Enhanced Defect Detection and Characterisation by Signal Processing of Ultrasonic Array Data

The full data set from an ultrasonic array comprises the time-domain signals from every possible transmitter-receiver pair in the array. The possibility of experimentally obtaining the full data set that is afforded by some new array controller systems massively increases the potential of ultrasonic arrays. Firstly, any conventional array imaging procedure (e.g. B-scans, angular scans, dynamic depth focusing) can be performed by post-processing the information in the full data set. More importantly, the full data set enables imaging techniques to be developed that have no direct counterpart in conventional ultrasonics. For example, the authors have previously shown that the total focusing method (TFM) outperforms other standard linear processing techniques when imaging a point reflector. This is because the TFM represents the complete array being focused in transmission and reception at every point in an image. In this paper two advanced versions of the TFM are described. The first called the vector TFM (VTFM) algorithm allows the angular reflectivity characteristics of any point in a sample to be obtained. This has obvious advantages for defect sizing and classification. The VTFM algorithm probes the scattering behaviour of a defect over a range of angles. The response is displayed as a vector that represents defect orientation and a scalar quantity that represents the specularity of the reflector. Both simulated and experimental results are shown from which the orientation of small reflectors is determined and visualised. The second algorithm is termed the diffraction TFM (DTFM) and uses a pair of arrays in a similar manner to time-of-flight diffraction (TOFD) systems. The immediate advantage of DTFM over TOFD is the ability to produce a B-scan image without moving the transducers. Potential additional advantages are superior sensitivity and the ability to overlay pulse-echo information from back-scattered signals with pitchcatch information from diffracted signals. Experimental results are presented from an industrial weld sample containing a number of realistic defects.