TRANSACTIONS OF TIANJIN UNIVERSITY
2016, 22: 555-562 DOI 10.1007/s12209-016-2825-5
Effects of Vortex Generator Jet on Corner Separation/Stall in High-Turning Compressor Cascade* Liu Huaping(刘华坪)1,Li Deying(李得英)1,2,Chen Huanlong(陈焕龙)1, Zhang Dongfei(张东飞)1 (1. School of Energy Science and Engineering, Harbin Institute of Technology, Harbin 150001, China; 2. Beijing Electro-Mechanical Engineering Institute, CASIC, Beijing 100074, China) © Tianjin University and Springer-Verlag Berlin Heidelberg 2016
Abstract:The effects of the vortex generator jet(VGJ)attached at the endwall on the corner separation/stall control are investigated by numerical simulation in a high-turning linear compressor cascade. The results show that the corner separation could be reduced significantly, which results in a wider operation range as well as a more uniform exit flow angle and total pressure profile. At the near-stall operation point, the maximum relative reduction of the total pressure loss is up to 32.5%,, whereas the jet mass ratio is less than 0.4%,. Based on the analysis of the detailed flow structure, three principal effects of the VGJ on the endwall cross flow and corner separation are identified. One is to increase the tangential velocity component opposite to cross flow, thus inhibiting the endwall secondary flow near the jet exit. The second is to suppress the pitchwise extension of the passage vortex as an air fence. The third is to sweep the low energy fluids towards the mainstream on the up-washed side and to transport the mainstream fluids to the endwall to reenergize the boundary layer on the down-washed side. Keywords:vortex generator jet; corner separation; high turning compressor; loss reduction
The three-dimensional flow separation in the corner between the endwall and the suction side is almost an inherent flow phenomenon in the compressor cascade, which is detrimental to aerodynamic efficiency and stability. Especially, at the off-design point with large incidence, the separation would expand abruptly, which results in the corner stall and limits the operation range of the compressor. In order to obtain a higher efficiency and wider operation range, numerous flow control technologies have been proposed to reduce the corner separation/stall. One effective method to constrain the corner separation/stall is the application of the 3D cascade passage design such as swap/dihedral blade and non-axisymmetric contoured endwall. Sasaki and Breugelmans[1] as well as Fischer et al[2] have shown that the swap/dihedral blade could force the low energy fluids to migrate towards the midspan, thus the loading along the span is redistributed and a loss reduction is obtained. The applications of the non-axisymmetric contoured endwall to the
secondary flow control and loss reduction are performed by Harvey[3,4] and Yoichi et al[5]. Their results indicate that the cross flow and the corner separation/stall are suppressed successfully. An alternative method to deal with this flow phenomenon is the use of the flow control technologies such as boundary layer suction and blowing, which has been investigated extensively by Peacock[6], Edward et al[7], Zhou et al[8] and Zhang et al[9]. The basic principles of suction and blowing are to remove the low energy fluids and to add momentum to the low energy fluids, respectively. In recent years, increasing studies of flow control using vortex generators (VGs) in the turbo-machinery have been presented. Generally, there are two kinds of VGs using either passive or active mechanisms. The applications of the passive VGs, which consist of thin solid strips or airfoils, are reported by Hergt et al[10,11], Ortmanns et al[12], Qi et al[13] and Wu et al[14]. The total loss is reduced obviously, but parasitic losses are also generated by the passive VGs. In contrast, the active vortex
Accepted date: 2016-05-24. *Supported by the National Natural Science Foundation of China (No. 51306042). Liu Huaping, born in 1983, male, Dr, lecturer. Correspondence to Liu Huaping, E-mail:
[email protected].
