Near infrared reflectance spectroscopy and multivariate analysis to monitor reaction products during biodiesel production

In agreement with the principles of green chemistry, near infrared spectroscopy (NIRS) allows multi-component analysis in a fast and nondestructive way, without requiring complex pre-treatments, being a safe, clean and energy saving technique. In this work, a preliminary study to develop near infrared calibration models to predict methyl esters (ME) yield, monoglycerides (MG), diglycerides (DG), triglycerides (TG), free glycerol (FG) and total glycerol (TotalG) content in biodiesel has been carried out. These parameters are considered key factors to determine biofuel quality, its commercialization and to study and monitor the transesterification reaction. For this purpose, samples of biodiesel produced from three different vegetable oils (maize oil, sunflower oil and olive–pomace oil) were analyzed following the EN14103 and EN14105 European standards as reference methods. NIRS calibration equations were validated with a group of validation samples. The mean spectra showed that the main variability on biodiesel NIR spectra occurred around 1700 and 2300 nm. Moreover, the principal components analysis (PCA) applied to the spectra revealed the grouping of samples according to the type of oils used for biodiesel production. The standard deviation of the prediction (cross validation) errors (RMSEPCV) of the calibration models and the standard deviation error (RMSEP) of the validation set resulted similar to the measurement errors (intra lab SELr) and repeatability (inter lab SELR) of each analyte. Results confirm the accuracy of the developed NIRS models for determination of glycerides content and methyl esters yield in biodiesel.