Prediction of properties of diesel/biodiesel blends by infrared spectroscopy and multivariate calibration

Partial least squares models (PLS) using near and middle infrared spectrometry were developed to predict quality parameters of diesel/biodiesel blends (density, sulphur content and distillation temperatures). Practical aspects are discussed, such as calibration set composition; model efficiency using different infrared regions and spectrometers; and the calibration transfer problem. The root mean square errors of prediction, employing both regions and equipment, were comparable with the reproducibility of the corresponding standard method for the properties investigated. Calibration transfer between the two instruments, using direct standardization (DS), yielded prediction errors comparable to those obtained with complete recalibration of the secondary instrument. (c) 2009 Elsevier Ltd. All rights reserved.