Ultrasonic inspection of austenitic and dissimilar welds

Ultrasonic testing requires materials that are acoustically isotropic. The sound propagation is assumed to be non-dispersive and independent of its direction. Finegrained steel qualities meet these requirements, for example. However, austenitic weld materials are highly anisotropic due to their dendritic structure produced by the cooldown process during welding. State-of-the-art ultrasonic inspection of stainless steel (austenitic) welds demands special procedures with component specific qualifications for both, the inspection technique and the inspection personnel. Funded by the German Nuclear Safety Research Program, we obtained a profound understanding of sound propagation in weld structures by elasto-dynamic simulation. Nevertheless, even the design of problem adjusted transducers and phased array techniques have not resulted in the reliable assessment of weld quality as required by standards and regulations. The "Sampling Phased Array" technique allows the acquisition of time signals (Ascans) for each individual sensor element of the array; the "SynFoc" image reconstruction algorithm solves the limitations dictated by the sampling theorem. Consequently, the array elements may be distributed along an optimized array aperture for high-resolution imaging. The reconstruction considers the sound propagation from each image pixel to the individual sensor element of the aperture. We use experimentally verified model structures of the welds for elasto-dynamic calculations, which can be analyzed with reconstructed A-scans, backwall signals for example. If necessary, the structure can be adjusted until we presume its correctness. Next, we add the acquired phase-corrected A-scans, which represent the actual sound propagation in the structure. This technique, in combination with "Sampling Phased Array " and "SynFoc", represent the "Reverse Phase Matching" principle. We present first results on anisotropic structures demonstrating the progress achieved using this technique. Future work is directed to high-speed computation of optimized elasto-dynamic algorithms to support assessing the local structure of welds, which may be different from the original model due to weld repair, for example.