The effect of spectrum loading variations on crack growth were evaluated analytically and experimentally using C-15, KC-10A, and DC-10 transport aircraft wing loads spectra and 7475, 2024, and 7075 aluminum alloys. A total of 134 spectrum variations were generated in the following categories: (1) baseline spectra; (2) mission mix; (3) sequence of missions; (4) individual flight length; (5) flight segments; (6) exceedances spectra; (7) design stress level or usage severity; (8) valley/peak coupling; (9) low-load truncation; (10) high infrequent loads; (11) clipping of large loads; (12) miscellaneous variations, including the effect of flight loads alleviation system; and (13) combined variations. Spectra were generated as random cycle-by-cycle, flight-by-flight loading sequences. Crack growth tests, with cracks starting as through-thickness or corner cracks on one side of a hole, were performed on 47 of these spectra.

Analysis to test correlation shows 40 percent or better accuracy 78 percent of the time. However, different crack growth models had to be used to match the three aircraft baseline spectra: linear (C-15 and DC-10) and modified generalized Willenborg (KC-10A). Crack growth due to spectrum loading, with respect to loads interaction effects, appears to be a function of retardation and acceleration phenomena. An analysis model, incorporating both of these phenomena, is needed to properly predict spectrum variation effects.

Largest effect on crack growth life due to spectrum variations, as measured in flight hours, was due to mission mix, flight length, design stress level or usage severity, high infrequent loads, load alleviation system, and a change from a wing type to a vertical tail type spectrum. Using spectra variations that could be expected in service, fleet-wide crack growth variations by factors in the neighborhood of 100 and 10 could be experienced, depending on whether a short-term or a long-term spectrum variation is considered.