PSI - Issue 19

Masanori Nakatani et al. / Procedia Structural Integrity 19 (2019) 294–301 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

299

6

3.3. Fracture surface

Figure 6 shows the FE-SEM images of fatigue fracture surface near crack initiation site. As shown in Fig. 6(a), radial marks were observed and this suggests that the fracture was caused by a single crack in polished specimen with HIP. On the other hand, it is difficult to define the crack initiation site for the surface roughness of as-built specimen with/without HIP. The existence of ratchet mark indicates that the multiple cracks were simultaneously initiated at the rough surface in the case of as-built specimens. It is considered that the surface roughness of as-built AM contributes to fatigue fracture as shallow and wide defect or circumferential notch. In addition, the defect was not observed on subsurface near crack initiation in the as-build specimen without HIP. This suggests that the surface roughness is a determinant factor in a degradation of fatigue strength for as-built specimen irrespective of HIP. It is noted that the defect also should contribute to fatigue fracture if there is defect near subsurface. In this study, it was revealed that the surface roughness of three-dimensionally complex surface morphology influences the fatigue strength of as-built AM specimen. Here, we discuss the evaluation method for the effect of the surface roughness in AM. As described firstly, the effect of small defect on the fatigue limit can be evaluated using √ area parameter model (Murakami (1994), Murakami (2002)). It has been reported that the √ area parameter model can be adopt to the fatigue limit prediction for machined specimen with artificial surface roughness (Murakami (2002)). The maximum height values in that literature ranged from 20.5 to 74  m. On the other hand, the S z values of as-built specimen fabricated by EBM and DMLS are over 88  m. Therefore, the applicability range of √area parameter model is necessary to be confirmed. Some of authors conducted the fatigue tests for polished specimen without HIP which fabricated by EBM and DMLS as same as this study and investigated the relationship between  K and √area as shown in Fig. 7 (Masuo (2018)).  K value for a crack with arbitrary shape was calculated using a followed equation. Δ = 0.65Δσ√ √ (2) Here,   is stress range (= 2  a , unit: MPa), √ area is square root of projection area of defect observed at crack initiation site (unit:  m). The specimens which have runout for N = 10 7 cycles were tested again at a higher stress and the size of defects at fracture origins was identified. The data agreed with the √ area parameter model indicated by a solid line. In Fig. 7, the dashed horizontal line corresponds to the threshold stress intensity factor range for long crack in a forged Ti-6Al-4V alloy;  K th = 10 MPam 1/2 (Oberwinkler (2010)). Thus, the upper limit of √ area parameter model in tested sample are expected to be about 230  m. On the other hand, the √ area calculated from  w and HV using √ area parameter model (Eq. (3)) are listed in Table 3. w = 1.43( + 120)/( √ ) 1/6 (3) 3.4. Effect of surface roughness

Fig. 6 FESEM images of fatigue fracture surface near crack initiation site.