Trans. Tianjin Univ.
generator jets(VGJs), in which the vortex is generated by the jet from the inclined hole mounted inside the wall, are controllable devices. The VGJs could be actuated and switched off independently according to different flow conditions. Moreover, less or even no additional losses are produced while the VGJs are turned off. Therefore, increasing investigations of flow control using steady and unsteady VGJs are performed by different researchers[15-20]. They presented the investigation of separation control in the low-pressure turbine blade using VGJs. The effects of the VGJs on the compressor cascade performance are reported by Zheng et al[21], Gmelin et al[22] and Zander et al[23]. The effects of the jet parameters, e.g., the jet-to-inflow mass flow ratio, actuation location, orientation angle and frequency, are extensively investigated. It has been validated that VGJs could improve the cascade performance significantly by these studies. However, considering the manufacturing constraints and the operational reliability of the flow control techniques, the small thickness of the compressor blade is a critical limitation for the application of the VGJs on the suction side, which requires a complex system of pipes inside the blade, as shown by Simon and Howard[24]. Thus, it is more potentially practical to install the VGJs on the endwall in terms of manufacturability and feasibility. In this paper, a single endwall VGJ is utilized to constrain the cross flow and suppress the corner separation at the near-stall operation point in a high-turning compressor cascade. An extensive numerical investigation is conducted to validate the potential of VGJ for the operation range enhancement and the loss reduction. The development of the induced vortex and its interaction mechanism with the secondary flow are further discussed in detail.
1
Numerical procedures
1.1
Cascade geometry A high-turning linear compressor stator cascade with a camber angle of 60° and axial chord of B=120 mm is used in the numerical investigation. The aspect ratio and the solidity are 0.83 and 1.28, respectively. The Mach number at the cascade inlet is 0.25 and the Reynolds number of the blade chord is around 6.7×105. The details of the cascade performance have been reported by Chen[25]. The dimensions of the vortex generator jet(VGJ) and its orientation are schematically shown in Fig. 1. The —556—
Vol.22 No.6 2016
Fig. 1
Schematic of VGJ in the cascade
VGJ is mounted at a chord length of xj/B= -10%, upstream of the leading edge and the pitchwise distance from the suction side is yj/t=10%, of the pitch. The air is injected through a pipe with a diameter of 5 mm and a length of 20 mm, where the pipe length could ensure the jet direction. Based on the pre-study to obtain the optimal effect of loss reduction, a pitch angle of α=20° and a skew angle of β=0° are fixed in this paper. Concerning the real compressor, one of the most practicable methods to generate the jet is inducing the air from the downstream higher pressure stage, it is more reasonable to impose the total pressure rather than the velocity at the jet inlet. Therefore, a total pressure of 104 825 Pa relative to the atmospheric pressure is applied at the pipe inlet, which is a little higher than that at the cascade inlet. The ratio of the velocity of the jet to that of the incoming flow is approximately 1.2, and the mass flow rate is 0.4%, of the mainstream. 1.2 Computational approach The three-dimensional steady viscous Reynoldsaveraged Navier-Stokes equations are solved using the commercial CFD package ANSYS CFX-14.0. Fig. 2 shows the schematic of the computational grid used in the present study. The computational domain extends 1.5B upstream and 3B downstream from the leading edge. Grid refinement is implemented at the insertion of the pile and the endwall. The total pressure profile tested in the experiment by Chen[25] is imposed to the inlet of the computational domain, where the mass-averaged total pressure is about 103,825 Pa. The total temperature at the inlet is 300 K. A fully developed turbulent flow with a turbulent intensity of 5%, is defined, which is a good approximation for the case with the experimental Reynolds number. A back pressure of 101,325 Pa is used at the cascade outlet. The endwall and the blade surface are adiabatic. The symmetry boundary condition is applied at the midspan.
Liu Huaping et al: Effects of Vortex Generator Jet on Corner Separation/Stall in High-Turning Compressor …
Fig. 2
Fig. 3 Pitch-averaged total pressure loss along the blade height at the outlet
Computational grid
To simulate the severe separation in the high-turning compressor cascade, five different turbulence models are tested and validated against the experimental data for the case at the incidence of i=0°. They are standard k-ε, RNG k-ε, shear stress transport(SST), k-ω and BSL. Since the turbulence models have specific requirements for the near-wall grid spacing, different grids are prepared correspondingly. The total number of the computation grids and the Y+ values are shown in Tab. 1. Tab. 1
Trans. Tianjin Univ.
