Numerical modelling of strain-localisation in compressed snow samples

Snow is a highly porous, rate-dependent, and strain-softening material whose mechanical behaviour is strongly influenced by sintering and microstructural evolution. Among the various deformation patterns it can exhibit, compaction bands, which are narrow zones of localised compressive deformation, play a critical role in processes such as the initiation of slab avalanches. This study focuses on modelling the onset and propagation of compaction band localisation in snow using an elasto-visco-plastic constitutive model developed by the authors within the framework of continuum mechanics. The model accounts for sintering effects through a time and deformation dependent internal variable, allowing for the simulation of strength degradation due to bond breakage. Numerical implementation was carried out in Abaqus/Standard via a user-defined material subroutine. The model was validated against laboratory experiments on confined compression on snow samples, reported in literature and reproducing key features such as band formation, downward propagation, and strain localisation. Parametric studies show that strain rate significantly influences the occurrence of localisation, with low rates leading to homogeneous deformation. Results demonstrate the model’s ability to capture the complex elasto-softening-hardening response of snow and highlight its potential for simulating real-world snow failure mechanisms, including those relevant to avalanche initiation. Although the model is able to reproduce the formation and propagation of compaction bands, some limitations remain. These concern not only the physical description of band reflection and the prediction of propagation velocity, but also numerical aspects related to the stability and robustness of the current implementation.