Principal Investigator Martin Bazant
Image: experimental photograph of an initially flat interface between green and yellow balls draining toward a small hole, between two glass plates.
In the Dry Fluids Group, we are developing new mathematical models and simulation techniques for particle dynamics in slow granular flows, e.g. during drainage from a silo, in coordination with real particle-tracking experiments. These non-equilibrium statistical models will describe flows at more coarse-grained level than molecular dynamics, but at a much finer scale that continuum models (which only describe the mean velocity). In some sense, the goal is to develop an analog of the random-walk theory of Brownian motion in gases for the very different case of granular drainage, which requires taking into account the cooperative nature of particle dynamics in dense random packings. A related goal is to develop new accelerated multiscale algorithms for simulations of realistic large-scale granular flows by combining the statistical model with occasional, brief, and localized relaxation taking into account inter-particle forces. We hope that such algorithms may also be applied to other amorphous materials, such as supercooled liquid and glasses.
We have created the MIT Dry Fluids Laboratory to do accurate particle-tracking experiments using a high-speed digital video camera and computer image processing. Our experimental data unambiguously rejects existing statisticals model of drainage, while strongly supporting a new model we proposed (before the experiments), based on the idea that diffusing ``spots'' of free volume cause cooperative particle diffusion.
We have been doing simulations with the Spot Model to better understand its strengths and weaknesses: In our experiments, we have observed the spatial velocity-velocity correlations and mean velocity profiles predicted by the model, but the model typically leads to unphysical density fluctuations. We are also collaborating with Sandia National Lab on molecular dynamics simulations of granular flow as another source of "experimental data" for testing theoretical ideas. The simulations are performed in the Applied Mathematics Computational Laboratory.
In addition to advancing the basic science of granular materials, an equally important objective is to provide mathematical models and simulation tools for the engineering of pebble-bed nuclear reactors. We are working closely with MIT nuclear engineers on the Modular Pebble-Bed Reactor.