[3C4] Precision simulation of UT defect echo scattering
L Bergbreiter¹,², H Mooshofer¹ and C Grosse²
¹Siemens AG, Germany
²Technical University of Munich, Germany
The aim of ultrasonic testing is not only to locate defects but to size them for further evaluation of the component’s condition. One approach to sizing defects is to compare the defect echoes with echoes from reference reflectors, for example disk-shaped reflectors. In this work, we simulate elastodynamic scattering of a plane wave under different angles of incidence using the grid-based elastodynamic finite integration technique (EFIT) combined with Auld’s reciprocity theorem (ART), which simulates all wave modes present in materials with linear wave propagation. All grid-based simulations are necessarily affected by discretisation errors, resulting in inaccuracies in the scattered echoes. On the other hand, the duration of the simulation is proportional to the grid size to the power of four (space+time) so that, in the interest of computability, the grid size must not be chosen finer than is necessary to achieve the targeted accuracy. In our approach, the deviation between the real reflector surface and discretised reflector surface is minimised to achieve an optimised grid. This fact and the use of the ART leads to feasible simulation times and memory demand. Therefore, the total amplitude inaccuracy due to numerical errors and perturbations in the simulation is determined to be less than 1%.
To study the angle dependency of reflectivity, we rotate the incidence angle of a plane wave relative to the defect instead of rotating the defect. This avoids inducing roughness of the discretised defect surface due to staircase. In this context, two different stimulation methods are investigated to excite a longitudinal plane wave in the far field. First, we use a stimulation via coupling plane, but this induces a backward propagating wave that leads to additional reflections at the simulation borders. This can be avoided by stimulation using a precalculated snapshot of the wave field.
To study the angle dependency of reflectivity, we rotate the incidence angle of a plane wave relative to the defect instead of rotating the defect. This avoids inducing roughness of the discretised defect surface due to staircase. In this context, two different stimulation methods are investigated to excite a longitudinal plane wave in the far field. First, we use a stimulation via coupling plane, but this induces a backward propagating wave that leads to additional reflections at the simulation borders. This can be avoided by stimulation using a precalculated snapshot of the wave field.
