Simulating creep of snow based on microstructure and the anisotropic deformation of ice

It is generally agreed that the creep of low-density snow is accompanied by drastic microstructural changes, which are a major source of macroscopic strain hardening. There is, however, no agreement about the dominating mechanism which mediates structural mobility at the microscale. A widely used model of creeping snow allows for intercrystalline deformations at the grain boundaries but neglects intracrystalline deformations in the grains. Here we show that the opposite scenario, which solely uses intra-crystalline deformations, while neglecting grain boundary sliding, is in better agreement with experiments. To this end we have conducted in situ, microtomography measurements of snow microstructure during creep experiments. 3-D tomography images are used to simplify the full microstructure to a 3-D beam-network. This reduces the number of degrees of freedom drastically, which enables us to carry out creep simulations by finite-element methods. We use Glen’s law for secondary creep of ice as the material model and account for the anisotropic creep behaviour of single crystals by assigning individual network strands a random orientation of the c-axis. The results suggest a separation of time scales between creep stress relaxations and slow microstructural changes make the key contribution to snow hardening. Although open-cell foam models clearly fail in predicting the observed viscosity–density relations, they are interestingly suggested as a potential limiting behaviour in our experiments.