About 3 years ago I asked why the larger grains rose to the top in a coffe jar. Lynn Bates referred me to a paper which described the good old 'Brazil Nut Effect'. I suspect that you could gain some insight by reading the thread and article within these forums. Not all segregation obeys the accepted norms. ■

My below comments assumes the flow regime acts independent of electro-static and other particle attraction mechanisms.

Movement of dry granular material will always suffer from fines perculating through course particles, by gravity, as the flow stream moves. The only way to reduce this phenomenon is to minimize the lateral flow as material is transported toward the exit point. Among such improvements might be multiple exit points that attempt to withdrawal uniformally the general bin level (i.e. do not allow lateral transport of the surface as occurs with any normal single point exit).

Another way would be to produce one or more long slot discharge openings attempting the same as the above.

You can seen this effect in any upscale 3-D DEM code. Most fines perculation mechanisms can be simulated. You do not need to study the effect with ultra-fine particles to see the behavior. We are assisting the study of segregation behavior using our DEM code ROCKY. Some of the effects can be seen on our website in: a) normal chute flow producing segregated material onto the receiving belt, b) graunlar flow segregation charging into bins as well as withdrawal from bins, and c) related granular streams as noted below.

A case in point, is the segregation problem found in HPGR feed hoppers, and their undesirable effects on grinding, presently under study by a large mining company and their consultants. ■

My last comment assumes the DEM code can handle a >10:1 particle size distribution in order to replicate the fines perculation. In addition, because the small particle fraction will account for the major particle count (i.e. >1000 small particles can fit into one major particle diameter when considering spherical shapes). Some prudence in designing the flow field is necessary.

You will need a DEM code that can handle millions of particles in the simulation program. If you wish the problem to be solved in a reasonable time period, think about investing in a 32 core computer or send the problem to Conveyor Dynamics, Inc. (CDI). We support DEM modeling using ROCKY with our many 12, 32, and 40 core machines.

See the latest posting on Youtube and our website with recent project excersizes:

http://www.youtube.com/user/TheROCKYDEM

http://www.conveyor-dynamics.com/rocky.htm ■

The simple answer to your first query is that it is not possible to totally stop segregation from taking place when handling free flowing bulk material that are of variable constitution, but it is usually practical to reduce the effect considerably. It is a large subject with many ramifications, as outlined in my book – ‘User Guide to Segregation’, published be The British Materials Handling Board at £ 30.00. Some tips are also given in an article ‘Using Inserts to address solids flow problems’ in Chemical Engineer, July 2010, PP 32 – 37.

Most segregation tends to occur during the filling of a silo through a variety of flow mechanisms, depending on the filling procedure. The zone order of emptying then mainly determines how the fractions come out. Mass flow will not eliminate segregation, but will mitigate it during the discharge of contents in the body section of the silo after flow is fully mobilised, but a velocity gradient in the converging section will preferentially empty the central region in the final period of discharge. Tall, funnel flow silos will also provide a re-mixing process during outflow from the body section after the drained repose cone has been established, but initial and terminal segregation are more pronounced

Improvements can be made by changing the filling system and/or modifying the flow regime during discharge. A range of techniques are available, but they require some experience to optimise. If you care to send the silo drawing and operational details to lyn@ajax.co.uk, I will be pleased to advise. ■