Time-of-flight diffraction - from semi-automatic inspection to semi-automatic interpretation

Time-of-Flight Diffraction (TOFD) is a recent but well-developed ultrasonic technique that is rapidly gaining prominence as the method of choice for many weld inspection applications. The data acquisition configuration lends itself conveniently to automation, and robotic scanning is routinely used. However, despite the mechanisation, the most time-consuming processes are still performed off-line manually. The trained operator utilises interactive software tools to visualise and calibrate the data, but the actual defect detection, sizing and characterisation are totally reliant on the operator's expert judgement. As the quantity and density of data increases, the effects of operator strain, visual fatigue and loss of concentration become more apparent, resulting in errors and inaccuracies. This combined with the growing pressure to complete the interpretation of inspection data in near real-time has underlined the need for robust and reliable interpretation aids to automate the routine aspects of the operators' tasks. While nothing can ever replace the expert operator, an interpretation aid can detect regions of interest and provide preliminary depth and sizing information, as well as an assessment of the defect class according to the adopted codes and standards. This can potentially increase operator interpretation throughput, reliability and consistency, resulting in a cost reduction. This paper proposes just such a scheme for semiautomatic TOFD interpretation. A number of signal and image processing tools have been specifically adapted for use with ultrasonic TOFD data. Results obtained have so far been promising in terms of accuracy, consistency, reliability and processing speed.