| Abstract: | Samples of commercial gasoline, from the National Program of Fuel Quality Monitoring of the National Petroleum Agency, were collected from gas stations located in the Midwestern state of Sao Paulo, Brazil, and analyzed by several physicochemical standard methods established by ANP Resolution no. 309. Also, important information related to tampering was analyzed with the marker solvent. Statistical analysis and exploratory chemometric were employed to discriminate the presence of markers of solvents in commercial gasoline. The results showed that statistical and chemometric parameters such as atmospheric distillation temperatures T10 and T90, RON, benzene and saturated and aromatic hydrocarbons satisfactorily describe the presence of marker solvent, usually with a probability exceeding 70%. Furthermore, after optimizing the SIMCA algorithm, sensitivity in the training set with cross‐validation leave‐one‐out (83.8%) and the set of prediction (77.1%) were revealed. The proposed method will become indispensable and recommended for discriminating samples of fuels for commercial applications in routine monitoring programs and quality control. Copyright © 2011 John Wiley & Sons, Ltd. |
