ABSTRACT
A number problems do exist with EMA. If the structure is excited with a fixed shaker then, assuming a two channel analyser is used, the accelerometer must be moved between measurements. Thus the structure is continually changing and this change can be significant in light structures. Aliasing can be a problem if the sampling rate is too low although a low pass filter on the sensor output cures the problem. Performing the FFT over a finite time interval leads to a problem called leakage. Leakage artificially increases the apparent damping in the structure and can produce large errors in the mode shape data. Another problem in correlating theoretical and experimental data is that the measured mode shapes will be complex because the structure will have complex damping mechanisms. Usually real modes are produced using the magnitude of the complex modes but allowing for the phase. The accelerometer and excitation locations should be at nodes of the FE model to ease the correlation process.
