This is a great idea and makes sense on so many levels. It is good to see it out there to get people talking. It is something that Gekko Systems has been preaching about and practicing for some time. Gravity pre-concentration for gold ores has been done by Gekko and has shown the benefits described exactly. Gekko has also done work on gravity pre-concentration for other minerals, which have shown promise and stand up to economic scrutiny.

Pre-concentration is the best if we have liberated gangue particles at that size range and if pre- concentration does not lead to loosing too many values as gangue at that stage.

There is no reason that pre-concentration shouldn't be adopted by all sites that have amenable ore types. Coarse liberating gold deposits, lead and zinc ores, tin etc.. Driving the effects of pre-concentration back further to the mining model and potential tailings discharge restrictions, coarse pre-concentration can have a significant effect on the NPV of a project.

Pre-concentration is as everyone agrees one of the best ways to minimise energy in milling and environmental unfriendly waste reduction where the waste can be dumped back in the pit instead of creating slimes ponds. There is much better technologies now available where pre-concentration can happen at even much larger particle sizes even if there is no liberation of the concentrate against the waste. 

In a gravity circuit, the total density of the particle is used for the separation, which creates problems if you do not have a perfect liberated sample. If a small piece of Gold or Tungsten for instance is stuck in a host rock sample it in most cases does not alter the total particle density enough to ensure that the particle will report to the concentrate fraction thus it will report to waste. 

STEINERT uses an X-Ray transmission process that scans and calculate the atomic density of every pixel in every rock thus it will still see a small piece of the target mineral in a bigger rock even if it is not liberated. This complete rock is then separated as product resulting in even lower losses of the target mineral. 

The advantages is that pre-concentration can happen at particle sizes of up to 90mm and normally not lower than 10-15mm due to loss in processing rate of these small particles. This is where the liberation problems start to become less of an effect for technology like the Gekko IPJ's for instance as Andrew mentioned above. 

In our latest test-work conducted on a Gold mine in Australia it showed that we could recover over 75% of the Gold in less than 8% of the total mass in particles ranging in size between 15mm-90mm. With figures like this, it will be crazy not looking at pre-concentration of these coarse size fractions and the feed to the current milling and flotation circuit is upgraded by 10 times (from 10g/t to 100 g/t).

It will not be long before most mines will adopt the process of coarse pre-concentration into their systems, they somehow are still reluctant to change but the economics and grade will speak for itself. With the lower commodity prices and pressure in the mining sector in general, every mine has to adopt to the best practices to reduce their OPEX and perform at optimal efficiencies or they will be out of business when the markets drop again or even lower currently.

If anyone would like to have more information on our processes or technology, just drop me a e-mail. We currently do all our stage-1 test-work for free to determine if we can sort your specific ore or not.

AAP in SA has just started commissioning a "proof of concept" trial rock sorting plant at a nominal 75-100 tph using XRF sorting with feed size range of 30-80mm; prepared by a contractor run screening/crushing plant. The plant is a proof of concept for potential ROM and low grade stockpile processing applications. The initial application will use measurement of copper and nickel to sort waste from value. typical target is to remove 15-25% of the ROM ore feed as a waste stream.

I have been think about your comments about 2 years ago about selecting "optimum" grinding media for your IsaMills. That not necessarily the most abrasion resistant nor the densest are the most suitable. The pure fully hard, high densityalumina beads working less well than (if I recall) zirconia/alumina. Can the benefit of mixed composition media have an association with the coefficient of friction of the grinding media against the milling product?
Old junior operators (me at UPM in 1974) had only two basic milling rules: " the smaller the balls the finer the grind" and "the load has to be thick (viscous) enough to stick between the balls".
This was recognized and is intuitive. A polished steel ball, with relatively low friction to the pulp load, will tend to let surrounding slurry disengage from its surface. If grinding media is less inclined to let slurry "slide" off it does it make it a "better" - more grinding per MJ -, medium?
I am asking you your thoughts

I guess my first comment, primarily we look for the lowest cost of ownership for size of ceramic media, TCO. This is determined in our small scale testing protocol, initially on a four litre Isamill, 1 hour, 2 hour and 8 hour testing, then if passing that hurdle comparatively against our database, proceed to 100 litre Isamill, 100 hour continuous testing on slurry. Thereafter depending on relative TCO, a commercial plant test may be scheduled. Can't have too many of those as it's potential disruptive to operations. A full scale test is necessarily quite long in duration, to ensure the previous load is not effecting the data.
One point on media size selection I can make is the relationship of averaged particle size, critical size build up and optimal media size, necessarily we are trying to grind the coarse size fractions in the feed slurry, so it's important that the selection bears this in mind, a small time interval of coarse anomalous feed during the sampling period must be considered. We have found that in practice.
With respect to rough media surface and it's impact, internal mill component wear is a factor to consider.

