Global air travel demand continues to grow, accompanied by a rise in carbon emissions, which have surged by 5.5% in 2024 alone. This growth underscores the need for decarbonization in the aviation sector and carbon-free fuels, such as liquid hydrogen (LH2), have emerged as promising solutions for future aircraft. Hydrogen is particularly attractive due to its zero carbon emissions at the point of use, and to its higher specific thrust than conventional propellants. However, when stored in the liquid state, its low working temperatures of about -253 °C pose stringent requirements on thermal insulation, boil-off management, pressure control, and overall system integration. While LH2 has been employed in space propulsion for decades, aviation introduces new challenges, such as longer mission profiles, lower nominal flow rates, and multiple usage cycles within the same day.
The COOL Pipe (Cryogenically Optimized and Operable LH2 Pipe) project, aims to experimentally characterize and to model the heat transfer, phase change mechanisms, and flow behavior of LH2 in aircraft distribution lines. As the hydrogen flow exits the storage tank, in either subcooled or near saturated liquid conditions, heat ingress through the pipe wall drives phase change along the transfer line. Depending on the flow conditions and the magnitude of the heat leak, various boiling regimes may develop, from nucleate boiling to film boiling.
This study presents the development of a one-dimensional thermo-fluid network model to predict these phenomena. The developed model resolves the coupled mass, momentum, and energy balances of the fluid, together with heat transport along the pipe walls. It relies on a limited set of closure parameters, primarily associated with the fluid–wall heat transfer coefficients. The identification of these closure parameters has historically relied on empirical correlations, with limited data points for cryogenic fluids, particularly for aviation-relevant conditions. In this work, the performance of state-of-the-art correlations on benchmark cases from the literature was first assessed. These results are then compared with those obtained from a data-driven identification strategy, in which the heat transfer coefficients are inferred through an implicit approach rather than explicitly extracted and fitted from experimental data.
In the proposed framework, the closure terms are expressed as parametric functions depending on the model’s thermo-hydraulic state and a set of tunable weights. The weights are then determined by solving a non-linear optimization problem, in which the discrepancy between model predictions and available data is minimized. The proposed methodology is demonstrated on a limited set of literature data, serving as a first validation step, and will be subsequently extended to aviation-relevant conditions using upcoming dedicated experimental datasets from the COOL Pipe project.

Keywords
Hydrogen|Modeling|Data-driven|Boiling|Heat transfer