ICF13A

13th International Conference on Fracture June 16–21, 2013, Beijing, China -5- overall the MB technique could reliably provide 100 m crack depth measurements. Fractography showed that cracks nucleated from 1) surface scratches; 2) particles or pores; and possibly 3) intragranular cracking. Many of these naturally nucleated cracks showed fairly rough fracture surfaces. Both corner and bore surface cracks, single and multiple were observed in various SENT coupons, as shown in Figure 5-b). In general, the small crack growth rate data showed large scatter bands and irregular ‘zigzag’ (acceleration/deceleration) growth. The irregular growth rate was mainly due to microstructure effects (such as grain size, orientation, and grain boundary) associated with the scattered residual stresses, and secondarily caused by the measurement resolution/accuracy. More data analysis showed that the absolute measurement error (~5 m) of the replication technique contributed only partially to the scatter band. Overall, the irregular ‘zigzag’ features were largely reduced after the crack size was over ~100 m or dc/dN was about 3.94E-7 in/cycle (1E-8 m/cycle). 1.E‐04 1.E‐03 1.E‐02 1.E‐01 1.E+00 0 50000 100000 150000 200000 250000 µm/cycle Cycles J407‐2F‐3 Growth Rate Curve da1/dN (Best measurement) Assumed error: ‐5 um (uniform) Assumed error: 5 um (uniform) Assumed error: 5 or ‐5 um (random) Assumed error: ‐5 or 5 um (random) 1. Miscrostructural effects are evident even considering the measuring errors 2. Microstructural effects are strong when da/dN is below 1E‐ 2 um/cycle (1E‐8 m/cycle) Figure 6. Typical surface small crack growth rate scatter (R= -0.3, MAX= 20 ksi/138 MPa) 3. Small crack stress intensity factor modeling A three-dimensional (3D) p-version StressCheck FE model was developed to calculate the stress intensity factors of the naturally nucleating cracks measured on the SENT test coupons [5]. The model was developed from a modified quarter-elliptical corner crack geometry at a hole. A useful feature of this model is that it was built as a parametric FE model, in which, as seen in Figure 7, various parameters can be assigned to the geometry of the crack, coupon, and mesh layers enclosing the crack front. Symmetry conditions can be assumed for surface crack modeling at a hole. The K- solutions were calculated at 96 points along the crack front, using the contour integral method with circular integration paths located between the first two concentric element layers. Eight elements were located along the crack front, with the extreme ones covering only the first and last 2 degrees to capture the peak stress intensity factor values close to the free surfaces. The analysis was performed with 8th degree polynomials. The process was automated using an Excel-based Visual Basic for Applications (VBA) program that could automatically configure the model, launch the analysis, and retrieve output results for a series of crack measurements in a specimen. Previous verification examples have shown that the improved NRC solutions, developed from the hole geometry version of the aforementioned model, are more accurate than the classical Newman-Raju solutions, and comparable to the Fawaz-Andersson results tabulated in AFGROW, which however do not cover cracks smaller than 10% of the plate thickness and do not include the surface crack case (which are now available in the NRC solutions) [5].

RkJQdWJsaXNoZXIy MjM0NDE=