1 Department of Wind Energy, Technical University of Denmark2 Test and Measurements, Department of Wind Energy, Technical University of Denmark3 Meteorology, Department of Wind Energy, Technical University of Denmark4 Università degli Studi di Padova
Dynamic behavior of a modified blade fitted onto a small 1 kW vertical-axis wind turbine is studied by two different approaches: Classical modal analysis (EMA) is carried out to validate the results of Operational Modal Analysis (OMA). In traditional modal analysis (EMA) one axis accelerometers are mounted at different points on the projection of the centroid line of the blade structure. Measurements are set up in PULSE LabShop software (product of Brüel & Kjær Company). In each measurement set, one reference point is subjected to an impulse force, and acceleration responses are recorded at three different points. This process continues until the data set contains all the points with their degrees of freedom. Finally the frequency response function (FRF) is obtained for all points, and the natural frequencies and the mode shapes are estimated by peak picking method. Operational Modal Analysis (OMA) is the second approach used in this project in parallel with stereo vision technique. In this method, only the output is required to be measured; actually the input is random and unknown. In this experiment markers are put on the blade centroid projection line (the same place as the accelerometer positions). The 3-D point deflections are monitored in time using stereo vision. Integration is not required for transforming acceleration to deflection in mode shapes identification because we will get deflection directly in this method. Two identical cameras take pictures of the blade and markers while it is excited by random and wind forces. The cameras are programmed in LabView to take pictures at the same time with 180 fps and store them on a high speed hard disk. The output deflection will be investigated in frequency domain by peak picking method, and then AR (Autoregressive) model is applied to describe the structure in time domain. Results of OMA and EMA show good agreement.