An experimental analysis of fatigue crack growth under random loading

An overall 235 tests involving load histories representing seven stationary Gaussian random processes and featuring different power spectral density functions [psd, S((omega))] of different shapes for the loads and/or different root mean square were conducted with a view to determining the influence of various loading parameters (viz. history length, stress level and bandwidth) on variability in crack growth life. Tests were carried out on 2024-T351 aluminium compact tension (CT) specimens. The results obtained suggest a strong influence of history length (viz. the number of cycles in each history) on variability in crack growth life. The number of cycles a crack takes to grow from an initial value to a final one at a given load level is strongly dependent on the bandwidth of the process that causes the crack to grow. Moreover, several statistical distributions fit the test data reasonably well.