What the EPC acronym covers

EPC stands for Engineering, Procurement and Construction. In mineral processing it means a single contractor takes your ore and target output and delivers a working plant:

• Engineering – ore sampling and metallurgical testwork, flowsheet development, mass balance, equipment sizing, and full plant and civil design.
• Procurement – sourcing or manufacturing every machine in the flowsheet, from crushers and mills to flotation cells, separators and the dewatering circuit.
• Construction – civil works, mechanical and electrical installation, and commissioning until the plant runs to specification.

What the +M+O adds

EPC+M+O extends the scope past commissioning into the period where most projects actually struggle – the first months of production:

• Management – production management support, process optimization and troubleshooting as feed is introduced.
• Operation – operator and maintenance training, plus assistance through ramp-up until the plant hits its design grade and recovery.

This matters because a plant that is built correctly can still underperform if operators are not trained or the circuit is not tuned to the real ore. EPC+M+O keeps the designer involved until the numbers are met.

EPC vs traditional multi-contractor delivery

How a project typically runs

1. Ore testing. Representative samples are tested for grade, mineralogy, hardness and recovery by method. This is the foundation of every later decision.
2. Flowsheet and plant design. The metallurgical results drive the flowsheet, then a mass balance, equipment list and civil layout. Typical engineering takes a few months depending on plant size.
3. Manufacturing and procurement. Equipment is built, commonly over several months, and inspected before shipment.
4. Construction and commissioning. Civil works, installation, electrical, and wet commissioning until the plant reaches design throughput.
5. Ramp-up and operation. Operator training and process tuning until target grade and recovery are sustained.

Why ore testing comes first

Everything in an EPC project flows from the metallurgical testwork, which is why a serious provider insists on it before quoting equipment. The testwork establishes head grade and mineralogy, the recovery achievable by gravity, flotation and leaching, the Bond work index that sizes the grinding mill, and reagent consumption. Skipping or shortcutting this step is the most common cause of plants that miss their design grade or recovery. With good testwork, the flowsheet, equipment sizes and expected metallurgy are all anchored in data rather than assumption, and the owner gets a realistic production forecast before committing capital.

Who it is for

EPC+M+O suits owners who want one accountable partner and a predictable path to production, especially first-time developers, remote sites, or projects in regions where assembling a multi-contractor team is impractical. It is also valuable where the ore is variable and the flowsheet must be designed from testwork rather than copied from another mine.

What to prepare before engaging a provider

• A representative ore sample – ideally covering the grade and mineralogy variation across the deposit, not just the best material.
• Target throughput – tonnes per day and expected operating hours, with any planned expansion.
• Site conditions – water and power availability, climate, access and elevation, all of which shape the design.
• Product and offtake requirements – the concentrate grade or dore specification your buyer demands.

What is included in the scope

A complete EPC+M+O scope typically covers metallurgical testwork, process and detailed plant design, civil and structural design, equipment manufacturing and procurement, transport and logistics, on-site installation and electrical work, wet and dry commissioning, operator and maintenance training, spare-parts packages, and ramp-up support to design capacity. Clarifying the exact boundary – what the owner provides locally (civil labor, utilities, tailings facility) versus what the EPC provider delivers – avoids the gaps that plague split contracts. A clear scope matrix agreed up front, listing each work package and who owns it, is one of the simplest ways to keep a project on schedule and on budget.

What single-source delivery looks like in practice

Because one provider designs the flowsheet and manufactures the equipment, every machine is sized to the duty rather than bought off a generic spec. A gold project, for example, integrates crushing, a grinding circuit, gravity and gold extraction stages that all balance to the same tonnage. A complete build like a CIP gold processing plant arrives as one coordinated package rather than a parts list to assemble. The same applies to copper, lithium, iron and other ores across the full equipment range.

Xinhai’s EPC+M+O model

Xinhai has delivered mineral processing projects under the EPC+M+O model for over 18 years, with in-house metallurgical testing, plant design, a manufacturing works in Yantai, and field teams for installation, commissioning and operator training. Capacities are configurable from small-scale plants to several thousand tonnes per day, sized to your tested ore. To see the full scope and how a project is staged, read the EPC+M+O services page, then send your ore details through the contact page to begin testwork.

How the two approaches differ

In dry magnetic separation, dry ore passes over or through a magnetic drum or roll; magnetic particles cling to the drum surface and are carried away from the non-magnetic stream. No water is used. In wet magnetic separation, ore is fed as a slurry; the magnetic fraction is captured on a rotating drum in the magnetic field and washed off as a clean magnetic concentrate while gangue exits with the water.

Particle size is the deciding factor

The single most important variable is particle size. Dry separation works well on coarse particles, roughly above 1-2 mm, where individual grains move freely and do not clump. As particles get finer, they agglomerate, trap dust and behave erratically in a dry field, so separation efficiency falls. Wet separation handles fine particles far better because the slurry keeps them dispersed and mobile, which is why finely ground magnetite is almost always processed wet. Most iron ore must be ground fine to liberate the magnetite, so the bulk of magnetite beneficiation worldwide is wet.

Side-by-side comparison

When to choose wet magnetic separation

Wet separation is the default for upgrading finely ground magnetite to a high-grade concentrate, often 60-68% Fe depending on the ore. It produces a cleaner product with higher recovery on fine feed, avoids dust entirely, and integrates naturally with a wet grinding and classification circuit. The trade-offs are water consumption and the need to dewater the concentrate afterward. A high-intensity wet drum magnetic separator is the workhorse here; explore the full magnetic separators range for field-strength options.

