This thesis aims to propose a possible method for locating and characterising laminar defects in composite beams. Currently, some structural health monitoring techniques do exist, using flexural vibration methods by detecting a change in natural frequencies or mode shapes. However, they do not easily give information about its location and extent. Chang & Chen (2003) illustrated that it was theoretically possible to locate the damage in a cracked beam by using a spatial wavelet transform on the mode shape data. The wavelet transform is a signal processing tool which has undergone a large amount research during the last twenty years. Using a wavelet transform on spatial data such as mode shapes, has the property of plotting a singularity where there is a discontinuity in the mode shape data.
The test beam was modelled as a three segment Timoshenko beam where the middle element represented a change in properties to simulate the damage. The theory used to develop this model was described and using the mode shapes that were plotted from the simulation model wavelet analysis was done. To perform the spatial wavelet analysis the biorthogonal splines wavelet was chosen as the mother wavelet. Using wavelet factors of (3,3) to (3,7) for the mother wavelet the analysis showed that it was possible to locate the defect theoretically as clear singularities were noticed in the decomposition that correlated to the location of the defect.
For the experiment carbon fibre reinforced laminate beam specimens with unidirectional fibres were used. A free- free boundary condition was set up for the beam by placing foam padded supports at the nodes of the vibrating beam and the vibration was excited by using a piezoelectric transducer attached to the end of the beam. The mode shapes were recorded using a laser vibrometer to measure the response along the beam. Three cases of damage were investigated on the vibrating beam. Two of the damage cases consisted of attaching brass masses to the beam causes a simulated inertial damage on the beam. The wavelet transform of the mode shapes from these damage cases did not indicate the location of the damage at all. The third damage case was a low velocity impact damage that had been initiated into the beam using an impact hammer. The first and second mode shapes didn’t indicate the location of the damage, however the third mode shape seemed to indicate the location of the damage when analysed by the wavelet transform.
It was recommended that for future experiments, use must be made of a smoothing function to try and reduce the sensitivity of the method to noise. Also further experimentation must be done on beams containing real damage.