Application of Computational Aerodynamic Analysis and Optimization in a Multi-Fidelity Distributed Overall Aircraft Design System

• Anwendung rechnergestützter aerodynamischer Analyse und Optimierung in verteilten, mehrstufigen Flugzeuggesamtentwurfssystemen

Gu, Xiangyu; Stumpf, Eike (Thesis advisor); Rizzi, Arthur (Thesis advisor)

Aachen (2017)
Dissertation / PhD Thesis

Dissertation, RWTH Aachen University, 2017

Abstract

In this thesis, a multi-fidelity distributed aircraft design process using fully automated Computational Fluid Dynamic (CFD) analysis and optimization is developed aiming to provide better prediction of aircraft characteristics in the early aircraft design stages for both conventional and unconventional aircraft configurations. In order to facilitate the application of CFD analysis in the extensive recursive design process, a fully automated CFD analysis process based on unstructured mesh technology is developed. The process, which includes geometry representation, mesh generation and flow simulation is implemented in the DLR distributed design environment and makes use of the DLR centralized data format CPACS in order to take advantage of other disciplinary tools available in the DLR. The process is verified with well-studied configurations. The results match well with reference wind tunnel and computational data which indicates the automated process is able to provide a reliable aerodynamic performance prediction for both inviscid and viscous flow.In order to tackle the wing shape design variables of the more complex geometry representation required by the CFD analysis, aerodynamic optimization is used to enhance the aerodynamic efficiency for the configuration initialized by the conceptual design tool. A Python script is developed in order to automate the setup and execution of the aerodynamic optimization as well as the results propagation. In this study, Multi-Disciplinary Analysis (MDA) and Multi-Disciplinary Optimization (MDO) are adopted to capture the dependency between different disciplines and trade off the discipline performance to support aircraft design. An MDA process using the automated CFD analysis process and aerodynamic optimization is developed to estimate the aircraft's synthesized characteristics. The MDA process is applied to a conventional aircraft configuration. The results show the process is able to provide a reliable estimation of the aircraft characteristics.At last, a distributed architecture is adopted for an aircraft MDO application. The design variables are decomposed into typical wing planform design variables in conceptual aircraft design and the wing shape design variables. Gradient based optimization is chosen to support the aerodynamic optimization with respect to the wing shape variables. The surrogate model method is chosen to handle the wing planform variables. The MDO process is applied to both a conventional aircraft and an unconventional configuration. The results show the implemented MDO process is able to handle the complicated tradeoff between different aircraft design disciplines for both conventional and unconventional aircraft configurations.