Determinazione delle proprietà del petrolio tramite spettroscopia ¹H-NMR e analisi chemiometrica

Currently, the refiners have to cope with an increasing demand of high quality distillates of tighter specifications, while facing the need to process increasingly heavier and poorer quality crudes. Furthermore, the recent trends to process heavy and lower cost crudes spurred the interest on residue upgrading, normally referred to as bottom of the barrel processing, as a means to fully exploit the crudes’ heavy fractions. To reach these goals, refiners are looking at optimizing production and improving selection of crudes and crude mixtures to be processed which can only be achieved through a detailed knowledge of the composition and quality of feed and finished products. Generally, to know the yield and the quality of the crude oils, it is necessary to perform an atmospheric distillation and to characterize their fractions at laboratory level (crude assay). In order to characterize the crude residue, a vacuum distillation is needed and the physical-chemical properties of the resulting fractions must be measured. The time required for these characterization measurements is between 2 and 3 weeks, the amount of sample needed is large, up to 10 Kg, and the overall routine is generally expensive in terms of instruments, man-hours and consumables. Furthermore, crudes quality is not constant and tends to fluctuate over time and with field. To ensure technically, economically, and environmentally proper refining operations there is a substantial need for fast analytical methods for characterization of crude oils and their residua. During this PhD research, as a fast and viable alternative, 1H NMR spectroscopy, combined with chemometrics, has been used. To exploit the full information content of the NMR data acquired on complex systems, as crude oils, chemometric regression techniques as partial least squares (PLS) have been developed. The proton NMR spectra of neat crude samples have been recorded using a 300-MHz laboratory spectrometer. PLS regression models for prediction were built using analytical data obtained at the SARTEC laboratory. The attention was focused on the study of the most common crude assay properties such as API gravity, TBP distillation yields, KUOP factor, pour point, total acidity number and sulfur, carbon, hydrogen and nitrogen content. Successively, it was proved that also other important physical-chemical parameters can be predicted using the NMR. Among these asphaltenes, carbon residue, metal content, and especially the breakdown of aromatic classes content which, despite their impact on process, have been overlooked in previous researches. In addition, this thesis, demonstrates the possibility to predict most of the properties of atmospheric and vacuum distillation residua directly from the 1H NMR spectra of crudes. The speed of this method and most of all the possibility to avoid the distillation of the sample prior to analysis, opens to the opportunity to use real time property data of residua to evaluate process strategies and to assess the impact on downstream processes. Chemometric modelling was carried out for the most important property of residua, i.e. yields, density, viscosity, sulfur content, asphaltenes, carbon residue, nitrogen, hydrogen and carbon content.