When to choose dry magnetic separation

Dry separation makes sense when water is scarce or expensive, when the feed is naturally dry and coarse, or as a pre-concentration step that rejects waste before grinding, saving energy on the milling stage. Dry roll and drum separators are common for coarse magnetite cobbing and for some weakly magnetic ores at high field intensity. The main challenges are dust management and generally lower concentrate grade, which may need a wet cleaning stage to finish. The dry magnetic separator suits these coarse and arid-site duties.

Field intensity: LIMS vs WHIMS

Beyond wet versus dry, the field strength must match the ore’s magnetism. Strongly magnetic magnetite is recovered on low-intensity magnetic separators (LIMS), typically permanent-magnet drums of a few thousand gauss, which are efficient and cheap to run. Weakly magnetic iron minerals such as hematite, limonite, ilmenite and martite need high-intensity or high-gradient separators (WHIMS/HGMS) operating at much higher field strengths to be captured at all. Getting this wrong is a classic failure: a LIMS drum will leave most of a hematite ore in the tailings. Magnetic susceptibility testing on the actual ore tells you which class of separator and what field strength you need before any equipment is bought.

Grind size, liberation and middlings

Magnetic separation can only sort liberated grains. If magnetite is locked to silica gangue, a coarse grind leaves composite middlings that report partly to concentrate (diluting grade) and partly to tailings (losing iron). Grinding finer improves liberation and grade but costs energy and pushes the duty toward wet separation. The economic grind size is the point where extra liberation no longer pays for the extra milling – usually found by grinding a series of samples and running each through the separator. Multiple cleaning stages then lift grade by re-treating the rougher concentrate. Browse the grinding equipment range, since mill sizing and separator choice are decided together.

Many iron ore plants use both

A common high-efficiency strategy combines the two. Dry magnetic separation runs first as a pre-concentration step on coarse, crushed ore to reject barren gangue cheaply before grinding. The pre-concentrate is then ground and processed by wet magnetic separation to reach final grade. This sequence cuts grinding energy (you only grind ore that contains iron) while still achieving a clean, high-grade concentrate. The wet concentrate then goes to dewatering before shipment.

A typical magnetite flowsheet

1. Crush ore and run coarse dry magnetic separation to reject waste.
2. Grind the pre-concentrate to liberation size.
3. Upgrade with wet drum magnetic separation, often in roughing and cleaning stages.
4. Dewater the concentrate with a thickener and filter.

Match the separator to the ore and the site

The wet vs dry choice comes down to particle size, target grade, and how much water the site can supply. Fine magnetite for a high grade goes wet; coarse pre-concentration and water-scarce sites go dry; and many plants stage both. Xinhai sizes magnetic separators from your ore’s magnetic susceptibility and liberation size, and designs the surrounding grinding and dewatering circuit under one EPC+M+O contract. Send your ore details and target Fe grade through the contact page for a recommendation.

The dewatering chain: coarse to fine water removal

Water removal gets harder and more expensive the drier you go. A sensible circuit removes the easy water first by sedimentation, then tackles the bound water by pressure or vacuum filtration. The typical sequence is thickener then filter, with cyclones sometimes splitting coarse sand for separate dewatering. Each stage has a job:

• Thickening: bulk water recovery and slurry densification, the cheapest water you will ever remove.
• Filtration: final moisture reduction to a handleable or stackable cake.

Thickeners vs filter presses vs vacuum filters

Thickeners: the workhorse first stage

A deep cone thickener uses gravity sedimentation, aided by flocculant, to settle solids into a dense underflow while clarified overflow water returns to the plant. A deep-cone or high-rate design produces underflow at 45-65% solids and reclaims the bulk of process water, often 70-85% of incoming water, which is decisive in arid regions. Thickeners are the cheapest water removal per tonne and almost always come first. Where space allows, a thickener alone can feed a conventional tailings dam; where dry stacking is required, it pre-densifies feed for filtration and dramatically cuts the load on the downstream filter. See the full thickening and dewatering range for sizing options.

Thickeners are sized on settling-flux testwork, not rules of thumb, because settling rate and achievable underflow density depend heavily on particle size and flocculant response. The flocculant itself is a key operating cost and a key performance lever: the right type and dose can double the settling rate and add several percent to underflow solids, while an under-dosed or poorly mixed feed produces a dilute underflow that overloads the downstream filter. A deep-cone design pushes underflow density toward the high end, approaching paste consistency for some ores, which is why it is favored where dry or paste tailings are the goal.

Filter presses: the route to dry stack

A plate-and-frame filter press clamps filter cloths between plates and forces slurry through under pressure, producing a firm cake at 75-85% solids that can be trucked and stacked without a dam. The press operates in batches: fill, pressurize, optional membrane squeeze and air blow, then discharge. It delivers the lowest residual moisture and the clearest filtrate, which is why it is the standard for dry-stack tailings and for high-value concentrate where every percent of moisture costs freight. The trade-offs are higher capital cost, cloth replacement and batch cycle management.

Vacuum filters: continuous concentrate dewatering

A disc vacuum filter draws slurry onto rotating discs under vacuum, forming and discharging cake continuously at roughly 80-88% solids. Continuous operation suits steady, high-tonnage streams such as iron or copper concentrate, and ceramic-disc versions cut energy use sharply versus conventional cloth designs. Vacuum filters generally leave slightly more moisture than a pressure filter and depend on a reliable vacuum system, but their continuous output and lower per-tonne energy make them attractive where a stackable but not bone-dry cake is acceptable.

