Comprehensive insights into advanced predictive modeling for low density polyethylene (LDPE) pyrolysis: Experimental...

Low Density Polyethylene (LDPE) waste is a significant challenge for both the environment and industrial management, and this can be converted into valuable fuels by thermochemical processes such as pyrolysis. This study explores comprehensive insights into thermal analysis, pyrolysis mechanism evaluation and advanced predictive modeling to explain the complex degradation behavior of LDPE. Thermal behaviour of LDPE was conducted by Thermogravimetric Analyser (TGA) in the nitrogen atmosphere at heating rates of 2.5, 5, 7.5, and 10 °C/min from room temperature to 1000 °C. For the kinetic analysis, iso-conversional (model-free) methods like Friedman, Kissinger-Akahira-Sunose (KAS) and Ozawa-Flynn-Wall (OFW) methods were applied. Furthermore, the multicomponent Distributed Activation Energy Model (DAEM), which identified two pseudo-components (PC1, PC2), was implemented to analyze the multiple reactions that occur during the pyrolysis of LDPE. Additionally, a pyrolysis experiment was conducted at 500 °C in the fixed-bed reactor. The pyrolysis yielded approximately 48.75 % pyro-oil and 51.25 % non-condensable gases. Findings from Gas Chromatography-Mass Spectrometry (GC-MS) revealed the various compounds in pyro-oil which include the aliphatic and aromatic hydrocarbons, alcohols, ethers and esters. The study further validated the pyrolysis findings through various predictive models, which include the Artificial Neural Networks (ANN), Classification and Regression Trees (C&RT), K-Nearest Neighbors (KNN) and Boosted Regression Trees (BRT). This study demonstrates that machine learning proves to be an effective and reliable approach for modeling LDPE pyrolysis kinetics. This work addresses a significant gap in the literature on LDPE and thus provides a framework that integrates kinetic modeling with the machine learning approach to advance development in the plastic waste conversion.