Aero-Structural Sensitivity Computation Workflow Using the CFD Software by ONERA, DLR and AIRBUS

Abstract

DLR is continuously developing a multidisciplinary optimization suite to accelerate the aero-structural design of aircraft. Emerging challenges such as integrating a large number of constraints based on high-fidelity aerodynamic analyses and advanced turbulence models into optimization, necessitate extending the feature set of the optimization suite and the available solvers. To meet these and other requirements, the next generation computational fluid dynamics (CFD) software by ONERA, DLR and AIRBUS (CODA) is being developed. A key feature of CODA is its algorithmic differentiation, pursued throughout the entire implementation. This capability is particularly advantageous for use in gradient-based optimization, as it enables exact differentiation in direct and adjoint mode for all implemented numerical discretizations, boundary conditions and turbulence models. In this work, the coupled aero-structural gradients are computed using the discrete direct method. Although direct methods are less efficient when the number of design variables exceeds the number of cost functions, they prove to be efficient for problems with many CFD-based constraints, such as local flow feature constraints. The consistency between the direct differentiation as well as the adjoint mode for rigid aerodynamic gradient computation is demonstrated by nearly identical optimization histories and designs. Furthermore, aero-structural coupled aerodynamic shape optimizations are performed. Finally, the methods are demonstrated on the fully coupled shape and structural sizing optimization of the industrially relevant benchmark test case DLR-F25, featuring a flexible high-aspect ratio wing. This paper presents the first aero-structural coupled gradient computation and optimization workflow that uses the CFD solver CODA, performed in conjunction with the structural solver, loads model and optimization framework Lagrange of Airbus Defence and Space.