PSI - Issue 25

6

Author name / Structural Integrity Procedia 00 (2019) 0 0–000

R. Baptista / Procedia Structural Integrity 25 (2020) 186–194

191

a)

b)

c)

d)

Fig. 2 Fatigue crack paths on a normal cruciform specimen, with (a) λ=1.0, φ=0º, α=0º, (b) λ=1.5, φ=0º, α=0º, (c) λ=1.0, φ=0º, α=30º, (d) λ=1.5, φ=0º, α=30º.

Applying the algorithm to non-proportional, out-of-phase loading paths, means the crack will propagate horizontally in the majority of the simulated conditions. Part of these results were already verified by Baptista et al. (2019) and experimentally confirmed by Lee et al. (2011). The only exception was obtained for λ = 1.5, φ = 30º, α = 30º, where the cracks propagated vertically. Table 1 shows the obtained fatigue life results and initial crack propagation directions predicted by the MTS criterion, for the different applied load paths. Regarding fatigue life, as shown by Misak et al. (2014), for an initially horizontal crack, FCG life is inversely proportional to λ . Our simulation results show that this behavior is valid for proportional in-phase and non-proportional out-of-phase loadings. For α = 0º and λ = 1.5, FCG lives are always higher then when λ = 1. Also, when a proportional out-of-phase ( φ =180º) loading is applied, this behavior is inverted fatigue life is proportional to λ , as experimentally determined by Lee et al. (2011). Considering an initially inclined crack ( α = 30º), no correlation between fatigue life and λ was determined. As mentioned, if λ = 1.5, φ = 0º, α = 30º or λ = 1.5, φ = 30º, α = 30º cracks will propagate vertically, along a perpendicular direction to the highest stress. The FCG life in both cases is lower than expected. When analyzing the influence of the load phase angle, it is clear, from Table 1, that FCG life is inversely proportional to φ . Again, the only exceptions were the beforementioned cases, where the cracks propagated vertically, leading to an inferior fatigue life. As φ increases, the loading path changes from a tension-tension case, where the crack will be closed during part of the loading cycle, to a tension-compression ( φ = 180º) case, where the crack is always open. Therefore, FCG life will decrease. If plotted against the biaxiality ratio B , one can see on Fig. 3 that FCG life is proportional to B . For λ = 1.5, φ = 0º, α = 0º, the initial B value was 0.96 rendering a fatigue life of 482304 cycles. While for λ = 1.5, φ = 180º, α = 30º, the initial B value was -2.15 rendering a fatigue life of 44834 cycles. The two mentioned exceptions are now clearly visible on Fig. 3. More work will be needed for fully understanding these two cases, but it is clear that T-Stress play an important role not only on crack propagation direction determination and stability, but also on FCG life.

Fig. 3 Normal cruciform specimen fatigue life as a function of T-Stress biaxiality ratio B .