Artificial intelligence-aided liquid hydrogen tank modelling for sustainable aviation

The aviation industry, as a significant contributor to CO₂ emissions, has been under growing pressure to adopt more sustainable and environmentally friendly solutions. Hydrogen, which produces zero carbon emissions upon oxidation, has emerged as a promising fuel for the future of this sector. However, a major challenge in developing hydrogen-powered aircraft is hydrogen’s low volumetric density, which necessitates large storage tanks to hold sufficient fuel for flight missions. These larger tanks add extra weight and reduce available space for passengers and cargo, emphasizing the need for optimal tank designs based on a range of critical parameters. Considering Gravimetric Efficiency (GE) as the targeted index to evaluate tank performance across different designs —specifically, the tank’s ability to store hydrogen relative to its own mass — this work has developed intelligent models that can evaluate the GE of liquid hydrogen tank based on more than 13 key design parameters. The models can analyse the influence of each independent parameter and accurately evaluate the GE of the tank within milliseconds, making them a useful framework for both evaluation and optimization tasks during tank design. In terms of the performance of the models, they demonstrated very close estimation to the constructed dataset with a size of 790,272 points, where the least accurate model estimates the GE with 1.72 % Mean Absolute Relative Error and 99.15 % R-squared. These results underscore the strong potential of AI-based approaches as versatile, fast, and accurate tools for designing hydrogen tanks in future aircraft, offering more generalizable solutions than traditional analytical models while maintaining high prediction accuracy.