Determination of physicochemical properties of biodiesel and blends using low-field NMR and multivariate calibration

The methods specify by regulatory agencies for determination of the physicochemical properties of biodiesel can be laborious and expensive, so that the development of alternative methodologies represent a major breakthrough. Thus, the low-field NMR (Nuclear Magnetic Resonance) is an advantageous option because is nondestructive and reduce the economic cost and time consumption.1,2 In this study, Partial Least Squares (PLS) regression method was used to create models that correlated the decay curves of CPMG (Carr–Purcell–Meiboom–Gill) signal, obtained in an equipment of low-field NMR, with the kinematic viscosity, the density and the refractive index of biodiesel samples and their blends. Sixteen oilseeds were utilized to synthesize the biodiesel, diversifying between edible and inedible oils, which together with the binary blends totalize forty one samples. The Figure 1 exhibits the correlation between reference and predicted values by PLS. Table 1 shows the statistical parameters that characterize the resulting models. To perform the regressions were used four or five latent variables (LV) and the captured variance was greater than 99.4% in all cases. Furthermore, the strong coefficients of determination (R2) for calibration sets demonstrate the high linearity of the models and the low values of Root Mean Square Error of Calibration (RMSEC) and prediction (RMSEP) indicate their accuracy. These results reveal the built models are very satisfactory and can predict the quality parameters of biodiesel and blends based in CPMG data with fairly efficacy.