Fatigue and Crack Growth in 7050-T7451 Aluminum Alloy Under Constant- and Variable-Amplitude Loading

Abstract

The 7050 aluminum alloy is used in many aerospace structural applications. Previous studies have identified that fatigue cracks develop very rough crack-surface profiles, which cause very high crack-closure levels due to a combination of plasticity, roughness and debris. Previously, tests were conducted on compact, C(T), specimens to generate crack-growth-rate data from threshold to near fracture over a wide range in stress ratios (R). New threshold testing methods, based on compression precracking, were used to generate the data in the near-threshold regime. The plasticity-induced crack-closure model, FASTRAN, was used to correlate the data over a wide range in stress ratios and crack-growth rates from threshold to near fracture. To account for the very high crack-closure levels, a very low constraint factor, like plane-stress conditions, had to be used in the model. In addition, the crack-opening loads were measured during these tests using a local strain-gauge method to generate another ΔKeff-rate curve. These two curves differed only in the near-threshold regime. Herein, fatigue-crack-growth tests were conducted on C(T) specimens under spike overloads and simulated aircraft spectrum loading. Fatigue tests were also conducted on single-edge-notch bend (SEN(B)), specimens over a wide range in loading conditions (constant amplitude and three aircraft spectra). All specimens were machined from a single forged block of 7050-T7451. However, no residual stresses were measured in both the SEN(B) and C(T) specimens. Two European standard spectra were used, but modified to have only tension-tension loading. The purpose of this paper was to evaluate the two different effective stress-intensity factor curves for making crack-growth and fatigue-life predictions. Small-crack theory was used to make fatigue-life predictions using inclusion-particle sizes from the literature. Fatigue predictions on the SEN(B) specimens agreed fairly well (±30%) using a 12-micrometer semicircular initial flaw located at the semicircular-edge notch under all loading conditions, except the model was unconservative (factor of three) on one of the severe aircraft spectra (Mini-TWIST+, Level 1). For the C(T) specimens subjected to single-spike overloads, the life-prediction code also produced much more retardation than observed in the tests. However, the predicted crack-length-against-cycles under the Mini-Falstaff+ spectrum were only about 15% longer than the tests. The discrepancy under the single-spike overloads and the severe aircraft spectra was suspected to be caused by the low constraint factor and/or crack paths meandering around overload plastic zones. Ideally, a roughness-induced crack-closure model; in addition to the plasticity model, would be needed to obtain more reasonable results.