13th International Conference on Fracture June 16–21, 2013, Beijing, China -8- experimental observations. Hence, both experimentally observed notch life sensitivity trends at low and high notch severity values are captured simultaneously with a single parameter variation. Figure 10. Domain for area averaging 2.0 tk notch using (a) surface-sweep method and (b) circular-sweep method. Figure 11. Surface-sweep nonlocal average OD parameter for OP TMF under 500 MPa net section stress for increasing area domain. Optimal areas for predicting 1 tk / 1.3 tk and 2 tk / 3 tk OP TMF notch life sensitivity trends were found to be 0.90 mm2 and 0.77 mm2 for surface-sweep and circular-sweep domains, respectively. The sweeping length scale, L, is shown for each method and notch geometry in Table 1. Table 1. Sweeping Length Scales for Optimized Area Domains The surface-sweep scale is smaller than the circular-sweep scale as the sweeping area is performed over the entire notch surface length. The length scale changes to preserve the same size area for averaging the damage parameter in each notch geometry. Relevant length scales here are larger than the 200 μm length scale for point and line methods utilized by Moore and Neu for isothermal fatigue of the some notch geometry and material [16]. Naik et al. [14] found the line average domain to be between 20 μm and 100 μm for notched Ti-6Al-4V with stress concentrations varying between 2 and 4. For a notched Al alloy, Susmel and Taylor [12] utilized a 244 μm length scale for implementing a point method approach and a 154 μm length scale for a line method approach. In the same study Susmel and Taylor also investigated low-carbon steel and utilized 492 μm and 396 μm scales for point and line methods, respectively. It is critical to note however that the length scales obtained from the literature were determined from a line-method approach, whereas here an area method approach is utilized. As such it is expected that the scale will be larger here. Additionally, as the columnar grain size is fairly large in this alloy, and hence the relevant length scale can be expected to be on the same order as the grain size (0.2 mm to 1.0 mm). The correlation of the area averaged damage parameter with experimental OP TMF life is shown in Fig. 12. Both averaging methods improve the life predictability as compared against the local method and critical area nonlocal approaches presented previously. With the exception of one life
RkJQdWJsaXNoZXIy MjM0NDE=