Application of machine learning to model waste energy recovery for green hydrogen production: A techno-economic analysis

This paper presents an innovative energy recovery approach for hydrogen production in landfill waste disposal plants. The proposed scenario integrates water electrolysis with direct methane reforming into hydrogen at a moderate temperature (500 °C) and incorporates a supercritical CO₂ heat pump. This design achieves reforming without relying on external heat sources, enhancing the system's efficiency. Additionally, the study applies machine learning to model landfill gas with a focus on energy recovery potential. Various machine learning algorithms are assessed for accuracy, and the highest-performing models—achieving R-squared values between 92 % and 99%—are benchmarked against existing landfill models, demonstrating improved precision. The landfill model developed in the initial phase serves as input for the energy model. Results suggest that the levelized cost of hydrogen production could be below 2 €/kg H₂ at stack level, aided by internal energy recovery mechanisms that increase production rates. At 500 °C, the methane conversion efficiency aligns closely with that of conventional systems, making this approach a viable and cost-effective alternative.