Further, agree that the golden rules, you mention still apply, but with the critical size caveat, a build up will negate the efficiency gain of smaller media, we have seen this in full scale results. In addition it's viscosity that's the key rather the slurry density. With respect to specific gravity of media, the machine you employ will have direct bearing, drive size and ceramic loading are related to sg; we are using sg 3.8 for Isamilling, sg of ZTA. This also works for SMDs. Best current TCO, note also production losses due to upsets works against using high cost zirconia, high specific gravity medias, although they give great wear results in testing.

Hmmm, we have drifted off the pre-con discussion back into fine comminution energy minimisation.

Fragmentation optimisation, both PSD and MJ, is nevertheless a most important part of the pre-concentration process. It is essential that as-mined ore which may be amenable to a pre-con route must be broken to liberate gangue but to minimise production of material that is uneconomic to put through the selected pre-concentration process.
Fragmentation as a feed preparation for pre-concentration.
If an individual particle sorting route is suitable then conventional compression crushers may be acceptable as such sorters have an economic bottom size 'cut-off' of (perhaps) 10 mm and crushing to (say) 25 or 30 mm may produce only a reasonable proportion of < 10 mm.
If we have to consider magnetic or densimetric preconcentration (as has been the focus in the past) then fragmentation to 100% < 8 mm might be a target. To produce such fragmentation without producing a lot of < 0.5 mm is a much more difficult problem.

For information, can some one give some case studies where pre concentration,under what conditions was adopted. One has to see the cost of pre concentration vis a vis the economic gain got by pre -concentration.I am sure one should explore the possibility but if the feed grades themselves are low, it is obvious the one has to be crush and grind for good liberation.Whether it is pre or post depends on ore and liberation.

I did a study on pre-concentration on one of Amplats' operations and I believe the plant was built by Batemans. Mt Isa Pb/Zn is probably a most notable base metal pre-con. Somebody, somewhere will give you whatever detail is not confidential about Amplats' pre-con operations. Other Merenski operations also do DM pre-concentration.
A couple of WA base metal operators have used DM pre-con recently for some of their ores but the route was not economic for all their resources. We have done many studies. An early extraordinary example was the DMC separation of hi-uranium Witwatersrand ore from ROM. This route could produce an adequately U-barren lowe density fraction that was ground and cyanided.
The "heavies" were ground and oxidising acid leached for U then neutralised and CN for gold - a "conventional" reverse leach.
I think 3 such plants were constructed.
I will have to do a bit of digging for more examples actually constructed.
Nepean Nickel - long-since closed down.
Can we call the densimetric upgrading of chromitite and chromite ores before smelting preconcentration? If so a lot of Zim and RSA chrome resources include DM or spiral pre-concentration.

Preconcentration or ore upgrading is common practice for the UG2 ROM stream, dense media separation is used, it was first done on development ore at Union mine in the nineties and is currently employed by AQP at its Kroondal operations. Crushed and screened ore from LP mechanized mining is upgraded by the removal of approximately 1/3 of the tonnage at a very low reject grade. Ore upgrading using rock sorting is being trials by AAP, sorting by XRF, rejecting waste in a 30-80mm stream using Cu, Ni and Cr values.

The study I did was for Ivan Plant after it was converted to UG2. I then had a long assignment in Indonesia and when I got back Batemans had been awarded the Union Plats pre-conc job.
We should also note that hand sorting was a most effective pre-concentration (if you have enough labour). Many of the elderly operations I did audits on had quite sophisticated hand sorting facilities.How good they were is debatable as the "sorted waste" dumps changed their names to "strategic reserve stockpiles" in the early '90s.
I have recently done some work on XR transmission lump IO sorting the quality of separation (translated into partition curve was not as good as we could achieve by DM.
The XRF route (as you are testing) sounds a much more specific approach. I wonder if you will be able to publish any results.
Inevitably the pgm industry has much greater research budget than IO so you are probably where the inovation is going to come from.

