The crack propagation law was derived from the S–N data in the very high cycle fatigue of a bearing steel. The propagation rate, da/dN (m/cycle), of surface cracks was estimated to be a power function of the stress intensity range, ΔK (MPa√m) with the coefficient Cs = 5.87 × 10−13 and the exponent ms = 4.78. The threshold stress intensity range was 2.6 MPa√m. The crack propagation from internal inclusions was divided into Stages I and II. For Stage I, the coefficient of the power law was C0 = 3.44 × 10−21 and the exponent m0 = 14.2. The transition from Stage I to II took place at ΔK = 4.0 MPa√m. For Stage II, the coefficient was Ci = 2.08 × 10−14 and the exponent mi = 4.78. The specimen size and loading mode did not influence the surface fatigue life, while the internal fatigue life was shortened in larger specimens and under tension–compression loading. For ground specimens, the surface fatigue life was raised by the compressive residual stress, while reduced by the surface roughness introduced by grinding. For shot‐peened specimens, fatigue fracture did not take place from the surface because of a high surface compressive residual stress. The internal fatigue life was reduced by the tensile residual stress existing in the interior of the specimens.

No abstract

No abstract

No abstract

No abstract

No abstract

Near‐threshold fatigue crack propagation tests were performed on circumferentially precracked round bars of a medium carbon steel under torsional loading. The crack propagation rate decreased with crack extension, because of the shear contact of crack faces. The crack propagation rate without the influence of crack‐surface contact was determined by extrapolating to zero crack extension the relationship between the crack propagation rate and crack extension. The applied stress intensity factor range was divided into two parts: one was the effective value responsible for crack growth and the other was the value corresponding to crack‐tip shielding. The resistance‐curve method was used to predict the fatigue limit for crack initiation and fracture. The R‐curve was constructed using the experimentally determined threshold value of the stress intensity range, which was the sum of the threshold effective stress intensity range and the threshold shielding stress intensity range. The threshold effective stress intensity range was constant. The R‐curve was independent of the precrack length and specimen dimensions. The predicted values agreed well with the experimental results.

No abstract