Determinação do teor de sebo em óleos para produção de biodiesel utilizando espectroscopia no infravermelho próximo

Abstract

The quality of biodiesel is directly related to the composition of its raw materials, with the mixture of vegetable oils and animal fats being common for adjusting properties. This work investigated the predictions of using a portable NIR spectrometer to quantify the bovine tallow content in mixtures with four types of vegetable oils: cottonseed, corn, palm, and soybean. Thirty-one samples with tallow concentrations ranging from 0% to 95% were tested using a spectral region of 900 to 1700 nm. Exploratory PCA analysis did not demonstrate logical clustering related to tallow content, and Partial Least Squares (PLS) regression failed to construct a predictive model. It is concluded that the portable NIR was not efficient for this application under scientific conditions. The absence of predictive dispersion was attributed to the combination of two limiting factors: the statistical scarcity of the number of samples and the limitation of the equipment's spectral range, which does not cover the entire near-infrared region.