Discussion on surface crack growth criteria under cyclic loading

Abstract

Various approaches are known concerning the transfer of test results obtained on standard specimens with through-the-thickness cracks under cyclic loading to structures with surface flaws [1][2][3]. Within the framework of linear elastic fracture mechanics, the range of the local (AK), averaged (AI~) and effective (AK) values of the stress intensity factor are used as surface crack growth criteria [3]. Meanwhile, the known investigations do not allow one to determine recommendations as to the choice of surface crack growth criteria under cyclic loading (see for instance [3]). In the present paper, based on the original experimental data_, the authors performed a comparative analysis of the applicability of the AK, AK and AKo values as surface crack growth criteria.The surface flaw propagation was studied on plates of length L = 220 mm, width 2W = 116 mm and thickness h = 30 mm made of pressure vessel steel 15Kh2MFA (cy02 = 584 MPa, ~, = 700 MPa, 5 = 21 percent, V = 74.6 percent [4]). Basic fatigue crack growth properties of the material were determined on standard single-edge-notch specimens 210x50x25 mm in size. The experiments were carried out in four-point bending fatigue at the frequency of 10 Hz and the stress ratio R = 0.32 in air, at room temperature. The value of fatigue crack growth rate in testing varied within the Paris region of the fatigue crack growth curve. The test conditions for standard specimens satisfied the known requirements which restrict the size of the plastic zone at the crack tip and the nominal stress level. The surface-cracked plates were tested in the nominal stress variation range of ~.om = (0.38...0.9) ~0.2.In order to initiate surface cracks, notches of 0.5 to 5 mm deep were introduced at a specimen mid section by using the electro-spark technique. The beach mark method was applied for monitoring the crack geometry. For this purpose the load amplitude was reduced by 50...70 percent while the maximum load level was kept constant; depending upon the fatigue crack growth rate the beach mark procedure held from N = 3000 to 10000 cycles. After final fracture the crack shape, size and growth rate were determined based on the beach mark measurements. Figure 1 presents a photograph of the specimen fracture surface containing beach marks. Evidently, the real surface crack front is well approximated by a semi-ellipse, a fact also noted by the authors of [3,[5][6][7]. Insignificant crack front deviations from the elliptical shape which may be observed near the free surface of a plate are explained by a comparatively high level of plastic deformations in Int Journ of Fracture 46 (1990) R26 the vicinity of the corresponding point of the crack front, as well as by the variations of strain constraint and local crack resistance properties of a material over the specimen thickness and by a redistribution of stresses due to plastic strains [7]. Generally the crack front location is described by the infinite number of degrees of freedom. However, Wu [8] determined that in predicting semi-elliptical ...