Bilevel high-fidelity aero-structural optimization

Abstract

This thesis investigates the applicability of bilevel formulations for coupled aerostructural design optimization of aircraft wings. Bilevel optimization formulations separate an optimization problem into a system-level and lower-level optimization problem, allowing the lower-level problems to be solved independently of each other, while or the commonly used monolithic formulations, interdisciplinary gradient exchange is required , which is prohibited by existing disciplinary organizational structures. Thus, these formulations have recently become of particular interest for the aerospace industry. To expand the understanding of bilevel formulations, this thesis adapts an existing monolithic optimization process to enable the utilization of bilevel processes. The implementation is carried out in the novel optimization framework GEMSEO to facilitate the reusability of processes across different formulation types. The efficacy of the adapted process is evaluated through a comparative analysis with the results obtained from a multidisciplinary feasible formulation. This analysis is performed by testing the adapted process for a wing profile on the simple transonic wing and the DLR-F25 test cases. The findings indicate that bilevel formulations of various types can attain mission fuel burn reductions comparable to those of the multidisciplinary feasible formulation, at approximately 2.5% and 4%, respectively. Despite the similarity in the objective function, the optimized designs exhibit slightly different design characteristics, with the bilevel optimized design tending to be more aerodynamically optimal. To assess the robustness of the bilevel formulations, a starting point variation of the simple transonic wing is performed. Certain bilevel formulations show a dependence on the starting point. However, the use of an implicit wing root bending moment target successfully mitigates this tendency. The results demonstrate the viability of bilevel formulations for coupled aerostructural optimization, enabling industry to implement multidisciplinary optimization while respecting organizational constraints for a distributed aircraft design process.