An upscaling approach for micromechanics based fatigue: from RVEs to specimens and component life prediction

A novel approach has been developed to estimate fatigue life at the specimen/component level from the simulation of relatively small Representative Volume Elements (RVE) of the polycrystalline microstructure. This technique estimates the statistical distribution of fatigue lives under general multiaxial loading conditions accounting for both microstructure and specimen size. The model relies on computational homogeneization where the crystal behavior follows a crystal plasticity model and simulations are performed using a FFT based solver. The simulation of the cyclic response of a set of different RVEs provides the statistical distribution of Fatigue Indicator Parameters for that RVE size, which can be accurately represented by an extreme value distribution as the Gumbel distribution. This distribution is upscaled to the actual size of the specimen or component of interest using a weakest link approach and is finally transformed into a distribution of fatigue lives using a simple fatigue-life expression fitted with experiments. The framework proposed estimates fatigue lives of specimens with millions of grains from the results obtained with RVEs containing only hundreds of grains and is able to reproduce the specimen size effect on fatigue life. The approach is first numerically validated and then used to predict the statistical distribution of fatigue life of a polycrystalline Ni-based superalloy showing very accurate results.