Comparison of two pre-fuel cell carbon capture systems employing methane decomposition and hydrogen membrane: Regression-based machine learning and optimization approach

This study examines two systems designed to operate with zero carbon dioxide emissions, each utilizing different methods for heat generation to support methane decomposition for hydrogen production. In the first system, heat is generated through carbon combustion, while in the second system, hydrogen combustion is used. Both carbon and hydrogen are produced within their respective systems and serve as fuels for heat generation. Machine learning techniques are applied to both systems, enabling precise prediction of their performance and costs over the operational cycle. To maximize both efficiency and economic feasibility, a dual-objective optimization strategy is utilized. The results indicate that the H-MD system achieves a higher energy efficiency of 76.32 % and a lower Life Cycle Cost (LCC) of $3.19 million, making it the more cost-effective and thermodynamically efficient option. Conversely, the C-MD system delivers a higher net power output of 350.7 kW but requires greater heat input (94.2 kW) and exhibits lower efficiency (64.62 %) with a higher LCC of $3.73 million. Despite these differences, both systems provide clean, zero-emission energy solutions, with the best selection contingent upon specific operational parameters, including economic considerations, efficiency targets, and the availability of heat sources.