Semi-quantitative and qualitative XRF analyses of alternative and renewable second-generation solid biofuels: Model development and validation

Abstract

New analytical laboratory tools for qualitative analysis of alternative and renewable solid biofuels have been developed. The primary target of this research was to develop and then validate a rapid method for semi-quantitative and qualitative analyses of the second-generation solid biofuels. X-ray fluorescence spectroscopy (XRF) was used in combination with a two-step, multivariate modelling procedure. First, soft independent modelling of class analogies (SIMCA) and classification and regression trees (C&RT) were applied to develop and validate the classifier, which enabled different biomass origins (agrarian biomass, forest biomass and furniture waste) and different possible sources of contamination (plastic, fossil fuels and lignin-cellulose after biomass acid hydrolysis) to be distinguished. Next, the model attempted to predict the concentration of individual components using partial least-squares regression (PLSR) models. In our study, we compared C&RT and SIMCA, and the classification models (algorithms) constructed by the C&RT method were characterised as having better properties than those based on SIMCA. The C&RT classification algorithm was able to predict the origin of biomass sources with a non-error rate greater than 95%. For predictions of the addition type, the non-error rate was greater than 91%. The developed methods can rapidly and adequately determine (qualitatively) the origin of biofuels and indicate possible sources of contamination.