Hamza Er-rebyiy

ENSAJ, Morocco

Title: Digital Twins for Green Hydrogen Production: From Technology-Resolved Comparison (AWE, AEM, PEMWE, SOEC) to Magneto-Tomography Coupling

---

Biography

Hamza Er-rebyiy is a PhD candidate in energy systems with strong expertise in green hydrogen production, specializing in Proton Exchange Membrane (PEM) and alkaline electrolysis technologies. At the National School of Applied Sciences (ENSAJ), he is responsible for the installation and operation of both electrolyzer systems, complemented by practical experience in energy management and microgrid integration. His expertise combines advanced modeling and optimization techniques with hands-on experience working with real industrial systems. His research focuses on the design and optimization of digital twins for hydrogen microgrids, integrating artificial intelligence (AI)-based control strategies, techno-economic analysis, and real-time Energy Management Systems (EMS). Through his research, Hamza aims to contribute to the development of next-generation planning and optimization tools for sustainable hydrogen production, storage, and grid integration.

Abstract

Water electrolysis powered by renewable electricity is the most scalable route to green hydrogen, yet industrial deployment remains limited by accelerated component degradation, poor response to fluctuating renewable input, and the inability to observe the internal state of an operating cell. Digital twins (DTs) are widely proposed to address these challenges, but existing reviews rarely compare implementations across the four electrolysis pathways—alkaline (AWE), anion exchange membrane (AEM), proton exchange membrane (PEMWE), and solid oxide (SOEC)—or examine their common structural limitation. Using a PRISMA-based Scopus search (2021–2026) with eligibility screening, we show that PEMWE twins are the most mature, combining multi-physics simulation with physics-informed neural networks (R2 = 0.95), while AWE remains largely electrical, AEM relies on offline data-driven surrogates, and no validated integrated twin yet exists for AEM or SOEC. Critically, every implementation reviewed is fed by the same narrow terminal measurements (voltage, current, bulk temperature, flow rate), leaving the internal state structurally underdetermined, a condition we term observational poverty. We propose coupling digital twins with magneto-tomography, which reconstructs internal current distribution non-invasively from boundary magnetic-field measurements (Hall-effect sensor arrays with a Biot–Savart inverse model, about 5% field agreement), through a hybrid architecture combining reduced electrochemical models with an online self-calibrating machine-learning layer. Bibliometric searches on Web of Science and Scopus confirm this coupling has not yet been reported, revealing an open, experimentally tractable research direction relevant to fault localisation, severity quantification, and remaining-useful-life prediction, with SOEC requiring high-temperature sensing above 500 ?C.