Multi-Objective AI Optimization of Plastic Waste Pyrolysis Integrating Energy Return on Investment for Circular...

Area of Science:

• Chemical Engineering
• Materials Science
• Artificial Intelligence

Background:

• Plastic waste accumulation necessitates sustainable recycling solutions.
• Thermochemical recycling via pyrolysis offers promise but requires optimization for energy efficiency and system-level sustainability.
• Conventional optimization often overlooks energy efficiency and overall sustainability.

Purpose of the Study:

• To develop a machine learning-enabled, surrogate-assisted, multi-objective artificial intelligence (AI) optimization framework for plastic pyrolysis.
• To maximize product recovery and minimize energy consumption by integrating Energy Return on Investment (EROI) and Higher Heating Value (HHV).

Main Methods:

• Utilized a dataset of 312 experimental cases for polyolefins, PET, nylon, and mixed plastics.
• Trained machine learning algorithms including polynomial regression, Gaussian process regression, and Random Forest.
• Employed Pareto front analysis with NSGA-II and a conditional variational autoencoder (GenAI) for enhanced exploration of operating regions.

Main Results:

• Random Forest model showed superior predictive accuracy for oil yield, HHV, char formation, and EROI.
• Optimal reaction conditions identified as moderate severities (400-450 °C, 40-70 min) for maximizing net energy and minimizing residues.
• Energy-positive configurations demonstrated superior techno-economic and life-cycle assessment outcomes compared to high-yield scenarios.

Conclusions:

• The developed AI framework provides a data-driven methodology for optimizing plastic pyrolysis.
• The study aligns polymer pyrolysis optimization with circular economy and energy sustainability objectives.
• Energy-positive configurations are key for achieving economic viability (IRR > 15%) and environmental benefits (CO2 reduction up to 47%).