Abstract
The prediction of hydrogen assisted cracking in metallic components benefits from numerical modelling of the coupled physical phenomena involved in embrittlement mechanisms. For medium-strength steels, dimpled features are still observed despite some macroscopic embrittlement in hydrogen environments. The nucleation, growth and coalescence of voids is influenced by hydrogen concentration, and the Hydrogen-Enhanced Localised Plasticity mechanism (HELP) is expected to operate. Within this context, the present work presents a hydrogen-informed GTN model (Gurson-Tvergaard-Needleman) where hydrogen redistribution is simultaneously solved to account for transient effects. The coupled framework implements: (i) hydrogen transport model considering trapping and stress-drifted diffusion; (ii) hydrogen-enhanced void nucleation considering decohesion mechanisms and trapping sites; (iii) hydrogen-modified void growth from unit cell simulations; (iv) realistic coalescence criterion based on Thomason’s condition; (v) non-local modification of the GTN model. The model is implemented through user subroutines in the commercial Finite Element software Abaqus, that are validated through single-element simulations. Numerical schemes are discussed and the implicit formulation used in previous works is adapted to Abaqus Explicit to avoid convergence issues. Additionally, notched tensile specimens with different stress concentration factors are simulated under different hydrogen concentrations and strain rates. Results show that the coupled model can reproduce macroscopic hydrogen embrittlement, i.e. failures at lower elongations, by simulating the effects of hydrogen on ductile features. Furthermore, this transient model is capable of predicting the stronger embrittlement effect at low strain rates. It is demonstrated also that non-local modification circumvents the mesh dependence of classical GTN formulations by incorporating a length scale. The physical meaning of this length scale and its relationship with diffusion distances are analysed. Additionally, the limitations of the present damage modelling approach in contrast to brittle failure models, e.g. Phase Field or Cohesive Zone Models, is also discussed in terms of hydrogen embrittlement micro-mechanisms.