• Harvey, G.
• Mulholland, Anthony J.
• Cunningham, Laura
• Bird, C.

Abstract

The non-destructive testing of austenitic steel welds using ultrasound is of vital importance for assessing safetycritical structures such as those found in the nuclear industry. The internal geometry of these welds is heterogeneous<br/>and highly scattering and this makes it dicult to detect and characterise any defects within them. To help overcome these diculties the use of ultrasound transducer arrays and the associated Full Matrix Capture is becoming more widespread. There is a need therefore to develop post-processing algorithms that best utilise the data from such devices. This paper considers the use of a time-frequency domain method known as the Decomposition of the Time Reversal Operator (DORT) method. To develop this method and to demonstrate its ecacy in tackling this problem a series of simulated data sets are used. The simulated data is generated using a finite element method (PZFLEX) with the heterogeneous internal microstructure of the weld being given by previous Electron Backscatter Diraction measurements. A range of artificial flaws are then inserted into this geometry. By varying the flaw size and type a comparison is conducted between the DORT method and the Total Focusing Method (TFM) and<br/>their relative ability to perform flaw detection assessed. Importantly, however, the DORT method relies on a Singular Value Decomposition in time and frequency space and this spectral information contains information about<br/>the flaw size and shape.

Topics

• impedance spectroscopy
• microstructure
• steel
• defect
• ultrasonic
• decomposition