Fatigue and fracture testing and analyses were performed on four engineering materials: a low-strength aluminum alloy (D16CzATWH), a high-strength aluminum alloy (Al7050-T7351), a low-strength steel (A36 steel), and a high-strength steel (9310 steel). Large-crack testing included compression precracked constant amplitude and compression precracked load reduction over a wide range of stress ratios. Single- and multiple-spike overload tests were conducted on some of the materials. Fatigue and small-crack testing were also performed at constant amplitude loading at a constant load ratio on the newly designed single edge notch bend specimen. Using the FADD2D boundary element code, two-dimensional stress analysis was performed on the new specimen to determine the stress intensity factor as a function of crack size for surface and through cracks at the edge notch. Collected fatigue crack growth rate data was used to develop a material model for the FASTRAN strip-yield crack growth code. FASTRAN was used to simulate the constant amplitude and spike overload tests, as well as the small-crack fatigue tests. The fatigue crack growth simulation results have shown that both low-cycle and high-cycle fatigue can be modeled accurately as fatigue crack growth using FASTRAN and that FASTRAN can be used to accurately predict the acceleration and retardation in fatigue crack growth rates after a spike overload. The testing has shown that the starting fatigue crack growth rate of any load-shedding test has significant influence on load history effects, with lower starting rates yielding lower crack growth thresholds and faster rates. Through inspection of fatigue surfaces, it has been shown that beveling of pin-holes in the crack growth specimens is necessary to ensure symmetric crack fronts and that the presence of debris along the fatigue surfaces can cause considerable crack growth retardation.

In the past, the disagreement of near-threshold fatigue-crack growth (FCG) rate data generated from constant Kmax tests, high load ratio (minimum to maximum load) constant R tests, and ΔKeff based data was a mysterious issue. Because of the disagreement, a variety of test or analysis methods were created to correlate FCG rate data. It was suspected that the ASTM threshold test method using load reduction was inducing remote crack closure due to plastically deformed material, which caused elevated thresholds and slower rates than steady-state behavior. The first goal of this study was the development of a test method to eliminate remote closure during threshold testing. In order to avoid/minimize remote closure effect, compression-precracking methods were used to initiate a crack from a starter notch on compact specimens. Two materials with different fatigue crack surface profiles (flat or very rough) were tested and the results generated from the conventional ASTM precracking method and the compression-precracking test method were compared. In order to understand the disagreement of near-threshold data, crack-opening load measurements were performed from locally (near crack tip) installed strain gages instead of the remote gage (i.e., back face gage). Some careful specimen preparations were performed to avoid out-of-plane bending, to maintain straight crack fronts, and to ensure testing system linearity. It was known that remote gages, such as crack-mouth- opening-displacement-gages were insensitive to measuring load-strain records near threshold. By using local gages, the crack closure effects were clearly observed even in high load ratio (R) tests, like or higher than R = 0.7, and constant Kmax tests, which were believed to be crack closure free. By measuring load-reduced-strain records from local gages, crack-opening loads were able to correlate FCG rate data and showed that ΔKeff-rate data was unique for a wide variety of materials. By comparing (ΔKeff)th values, it may provide reasonable guidance for the material resistance against FCG. Because of “high R crack closure”, some theories considered in the past may need to be reconsidered. First, constant Kmax tests are not entirely crack-closure free. Second, there is no critical load ratio, Rc, to indicate the transition from crack-closure affected to crack-closure free data, and Kmax effects that appear in ΔKth-Kmax relations. Research has shown that the three dominate crack-closure mechanisms (plasticity-, roughness- and debris-induced crack closure) FCG rate behavior in the threshold regime from low to high load ratios.