Espectroscopia na região do infravermelho próximo com instrumento portátil para caracterização de petróleos

Abstract

Due to problems of theft and adulteration of crude oils and fuels, this work developed analytical methodologies based on multivariate data analysis (classification and regression algorithms) of NIR spectra obtained with a portable spectrometer. The focus was to characterize crude oils in terms of their physicochemical properties, classify them by origin and discriminate adulterants in crude oils and fuels. The three objectives were achieved. There was a limitation in the classification by origin because of the absence of the fingerprint region (between 1,300 and 900 cm-1 ). However, the instrument demonstrated the efficiency in discrimination and quantification of adulterants with linear models such as PLS-DA and PLS, respectively. Finally, although the region between 6,000 and 4,000 cm-1 is important for estimating chemical characteristics in PLS models, because it presents the combinations and first overtones of SH, NH, CH, OH bonds. The SVR model transposes this absence in the spectra obtained with the portable spectrometer. The SVR is able to model the information present in the range between 6,000 and 11,000 cm-1 with similar or better precision to estimate characteristics, such as sulfur, nitrogen, and asphaltenes contents, and API gravity. In this way, the portable spectrometer demonstrates desirable efficiency to analyze petroleum and its derivatives with the ease of portability for in-situ applications.