PSI - Issue 75

Alberto Visentin et al. / Procedia Structural Integrity 75 (2025) 593–601 Alberto Visentin, Alberto Campagnolo,Vittorio Babini, Giovanni Meneghetti/ Structural Integrity Procedia (2025)

600

8

1500

Δσ A,50% = 214 MPa N A = 2 ∙10 6 cycles Scatter Index (2.3% - 97.7%): T σ = 296/156 = 1.90 Slope k = 3.0

PSM design scatter band for steel joints ( λ = 0 )

1000

∆σ eq,peak [MPa]

296

Testing conditions: As-welded (AW) Failure: Weld toe (saddle region)

214

200

Global solid model - Model 1 (Fig. 2a) Global shell model + solid submodel - Model 2 (Fig. 3) Global shell model + solid submodel - Model 3 (Fig. 3)

156

100

1.00E+04

1.00E+05

1.00E+06

2.00E+6

Number of cycles to crack initiation, N init

Fig. 4. Fatigue strength assessment of CHS-to-SHS welded joints according to the PSM. Comparison between the PSM-based fatigue design scatter band for uniaxial local stress (λ = 0) and the experimental fatigue results by (Gandhi and Berge 1998) re-evaluated in terms of the equivalent peak stress. The design scatter band was calibrated in (Meneghetti and Lazzarin 2011) and is not fitted on the experimental data shown in this figure. 4. Conclusions The Peak Stress Method is an engineering approach for the fatigue strength assessment of welded structures which takes advantage of the linear elastic peak stresses calculated at the weld toes and weld roots, by means of FE models meshed with 2D or 3D solid elements. In this work, the applicability of the PSM combined with shell finite element models has been investigated to furtherly reduce the size of FE models if compared to solid finite elements and to rapidly analyze large-scale structures. First, a numerical procedure has been proposed to integrate a global shell model of the full welded structure with local solid submodels of critical regions, to which the PSM is applied. The stiffness contribution due to the weld bead has been accounted by adopting two empirical formulae proposed in the literature. Then, the proposed numerical procedure has been applied to a case study, relevant to the fatigue strength of a CHS brace welded to a SHS chord made of structural steel and subjected to axial loading. The equivalent peak stress distributions obtained through the proposed shell-to-solid submodelling approach have been compared with those derived from a global solid model of the full structure. The comparison highlights the effectiveness of the proposed strategy in balancing computational efficiency and fatigue assessment accuracy, the deviations being always below 15% and the time saving for solving the shell-solid FE models instead of a full solid FE model being greater than 50%.