Is ore sorting the future? Part 1- The Technology
Can it be the technology to deliver more metal from exiting mines and undeveloped sub-economic deposits?
This piece is a little bit different as we are taking a technology rather than a deposit or geology centric perspective. The geology will quickly enter the fray as it always does.
A technological solution to imperfect orebodies
Periodically a company will make a discovery or start evaluating a historical resource and after some period, talk of “ore sorting” will begin. Generally, this is a very strong indication that there is something imperfect about the deposit. Most often this means that the grades are too low and/or the orebody is too small to support the costs of building a mine. Sometimes it can mean that some parts of the rock need to be separated from the desirable portions before milling and flotation/leaching, but in all cases, it means additional (undesired) complexity.
An interesting observation that one can make after studying several companies that have pursued ore sorting is that it is only ever considered late in processing studies. No one drilling out a high-grade discovery pursues ore sorting.
On the face of it this can be perplexing as the promise of ore sorting is one of reduced capital costs due to smaller processing plants, less environmental risks brought on by dry stack tailings, smaller footprints, less energy and generally more favorable economics. In all but a few special cases ore sorting does not deliver this. By identifying why this is the case we can then hope to be able to predict if (and when) it might change and be ready to profit from it. But before we get so bold, let’s look at what ore sorting is.
The history of ore sorting
Leaving out any official definitions, ore sorting is the act of intercepting the broken ore and host rocks and separating the two before subsequent concentrations processes or before the product is sold. In many accounts of historical deposits mined during the 1800’s, particularly supergene ones, production numbers are given as X number of tons at X concentration % metal of “Hand dressed ore”. Hand dressing in this case is the act of ore sorting, most often by what were known as “Pickey Boys” pictured below.
“Pickey boys” sorting copper ore at the Moonta-Wallaroo mines, South Australia.
One might be surprised to find out that this still happens today! At the ultra-high-grade Hishikari mine owned by Sumitomo, the ore (white veins) contrast strongly against the dark host rock and as a result significant material can be rejected visually and by hand from further processing.
Coarse ore sorting on a conveyor belt at the Hishikari mine, Japan.
Hishikari deposit Low-Sulphidation vein in development heading. Mine employee for scale.
This ore does not travel to a conventional processing plant as one might imagine, but rather it is moved by barge to one of the companies copper smelters where it is used as high-grade silica flux in the smelting process. As a result, the ore, in its entirety serves as a valuable consumable for the copper smelter and in the process the gold is recovered free of charge. A brilliant move by the company.
Despite this ingenuity when we speak of ore sorting in the modern sense this is not what we are referencing. Ore sorting is performed by machine sensors on conveyor belts in one of two ways as a) bulk sorting or b) particle sorting.
Bulk Sorting
Bulk sorting seeks to measure the concentrations of ore as it passes below a sensor on a conveyor belt (most often). Popular in this case are either PGNAA, PFTNAA and MR sensors. Examples of others include XRF that has also been placed in shovel buckets. These sensors are used most often to guide the processing plant as to the grade of the material they are receiving. Theoretically, the end of a conveyor belt carrying this ore can be fitted with a diverter box that would separate the material into either ore or waste taking into considerations some threshold. However, widespread adoption of this has not transpired. There are number of reasons but primary among them is the expectation that the material coming out of the pit and passing through the primary crusher should already be “all ore” and the waste should be landing elsewhere. Moreover, the “dwell time” or the time that the sensors need to take a reading is on the order of 30 seconds. In a medium sized copper mine, processing 2600 tons per hour, this equates to 21.66 tons per reading. On this scale of measurement, one would expect (unless someone has buggered up bad) that whatever material was not ore would be lost in all the material that is ore.
Conveyor Belt
Another limitation of bulk sorting is its inability to measure precious metal concentrations (Au, Ag, Pt, Pd) associated with the primary economic base metals at many mines. No sensors deployed in this way can measure these metals reliably at concentrations at which they are economic. Couple that with instrument error and what you have are a bunch of systems capable of instructing processing plants but unable to serve as the separators of ore and waste in many cases.
Some providers have gone as far as to produce mobile, in-pit, bulk sorter testing equipment. These are a new addition to the market, and I am uncertain if they are intended to demonstrate usefulness that could then be scaled up to the size of larger throughput systems or used as is. Either way they make for a practical approach to the problem.
Particle sorting
X-ray particle sorter promotional video from Tomra at a Austrian tungsten mine
Particle sorters on the other hand achieve excellent separation of (as the name implies) individual particles. They are quite remarkable machines really. Adapted from the recycling industry, particle sorters employ a sensor (RGB, Hyperspectral, XRT, Gamma) to image ore pieces as they travel on a conveyor belt and fly off it in a predictable trajectory. A computer then decides if each particle is ore or waste and interjects their trajectory with an air cannon to land in a different bin. Typically, ore is intercepted and the waste falls through. The results of this technology can be remarkable as the Mt. Carbine tungsten mine indicates.
