Issue 41

J.A.O. González et alii, Frattura ed Integrità Strutturale, 41 (2017) 227-235; DOI: 10.3221/IGF-ESIS.41.31 233 Figure 10 : FCG rates da/dN and crack opening ratios K op /K max measured under quasi-constant {  K  20MPa  m , R  0.1 } loading conditions by the four redundant techniques (near and far-field strain gages and DIC-based COD and strain fields) along the crack path in the thick DC(T) specimens ( t  30mm ), supposedly under plane strain conditions. The images collected for the DIC analyses were processed by the VIC-3D software using a 35-pixel subset window size, an 8-pixel step size, a 15-pixel strain window size, and normalized sum of squared differences cross correlation functions. Since the load was applied in the vertical direction, the v -displacement and the corresponding ε y strain map were used to identify K op . A pair of symmetrical points was located along the crack faces at 2mm behind the crack tip to obtain crack opening displacement (COD) measurements from the v -displacement field, see Fig. 5(a), whereas the strain history in the y -direction was obtained from a point located at 1mm ahead the crack tip, see Fig. 5(b). Notice that the data points around the crack faces and very near the crack tip were excluded from these analyses, to avoid their intrinsically high noise level. No significant difference was observed in the FCG rates measured in all specimens. Indeed, in all of them it was found that da/dN  10  5 mm/cycle , albeit in the thinner specimens the crack grew under pl-  and in the thicker ones under pl-  conditions, showing that, at least in those tests, the FCG rates are not dependent on the dominant stress state around the crack tip. Moreover, notice that the K op /K max behavior is not identical in all specimens tested under pl-  or pl-  conditions, see Figs. 9 and 10 once again. This indicates that K op is not a property of the geometry/load pair. Instead, it can vary in nominally identical specimens submitted to equal loading conditions not only with the relative crack size a/w , but it can also depend on local details along the crack path, probably because it is also affected by non-PICC mechanisms. This K op variation can also be seen as still another reason to question the blind use of models that suppose that  K eff can always be assumed as the one and sole FCG driving force in all fatigue problems. Finally, to confirm those statements, yet other two tests were performed in similar thin and thick DC(T) specimens, but under slightly different {  K  15MPa  m , R  0.1 } quasi-constant loading conditions, see Figs. 11 and 12. The very same trend, namely K op /K max decreases as a/w increases, measured as described above by the same redundant DIC and compliance techniques, indicates that this behavior is indeed representative of the tested material. Notice that the quasi-constant da/dN ratios measured in those tests are smaller than the ratios measured in the former tests performed under a higher  K , exactly as expected. Notice as well that, although the thick specimen in this new test is thinner than the thicker specimens used in the former tests, it still obeys the plane-strain requirements due to its lower K max . C ONCLUSIONS imple and easily reproducible tests were used to experimentally check if the actual fatigue crack driving force is indeed the effective stress intensity range  K eff  K max  K op , as defended by many fatigue experts. First, a fatigue crack with an initially straight front was propagated with part of this front closed by bending loads, to show that even if some parts of the crack front remain closed, its opened parts can grow by fatigue. This strong evidence indicates S