Calculation the ultrasound scattering to improve the detection of crack-like flaws in austenitic welds. Phase 1. Calculation the ultrasound scattering for 2D-weld seam models. Final report

The simulation of ultrasonic nondestructive testing, in particular weld inspection of nuclear power plant components, represents an important aspect for test planning. With the help of simulations it is possible to determine and optimize important testing parameters like wave mode, frequency, transducer type and inclination angle before the actual measurement is performed. For the calculation of A-, B- and C-scans several simulation techniques have been used so far, e.g. ray tracing, point source synthesis, Gaussian wave packets, CIVA and EFIT (Eiastodynamic Finite Integration Technique). However, all these codes do not take the scattering by the polycrystalline microstructure explicitly into account. In particular, for strongly scattering media like the austenitic base material and the weld area, this interaction needs to be considered. One of the main motivations of this project was to close this gap in current simulation methodology. The goals were 1. to calculate the relevant scattering coefficients analytically in order to determine the smallest detectable size of circular disc defects in different microstructures; 2. to expand the EFIT-code in such a way so that not only defect echoes and backwall echoes, but also time-domain grain scattering in A-Scans can be incorporated; 3. to support and validate theory and simulation by appropriate experiments. Thus, a better flaw detection sensitivity and an improved interpretation of echo indications were intended. In Phase 1 of the project the mentioned objectives were pursued for three different base materials, fine-grained copper (mean grain size of 30 μm), coarse-grained steel casting (50 μm) and coarse-grained austenite (100 μm) for frequencies between 4 and 20 MHz. The scattering coefficients and smallest detectable circular disc flaws were determined by using an adapted wave equation. In the enhanced EFIT-code the polycrystalline microstructure was included on the basis of Voronoi and Laguerre tessellations. The new EFIT-code is able to model 2-D and 3-D problems, P- and S-waves, normal and angular radiation and to calculate A-Scans for immersion and contact technique with plane and focused transducers. The results of a comparison between experiments and theory/simulation are quite promising: The analytically calculated smallest detectable circular disc flaws are slightly smaller than the measured values. For frequencies between 4 and 10 MHz the deviations are between 5 and 40%; for frequencies between 15 and 20 MHz the deviation is approximately 50%. The A-Scans calculated by the expanded EFIT-code are in a good agreement with the measured signals. The amplitudes of defect and backwall echo also agree very well. The backscattering contributions in the simulations show the same behavior than those in the experiments. The deviations in signal-to-noise ratio between measurement and simulation are smaller than +/- 3 dB. Phase 2 of the project with the main focus on austenitic weldings and columnar crystals microstructure is granted.