Gold Ore Processing Equipment and Software Compared

Gold ore processing equipment is chosen by the ore, not by the brand. Cyanide leaching has been the main gold extraction method since the 1970s, usually in a heap leach or a carbon-in-leach (CIL) mill, after crushing and grinding (Society for Mining, Metallurgy & Exploration briefing). Everything else on this page either feeds that step, protects it or replaces it for ores it cannot treat. Recovery shows how much the ore matters: in 2025 Barrick reported 86% recovery through the roasters at Nevada Gold Mines, 80% through its autoclaves, 78% through its oxide mill and 75% at Pueblo Viejo (Barrick Annual Report 2025).

This guide is written for mining business owners and plant decision-makers. It compares the main equipment and software choices stage by stage, says what each is best for, gives dated evidence, and lists what to check before you buy. It is not a ranking from best to worst, because a gravity circuit and an autoclave solve different problems.

How we chose and compared these tools

We followed the path of the ore through a plant: crushing and grinding, pre-concentration, gravity recovery, leaching, treatment of hard-to-leach ore, non-cyanide options, and the software that watches and controls the plant. For each stage we picked the technology a mine owner is most likely to be quoted, then looked for figures from operators’ annual reports, peer-reviewed or conference papers and test programmes.

• Evidence first: every number below has a named source and a date. Where the only figures are a company’s own test results, we say so.
• Ore-specific results: a result from one ore body does not transfer to yours. Treat all percentages as examples, and ask any supplier for test work on your own samples.
• No vendor price lists: equipment prices are not listed, because they vary by project and supplier. The calculator at the end lets you test the value of a recovery gain with your own numbers.
• Terms explained: grade is the amount of gold in the rock, in grams per tonne (g/t). Recovery is the share of that gold the plant captures. A flowsheet is the planned sequence of processing steps.

Gold ore processing equipment and software compared at a glance

Which flowsheet fits your ore

Start with the ore, not the equipment catalogue. Six questions narrow the choice before you ask for any quote.

Infographic

Gold ore path through a plant in 5 steps

1. 1Crush and grindHPGR or SAG mills reduce the rock to size.
2. 2Sort or concentrateSorters reject waste; gravity takes coarse gold.
3. 3Treat refractory oreRoast or autoclave sulfide ore if needed.
4. 4Leach and captureCyanide or thiosulfate dissolves gold; carbon or resin captures it.
5. 5Recover and refineGold is stripped, melted into doré and refined.

1. Is some of the gold coarse? If so, put a gravity circuit in the grinding loop first. Gold is very dense, so it separates cleanly from host rock (ALS Metallurgy).
2. Is the ore hard and competent? Consider HPGR, either as the main crusher or to take pebbles out of an overloaded SAG (semi-autogenous grinding) mill circuit.
3. Does a large share of the rock carry no gold? Test sorting before you size the mill.
4. Is the gold locked in sulfide minerals? That is refractory ore, and direct cyanide leaching will recover too little. It needs roasting, pressure oxidation (autoclave treatment under heat and pressure) or another pre-treatment.
5. Does the ore contain carbon, or copper that eats cyanide? Carbonaceous and preg-robbing ores (ores that pull dissolved gold back out of solution) call for a different circuit design or a different leach reagent.
6. Is the tonnage large and the grade low? Heap leaching may suit it, but the choice should rest on a cost comparison and not on a rule of thumb.

That last point has a long record. A 1984 AusIMM case study of small Australian deposits near 2-3 g/t found that the common habit of assuming heap leach below some grade and carbon-in-pulp above it could mislead, and it ran a worked comparison for a 3-million-tonne deposit at 3 g/t instead (Battaglene, Noakes and Willsteed, 1984). The same logic applies to every choice here: test the ore, price the alternatives, then decide.

The flowsheet is also where cost is set. Our video follows one nugget through every step, from crushing to refining.

