PSI - Issue 38

Jinchao Zhu et al. / Procedia Structural Integrity 38 (2022) 621–630 Author name / Structural Integrity Procedia 00 (2021) 000 – 000

622

2

1. Introduction A wide range of engineering structures rely on welded joints for proper structural integrity. These joints are often the limiting factor for the fatigue strength of such structures. Studies (Kainuma et al. (2008), Balasubramanian et al. (2000), Lee et al. (2009), Barsoum et al. (2008)) have been carried out to assess the influence of the weld geometry on the fatigue strength. It is found that, the weld size (Kainuma et al. (2008), Balasubramanian et al. (2000)), weld toe radius and toe angle (Lee et al. (2009)) have a major influence on the fatigue properties. Detailed 3D finite-element analysis (FEA) of welds have been performed to assess the influence of geometrical variability, see e.g. (Hultgren et al. (2021) 1 , Balasubramanian et al. (2020)). These methods are often based on the Weakest-link assumption (Hultgren et al. (2021) 2 , Sandberg et al. (2017)). However, performing such detailed analysis is generally computationally prohibitive, especially for practicing engineers. Therefore, a widely used simplified and computationally efficient fatigue assessment method is based on the notch stress (NS) method. The notch stress is computed by FEA with a certain assumed notch radius and an idealized geometry (Hobbacher et al. (2016)). The notch stress range is obtained as the product of the applied nominal stress range and the stress concentration factor (SCF) due to the local notch. Barsoum et al. (2008) computed the SCF using a reference weld toe radius of 1 mm and the mean value of all measured toe angles. It was found that the characteristic strength, i.e. stress range at 2 million cycles and 97.7 % survival probability, is roughly 15% lower than the IIW FAT-value of 225 MPa (Hobbacher (2016)). This difference is attributed to the high scatter in the local weld geometry. Pedersen et al. (2010) determined the SCF by using a reference radius of 1 mm and flank angles of 45° for fillet welds and 30° for butt welds. It was shown that the specimens with poor local weld toe profile could fall below the FAT 225 curve and FAT 200 was therefore proposed to give the same safety as observed in the nominal stress system. Rohani et al. (2021) found that the notch stress approach with FAT 225 shows nonconservative assessment of fatigue strength when idealized geometry with a reference radius of 1 mm is applied. These studies demonstrates that the choice of the idealized weld geometry in the notch stress analysis is still not clear, especially when the geometrical variations are large. The major limitation of the aforementioned studies is that the variability in geometrical weld parameters is not included in the fatigue assessment. Instead, the notch stress analysis is based on only one idealized geometry and a FAT-value based on recommendation. However, the local geometrical parameters generally vary stochastically along the weld. This is due to stochastic variabilities in the manufacturing process parameters that can hardly be eliminated (Barsoum et al. (2008), Mansour et al. (2019)). It has been shown that these geometrical variations directly affect the scatter in fatigue life (Barsoum et al. (2008)). To address this limitation, it is essential to include the local geometrical variations when assessing the fatigue life. In this work, we use Monte Carlo (MC) simulation (Mansour et al. (2018), Mansour et al. (2016)) to study the influence of these stochastic variations on the probability distribution (Mansour et al. (2014)) of fatigue life. A non-load carrying fillet cruciform joint is studied based on two notch stress approaches with different reference toe radius and FAT values: r ref = 1 mm with FAT 225 (Hobbacher (2016)) and r ref = r actual + 1 mm with FAT 200 (Fricke (2012)). Both approaches are compared with respect to their fatigue life prediction as well as the predicted influence of geometry variability. 2. Fatigue life assessment 2.1. Local geometry and assumed variability The geometry of the studied cruciform specimen and illustration of the geometrical parameters are shown in Fig. 1(a) and Fig. 1(b), respectively. The mean values of the weld leg length, toe angle and toe radius are 10 mm, 45° and 2 mm, respectively. The corresponding standard deviations are 0.5 mm, 9°, 1 mm. This corresponds to coefficient of variations (CoV) of 5%, 20% and 50%, which is in accordance with measurements in a previous study (Barsoum et al. (2008)) on robotic metal core arc welding (MCAW).