A flexible robotic cell for in-process inspection of multi-pass welds

Lines, David and Javadi, Yashar and Mohseni, Ehsan and Vasilev, Momchil and MacLeod, Charles Norman and Mineo, Carmelo and Wathavana Vithanage, Randika Kosala and Qiu, Zhen and Zimermann, Rastislav and Loukas, Charalampos and Foster, Euan and Pierce, Gareth and Gachagan, Anthony (2020) A flexible robotic cell for in-process inspection of multi-pass welds. Insight: The Journal of the British Institute of Non-Destructive Testing, 62 (9). pp. 526-532. ISSN 1354-2575

Text (Lines-etal-Insight-2020-A-flexible-robotic-cell-for-in-process)
Accepted Author Manuscript
License: Creative Commons Attribution-NonCommercial 4.0 logo

Download (970kB)| Preview


    Welds are currently only inspected after all the passes are complete and after allowing sufficient time for any hydrogen cracking to develop, typically over several days. Any defects introduced between passes are therefore unreported until fully buried, greatly complicating rework and also delaying early corrections to the weld process parameters. In-process inspection can provide early intervention but involves many challenges, including operation at high temperatures with significant gradients affecting acoustic velocities and, hence, beam directions. Reflections from the incomplete parts of the weld would also be flagged as lack-of-fusion defects, requiring the region of interest (ROI) to adapt as the weld is built up. The collaborative SIMPLE (SIngle Manufacturing PLatform Environment) project addresses these challenges by incorporating robotic inspection within a robotic tungsten inert gas (TIG) welding cell. This has been accomplished initially with commercial off-the-shelf ultrasonic phased arrays, but is flexible enough to adapt to future developments with solutions suitable for higher temperatures. The welding and inspection robots operate autonomously. The former can introduce deliberate defects to validate the latter, which uses 5 MHz 64-element phased arrays on high-temperature wedges to generate sector scans after each weld pass. The results are presented, confirming that the challenges have been addressed and demonstrating the feasibility of this approach.