To me, pre-concentration is all about the definition of ‘coarse’ and when we can reject gangue without incurring excessive metal losses.

What constitutes coarse is largely ore texture-dependent (i.e. grain size/clustering on a micro-scale, and the ability to use ore zones or a characteristic host mineral to discriminate on a macro-scale). For a given mineral texture, the practical feasibility for a given beneficiation technology is a strong function of the largest average particle size at which the mineral property of interest in ore-bearing and gangue-only particles are sufficiently different to confidently allow discrimination of one from the other. For example, at what particle size is the likelihood sufficient that a sulphide mineral is exposed enough to allow discrimination by flotation? Similarly, at which point are gangue and ore-bearing particles likely to be sufficiently different in density to allow gravity/DMS concentration?

I would say the key point here is defining ‘coarse’ for a given pre-concentration technique and ore, and realizing that we do not need >90% or so liberation before we can concentrate. However, what shouldn't be overlooked is that a consistent procedure is needed to test different options, and that it is important to convince managers and operators to take the risk to move away from (possibly) more accepted beneficiation methods.

Silver Standard's Mina Pirquitas in northern Argentina, mines and processes an ore body that is part of the Bolivian silver-zinc metallogenic belt. 

The flowsheet is three stage crush to -9mm, Gekko IPJ plant, the ball milling to silver and zinc floation.
The Gekko IPJ plant consists of wet screneing at 2mm, ahead of three lines of Rougher+Scav IPJs.
The -2mm passess directly to grinding, the -9+2mm is jigged. Jig tails assay similar to final flotation tailings, mass about 30% of jig plant feed.
There is a technical paper on Gekko's website.

Liberation size is what concerns most people in pre-concentration. This is true to any gravity concentration (be it a Jig, DMS, Spirals etc..), but the use of a XRT unit actually overcomes the problems in a unique way. Where Density separation takes the "total" density of a particular rock and then places it somewhere on a separation graph for separation, a XRT unit looks at a rock in a different way. The XRT can "see" a small particle with a higher density in a larger rock with a low density and then we can select it as a concentrate even on this small speck's properties. This is totally impossible for gravity concentration as the rock's total density is still alter too little to place it on a different area on the density separation curve.
So to answer the size question for pre-concentration, it is only limited to liberation in Density separation, but STEINERT have successfully pre-concentrated numerous large particle sizes of up to 90mm where the 80% liberation of the ore is at a much lower size of -6mm.

Here is a link to one project where we have applied the sorter in waste rejection in the crushing circuit. The particle sizes treated here is -40mm +15mm, and we have also included coarse fraction processing being -90mm +40mm on this material.

http://www.steinertglobal.com/au/en/products/sensor-sorting/steinert-xss-x-ray-sorting-system/

Is there a visible difference between mineral and gangue? If so you can probably do something
Unfortunately there is no universal rule of thumb. The mineralised component of ROM must have a distinguishable and instantaneously measurable characteristic difference to the gangue component.
The “gangue” can be “over-break” or mining dilution or may be low grade material within the ore zone, e.g. chert/quartz banding in a macro or mezzo BIF.
Gideon van Wyk has described what the individual particle sorting pre-con suppliers can offer – colour, texture, XRT, fluorescence and other such “diagnostics”.
Somebody described the work developing by performing XRF for Cu/Ni on individual particles. This must be considered the ultimate in direct characteristic recognition, actually identifying the metals that you want (or are part of what you want).
Most of us have to make do with a characteristic that is more easily used and only has an indirect link with the values to be recovered – the classic being density. Metalloid minerals and “valuable” metal oxide mineral tend to have higher density than siliceous gangues.
Those who have used IPJ as the densimetric sorter to reject coarse gangue before milling and flot. If the density differentials are sufficient then jigging is an elegant route. This is not often the case.
High precision densimetric separation can be used to perform separations that would result in unacceptable misplacement of values in reject or rejects in product if attempted with any design of jig.
Only an amount of testwork will evaluate if a preconcentration can be the economic process route. The individual particle sorter industry will do a preliminary evaluation on a fairly small sample – followed by a near-pilot scale program.
The densimetric routes require size and pyknometric characterisation determinations on the ore to be evaluated. Modelling of the density/grade characteristic (for jig or DM process) can give an accurate prediction of performance.

