Determination of Petroleum Properties by Fiber-Optic Fourier Transform Raman Spectrometry and Partial Least-Squares Analysis

Abstract

The petroleum products unleaded gasoline and reformate have been analyzed by Fourier transform (FT) Raman spectrometry with the use of a fiber optic cable that allows a measurement to be carried out remotely from the spectrometer. Samples were contained in glass cuvettes, and collection of high-quality spectra was simple. Fuel properties such as research octane number (RON), motor octane number (MON), density, benzene content, and flexible volatility index (FVI) were determined by application of the partial least-squares multivariate statistical approach. Analytically useful calibrations ( R2 > 0.97) were obtained for all of the significant fuel properties studied. Cross-validation results were, as expected, worse than the calibration results but still indicated the usefulness of the method. Standard errors of prediction with the use of cross validation for models that contained 14–28 samples include: MON = 0.12, RON = 0.16, benzene % = 0.09, and density = 0.0018. These results demonstrate that it is feasible to undertake on-line analyses of petroleum fuels by fiber-optic FT-Raman spectrometry.