PSI - Issue 19

Yukitaka Murakami et al. / Procedia Structural Integrity 19 (2019) 113–122 Yukitaka Murakami et al. / Structural Integrity Procedia 00 (2019) 000–000

119

7

Fig. 9 Schematic illustration of crack growth behavior for surface rack and internal crack.

7. Statistics of extremes analysis of defects toward standardization Since fatigue strength is the most important parameter for structural integrity, we need to consider the effect of defect, surface roughness and their statistical properties in the structural integrity standard. The first step is quality control of basic processing by sample coupons. By cutting the sections of a coupon, we can obtain the statistics of extreme values of defects (Fig. 10 (a) and (b)). This step is very similar to the method already introduced to the inclusion rating of commercial steels such as bearing steels (Murakami (2002)).

Fig. 10 Statistics of extremes of defects of in AM coupons. (a) Data acquisition by coupon, (b) Statistics of extremes of defects in coupons.

The second step is to obtain the statistics of extremes of defects at fatigue fracture origins of fatigue specimens (Fig. 11(a) and (b)). These data are useful for fatigue design of actual parts. Figure 11 (b) compares the statistics of extreme values of defects observed at fracture origins of specimens manufactured by three different machines. We can say that the quality of the material manufactured by the new machine is higher than those of the old machines. The third step is to obtain the statistics of extremes of defects at fatigue fracture origins of actual parts. The image is illustrated in Fig. 12. The failure tests by overloading are necessary for data acquisition to predict the lower bound of fatigue strength of actual parts (Murakami et al (2017)). By these steps, we can confirm the quality of processing and actual parts. In reality, the critical location of real components is very often at stress concentration. Such critical locations exposed to high stress concentration is actually limited to a small volume compared to the total volume of the component. This nature of statistics reduces the probability of presence of large defects at such a stress concentration cite.