▶ Follows one gold nugget from rock to ring: crushing, cyanide leaching and carbon, heap leaching, electrowinning and refining (5:27). · 5:27

1. High-pressure grinding rolls (HPGR)

Best for: hard ore and plants where the grinding circuit limits throughput.

Chart

Modelled HPGR energy saving vs a SAG-ball mill circuit

Percent lower energy per tonne, two hard-ore cases; equipment only vs whole circuit

An HPGR squeezes rock between two counter-rotating rolls. Weir says the technique produces more fines, which reduces the energy needed for later milling (Mining Technology). A refereed paper in the Journal of the Southern African Institute of Mining and Metallurgy modelled four complete circuits for two hard-ore cases and compared HPGR plants with SAG-ball-crusher (SABC) plants. Counting only the comminution equipment (crushing and grinding), the HPGR circuits used 25.1% and 30.2% less energy. Once conveyors, screens, pumps and dust systems were added, the whole-circuit saving fell to 7.7% and 18.4% (Rosario and Hall, 2010). The paper also expects lower steel use, because HPGR circuits drop the SAG mill grinding balls.

That is the figure to remember when a supplier quotes an energy saving: ask whether it covers the machine or the whole circuit. The same paper lists gold plants that had been designed around HPGR by 2010, including Boddington and Bendigo in Australia and Zapadnoe in Russia.

A 2026 example shows the other use. Weir has been ordered to supply two Enduron HPGRs to Zijin’s Akyem mine in Ghana, where the SAG and ball-mill circuit is described as struggling with harder ore. The rolls will crush pebbles to clear a bottleneck, and Weir expects higher throughput and recovery (Mining Technology, order recorded in Q3 2026). Those benefits are the supplier’s expectation, not a measured result.

Pros: lower modelled energy per tonne, less grinding steel, and a way to unblock an existing mill. Cons: the saving depends on the ore and on the extra equipment around the rolls, and a very wet or clay-rich feed needs separate testing.

2. XRT ore sorters

Best for: ore where gold travels with a denser mineral such as sulfides and the rest of the rock is barren.

Chart

Mass pull and gold recovery by particle size at Cariboo

Mass pull (share of feed kept) against gold recovery, three size fractions

An X-ray transmission (XRT) sorter scans each rock as it falls past a sensor and fires a jet of air to eject the ones that look like waste. The aim is to send less rock to the mill, so the mill sees a higher grade, or to treat more ore through the same mill. A conference paper on sensor-based sorting of AngloGold Ashanti ores reported that gold recovery above 95% looked possible at a 40% mass pull (the share of feed that is kept) using one or two stages of pre-concentration (Dumont, Lemos and Robben, 2019).

A recent public bulk test came from the Cariboo Gold Project in British Columbia, run by the Saskatchewan Research Council with TOMRA Sorting. About 80 tonnes of material went through a production-scale TOMRA COM 1200 machine. The 10-30 mm and 30-70 mm fractions kept 89% and 84% of the gold at mass pulls of 45% and 58%, and a previously untested 6-10 mm fraction kept 92% at a 41% mass pull. The company says the sorter can divert marginal material for roughly C$1-2 per tonne (Osisko Development news release, 7 July 2025; the same-day correction changed the 6-10 mm recovery to 92%). Osisko Development is now named Osisko Gold.

The same release carries the caution. About 76% of the gold in the test sample was estimated to sit in particles of 10 mm or smaller, which bypassed the sorter in the main test, against 45% in the 2025 feasibility study. The sorter only works on the coarse fraction, so the real benefit depends on how your ore breaks.

Pros: rejects waste early and can open up mill capacity. Cons: needs a representative bulk test, and ore that breaks into fines gets less from it.

3. Gravity concentrators for coarse gold

Best for: ore with coarse or well-liberated gold that you want to capture before any chemicals.

