Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
The accuracy of flutter or forced response analyses of turbomachinery blade assemblies strongly depends on the correct prediction of the unsteady aerodynamic loads acting on the vibrating blades. For unsteady linearized CFD solvers, the quality of the steady-state flow solution constitutes the basis for efficient and accurate CFD unsteady computations. This paper presents the steady-state numerical and experimental results of an annular transonic compressor cascade dedicated to aeroelastic investigations. The measurements were performed in an annular test facility for non-rotating cascades. For both a subsonic and a transonic flow condition, the steady-state flow field measured was simulated and the steady-state static pressure distributions measured on the blade surface were compared to the computational predictions. Especially for the transonic flow condition, results show that the nonlinear effect induced by the presence of a shock in the blade channel requires a detailed computational model taking into account the geometrical features of the experiment. The simulations were performed with two different meshing setups. For a first setup omitting the geometrical complexity of the experimental model and only including the blade passage, results highlight that the steady-state blade loading is not predicted correctly. With a second computational setup including the cascade’s detailed geometry, the relevant physics of the experiment are captured and the prediction accuracy is improved. The presence of leakage flows that arise due to several cascade’s slits and cavities was identified and their impact on the main flow field is discussed. For both flow regimes, the computational setup, the boundary conditions imposed as well as the characterization of the physics associated to the experiment are analyzed to define the optimal level of modeling required to ensure a robust mean flow solution to the initialization of the unsteady CFD procedure.
Naifu Peng, Ho Ling Li, Yue Yang