The necessary trend towards higher efficiencies and at the same time lower specific fuel consumption brings the design of high-pressure compressors to ever higher pressure ratios and thus leads to highly loaded designs. This highly loaded design will become even more sensitive to the various influencing parameters, like shock-boundary layer interaction or wake boundary layer interaction, so far than today’s conventional designs. This highly complex flow behavior makes both the prediction of existing compressors and the design of new high-efficiency compressors difficult and inaccurate, since the design tools available today take these phenomena into account only insufficiently or not at all. This means that even if the overall compressor behavior with regards to integral evaluation parameters, such as efficiency and pressure rise is predicted correctly, this is not a reliable indicator of the correct physics having been captured. Due to limitations of current turbulence and transition models, the prediction of the above described effects occurring in individual blade rows is often inaccurate, and the overall integral solutions tend to contain a level of model error cancellations.
To improve these models high-resolution modeling data is required. With the constantly-growing computing power available, scale resolving simulations like LES (Large Eddy Simulation) are becoming a feasible tool to generate this validation data. They can provide significantly more complete aero-thermal information and insights into the complex flow phenomena, compared to ever more complex experimental environments that suffer due to limited probe access within the increasingly more complex test rigs for which cost is increasingly a limiting factor. Over the last years LES has already been successfully applied to predicting a variety of turbomachinery problems, such as aerodynamic behavior of high-pressure and low-pressure turbine cascades. Successful LES studies have also included compressor cascades and the influence of different kinds of turbulence phenomena like laminar-turbulent boundary layer transition due to incoming distorted flow or shocks. Besides providing new insights into the underlying physical mechanisms, the generated high-fidelity databases can be used to improve the predictive accuracy of RANS models. In this way, high-fidelity simulations like LES of turbomachinery flows can directly affect industrial design iterations by helping to generate more accurate reduced order models for instance RANS.