Prediction of elemental composition, water content and heating value of upgraded biofuel from the catalytic cracking of pyrolysis bio-oil vapors by infrared spectroscopy and partial least square regression models

Abstract

The elemental composition, heating value and water content, are important properties to be characterized for pyrolysis bio-oils, providing information on their quality. These properties are mainly determined according to ASTM standards by using three different analytical techniques requiring time and cost. This research was focused on a simple method to determine the weight content of carbon, hydrogen, oxygen and water as well as the heating value, by Fourier transform infrared spectroscopy (FT-IR) using models based on partial least squares regressions (PLS). Samples were classified into two sets according to Kennard-Stone algorithm. The first set of samples was used to develop the calibration models for each physical parameter, where the number of latent variables was determined by full cross validation procedure. The second set of samples was employed as an external prediction set, assessing the quality of the models. External predictions confirmed that robust models were developed since elemental analysis, heating value and water content of the upgraded biofuels obtained by the catalytic cracking of pyrolysis bio-oil vapors could be determined with good predictive ability with a root mean square error of prediction of carbon content = 0.963 wt.% (R2 = 0.836, range = 70.9–78.8 wt.%), hydrogen content = 0.101 wt.% (R2 = 0.815, range = 8.01-8.75 wt.%), oxygen content = 0.910 wt.% (R2 = 0.873, range = 12.3–20.9 wt.%), water content = 0.416 wt.% (R2 = 0.829, range = 2.79–7.24 wt.%) and heating value = 0.539 MJ kg−1(R2 = 0.874, range = 31.9–36.9 MJ kg−1) by chemometric tools joined to medium infrared spectrum.