170 likes | 268 Views
EWEA 2011, March 14.-17. 2011 Brussels, Belgium. EXPERIMENTAL INVESTIGATION OF DYNAMIC LOAD CONTROL STRATEGIES USING ACTIVE MICROFLAPS ON WIND TURBINE BLADES O. Eisele, G. Pechlivanoglou, C.N. Nayeri, C.O. Paschereit Hermann Föttinger Institute (ISTA), TU Berlin, Germany. Contents.
E N D
EWEA 2011, March 14.-17. 2011 Brussels, Belgium EXPERIMENTAL INVESTIGATION OF DYNAMIC LOAD CONTROL STRATEGIES USING ACTIVE MICROFLAPS ON WIND TURBINE BLADES O. Eisele, G. Pechlivanoglou, C.N. Nayeri, C.O. Paschereit Hermann Föttinger Institute (ISTA), TU Berlin, Germany
Contents - Motivation - Test Model Configuration - Wind Tunnel & Force Measurement Setup - Experiment Description - Direct Inverse Control - Controller Design - Results - Conclusion
Wind Gusts Tower Shadow Wind Shear Unsteady aerodynamic loads Gravitational Effects Yaw Misalignment Motivation → Large blade deflections → Reduction of the blade lifetime due to fatigue
Motivation • Aim: Reduction of unsteady aerodynamic loads • Solution: Local control surfaces along the span of WT-blades • Adaptation of the aerodynamic characteristics of the blade • Common Solutions: Deformable flaps, Microtabs, rigid flaps • Problems: sensors, controllers required • Scope of the Project: • Evaluation of dynamic lift load reduction potential using rigid TE-Microflap • PID–Control vs. Direct Inverse Control
Test Model Configuration • Airfoil: AH 93-W-174 • Chord: 60cm; Span: 154cm • Plain rigid flap, hinged at TE • Flap-chord: 1.6%c • Flap-thickness: 0.3%c • Max. flap deflections: • 56.6° to pressure side • 74° to suction side • Actuation with digital servos Max. 74° Max. 56.6° Trailing Edge
Wind Tunnel & Force Measurement Setup • Closed loop wind tunnel placed at ISTA/HFI TU-Berlin • Test section: 2 x 1.41 m² • Nozzle contraction ratio: 6.25 : 1 • Test model mounted on an external 6-component wind tunnel balance
Experiment Description The Scenario: • Airfoil under arbitrary pitching motion in the wind tunnel • Controller determines flap deflection to achieve the reference lift • Sampling Rate: 20Hz • Reynolds number: 10⁶
Experiment Description • AoA-Signal generated from white noise sequence • Mean: 7°; Amplitude: 3° • Pitching rate: 2.2°/sec
Direct Inverse Control • The system to be controlled can be described by: • The inverse model: • The inverse controller: • The function g'-1 is obtained by teaching a neural network based on measured data
Controller Design Direct Inverse Controller PID - Controller • Controller design with NNCTRL-Toolkit • 8000 data samples from closed loop experiment • Teaching: 6500 samples • Validation: 1500 samples • Optimization of the neural network architecture • Final network: • Discrete version of: • Manual tuning: • Step change in reference lift • Observation of measured lift • First estimation: • Ziegler Nichols Method • Fine tuning
Controller Design Validation of the Inverse Model: • Predicted control signal very close to the control signal applied by PID-Controller
Results PID-Control: Time Series
Results PID-Control: Statistical Quantities • Load Reduction Potential: 70%
Results Direct Inverse Control: Time Series
Results Direct Inverse Control: Statistical Quantities • Load Reduction Potential: 36.8%
Conclusions • High potential for dynamic lift load reduction using TE-microflaps • In case of PID controlled microflap: 70% • In case of DIC controlled microflap: 36.8% • Unstable behavior of DIC, very active control signal • High performance of neural networks for dynamic system modelling • Further neural network based control approaches proposed
EWEA 2011, March 14.-17. 2011 Brussels, Belgium EXPERIMENTAL INVESTIGATION OF DYNAMIC LOAD CONTROL STRATEGIES USING ACTIVE MICROFLAPS ON WIND TURBINE BLADES O. Eisele, G. Pechlivanoglou, C.N. Nayeri, C.O. Paschereit Hermann Föttinger Institute (ISTA), TU Berlin, Germany THANK YOU VERY MUCH FOR YOUR ATTENTION...