Mechanism-driven integrated endwall/blade profiling for flow separation control: Optimization, flow physics, and empirical design

Corner separation is a critical flow phenomenon in highly loaded compressors, as it significantly degrades aerodynamic performance. To mitigate the associated flow losses, the manuscript proposes an innovative integrated profiling approach, which is developed based on the conventional end wall profiling and three-dimensional blade profiling method. The new profiling method enables integral construction from the suction surface, via the end wall, to the pressure surface. It not only facilitates smooth transition between the end wall and blades but also enhances the flexibility of the profiling space. First, a new optimization platform based on MATLAB and Python has been developed. The platform integrates hyperparameter optimization for surrogate models and the selection of multi-objective optimization algorithms, effectively addressing the high-dimensional challenges in the optimization process. Second, based on the novel profiling design method and optimization platform, a series of optimized cases were obtained. Compared with the baseline cascade, the optimal shape exhibits a reduction in the loss coefficient by 3.99% and 2.96% on the OP2 and OP1 conditions, respectively. Third, combined with quantitative loss source analysis and qualitative flow analysis, the in-depth flow control mechanism of the optimized profiling is revealed. The integrated profiling mitigates corner separation by modulating the intensity of lateral secondary flows and adjusting the direction of local secondary flows. Finally, based on the secondary flow regulation mechanism and its primary action regions, the design scope of the integrated profiling is narrowed, and a new empirical profiling method is proposed. The empirical method adopts a key profiling unit superposition strategy, which features rapid profiling and high effectiveness. Numerical simulations demonstrate that the empirical profiling method achieves a loss coefficient reduction of up to 5.17%. It not only significantly shortens optimization time but also exhibits excellent performance. This study contributes to advancing the development of corner separation control technology, optimizing the compressor design process, and providing a practical solution for improving compressor performance.