Gold is far denser than the rock around it, at about 19.3 against 2.8-4.0, so it can be separated by spinning the slurry in a centrifugal concentrator. In industrial plants these units normally sit inside the grinding circuit and treat the cyclone underflow or the mill discharge. ALS Metallurgy runs both Knelson and Falcon machines in its labs (ALS Metallurgy).

The test that sizes the circuit is the gravity-recoverable gold (GRG) test. ALS describes the standard version as a 50-100 kg sample passed through a Knelson concentrator at three successive grind sizes, from 850 micrometres down to 75 micrometres, with each concentrate and the final tailings assayed. It warns that bench tests tend to over-recover compared with a full-scale plant, and that treating concentrate by intensive cyanidation also dissolves fine, partly liberated gold, which can overstate the GRG figure.

Pros: low chemical use, early cash from coarse gold, and a result you can measure on a sample before buying. Cons: it captures only the gold that is liberated and coarse enough, so it is usually a first step and not the whole flowsheet.

4. Leaching circuits: CIL, carbon-in-pulp and heap leach

Best for: most oxide and easily leached ore once it is ground or stacked.

Cyanide leaching dissolves gold in a weak solution, typically 100 to 500 parts per million cyanide (0.01% to 0.05%). Two arrangements dominate: milling with carbon-in-leach tanks, and heap leaching (SME). Barrick defines heap leaching as broken ore stacked on sloping impermeable pads with a weak cyanide solution applied continuously, and the gold-bearing solution then collected. In a CIL circuit, carbon sits in the leach tanks and picks up the dissolved gold as it forms.

Cyanide use is regulated and audited. SME notes that the International Cyanide Management Code requires tailings solutions to stay below 50 ppm cyanide to protect wildlife, that many mines achieve less than 10 ppm, and that certified operations are audited every three years by independent professionals. The mining industry uses about 6% of the cyanide produced worldwide.

The main equipment risk is the ore itself. Pueblo Viejo, which Barrick operates, produced 379,000 ounces in 2025, in the lower half of its 370,000-410,000 ounce guidance, and Barrick blamed lower CIL recovery from higher than planned copper and preg-robbing ores in the feed blend (Barrick Annual Report 2025). Recovery there fell from 79% in 2024 to 75% in 2025.

We cover the chemistry in more detail in our guide to gold extraction by the cyanide process, and the earlier steps in our overview of gold ore processing methods and gold separation.

Our tips for sustainable gold processing cover water, reagent and energy use around these circuits.

Pros: a proven method with a long record. Cons: recovery drops when carbon, copper or tight sulfide locking appear in the feed, and a cyanide plant needs a tailings and water plan.

5. Roasters and autoclaves for refractory ore

Best for: sulfide ore where the gold is locked inside pyrite or arsenopyrite and plain cyanide leaching recovers little.

Chart

Reported gold recovery by plant or processing route, 2025

Percent of gold recovered; Nevada Gold Mines at Barrick’s 61.5% share, Pueblo Viejo at 60%

Barrick’s glossary defines the two routes. Roasting treats sulfide ore with heat and air or oxygen-enriched air to oxidise the sulfides and remove other elements such as carbon, antimony or arsenic. An autoclave applies high temperature and pressure to turn refractory sulfide into ore that leaches. Either way, the gold is leached afterwards.

Nevada Gold Mines reports all three routes in one portfolio, which allows a rough comparison. In 2025 its roasters treated 5.26 million tonnes at 86% recovery, its autoclaves 4.24 million tonnes at 80%, and its oxide mill 7.68 million tonnes at 78%, all at Barrick’s 61.5% share (Barrick Annual Report 2025). Different ore goes to different plants, so these figures do not say that a roaster beats an autoclave on the same rock. They show what each route achieved on the ore it was given. Nevada is also the centre of US gold output, with about 64% of 2025 production (USGS Mineral Commodity Summaries 2026).

Pros: turns ore that would otherwise be waste into payable gold. Cons: high capital cost, off-gas or oxygen handling, and strong dependence on mineralogy. Always run pilot-scale tests on your own ore, because published results vary widely between deposits.

