How to Model Granular Flow

12/9/2008 by Machine Design Staff

Suppose a hopper is filled with a granular material such as sand, and you need to compute the force or stress on the hopper walls. If static, this problem is straightforward to solve using FE software because the nonmoving material can be treated as a solid. But attempting to simulate the unloading of the hopper would complicate things beyond the capability of FE or even of multiphysics programs due to the transient load changes.

In analyzing the hopper unloading, one task might be to find the velocity profile of the flow field. The movement of a fluid can be measured by sensors, and that of solids by strain gages, but particles are a different story. That’s because each particle is solid and so has properties such as cohesion. But in bulk, the material flows like a liquid.

Clipping plane placed in central axis of a hopper allows analysis of material flow in the central section of hopper

It is possible to examine particle flow with a camera. However, cameras give no useful data on why grains behave as they do. This makes it difficult to improve the designs of complex machinery, conveyors, chutes, buckets, and draglines for the bulk handling of materials such as rocks, pharmaceutical powders, and irregularly shaped pieces of scrap. In reality, designers of such equipment are often forced to make and test numerous physical prototypes.

Fortunately, powerful hardware such as four and eight-core computers are now available for under $10,000. This, along with EDEM software for what is called “discrete element modeling,” takes simulation of granular flows out of the hands of academia, where it has been for the past 30 years, and puts it squarely on the desktops of engineers.

“The big difference between discrete and finite-element or finite-volume methods is that particles are used for computing polynomials instead off meshes,” says John Favier, CEO of DEM Solutions, Lebanon, N.H., the developers of EDEM. “We only use a mesh to represent the surface of the geometry, while representative particles are modeled individually, which is where the term ‘discrete’ comes from. So less computing time is involved than with FE.”

For simulations, users import a CAD model of their machine into EDEM in IGES, STEP, or a native CAD file format, where it is converted into a surface mesh. Included in the program is a tool for designing irregularly shaped particles because real particles are often not perfect spheres. A CAD model of a particle can also be imported to allow accurate visualization of the material. In EDEM, users define particle properties such as material, size, and shape, as well as interaction properties such as friction. Additional interaction physics such as cohesion between contacting elements can also be included.

“The software database contains a range of standard materials,” says Favier. “It also includes properties to define materials such as coal or plastic pellets. Users with material know-how have an advantage in being able to perform useful simulations. For now, users tend to keep information to themselves, but in time, I think we will see a lot of the data published. In any case, the software’s default values should provide enough to get going.”

Say the imported CAD model was an excavation bucket, for example. Users would create a pile of representative particles through which the bucket will move. Users define the bucket geometry movement to recreate loading cycle as particles are scooped up from the pile. The software calculates this information at high resolutions of space and time, collecting data on where every particle has gone and on every contact made with the machinery. Under the hood, a lot of the computation is based on F = ma, so results include the change of force on the machine over time.

This knowledge is critical in applications such as wear analysis. Output includes animations and color-coded illustrations of the particles’ velocities that let users see regions of greatest machinery wear and where particles stick to walls. Users can then redesign components for a longer-lasting and more efficient machine.

Article edited by Leslie Gordon, Sr Editor, Machine Design

Article reprinted by permission of Penton Media, publisher of Machine Design