PSI - Issue 53

Camilla Ronchei et al. / Procedia Structural Integrity 53 (2024) 112–118 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

116

5

, 1 af  − are displayed in Fig. 4. Note that, if the eq,a N - a C point lies outside the ellipse which represents the fatigue endurance condition of Eq. (4), a failure condition is estimated, whereas a run-out condition is predicted if such a point is inside the elliptical domain.

150

150

150

(b)

(c)

(a)

failure run-out

failure run-out

failure run-out

100

100

100

50 SHEAR STRESS AMPLITUDE, C a [MPa]

50 SHEAR STRESS AMPLITUDE, C a [MPa]

50 SHEAR STRESS AMPLITUDE, C a [MPa]

0 150 EQUIVALENT NORMAL STRESS AMPLITUDE, N eq,a [MPa] 0 50 100

0 100 150 200 EQUIVALENT NORMAL STRESS AMPLITUDE, N eq,a [MPa] 0 50

0 150 EQUIVALENT NORMAL STRESS AMPLITUDE, N eq,a [MPa] 0 50 100

Fig. 3. Shear stress amplitude vs equivalent normal stress amplitude for: (a) cyclic axial loading; (b) cyclic torsional loading with R =-1; (c) cyclic torsional loading with R =0.1, by employing w  and w  as fatigue limits.

150

150

150

(a)

(c)

(b)

failure run-out

failure run-out

failure run-out

100

100

100

50 SHEAR STRESS AMPLITUDE, C a [MPa]

50 SHEAR STRESS AMPLITUDE, C a [MPa]

50 SHEAR STRESS AMPLITUDE, C a [MPa]

0 100 150 200 EQUIVALENT NORMAL STRESS AMPLITUDE, N eq,a [MPa] 0 50

0 150 EQUIVALENT NORMAL STRESS AMPLITUDE, N eq,a [MPa] 0 50 100

0 150 EQUIVALENT NORMAL STRESS AMPLITUDE, N eq,a [MPa] 0 50 100

Fig. 4. Shear stress amplitude vs equivalent normal stress amplitude for: (a) cyclic axial loading; (b) cyclic torsional loading with R =-1; (c) cyclic torsional loading with R =0.1, by employing , 1 af  − and , 1 af  − as fatigue limits.

For cyclic axial loading (Figs 3(a) and 4(a)), both experimental failure and run-out conditions are correctly predicted, independent of the pair of fatigue limits considered in the procedure. Regarding cyclic torsional loading with 1 R =− (Figs 3(b) and 4(b)), better predictions are obtained by using w  and w  , since all the experimental failures are correctly simulated. Finally, for cyclic torsional loading with 01 R . = (Figs 3(c) and 4(c)), the present procedure in terms of w  and w  allows to correctly estimate all the experimental run-outs and failures, whereas some failures are not correctly predicted when the experimental fatigue limits are employed in the procedure. For finite life fatigue data, the theoretical results in terms of fatigue lifetime, f N , are compared to the experimental ones, exp N , by considering both the pair of experimental fatigue limits and the computed ones (see Fig. 5). In particular, it can be observed that, when , 1 af  − and , 1 af  − are employed (Fig. 5(b)), about 40% of results falls into scatter band 2 (see the dashed lines in Fig. 5), whereas about 66% into scatter band 3 (see the dot-dashed lines in Fig. 5). However, the results outside the above scatter bands are on the non-conservative side of the plot. On the other hand, with w  and w  (see Fig. 5(a)) , it is possible to obtain better estimations since the results fall into scatter band 2 and 3 for about 56% and 89% of cases, respectively.