A Knowledge-based Engineering Application for Wing–Nacelle–Intake Integrations in Fuel Cell Aircraft

Abstract

This thesis presents the development and assessment of WIN-Gen (Wing–Intake–Nacelle Generator), a knowledge-based engineering tool for the parametric generation and evaluation of outer mould line (OML) geometries for wing-nacelle-intake (WNI) configurations of fuel cell-powered and electrified air craft. The work addresses the growing need for systematic propulsion integration studies enabled by the design freedom inherent to fully electrified propulsion systems. WIN-Gen enables the automated generation of a wide range of WNI configurations and supports early stage assessment of geometric feasibility and indication of aerodynamic behaviour. Using a Model Based Systems Engineering (MBSE) approach, stakeholder needs identified from literature are translated into a structured and verifiable set of system requirements, which guide the tool architecture and functionality.
A key methodological contribution of this thesis is the introduction of the parametrisation-as-composition paradigm. In this approach, the parametrisation of a component is explicitly decoupled from its geometric definition, allowing multiple parametrisation strategies to be interchanged without modifying the underlying component model. Used extensively throughout this work, this significantly improves modularity, reusability, and extensibility of the design framework and enables systematic expansion of the tool to new parametrisations, components and integration strategies. In addition, this work introduces a novel parametric blending strategy, the BlendSolid algorithm, to address the challenge of smoothly integrating complex geometric components. The algorithm enables quasi-tangent blending between low to medium-complexity solids while maintaining intuitive and fully parametric control over the intermediate blending surface. This capability is essential for the robust integration of wings, nacelles, and intakes and forms a critical enabler for automated generation of integrated WNI geometries.
The tool provides fully parametric modelling of the wing, nacelle, and scoop intake and allows independent component design and analysis prior to integration. Through extensive geometric verification, WIN-Gen demonstrates robust behaviour during manual design, where configurations can be interactively corrected. In automatic design generation, a parametric robustness of approximately 30–35% is achieved across a large and realistic design space, while still enabling the generation of a diverse set of distinct and valid geometries. The primary sources of failure are identified as limitations in Gordon surface divergence, geometric blending limitations, and implemented modelling methodology. A geometric validation study reproducing a reference wing-nacelle configuration (TUD-PWF) shows excellent agreement, with an geometrical distance RMS deviation of 0.13% and a maximum deviation of 0.49%. Preliminary aerodynamic validation indicates promising agreement in lift characteristics and local pressure distributions. However, the fidelity of automatically generated aerodynamic results is currently limited by surface meshing quality which is out of scope in this work, indicating the need for further study.
Overall, WIN-Gen constitutes a comprehensive and extendable parametric framework for wing-nacelle intake configurations, capable of generating a wide range of OML designs for fuel cell and electrified propulsion concepts. Nevertheless, limitations remain in the modelling of fully blended and highly integrated WNI concepts, the maturity of the aerodynamic analysis framework for propulsion integration, and the inclusion of innovative intake configurations. Owing to its inherently extensible architecture, targeted future work is expected to address these limitations. Recommended developments include improved geometric algorithms, a comprehensive meshing and aerodynamic simulation study, an expanded design validation effort, and further enhancement and extension of component-level modelling capabilities.