The ceramic-disc variant deserves a note because it changes the economics. Its microporous ceramic plates hold vacuum within the plate, so only a small vacuum pump is needed and air is not drawn through the cake as in a conventional cloth filter. The result is markedly lower power per tonne and very clear filtrate, at the cost of careful plate maintenance and acid cleaning to prevent blinding. For steady concentrate streams, the energy saving over the plant life can be substantial, which is why ceramic vacuum filters have become common on iron and copper concentrate duties.

How to choose and sequence

Start from the discharge requirement and work backward. If a permitted dam is available and water recovery is the goal, a thickener may be enough. If dry stacking is mandated, plan for a thickener plus filter press. For continuous concentrate at high tonnage, a thickener plus vacuum filter is often the lower-cost continuous option.

• Always thicken first. Feeding a filter at 50%+ solids instead of 25% can halve filter area and cost.
• Match flocculant to ore. Settling rate and underflow density depend heavily on flocculant type and dose; bench-test before sizing.
• Mind the fines. Clay-rich tailings settle slowly and filter slowly; a deep-cone thickener and paste-capable press handle them better.
• Recover water deliberately. Pair dewatering with slurry pumps and a closed water loop to minimize freshwater make-up.
• Plan for variability. Tailings characteristics shift as the orebody and grind change, so size equipment with headroom rather than at the average case.

One more consideration is increasingly decisive on new projects: the regulatory and closure picture. Dry-stack tailings remove the standing-water dam that drives the worst tailings-failure risks, and many jurisdictions now favor or require filtered tailings for new permits. That regulatory pressure, combined with the water-recovery savings, is why the thickener-plus-filter route is steadily displacing conventional dam-only schemes even where a dam would be technically adequate. Factoring closure cost and permitting risk into the comparison usually tilts the economics further toward filtration than capital cost alone suggests.

Dewatering is a system, not a single machine. Because Xinhai delivers the full circuit under one EPC+M+O contract, the thickener, filter and water-return are sized together against your actual tailings sample rather than bolted on at the end.

What you are trying to achieve

The target is a chemical-grade concentrate, conventionally 6% Li2O (5.5-6.5% range), with iron kept low because Fe2O3 is a penalty in converter feed. Run-of-mine spodumene ore commonly assays 0.8-1.5% Li2O, so the plant must reject a large mass of gangue, mainly quartz, feldspar and mica, while not losing the relatively heavy, brittle spodumene grains. Two properties make this possible: spodumene is denser (SG ~3.1-3.2) than the silicate gangue (SG ~2.6), and its surface can be selectively floated after careful conditioning.

The mass balance is unforgiving. Upgrading a 1% Li2O ore to a 6% concentrate means the concentrate is only about a sixth of the feed mass at best, so a large tonnage of gangue must be rejected cleanly without dragging spodumene to tailings. Two complications make spodumene harder than a textbook density or flotation separation. First, spodumene weathers to less recoverable forms near surface, so feed mineralogy varies with depth. Second, the surface chemistry of spodumene and the feldspar it must be separated from is similar, so flotation demands tight control of pH, conditioning and reagent dosing. These realities are why testwork, not a generic flowsheet, drives the design.

The flowsheet, stage by stage

1. Crushing

Three-stage crushing typically reduces ore to below 10-12 mm. A jaw crusher takes the primary duty and a cone crusher handles secondary and tertiary reduction. Spodumene is brittle, so crushing is kept controlled to avoid over-generating fines that are harder to treat by density.

2. Grinding and classification

Ore is ground to liberate spodumene from gangue, usually to a flotation feed around 65-80% passing 150-200 micron. A wet ball mill in closed circuit with a hydrocyclone or classifier controls the product size. Over-grinding is avoided because ultrafine spodumene floats and separates poorly.

3. Dense-media separation (optional, for coarse feed)

Where the ore liberates coarse, dense-media separation upgrades the coarse fraction cheaply by exploiting the SG difference, rejecting a large mass of light gangue before grinding and shrinking the flotation plant. DMS commonly produces a coarse 4-6% Li2O pre-concentrate and is a major cost lever when liberation allows it.

4. Desliming, mica and iron removal

Fines (slimes) are removed before flotation because they consume reagents and depress selectivity. Mica is floated off or removed ahead of spodumene flotation, and magnetic separation pulls iron-bearing minerals to protect concentrate grade. A wet drum magnetic separator is the standard tool for iron removal; see the full magnetic separation range.

5. Spodumene flotation

The deslimed pulp is conditioned, typically at elevated pH with a fatty-acid collector after surface activation, and spodumene is floated away from quartz and feldspar in a bank of cells. A mechanical flotation machine rougher, scavenger and cleaner train produces the final 6% Li2O concentrate. Flotation is essential for the fine fraction that DMS cannot treat. Explore the flotation equipment options for circuit sizing.

Conditioning is the make-or-break step. Spodumene surfaces are activated, often with a cation such as calcium and at high pH, before a fatty-acid or hydroxamate collector is added, so that spodumene floats while feldspar and quartz stay depressed. Reagent dosage, conditioning time and water quality all shift the selectivity, and small changes can swing the grade-recovery balance noticeably. Because the separation is this sensitive, the rougher concentrate is almost always cleaned in two or three stages, with cleaner tailings recirculated, to lift grade to specification without throwing away recoverable lithium.

6. Dewatering

The concentrate is thickened and filtered to a shippable moisture. A thickener recovers process water and a filter produces cake, while tailings are dewatered for storage. Water recovery matters because many lithium projects sit in arid regions.

