There is a lot of hype going around on the recovery of fines via gravity. To be sure it is an important target, and yes we can perform it, but make no mistake, it is only the one end of the spectrum.
Fines recovery is almost invariably associated with tailings, the values are there because they have resisted processing. But why are they there in the first place? Like it or lump it, processors have been unwittingly creating difficulty for themselves for a very long time. You see, minerals that are gravity targets have higher SG’s than their hosts, and in the majority of cases they are more brittle than their hosts too, so why is anyone surprised that they turn up in the finer classes in a comminution system? And if that system employed closed circuit classification that returned the dense as well as the coarse particles to the mill, then those particles turn up several size classes down. In processing ‘fine’ and ‘coarse’ are subjective terms, so let me define them in context of this article:
Tailings stockpiles are historical, and of course we have put energy into developing processes to deal with the fines, but in this GOIA I want to spend some time at the other end of the spectrum and see if we can escape this unfortunate problem in the future.
A very large number of the minerals we target occur in massive or large crystalline form, various sulphides, malachite and wolframite to name a few. So at very coarse size, even up to 2mm, there will be a high proportion of liberated grains present. Provided these are below 2mm in size the Knelson CVD concentrator is able to capture them with relative ease. In fact, even a bucket would do a fair job. If, however, we willfully destroy all those easy targets and send them down to the bottom end of the spectrum, then the task becomes progressively more difficult as the gold curve below attests:
The reason for the difficulty is not hard to understand when we consider the very large number of particles that are generated at progressively finer size. Complete separation requires that all the targets get to the concentrate zone whilst all the gangue is ejected and to do this the targets must progress through the gangue and find their refuge before the slurry mass exits the concentrator. At the enhanced G forces employed for fine separation the time frame is vastly compressed, magnifying the difficulty of the sorting task. With these huge numbers of particles present the probability of random displacement and path collisions adds to the difficulty of the task.
It is therefore not difficult to understand the logic in making a separation at as coarse a size as possible within the mineralogical constraints, and in practical terms the Knelson CVD can handle anything below 2 millimeters. Of course, not all the target may be wholly liberated, but remember that we can target the semi’s too, provided there is sufficient SG differential. The pictorial below illustrates the case with a sulphide at SG 4.6 in a gangue of SG 2.7. Of course it would be dependent upon several factors, but we have attained separations with an SG differential of 0.5, so particles of 25 to 30% by volume of target are fair game. Obviously the purity may then be impaired, but comminution of a small mass is far more manageable, controllable and efficient than it would have been with all that gangue coming along for a free ride.
Ah yes, I can hear the chorus – “what if our target is finely disseminated?” Well, if it’s a fine dissemination throughout the host gangue, then obviously this isn’t for you. But if the dissemination is in a dense species, for example gold in sulphides, then we would target the sulphides and then subject them to a secondary process – Two Stage Processing. With the advent of micro grinding the scope for a whole gravity route may become feasible and we have several plants in Africa which do just this, but with the pre-concentration at this stage via flotation.
Primary comminution becomes very important if we are going to consider coarse separations, the concept of in-circuit screening is becoming really topical, especially in the South American float territory. And whereas conventional hydrocyclone classified circuits aimed for a grind, a screen classified circuit aims for a liberation size.
You may have noticed the careful reference to target particles rather than concentrate and the avoidance of the terms grade and recovery. The bottom line, my fellow processors, is that those very restrictive terms are a mere consequence of what this game is all about – particulate separation!
