Multiaxial fatigue response of normalized 1045 steel subjected to periodic overloads: Experiments and analysis.

摘要

During the past decade it has been observed that periodically applied overloads of yield stress magnitude can significantly reduce or eliminate crack closure under uniaxial or mode I loading. In this thesis the results of a series of biaxial tension-torsion experiments that were performed to evaluate the effects of overloads on the fatigue life of smooth tubes constructed of normalized SAE 1045 steel are presented. Constant amplitude and periodic overload experiments were conducted under five different in-phase stress ratios, including uniaxial ($Y=sxy/sxx=0$), pure torsion ($Y=\infty$), and three intermediate ratios ($Y=12$, 1, and 2). In addition, under torsional loading, two different out-of-phase overloads were evaluated.; Periodically applied overloads of yield stress magnitude caused cracks to grow under crack face interference-free conditions. Strain-life curves were obtained by computationally removing the overload cycle damage from test results and calculating equivalent fatigue lives. A factor of two reduction in the fatigue limit was found at all strain ratios when these results were compared to constant amplitude results.; Another series of constant amplitude and periodic overload tests was conducted on notched axle shafts to evaluate the effects of overloads on a component. The effects of various kinds of bending overloads on torsional fatigue were found to be quite similar to those encountered in the $Y=0$ tubular testing.; Further, under in-phase loading, it was found that shear cracks initiated and grew longitudinally for $Y=0$, $12$, and 1, and for $Y=2$ loading shear cracks initiated on the maximum shear strain planes. These observations concerning the initiation plane were used to unify the constant amplitude data and to justify the use of the maximum shear criterion with the overload fatigue data. Parameter-life curves were developed using the equivalent life data and several common multiaxial damage parameters. It was found that the simple maximum shear strain criterion together with uniaxial overload data gave safe but not unduly conservative life predictions for all of these strain ratios.; Crack face interference-free crack growth curves were obtained for mode I and mode II crack growth and observations of cracking were combined with two models which predict changes in crack growth behavior. The models used were based on crack area increment and strain energy release rate criteria. Both of these models underpredict the shear crack length at which crack growth mode changes, but they do predict the general trends observed in the data in terms of changes in crack growth mode as a function of strain amplitude and load ratio.; Several different crack growth predictions were made. These included baseline predictions wherein the strain concentration profile is determined separately for each load ratio. Predictions were then made using the crack area increment and strain energy release rate criteria. The strain concentration profiles for shear and tensile crack growth were independently determined, and they were then used in the area and energy models to predict strain life curves for all of the stress ratios. The area and energy models gave good predictions of the experimental lives for all of the stress ratios, but the best predictions were provided by the baseline predictions.