Abstract
The phase field fracture method has quickly gained traction as a powerful numerical tool. Advanced cracking phenomena, such as crack branching, merging, initiation from arbitrary sites and complex crack trajectories, can be captured without convergence problems and on the original finite element mesh. We extend this success to hydrogen embrittlement [1]. A general chemo-mechanical framework is presented, suitable for any mechanistic interpretation upon an appropriate definition of the hydrogen-dependent fracture energy degradation law. We chose to particularise for the case of decohesion-driven embrittlement and ground the degradation law on atomistic calculations. In addition, strain gradient plasticity is used to provide a richer description of crack tip mechanics [2]. Model predictions reveal the critical role of plastic strain gradients in rationalising decohesion-based arguments and in capturing the transition to brittle fracture observed in hydrogen-rich environments.
The potential of the proposed phase field modelling framework is demonstrated by means of representative case studies. First, crack growth resistance curves are computed in a wide variety of scenarios, showing that the model appropriately captures the sensitivity to material strength, loading rate and hydrogen concentration. Secondly, model predictions are benchmarked against experiments on Ni-based alloys and ferritic and martensitic steels. Results reveal a promising agreement. Insight is also gained into the suitability of standardised experiments; slow strain rate testing (SSRT) is revisited, showing that subcritical crack growth compromises its validity. Finally, the capabilities of the modelling framework in enabling Virtual Testing are showcased. Large-scale multi-physics predictions are obtained for technologically-relevant applications. The phase field method easily enables capturing damage evolution from an initial distribution of defects, as measured from in-line inspection or other non-destructive techniques. More recent extensions, such as the consideration of multiple traps [3] or the modelling of fatigue damage [4] will also be discussed.