Prediction of the SARA analysis of Colombian crude oils using ATR–FTIR spectroscopy and chemometric methods

In this work eight chemometrics models to predict saturates, aromatics, resins and asphaltenes (SARA) composition of fifty Colombian crudes oils using Fourier transform infrared coupled to attenuated total reflectance (ATR–FTIR) spectra were developed. The samples were correlated by similarity using principal components analysis (PCA) with their ATR–FTIR spectra. The validation showed satisfactory results for the prediction of the SARA analysis of crude oils. For each SARA component, standard errors of prediction (SEP) for light samples were 1.9, 1.7, 1.3, and 0.4. For heavy samples SEP were 2.5, 1.7, 3.7, and 1.4. In all cases, the coefficients of correlation (R2) between the values of reference and those predicted by the models were superior to 0.95. The IR spectroscopy coupled with the ATR cell plus chemometric techniques provide an alternative way for the quantitative prediction of the wt% SARA group-types with minimal handling of the samples in a short period of time.

► Colombian crudes oils were separated according to similarity using PCA.
► ATR–FTIR spectra of crude oils were used to obtain predictive PLS model.
► PLS models predict SARA composition in Colombian crudes oils.