The challenge with particle sorters is that a) their throughput is comparatively low and b) not all ores are suitable. For example, the Mt. Carbine mine has ample capacity to feed many sorters and their expansion of sorting capacity has seen production rise.
Hunting down exact figures for sorter throughput is more difficult than I expected. For example this Resources and Reserves report for Mt Carbine states that “The treatment rate will be 80 tph to achieve a annualized throughput of of 525,600 tones”. It is not clear if this is in reference to a single or multiple ore sorters. A further announcement gives quarterly production figures of 132,654 tones in a quarter that calculates out to 60.7 tones per hour assuming 24 hour production. This is unrealistic and if we factor is some down time a hourly production figure of 80 tph becomes a good approximation.
For the sake of discussion we can make the generous assumption that it is 60-80 tph for a single ore sorter. From this we can observe that the half-million tones that this annualizes to, is in effect half the tones that some of the worlds biggest mines process on a daily basis. Nevertheless, this performance is transformational for deposits such as Mt. Carbine.
The deposits where particle sorters have proved successful share a few characteristics:
Their ores are suitable for particle sorting.
They are either restated or existing operations.
I am unaware of any deposits that were developed from discovery with ore sorters as components of their processing plants other than possibly some diamond mines.
Ore suitability
Ore suitability can be expressed as the efficiency with which minerals of economic interest, can be separated from uneconomic host rocks at some coarse crush size (generally higher than 10 mm). This is essential as the smaller the rock is needed to be crushed and ground, the more it costs (as per the laws of comminution), and any savings provided by sorting are lost. Compare for example the two ores below.
The first is Mt. Carbine Tungsten ore:
Mt Carbine high-grade intersect source
And the second one a intersect from a Equadorian Porphyry copper prospect. I chose this example, in particular, for its inclusion of a micro view of fine-grained sulphides. What this demonstrates is that the mineralization is more evenly distributed in the Porphyry example.
The first image on the other hand shows grey unmineralized wallrock containing a quartz vein with black, bladed Wolframite crystals within it (shown by the green squares). Both the wall rock and most of the quartz vein contain no mineralization whatsoever.
As far as particle sorters are concerned the difference between the two rock types is vast, in one (Tungsten ore) the desirable mineral is coarse and concentrated in distinct zones within the rock, in the other it is fine grained and evenly distributed. The distinction that can be made therefore in crushing both rocks to some size, say 50mm is that in the first case the total metal budget will be in just a few particles whereas in the second it will be evenly distributed. This makes one type of ore “sortable” and not the other.
Evidence of this can be seen it the particle sorting test at Mt. Carbine. Where an astounding best result upgrading ratio of up to 28.05 times was achieved. This in effect turned the waste piles in question from weakly mineralized rock to a very high-grade tungsten ore as 83.67% of the metal budget was recovered and some 96% of the material was removed.
Restarted or existing operations
By way of background, the Mt. Carbine mine operated until the mid-80’s when tungsten prices collapsed, and it closed. It restarted some 10 years ago through a staged initial reprocessing of the tailings, followed by particle sorting of the waste piles and finally a dewatering and open pit start up.
The possible ore sorter streams at Mt. Carbine span open pit operation, low-grade stockpiles and “optical ore sorter rejects”. The latter are the products of previously used and less efficient optical ore sorters.
Mt Carbine ore sorter product
Another example of an ore sorter deployment is the former Antas North mine of Oz minerals in the Carajas district of Brazil. Here at the end of the mines life the company was left with about a million tons of low-grade material and the prospect of running it through ore sorters was examined to extract value ahead of closure. This may have included further plans to assess the peripheries of the deposits and additional pit pushbacks to access more marginal material should the method have proven feasible.
However, it’s not clear if this ever happened but judging by the mentions of ore sorting in both the resources and reserves statement and the annual report in 2021 and its absence in the following year, it is possible that this pursuit was not one of high economic appeal.
Low grade surface stockpile description and ore sorter performance
Finally, in the video above the gentlemen explains the mines use of ore sorters as being necessary due to its age and the exhaustion of high-grade ore and the need to process low-grade material.
I will attempt to unravel in Part 2 the reasons behind why ore sorting does not feature prominently in new mine developments yet has demonstrated transformational effects in existing and historical operations.
Part 2-Preview
Part 2 will be a deep dive examination of the economic impacts of ore sorting. The key objective of these technologies is to interject large amounts of low grade or unmineralized material before they need to be pushed through expensive secondary crushing circuits. Below as a teaser I have included the power consumption statistics from the Gruyere mine feasibility study, not because it is a ore sorting mine but because it demonstrated the massive escalation in cost from primary crushing to grinding.
Power capacity and consumption by purpose at the Gruyere Au mine
This is the value that ore sorting provides. The ability to reduce milling costs. It does not put any extra metal into the rock and therefore its economic impacts have a upper bound. More in Part 2.
Cheers,
CC
Disclaimer
Nothing here is financial advice, comes with no warranty and I take no responsibility for anyone’s financial decisions.