6. Options without cyanide: thiosulfate leaching

Best for: carbonaceous ore that holds dissolved gold back, or sites where cyanide cannot be used.

Barrick became the first producer to commercialise thiosulfate recovery of gold at Goldstrike in Nevada, with the first 107-ounce doré bar poured in November 2014 after more than 20 years of development and a US$620 million project. The circuit keeps the crushing, grinding and autoclave stages, then replaces the CIL circuit with seven tanks containing calcium thiosulfate and resin (Canadian Mining Journal, May 2015). Coverage of the CSIRO work in June 2015 said it was ramping up to recover gold from four million tonnes of stockpiled ore that traditional methods could not treat economically (Mining.com).

CSIRO has also developed its own thiosulfate-based reagent aimed at places where cyanide cannot be used and at small, high-grade deposits, tested in vat, heap and in-situ formats. A demonstration plant treating 50 tonnes of ore per vat ran successfully for more than six months from March 2018 (Dai and others, 2019).

Pros: avoids cyanide and handles ore that cyanide struggles with. Cons: far fewer plants have run it, and reagent consumption and chemistry control take careful engineering.

7. Process control and digital twin software

Best for: plants that already record reliable data and see their ore change from week to week.

The software layer does three jobs: it records what the plant is doing, it predicts what will happen if a setting changes, and it adjusts settings automatically. A digital twin is a model of a plant that predicts a target, such as gold lost to tailings, from the control settings. A 2019 AusIMM paper built a physics-plus-AI twin of a CIL plant from real data at Barrick’s Cortez plant. Its predictions reached an r-squared (a 0-to-1 score of how closely the model matches real results) near 0.9, and the authors point to real-time cyanide dosing as a use: keep enough cyanide in the circuit to avoid recovery loss, without overdosing and then paying to treat the tailings to meet permit limits (Bhongale and Moyo, 2019).

The need is real. Barrick’s 2025 report attributes Pueblo Viejo’s lower fourth-quarter output to lagging recovery performance of historically stockpiled material in its flotation, autoclave and CIL circuits. When feed changes that much, a plant model and a feed forecast can help the operator react. Our piece on mining process optimization looks at equipment performance data, and our guide to the gold mining value chain places the plant in the wider chain.

Pros: targets reagent and energy use, and gives early warning when feed changes. Cons: the model is only as good as the sensors and assays behind it, and a twin built for one plant does not move to another. Treat any vendor percentage as a claim to test in a trial; the only study cited here is a research model of one plant.

▶ Covers AI apps for finding targets, planning the mine and running the site; the plant-control tools above are one part of the third job (3:52). · 3:52

8. Tailings, energy and traceability records

Best for: owners who sell to refiners or lenders that audit the supply chain.

London’s LBMA requires every Good Delivery refiner to pass an independent third-party audit every 12 months under a programme that follows the five-step OECD due diligence framework (LBMA Responsible Sourcing). The programme is built to show that refined gold was responsibly sourced, so tailings, energy and fuel records and the weight and assay of each lot are the plant data a buyer may request.

The LBMA page names no particular software, so the choice is yours. Farmonaut does not make or sell processing equipment.

Where satellite screening fits before the plant

Plant design starts with the orebody, and the orebody starts with exploration. Satellites cannot see gold itself, but they can map the iron stains, clay minerals and faults that surround a gold system, which helps decide where to drill. That tells you where the ore may be and what kind of rock it sits in, long before you choose between a gravity circuit and an autoclave.

Farmonaut screens licence areas from satellite data and returns target zones with georeferenced files, usually in 5 to 20 business days. The results are exploration targets, not a mineral resource, and they do not replace drilling, sampling or metallurgical test work on real ore.