DMS vs flotation: where each fits

Most modern hard-rock plants combine the two: DMS handles the coarse, well-liberated fraction at low cost and flotation recovers the fines, giving the best overall recovery. The right split depends entirely on liberation, which is why a metallurgical test program comes first. As a rule of thumb, the coarser and better-liberated the spodumene, the more of the upgrade work DMS can do cheaply, shrinking the flotation plant; finely intergrown ores push more mass into flotation and raise reagent cost. A heavy-liquid separation test on sized fractions quickly shows how much DMS can achieve before any flotation testing begins.

Typical performance and the role of testing

• Feed grade: 0.8-1.5% Li2O run-of-mine.
• Concentrate grade: 5.5-6.5% Li2O, with Fe2O3 controlled below the converter penalty.
• Overall recovery: typically 65-85%, depending on fines content and liberation.
• Key losses: ultrafine spodumene to slimes and unliberated middlings.

Because reagent scheme, grind size and the DMS/flotation split are all ore-specific, Xinhai begins every lithium project with bench and pilot testing, then designs the complete spodumene processing plant under an EPC+M+O contract so comminution, separation and dewatering are balanced to your deposit rather than assembled from generic units.

How each one works

A spiral classifier uses gravity settling in an inclined tank. Coarse particles settle and a slow-turning spiral rakes them up the incline to the mill feed, while fine particles overflow with the water at the lower end. It is a mechanical, low-speed device.

A hydrocyclone uses centrifugal force. Slurry is pumped tangentially into a conical body; coarse particles are thrown to the wall and report to the underflow (back to the mill), while fines exit the top as overflow. It has no moving parts but needs a feed pump.

Head-to-head comparison

When to choose a spiral classifier

Spiral classifiers shine where the separation is coarse and the operation values simplicity. They are forgiving, easy to operate, tolerant of surges, and need no feed pump, which makes them popular on small and medium plants and ahead of gravity concentration where a clean coarse overflow helps. The high-weir design suits coarser cuts; submerged designs handle finer overflow. The trade-offs are a large footprint, lower capacity per unit, and an inability to make very fine cuts. See the high weir spiral classifier for a typical coarse-duty unit.

When to choose a hydrocyclone

Hydrocyclones dominate modern, larger grinding circuits because they make a sharp, fine cut at high capacity in a tiny footprint, and a cluster of cyclones scales easily by adding units. They give tight control over circuit product size, which matters when downstream flotation or leaching needs a consistent P80. The costs are a feed pump (energy) and routine replacement of the apex and vortex finder liners, which wear in abrasive duty. For fine classification and high tonnage, the hydrocyclone separator is usually the better fit.

What drives a hydrocyclone’s cut size

A cyclone’s separation is tuned, not fixed. Cut size drops (gets finer) with smaller cyclone diameter, higher feed pressure and lower slurry density, and rises with a larger apex (spigot) opening. Operators adjust apex and vortex finder sizing to hold the target cut as ore and tonnage vary. This tunability is a real advantage, but it also means cyclones need a stable feed from a well-controlled pump and sump – surging feed gives a wandering cut. Apex wear gradually coarsens the cut, so liner condition is part of routine control, not just maintenance.

Capacity, wear and operating cost

Per unit of floor space, a hydrocyclone moves far more slurry than a spiral, and capacity scales simply by adding cyclones to a manifold or cluster. That density of throughput is why large concentrators favor them. The offsetting costs are pumping energy and wear-part replacement: apex and vortex finder liners in abrasive iron or hard-rock duty may need changing on a regular cycle. Spiral classifiers carry the opposite profile – low energy and slow, predictable wear on the spiral flights and tank liner, but a large footprint and modest capacity that make them impractical to scale to high tonnage. For small operations, the spiral’s lower complexity and absence of a feed pump often win on total cost of ownership, and its slow-moving mechanism is easy for less-experienced crews to run and maintain in remote locations where spare cyclone liners may be hard to source.

Quick selection guide

• Pick a spiral classifier if: the cut is coarse, the plant is small to medium, you want no feed pump and minimal maintenance, or you are feeding a gravity circuit.
• Pick a hydrocyclone if: you need a fine cut size, high capacity, a small footprint, or tight product-size control ahead of flotation or leaching.
• Consider both: some circuits use a spiral for coarse primary classification and cyclones for fine secondary control.

It is really a circuit decision

The classifier choice cannot be made in isolation from the mill it closes. Cut size sets the circulating load, which sets the effective mill capacity, which feeds back into mill sizing. A hydrocyclone making a fine cut will recirculate more coarse material and let the mill grind finer; a spiral making a coarse cut suits a coarser target. If you are still sizing the mill itself, see our guide on choosing a ball mill, and browse the full classifiers and hydrocyclones range alongside the grinding equipment so the two are matched.

Getting the cut size right

Specify classification from the product size your recovery process needs, then work back to cut size, circulating load and unit count. Xinhai sizes classifiers as part of the integrated grinding circuit under one EPC+M+O contract, so the classifier, mill and pump are balanced to your ore and target tonnage rather than picked from a catalog in isolation. Send your flowsheet and target P80 through the contact page for a sizing recommendation.

Match the plant to the ore first

Before any equipment is bought, two questions decide the whole design: is the gold free-milling or refractory, and is it coarse or fine. Free-milling, coarse gold can be recovered largely by gravity alone, which is the cheapest plant to build and run. Finely disseminated or sulphide-locked gold needs leaching, which adds tanks, reagents and an elution circuit but lifts recovery from perhaps 40-60% (gravity only) to 88-95% (gravity plus CIP). A 1-2 tonne metallurgical test on a representative sample is the single best investment you can make before ordering steel.

