Issue 37

K. Yanase et alii, Frattura ed Integrità Strutturale, 37 (2016) 101-107; DOI: 10.3221/IGF-ESIS.37.14

According to the previously reported results [3-7],   0.18 k holds for carbon steels, Cr-Mo steel, ductile cast irons, and high tension brass. If the effect of biaxial stress is negligible (i.e.,  0 k ), the torsional fatigue limit is solely determined by the major principal stress as follows (cf. Fig. 6):



HV area

1.43(

120)









(3)

w

w

1/6

(

)

On the other hand, Schönbauer et al. proposed that the fatigue limit in the presence of a large defect or a long crack can be estimated by the following equation [2]:

 2 0.65 K 

th,lc





(4)

w



area

th,lc ΔK =6.7MPa m for the investigated 17-4PH steel at R = ˗ 1 was

where the threshold stress intensity factor range of

determined by Schönbauer et al. [9]. Fig. 7 shows the relationship between the shear stress amplitude  a and area . When a specimen endured the number of cycles N = 1.2  10 7 , it was regarded as a run-out specimen. As shown, the prediction for fatigue limit with k = 0 (Fig. 7(b)) renders higher correlation to the experimental data than using k = ˗ 0.18 (Fig. 7(a)). Further, for the torsional loading, fatigue limit can be reasonably estimated by the two prediction lines:    0.6 (1.6 ) a HV and the threshold for long crack (cf. Eq. (4)).

400

400

a  = 0.6  (1.6HV)



"1‐hole defect"

a  = 0.6  (1.6HV)



"1‐hole defect"

350

350

"2‐hole defect"

150 Shear stress amplitude,   a  (MPa) 200 250 300

150 Shear stress amplitude,   a  (MPa) 200 250 300

"2‐hole defect"

 area parameter model with "k = 0"             

"3‐hole defect"

"3‐hole defect"

 area parameter model  with "k = ‐0.18"

Threshold for long crack

Threshold for long crack

Failure Run‐out Prediction for fatigue limit

Failure Run‐out Prediction for fatigue limit

100

100

10 1

10 2

10 3

10 1

10 2

10 3

 area (  m)

 area (  m)

(a) Prediction with k = ˗ 0.18 (b) Prediction with k = 0 Figure 7: Relationship between shear stress amplitude  a and area .

In Fig. 8, fatigue limit for torsional loading and tension-compression loading is compared [2]. It is noted that, in the tension-compression data, various types of defects are considered (e.g., drilled hole, circumferential notch, corrosion pit). As shown, the respective fatigue limit can be reasonably estimated by considering the major principal stress (i.e., k = 0) and area . The torsional fatigue limit becomes insensitive to the defect when  100μm area , which is in strong contrast

105