A stored energy criterion for fatigue crack nucleation in polycrystals

Abstract A series of microstructurally-differing, large-grained, notched, polycrystal BCC ferritic steel bend test samples have been analysed to extract the experimentally observed sites of fatigue crack nucleation together with the numbers of cycles to cause crack nucleation. The samples have been modelled with explicit representation of both grain morphologies and crystallographic orientations using crystal plasticity which has enabled a detailed assessment to be made of key microstructure-level quantities such as accumulated slip, slip rate, and densities of both statistically stored and geometrically necessary dislocations local to the experimentally observed sites of crack nucleation. These quantities when considered independently have not been found to correlate with experimentally observed cycles to nucleation. A new criterion for fatigue crack nucleation has been introduced in which a critical stored energy density, Gc, is argued to be necessary in order for crack nucleation. The rate of stored energy density determined at the sites of crack nucleation has been shown to correlate well with experimental measurement of cycles to nucleation, and the number of cycles to cause fatigue crack nucleation, for the samples for which such measurements are available, is well predicted. The criterion enables prediction of cycles to crack nucleation for all of the experimental samples and has been shown to demarcate correctly the crack nucleation lives observed over the range of differing experimental microstructures.