Two diagnostic tests are worth commissioning early. A gravity-recoverable-gold (GRG) test tells you how much of the gold can be won by gravity alone, which sets the size of the gravity stage and the cyanide you can avoid. A bottle-roll cyanidation test on the gravity tailing tells you the leach recovery and reagent consumption to expect. Together they convert the vague question of which plant into hard numbers: recovery, reagent cost per tonne and the split between gravity and leach. Skipping this step is the most expensive shortcut in small-scale gold, because it tends to surface only after the plant is built and underperforming.

The equipment list, stage by stage

1. Crushing and feeding

Run-of-mine ore is reduced to a millable size, usually below 15-25 mm. A small plant typically uses a single-stage or two-stage crusher with a feeder ahead of it.

• PE jaw crusher for primary reduction; a PE-400×600 handles roughly 15-60 t/h.
• Vibrating grizzly feeder to meter ore and scalp fines.
• A spring cone crusher as a second stage if a finer, tighter product is needed.

2. Grinding and classification

Grinding liberates the gold, typically to 60-80% passing 75 micron for leaching. A closed circuit of mill plus classifier keeps the product size consistent.

• Wet ball mill sized to your tonnage and target grind.
• Spiral classifier or a hydrocyclone to close the grinding circuit.

3. Gravity recovery

Free gold should be captured as soon as it is liberated, before it dissolves slowly or is lost. A gravity stage is cheap insurance and often recovers a third or more of the gold.

• Centrifugal concentrator in the grinding circuit to catch fine free gold.
• Shaking table to clean the gravity concentrate to a smeltable grade.

4. Leaching and recovery (for refractory or fine gold)

The gravity tail goes to a cyanide leach. A CIP plant adsorbs dissolved gold onto activated carbon in a train of agitated tanks, then an elution and electrowinning system strips the carbon and recovers gold for smelting. For the choice between CIP, CIL and heap leach, see our guide to CIL vs CIP vs heap leach.

A small leach circuit is usually a string of six to eight agitated tanks giving a total residence time of 18-30 hours, sized so the gold is fully dissolved and adsorbed before the pulp leaves the last tank. Lime is added to hold pH around 10-10.5, air or oxygen is sparged to supply the dissolved oxygen leaching needs, and carbon is advanced counter-current to the pulp so the loaded carbon is drawn from the first tank. Getting the tank count and residence time right is what separates a plant that hits its recovery target from one that chronically leaves gold in the tailing.

Indicative budget by plant size

The ranges below are equipment-only and indicative; they exclude civil works, freight, duties and installation. Frame them as starting points to scope your project, not quotes.

Hidden costs to budget for

• Civil and installation: often 20-40% on top of equipment cost.
• Power and water: grinding dominates power draw; secure supply or a genset early.
• Reagents: cyanide, lime and carbon are ongoing operating costs in a leach plant.
• Tailings and water return: a thickener pays back in recovered water and is increasingly required for permitting.
• Spares and wear parts: mill liners, balls, crusher jaws and pump parts.

Phasing a project on a tight budget

Cash-constrained projects often build in phases. A common approach is to commission a gravity-only plant first to generate early cash flow from the free gold, then add the leach circuit once the deposit and the cash position are proven. This works only if the layout is planned for the leach from the start, with tank pads, power and water sized for the final configuration; retrofitting a leach into a plant laid out for gravity alone wastes money. Plan the full flowsheet on paper even if you build it in stages.

Build it as one system

The most common small-plant mistake is buying machines piecemeal that do not balance, leaving a bottleneck that caps throughput. Because Xinhai supplies turnkey mineral processing plants under an EPC+M+O model, the crusher, mill, gravity and leach stages are sized against one another and against your ore test, and the package includes installation, commissioning and operator training so the plant reaches nameplate faster.

Know where the money goes

Before optimizing, understand the typical split. Grinding usually consumes the largest share of plant energy, often 40% or more. Cyanide and lime are the biggest reagent costs, with carbon, flocculant and grinding media adding up. The table below shows the main cost drivers and the most effective lever for each.

Lever 1: recover free gold by gravity first

The cheapest gold to recover is the gold you never leach. Installing a centrifugal concentrator in the grinding circuit captures coarse free gold as soon as it is liberated, before it has to be dissolved. A shaking table cleans that gravity concentrate to a smeltable product. Removing free gold ahead of the leach cuts cyanide consumption, shortens leach time and reduces the gold inventory tied up in carbon. For many free-milling ores this single change is the biggest cost reducer available. See the gravity concentration range for sizing.

Lever 2: grind to liberation, not finer

Over-grinding burns power and generates slimes that slow settling and raise reagent demand, while under-grinding leaves gold locked and uextractable. The optimum, often 70-80% passing 75 micron for free-milling ore, comes from liberation testwork. Running a closed grinding circuit with a well-tuned hydrocyclone keeps the product size consistent and avoids both over- and under-grinding. A few micron of unnecessary fineness can add measurable cost per tonne across the life of a plant.

Lever 3: dose cyanide and oxygen to demand

Cyanide is consumed not only by gold but by cyanicides such as copper and reactive sulphides. Measuring free cyanide and titrating to a target, rather than running a fixed high addition, prevents the costly habit of over-dosing for safety margin. Likewise, gold dissolution needs dissolved oxygen; matching aeration or oxygen injection to the actual leach kinetics in the early tanks, where demand is highest, speeds leaching and lets you hold cyanide lower. Modern leaching agitation tanks with efficient aeration help here.

