Optimising pre-reforming for quality r-SOC syngas preparation using artificial intelligence (AI) based machine learning (ML)

Reaching CO₂ reduction goals for 2030 and beyond involves cross-sectoral information sharing and transdisciplinary research collaborations. The International Future Laboratory for Hydrogen Economy at the Technical University of Munich (TUM) has proposed a network model for an international green hydrogen economy based on three pillars: electrically assisted gasification, reversible solid oxide cells, and biocatalytic synthesis. Advanced artificial intelligence (AI)-based machine learning approaches are utilised to evaluate and optimise SOC-ready clean syngas preparation. Using DoE, the pre-reforming process of various syngas constellations has been evaluated. The possibility of developing and simulating the complex thermochemistry of syngas-containing fuel using a CFD model of a pre-reformer is approved. The model is utilised to create data for the successful construction and training of an AI-based machine-learning model. A multi-regression study illustrates the interactions between process factors in order to comprehend and improve the fuel outlet conditions of the pre-reforming process, which are essential for the fuel inlet region of solid oxide cells. The ML model and numerical CFD predictions are in excellent agreement showing less than 1% error. It was possible to precisely define the operating parameters for three distinct syngas configurations in order to achieve 500 °C fuel outlet conditions. Consequently, the results contribute to a better knowledge of the preparation of high-quality, pure syngas, furthering sustainable research and r-SOC operation in a safe, consistent manner.