Solution to the problem of low sensitivity of crack closure models to material properties

The paper focuses on differences between crack closure obtained by numerical models and by experimental fatigue crack growth rates, namely for three different steels (bainitic steel, predominantly pearlitic steel and additively manufactured austenitic stainless steel). The experimental data revealed a load ratio effect different from that predicted by the most used plasticity-induced crack closure (PICC) models. The term "irreversibility of plastic deformation" was proposed in this work to be used as a material property to estimate how strong the PICC effect would be in a material. Two basic phenomena, which are usually omitted in other models, were considered in the explanation: (i) cyclic softening/hardening, (ii) brittle microcracking at the maximum load. The strip-yield model has the ability of conducting fast and accurate simulations under the variable-amplitude loading. It was demonstrated that the results of this model are practically independent of the material properties. This is caused by the consideration of only monotonic elastic-plastic material properties. To simulate real load ratio effects, the parameter beta in the strip-yield model (constraint factor in compression) is proposed to be used as a variable. It enabled a generation of different ratios of monotonic and cyclic plastic zones, which in turn helped to reproduce the crack closure values observed experimentally. Discussion and explanations were provided regarding the material microstructure. The proposed approach considers unequal tensile and compressive yield stresses caused by the different irreversibility of plastic deformation, which explains the dissimilarities in the load ratio effect observed in the investigated materials. It can also improve the accuracy of residual fatigue life estimations.