Uso de FT-MIR e calibração multivariada por MCR-ALS e SVR na determinação do teor de bioquerosene de macaúba e palmiste em misturas com querosene de aviação

Abstract

Aviation kerosene, commercially known as QAV, is a fuel derived from non-renewable energy sources used mainly in the airline industry. According to data provided by International Air Transport Association - IATA, the use of this fuel by this sector generates approximately 2% of carbon dioxide emissions to the atmosphere. These emissions represent a global concern in the 21st century and may increase with the expansion of the aviation industry. Thus, it is necessary to use environmentally sustainable fuel as an alternative to fossil fuel. Aviation biokerosene (BioQAV) is a fuel derived from sources of natural origin and can be animal or vegetable. In this work, the oilseeds macaúba and palmiste were used, which are not used in human consumption for the production of biokerosene and the analysis by spectroscopy in the medium infrared combined with the methods of Resolution of Multivariate Curves by Alternating Minimum Squares (MCR-ALS) and Regression by Support Vectors (SVR) to determine its contents in mixtures with kerosene in the concentration range from 1.00 to 70.00 (% v / v). For Support Vector Regression (SVR) models of biokerosene / kerosene mixtures, the results for macaúba biokerosene (BioQAVM) obtained values of RMSEP = 0.81 (% v / v), RMSEC = 0.06 (% v / v), C = 997.23, ε = 0.064 and R = 0.999953, for palm kernel biokerosene (BioQAVP) the values of RMSEP = 0.78 (% v / v), RMSEC = 0, 20 (% v / v), C = 915.08, ε = 0.063 and R = 0.999993. The results indicate the SVR as the best tool for multivariate calibration applied to biokerosene data obtained through spectroscopy in the medium infrared region, in order to quantify more quickly, directly and at a lower cost, the biokerosene contents of macauba and palm kernel in the samples. analyzed. For the MCR-ALS models, the recovery of the spectra is a tool that provides meaning and chemical interpretation of the results. It can be seen that the optimal number of main components to describe the models were 3 components. Thus, the rank of the data matrix for each BioQAV sample is 3, that is, the spectra of the samples can be obtained through a linear combination of three individual spectra. In practice, the rank of a matrix corresponds to the number of chemical species present in the mixture.