After each period of 500 thermal cycles, the specimens were removed from the furnace and investigated for cracking at a low magnification (10 to 60x). A rough estimate of crack initiation lifetime was defined as when a full-face hairline crack was visible at the base of the notch. Extrapolations of the long crack growth data were used to further refine when initiation was to have occurred. Any subsequent crack growth after initiation was measured on each side of the specimen. Through crack depth was taken as the average of these two measurements. TEST RESULTS Times to crack initiation in the notched specimens are shown graphically in figure 1. Unlike the authors’ previous work [5], where elastic stress amplitude was used, times to crack initiation are plotted against the Neuber pseudostress amplitude ( ). The pseudostress is defined as the local strain amplitude at the notch ( ε a S' a) multiplied by Young’s Modulus (E). S Ea a ε = ' (1) When the behaviour of the material remote from the notch is predominantly elastic, the pseudostress amplitude can be related to the nominal stress amplitude (Sn) and the local stress amplitude ( σa). The final relationship, outlined in work by Prater and Coffin [7], is reproduced below: ( ) a f n a k S S σ 2 ' = (2) Here kf is the notch fatigue factor, determined from a notch sensitivity analysis as used in [5]. A maximum value of 5.0 has been established for kf in accordance with a “worst case notch” analysis. Use of equation (2) also requires knowledge of the cyclic stress strain relationship for the material. Prater and Coffin [7] provide data for carbon steel that has been adapted for use in this investigation. Figure 1 also shows a curve based on ASME fatigue data. This curve is slightly different to the published design curve as shown in Section VIII of the ASME Code [2]. The design curve has a built-in conservatism of 2 times on stress and 20 times on lifetime on the fatigue data, whichever is the most conservative. Crack growth results are shown on figure 2. Stress intensity factors have been plotted against the crack growth rate using the R-ratio to categorise the data. Both water and air environment crack growth curves for carbon steel, taken from the ASME Boiler and Pressure Vessel Code [3], are also plotted for comparison. DISCUSSION Crack Initiation As shown in Fig. 1 and reported in [5], the application of a primary load displays little or no effect on crack initiation lifetime within the limitations of experimental variance. The same can be said for reducing the dissolved oxygen level from 8ppm to 2ppm. In all of the test cases, the ASME fatigue curve seems to provide a good approximation of the relation between number of cycles to crack initiation and notch pseudostress amplitude. However, any attempt to use the ASME curves to predict crack initiation should be approached with care. This is because the ASME fatigue curves are based on uniaxial strain controlled testing of small cylindrical specimens. Failure in one of these specimens is defined as when complete fracture occurs. It is then assumed that the failure of these small specimens is equivalent to the initiation of a small crack in a larger structure.
RkJQdWJsaXNoZXIy MjM0NDE=