Real-time weld defect identification

19 May 2022

A technique that allows defects in fusion welds to be detected in real time will save time and cost in the nuclear sector and beyond. By Tony Burnett

THE AWESIM (AUTOMATED WELDING EQUIPMENT System Inspection and Monitoring) project was borne out of curiosity and a classic “what if” question: what if we could detect defects in fusion welds in real time? Challenging traditional approaches is never easy, so asking the question was a courageous step, as welding and non-destructive testing of a completed welded component have traditionally been separate and sequential activities.

Detecting defects as they are formed offers substantial cost, quality and schedule benefits to industry. In fusion welding it would reduce the length of the feedback loop, from a weld being formed to knowing that it is sound, from hours or days to minutes (depending on weld complexity and component size).

The challenges for the nuclear industry in manufacturing high integrity complex components using fusion welding techniques are significant. Recent examples where problems with fusion welds have led to substantial costs and delays include the Flamanville EPR, where a range of weld inspection and certification issues have severely delayed the start of power generation and increased costs.

A system capable of real-time detection of defects in fusion welds could be a valuable tool for reducing the costs of manufacturing high value components in the nuclear new build programme. This recognition set the research path and goals.

Collaboration and AWESIM

Work on the project began in 2015/16, when the University of Sheffield’s Nuclear Advanced Manufacturing Research Centre (Nuclear AMRC) began investigating advanced welding and forming technology and the University of Strathclyde’s Advanced Nuclear Research Centre (ANRC) started investigating whether phased array ultrasound inspection could be undertaken effectively at the same time as the weld is formed. These complementary research themes developed in parallel and a range of UK government funding provided to academic institutions was used to increase the technical readiness of the technology.

The parallel development efforts resulting in collaboration between the two research centres for the first time during the ‘SIngle Manufacturing Platform Environment’ (Simple) project. Simple was funded by the UK government Department for Business, Enterprise and Industrial Strategy (BEIS) as part of the Energy Innovation Programme, Advanced Manufacturing and Materials Call, Phase 1 call in 2016. This phase was successfully completed in 2018 and yielded promising avenues of development.

In late 2019, the research centres sought out sponsors of the technology from among the industrial members of the research centres. Cavendish Nuclear Ltd, a wholly owned subsidiary of Babcock International Group, took the lead and formed a consortium that included Doosan Babcock Ltd, Nuclear AMRC and ANRC. Subcontract suppliers PEAK NDT Ltd, a specialist non destructive testing supplier (categorised in the UK as a ‘small or medium sized enterprise, SME) and Frazer Nash Ltd completed the consortium.

The consortium bid for funding from BEIS as part of the £505m Energy Innovation Programme through the Advanced Manufacturing and Materials Phase 2b call. BEIS provided £1,348,000 of funding to Cavendish Nuclear and a further £726,000 of industrial support was provided by the industrial partners in the consortium to complete the project.

The plan was to demonstrate a single fully integrated industrial-sized system capable of detecting weld defects in real-time a year after project kick-off.

The team choose the Nuclear AMRC facility in Rotherham for the demonstration. A single use case was selected as the basis of the demonstration — a circumferential weld in a low carbon alloy steel pipe 406mm O/D, with walls 16mm thick, and a ‘V’ groove weld preparation.

Owning and delivering AWESIM

Work started on the AWESIM project in April 2020, two weeks after the start of the first lockdown caused by the Covid-19 pandemic. Adaptation was essential. The technical kick-off meeting for forty-seven participants over Microsoft Teams was a first for us all. Oddities became the norm — like test rigs having their own web addresses so that they could be invited to meetings.

As infection rates ebbed and flowed, and government restrictions changed, it became inevitable that the original delivery plan had to change. It was not possible to demonstrate the technology on a single platform, owing to travel and social distancing restrictions. A ‘plan B’ was needed urgently and the team came up with a solution. Two autonomous and robotically-enabled demonstrators were built, one at each research centre, joined virtually by a software emulator. A shared common workpiece physically travelled by courier between the centres to demonstrate that both the weld process monitoring and NDT inspection system elements could detect the same defects.

The change was significant, extending the project to December 2021. Costs increased as well, but were controlled by the research centres’ abilities to acquire additional equipment and resources, and the goodwill of the industrial partners who continued with the engagement.

AWESIM outcomes

To measure whether a system can detect a weld defect in real-time, it is essential to have the means to create a quantifiable and repeatable defect when it is needed. This might sound easy but it is not. Modern autonomous or robotic welding centres are designed to minimise the occurrence of weld defects. Trying to get such systems to create a defect of a measurable, predictable size when needed took considerable ingenuity.

Once this challenge had been met it was possible to test the technology to find engineered defects of different forms, such as root weld failure, lack of side wall fusion, inclusions and porosity.

Nuclear AMRC delivered a weld process condition monitoring system that brought together the processed output from four dissimilar sensors to reach real-time decisions as to whether a fusion weld defect was present or not. The sensors included a laser weld profile sensor, a high definition camera, an acoustic sensor and a weld process power monitoring sensor.

While the sensors were commercially-off-the-shelf components, the data analysis software was not. Bespoke data analytics and image processing methods involving neural networks and machine learning algorithms were developed. Aggregation of the processed data sets was used to confirm in real time whether a defect was present.

ANRC continued with its development of a phased array ultrasound sensor capable of operating at surface temperatures up to 350°C. A combination of hardware and software development took place. A roller probe configuration using a tyre made from a custom temperature resistant polymer, cooled internally, was chosen and developed. This meant no acoustic couplant was needed and the forced cooling ensured the sensor could operate without restriction — 24 hours per day, seven days per week if needed. Compensation for the variation in the speed of sound in steel at temperature was factored into the full matrix capture (FMC) phased array ultrasound testing (PAUT) measurement approach for the freshly deposited weld bead. It was shown that the prototype sensor and software suite can detect, locate and size defects with the same accuracy as similar measurements carried out at ambient temperatures. By avoiding the need for a liquid acoustic couplant being applied to the surface of the component being inspected, time and cost savings can be realised.

The team showed that the real-time capability of the technology gave operators a higher degree of confidence that the welding is ‘right first time’, improving schedule certainty while maintaining quality. Abortive welding was avoided, as the welding process could be stopped as soon as an out-of-specification weld was found. This process-control step leads to increased productivity and optimisation of the energy and materials used, improving sustainability. Additional benefits can be realised by reducing the dependence on radiographic weld inspection, further saving time and reducing hazards to operators by reducing radiation risks.

The team were able to successfully demonstrate the AWESIM technology. They performed real-time detection of defects in September 2021 to representatives from BEIS, the Office for Nuclear Regulation and the Environment Agency face-to-face. Realisation of many of the benefits identified before the project began is in sight. Patent applications were prepared covering all elements of the technology and patents have been filed.

Next steps

Planning is under way to develop the technology for commercial deployment. The consortium members are developing the technology to follow up on secondary opportunities identified, but not worked on, in AWESIM.

The AWESIM project resulted in a technology that offers a potentially disruptive step-change in the technology used for welding and NDT of high integrity components.

The benefits to the nuclear industry are clear. The technology is being applied to the future nuclear new- build programmes (Sizewell C, UK SMR, advanced modular reactors and fusion reactors). Although the technology was developed using government funds and focused on nuclear industry applications, it could apply to any industry that requires high-integrity welded fabrications.

Author information: Tony Burnett is Head of Innovation and Technology at Cavendish Nuclear Ltd

AWESIM project test rig at Nuclear AMRC, University of Sheffield
Temperature compensated phased array ultrasound sensor at ANRC, University of Strathclyde

Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.