A Centrifugal Separator is a rotating-bowl device that uses high g-force to separate slurry particles by density, pulling heavier minerals outward against the bowl wall while lighter gangue overflows out the top. The Knelson Concentrator, used in gold rooms across operations like Barrick's Goldstrike and Newmont's Boddington, is the textbook example. The mechanism solves the problem of recovering fine, free gold and other dense minerals that escape jigs, spirals, and shaking tables. A well-tuned unit recovers 60-95% of liberated gold finer than 100 µm at feed rates up to 1000 t/h on the largest models.
Centrifugal Separator Interactive Calculator
Vary bowl speed, target g-field, and particle properties to size the effective separation radius and see the centrifugal field response.
Equation Used
The calculator uses the separator g-field relation: radial acceleration equals radius times angular speed squared. Enter the desired g level and bowl rpm to find the effective bowl radius needed at the active separation band. The capture index is normalized to the article example of 50 um gold at SG 19 in a 300 g field.
- Radius is the effective active separation radius in the rotating bowl.
- Target g-field is radial acceleration divided by standard gravity.
- Particle capture index is a simple normalized density-size-g indicator, not a full recovery model.
- Fluidization water, slurry viscosity, feed rate, and riffle geometry are not included.
How the Centrifugal Separator Actually Works
A Centrifugal Separator spins a conical or cylindrical bowl at speeds that generate 60 to 300 g of radial acceleration. Slurry enters axially at the bottom, climbs up the inside wall, and the dense particles get pinned outward into a series of riffles or rings while the lighter material flows out over the top lip. The whole point is to compress weeks of natural gravity settling into a few seconds — you are not separating by size, you are separating by density. That is why a 50 µm gold particle (SG 19) reports to concentrate while a 200 µm quartz particle (SG 2.65) gets carried out as tails.
The geometry matters. In a Knelson, the bowl is fluted with concentrated rings, and pressurised back-water (called fluidization water) injects through small holes in each ring at 0.5 to 1.5 psi. That water keeps the heavy bed loose enough to let new dense particles bump in and displace lighter ones already trapped — without it, the rings pack solid in 30 seconds and the unit chokes. Pressure off by 0.3 psi and you either lose gold to tails (too high) or compact the bed and stop accepting feed (too low).
If bowl speed drifts low, you lose fine gold below 75 µm because the radial g-force can't overcome viscous drag. Push speed too high and the riffle bed scours — you'll see gold reporting to tails on the assay. Common failure modes are worn polyurethane bowl liners (lifespan around 2000-4000 hours on hard ores), seized fluidization water solenoids, and feed-tube wear that disrupts the axial entry pattern.
Key Components
- Rotating Bowl: The conical or cylindrical vessel that spins at 400-2000 RPM to generate 60-300 g. Knelson XD bowls run polyurethane liners replaced every 2000-4000 hours; Falcon SB bowls are smooth-walled stainless. Concentricity matters — runout above 0.5 mm causes vibration and uneven concentrate banding.
- Fluidization Rings (Knelson) or Smooth Wall (Falcon): Knelson rings are machined with holes typically 1.0-1.5 mm diameter at 8-12 mm spacing. Falcon SB uses no fluidization rings — the smooth wall relies on a flowing-film effect at higher g (200-300 g). Hole drift above 0.2 mm starts losing fine particles.
- Fluidization Water Manifold: Delivers clean process water at 0.5-1.5 psi backflow through the bowl rings. Flow rates run 7-150 L/min depending on machine size (KC-CD12 up to KC-XD70). Solenoid-actuated for batch flush cycles every 30-60 minutes in semi-continuous units.
- Drive Motor and Variable Frequency Drive: Typically a 5-150 kW motor with VFD letting the operator dial g-force from 60 to 200+ g. The VFD ramp-up rate matters — too fast and you spike the bearings, too slow and concentrate banding shifts during start-up.
- Feed Tube and Distributor Cone: Brings slurry axially into the bowl base and distributes it evenly outward. Wear here is constant — silica sand at 30% solids will eat a mild-steel feed tube in 200 hours. Most operators run ceramic or chromium-carbide-clad tubes.
- Concentrate Discharge Valve: Pneumatic pinch valve or rotary discharge that empties the trapped concentrate at a programmed cycle. Cycle times are tuned to ore — high-grade gold ore might flush every 15 minutes, low-grade tantalum every 4-8 hours.
Real-World Applications of the Centrifugal Separator
Centrifugal Separators dominate fine-particle gravity concentration anywhere a density contrast above 2.5 exists. They earn their place in gold rooms, tin and tungsten circuits, tantalum recovery, and increasingly in rare-earth and critical-mineral plants. The choice between a Knelson and a Falcon usually comes down to particle size, feed grade, and whether you need fluidization to deal with sticky clays.
- Gold Mining: Knelson KC-XD48 concentrators in the gravity room at Newmont's Boddington mine in Western Australia, recovering free gold from cyclone underflow ahead of the CIL circuit.
