This paper presents a closed-loop active flow control strategy to reduce the velocity deficit of the wake of a compressor stator blade. The unsteady stator-rotor interaction, caused by the incoming stator wakes, generates fast changes of the rotor blade loading, affecting the stability and the performance of the overall compressor. Negative effects will be seen likewise when unsteady combustion concepts, such as a pulsed detonation, produce upstream disturbances. Furthermore, the periodic unsteady flow leads to additional undesired effects such as noise and blade vibrations. A controlled reliable manipulation of the stator wake is a way to handle these issues. Therefore, investigations on wake manipulation with trailing-edge blowing were carried out on a new low-speed cascade test rig. Detailed information about the wake profile is obtained by five-hole probe measurements in a plane downstream of the cascade for the natural and the actuated flow at a Reynolds number of 6×105. These measurements show a significant reduction of the wake velocity deficit for the investigated actuator geometry with an injection mass flow of less than 1% of the passage mass flow. Based on these results a position in the wake was chosen which is representative for the actuation impact on the velocity deficit. There, a hot-wire-probe measurement serves as the controlled variable. A family of linear dynamic black-box models was identified from experimental data to account for nonlinear and unmodelled effects. Static nonlinearitiy was compensated for by a Hammerstein model to reduce the model uncertainty and get a higher controller performance. To handle off-design conditions, a robust controller working in a range of Reynolds numbers from 5×105 to 7×105 was synthesized. The task of the controller is to rapidly regulate the controlled variable to a reference velocity by changing the blowing amplitude. The synthesized robust controller was successfully tested in closed-loop experiments with good results in reference tracking for pulse series up to 20 Hz. This translates into a much higher frequency when scaled to the dimension of a real machine.
