Efficient testing of weld seam models with radii of curvature in the millimeter range using laser ultrasound

Laser ultrasound is a widely used tool for industrial quality assurance when a contactless and fast method is required. In this work, we used a laboratory setup based on a confocal Fabry-Perot interferometer to examine weld seam models. The focus was placed on small samples with curved surfaces (small in the sense that the radius of curvature is comparable to the largest ultrasonic wavelength) and on efficient ways to detect the presence and volume of process pores, with the goal to transfer this method to industrial applications. In addition to this experimental method for investigating welds, a numerical method that models the experimental setup was implemented in MATLAB. For this purpose, first the thermal effects of the excitation process were taken into account by solving the thermal diffusion equation with an explicit scheme. Then, the elastodynamic equations were solved using the Elastodynamic Finite Integration Technique, taking into account the stresses induced by the excitation process. The B-Scans generated with this numerical model were compared with experimental B-Scans for simple test cases and good agreement was found. In a next step, the additional structures in the B-Scans resulting from air inclusions were identified and investigated with both methods using flat test specimens at first. Besides the direct echoes, structures from skimming surface waves and multiple reflections were visible. These additional structures are unwanted in defect reconstruction methods like the Synthetic Aperture Focusing Technique (SAFT) as they would lead to artifacts. In samples much larger than the largest ultrasound wavelength, however, these unwanted structures are still negligible in amplitude or can be well separated temporally, but for small samples this is no longer the case. As a result, reconstruction methods based on direct echoes like SAFT are difficult to apply. For many industrial applications, the reconstruction is not decisive at all, but only the knowledge of the total volume of process pores (TVPP). It is shown with both experimental and numerical methods, that this TVPP can be estimated from the variation in the B-Scans from various small weld seam models.