The two reagents interact, which is where real savings hide. Gold dissolution depends on both free cyanide and dissolved oxygen, and if oxygen is the limiting factor, adding more cyanide does nothing but raise cost and feed cyanicides. Many circuits are oxygen-starved in the first one or two tanks where leaching is fastest; supplying oxygen there, by sparging pure oxygen or by improving impeller aeration, lets the same recovery be reached at a lower cyanide concentration. A simple program of measuring dissolved oxygen and free cyanide profiles down the tank train usually reveals where reagent is being wasted, and the fix is operational rather than capital.

Lever 4: manage pH efficiently

Lime is added to hold pH around 10-10.5, which prevents cyanide from escaping as toxic HCN gas and protects against acid loss. Over-liming wastes reagent and can passivate gold surfaces; under-liming loses cyanide and creates a safety hazard. Efficient lime slaking and pH control instrumentation pay back quickly by keeping addition to what the circuit actually needs.

Lever 5: recycle water and carbon

• Water: a thickener returns most process water and, in a CIL/CIP plant, residual cyanide with it, lowering both make-up water and reagent make-up. This is decisive in arid regions.
• Carbon: efficient elution and regeneration in a well-run elution and electrowinning system keeps carbon activity high so less make-up carbon is needed.
• Tailings detox and reuse: recycling cyanide-bearing solution reduces fresh cyanide demand where regulations allow.

Lever 6: control grinding media and liner wear

Grinding is not only the largest energy consumer; steel media and mill liners are a steady consumable cost that often goes unmanaged. Matching ball size and charge to the feed, keeping the mill at its optimal filling and choosing liner profiles suited to the ore all reduce steel consumption per tonne and stabilize the grind, which in turn stabilizes downstream leach performance. Erratic grind size forces operators to over-dose reagents as a buffer, so tightening grinding control quietly lowers reagent cost as well as media cost. It is a good example of how energy and reagent savings are linked rather than independent.

CIL vs CIP and the cost picture

Circuit choice affects cost too. CIL combines leaching and adsorption in the same tanks, suiting ores with preg-robbing carbon, while CIP separates them and can be cheaper to operate on clean ores. The right choice depends on ore behavior; our guide to CIL vs CIP vs heap leach covers the trade-offs in detail. Either way, a properly sized gold extraction circuit with the right tank count avoids the cost penalty of carbon attrition and incomplete leaching.

Put it together

None of these levers requires exotic technology; they require a flowsheet designed for the ore and instrumentation to dose to demand. Combined, gravity pre-recovery, optimized grind, demand-based cyanide and oxygen, efficient pH control and water and carbon recycling commonly trim 10-30% off reagent and energy cost per ounce. Because Xinhai designs and builds the full circuit under an EPC+M+O contract, these efficiencies are built into the plant from the testwork stage rather than retrofitted later.

Start with the grinding duty

Three numbers anchor every ball mill specification:

• Throughput – dry tonnes per hour the mill must process at steady state.
• Feed size (F80) – the 80% passing size of the mill feed, typically the crusher or SAG product, often 6-20 mm for a ball mill.
• Product size (P80) – the 80% passing target needed by flotation, leaching or gravity, commonly 75-150 microns, finer for refractory or fine-grained ores.

The energy to go from F80 to P80 is estimated from the Bond ball mill work index (kWh/t), measured on your ore. That energy, multiplied by throughput, gives the mill power draw, which sets the mill size. A hard ore at 15-18 kWh/t needs a much bigger mill than a soft one at 8-10 kWh/t for the same tonnage.

Wet or dry milling?

Most wet concentrators grind wet because it suits downstream flotation and leaching, gives finer products at lower energy, and controls dust. Dry milling is reserved for moisture-sensitive products, some industrial minerals, and arid sites where water is scarce. If you are grinding ahead of cyanidation or flotation, a wet ball mill is almost always the right call. For dry fine grinding of non-metallics like limestone or barite, a Raymond roller mill often beats a dry ball mill on energy.

Overflow vs grate discharge

Overflow mills are simpler and give a finer, cleaner product, ideal as the final grinding stage before flotation or leaching. Grate-discharge mills hold a lower slurry level and discharge faster, giving higher throughput and a slightly coarser product, which suits the primary stage or where overgrinding must be avoided. As a rule of thumb, use grate discharge for coarse primary grinding and overflow for fine regrind or single-stage duties.

Sizing media, liners and the motor

Once the mill diameter and length are set, three wear-and-power choices follow:

• Grinding media: ball charge is typically 30-40% of mill volume. Top ball size scales with feed size and ore hardness; a mixed charge maintains an efficient size distribution. Steel consumption commonly runs 0.5-1.5 kg/t.
• Liners: rubber liners suit fine grinding and lighter media; steel or composite liners handle coarse, abrasive duties. Liner profile affects lifting action and energy efficiency.
• Drive: mill speed is set around 70-80% of critical speed. The motor is sized to the calculated power draw plus a margin for ore variability.

Where the mill sits in the circuit

A ball mill rarely works alone. It is paired with a classifier so coarse material is returned for regrinding while fines pass on, forming a closed circuit. Whether you close the circuit with a high weir spiral classifier or a hydrocyclone affects the achievable product size and circulating load. Browse the full ball mills and grinding machines range to match the mill to your front-end crushing and classification.

Open vs closed circuit and circulating load

Almost all production ball mills run in closed circuit, where oversize from the classifier returns to the mill feed. The circulating load – the ratio of recycled coarse material to fresh feed – is typically 200-350% in a well-tuned circuit. A higher circulating load lets the mill grind at a coarser internal size distribution, which is more energy-efficient and reduces overgrinding, but it demands more pumping and classifier capacity. Open-circuit grinding (no return) is simpler but produces a wider size distribution and is generally reserved for coarse or single-pass duties.

