Measured fatigue crack growth increments versus predictions for small cracks in 7XXX aluminium alloys

Whilst crack growth retardation following the application of an overload during variable amplitude loading is a well-known spectrum effect, the order of smaller amplitude cycles can also affect crack growth. One method of accounting for such spectrum effects in crack growth prediction algorithms is by correcting the estimated effectiveness of each cycle or group of cycles. Such predictions are usually verified by comparison to demonstrated total fatigue lives. These corrections have been shown to be suitable for cracks that are at least several millimetres long. However, for physically small cracks: here defined as less than a few mm in depth, such predictions are often far less accurate. In the case of aluminium alloy aircraft structures, where cracks typically grow approximately exponentially from small discontinuities, these predictive algorithms should also be evaluated for their ability to predict small crack growth rates, rather than total life alone. To examine the more subtle crack growth acceleration and retardation effects from individual load cycles in typical aircraft structural 7XXX aluminium alloys, this paper presents the results of quantitative fractography measurements of very small fatigue crack growth increments produced under the application of specially engineered spectra. One such spectrum that contained short segments of constant amplitude loads of various mean stresses, was designed to examine the extent of crack growth retardation, such as from closure, at a range of small crack lengths. Other spectrum effects were also explored through use of test spectra with constant amplitude cycles that were either separate to or embedded within small segments of variable amplitude loading. The local growth produced in each case was then measured and compared to predictions produced by two closure/retardation correction methods; based on wake and plastic zone related phenomena. The results showed that the prediction methods, although they made a