(a)Experimental result[25]
Details of computation grids
Turbulence model
Number of grids
First cell on the wall
Standard k-ε and RNG k-ε
900 000
y =30—60
SST, k-ω and BSL
1 400 000
y <1
+
+
The pitch-averaged total pressure loss along the blade height is shown in Fig. 3. It suggests that the most obvious difference appears in the region from z/H=10%, to 25%,. Among these turbulence models, the k-ω model overestimates the value of the loss and the SST model presents the smallest loss. The standard k-ε model provides the closest results to the experimental data below z/H=30%,, while the loss near the midspan is underestimated slightly. The total loss coefficient at the outlet predicted by the standard k-ε model and the experiment are 0.067,5 and 0.067,8 respectively, therefore, the discrepancy is only 0.5%,. The limiting streamlines on the suction side are further provided in Fig. 4. It is widely accepted that almost no Reynolds-averaged turbulence model could accurately simulate the flow details absolutely for the severe threedimensional separations in the high-turning compressor cascade, as presented by Chen et al[26] and Zhang et al[27]. In this work, the comparisons of limiting streamlines between the numerical and the experimental results also show that slight differences in the separation region are detected. Nevertheless, the scope and the start point of the separation are well predicted by the numerical simulation using standard k-ε model.
(b)Numerical result
Fig. 4
Limiting streamlines on the suction side
Overall, it could be concluded from Fig. 3 and Fig. 4 that the standard k-ε model overcomes the others by reaching the closest results to the experimental data at the design point. Then the same numerical method is used to simulate the flow field at the near-stall operation points, which are characterized by the serious separation similar to the design point, thus enabling to quantify the influence of the flow control on the loss behavior and cascade performance over the operation range with comparative accuracy, as utilized by Chen et al[26] and Feng et al[28].
2
Results and discussion
2.1
Flow field To demonstrate the influence of the VGJ on the flow field, the limiting streamlines on the suction side and the endwall are presented in Fig. 5. For the baseline case with the incidence of i=5° in Fig. 5(a), severe flow —557—
Trans. Tianjin Univ.
Vol.22 No.6 2016
separation occurs on the suction side. It originates in the corner at about 40%, of the axial chord and takes up 40%, of the blade height near the trailing edge. With the incidence increasing to 6°, as shown in Fig. 5(b), the separation line shifts upstream and the separation region almost extends to the midspan on the rear part of the suction side. Moreover, the flow reverse is also observed on the endwall, which indicates that the corner stall occurs, thus generating significant aerodynamic blockage and loss. The corner stall is further enhanced at i=7° in Fig. 5 (c). Furthermore, the flow topologies for the typical “corner separation” in Fig. 5(a) and “corner stall” in Fig. 5 (b) and (c) respectively are also shown by the red solid lines and dashed lines on the wall. The flow patterns coincide fairly well with the models proposed by Lei et al[29] as well as Yu and Liu[30], which also implies that the numerical method employed in the current study could reveal the important flow structures and mecha-
(a)Baseline case, i=5°
(d)With VGJ, i=5°
Fig. 5
(b)Baseline case, i=6°
(e)With VGJ, i=6°
(c)Baseline case, i=7°
(f)With VGJ, i=7°
Limiting streamlines on the endwall and the suction side
The comparisons of the axial vorticity between the baseline case and the case with a VGJ at i=5° are shown in Fig. 6. The fluid pathlines of the jet are colored by purple. It is noticeable that a strong streamwise induced vortex(IV)with the opposite rotating direction to the passage vortex is generated by the interaction between the jet flow and the incoming flow of the cascade. In addition, due to the high vorticity of the streamwise vortex, a counter-rotating vortex(CV)is induced on the top of this vortex. —558—
nisms in the cascade. In Fig. 5(d) and Fig. 5(e), it is noticeable that the separation line on the suction surface shifts downstream by using VGJ. Especially, the secondary flow topology on the endwall is changed distinctly compared with the baseline cases. Near the leading edge, the jet introduces additional tangential momentum into the flow opposite to the endwall cross flow, thus inhibiting the endwall secondary flow, which should be attributed to one of the effect mechanisms of VGJ. In the rear part of the passage, the flow separations are remarkably reduced. At i=6°, no reverse flow is observed on the endwall, hence the onset of the corner stall is prevented. So similar flow topologies are observed on the wall at i= 5° and 6°. When the incidence is increased to i=7°, slight corner stall occurs near the trailing edge. Therefore, it could be concluded from Fig. 5 that a wider operation range is obtained by using endwall VGJ.
