Thermokinetic, Thermodynamic, and Pyrolysis Analysis of Agro-Industrial Biomass for Bioenergy: Insights from Isoconversional Models and Neural Networks

Abstract

The growing demand for renewable energy and sustainable waste management has driven interest in the thermochemical valorization of agro-industrial residues. This study investigated the thermokinetic, thermodynamic, and pyrolysis behavior of four lignocellulosic biomasses—coconut shell, orange pomace, banana pseudostem, and compost barn—for bioenergy applications. Thermogravimetric analysis (TG/DSC) and kinetic modeling using isoconversional methods and multilayer perceptron (MLP) neural networks were employed. Kinetic, pyrolysis, and thermal parameters—pre-exponential factor, activation energy, reaction order, initial devolatilization and devolatilization completion temperatures — were determined, along with thermodynamic parameters such as entropy (ΔS), Gibbs free energy (ΔG), and enthalpy (ΔH). All biomasses exhibited typical lignocellulosic degradation profiles, with predominant exothermic events. Activation energy varied widely (140.32–458.35 kJ·mol⁻¹), with banana pseudostem showing the lowest values, indicating easier energy conversion. Banana pseudostem exhibited the highest thermal decomposition index (2.84 × 10^− 6%²·min⁻²·°C⁻³) and the lowest initial devolatilization temperature (183.70 °C). Orange pomace showed the highest ΔH (892 kJ·mol⁻¹) and ΔS (1.31 kJ·mol⁻¹·K⁻¹), suggesting high thermal stability. ΔG values remained relatively constant across conversion degrees. MLP neural networks demonstrated superior performance (R² = 0.99) compared to traditional models in predicting TG curves. These findings support the efficient selection and modeling of biomass for thermal processes and reinforce the potential of AI-based tools in optimizing bioenergy applications.