Automated Inspection and Imaging Method for Insertion-type Nozzle-to-Shell Fillet Welds Based on Flexible Ultrasonic

Insertion-type nozzle-to-shell fillet welds are critical load-bearing structures in pressure vessels, whose welding quality directly impacts equipment safety. Traditional ultrasonic testing (UT) methods are limited by the curvature of the nozzle's inner wall and complex sound fields, leading to difficulties in defect identification and localization, as well as high miss-detection rates. This paper proposes an innovative automated inspection and imaging method that integrates a flexible ultrasonic phased array probe, a dedicated mechanical scanning device, and intelligent data stitching software. Firstly, an accurate mathematical model for beam focusing in a water-steel bilayer medium is established. By calculating the delay times for each array element, precise focusing of ultrasonic waves in the fillet weld region of complex curved components is achieved, significantly improving the signal-to-noise ratio and detection sensitivity. Secondly, to address the operational scenario requiring multiple axial-circumferential composite scans, a dedicated data stitching algorithm and software are developed. This software automatically stitches and converts the discrete, slanted-line raw C-scan data into an intuitive and complete two-dimensional unfolded image, enabling precise defect localization and quantitative analysis. Experiments and applications demonstrate that this method can stably inspect nozzles of various specifications with inner diameters ≥150 mm. Its defect detection rate and quantitative accuracy are significantly superior to conventional methods, providing a complete solution for the automated and highly reliable non-destructive evaluation of insertion-type nozzle-to-shell fillet welds