Single-stage vs two-stage grinding

A single ball mill can take crusher product directly to final size on softer ores or modest tonnages. For hard ores, fine products, or large throughput, a two-stage layout – a primary mill (often grate-discharge) followed by a secondary overflow mill, or a SAG mill ahead of a ball mill – spreads the duty and gives better control. The split between stages is set so neither mill is the sole bottleneck. This is decided during flowsheet design from the work index and target P80.

Common mistakes when specifying a mill

• Sizing on nameplate, not ore. Two ores at the same tonnage can need very different mills if their work indices differ. Always test the actual ore.
• Ignoring ore variability. Hardness changes with depth and zone. A motor sized only for average ore stalls on the hard fraction; build in a margin.
• Mismatching the classifier. A mill is only as good as the circuit it closes. Undersized cyclones or pumps cap real throughput regardless of mill size.
• Wrong liner for the duty. Steel liners in a fine-grind rubber-liner duty waste energy; rubber in a coarse abrasive duty wears out fast.

A practical selection checklist

1. Confirm throughput and operating hours per year.
2. Get F80 and P80 from your flowsheet and recovery requirements.
3. Run a Bond work index on representative ore.
4. Calculate power draw and select mill diameter and length.
5. Choose wet/dry and overflow/grate to suit the duty.
6. Specify media, liners and drive; confirm circulating load with the classifier.

Xinhai sizes mills from ore testwork rather than catalog defaults, with capacities configurable from small pilot units up to large production mills. Because we also design the surrounding crushing and classification circuit under one EPC+M+O contract, the mill, motor and wear parts are matched to your actual ore and target tonnage.

How gravity separation works

Every gravity device exploits the same principle: in a moving film of water, dense particles settle, lag or report differently than light particles. The usable density contrast is commonly expressed by the concentration criterion, (SG heavy – 1) / (SG light – 1). A value above about 2.5 means easy separation at almost any size; between 1.75 and 2.5 separation is feasible above roughly 0.15 mm; below 1.25 gravity alone struggles. Gold (SG 19), cassiterite (SG 7) and chromite (SG 4.5) separate readily from silica (SG 2.65); fine coal and some oxidized ores do not.

The two variables that decide everything

• Particle size: fine feed needs gentle, film-flow devices; coarse feed needs pulsing or tumbling action.
• Throughput vs. grade: high-tonnage roughing favors spirals and jigs; final cleaning to a sellable concentrate favors tables and centrifugal bowls.

The four main gravity devices compared

Shaking tables

A gold shaker table uses a riffled deck with asymmetric horizontal motion and a thin cross-flow of water. Dense particles migrate along the riffles to the concentrate end while light gangue washes over the edge. Tables give the sharpest separation of any gravity device and produce a clean, high-grade concentrate, which is why they are almost always the final cleaning stage in a small gold plant. The trade-off is low unit capacity, so tables are usually fed from a pre-concentrating device rather than raw ore. Browse the full range of gravity concentration equipment to see how tables pair with upstream units.

Spiral chutes

The spiral chute separator has no moving parts: pulp flows down a helical trough and centrifugal plus gravitational forces band the minerals across the channel, where splitters cut concentrate, middling and tailing. Spirals shine on fine, high-tonnage duties such as iron ore, chromite, ilmenite and zircon sands. They are cheap to run, easy to operate and scale simply by adding starts, making them a favorite roughing device ahead of magnetic or table cleaning.

Jigs

Jigs pulse water up through a bed of particles so dense grains stratify to the bottom and report through the screen. Because they handle coarse feed, jigs are ideal for alluvial deposits, coarse gold, tin and tungsten, and for pre-concentrating run-of-mine before grinding, which cuts downstream mill load. They tolerate variable feed and require little water relative to tonnage.

Centrifugal concentrators

A centrifugal concentrator applies many times the force of gravity in a fluidized, rotating bowl, capturing fine free gold that tables and spirals lose. Installed in or after the grinding circuit, it recovers gold as soon as it is liberated, often lifting overall plant recovery by several points and reducing the gold tied up in circulating load. It is the standard answer to the question, where does my fine gold go.

Centrifugal bowls run in two modes. Batch units periodically stop to rinse a high-grade concentrate and are common in small plants and for cleaning duty, while continuous units discharge concentrate on a timed cycle without stopping the feed, suiting larger throughputs. The captured concentrate is small in mass but high in grade, so it is almost always cleaned on a shaking table before smelting. Because the bowl applies tens of g, it recovers gold down to roughly 10-25 micron that a table would lose, which is exactly the fraction that otherwise reports slowly, or not at all, to a downstream leach.

Where gravity sits in the flowsheet

In most plants gravity is not the only step. It is a fast, cheap pre-concentration stage that pulls coarse and free values early, leaving a smaller, upgraded stream for flotation or leaching. A typical gold flowsheet runs the milled pulp through a centrifugal concentrator and table to recover free gold, then sends the gravity tail to a flotation circuit or a CIL/CIP leaching plant. This hybrid approach maximizes recovery while shrinking cyanide and reagent demand. For the full picture of how recovery routes combine, see our guide to gold recovery methods compared.

Wear, water and operating cost

Part of gravity’s appeal is low operating cost, but each device has its own wear and consumable profile worth budgeting. Spirals have no moving parts, so wear is limited to the polyurethane or rubber lining of the trough and the splitters, replaced periodically. Tables wear the deck surface and riffles and depend on a reliable head-motion drive. Jigs wear screens, diaphragms and the drive mechanism, and consume ragging where used. Centrifugal bowls wear the fluidization fittings and the bowl liner. None of these approaches the reagent and energy bill of flotation or leaching, which is precisely why gravity is placed first wherever the ore allows it.

