We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Thermal Degradation of Palm Fronds/Polypropylene Bio-Composites: Thermo-Kinetics and Convolutional-Deep Neural Networks Techniques
Summary
Researchers used thermo-kinetic analysis and convolutional deep neural networks to study the thermal degradation of palm frond/polypropylene bio-composites, demonstrating that combining natural fiber with plastic waste can produce materials suitable for environmentally conscious industrial applications aligned with global sustainability initiatives.
Identifying sustainable and efficient methods for the degradation of plastic waste in landfills is critical for the implementation of the Saudi Green Initiative, the European Union's Strategic Plan, and the 2030 United Nations Action Plan, all of which are aimed at achieving a sustainable environment. This study assesses the influence of palm fronds (PFR) on the thermal degradation of polypropylene plastic (PP) using TGA/FTIR experimental measurements, thermo-kinetics, and machine learning convolutional deep learning neural networks (CDNN). Thermal degradation operations were conducted on pure materials (PFR and PP) as well as mixed (blended) materials containing 25% and 50% PFR, across degradation temperatures ranging from 25 to 600 °C and heating rates of 10, 20, and 40 °C·min-1. The TGA data indicated a synergistic interaction between the agricultural waste (PFR) and PP plastic, with decreased thermal stability at temperatures below 500 °C, attributed to the hemicellulose and cellulose present in the PFR biomass. In contrast, at temperatures exceeding 500 °C, the presence of lignin retards the degradation of the PFR biomass and blends. Activation energy values between 81.92 and 299.34 kJ·mol-1 were obtained through the application of the Flynn-Wall-Ozawa (FWO) and Kissinger-Akahira-Sunose (KAS) model-free methods. The application of CDNN facilitated the extraction of significant features and labels, which were crucial for enhancing modeling accuracy and convergence. This modeling and simulation approach reduced the overall cost function from 41.68 to 0.27, utilizing seven hidden neurons, and 673,910 epochs in 13.28 h. This method effectively bridged the gap between modeling and experimental data, achieving an R2 value of approximately 0.992, and identified sample composition as the most critical parameter influencing the thermolysis process. It is hoped that such findings may facilitate an energy-efficient pathway necessary for the thermal decomposition of plastic materials in landfills.