The latest study published in Engineering has unveiled a groundbreaking approach to advancing green ethylene manufacturing, with profound implications for global sustainable chemical production. Conducted by a collaborative team from the University of Sheffield and Southeast University, the research introduces an innovative physically consistent machine learning (PCML)-based hybrid modeling framework for the steam thermal cracking process, which is one of the most energy-intensive and carbon-emitting operations in the petrochemical industry.
Multi-objective optimization has proven to be a promising method for balancing economic benefits and CO2 emissions in several chemical engineering processes. However, when this optimization method is applied to steam cracking processes, modelling faces enormous challenges. Traditional first-principles models offer strong interpretability but are too computationally demanding for optimization because of the complex heat transfer, chemical reaction, and coke formation in the steam cracking process, while purely data-driven models lack the physical robustness needed for reliable decisions.
The research was carried out by an international team led by Professor Meihong Wang from the University of Sheffield and Professor Xiao Wu from Southeast University. This study developed a PCML-based hybrid model that significantly reduced the computational demand for multi-objective optimization from 19.2 hours to 77 seconds. The optimization results demonstrate that dynamic adjustment of operating parameters in response to coke formation can simultaneously enhance profitability and reduce CO2 emissions. Notably, the research reveals that a 28.97% reduction in annual profit could lead to a substantial 42.89% decrease in annual CO₂ emissions. The proposed multi-objective optimization framework comprehensively considers the entire operational cycle of steam cracking process, incorporating the environmental impacts of the decoking process—an aspect frequently neglected in existing studies.
The key findings of this study highlight the great potential for green ethylene manufacturing based on artificial intelligence through modeling and optimization approaches. This study will be important for industrial practitioners and policy-makers.
The paper "Toward Intelligent and Green Ethylene Manufacturing: An AI-Based Multi-Objective Dynamic Optimization Framework for the Steam Thermal Cracking Process," authored by Yao Zhang, Peng Sha, Meihong Wang, Cheng Zheng, Shengyuan Huang, Xiao Wu, Joan Cordiner. Full text of the open access paper: https://doi.org/10.1016/j.eng.2025.06.045.