PSI - Issue 34

Camilla Ronchei et al. / Procedia Structural Integrity 34 (2021) 166–171 C. Ronchei, S. Vantadori, D. Scorza, A. Zanichelli, A. Carpinteri / Structural Integrity Procedia 00 (2021) 000 – 000

170

5

The comparison between experimental,

exp N , and theoretical,

f N , fatigue lives for both vertical and diagonal

AM specimens is shown in Figs 2(a) and 2(b), respectively, where the solid line indicates f

exp N N  . Moreover, the

dashed lines correspond to

f exp N N equal to 0.5 and 2 (scatter band 2x), whereas the dash-dot lines correspond to

f exp N N equal to 0.33 and 3 (scatter band 3x). In particular, the following remarks can be made:  for vertical AM specimens (Fig. 2(a)), 53% of the results fall within the scatter band 2x, whereas 88% of the results fall within the scatter band 3x;  for diagonal AM specimens (Fig. 2(b)), 64% of the results are included into scatter band 2x, whereas 82% of the results are included into scatter band 3x. When the theoretical results are located outside the above scatter bands, they are on the conservative side.

10 7

(b)

(a)

10 6

10 5

N exp [cycles] 10 3 10 4

N exp [cycles]

Tension Torsion B. in phase B. out-of-phase

Tensile Torsion B. in phase

10 2

N f [cycles] 10 2 10 3 10 4 10 5

10 6

10 7

N f [cycles] 10 2 10 3 10 4 10 5

10 6

10 7

Fig. 2 Comparison of experimental and theoretical fatigue lifetime: (a) vertical and (b) diagonal AM Ti-6Al-4V specimens

Finally, further considerations on the accuracy of the present criterion are drawn by computing the value of the mean square error, RMS T , as was proposed by Walat and Łagoda (2014) . In particular, according to the root mean square error method, the value of RMS T is equal to 1 when a perfect correlation exists between experimental and estimated fatigue lives; moreover, the RMS T value increases with decreasing accuracy of the criterion. Fig 3 shows the RMS T value computed for the different loading conditions here examined: tensile, torsional and biaxial loading. Note that the RMS T results do not take into account the build direction, since no evident differences in terms of fatigue behaviour are observed between vertical and diagonal specimens. From such a graph, it can be observed that better estimations are obtained for tensile and biaxial fatigue loading (with a RMS T value equal to 1.54 and 2.76, respectively), whereas the RMS T value is slightly greater than 3 for torsional loading. It is worth pointing out that the accuracy of the present criterion is significantly affected by the large dispersion of experimental data, typical of AM metals under fatigue loading. 5. Conclusions In the present research work, the fatigue behaviour of AM Ti-6Al-4V specimens under both uniaxial and biaxial cyclic loading has been theoretically investigated. Fatigue tests available in the literature have been simulated through a critical plane-based fatigue criterion proposed by Carpinteri et al. (2015). The fatigue life has been computed by using an equivalent strain amplitude (also named fatigue damage parameter) together with the tensile Manson-Coffin curve. The accuracy of such a criterion has been evaluated by comparing the theoretical results (in terms of fatigue life) with the experimental ones.