Screening analysis to detect adulterations in Brazilian gasoline samples using distillation curves

Data on evaporated fraction temperatures obtained from the distillation curves of Brazilian gasolines were employed to identify adulterated samples by applying the Soft Independent Modeling of Class Analogy (SIMCA) model. Since 1977, different quantities of ethanol have been added to the gasoline sold in Brazil. The ethanol concentration is established by Brazil's Petroleum Regulatory Agency. Ethanol and heavy solvent prices are lower than that for gasoline, so they are often added to the fuel for purpose of gasoline adulteration. Presence of these species in high or low concentration creates modifications in distillation curves. The SIMCA chemometrics method was selected due to its offering the possibility of sample classification into to the defined class. The training set for the SIMCA model consisted of 34 samples and test set consisted of 16 gasoline samples, eight of which were adulterated. Five principal components were necessary to explain 95% of the total variance. All test set samples were correctly classified while four samples of the training set, which were in conformity with legislation. were misclassified as contaminated by SIMCA. (C) 2003 Elsevier Ltd. All rights reserved.