Kolloquiumsvortrag, Prof. King,TU Berlin / am 09.07.2018
09.07.2018 von 17:15 bis 18:00
Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel
Titel: Closed-loop active flow control of repetitive disturbances in a linear stator cascade
Abstract: This presentation demonstrates the mitigation of disturbances in a periodically throttled compressor stator cascade by closed-loop active flow control. The work is motivated by the replacement of a constant pressure combustion of a gas turbine with the significantly more efficient constant volume counterpart. One way of implementing a constant volume combustion is a pulse detonation engine. In such an engine, the heat addition takes place in a discontinuous, detonative manner while the inlets to the combustion tubes are closed. As a result, periodic pressure waves will be transmitted upstream into the compressor. A successful exploitation of this new combustion system therefore strongly depends on the ability to dampen negative effects resulting from these disturbances. To this end, active flow control is applied based on two different actuators for the sidewalls and the blade. Blowing fluidic switches are used to excite flow modes to decrease detrimental corner vortices and premature flow separation. For closed-loop control, the repetitive nature of the disturbances can be utilized to increase control performance. Three concepts will be considered here and compared. In iterative learning control (ILC), either in a frequency-based or time-based formulation, un-modeled nonlinear effects and disturbances are learned over the cycles to some extent and used to adapt the control inputs. As an alternative, repetitive model predictive control (RMPC) is introduced which more directly makes use of remaining control errors of a last cycle. RMPC and one of the ILC formulations, which solve respective optimization problems on-line, outperformed the considered frequency-based version of classical ILC with respect to control performance. However, the frequency domain ILC variant might benefit from its robustness features.