Hydrogen system design presents unique engineering challenges, particularly in high-pressure storage and refueling applications where transient thermodynamic and fluid dynamic effects play a critical role in safety and performance. Accurately simulating hydrogen tank filling requires capturing rapid pressure changes, temperature gradients, compressibility effects, and heat transfer interactions — phenomena that traditionally demand high-fidelity transient CFD simulations with significant computational cost and engineering effort.

As hydrogen infrastructure scales globally, the ability to evaluate tank filling strategies, thermal management approaches, and safety margins earlier and more frequently in the development cycle is becoming essential.

This session presents a Physics AI approach to transient hydrogen tank filling simulation. By combining high-fidelity simulation data with physics-informed machine learning, it becomes possible to predict transient pressure and temperature evolution in hydrogen storage systems with dramatically reduced computation times, while maintaining alignment with validated physical models.

The presentation will explore:

·       The core physical challenges of hydrogen tank filling simulation, including compressible flow, heat transfer, and material considerations

·       How physics-informed AI models can be trained on high-fidelity CFD results to create fast, trustworthy surrogate models for transient analysis

·       How accelerated computing and physics AI frameworks enable scalable training and deployment of such models

·       Practical use cases for evaluating filling protocols, thermal management strategies, and design variants in hydrogen storage systems