Degradation modeling of polymer electrolyte membrane water electrolyzers for hydrogen production: motivation, status, and strategies

Abstract Polymer electrolyte membrane water electrolyzers (PEMWEs) are an important fixture in the generation of green hydrogen. This paper studies the emerging trends in degradation modeling of PEMWEs. The ability to predict durability and degradation in PEMWEs is key to optimizing their control, design, maintenance, safety monitoring, performance, and lifespan. For the purposes of fault detection, bond graph modeling is a novel approach allowing for precise, expandable monitoring and isolation. Future fault detection work should evaluate the system health by considering elevated degradation rates, as unchecked voltage rise can reduce the useful lifetime of PEMWEs. Performance modeling techniques, such as overvoltage fitting and equivalent circuit models, are commonly employed for diagnostics in cell durability experiments, but both suffer when comparing polarization and electrochemical impedance spectroscopy data. They offer indications of the degradation source but must otherwise be complemented by destructive analyses. Moreover, multi-physics and kinetic modeling are growing trends that seek to explain the dissolution of the ionomer content or catalyst material in membrane electrode assemblies. They have been used to a limited extent in diagnostics studies to explain the differences between the durability of different membranes, but are more notable in prognostics studies. Most recently, artificial intelligence and machine learning techniques—which are increasing rapidly in complexity and robustness—have been used to predict cell performance, adjust residual thresholds in monitoring studies, and increase sensor redundancy in cases of non-critical faults. Control techniques and techno-economic analyses have additionally benefited from the consideration of degradation, although in these fields the models employed have been rudimentary. Considerable progress has been made in modeling PEMWE degradation, but there is still work to be done to understand the impact of dynamic performance, predict long-term performance loss, and bridge the gap between model electrode studies and real-world membrane electrode assemblies.