Against the background of the high development status of modern axial compressors, a further performance enhancement is linked with the extension of the design space in the development process and the concentration on the essential loss mechanisms in the compressor. The performance of a compressor cascade is considerably influenced by secondary flow effects in the near endwall region, since up to 50 percent (for low aspect ratio) of the losses in the bladed channel of a turbomachinery are linked to the endwalls. In this context the application of non-axisymmetric profiled endwalls provides a potential for compressor improvement. The paper presents the detailed experimental and numerical investigation of controlling the endwall cross flow in a compressor cascade. The general approach is based on a boundary layer fence arrangement, whose application on the compressor endwall works as a non-axisymmetric endwall contour. This non-axisymmetric endwall modification constrains the interaction of the endwall cross flow with the suction side boundary layer, thus the onset of the corner separation is delayed and a significant loss reduction of 8 percent is achieved. The experiments were carried out in a linear compressor cascade at the high-speed cascade wind tunnel of the DLR in Berlin at peak efficiency (design point) and off-design of the cascade at Mach number M = 0.67. Furthermore, high fidelity 3D-RANS flow simulations were performed in order to analyze the complex blade and endwall boundary layer interaction. The combined consideration of experimental and numerical flow pattern allows a detailed interpretation and description of the resulting flow phenomena.

The large secondary flow area in the compressor hub-corner region usually leads to three-dimensional separation in the passage with large amounts of total pressure loss. In this paper numerical simulations of a linear high-speed compressor cascade, consisting of five NACA 65-K48 stator profiles, were performed to analyze the flow mechanism of hub-corner separation for the base flow. Experimental validation is used to verify the numerical results. Active control of the hub-corner separation was investigated by using boundary layer suction. The influence of the selected locations of the endwall suction slot was investigated in an effort to quantify the gains of the compressor cascade performance. The results show that the optimal chordwise location should contain the development section of the three-dimensional corner separation downstream of the 3D corner separation onset. The best pitchwise location should be close enough to the vanes’ suction surface. Therefore the optimal endwall suction location is the MTE slot, the one from 50% to 75% chord at the hub, close to the blade suction surface. By use of the MTE slot with 1% suction flow ratio, the total-pressure loss is substantially decreased by about 15.2% in the CFD calculations and 9.7% in the measurement at the design operating condition.

An experimemtal investigation in a high speed compressor cascade has been carried out to show the effect of different types of secondary flow suction. In order to get deeper insight into the separated three dimensional flow topology and to determine appropriate suction positions, numerical simulations are performed additionally for the baseline cascade. To obtain the flow solution, an implicit, pressure based solver, elaN3D (by ISTA TU Berlin), is employed in steady RANS mode, whereby the Menter SST-k model is used for turbulence treatment. Both investigations are conducted at Mach number Ma = 0.67 and Reynolds number Re = 560.000. The aerodynamic design condition is used. The examined cascade consists of NACA65-K48 type vanes. The experiments include measurements with four different types of suction geometries plus reference measurements. Total pressure and flow angle measurements in the wake show the flow deflection, total pressure loss and the rise of the static pressure of the cascade. The best suction geometry follows the design of R.E. Peacock, designed for low Mach number cascades, with small changes. Using a maximum suction rate of 2% of the main flow the total loss coefficient was reduced by 23%. In this case the stage efficiency — calculated with a reference rotor — is increased by almost 1%. The vacuum pump energy consumption has been taken into account for this calculation. In another case the suction geometry has been chosen in a way that the suction slot is placed along the sidewall from suction side to pressure side following the wall streamlines. With an increased suction rate of 5% of the main flow, the vortex system in the passage is eliminated and the total loss coefficient is decreased to 0.055, which equals to a decrease of 37%. Taking into account that compressors in aero-engines provide bleed air for the plane’s air system, enormous efficiency increase is possible. For this the air bleed valves need to be redesigned.

An experimental study is presented in which passive and active flow control are combined in a way that they complement and support one other. Secondary flow control using boundary layer fences is combined with a boundary layer suction in a compressor cascade at high Mach numbers. Inflow Mach number of 0.67 and Reynolds number (based on chord length) of 560.000 assure realistic conditions. The cascade, equipped with five stator vanes of NACA65 K48 type is used in an ambient condition measurement environment. Pressure measurements form the basis of the experimental investigations, flow visualization is used to obtain insight into the topology of the flow field. The boundary layer fences installed on the suction side of the vanes create a region of low-loss two dimensional flow in the center of the passage. A region of high flow loss is generated at the side wall between wall and BL fence. This region is treated with through-wall boundary layer suction as used in previous investigations. This helps stabilize the flow near the wall and prevent large separated areas. The total pressure loss is reduced remarkably and the outflow becomes more two-dimensional compared to the reference measurement and even compared to the measurement with suction applied without BL fences. The application of boundary layer fences on flow-suction experiments allows obtaining the same loss reduction gains by using lower amounts of suction.

Numerical and experimental results for a high-speed compressor cascade with secondary flow suction are presented. Steady flow suction of low momentum fluid from the back flow region in the corner between end wall and vane is considered in order to diminish the corner separation. Investigations are performed at the design point with an inlet Mach number of 0.67 and a Reynolds number of 560,000 based on axial chord and inlet velocity. The steady Reynolds-Averaged Navier-Stokes simulations are evaluated against data from the accompanying experiment collected with pitot tubes and Conrad angle probes. Laminar separation bubbles on both suction and pressure surface are observed. Thus, transition from laminar to turbulent flow is respected in the simulations. The uncontrolled base flow case and various suction ratios (ratio of drawn to passage mass flow) are exploited. Additionally, the position of the slot is varied numerically. It is found that relocation of the slot slightly away from the suction surface improves the performance of the flow suction.