It is now well known that an overload applied during a metal fatigue crack growth experiment causes crack growth retardation [1,2,3]. In addition, underloads subsequent to the overload tend to reduce the retardation phenomena [1]. It has been reported that time at small underload does not seem to play a significant role in its influence on the retardation behavior [3]. The data presented here, however, where the underload is greater, do show such a time effect. This question is of particular significance in terms of relating accelerated laboratory tests to the actual loading seen during the service lifetime of a structure designed to the fatigue crack growth behavior of the material.Fatigue crack growth experiments were carried out on an Al 2219-TdSl (Oy = 50 ksi) compact tension specimen (B = 1 in, W = 4.25 in, H = 2.09 in). Baseline tension-tension loading consisted of a I0 Hz sine wave whose mean value and amplitude are varied so that at each point where an overload is applied (4 points per sample, spaced 3/8 in apart), Kmi n = 4 ksi/in and K = 12 ksi/in. The overload is suchTestlng is done on an MTS clo~p electro-hydraulic testing machine, in room air with relative humidity between 30% and 55%. Crack depth is plotted versus number of cycles. The extension of the steady-state growth lines before and after the overload should be roughly parallel, with a separation of a certain number of cycles which we will call the delay.The relative crack depth is measured by the attenuation of a transmitted longitudinal ultrasonic wave, transmitted and received by two 2.25 MHz piezoelectric transducers.The received signal is converted to a DC signal and is plotted against load. This plot is not only a means of monitoring crack growth, but also shows the crack closure effect reported in [2,3,4], that is, that the crack effectively closes at loads below some Pc > Pmin" An underload of varying magnitude and duration was applied after each overload.Each test was repeated three times, never twice on the same sample or at the same crack length. Results are summarized in Figs. l and 2.The extent of the retardation delay decreases with the time at zero load (Fig. i) and also with the magnitude of the underload (Fig. 2). The significant result shown here is the influence of holding time at zero load. A decrease of about 50% in retardation delay is caused by an underload of about 2 sec. Extending the hold time to 24 hr reduces this by about another 30%. The impact on design philosophy would be even greater if the 2 sec hold time represents the accelerated tests and the 24 hr represents the real loading history. 13 (1977) 248
Int Journ of FractureUsing the crack closure model [2,4], what happens is that when baseline loading is resumed, the crack is not fully open and the crack growth rate is reduced and will not return to the steady-state value until the two rough crack surfaces created during the overload become smooth through cyclic contact. During the underload, these two surfaces are smoothed as they are pressed together. Time ...
