Enhancing Techno-Economic Assessments in Aeronautic Product Development with Systematic Uncertainty Management

Kurzfassung

This thesis investigated the enhancement of transparency and reproducibility in technoeconomic assessments (TEAs) for aeronautical product developments when input parameter uncertainties are present. The primary objective was to overcome identified barriers in the adoption of a systematic uncertainty management methodology. These included methods for the separation of relevant and negligible uncertainties, the application of Dempster-Shafer Theory of Evidence (DSTE) under data scarcity, as well as the combination of epistemic (knowledge-based) and aleatory (variability-based) uncertainties. By linking these barriers with systematic and comparative analyses, the findings of this dissertation provide a robust framework for effective uncertainty management in TEAs, promote the field of innovative aeronautic product development, and improve decision-making processes under uncertainty. To illustrate the developed uncertainty management methodology, a recurring case study on the lifecycle-based TEA of Hybrid Laminar Flow Control (HLFC) was utilized, drawing on information from two European projects. This case study served as a realistic and interdisciplinary example to demonstrate the quantification of input and output uncertainties, as well as other UQ methods addressed in this thesis. A significant contribution of this dissertation was the investigation of the strengths and weaknesses of various Global Sensitivity Analysis (GSA) techniques, which quantify the individual criticality of parameter uncertainties. Unlike conventional approaches that often select GSA methods without clear criteria, this research systematically assessed their capabilities, interpretability, and computational efficiency. The identified and partially significant differences underscore the necessity for an informed and context-specific selection of GSA techniques. Additionally, the Python package dste was developed to address the need for user-friendly programming toolboxes for handling DSTE-based UQ. Related analyses demonstrated the capabilities of the package and discussed the application of DSTE through systematic expert interviews and theory-specific UQ metrics. Furthermore, the associated interpretation difficulties, particularly concerning the recipients of the TEA, and the challenges related to computational efficiency were examined. The research also explored methods for combining epistemic and aleatory uncertainties and proposed a novel approach that integrates DSTE-based and probabilistic UQ approaches using nested Monte Carlo simulations. This approach enhances interpretability and computational efficiency compared to a purely evidence-theoretic approach and provides a nuanced representation of uncertainties. Decision-makers benefit from clearer insights through understandable visualization and straightforward interpretation, while users can derive tailored recommendations due to the clear separation of epistemic and aleatory effects. Additionally, this approach offers repeatability, allowing UQ to be consistently applied and repeated throughout the product development process as new information becomes available.