Laboratory customized online measurements for the prediction of the key-parameters of biomass quality control

Abstract

The European target of ensuring reliable and sustainable energy has led to the increase in biofuel demand. This growth makes necessary the check of the product quality in order to prevent environmental and technical problems during combustion. Technical standard EN ISO 17225 divided the different biofuels into quality classes on the basis of their chemico-physical characteristics and the origin and source. In addition, they define the laboratory methodologies to be performed. These conventional analyses can determine these quality parameters but they are lengthy and expensive compared to the real need of the market. In this study, Vis-NIR spectroscopy coupled with partial least squares regression was used to predict the most important chemical-physical parameters of woodchip and pellet samples as a possible alternative to the conventional laboratory analysis. The results showed the possibility to use spectroscopy to obtain information about biofuel quality. In detail, moisture content and net calorific value of woodchip samples were predicted with RMSEP of 3.78% (r2(pred) = 0.97) and RMSECV of 0.37 MJ/kg (r2(CV) = 0.92), respectively. Ash content and gross calorific value of pellet samples were predicted with RMSECV of 0.44% (r2(CV) = 0.81) and 0.20 MJ/kg (r2(CV) = 0.78), respectively, while ash content and gross calorific value on ground pellet samples were predicted with RMSECV of 0.47% (r2(CV) = 0.78) and 0.19 MJ/kg (r2(CV) = 0.80), respectively. The best results were obtained considering only the near infrared region of the electromagnetic spectrum, suggesting that the visible part is not influential for the prediction of the parameters of this study. Having such a rapid and economic tool will be fundamental for the biofuel processors in order to check different quality characteristics of the products directly in real time without the time delay of the laboratory analysis and complications of sampling representation.