Correct!  All minerals are different and test-work is the only way to clearly give a indication if any process will work or not. As for the quick check, we normally only require to scan about 20-40 rocks using various separation techniques and from this we can gain a reasonable accurate recovery of the target minerals. On material where density is used for separation, we normally get a complete range of particles and number, weigh, scan, calculate density and photograph each rock in the range. We can then group certain densities together as concentrate and waste, and calculate grade and recovery (this process is mainly used on Iron ore and coal).
All our first stage test-work is free, and we only charge for bulk samples or pilot campaigns, to recover actual cost incurred.
We also have mobile plants available that we can install on a clients site for proof of concept and these plants are totally self sustained.

I have organised work by you and your competitors in the past. I am very aware of your techniques.

We do IPP on lumps up to 150 mm and down to 6 mm and fraction-ate into as many as 38 very accurate density fractions. For large lumps it takes a very large mass of ore. For 6 to 8 mm only 1 few hundred grams. It is sufficient mass to get a statistically valid number of particles.

Sub-6 mm we get work done by others by HLS using Cerichi sol'n. UGH!

Some basic information like method of mining(where large proportion of overburden(waste material with no value) is mixed with ore/ and the grade of R.O.M will give a "thumb of rule" indication whether we can even consider sorter. 

I think we should talk the details to the operators who want to pursue it 

The non-ferrous metals mining industry has to delete ROM ore dilution from its operations. Why hasn't it? The key lies, as always, in the economics, which are missing in Adrian's slides. I also believe the degree of upgrading suggested is too optimistic. A rule of thumb that 3 tons of ROM ore at grade X could yield 1 ton of pre-concentrated mill feed at a grade of 2X is much more realistic, using sensor based machine sorting test results.

Let's take the example of ROM copper ore at 0.2% Cu, that's 4 lbs Cu metal per ton. What's the value of copper in the ground? Let's be generous and assume $3/lb for this calculation, with no losses in processing. So we have $12/ton to work with, which rules out U/G mining. Open pit ROM ore could (still) be presented to a nearby Mill for $2/t, Mill cost could be $6/ton, leaves $4/ton ROM ore containing 4lbs Cu for refining. That's $1/ lb copper. The reader is welcome to put in his /.her own numbers, but investors better hope there is some gold or moly with the copper for profits, or that the copper is there as leachable oxide.

Can this scenario be saved by pre-concentration of the Mill feed, assuming sulphide ROM ore? Let's assume it is indeed possible to get 90% of the copper in 40% of the mass (Mill feed). That's 3.6 lbs of copper in 0.4 ton Pre-Concentrated Mill Feed. (P-CMF). Costs: $2 for mining, plus $1.20 for disposing of the P-CMF reject as a separate stream, $2.40 for milling (smaller Mill, relative higher costs?) = $5.60 for 3.6 lbs of copper. Add copper refining cost at $3.60/lb Cu gives $9.20 total, a saving of 0.80/ ton ROM to do the pre-concentration. 

Where can one buy a ROM pre-concentrator to do this job for $.80/t ROM?

Bulk sorting can be done in a number of ways - one example is screening by size as in the presentation. Another is diverting increments based on real time through belt conveyed grade measurement (say on a 1-2 minute increment). Where grades are over 0.05-0.1% it is a simple and inexpensive approach (<0.15/t on primary crushed ROM material) that can be used for upgrading mill feed quality in base and ferrous metals applications. It has been used for many years for diverting product quality material to bypass beneficiation in coal and iron ore, but is equally suited to measuring base metal grades as evidenced at Sepon Cu and Mt Isa Zn-Pb. Diverted material can be upgraded by screening, particle sorting, HMS, etc, if economic to do so. A combination of processes and technologies is likely to give the optimum result. Real time grade measurement appears to be an integral part of the solution to provide an improved level of control.

As a Mineral Engineer I agree pre-concentration has to be explored before an ore is subjected to processing. But how much we can eliminate and at stage we should do this and cost benefit analyses have to be noted before we decide the circuit.

