From mechanism empirical to physics-informed: a comprehensive review of multiaxial non-proportional low-cycle fatigue life prediction

Purpose - During the service life, engineering materials often suffer from alternating loading from different directions simultaneously, causing unexpected multiaxial fatigue damage. In response to this issue, scholars measure the differences in different loading paths through quantitative evaluation of multiaxial non-proportional loading. However, how to accurately describe the effects of multiaxial non-proportional loading remains a key concern. Design/methodology/approach - This review introduces the influence of multiaxial non-proportional loading on the cyclic deformation and fatigue life of materials, followed by the development of life prediction models, which covers the evolution from the equivalent strain criteria, to the energy method and critical plane approach, and to the latest data-physics fusion-driven methods. Findings - From the perspective of material damage, a series of methods including equivalent strain, strain energy density, and critical plane method have been developed for accurately evaluating the multiaxial fatigue life of structures. With the development of data-driven algorithms, a series of physical-informed neural networks have also been developed based on these empirical models to obtain more accurate prediction results. In future research, studies that integrate physical mechanisms with data-driven methods can provide reliable results for multiaxial fatigue life prediction. Originality/value - This review introduces multiaxial fatigue life prediction models, including classical methods and data-driven methods. It provides a reference for further research into the theories of multiaxial fatigue life prediction.