Co-Pyrolysis of Banana Pseudostem and Plastic Waste by Thermogravimetry: Kinetics and Modeling with Artificial Neural Networks

This study investigates the synergy and performance of copyrolysis between banana pseudostem (BPS) and plastic residues─polyethylene terephthalate (PET) and polypropylene (PP)─using thermogravimetric analysis and artificial neural network (ANN) modeling. Results indicated that all parameters characterizing copyrolysis increased with plastic proportion in the mixture (initial temperature, maximum weight loss rate, and final residual mass). Optimal synergy was achieved at 30 K/min for 25% PET/75% BPS blend, yielding the lowest activation energy (139.7 kJ/mol), and at 10 K/min for the 50% PP/50% BPS blend (133.1 kJ/mol). The ANN model demonstrated high predictive accuracy (R2 above 0.999), with predicted and experimental weight loss data closely aligned with those obtained experimentally, and activation energy results showed less than 5% relative error from experimental data in most cases. These findings underscore ANN’s reliability in generalizing weight loss data and activation energy predictions across various conditions, offering a novel and efficient approach to optimizing pyrolysis processes compared to traditional methods.