CFDOpt: A Python Generic and Modular Shape Optimization Platform Driven by High Accuracy CFD Analysis Code - S-Duct and 2D/3D wing optimization cases -Inverse design optimization for 2D airfoils

摘要

This paper exposes the development of a python optimization in-house platform CFDOpt driven by the accurate RANS code NES. The program is totally modular in that sense that the treatment for the geometry parameterization, the mesh generation, the objective function estimation, as well as the module for optimization, are totally decoupled from the CFD solver. This high level of modularity is a central key point in this development that permits to easily handle any part of the aerospace platform. Several applications are exposed, considering successively the problem of 2D airfoils drag reduction, internal flow distortion and swirl minimization, 3D wings drag reduction.