13th International Conference on Fracture June 16–21, 2013, Beijing, China Case 1: a0=9.32 µm, σa=360 MPa, R=48 µm, a= 2.4 mm, Nexp=5.5×108 cycles on a pre-corroded specimen α aint/a0 Naint-a0 Na0-ai Nai-a 100 0.9 115,801 44,151 4,507,080 0.94 13,163 0.97 2,300 Case 2: a0=41.5 µm, σa =160 MPa, R=300 µm, a= 2.6 mm, Nexp=1.83 ×108 cycles on specimen tested under sea water flow α aint/a0 Naint-a0 Na0-ai Nai-a 100 0.9 515,466 168,014 10,258,700 0.94 58,593 0.97 10,237 Table 2: Hemispherical surface crack growth vs. experimental fatigue life, for x=3, and different aint/a0 ratios. Microscopic observations of the cylinder surface showed that R5 steel was damaged by fatigue – corrosion in the areas where the stress amplitude was higher than 56 MPa, a few corrosion traces only were observed where the stress amplitude was smaller (Figure 9d). The final fracture occurred where the stress amplitude was between 112 and 120 MPa. This shows an experimental evidence of the coupling between the corrosion process and cyclic loading, even at ultrasonic frequency (20 kHz), since below 56 MPa without any sea water flow there is no failure after 2×108 cycles (see the SN curves Figure 3). Futhermore, three additional tests were carried out on VHCF specimens under the same conditions than all the other tests but, without any cyclic loading. Only the sea water flow was applied on the specimen surface with the same flow rate than for the tests with cyclic loading. The aim of these tests was to quantify the corrosion damage after the same duration than 107 cycles (8mn20s at 20 kHz), 108 cycles (1h23mn20s at 20 kHz) and 109 cycles (13h53mn20s at 20 kHz). No corrosion trace was observed by SEM after 8mn20s. After 1h23mn20s only a few traces of corrosion appear but not pits were observed. Corrosion pits were observed at the specimen surface after 13h53mn20s. The typical size of these pits was varying between 30 and 60 µm (Figure 11) this is significantly smaller than under simultaneous cyclic loading and sea water flow (Figure 8). The same pit size was observed under sea water flow with a stress amplitude of 240 MPa after 5×107 cycles. This corresponds to 42 mn only at 20 Khz! This proves that there is a coupling between corrosion and cyclic loading. Because of such coupling, the question of the roughness effect occurred. Complementary fatigue tests were carried out under sea water flow at a stress amplitude of 250 MPa on the same VHCF specimens but with a polished surface (Ra=0.1 µm) to investigate a possible roughness effect. There was no evidence of the surface roughness effect under sea water flow (Figure 3): the fatigue life is in the scatter band of other experimental data (with Ra=0.6 µm). This work shows experimental evidence of the coupling between corrosion and cyclic stress, even at 20 kHz while some authors think that fatigue corrosion interaction is not active when the loading frequency is high because of the long characteristic time of the corrosion process compared with the loading period. The observations of this study tend to show that the number of loading cycles (not the time) is a key factor in the fatigue crack initiation phenomenon under sea water flow. This has been recently confirmed by El May et al. [17] on a martensitic stainless steel in NaCl aqueous solution in HCF regime with in-situ fatigue test at around 100 Hz. These authors showed that PSB due to cyclic loading break the passive layer and then corrosion pits appears before crack initiation.
RkJQdWJsaXNoZXIy MjM0NDE=