Data-driven automatic validation of phased-array ultrasonic-testing data acquisitions of welds in thick austenitic stainless-steel plates for the ITER vacuum vessel manufacturing

• Phased-array ultrasonic testing is used to assess weld quality during manufacturing. • Complex and long PAUT scanning can result in invalid acquisitions and workshop disruptions, requiring rescanning. • PAUT acquisition validity should be confirmed before processing, to increase efficiency. • Three validation checks are possible regarding missing data, coupling and thickness. • Automatic PAUT acquisition validation can save time and resources, increasing efficiency in 99 %, in complex manufacturing. Phased-Array Ultrasonic Testing (PAUT) is a commonly used technique for rigorous non-destructive testing (NDT) of metallic welds and composite material welds, since it is known for its sensitivity and precision, mainly used in highly demanding industrial applications. Nevertheless, its interpretation is time-consuming and requires a skilled certified inspector, according to ISO 9712. The complexity of the PAUT output interpretation depends on many factors such as the type of material, the length of the welded area and the number of beams configured to be transmitted by the probe at each location. During the weld quality check, one of the main tasks of the inspector as part of the interpretation is to ensure that the acquisition is valid. Unfortunately, validation can only be done manually during the inspection and processing phase of the weld PAUT output, which can be time consuming. In the manufacturing of large and complex components with special materials, late validation of PAUT acquisitions results in large cost overruns and delays. This paper presents an automatic early detection analysis mechanism, which differentiates valid from invalid PAUT weld acquisitions based on the analysis of acquired data. This data analysis has been translated into a successful development, which can be used live in a manufacturing workshop for mechanical components. The International Thermonuclear Experimental Reactor (ITER) Vacuum Vessel has been selected as a case study for the adequacy of this technique to welding, as well as to improve time and cost efficiency of PAUT. This paper brings a new data-driven methodology regarding PAUT acquisition validation, covering a gap in technical-scientific literature, by finding the most important and common features to valid PAUT acquisitions.