13th International Conference on Fracture June 16–21, 2013, Beijing, China -8- is about 2.5. According to the literature, no crack turning was observed at k = 0, 0.5 and 1 [11], but crack will turn when k > 2 [10]. Therefore there is more crack deviation in the test than predicted. We have used the weld metal da/dN data for the life prediction presented in Fig. 4a, in which residual stress effect is included in the coupon tests to acquire the da/dN data. However, we have not used the combined lateral stresses ( x_applied + x_residual) in the calculation of the crack growth driving forces (KI, KII, and T-stress); this could affect the magnitude of crack turning. Further, although residual stress effect is reflected in the measured crack growth data of welded M(T) sample with a same transverse weld, because the cruciform specimen has different geometry and dimension, welding induced residual stresses in the cruciform specimen could be different from those in the M(T). 4.4.4. Effect of weld metal microstructure and anisotropic material property Although mixed mode loading is believed as the primary cause of loss of directional stability by many researchers, microstructure changes in the weld metal also contribute to the crack directional instability. In the test specimens, initial crack plane was 5 mm from the weld nugget centre meaning that material properties are different for each side of the crack plane, i.e. weld nugget property on one side of the crack and TMAZ/HAZ on the other side. Furthermore, alloy 2198-T8 has highly anisotropic microstructures [24-25]. According to Yates et al. [14], highly anisotropic microstructures can also lead to significant changes in crack orientation, but more often in a zigzag pattern maintaining the overall mode-I trajectory. Crack path deviation triggered by material property irregularity can be enhanced by the applied biaxial loads as demonstrated by the test results in Fig. 4b. Test measured crack growth route shows a zigzag path when a < 70 mm [18]. However, predicted crack deviation is in one direction. Crack turning angle is determined by the sign of KII, which is function of applied stress and geometry but not the material property. It is worth noting that crack deviation was not observed in M(T) under uniaxial loading with weld perpendicular to the loading direction, but observed in the cruciform specimen under biaxial loads; this is due to the influence of KII and T-stress. 4.4.5. Other influential factors Welding-induced distortion was significant in magnitude [18]. In the tests initial distortion was corrected by applying clamping forces at the specimen’s four loading ends. This will affect the initial stress condition in the specimen. The initial distortion was not completely illuminated by clamping. However, the initial out-of-plane deformation and clamping force induced initial stresses were not modelled. This could affect the accuracy in the predicted crack growth rates and trajectory. Test specimens involved two material rolling directions in relation to the welding and crack growth direction, showing that the material rolling direction affects the crack growth path. This is caused by material anisotropy [24-25], weld microstructure change, and residual stresses in both material directions. However, these effects are not modelled. Research also shows that crack location within weld, e.g. in the nugget centre or the HAZ, will also influence the crack growth rates considerably [26].
RkJQdWJsaXNoZXIy MjM0NDE=