A criterion for omission of variable amplitude loading histories

Abstract The standard random loading sequence EC9 and the cycle-by-cycle sequences TWIST and MiniTWIST have been edited by a procedure based on notch or local (hot spot) stress, and the results of tests on specimens of FE 690 and AIMg4.5Mn using original and edited spectra are compared. Omission of small cycles below a filter level of about 50% of the materials' constant amplitude endurance limit was found to be ‘allowable’; this figure was confirmed by a variety of literature data. Using this level, the crack initiation lives (compared with the fatigue life under the unedited sequence) were longer by 10–30% for EC9 and by 20–55% for TWIST and MiniTWIST. In addition, for the latter two spectra, the crack propagation lives were longer by up to 90% with a detrimental influence of higher loading levels. Total life of the notched aluminium specimens increased by 30–70% using a filter level of 50% of the material's endurance limit.