PSI - Issue 22

6

Y. Ai et al. / Structural Integrity Procedia 00 (2019) 000 – 000

Yang Ai et al. / Procedia Structural Integrity 22 (2019) 70–77

75

250

10 3

Al 7075-T6 Estimated Kt =2 tests

Al 7075-T6 Estimated Kt =4 tests

200

10 2 Maximum nominal stress  max /MPa

50 Maximum nominal stress  max /MPa 100 150

10%

50%

10%

90%

50%

90%

10 3

10 4

10 5

10 6

10 7

10 2

10 3

10 4

10 5

10 6

10 7

(a) P-S-N curves of Kt =4 specimen Number of cycles to failure, n

(b) P-S-N curves of Kt =2 specimen Number of cycles to failure, n

800 1000

Al 7075-T6 Estimated Kt =1 tests

600

200 Maximum nominal stress  max /MPa 400

10%

50%

90%

10 2

10 3

10 4

10 5

10 6

10 7

Number of cycles to failure, n

(c) P-S-N curves of Kt =1 specimen Fig. 2 Predicted P-S-N curves for Al 7075-T6 specimens with different geometries

Shown from Figs. 1-2 that, the estimated P-S-N diagrams agree well with experimental results, both in terms of median curve and probabilistic scatter bands, which proves the proposed model describe well on the geometrical and statistical size effect. For the two alloys, the predicted results derived from different combinations verify the relationship between fatigue life and equivalent highly stressed volumes. Under the same maximum local stress, the specimens with larger highly stressed volume have shown lower fatigue life. Thus, this probabilistic procedure is a powerful tool to complete the transfer between structures with different geometries and volumes. 4. Conclusions In this paper, by combining the concept of HSV with Weibull distribution, a probabilistic model considering both effects of notch and size is proposed for transferring fatigue life distribution of specimens with different geometries. The following conclusions can be summarized: (1) A special relation under the assumption of equivalent highly stressed region is introduced to take both geometrical and statistical size effects into account. Moreover, a dynamic model coefficient considering different maximum local stresses is elaborated to characterize the relationship of different equivalent highly stressed volumes on fatigue life. (2) Using the highly stressed node method of FE analysis under the stress values within 80% of the evaluated maximum stress, the highly stressed volume of two alloys is estimated. Results indicate that experimental data of all predicted specimens are almost within the estimated P-S-N curves with 90% survival probability derived from those of reference specimens, which provide well correlations between model