PSI - Issue 77

Georg Veile et al. / Procedia Structural Integrity 77 (2026) 348–356 Author name / Structural Integrity Procedia 00 (2026) 000–000

350

3

Fig. 1 Comparison of the experimental and predicted fatigue life with an idealised fictive radius of 1 mm (left) and scanned weld surfaces (right) from (Veile et al. 2025).

This raises the question of the cause of increased scatter. Especially gradient based FDP showed an increase in scatter. Thus, suggesting a potential causative in gradient effects. Simultaneously, the effect on gradient determination can be investigated for idealized weld geometries. For this reason, the influence of gradient determination on the prediction is investigated in this work. 2. Methodology The investigated specimens in this work resemble nuclear grade weld joints of reactor internals (Central Rod Guide Assembly Support). These components are subject to environmentally assisted fatigue. For this reason, austenitic stainless steels (AISI 347 and 304 L) as well as the nuclear grade weld metal ER 347 were used to manufacture component like specimen. For more info, reference is made to [3,8] regarding the experimental setup, FEA, material models, as well as [10] regarding the load state of reactor internals. Here one can find a detailed description of the experimental setup for the displacement-controlled fatigue tests with a displacement ratio of -1. As failure a force drop of 25% was set to stop the experiments. No safety factor was applied to the fatigue life assessment in this work. This works focus is the fatigue life assessment of weld joints by also implementing the real weld joint topology into FEA. As material model an elastic plastic multilinear-kinematic hardening model was implemented for the weld and base materials.

Fig. 2 Methodology to implement scanned weld joint topologies into FEA using NURBS and the sub-modelling technique.

In Fig. 2 the methodology to implement scanned weld joint topologies into FEA is illustrated. For a detailed description of the scan procedure reference is made to [3,8]. The scan was conducted with GOM’s ATOS system with an accuracy of min. 12 µm [11]. The scan was polygonised and imperfections like holes were corrected manually. Subsequently, the discontinuous polygonised surface was converted into continuous surfaces using Non-Uniform Rational B-Splines (NURBS) patches. For comparison, the weld radii were also idealized using a fictious radius of 1