(a)Baseline case
Fig. 6
(b)With VGJ
Distribution of axial vorticity in the cascade at i=5°
Liu Huaping et al: Effects of Vortex Generator Jet on Corner Separation/Stall in High-Turning Compressor …
Fig. 7 presents the schematic of the interaction mechanism between the induced vortex and the secondary flow in the near endwall region. On the front part of the endwall, an obvious separation line is observed at the bottom of the up-washed region between the induced vortex(IV)and the passage vortex(PV), which indicates that the low energy fluids near the separation line gather together and then are transferred to the mainstream. In the down-washed region near the suction side, the mainstream fluids are swept towards the endwall. As a result, the boundary layer close to the suction side is reenergized and the abilities of the fluids against the adverse pressure gradient are enhanced in the corner, which contributes to the reduced flow separation in the rear part of the passage as shown in Fig. 5. Along the streamwise direction, both the passage vortex and the induced vortex are deflected towards the suction side by the cross flow pressure gradient. Hence the separation line on the endwall is bended and eventually intersects with the suction side. Downstream of the intersection point, the low energy fluids generated on the endwall interact with the boundary layer on the suction side, thus generating significant loss. Owing to the migration of the low energy fluids on the endwall, the induced vortex is forced to rise away from the endwall. Meanwhile, the cross flow and the migration of the passage vortex are blocked and delayed by the counteractive effect of the induced vortex, which suggests that the induced vortex here works as an air fence. As developing downstream, the core of the vortex expands while the intensity is decreased and then dissipates completely near the trailing edge, which is attributed to the mixing processes between the induced vortex and the boundary layer.
Trans. Tianjin Univ.
of the outlet of the jet compared with the baseline case. Furthermore, since the endwall secondary flow is suppressed and the boundary layer near the suction side is reenergized by the induced vortex, the axial velocity in the corner region is increased significantly. Therefore, the backflow near the suction side in the rear part of the passage is appreciably reduced in comparison to the baseline case. However, in the up-washed region, an axial velocity deficit is examined, which is associated with the strong mixing process caused by the transportation of the low energy fluids to the mainstream.
(a)Baseline case
Fig. 8
(b)With VGJ
Distribution of axial velocity in the cascade at i=5°
2.2
Loading of the cascade Fig. 9 illustrates the pitchwise distribution of the static pressure coefficient on the endwall at different axial chord sections, where y/t=0 denotes the suction side and y/t=1 denotes the pressure side. Compared with the baseline case, the momentum enhancement in the endwall boundary layer by the induced vortex results in a lower pressure on the endwall at x/B=20%, of the axial chord, which is more obvious near the suction side. A similar trend is observed at x/B=50%, except the region from y/t=15%, to 25%, of the pitch, corresponding to the up-washed region shown in Fig. 5. At x/B=80%, of the axial chord, the reduction of blockage leads to an enhanced flow diffusion; therefore, a higher pressure is obtained.
Fig. 7 Schematic of the interaction mechanism between induced vortex and cross flow at i=5° Fig. 9 Pitchwise distribution of static pressure coefficient The distribution of the axial velocity at different axon the endwall at i=5° ial chord sections is shown in Fig. 8. In the front part of The static pressure coefficients at different blade the passage, while the air is injected to the passage at a high speed, a lager axial velocity is obtained downstream height are shown in Fig. 10. It is clearly evident that the
—559—
Trans. Tianjin Univ.
