Gaining control over laser welding
A laser process head with real-time weld control has been demonstrated, after a four-year EU project. Daniel Lloyd, development engineer at Laser Optical Engineering, reports on the work
The Manufacturing Technology Centre (MTC) in Coventry has demonstrated a laser process head designed to close the loop on real-time weld control. The project, Real-time dynamic control system for laser welding (Radicle), received funding from the European Union’s Horizon 2020 programme to enable the design, manufacture and testing of the tool over a four-year period.
Project co-ordinator, Mark Holden, said: ‘These demonstrations have come together through great
pan-European collaboration involving companies from the UK, Sweden, Finland, Switzerland and Italy.’
While the project was funded by the EU, it was driven by a group of leading manufacturers including GKN Aerospace and GE Power. Clive Grafton-Reed, from Rolls-Royce, commented: ‘It has been great to be involved in this project from the start. Weld quality is critical in aerospace applications, and being able to implement a tool like this will give us even greater confidence in the laser welding process. This will help make us more efficient at manufacturing by reducing the need to x-ray large components, resulting in a saving on the re-work.’
Giuseppe D’Angelo, from Centro Research Fiat, noted: ‘While we make cars, weld quality is absolutely critical for us as well. Being able to make fewer welds of better quality would enable us to optimise vehicle strength and reduce weight. These are both essential for modern drivers across all of Fiat’s brands.’
At the outset, the project concept was to improve efficiency by reducing both the amount of material wasted as a result of poor welds, and the energy required for rework. These aims are not just limited to the partners involved, but all welding applications.
The core of the process tool demonstrated was produced by Swedish laser system firm Permanova, with additional technology and sensors provided by Laser Optical Engineering (LOE) and Bit Addict, and the machine learning for control by the Technical Research Centre of Finland (VTT).
Anna Wallner led the project for Permanova, and described the collaborative process of integrating the bespoke welding tool. She said: ‘While we have produced many laser heads for a broad range of customers, Radicle is increasing their data capture and handling capabilities very much in line with Industry 4.0. Using our optical knowledge we have specified wavelengths to enhance process integration.’
The processing head develped through the Radicle project
This meant that Permanova had to work closely with project partners Bit Addict and LOE to identify the key wavelengths for the sensors and look at other commercial systems that customers want to use in the future. Wallner commented: ‘Getting optical components with appropriate wavelength and power specification was challenging but we managed it.’
LOE worked with MTC and The Welding Institute (TWI) to identify the key emissions from the weld that could practically be monitored. John Tyrer, MD of LOE, said: ‘While there are a lot of academic studies, many of these use sensors and setups which cannot be translated into practical industrial systems. We had to find ways of gathering the emissions from the weld keyhole that tell us what is happening and allow predictions to be made in less than ideal environments.’
As a result, LOE has developed a sensor pack that can be mounted onto the Permanova head to monitor a number of optical emissions, both in image and wavelength form at high speed, as well as the acoustic emissions. This sensor pack sits alongside a camera which Bit Addict uses for seam tracking. Niclas Wikstrom, of Bit Addict, said: ‘While seam tracking has been approached in a number of ways in the past, we feel that the system we have designed has a number of advantages. It is fully integrated with the head, requiring less maintenance and alignment, while the use of a camera and illumination allows us to simultaneously view the keyhole and weld pool, something most other systems can’t do.’
Olli Nurmi, of VTT, was faced with the challenge of making sense of the data generated by the various sensors, which was done using machine learning. Nurmi said: ‘The amount of data by itself isn’t huge, but it’s generated in a number of ways, which makes formatting it for analysis difficult. It then needs mapping to weld quality data, something that doesn’t seem to have been done before.’
The weld quality data has been produced by MTC and TWI in a significant package of work to standards set by the end users. Nic Blundell, of MTC, and technical lead for Radicle, explained: ‘Each end user – indeed each potential application – has its own criteria and requirements, and this can initially seem an overwhelming challenge. However, the welder understands how the selection and control of robust parameters can realise such applications. The trick is in getting an automated system able to apply that same level of understanding, reasoning and response. Radicle has been an opportunity to identify and push the boundaries of process automation and to support a very genuine business driver.’
Chris Allen, of TWI, also pointed out that while many companies have their own process for welding ‘this is typically proprietary and not shared. As a result, process windows can be very narrow and poorly understood. This project has allowed us to expand these maps for a number of materials, helping laser welders better understand process sensitivity and tolerances.’
Tony Pramanik, of the MTC, said: ‘While MTC and TWI both use similar lasers, because of differences in the beam delivery systems, the process parameters and resulting welds vary. It’s been really interesting to see the effects of these differences in the signals we’ve been gathering.
‘Being able to correlate features in the welded parts and the signals has really helped us understand the effects of the process changes in weld parameter tolerances.’
André Cereja, of EWF, has been leading the effort to help inform potential users of the project. ‘We have been developing a new training programme which will be delivered through our accredited centres to welding operators and engineers. It’s been really interesting seeing the exciting developments led by the other project partners.’
The project is now ending its EU-funded development period, and the partners are exploring ways to continue to develop the system to a fully commercial product. As Matthias Hoebel, of GE, said: ‘Several companies have been trying to create this system and have come close. This tool and the knowledge developments have progressed the state of the art. We look forward to completing the developments and using the system in our factories.’