Composite model-based signal and image processing algorithms for semi-automated crack characterisation

Ultrasonic NDE (Non-Destructive Evaluation) of industrial materials is an important and cost-effective way of monitoring plant integrity. The NDE requirements are particularly stringent in heavily legislated nuclear industry. With a large aging fleet and new built on the way it is becoming harder and harder to train a required number of suitably qualified personnel to conduct ultrasonic inspections. Also, the process often calls for performance of repetitive and therefore boring tasks, increasing chance of human error. A lot of data collection is now automated but not data interpretation. Funded by TSB, Sound Mathematics Ltd., with 20 years of experience behind them in mathematical modelling of ultrasonic inspections, has recently developed a code, which is capable of generating automatically and in real time several possible versions of inspection reports, and thus addresses this issue. The code could be used for detecting, locating, sizing and identifying orientation of a large planar crack. We now wish to investigate whether similar codes can be created for characterising other major types of cracks described in the British Standards: Ultrasonic Examination.