Nucleation, growth and coalescence of multiple cavities at a grain-boundary (Q704342)

This paper is devoted to presentation and verification of a cavity element which directly models the cavity nucleation, growth and coalescence on a grain boundary. A material model of the grain boundary with cavities is constructed, using the classical finite-element scheme. A variational principle, being the basis of the finite element model, is obtained in consideration of total free energy and matter conservation law. The variational principle leads to a set of linear simultaneous equations from which all six degrees of freedom for each cavity element can be solved. Unlike the smeared-out model, individual cavities and their full interactions are taken into account in the cavity element. The cavities can exchange matter with each other, leading to cavity ripening. Each cavity can shift along the grain boundary as well as grow or shrink. The cavity nucleation and coalescence are included into the model by remeshing the grain boundary at a nucleation or coalescence event. It is proposed that the cavities can be nucleated or coalesced anywhere on a grain boundary, leading to a non-uniform distribution of cavities with various sizes. A series of cavity elements are assembled together to construct a grain boundary. The cavity element model, considering the full interaction between the cavities, includes surface and grain boundary diffusions. The model excludes power-law creep as a contributing mechanism for cavity growth, limiting to materials which are subjected to low stresses (less than \(10^{-3} G\), where \(G\) is the shear modulus). The cavity element is first verified using a series of cases for which either the solution, or the trend of the solution is known. It is then compared to the smeared-out model under the condition that solid state diffusion is the only rate controlling mechanism. It is shown that the two models only agree with each other either at unrealistically high level of stress or extremely low rates of nucleation. Finally, it is investigated how the relative diffusivity affects the damage evolution, separation velocity and overall rupture time, using the finite-element model. This effect is only significant at a relatively high level of applied stress.