Self-diffusion in dense granular shear flows

TL;DR

This study experimentally investigates diffusivity in dense granular shear flows, revealing anisotropic diffusion influenced by force networks, shear rate, and boundary effects, with simulations confirming key diffusive behaviors.

Contribution

The paper provides the first detailed experimental analysis of anisotropic diffusivity in dense granular shear flows, highlighting the role of force networks and shear rate effects.

Findings

Diffusivity is proportional to local shear rate, D ≈ γ̇a².

Diffusivity is higher along the mean flow direction, approximately twice as large as perpendicular.

Force networks suppress diffusivity along their direction, creating anisotropy.

Abstract

Diffusivity is a key quantity in describing velocity fluctuations in granular materials. These fluctuations are the basis of many thermodynamic and hydrodynamic models which aim to provide a statistical description of granular systems. We present experimental results on diffusivity in dense, granular shear in a 2D Couette geometry. We find that self-diffusivities are proportional to the local shear rate with diffusivities along the mean flow approximately twice as large as those in the perpendicular direction. The magnitude of the diffusivity is D \approx \dot\gamma a^2 where a is the particle radius. However, the gradient in shear rate, coupling to the mean flow, and drag at the moving boundary lead to particle displacements that can appear sub- or super-diffusive. In particular, diffusion appears superdiffusive along the mean flow direction due to Taylor dispersion effects and subdiffusive along the perpendicular direction due to the gradient in shear rate. The anisotropic force network leads to an additional anisotropy in the diffusivity that is a property of dense systems with no obvious analog in rapid flows. Specifically, the diffusivity is supressed along the direction of the strong force network. A simple random walk simulation reproduces the key features of the data, such as the apparent superdiffusive and subdiffusive behavior arising from the mean flow, confirming the underlying diffusive motion. The additional anisotropy is not observed in the simulation since the strong force network is not included. Examples of correlated motion, such as transient vortices, and Levy flights are also observed. Although correlated motion creates velocity fields qualitatively different from Brownian motion and can introduce non-diffusive effects, on average the system appears simply diffusive.