In the coming decades, the aviation sector must undergo significant transformations to meet the ambitious targets for CO₂, NOx, and noise emission reduction. Therefore, major research efforts are being undertaken to develop future, revolutionary aircraft that will satisfy environmental requirements, can be operated economically, and will meet the needs of a growing market. Such revolutionary aircraft models will require a complete redesign of current aircraft. These designs leverage novel technologies to reduce structural weight, increase propulsion efficiency, or enable sustainable energy supply systems. A major challenge in the development process of such systems is the sparse database and the lack of knowledge about future technology developments. Simulation models are commonly used to assess the potential and performance of novel systems in an early design stage. However, especially in early design phases, there are large uncertainties in model parameters or in the model form itself, and thus dedicated techniques are required to obtain reliable and robust model predictions. To address these challenges, this thesis develops state-of-the-art uncertainty quantification (UQ) techniques for simulation models of future sustainable and energy-efficient aircraft technologies. Incorporating UQ techniques into the design process enables better model understanding and can guide future research efforts. This thesis presents four studies at the intersection of UQ and aircraft technology applications: First, the mission range of a hybrid-electric aircraft is investigated in the presence of uncertain energy system parameters. Conservative range estimates are predicted for different CO₂ reduction targets using probability bounds analysis. Second, the potential of boundary-layer ingesting engines is analyzed based on a parallel compressor model. A global sensitivity analysis reveals the primary sources of uncertainty, and both variance-based as well as moment-independent studies, yield consistent results. Third, in the field of feedforward gust load alleviation, a novel Bayesian approach for wind field reconstruction is proposed. The reconstruction problem is formulated as a Bayesian inverse problem and automated hyperparameter learning is leveraged to recover the important frequency content of the wind field. Finally, shape-adaptive compressor blades are optimized by determining the optimal position and orientation of piezoelectric actuators. In this context, a robust optimization problem is solved and a novel mono-level approach for surrogate-based robust optimization is introduced. For each application, the UQ studies provide valuable insights into the model behavior and enable engineers to further improve the systems and better predict their potential. This thesis helps bridging the gap between available UQ methods and engineering applications, paving the way for these new technologies to be incorporated into future aircraft.
Julius Schultz (Tue,) studied this question.