A combined analysis of the drying and decomposition kinetics of wood pyrolysis using non-isothermal thermogravimetric methods

Abstract

In this paper, we evaluate the pyrolysis of wood biomass as a combination of drying and thermal decomposition via the Page and modified Page models for drying kinetics as well as the Friedman and Vyazovkin methods for solid-state decomposition kinetics. This approach was applied to data obtained from thermogravimetric analysis of three wood species (spruce, pine, and birch) at 5, 20, and 30 K/min temperature programs.According to the Page model, the average activation energies for spruce, pine, and birch wood between 30 and 150 °C were 12.87 ± 1.08, 13.32 ± 0.48, and 11.61 ± 0.59 kJ/mol, respectively. While all activation energies fell between 11.0 and 14.5 kJ/mol, the modified Page model predicted slightly higher energies, with an average absolute difference of 6.4% from Page's predictions. The activation energies and pre-exponential factors predicted by both models were lower at low heating rates, with the pre-exponential factor yielding significantly large differences between 5 and 30 K/min. These results showed that drying kinetics were significantly affected by heating rates. In addition, the goodness-of-fit analysis revealed that both models were reasonably accurate when predicting wood drying kinetics.For the analysis of solid-state decomposition kinetics, a comparison of Friedman's linear differential method (FR) and Vyazovkin's nonlinear integral method (NLN-INT) was conducted at temperatures higher than 150 °C. In contrast to the NLN-INT method, the FR method predicted activation energies slightly higher, with an average absolute difference of about 8.4%. Evaluation of the relative errors revealed that both the FR and NLN-INT methods performed similarly. However, the Friedman (FR) method provided a reasonable fit to multistep decomposition kinetics through the simultaneous estimation of activation energies and pre-exponential factors. Nevertheless, the activation energies estimated by both the FR and NLN-INT methods were unreliable at conversions of α < 0.15 and α > 0.85. Validation of the kinetic results was conducted with differential thermogravimetric data at a heating rate of 5 K/min.