Espectroscopia UV-Vis para avaliação de biodiesel e misturas biodiesel/diesel

Abstract

Biodiesel is a fuel derived from renewable sources and can be obtained from vegetable oils, animal fat until residual oils. In Brazil commercialization of biodiesel occurs in blends, being added 5 % to diesel (B5). Biodiesel to be marketed must follow a series of standards established by Brazilian National Agency of Petroleum, Natural Gas and Biofuels (ANP). However alternative methodologies are being developed for the purpose and quality control of biodiesel. The objective of this work is to build multivariate models, in first moment to classify samples of biodiesel produced from different oils, in the second moment to predict the amount of biodiesel blended with diesel, and finally to classify mixtures biodiesel/diesel adulterated with soybean oil. In all cases molecular UV-Vis spectroscopy was used. An exploratory analysis was used to determine the behavior of the samples under study, for this we used the Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA). To classify samples biodiesel and blends biodiesel/diesel recognition techniques were used such as Supervised Independent Modeling of Soft Analogies between Classes (SIMCA) and Linear Discriminate Analysis technique associated with variable selection Successive Projections Algorithm (SPA-LDA). To predict the amount of biodiesel blended with diesel was used variable selection Successive Projections Algorithm (SPA) and Stepwise (SW), techniques associated with Multiple Linear Regression (MLR) and compared the values with Partial Least Squares (PLS). Based on the results obtained, it is possible to carry out the classification of samples of biodiesel from the oil source and the samples contaminated with soybean oil with 100% success. For predicting biodiesel content in blends biodiesel/diesel results were further adjusted using the SPA-MLR, where the average error of prediction was 0.57 %.