Uso da Espectroscopia de Infravermelho Médio e Métodos Quimiometricos MCR-ALS e SVR no Controle de Qualidade de Biodiesel Metílico de Dendê e Macaúba

Abstract

Among the analyses performed by the National Petroleum Agency (ANP) on S10 diesel, the largest nonconformity was the biodiesel content in the diesel. To ensure the quality of S10 diesel, this study proposes a method to determine the biodiesel content and adulteration in diesel, using the mid-infrared spectroscopy technique combined with chemometric methods such as Multivariate Curve Resolution by Alternating Least Squares (MCR-ALS) and Support Vector Regression (SVR). All samples were prepared in the laboratory and the pure diesel was donated by Transpetro. The results of the determination of the methyl palm oil biodiesel content in diesel blends, using MCR-ALS, showed an RMSEC of 0.52%, RMSEP of 0.65% and a correlation coefficient of R2 of 0.9970. The determination of the methyl biodiesel content of macauba in diesel obtained RMSEC of 0.32%, RMSEP of 0.46% and R2 of 0.9997. The determination of adulteration of palm oil biodiesel with lubricating oil obtained RMSEC of 0.65%, RMSEP of 0.81% and R2 of 0.9998. The determination of adulteration of macauba oil biodiesel with residual oil obtained RMSEC of 0.34%, RMSEP of 0.52% and R2 of 0.9998, all these results demonstrating the high precision and reliability of the models in the identification of content and adulterations using (MCR-ALS). In Support Vector Regression (SVR), the determination of the methyl biodiesel content from palm oil in diesel using SVR obtained RMSEC of 0.11%, RMSEP of 0.15%, RMSECV of 0.31% and R2 of 0.9998. The determination of the methyl biodiesel content from macaúba in diesel with SVR obtained RMSEC of 0.16%, RMSEP of 0.24%, RMSECV of 0.22% and R2 of 0.9994. The determination of adulteration of palm oil biodiesel with lubricating oil using SVR obtained RMSEC of 0.12%, RMSEP of 0.58%, RMSECV of 0.29% and R2 of 0.9965. The determination of adulteration of macauba biodiesel with residual oil and lubricating oil using SVR obtained RMSEC of 0.20%, RMSEP of 0.28%, RMSECV of 0.29% and R2 of 0.9994. Both models demonstrated good precision and robustness in detecting adulterations.