Vol.22 No.6 2016
loading of the blade in the near-endwall region is remarkably affected by the VGJ. At z/H=2%, and 20%, of the blade height, the pressure on the suction side from the leading edge to about x/B=60%, of the axial chord is reduced while it is increased near the trailing edge. However, only slight changes are observed on the pressure side compared with the baseline case. Overall, it could be deduced from Fig. 9 and Fig. 10 that the influence of the VGJ on the loading of the cascade mainly lies in two aspects. Firstly, in the front part of the cascade passage, while the endwall boundary layer is reenergized, the flow acceleration is enhanced and thus the loading of the blade is increased. Secondly, owing to the reduction of blockage in the rear part of the passage, a higher diffusion is obtained. Fig. 11 shows the distribution of the pitch-averaged exit flow angle along the blade height at the outlet, where the angles below zero and above zero represent the overturning and under-turning of the flow respectively. At i= 5º shown in Fig. 11(a), the flow turning at the midspan is
(a) z/H=2%,
Fig. 10
almost the same as that without VGJ. From z/H=10%, to 35%, of the blade height, the under-turning of the flow is reduced significantly. The maximum increase of the exit flow angle is approximately 7° at z/H=18%,. In the nearendwall region below z/H=10%,, the over-turning angle is increased slightly. Consequently, the loading of the cascade is improved dramatically. Moreover, it is noticeable that a more uniform spanwise exit flow angle is obtained by employing the VGJ. Therefore, it could be deduced that a favorable incoming flow angle profile is provided for the downstream cascade in the multi-stage compressor. Fig. 11(b)implies that for the baseline case with the incidence of i=6°, the corner stall leads to severe underturning from z/H=15%, to the midspan. While the VGJ is applied, the separation reduction provides a considerable enhancement of flow turning. The maximum increase of the exit flow angle is about 10°, which appears at about z/H=20%, of the span. As a result, a higher loading and a favorable exit flow angle are gained at this operation point.
(b) z/H=20%,
(c) z/H=50%,
Distribution of pressure coefficient on blade surface at i=5°
(a) i=5°
(b) i=6°
Fig. 11
2.3
Pitch-averaged exit flow angle
Loss reduction Fig. 12 presents the total pressure loss in the cascade. Corresponding to the axial velocity deficit in the up-washed region as shown in Fig. 7, the losses in this region are increased while the VGJ is used. Nevertheless, in the corner region between the suction side and the endwall, it is clear that the scope of the region with high total pressure loss is reduced dramatically compared with the baseline case. The pitch-averaged total pressure losses at the outlet of the cascade for different incoming flow incidences are —560—
provided in Fig. 13. In order to take the mass and the kinetic energy of the jet into account, the total pressure loss coefficient is calculated as follows:
(m0 P0* m1 P1* ) (m0 m1 ) P2* m0 ( P0* P0 ) m1 ( P1* P1 )
where m is the mass flow rate; P* and P are the massaveraged total pressure and the mass-averaged static pressure, respectively; the subscripts 0 and 2 represent the inlet and the outlet of the cascade, and the subscript 1 denotes the inlet of the jet pipe.
Liu Huaping et al: Effects of Vortex Generator Jet on Corner Separation/Stall in High-Turning Compressor …
From Fig. 13(a), at i=5°, the difference between the case with and without VGJ is almost negligible near the midspan, while from the endwall to z/H=40%, of the blade height, the losses are reduced remarkably by the use of VGJ. For the baseline case at i=6°, the corner stall results in significant losses in the cascade. In contrary to this, since the onset of the corner is prevented by the induced vortex, the losses are reduced substantially along the entire blade height. Therefore, the benefit of the VGJ to the loss reduction in the corner region overcomes the loss increase in the up-washed region, which results in a considerable total loss reduction. Furthermore, Fig. 13 also indicates that a more uniform total pressure along the blade is obtained, and then the total pressure distortion for the downstream cascade inlet could be reduced significantly.