I do feel that one has to investigate whether an ore has the chatacteristics for pre-concentration depending on liberation phenomena ; many times, pre concentration, if the quantity one can reject is small and in addition if some values are lost during pre-concentration, one has to evaluate carefully in incorporating a circuit for pre- concentration

I'm not aware of sulphide/gold/pgm ores that have allowed 90% of the contained value(s) to be concentrated in 40% of the mass by screening. If this occurred with any frequency, it would be standard procedure to check for that processing option in the mine development stage. 

I have the same comment with regard to belt conveyed grade measurement. I cannot conceive that a veined sulphide ore body can be blasted, scooped, trucked, dumped on a belt and then still allow to be diverted in 1-2 minute increments while yielding 90% of the valuables in 40% of the mass. The industry has not even proven that their ore blocks are characterized to that level of accuracy. 

I am currently trying to get industry support for that by redrilling an already defined (by one drillhole) sulphide ore block (prior to mining, obviously). Redrill the block by three new holes and then cut the new core in, say 2" lenghts, each piece to be analyzed by XRF for valuables. This will give approx. 1000 measurements, to yield a much more precise value for the entire block. Want to take bets on how well it matches the current one-hole value? The results will also allow a more dependable insight in how well this block could be pre-concentrated. Success to date: Nil. The industry is not interested in finding inconvenient truths; the bankers would go nuts if the certainty of grades and tonnages would become questionable. 

I agree that a lot of work is needed to be able to achieve major improvements in pre-concentration - no argument. Technologies and processes that currently exist to make incremental improvements should also be considered. Measuring in usable increments (smaller than a haul truck) can be done relatively cheaply and without major process changes. Truckloads of "waste" can be diverted and that saving extends all the way to reduced tailings generation. Diverted waste is replaced with ore, so average feed grade to the mill increases. It is an incremental form of pre-concentration.

Maybe geometallurgical domaining should include a parameter that indicates pre-concentration factor, a function of in-block grade variability. Ore blocks can be flagged for separate processing paths that include pre-concentration. Within a mine it should be possible to distinguish lower grades in blocks due to dilution (higher pre-concentration amenability), from lower grades due to more sparsely disseminated mineralisation (lower). There may be no need to process all the ore through a pre-concentration stage. Orebody knowledge is critical. Many sites don't understand it enough to be able to consider these evaluations.

I find it absolutely fascinating that geo-metallurgical modelling is once again being touted as the great savior. What about knowing your ore body, noting how the plant performs when the ore is coming from a particular area. The idea of preconcentration is as old as the hills and any metallurgist worth his salt investigates the potential for preconcentration early in the test program, nothing new here except some of the machines available to do the preconcentration.

What is being suggested is exactly along the lines of your second comment. Isn't geometallurgy the characterization of ore based on its processing performance (determined from metallurgical testing)? I thought geometallurgy was all about knowing your orebody and using known performance to ensure the metallurgical characteristics were considered in mine planning rather than just tonnes and grade.

The one point raised in the presentation that I found most interesting was the admission that as we have moved from selective mining to "bulk low cost mining" we now need new devices to remove the dilution that our new mining method ensnares! - it is a matter of costs but perhaps not the owners are looking for that bigger is cheaper.

The presentation's data would be more applicable to precious metals than copper and other base metals. The gangue vein between two copper ore veins is seldom as big as the copper vein, if it was would we call the total mineral block "Ore"?

The biggest concern I have from the presentation is that more feed to the mill means less selectivity, and thus falling recovery. To me this means more sulphides to tailings and thus MORE ACID MINE DRAINAGE.

I would commend my employer BBA for being VERY concerned about maximising recovery, and thus reducing AMD. The message seems to be select your process design engineer carefully if you want to maximise recovery, minimise extraction cost (including mine costs) and minimise "rehabilitation or disposal costs"!!

Don't accept the mass produced "standard circuit" but hire an engineer who knows the problems. Yet as we all know, retirement is reducing the qualified experts!.

I like your last comment-"hire a qualified expert". Most of the circuit selections are by word of mouth/as spoken by equipment manufacturer/or they are doing that and so it should work for me/lowest cost). My intention is not to let down any one; what happened has happened; no regrets-----can we turn the page and write a new one.
I have seen, in many times,the process expert plays a very small role while others take the final call.
Ores are site specific and have different characteristics and let process man decide the simplest process flow sheet please.

