ICF10B

intermediate baseline cycles were previously programmed for each test. The crack growth rates were determined by the secant method. Load-displacement behaviour was monitored at all crack measurements for each of the tests using a pin microgauge. The gauge pines were placed in the two drilled holes of 0.5 mm diameter located above and below the centre of the notch. The distance between these points was 3.5 mm. In order to collect as much load-displacement data points as possible during a particular cycle, the frequency was reduced to 0.5 Hz. From the load-displacement records, variations of the opening load Pop were derived using the technique known as maximisation of the correlation coefficient [16]. This technique involves taking the upper 10% of the P-ε data and calculating the least squares correlation coefficient. The next data pair is then added and the correlation coefficient is again computed. This procedure is repeated for the whole data set. The point at which the correlation coefficient reaches a maximum can then be defined as Pop. The fraction of the load cycle for which the crack remains fully open, parameter U, was calculated by the following equation: min max op max P P P P U − − = (2) The values of the effective stress intensity factor range, ∆Keff, are given by the expression: K K K U K op max eff ∆ = − = ∆ (3) RESULTS AND DISCUSSION Figure 2(a) illustrates the typical transient crack growth behaviour obtained when a specimen is subjected to a single tensile overload, a high-low block and a low-high block in a constant ∆K test. In this figure the crack length from the step load event, a-aO, is plotted against the number of cycles from the point of load variation, N-NO, where aO and NO are the crack length and the number of cycles at which the change in load is applied, respectively. Generally, the magnitude and extent of retardation are quantified by the crack growth increment affected by the step in load, ∆aO, and by the delay cycles, ND. ∆aO is the crack growth distance between the point of load variation and the one at which the crack growth rate reaches the steady-state level corresponding to ∆K2. ND is the difference between the number of cycles at which growth to steady state level ∆K2 is achieved and the number of cycles that would occur for the same loading conditions and crack length if no load variation was applied. Figure 2(a) shows that there is a brief initial acceleration of crack growth rate immediately after the overload. The subsequent crack growth rate decreases until its minimum value is reached, followed by a gradual approach to the level of the baseline steady-state. The observed behaviour is usually referred to as delayed retardation of crack growth. The effect of the high-low block is similar to that observed for the peak overload. However, for this load sequence, the retardation is always immediate and not preceded by the acceleration phase. The low-high sequence produces an acceleration of crack growth rate, above the steady state level expected for the high block, followed by a gradual reduction to the corresponding steady-state level ∆K2. These trends are consistent with the behaviour normally reported in the literature [3,5,17]. As expected, the magnitude and extent of crack retardation are higher for the high-low block than for an equivalent single tensile overload. For the overload ∆aO=0.65 mm and ND=12900, while for the high-low block ∆aO=2.64 mm and ND=277500, representing an increase in life of approximately fourteen times. Also, the minimum value of da/dN reached during the retardation phase decreases from 0.32 to 0.009 relative to the constant amplitude da/dN baseline level. Moreover, the crack retardation is even higher for the high-low sequence with ∆K1=9 MPa m1/2 and ∆K2=6 MPa m1/2, than for a 100% overload (∆a O=2.52

RkJQdWJsaXNoZXIy MjM0NDE=