Turbo-Machinery Design based on Multi-Physics Fluid-Structure Optimization

Automatized Aerodynamic Design via Optimization

Improving efficiency of fan systems is of great interest in the automotive industry. Turbomachine optimization is becoming a key technique to improve fan systems guarantying good performances in terms of aerodynamics, structure and vibration as well as the reduction of noise.
This paper aims to show the geometric parameterization and mesh morphing capability of a large number of parameters for the design and optimization of turbomachine. It could apply for a wide range of cases as demonstrated by realistic examples from industry. It also shows the use of multi-physics approach such as fluid-structure interaction in the design and optimization of fans. Up to fifteen parameters such as chords, stagger and sweep angles, camber, profile thickness have been used on a single blade to control the blade geometry. These independent parameters are set on an initial mesh, and a morphing technique drives the blade modification regarding the parameter combination. This allows designing various types of complex blade shapes with smoothness mesh quality verification during the morphing process. The results are discussed with particular attention to the static deformations under centrifugal effect, the eigenmodes as well as the displacements generated. The shape deformations induced by these parameters variations is transposed to the Computational Fluid Dynamic (CFD) mesh. Steady and unsteady RANS simulations are performed to predict the aerodynamic performances and to extract the pressure field variation in time on the fan. A structural analysis can then be conducted in order to calculate the centrifugal forces and the harmonics for the structural dynamic response of the rotor. The structural noise and its sound radiation is then evaluated.
Results obtained have allowed evaluating the set of parameters that influence the structural noise on the blade. The optimization process is then conducted by performing an automatic differentiation of the selected relevant parameters which enables to build database containing high order values of these parameters for decision support.

Keywords: Mesh morphing and parametric models for the rotor-stator structure, optimization, turbomachine, Eingenmodes calculation in a rotational domain with centrifugal and aerodynamic loadings, vibro-acoustics; Unsteady CFD simulation of fans.