Numerical simulation tools are used in snow science to predict avalanche runout distances. The results are applied to plan and develop avalanche mitigation strategies to protect mountain communities. Fluid dynamics equations describing mass and momentum conservation of the flowing snow are the basic physical elements of the numerical models. Advanced models now include additional balance equations describing the production and decay of kinetic energy associated with random granule motion. Because the avalanche is running on a hard surface (the ground) and contains a free upper surface (the flow height), shearing at the base or within the flow necessarily implies the production of random fluctuation energy and the expansion of the flow volume. The expansion is often rapid, leading to explosive eruptions of ice-dust plumes at the avalanche front. In this presentation we present a numerical model of this process, explaining the mechanics of the volume expansion and the role of random fluctuation energy. The process is mediated by the granular properties of the snow clods, including granule size, temperature and moisture content. Of particular importance is the material cohesion. Cohesion not only modifies the constitutive relation for shearing, but also mediates the energy fluxes associated with the fluctuation energy. To demonstrate our modelling approach we present an example simulation of a real snow avalanch in general three-dimensional terrain.