We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Thermal Decomposition of Date Seed/Polypropylene Homopolymer: Machine Learning CDNN, Kinetics, and Thermodynamics
Summary
Researchers studied the thermal decomposition of date seed and polypropylene plastic composites using both experimental methods and machine learning models. They found that mixing date seeds with polypropylene lowered the temperature needed to break down the plastic and that neural network models could accurately predict the decomposition behavior. The study suggests that agricultural waste like date seeds could help make plastic waste recycling through pyrolysis more energy-efficient.
The buildup of abandoned plastics in the environment and the need to optimize agricultural waste utilization have garnered scrutiny from environmental organizations and policymakers globally. This study presents an assessment of the thermal decomposition of date seeds (DS), polypropylene homopolymer (PP), and their composites (DS/PP) through experimental measurements, machine learning convolutional deep neural networks (CDNN), and kinetic and thermodynamic analyses. The experimental measurements involved the pyrolysis and co-pyrolysis of these materials in a nitrogen-filled thermogravimetric analyzer (TGA), investigating degradation temperatures between 25 and 600 °C with heating rates of 10, 20, and 40 °C.min-1. These measurements revealed a two-stage process for the bio-composites and a decrease in the thermal stability of pure PP due to the moisture, hemicellulose, and cellulose content of the DS material. By utilizing machine learning CDNN, algorithms and frameworks were developed, providing responses that closely matched (R2~0.942) the experimental data. After various modelling modifications, adjustments, and regularization techniques, a framework comprising four hidden neurons was determined to be most effective. Furthermore, the analysis revealed that temperature was the most influential parameter affecting the thermal decomposition process. Kinetic and thermodynamic analyses were performed using the Coats-Redfern and general Arrhenius model-fitting methods, as well as the Flynn-Wall-Ozawa and Kissinger-Akahira-Sunose model-free approaches. The first-order reaction mechanism was identified as the most appropriate compared to the second and third order F-Series solid-state reaction mechanisms. The overall activation energy values were estimated at 51.471, 51.221, 156.080, and 153.767 kJ·mol-1 for the respective kinetic models. Additionally, the kinetic compensation effect showed an exponential increase in the pre-exponential factor with increasing activation energy values, and the estimated thermodynamic parameters indicated that the process is endothermic, non-spontaneous, and less disordered.