▶ Why no satellite sees gold directly, and how alteration halos are mapped to guide exploration (2:00). · 2:00

What a recovery gain is worth

Every item on this page comes down to one question: how much more gold leaves the plant, and what does that cost? Kitco showed a gold bid of US$4,157.70 per troy ounce on 15 October 2026 at 2:50 p.m. EDT (Kitco). At that price one extra percentage point of recovery on a plant treating 1 million tonnes a year at 2 g/t is worth about 643 ounces a year. The calculator below lets you test your own tonnage, grade, recovery and price, and subtract the extra operating cost of the equipment.

What to ask a supplier before you buy

• Has the unit been tested on a bulk sample of my ore, and who ran the test?
• Is the quoted saving for the machine alone or for the whole circuit?
• What does the recovery or throughput guarantee cover, and what happens if the feed changes?
• What are the wear parts, spares and reagent needs per tonne?
• How much water and power does it use per tonne, and what tailings does it create?
• Where has it run for at least two years on similar ore, and can I speak to that operator?
• Which data must my plant record for the software to work?

Get written answers, then compare them against the evidence on this page.

For a closer look at a related question, see our guide: Copper Mining Software: 12 Tools Compared by Job.

Frequently asked questions

What equipment is used to process gold ore?

A typical plant uses crushers and mills to break the rock, often with an HPGR or SAG mill, then gravity concentrators for coarse gold, then leaching tanks with carbon or a heap leach pad. Refractory ore adds a roaster or autoclave first. Gold is then recovered from carbon, melted and refined (SME; Barrick).

What is the most common way to extract gold from ore?

Cyanide leaching. SME says it has been the dominant gold extraction technology since the 1970s, and that heap leaching and carbon-in-leach milling are the two most common ways to use it.

What is the difference between CIL and heap leaching?

CIL grinds ore to a slurry and leaches it in tanks that also hold carbon to capture the gold. Heap leaching stacks crushed ore on a pad and sprays it with weak cyanide, with the gold-bearing solution collected below. Heaps suit large, lower-grade tonnages, but a 1984 AusIMM study shows the choice should be based on a cost comparison, not a fixed grade.

What is refractory gold ore?

It is ore in which the gold is locked in sulfide minerals so that cyanide cannot reach it. It needs pre-treatment, such as roasting or an autoclave, which Barrick defines as heat and air for roasting, and high temperature and pressure for the autoclave.

How much gold does a plant actually recover?

It depends on the ore and the route. Barrick reported 86% through roasters, 80% through autoclaves and 78% through the oxide mill at Nevada Gold Mines in 2025, and 75% at Pueblo Viejo.

Does an HPGR save energy in a hard-rock mill?

In two modelled hard-ore cases the whole HPGR circuit used 7.7% and 18.4% less energy than a SAG-ball mill circuit, and the comminution equipment alone used 25.1% and 30.2% less (JSAIMM, 2010). Your result depends on the ore and the circuit.

How much waste can an ore sorter remove?

In the Cariboo test, 42-55% of waste was rejected on the 10-30 mm and 30-70 mm sizes with 84-89% gold recovery, and 59% on the 6-10 mm size with 92% recovery (Osisko, July 2025). Fine particles under 10 mm bypass the sorter and can hold most of the gold.

Can gold be processed without cyanide?

Yes, for some ores. Barrick’s Goldstrike mine produced its first gold with a thiosulfate circuit in November 2014, and CSIRO has demonstrated a thiosulfate reagent in vat tests. Few plants run it, and the chemistry needs close control.

What is a digital twin in a gold processing plant?

It is a computer model of the plant that predicts results, such as gold lost to tailings, from the control settings. A 2019 AusIMM paper described a hybrid twin of a CIL plant at Barrick’s Cortez operation with an r-squared near 0.9.

What technology is used to track gold supply chains?

The LBMA Responsible Sourcing Programme requires London Good Delivery refiners to pass an independent audit every 12 months, following the OECD five-step due diligence framework. It names no required software.

Does Farmonaut sell gold processing equipment?

No. Farmonaut provides satellite exploration screening. It does not make or sell processing equipment.