Diesel oil discrimination by origin and type using physicochemical properties and multivariate analysis

The objective of this work is to provide models for the prediction and classification of B4 (volume fraction of 4% in conventional diesel) and diesel samples commercialized in the state of Minas Gerais – Brazil, with respect to its origins and types. Hence, we determined physical chemical parameters, such as, specific gravity, flash point and distillation curves for two distinct diesel samples of metropolitan and interior types from different refineries. Through PCA (principal components analysis), the first three principal components separated the samples by origin and types, with percentage of explained variance greater than 96%. For the LDA (linear discriminant analysis), the information in the invoice receipt of samples regarding origin and type were used. In the simultaneous prediction of origin and type, the percentage of correct classification was above 94%, considering the classification error, due to the similarity between some samples.