Automated Design Methodology of Turbulent Internal Flow using Discrete Adjoint Formulation

Abstract

An optimal shape design approach is presented for a subsonic S-shaped intake geometry using aerodynamic sensitivity analysis. Sensitivity analysis is performed for the three-dimensional Navier–Stokes equations coupled with two-equation turbulence models using a discrete adjoint method. For code validation, the result of the flow solver is compared with experimental data and other computational results. Through the study on turbulence models and grid refinement, results from several turbulence models are compared and the minimal number of grid points yielding an accurate numerical solution is obtained. And, the adjoint sensitivity code is also validated by the comparison with complex step derivative results. To obtain a sufficient and flexible design space, NURBS equations are employed as a geometric representation and a new grid modification technique, Least Square NURBS Grid Approximation is applied. With the verified flow solver, sensitivity analysis code and geometric modification technique, the optimization of the S-shaped intake is carried out and the enhancement of overall intake performance is achieved. In addition, the off-design performance of a designed S-shaped intake is tested to confirm the robustness of the current design approach. As a result, the capability and efficiency of the present design tools are successfully demonstrated in three-dimensional high Reynolds subsonic intake design.