Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
10141039 | Vibrational Spectroscopy | 2018 | 11 Pages |
Abstract
There are still few initiatives studying applications for the classification of biodiesel derived from a mixture of raw materials, whether pure (B100) or mixed with diesel. The present study aims to conduct a preliminary assessment on the classification of pure biodiesel blends, and their mixtures with diesel at a proportion of 10%, based on FTIR spectroscopy and multivariate analysis. The work is contextualized around the monthly raw material consumption of Brazilian regions along a year. From this work, it is possible to verify that FTIR analysis, combined with multivariate methods, can be applied to classify both pure biodiesel blends and those mixed with diesel. The PCA showed great potential for recognizing the data patterns, while the HCA are able to discriminate the B100 blends from different Brazilian regions and cluster the samples according to the region biodiesel profile. For the B10 fuel blends, the OSC-PLS-DA achieved 100% sensitivity and specificity and can be applied for the classification procedure of biodiesel/diesel blends.
Keywords
HATRPLS-RSIMCAHCAMIROSCNIRANPASTMPLS-DAnuclear magnetic resonancePCABiodieselPartial least squares discriminant analysishierarchical clustering analysisPrincipal component analysisMultivariate analysisNMROrthogonal signal correctionAmerican society for testing and materialsPartial least squares regressionFTIRInfrared spectroscopyFourier transform infrared spectroscopyMid-infrared spectroscopySVMSupport vector machineFuel blendsSoft independent modeling of class analogyNear Infrared spectroscopyChemometrics
Related Topics
Physical Sciences and Engineering
Chemistry
Analytical Chemistry
Authors
Victor Hugo J.M. dos Santos, VÃctor Z. Pestana, Jean S. de Freitas, Luiz F. Rodrigues,