Ultrasonic Array Inspection Optimisation for Non-Destructive Evaluation

Ultrasonic arrays have seen a dramatic increase in industrial up-take over recent years to the point where it seems possible that NDE array inspections will completely replace the present industry-standard single point measurements over the course of the next decade. The current NDE array approach is based heavily on a combination of rapid beamforming equipment, developed originally for medical ultrasound, and inspection strategies that follow approaches developed originally for single element probes. Whilst using an array in this way undoubtedly delivers good imaging in some circumstances, it is far from obvious how close to optimal it is. This project addresses the fundamental questions at the heart of NDE: how first to quantify and then optimise the performance of an inspection. The selection of parameters to quantify performance is critical and depends on the purpose of the inspection (e.g. defect detection or sizing) and will be informed by input from the industrial partners. A modelling framework will be developed that allows array inspections to be designed to optimise the chosen parameter (e.g. probability of detection, sizing accuracy). This optimisation framework will be based on rapid forward models, a sound understanding of the factors that most strongly influence the choice of array inspection configuration and a rigorous statistical methodology. This latter aspect is particularly important as some level of uncertainty is inherent in all NDE inspection: this can range from unknown defect orientation through to unknown velocity distributions. This modelling framework will not only allow for inspections of complex parts to be optimised, but the exploration of the relevant parameter space will inform current best practice and help in tasks such as choice of secondary inspection. Together these developments will produce a step change in the performance of arrays leading to improved inspection reliability, safer structures and ultimately reduced design conservatism.

The project is part of the UK Research Centre in NDE (RCNDE), the funding for which is earmarked by EPSRC for industrially driven research. The project is also supported financially by Rolls-Royce, Sellafield, BAE Systems, SERCO and EDF.

The thrust of the proposal is to develop a generic mathematical framework for the array optimisation process and demonstrate its application through a number of case studies supplied by our industrial collaborators (Rolls-Royce, EDF, Sellafield, BAE Systems and SERCO). We strongly believe that this parallel approach of progressing the underlying engineering science alongside applications is the way to maximise the impact of our work. The support from industry (£90,000 cask + in-kind) indicates that a clear industrial need exists. The collaborating companies have been involved directly in the development of the proposal. Three of the collaborators (Rolls-Royce, EDF and SECRO) are already experienced users of arrays and their involvement highlights the need felt by industry to further improve these inspections and develop arrays to the point where they can be deployed with confidence on the most challenging inspection problems. Note that in addition to financial support, Rolls-Royce have agreed to direct the efforts of an EngD student into rapidly transferring the results of this project into their organisation. BAE Systems and Sellafield are less experienced users of array techniques and are looking to explore where and when array can be best utilised. All five collaborating companies will be well positioned to bring the findings into use in their industries.

The primary route to communication of the results of the project to industry will be via the five collaborating companies. Additionally, the other industrial members of the RCNDE will be informed of the principal steps of progress, via the RCNDE annual review, and so will be able to take early advantage of the new capabilities. We also plan on running a workshop, including software demonstrations, at the 2 year point in the project to disseminate some of the more practical findings to a wider audience. The work will be presented at the main NDE conferences (e.g. annual Review of Progress in Quantitative NDE, USA; annual British Institute of NDT conference). Finally, we will publish in technical journals read by NDE practitioners as well as more academically focused journals (e.g. J. Acoust. Soc. Am.; IEEE UFFC; Proc. Roy. Soc.).

When the research is ripe for industrial implementation, we will present it at an RCNDE technology transfer event to both the other main industrial end-user members of RCNDE and also to the wider community of RCNDE associate member companies (both large global companies and SMEs). As most new NDE technology (software included) is manufactured by SMEs they form a particularly important part of the supply chain so this type of event is critical. These events are run with the express purpose of aiding the exploitation of findings delivered from the RCNDE.