Towards semi-automated non-destructive evaluation

L Fradkin,I Bell,R Taneja,P Fitzgerald,G Elston,D Lines,J Wharrie,V Zernov

doi:10.1088/1742-6596/457/1/012008

Abstract

A demonstrator has been developed showing feasibility of semi-automatic characterisation of large planar flaws in steel using ultrasonic transducer arrays. The unit is based on a real-time ultrasonic imager deploying National Instruments hardware and software, is connected to an IMASONIC linear phased array containing 128 elements and incorporates a novel flaw characterisation algorithm, which is a model-based variant of Total Focusing Method, taking into account undulations in inspection surface. It has been shown to process RF data collected in immersion reasonably fast and be capable of detecting and characterising with reasonable accuracy large planar defects.

Highlights

We have addressed a high level, long-term challenge of the ultimate deployment of an innovative ultrasonic unit for flaw characterisation and interpretation that would lead to cost effective and reliable semi-automatic NDE (Non-Destructive Evaluation)
While in order to carry out crack characterisation inspectors rely mostly on TOFD (Time of Flight Diffraction) approaches pursued by those who work towards automating crack characterisation can be broadly divided into pure signal processing procedures, such as those based on CS (Compressed Sensing) [2] or FMC [3], [4] which are best suited for dealing with specular reflections and more general and more time-consuming model-based data processing algorithms, such as those desciribed in [5] and [6]
The reported feasibility study has shown that simulating configurations which produce diffraction from defect edges usually leads to reliable defect sizing, even though further work is required to study and reduce probability of false indications

Summary

Introduction

We have addressed a high level, long-term challenge of the ultimate deployment of an innovative ultrasonic unit for flaw characterisation and interpretation that would lead to cost effective and reliable semi-automatic NDE (Non-Destructive Evaluation). Semi-automatic crack characterisation As mentioned at the end of the previous section, applying the intensity function (1) different points on the grid that surrounds a scattering point can pick up different portions of the same scattered pulse This can lead to a smudged image of the scatterer and maybe, to an error in its positioning. The resulting LabVIEW application first combines image and signal processing algorithms to implement a model-based modification of TFM to identify such spots It joins their centres by a straight segment to characterise the defect in terms of its orientation (in degrees) with respect to the inspection surface, extent (in mm) and depth (in mm), that is, the shortest distance from a defect end to a specimen’s surface (the inspection surface or else the backwall). 7mm/5 mm 9 mm/5 mm 5 mm /5 mm 8 mm/5 mm 3 mm/5 mm 6 mm/5 mm 4.5 mm/5 mm 7 mm/5 mm 0 mm/0 mm 1 mm /0 mm 1 mm /0 mm 5 mm /0 mm 0 mm/0 mm 2 mm/0 mm 0 mm /0 mm 2 mm/0 mm Figure 5. (a) Upper surface profile approximated by a 10th degree Polynomial and resulting raw image. (b) Upper surface profile approximated by a 0th degree polynomial (constant) and resulting raw image

Speed of execution

Conclusions