Enabling liquid hydrogen jet engines through thermal management research — a six-stage research programme running from first-principles architecture definition through to full engine system-level impact.
Liquid hydrogen propulsion is one of the leading pathways to net-zero aviation, but it fundamentally changes a jet engine's thermal balance. This DPhil asks: what thermal management architecture does a liquid hydrogen jet engine actually need, and can the surface heat exchangers within it be understood, validated, and optimised well enough to demonstrate a real improvement in engine efficiency?
This research sits deliberately at the intersection of hands-on aerothermal engineering and applied machine learning — each stage draws on both. The toggle below simply lets you emphasise whichever side is most relevant to you; every stage still contributes to both.
Each stage below builds directly on the one before it — from defining the thermal architecture, to experimentally testing and validating one heat exchanger geometry in detail, to optimising it computationally, to finally closing the loop back at engine level. Click any stage for the full detail.
Through an experimental / systems-engineering lens: this programme runs from system-level thermal architecture through hands-on facility development and rig testing to full engine-level impact — the highlighted stages below carry the most experimental and systems-engineering depth. Click any stage for the full detail.
Through a computational / machine-learning lens: the highlighted stages below are where this research generates real experimental ground truth, validates a physics-based model against it, and then uses machine learning to search a design space CFD alone could never practically cover. Click any stage for the full detail.
Define the thermal architecture, then size heat exchangers from a thermodynamic cycle model across the flight envelope.
Major modifications to an annular wind tunnel, enabling experimental testing under representative outlet-guide-vane installation flow.
Experimentally test the heat exchanger geometry under uniform and distorted (post-OGV) inlet conditions.
Validate a numerical CFD model against the experimental dataset, enabling design exploration beyond physical testing.
Gaussian Process surrogates from an adaptive DoE campaign; Bayesian Optimisation and NSGA-II searches cut evaluation time to milliseconds.
Represent the optimised designs at full engine level to quantify the resulting efficiency improvement.
Stage 3 of this research formed the basis of a paper presented at the AIAA AVIATION Forum 2026, with a journal version currently under review — see Publications. A further paper on Stage 5's surrogate-based optimisation is planned for ASME Turbo Expo 2027, with a journal version targeted for the ASME Journal of Turbomachinery.