A computational similarity-assisted multi-domain framework for conceptual engineering design

Doctoral thesis, 2025

In this thesis, a set of methods are proposed for assisting design engineers in exploring trade-offs among performance, manufacturability, and sustainability, already during the early design phase. These methods were developed in close collaboration with a Swedish aerospace company to mitigate the risk of costly redesigns later in development. Among the challenges addressed is how to capture and evaluate the necessary information while there is still enough freedom to make changes to the design. Through the application of a new flexibility metric, discrete design spaces often found in the concept phase can rapidly be explored, in search of competitive and future-compatible concepts. Enhanced function-means modeling is deployed to capture system-level interactions, enabling pre-embodiment consideration of trade-offs among performance, manufacturability, and sustainability. Automatic geometry generation is used to generate large batches of design variants that can be evaluated through various simulations. Finally, similarity metrics are leveraged to assist designers in staying within the well-understood regions of the design space, thus avoiding some of the risk of new concept development. These methods have been implemented into a design software suite, enabling them to flexibly be adjusted for different needs, and to be utilized by both designers, academics, and students.

For the aviation industry, these contributions serve as proposals for how to explore and integrate new design concepts. This has, in recent times, become of increased importance due to the ongoing climate crisis, which necessitates rapid development of more sustainable propulsion alternatives. Consequently, an improved understanding of how to efficiently integrate new technology is paramount.