Developing a Framework for the Design of Hydrogen Fuel Cell Supply Architectures

Kurzfassung

One of the solutions for reducing emissions from the aviation industry is the implementation of electric or hybrid-electric propulsion systems in the aircraft. The German Aerospace Center (DLR) Institute of Electrified Aero Engines is investigating the feasibility of using fuel cells in hybrid-electric propulsion systems to power regional aircraft. The aim of the thesis is to develop a framework for the design, optimisation, and assessment of hydrogen fuel cell supply architectures for an Solid Oxide Fuel Cell (SOFC) powered aircraft. In addition to this, the implementation of waste reduction methods using a turbine is examined by modelling the air supply system. Key components for a hydrogen fuel cell supply architecture include cryogenic storage tanks, vaporisers, pipes, heat exchangers, pumps, turbines and compressors. Three architectures were defined using these components. Two of them utilised the tank self-pressurisation as the driving force behind the hydrogen flow in the system, with the one extracting gaseous and the other extracting the liquid hydrogen from the tank. The final architecture employed liquid hydrogen extraction using piston pumps. Analytical models for sizing each of these components and modelling the fluid flow through them were implemented in Python scripts. These were integrated in an optimisation loop utilising a local simplex-based algorithm within the RCE integration environment. Each architecture was optimised for two objective functions, the first being for system mass, and the second for both system mass and net power recovered. The optimisation parameters are the geometrical parameters of the components. Its constraints were related to dimensional limitations set by the aircraft and the fluid properties, such as pressure drops and fluid phases, throughout the system. Preliminary component sizings from previous project phases set the intialisation point. Results indicated that the choice of objective function greatly impacts the converged system. For the mass optimisation, the optimiser minimised the mass of all the air side components. On the hydrogen side, a compromise is found between the mass of the component responsible for maintaining the pressure difference for the desired mass flowrate and the mass of the remaining components. When optimising for mass and power, the optimiser reduced air side pressure drops in order to maximise turbine power recovery and minimise compressor power, resulting in a heavier but more power-efficient system. The different architecture layouts primarily affected the system’s parasitic power, which includes additional power required to maintain tank and vaporiser storage pressure, as well as pump power. The implementation of a vaporiser in the liquid extraction architectures reduces this value, as less hydrogen needs to be vaporised in the tank and the mass of hydrogen in the vaporiser is less. Similarly, using a pump allowed for lower tank pressure, reducing both tank mass and power, leading to further reductions in parasitic power. These findings led to the selection of the pump-fed liquid hydrogen extraction architecture as the most suitable for the mission. Post-optimisation analyses included a sensitivity analysis to identify parameters with the largest impact on the converged system. The heat exchanger geometrical parameters had the largest impact due to their significant contribution to overall system mass and pressure losses, which affected other components like tanks and pipes. The analysis also shows that components related to the pressure drop, such as the pipes, become more important when optimising for mass and power. A mission analysis for tank self-pressurisation in the gaseous extraction architecture revealed substantial additional energy required for tank operation, corresponding to a 14.7% system mass increase if stored in lithium-ion batteries. Finally, two additional sets of optimisations were performed with different initialisation points to examine its effect on the results. Although the converged systems differed from baseline optimisations, similar trends were observed.