In the search for renewable, alternative propulsion technologies for use in aviation, hybrid SOFC-jet-engine architectures are seen as very promising. The critical performance characteristics for a fuel cell in such an application differ greatly from those currently in use, however: Volumetric power density, fuel utilisation, and operating temperature are key criteria. One way to increase these characteristics is to use novel stack architectures, such as monolithic or tubular arrays. There is a large variation in exact implementations of such architectures, primarily geometrically-- the size of fluid channels, skewedness, etc. As a first venture into this design- and optimisation space, a numerical model is developed that enables analysis for arbitrary three-dimensional geometries of SOFC stacks at steady-state conditions, implemented with ease of integration in mind. It is released as open source software for anyone to use. This model was extensively validated and verified, after which it was applied by performing sensitivity analyses on geometry parametrisations for three characteristic geometry classes: planar, tubular and corrugated monolithic. It was found that the stack geometry has only minor effects on the thermal performance, but strong implications on power density and fuel efficiency. Strong variations in the effects of similar changes exist between the different geometries. % Drawing % general conclusions for 'all' fuel cells is inadvisable, as it skews % expectations. Instead, individual studies around specific geometries % should be performed. The core design consideration is also more complicated than a mere trade-off of power density and fuel efficiency, with certain parameters yielding improvements in both factors due to strong coupling of disparate processes. Miniaturisation increases power density for all geometries considered, but it is seen to come with diminishing returns. The results also give insight into the process of stack design itself: as an indicator of volumetric power density, current density alone has proven to be an unreliable indicator, despite being used analogously as such in existing literature.