Selection checklist

• Run a particle-size and SG analysis first; size dictates the device.
• Coarse and free values: jig or centrifugal up front.
• Fine, high tonnage: spirals for roughing, tables for cleaning.
• Always confirm the concentration criterion before committing to gravity alone.
• Test on a representative sample; Xinhai’s lab can recommend a flowsheet sized to your ore.

The first question: how does the gold occur?

Before comparing methods, characterize the ore. A few questions decide the flowsheet:

• Is the gold coarse and free-milling, or fine and disseminated?
• Is it associated with sulfides (pyrite, arsenopyrite) or free in quartz?
• Is there carbonaceous material that would re-adsorb dissolved gold?
• What is the head grade, and what recovery does the economics demand?

A gravity-recoverable gold (GRG) test and a mineralogical scan answer most of this and point to the right combination.

Gravity concentration

Gravity exploits the high density of gold (about 19 g/cm3) versus gangue (2.6-3 g/cm3). It is the cheapest method to run, uses no reagents, and recovers coarse free gold early so it does not get lost or over-ground. Centrifugal concentrators capture fine free gold down to tens of microns, while shaking tables and spiral chutes handle coarser fractions and produce a saleable concentrate. The limitation is that gravity cannot recover gold that is locked inside other minerals or too fine to settle.

Typical placement: a centrifugal concentrator in the grinding circuit to scalp free gold, with a shaking table to upgrade the gravity concentrate. See the full gravity concentration range for spirals and jigs.

Flotation

When gold is fine and associated with sulfides, flotation collects the gold-bearing sulfides into a concentrate that is a small fraction of the original mass. This is ideal for refractory or sulfide-hosted gold: instead of leaching the whole orebody, you leach (or smelt) only the concentrate, cutting reagent and energy cost dramatically. Flotation needs reagents and good liberation, and works best on sulfide minerals rather than free gold in oxide ore.

Cyanide leaching

Leaching dissolves gold chemically with dilute cyanide and recovers it onto activated carbon (CIL/CIP) or by zinc precipitation. It reaches the highest recoveries, 90-96% on amenable ores, and handles fine, disseminated gold that gravity and flotation miss. The trade-offs are reagent cost, residence time and the need for careful cyanide management. Leaching is applied either to the whole milled ore or, more economically, only to a flotation or gravity concentrate. See gold extraction equipment for leach tanks and the gold room.

Side-by-side comparison

Reagents and control in flotation

Flotation performance hinges on reagent selection and grind size. Collectors such as xanthates render the sulfide surfaces hydrophobic so they attach to air bubbles; frothers stabilize the bubble film; and modifiers (lime, copper sulfate, depressants) control pH and selectivity. Grind size must liberate the gold-bearing sulfides without overgrinding into slimes that float poorly. A typical sulfide gold flotation runs at a P80 around 75 microns, with the optimum confirmed by bench testwork. Because flotation rejects most of the gangue, the resulting concentrate is often 10-30 times higher grade than the feed, which is exactly what makes downstream leaching cheap.

Why most plants combine all three

A well-designed gold plant rarely relies on one method. A common high-recovery flowsheet runs gravity inside the grinding circuit to pull out coarse free gold first (which is hard to leach and easy to lose), then sends the rest to flotation or directly to leaching depending on mineralogy. For sulfide ores, gravity plus flotation concentrates the gold into a small mass that is then leached. This staged approach maximizes overall recovery while keeping reagent consumption proportional to the gold-bearing mass, not the whole orebody.

A typical free-milling flowsheet

1. Crush and grind to liberation size (often 75-106 microns).
2. Gravity scalp coarse free gold with a centrifugal concentrator.
3. Leach the gravity tailings by CIL/CIP.
4. Recover gold from loaded carbon by elution and electrowinning.

A typical refractory/sulfide flowsheet

1. Crush, grind and run gravity for any free gold.
2. Float the gold-bearing sulfides into a concentrate.
3. Treat the concentrate (leach, or oxidize then leach).

Reading recovery numbers correctly

Be careful comparing the recovery figures for each method, because they measure different things. Gravity recovery is quoted as a share of total contained gold and is inherently limited to the coarse free fraction, so 20-60% is normal and not a weakness – it simply reflects how much of the gold is recoverable by density. Flotation recovery is quoted to concentrate, where 85-95% is typical, but that concentrate still has to be treated to produce metal. Leaching recovery is the closest to a true overall figure. The number that ultimately matters is plant recovery across the whole flowsheet, which a well-staged combination maximizes by sending each gold form to the method best suited to it.

Cyanide management is part of the design

Any cyanidation circuit must address safety and environment from the outset: pH is held around 10.5-11 with lime to keep cyanide stable and avoid hydrogen cyanide release, and tailings are detoxified before discharge. These requirements influence reagent cost and permitting, and are a reason flotation pre-concentration is attractive – leaching a small high-grade concentrate uses far less cyanide than treating the whole orebody.

Matching method to ore is the whole game

Choosing gold recovery methods is really about reading the ore correctly. Coarse free gold to gravity, sulfide-locked gold to flotation, fine and disseminated gold to leaching, and almost always a combination. Xinhai runs the gravity, flotation and leach testwork in-house, then designs the integrated flowsheet and supplies the equipment under one EPC+M+O contract. For the CIL vs CIP vs heap leach decision specifically, see our gold processing plant comparison.