Pre-concentration is not a new concept. It has huge potential to be utilized within green field flow sheets, however I believe to many metallurgist place too much focus on 'absolute' recovery and not 'economic' recovery. Not all ore types will be amenable, however the metallurgical test work needs to be performed to look for the coarsest liberation size of the valuable mineral. Once this is determined there are a number of tools that can be utilized including ore-sorting, screening, gravity, etc. Silver Standards Pirquitas mine is a good example where continuous gravity and screening was utilized on a -10mm crushed ore to recover +95% of the metal into ~66% of the mass, So basically ~33% of the mass was rejected directly to tail prior to the milling circuit. The benefits are not only evident in the downstream processing plant ( capex, grades, recovery), but this data can be used to optimize mining models by dropping cut off grades, accepting dilution and allowing lower grades to be mined. The potential is there we just need to ensure that it is considered an option in the trade off studies.

There are a few definitions of 'liberation size' which can be applied to pre-concentration: Coarsest size at which waste can be rejected at an acceptable loss of valuable mineral; coarsest liberation size at which valuable mineral can be recovered to a concentrate without excessive size reduction; coarsest grind size at which an acceptable concentrate grade can be produced. Once the ore is on the ROM pad, the key seems to be staged size reduction with upgrade between each stage, but this is likely to require more equipment with higher capital and operating costs.

I like to thank all for their comments. Understanding the geo-metallurgy of one's orebody is as important as the economics, seems to be the consensus. Geo-metallurgy is site specific, Not much I can do about getting exploration geologists to think as process metallurgists, but the economics of (machine) sorting are universal and can be studied off site.

As I tried to point out in my first comment, sorting sulphide ores by machine does not yield enough economic benefits in further downstream processing. The earlier work I did for Barrick Gold was (originally) aimed at removing carbonaceous material by selective heating with microwaves (and then using IR sensors and the standard air-blast-off-the-belt technique). That work was aimed at improving the gold recovery of preg robbing ores, a substantial economic benefit. However, I was never able to get selectivity against sulphides, which also heat up in a microwave field. I demonstrated that, depending on the ore, good selectivity for sulphide ore (rocks) was demonstrable. The economics of that process route were largely negative; R&D was stopped.

Two years ago I was asked me for a suggestion for a project on skill demonstration in electro-mechanical engineering. I suggested a sulphide ore sorter, based on MW heating, separating the hot sulphide rocks from the cold gangue, not by IR sensing and air blast technology, but by discharging the rocks on a belt, coated with a thermosensitive adhesive. Hot rocks would stick, cold ones not, so the different rocks would have different discharge characteristics from the belt. The idea went live and Steinert of Germany participated as a sorting machine producer. The machine was build in the September 2014 to May 2015 academic year and the final report is available on request. All of the work, concept, etc. are in the public domain. Conceptually, this approach will reduce sulphide ore sorting costs substantially, since this approach gets rid of the air blasting. Operating the air compressor is responsible for 70-80% of sorting machines operating costs.

Pre-concentration brings both processing benefits and good sustainability benefits with energy savings and reduced waste, particularly tailings and thus diminishing legacy issues. This shows the huge potential in successful pre-concentration, but also indicates the difficulties. Most operating plants and plants at the design stage have a big incentive to investigate existing pre-concentration processes and enabling instrumentation, but overall economics has usually been against it. Usually the boom-and bust scenario of the mining industry means that a long term sustainable view is not taken when the economics are calculated - high production short term with an inefficient plant is likely to give the 'best' result. This view exacerbates the ups and downs in the industry. Is there some way that the mining industry can be stabilized so that a long term view can be seen as the optimum, or is it in the interest of mining industry executives to over-produce in the good times and shut down when demand falls? In business terms, this strategy also removes most potential competition from the market.

Companies that can self fund a project are in a position to take a long term view allowing them to 'over capitalize' in order to get a better long term benefit. Projects which require financing have to satisfy the relatively short term repayment requirements of the financiers hence the resulting minimization of initial capital. Pre-concentration is an ideal all projects should pursue as far as practicable during the development phase but in the vast majority of cases, until new technologies become available, cold hard finance considerations will stop implementation.

In closing, this may help http://www.911metallurgist.com/geology-prospecting/technical-papers-about-blasting/