High-fidelity Numerical And Experimental Study Of Corner Separation In High-load Compressor

Corner separation is one of the main sources of flow losses in compressor,and it has a huge impact on the flow field stability.It is crucial to understand the corner separation for improving the high-loading compressor’s performance.However,digging the physical mechanism of corner separation is limited by RANS method and time-averaging experimental methods,as well as conventional data analysis methods.To solve the bottlenecks,high-fidelity numerical methods which are delayed detached eddy simulation(DDES)and detached eddy simulation(DES),referred as DES-series,are developed for complex flow fields of the turbomachinery;besides,an unsteady pressure experimental method is developed for measuring the unsteadiness in boundary layer.Advanced numerical and experimental measurement methods can resolve the flow field in corner separation region with high resolution both in in time and space.Advanced data mining technology called dynamic mode decomposition(DMD)is introduced into analyzing the flow field in the corner.The prime motivation of the thesis is re-analyzing the spatiotemporal evolution mechanism of corner separation with more advanced methods,and dig the causes of instability in the corner separation.The main contributions of this thesis are listed as follows:1.DDES and DES numerical simulation methods are proposed for turbomachinery flow field,which is characterized by complex geometry,high Reynolds number,and high adverse-pressure gradient.A gird-independence criterion considering unsteadiness is given according to those characteristics.Parallel algorithms including MPI and Open MP are developed to improve the efficiency.The "CFL number increment method" is proposed,which can ensure the numerical stability.The accuracy of the numerical method is verified using flow fields with high subsonic velocity,transonic velocity and adverse-pressure gradient.The comparison with various experimental results such as static pressure hole,oil flow and pressure sensitive paint shows that the DES-series method developed in the thesis is superior to RANS in fine capture of large separation flows2.Considering the aerodynamic characteristics of turbomachinery,including widespectrum,multi-scale,unsteady and highly nonlinear coupling,POD and DMD are innovatively applied to complex flow field analysis.The relationship between matrix factorization and decoupling of flow field is established.The comparison analysis shows that the advantage of POD is that it can extract the flow structure with different unsteady energy,while as for the complex coupling flow field,the DMD is more suitable.3.Considering geometric characteristics of the corner of compressor,which are narrow space,thin-wall blade and large turning angle,a "non-embedded measurement method" based on a micro-pressure-sensor is proposed.An unsteady pressure measuring strategy for compressor corner separation is developed using the non-embedded measurement method.The unsteady pressure measurement in the range of 1k Hz ~ 7k Hz in the corner realized after extracting the signals for low signal-to-noise ratio experimental data.4.Based on the developed high-fidelity numerical and experimental method,ample unsteady data in the corner region of high-subsonic compressor cascade is obtained.With the help of DMD data mining method,a complex flow field is decoupled spatiotemporally.The dominating fluctuation vortex structures in and out of the boundary layer in the corner separation region are detected,and the reason for the formation of these structures are also analyzed.Furthermore,the inherent unsteady behavior and the trigger of flow field instability in the corner separation is determine.