Progress in Reliable Detection of Near Surface Reflectors when Inspecting Anisotropic and Isotropic Material Using the Total Focusing Method

Load-bearing components made of composite laminates of several centimetre thickness, e.g. in wind turbine blades, are frequently used in the energy sector. These components are usually tested with conventional ultrasound techniques. A typical approach to increase energy penetration depth is testing with lower frequencies. This leads to a decrease of sensitivity and consequently to a reduced detectability of small defects compared to higher frequencies – especially for defects close to the surface. Another possibility is to use high excitation voltage or gain to improve penetration, but this also leads to a much more pronounced initial pulse with saturated or clipped a-scans resulting in loss of information. Consequently, the defects close to the surface are often indistinguishable to the initial pulse and the flaw is overlooked. In comparison to conventional ultrasonic testing, the Total Focusing Method (TFM) shows higher resolution of near surface defects using the same frequencies. The TFM can be adapted to anisotropic media by consideration of the direction dependent wave propagation. Therefore, sound paths not perpendicular to the surface, which show less clipping, can be used for imaging. In this contribution we discuss approaches to improve the detectability of defects close to the surface in carbon fibre reinforced plastic (CFRP) and aluminium using Full Matrix Capture (FMC) and TFM. As a result, in CFRP flaws with a depth of 0.9 mm and above can be detected. The presented methods also improve the signal to noise ratio of near surface defects in the TFM reconstructions up to 4 dB. The first approach filters the FMC pulses in the wavenumber-frequency domain which reduces the afore-mentioned disturbances in the time domain signals and thus improves the detectability of near surface defects. The second approach is based on a maximum angle in the reconstruction step which reduces the entries of the information matrix based on location. This procedure is similar to taking the directivity function of each array element into account. Therefore, only time signals with a high signal to noise ratio are considered.