- Artisanal and Small-Scale Mining: iCON IGR-100 portable centrifugal concentrators replacing mercury amalgamation in small Peruvian and Colombian gold operations under planetGOLD programmes.
- Tin and Tantalum: Falcon SB2500 concentrators at Wodgina and Greenbushes-area tantalum operations, recovering coltan fines below 100 µm that escape spirals.
- Coal Fines Recovery: Falcon C-series concentrators at Australian and South African coal washeries, separating fine clean coal from pyritic ash in the -150 µm fraction.
- Platinum Group Metals: Knelson concentrators at Stillwater Mining in Montana for recovery of native PGM grains from milling circuits.
- Rare Earth and Critical Minerals: Falcon UF and SB units in pilot circuits for monazite and xenotime recovery at Lynas Mt Weld and various North American REE projects.
The Formula Behind the Centrifugal Separator
The single most important number for a Centrifugal Separator is the centrifugal acceleration generated at the bowl wall, expressed as multiples of gravity (g-force). At the low end of the typical operating range — around 60 g — you have a fluidized bed that is gentle on coarse, friable ores but loses fine gold below 50 µm. At the high end — 200 to 300 g, where Falcon SB units operate — you capture sub-20 µm particles but you also drag in middlings and dilute the concentrate. The sweet spot for most free-milling gold circuits sits at 90-120 g. This formula tells you what g-force you actually generate at a given bowl speed and radius, which is the number you tune against assay results.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| G | Centrifugal acceleration expressed as multiples of gravity (g-force) | dimensionless (×g) | dimensionless (×g) |
| r | Effective bowl radius at the concentrate ring | m | ft |
| N | Bowl rotational speed | RPM | RPM |
| g | Gravitational acceleration | 9.81 m/s2 | 32.2 ft/s2 |
Worked Example: Centrifugal Separator in a West African artisanal gold plant
A cooperative in eastern Ghana is commissioning a refurbished Knelson KC-CD20 (20-inch / 0.508 m bowl diameter, effective ring radius about 0.22 m) to recover free gold from alluvial tailings reprocessing. They want to know what g-force they generate at the manufacturer's nominal 60 Hz drive setting (about 720 RPM at the bowl), and how the recovery picture changes if they slow it down to handle a coarser feed or push it harder for fine values.
Given
- r = 0.22 m
- Nnom = 720 RPM
- Nlow = 500 RPM
- Nhigh = 950 RPM
Solution
Step 1 — calculate the nominal g-force at the manufacturer's 720 RPM setting:
127 g is the textbook Knelson sweet spot — heavy enough to pin gold finer than 50 µm against the rings, light enough that the fluidization water can still keep the bed loose. At this setting on a clean alluvial feed you should see 80-90% recovery of liberated gold above 25 µm.
Step 2 — check the low end of the operator's likely range, 500 RPM (used when feed is coarser or contains sticky laterite clay):
61 g is right at the floor of useful operation. You'll keep coarse gold above 100 µm easily, but anything finer than 50 µm starts reporting to tails because viscous drag in the slurry beats radial acceleration. Expect overall recovery to drop to maybe 55-70% on a typical alluvial size distribution.
Step 3 — check the high end, 950 RPM (operator pushing for fine micron gold):
222 g is Falcon SB territory, not Knelson. The fluidization water at 1.0 psi can't keep up with that acceleration — the bed packs, fines bypass the rings, and you start scouring concentrate already trapped. In practice the operator will see concentrate grade fall and bearing temperatures climb above 70 °C within an hour.
Result
At 720 RPM the KC-CD20 generates a nominal 127 g — the right operating point for free gold recovery from alluvial reprocessing. Across the range the unit goes from 61 g at 500 RPM (gentle, coarse-only, 55-70% recovery) through 127 g nominal (the design sweet spot, 80-90% recovery) to 222 g at 950 RPM (over-driven, bed-packing, falling concentrate grade). If your measured g-force differs from the calculated value, check three things in this order: (1) bowl-shaft runout above 0.5 mm from worn lower bearings will cost you 5-10 g of effective acceleration through wobble losses; (2) a slipping V-belt or worn VFD encoder feeding back wrong RPM is the most common cause of a calculated-vs-actual mismatch on refurbished units; (3) effective radius drift — if the polyurethane ring liners are worn down by 3-4 mm, your r is no longer 0.22 m and the whole calculation shifts.
