Premium
Raman spectroscopy for the discrimination and quantification of fuel blends
Author(s) -
Liu Zhe,
Luo Ningning,
Shi Jiulin,
Zhang Yubao,
Xie Chengfeng,
Zhang Weiwei,
Wang Hongpeng,
He Xingdao,
Chen Zhongping
Publication year - 2019
Publication title -
journal of raman spectroscopy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.748
H-Index - 110
eISSN - 1097-4555
pISSN - 0377-0486
DOI - 10.1002/jrs.5602
Subject(s) - raman spectroscopy , biodiesel , partial least squares regression , diesel fuel , principal component analysis , biofuel , analytical chemistry (journal) , fossil fuel , environmental science , biological system , materials science , chemistry , pulp and paper industry , mathematics , statistics , waste management , chromatography , optics , organic chemistry , engineering , physics , biology , catalysis
Biodiesel is an alternative energy source to replace fossil fuels and reduce the environmental pollution. Adding biodiesel in fossil diesel can increase the oxygen content (from fatty acid) and promote fuel to be burned more quickly and thoroughly. However, the biodiesel content criterion of different countries was diverse from each other. In this study, Raman spectroscopy was used as a tool in classifying fuel blends and quantifying biodiesel contents. For classifying the fuel blends, principal component analysis (PCA) method was employed, where 87.22% of spectral variation was characterized by the first two components PCA scores shows a clear discrimination between the pure fuels and mixture fuels. Meanwhile, for identifying and quantifying the blends of diesel and biodiesel, Raman spectroscopy analysis based on partial least squares (PLS) regression was conducted. Biodiesel mainly present three characteristic Raman regions corresponding to the spectroscopy of diesel. The C─H Raman region presents the better quantitative capacity than the C═C and C═O spectral regions. And the PLS regression built from C─H Raman spectral region in quantifying biodiesel contents presents a higher correlation coefficient and lower root mean square error for prediction. Furthermore, employing only C─H Raman region coupled with PLS regression for predicting concentration of biodiesel can reduce an order of magnitude of root mean square error compared with using three characteristic Raman spectral regions together. Our result show that Raman spectroscopy combined with PCA and PLS can identify fuels and biofuels for discrimination and quantitation.