenics and aromatics.

The method gave accurate results even for oils with a broad composition that were sourced from some of the refineries. In most cases, the results were in line with those obtained by ASTM methods. However, the researchers suggested that the accuracy could be improved even more by increasing the number of samples and developing models for crude oils with a specific range of properties. They concluded that the method "offers a very fast alternative to time-consuming laboratory methods, avoiding tedious sample preparation and elaborate spectral data pretreatment."