13th International Conference on Fracture June 16–21, 2013, Beijing, China With these three different regimes [12] the very quick propagation of short crack and the quick propagation of small crack compared to long crack were considered. The detailed equations are given in [12], they were applied to our data assuming that the initial crack has a length, a0, corresponding to the corner: da/dN=b. Figure 9: a) Longitudinal cross section of a specimen showing b) a fatigue crack initiation at a corrosion pit and the yellow highlighted old austenitic grains, c) illustration of the short fatigue crack and the grain boundaries. Two cases are shown in Table 2 with two experimental total fatigue lives Nexp; several additional cases are detailed in [12]. In the first case 99.1% of the fatigue life is due to initiation. For the second case, 94% of the fatigue life was consumed by the initiation phase. Furthermore, it has been shown in [12] that with a high value of α and large aint/a0 one obtains similar NTotal results than with a low α and small aint/a0, that is because with a higher α the crack does not grow as far due to the slope of the da/dN curve in the threshold region, then aint/a0 must be larger, and with a lower α, aint/a0 must be smaller. This calculations show that crack initiation dominates the fatigue life. Even if our calculations were carried out by using the crack growth curve in air, the conclusion is still valid because the crack growth rate for the R5 steel under sea water flow is higher than in air as shown Figure 8 and usually reported in literature [15]. Since the fatigue life is dominated by the crack initiation regime even in corrosive environment, additional experiments have been carried out to try to understand why this initiation period is significantly affected by sea water flow. 5. Additional experiments and discussion about the coupling between cyclic loading and corrosion To complete these tests an ultrasonic fatigue test has been carried out on a cylindrical specimen. Since during an ultrasonic fatigue test a stationary wave is applied on the specimen, all along this cylinder different stress amplitudes are applied. Figure 9B shows the distribution of the stress amplitude along the cylindrical specimen loaded in its central part with a stress amplitude of 120 MPa (highest value along the specimen). This corresponds to a mean fatigue life of 2×108 cycles. For observation, the specimen surface was divided in 13 zones of around 1 cm long, each one was loaded to a specific stress amplitude interval. By testing such a cylinder under sea water flow (Figure 9a), observation of the damaged areas allowed us to investigate the link between the stress amplitude and the damage due to the in-situ fatigue – corrosion process. The specimen failed after 7.37×107 cycles.
RkJQdWJsaXNoZXIy MjM0NDE=