(a)Baseline case
(b)With VGJ
Fig. 12 Distribution of the total pressure loss coefficient in the cascade at i=5°
3
Trans. Tianjin Univ.
Conclusions
The numerical investigation of corner separation/ stall control by the use of the vortex generator jet is carried out in this paper. The cascade flow patterns and the development of the streamwise vortexes are discussed in detail. It is shown that the presence of the induced vortex could change the secondary flow and improve the performance of the cascade significantly. The influence of VGJ on the endwall secondary flow could be attributed to three kinds of mechanisms:(1)The tangential velocity component opposite to the cross flow could inhibit the endwall secondary flow near the jet exit.(2)The induced vortex enhances the mixing between the boundary and the mainstream, including transporting the low energy fluids to the mainstream on the up-washed side and sweeping the mainstream fluids to the endwall on the down-washed side.(3)The induced vortex blocks the cross flow and passage vortex as an air fence. Due to the above effects, the corner separation and the loss are reduced significantly. At the near-stall operation point with the incidence of i=5°, the total pressure loss is reduced by 32.5%, in comparison with the baseline case. Moreover, the onset of the corner stall is delayed by the use of VGJ, which obtains a wider operation range. Additionally, the reduction of the passage blockage also contributes to a more uniform exit flow angle and total pressure along the blade height, which could provide a favorable incoming flow condition for the downstream cascade. Nomenclatures B — Axial chord length; H — Blade height; Cp— Pressure coefficient;
(a),i=5°
y+— y plus; i — Incidence; t — Pitch; x, y, z — Coordinates; α — Pitch angle; β — Skew angle; θ — Exit flow angle; ω — Total pressure loss coefficient; Ω — Vorticity.
References (b),i=6°
Fig. 13 Pitch-averaged total pressure loss coefficient along the blade height at the outlet
[1] Sasaki T, Breugelmans F. Comparison of sweep and dihedral effects on compressor cascade performance[J]. Journal of Turbomachinery, 1998, 120(3): 454-464.
—561—
Trans. Tianjin Univ. [2] Fischer A, Riess W, Seume J R. Performance of strongly bowed stators in a four stage high-speed compressor[J]. Journal of Turbomachinery, 2004, 126(3): 333-338. [3] Harvey N W. Some effects of non-axisymmetric end wall profiling on axial flow compressor aerodynamics. Part I: Linear cascade investigation[C]. ASME Paper, GT200850990, 2008. [4] Harvey N, Offord T. Some effects of non-axisymmetric end wall profiling on axial flow compressor aerodynamics. Part II: Multi-stage HPC CFD study[C]. ASME Paper, GT2008-50991, 2008. [5] Yoichi I, Toshinori W, Takehiro H. Effect of endwall contouring on flow instability of transonic compressor[J]. International Journal of Gas Turbine, Propulsion and Power Systems, 2008, 2(1): 24-29. [6] Peacock R E. Boundary-Layer Suction to Eliminate Corner Separation in Cascades of Aerofoils[R]. Aeronautical Research Council Reports and Memoranda, 1971-3663, 1971. [7] Edward P B, Dennis E C, Khairul B Z. Application of synthetic jets to reduce stator flow separation in a low speed axial compressor[C]. AIAA Paper, 2008-0602, 2008. [8] Zhou Y, Liu H X, Zou Z P et al. Boundary layer separation control on a highly-loaded, low-solidity compressor cascade[J]. Journal of Thermal Science, 2010, 19(2): 97104. [9] Zhang L X, Chen S W, Xu H et al. Effect of boundary layer suction on aerodynamic performance of high-turning compressor cascade[C]. ASME Paper, GT2013-94907, 2013. [10] Hergt A, Meyer R, Müllermm M W et al. Loss reduction in compressor cascades by passive flow control[C]. ASME Paper, GT2008-50357, 2008. [11] Hergt A, Meyer R, Engel K. Effects of vortex generator application on the performance of a compressor cascade[J]. Journal of Turbomachinery, 2013, 135(2): 21-26. [12] Ortmanns J, Pixberg C, Gümmer V. Stator endwalls numerical investigation of vortex generators to reduce crosspassage flow phenomena in compressor[J]. Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy, 2011, 225: 877-885. [13] Qi L, Zou Z P, Wang P et al. Control of secondary flow loss in turbine cascade by streamwise vortex[J]. Computers & Fluids, 2012, 54: 45-55. [14] Wu P G, Wang R G, Guo F F et al. Mechanism analysis of effects of vortex generator on high-load compressor cascade[J]. Journal of Propulsion Technology, 2016, 37(1): 49-56. [15] Gompertz K A, Pluim J, Bons J P. Separation control authority of vortex generating jets in a low-pressure turbine cascade with simulated wakes[C]. AIAA Paper, 2009-377, 2009. [16] Memory C, Gompertz K, Chen J P et al. Numerical simulations of vortex generating jets on low pressure turbine
—562—
Vol.22 No.6 2016 blades[C]. AIAA Paper, 2010-90, 2010. [17] Ralph J V, Mounir B I. Separation control on high lift lowpressure turbine airfoils using pulsed vortex generator jets[J]. Applied Thermal Engineering, 2012, 49: 31-40. [18] Florian H, Andreas F, Joerg R S. Modeling vortex generating jet-induced transition in low-pressure turbines[J]. Journal of Turbomachinery, 2014, 136(7): 071005. [19] Liu X M, Zhou H Y. Numerical investigations of flow separation control for a low pressure turbine blade using steady and pulsed vortex generator jets[C]. ASME Paper, GT2010-22587, 2010. [20] Liu X M, Wang X. Performance improvement of a low pressure turbine blade by using synthetic jets[J]. Journal of Engineering Applications of Computational Fluid Mechanics, 2011, 5(4): 445-458. [21] Zheng X, Zhou S, Hou A P et al. Separation control using synthetic vortex generator jets in axial compressor cascade[J]. Acta Mechanica Sinica, 2006, 22(6): 521-527. [22] Gmelin C, Zander V, Hecklau M et al. Active flow control concepts on a highly loaded subsonic compressor cascade: Resume of experimental and numerical results[J]. Journal of Turbomachinery, 2012-061021, 2012. [23] Zander V, Hecklau M, Nitsche W et al. Active flow control by means of synthetic jets on a highly loaded compressor cascade[J]. Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy, 2011, 225: 897-906. [24] Simon E, Howard H. Separation-control mechanisms of steady and pulsed vortex-generator jets[J]. Journal of Propulsion and Power, 2012, 28(6): 1201-1213. [25] Chen H L. Investigation on Aerodynamics Performance of Application Slot Injection Technology in Compound Lean Compressor Cascades[D]. Harbin Institute of Technology, Harbin, China, 2009(in Chinese). [26] Chen S W, Sun S J, Lan Y H et al. Numerical investigations on effects of inlet boundary layer characteristics on aspirated compressor cascade[J]. Journal of Mechanical Engineering, 2014, 50(14): 158-163(in Chinese). [27] Zhang H D, Wu Y, Li Y H et al. Investigation of the vortex structure and flow loss in a high speed compressor cascade[J]. Acta Aeronautica et Astronautica Sinica, 2014, 35(9): 1-12(in Chinese). [28] Feng Y Y, Song Y P, Liu H P et al. Effects of end-wall vortex generator jet on flow separation control in compressor cascade under different angles of attack[J]. Journal of Propulsion Technology, 2015, 36(1): 55-60(in Chinese). [29] Lei V M, Spakovszky Z S, Greitzer E M. A criterion for axial compressor hub-corner stall[J]. Journal of Turbomachinery, 2008, 130(3): 1-10. [30] Yu X J, Liu B J. A prediction model for corner separtion/stall[C]. ASME Paper, GT2010-22453, 2010. (Editor: Liu Wenge)