Kinetic and thermodynamic analyses of pyrolysis of finger millet (Eleusine coracana) straw through both model-free and model-based methods and application of ANN-based machine learning model to predict thermal degradation

Abstract

Present study provides crucial insights on thermal degradation of finger millet straw (FMS) at various stages of devolatilization and in-depth kinetic and thermodynamic analyses together with artificial neural network (ANN) modeling. A combined approach utilizing both model-free methods, viz. Kissinger-Akahira-Sunose (KAS), Flynn-Wall-Ozawa (FWO), Friedman, Starink, and Vyazovkin, and a model-fitting method, namely, distributed activation energy model (DAEM), was explored to analyze the kinetics. Results envisaged that average activation energy for FMS pyrolysis ranged between 167 and 175 kJ/mol. One-way ANOVA technique revealed no significant deviation in activation energies obtained through various methods. Master plots revealed that FMS pyrolysis followed first order (R1) for α < 0.5 and two-dimensional diffusion (D2), Ginstling-Brounshtein (D4), and power law (P4) models for α > 0.5. Estimated thermodynamic parameters (ΔH ≈ 164 kJ/mol and ΔG ≈ 169 kJ/mol) confirmed viability of FMS pyrolysis. ANN-based model yielded a test R² of 0.994 and low mean squared error of 0.684 for predicting weight loss during FMS pyrolysis.