When to Use a Centrifugal Separator and When Not To
Centrifugal Separators are not the only way to separate by density, and they are not always the right answer. Spirals, shaking tables, and jigs all still exist for good reason. Pick the wrong one for your size fraction and you'll either burn capital on the wrong machine or watch your recovery curve flatten.
| Property | Centrifugal Separator (Knelson/Falcon) | Shaking Table (Wilfley) | Spiral Concentrator (Humphreys) |
|---|---|---|---|
| Effective particle size range | 10-1000 µm (Falcon UF down to 3 µm) | 75-2000 µm | 75-3000 µm |
| Capacity per unit | 1-1000 t/h (KC-CD3 to KC-XD70) | 0.5-2 t/h per deck | 1-7 t/h per start |
| G-force / separation driver | 60-300 g radial | 1 g + asymmetric shake | 1 g + flowing film |
| Capital cost (relative) | High ($150k-$2M) | Low ($15k-$50k) | Low ($5k-$30k per start) |
| Operator skill required | Medium-high (VFD tuning, fluidization) | High (deck dressing is an art) | Low (set splitters and walk away) |
| Best application fit | Fine free gold, tantalum, PGMs | Final concentrate cleaning | Coarse pre-concentration of heavy minerals |
| Maintenance interval (liner life) | 2000-4000 hours bowl liner | Decking surface 6-18 months | Polyurethane liner 5+ years |
| Water consumption | High (fluidization water 7-150 L/min) | Medium (wash water across deck) | Low (no added water beyond slurry) |
Frequently Asked Questions About Centrifugal Separator
This is almost always concentrate ring saturation. The rings have a finite holding volume — once heavy minerals fill them, new dense particles entering the bowl simply displace previously captured ones back into the tails stream, and your recovery curve collapses.
The fix is shorter cycle times. If you're running a semi-continuous KC unit on a high-grade or sulphide-heavy feed, drop your flush interval from 60 minutes to 20-30 minutes. On heavy black-sand alluvial feeds with high magnetite content, some operators flush every 10 minutes. You can confirm this is the cause by sampling concentrate ring grade at flush — if grade is high but tails grade is also climbing through the run, saturation is the diagnosis, not a tuning problem.
Falcon SB will outperform a Knelson on truly fine gold below about 30-40 µm because the higher g-force (200-300 g vs 60-150 g) overcomes the viscous drag that pins fine particles in the slurry stream. The Falcon's smooth-wall flowing-film design also handles ultra-fine feeds without the bed-packing problem you hit when you push a Knelson above 150 g.
The catch — Falcon SB has a much narrower operating window. It's intolerant of feed surges, intolerant of coarse oversize above 600 µm, and runs as a batch unit so you need surge tanks. Knelson handles a sloppier feed and tolerates coarser oversize. Rule of thumb: if your target gold is mostly above 50 µm and your feed has variable size distribution, run a Knelson. If you're hunting 10-30 µm liberated gold or tantalum slimes, Falcon SB earns its place.
Pressure at the gauge is not the same as pressure at the rings. The most common hidden cause is partial blockage of the fluidization holes by fine pyrite, magnetite, or precipitated iron — this is especially bad on water sources with dissolved iron. The rings near the top of the bowl block first because they see the lowest local flow.
Pull the bowl during a planned shutdown and back-flush each ring with clean water. If you find more than 10-15% of the holes blocked or scaled, your gold-loss problem is mechanical, not metallurgical. Operators in West African laterite operations see this constantly because process water is high in dissolved iron and sets up scale within weeks.
Don't size on the headline tonnage. Size on the gravity-recoverable gold (GRG) test result and the cyclone underflow split you actually plan to feed to the unit. Most gravity rooms only treat 20-40% of the cyclone underflow stream, not the full flow, because passing 100% gives diminishing returns once you've captured the free gold.
For 80 t/h total throughput with a 30% bleed to gravity, you need a unit rated for ~25 t/h — that's a Knelson KC-XD30 or KC-XD40 with headroom. Verify by running a GRG test at SGS or ALS Metallurgy first; a feed with 60% GRG justifies the larger unit, while a 25% GRG feed means a smaller machine plus a shaking-table cleaner is more economic.
Two likely causes, both mechanical. First — bowl liner wear. As the polyurethane rings wear, the riffle profile flattens and the bed becomes less selective. You'll keep catching gold but you'll also catch more middlings, so concentrate grade drifts downward while overall mass pull rises. Check liner profile against a new spare; replace at 3-4 mm wear.
Second — feed dilution. If your slurry pump or sump level controller drifts and feed solids drop from the spec 40-50% solids down to 25-30% solids, the bed never builds enough density gradient to reject light minerals properly. Both problems show up as falling concentrate grade with steady or rising mass pull, which is the diagnostic signature.
Yes, and that's exactly the case the iCON i150 and IGR-100 portable units were designed for under UN Environment's planetGOLD programme. A small centrifugal unit will recover 70-90% of the free gold mercury would have caught, with no mercury exposure to the miner or downstream river system.
The honest catch — centrifugal recovery falls off on very fine gold below 25 µm, which is exactly where mercury was most effective. Operations switching from mercury usually need a downstream cleaner step (a small shaking table or a Falcon UF) to catch the fines that the centrifugal unit misses. Going centrifugal-only without that polish step is the most common reason small operators report disappointing results after the switch.
References & Further Reading
- Wikipedia contributors. Knelson concentrator. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
