Pyrolysis of <i>Phragmites australis</i> (common reed): experimental analysis, thermokinetics and modelling via deep neural networks (DNN)

This study provides a comprehensive analysis of the potential applicability of Phragmites australis (common reed) as a viable feedstock for pyrolysis operations. The analysis utilizes thermogravimetric data collected in accordance with the recommendations of the International Confederation for Thermal Analysis and Calorimetry (ICTAC) for kinetic modelling. Experimental measurements were conducted at heating rates of 5, 10 and 20 °C·min−1 across a temperature range of 25–1000 °C, with controlled sample weights maintained at 4.2 ± 0.1 mg. Compositional analysis confirmed the lignocellulosic nature of the sample, while thermal devolatilization studies indicated the evaporation of moisture content and light volatile matter within the temperature range of 25–162 °C, followed by the devolatilization of hemicellulose (192–386 °C) and cellulose (234–406 °C), with the slow degradation of lignin typically occurring above 424 °C. The application of a deep neural network (DNN) model effectively captured the complex characteristics of the thermogravimetric (R2~0.997), derivative thermogravimetry (R2~0.942) and conversion (R2~0.994) traces, following rigorous hyperparameter tuning and modelling adjustments, thereby ensuring that the DNN model learned a more generalized pattern. The utilization of the Coats–Redfern model-fitting kinetic method, which incorporates 26 selected solid-state reaction mechanisms, indicates that the predominant reaction mechanisms for the thermal degradation of Phragmites australis conform to the D3 diffusion (Jander), D5 Zhuravlev, Lesokin, Tempelman and A3/2 Avrami–Erofeev models. The overall activation energy, as estimated via the Flynn-Wall-Ozawa (FWO), Kissinger-Akahira-Sunose (KAS) and Vyazovkin models, is determined to be 109.90 kJ·mol−1, which decreases to 97.3 kJ·mol−1 with the inclusion of the Friedman differential kinetic method. Furthermore, the estimated values of the thermodynamic parameters necessary to achieve activated complexes suggest an endergonic, reactant-favoured and thermodynamically stable operation. It is anticipated that the combined experimental and modelling approaches employed in this study will facilitate the potential application of Phragmites australis as a viable bioenergy source, as well as its prospective use as a biocatalyst accelerator in the processing of bio-composite materials.