PSI - Issue 75

Carl-Fredrik Lind et al. / Procedia Structural Integrity 75 (2025) 519–529 Carl-Fredrik Lind et al./ Structural Integrity Procedia (2025)

527

9

Figure 7. Maximum principal stress distribution in shell element models.

6.2. Welded component A fully automated workflow was developed using the Master S-N curve method. The process combines structural stress computation with unsupervised clustering to identify weld lines and assess fatigue life efficiently. Weld nodes are first extracted from the FEM model based on mesh topology. For each node, coordinates, reaction forces, moments, and shell thicknesses are obtained from solver output (e.g., Abaqus NFORC). Structural stress is then calculated using the mesh-insensitive approach by Dong (2001), combining membrane and bending components. As shown in Figure 1, this is done by projecting nodal data into the local weld coordinate system and applying Eq.(5). To group nodes into weld lines, the DBSCAN algorithm (Ester et al., 1996) is applied. This unsupervised clustering method identifies spatially adjacent weld nodes without requiring pre-defined group counts. DBSCAN has been previously used in weld analysis by Ghanadi et al. (2024) and is effective for handling irregular geometries. Each weld groups fatigue life is then calculated using the Master S-N curve formulation (Dong et al., 2003), with the minimum nodal life in each cluster representing the welds critical value. Welds falling below a defined fatigue threshold are flagged for further review. Fig. 8 illustrates the automated grouping and evaluation, highlighting critical welds directly in the post processor. The method was implemented as a Python-based post-processor for FEM results (e.g., from Abaqus), integrating data extraction, stress computation, clustering, and fatigue life evaluation into a single workflow.

Figure 8.Maximum principal stress distribution in welded component.