Modeling of Localized and Structural Fatigue Damage Under Complex Multiaxial Loading

Abstract

Fatigue is the final test for many–if not all–properly designed metallic structures. Making appropriate assessments of progress towards fatigue failure, or a design’s resilience against it, however, is not a simple task. Given the extraordinary time and effort to carry out fatigue testing, there has been a lot a interest in fatigue modeling, but significant challenges remain both due the advent of new alloys as well as the large number as well as complexity of loading cycles seen by the part during service. Methods in the time domain were the first to be used to model the progress of fatigue. Current techniques (typically an amalgamation of notch correction, elastic-plastic modeling scheme, cycle count, and damage criteria) rely on many assumptions which may impair their applicability. Frequency domain methods are available, and these hold the appeal of rapid analysis over entire structures, but these methods are restricted in the class of responses that they can handle. These limitations influence the accuracy of analyses even in the cases in which modifications are made to the base model; for instance, modifications available in the literature made to allow elastic-plastic (EP) response as part of a spectral fatigue life calculation completely lose any load ordering effects, though it is known that the ordering of loading blocks can have a pronounced effect on the resulting life. Loss of load ordering is not a problem relegated to the frequency domain either; many of the commonly-used cycle counting techniques available in the literature require a posteriori rearrangement of the input data meaning the accuracy is tied to the chunk size processed, by design. Inability of many current models to handle input data upon receipt–adjusting structural fatigue damage predictions on-the-fly–can hurt more than their accuracy; it largely prevents their use in high-tech applications such as the production of a “digital twin" for, e.g., real-time diagnostics. In this work, we investigate the application of an accumulated micro-plastic work (AMW)-based criteria for fatigue calculations combined with a continuous-flow plasticity model, an approach which resolves many of the issues inherent in previously-used techniques. No cycle count or qualitative damage parameters are required, and the effects of load ordering will always be maintained. Through addition of an appropriate kinematic hardening model, it is shown how the continuous-yielding plasticity model combined with the AMW criterion can handle many different load cases with high accuracy. Finally, we combine this fatigue prediction model with specialized elastic-plastic continuous flow structural elements, demonstrating its capabilities with fatigue pre-dictions by comparing against Al-7075 dogbone specimens tested in the literature, and simulating several polymer microarchitected structures. By introducing these tools with varying levels of granularity and computational complexity, we can enhance the flexibility of the practitioner in carrying out fatigue analysis; this could make whole-structure fatigue analysis practicable in design or, conceivably, construction of real-time digital twins for metal aircraft health assessment.