TY - JOUR
T1 - Time domain-NMR combined with chemometrics analysis
T2 - an alternative tool for monitoring diesel fuel quality
AU - Santos, Poliana M.
AU - Amais, Renata S.
AU - Colnago, Luiz A.
AU - Rinnan, Åsmund
AU - Monteiro, Marcos R.
PY - 2015/4/16
Y1 - 2015/4/16
N2 - Time-domain nuclear magnetic resonance (TD-NMR) was explored as a rapid method for simultaneous assessment of the quality parameters in commercial diesel samples (B5 diesel-biodiesel blend). A principal component analysis (PCA) obtained with the relaxation decay curves revealed tight and well-separated clusters, allowing discrimination of the diesel samples according to the sulfur content: 10 (S10), 500 (S500), and 1800 (S1800) mg kg-1. Classification models based on the soft independent modeling of class analogy (SIMCA) showed a good discrimination power with a percentage of correct classification ranging from 90% (for S500 diesel samples) to 100% (for S10 and S1800 diesel samples). Partial least-squares regression (PLSR) was used to estimate the cetane index, density, flash point, and temperature achieved during distillation to obtain 50% of the distilled (T50) physicochemical parameters in the commercial diesel samples. The best PLSR models were obtained with two latent variables, providing a standard error of prediction (RMSEP) of 0.60, 2.37 kg m-3, 3.24, and 2.20°C for the cetane index, density, flash point, and T50, respectively, which represents the accuracy of the models. The results support the application of TD-NMR to evaluate the quality of B5 diesel, providing a simple, rapid, and nondestructive method for the petrofuel industry.
AB - Time-domain nuclear magnetic resonance (TD-NMR) was explored as a rapid method for simultaneous assessment of the quality parameters in commercial diesel samples (B5 diesel-biodiesel blend). A principal component analysis (PCA) obtained with the relaxation decay curves revealed tight and well-separated clusters, allowing discrimination of the diesel samples according to the sulfur content: 10 (S10), 500 (S500), and 1800 (S1800) mg kg-1. Classification models based on the soft independent modeling of class analogy (SIMCA) showed a good discrimination power with a percentage of correct classification ranging from 90% (for S500 diesel samples) to 100% (for S10 and S1800 diesel samples). Partial least-squares regression (PLSR) was used to estimate the cetane index, density, flash point, and temperature achieved during distillation to obtain 50% of the distilled (T50) physicochemical parameters in the commercial diesel samples. The best PLSR models were obtained with two latent variables, providing a standard error of prediction (RMSEP) of 0.60, 2.37 kg m-3, 3.24, and 2.20°C for the cetane index, density, flash point, and T50, respectively, which represents the accuracy of the models. The results support the application of TD-NMR to evaluate the quality of B5 diesel, providing a simple, rapid, and nondestructive method for the petrofuel industry.
U2 - 10.1021/acs.energyfuels.5b00017
DO - 10.1021/acs.energyfuels.5b00017
M3 - Journal article
SN - 0887-0624
VL - 29
SP - 2299−2303
JO - Energy & Fuels
JF - Energy & Fuels
IS - 4
ER -